Methods and Principles of Systematic Zoology - VePub

Loading...
McGRAW-HILL PUBLICATIONS IN THE ZOOLOGICAL SCIENCES E. J. Boell,

METHODS AND PRINCIPLES

CONSUL TING EDITOR

OF SYSTEMATIC ZOOLOGY

Baitsell . HUMAN BIOLOGY Breland · MANUA L OF CoMPAR ATlVE ANATOM Y Chapman · ANIMAL EcoLOGY Clausen . ENTOMO PHAGOU S INSECTS Goldschmidt . PHYSIOL OGICAL GENETIC S Hau pt . FUNDAM ENTALS OF BIOLOGY Hyman . THE INVERTE BRATES: PROTOZO A THROUG H

CTENOP HORA (Vol. I) THE lNVER1'E BRATES: PLATYH ELMINTH ES AND RHYNCH OCOELA (Vol. II) THE INVERTE BRATES: AcANTH OCEPHA LA, AscHELM INTHEs, AND ENTOPRO CTA (Vol. III)



Johannsen and Butt . EMBRYO LOGY m· INSECTS AND M YRIAPOU S Leach . FmrnTIO NAL ANATOM Y OF THE MAMMA L Metcalf and Flint · FUNDAM ENTALS of INSECT Ln·1~ Mitchell · GENERA L PHYSIOL OGY Mitchell and Taylor ·LABORA TORY MANl!AL OJ" GENERA L Purn101. 0GY Pearse · ANIMAL EcoLOGY Quiring . FUNCTIO NAL ANATm n OF THE v ERTEBRA 'l'ES Rogers · TEXTBO OK OF Co::v!PAR ATIVE PHYSIOI, OGY Rogers · LABORA TORY Ou·rLINE S IN COMPAR ATIVE PHYSIOL OGY Shull · EvoLUTI ON Shull · HEREDIT Y Shull, LaRue, and Ruthven · PRINCIP LES m· ANIMAL B101,0GY Shull, LaRue, and Ruthven ·LABORA TORY DIRECTI ONS IN PmNcrp1 ;i;;s ANIMAL BIOLOGY

ERNS T MAYR Curator of the Whitney-Rothschild Collections The American Museum of Natural History

E. GORTON LINSLEY A.ssociate Professor of Entomology and Chairman Department of Entomology and Para.~itology University of California

ROBE RT L. USIN GER Associate Professor of Entomology University of California Alternate Member, International Commission on Zoological Nomenclat1ire, Paris 1948 OF

Snodgrass . PRINCIP I,ES OF lNSI•lCT MORPHO LOGY Storer · GENERA L ZooLOGY Storer . LABORA TORY MANUA L FOR GENERA J, ZOOLOG Y weichert ' ANATOMY OF THE CHORDATES Weichert ·ELEME NTS oF CHoRDA TF; ANATOM Y Welch · LIMNOLO GY Wieman · AN INTRODU CTION TO VERTEB RATE EMBRYO LOGY Wolcott · ANIMAL BIOLOGY Wolcott and Powell ·LABORA TORY GumE IN ANIMAL B101.onY There are also the related series of JJfcGraw-Hill Publications in the Botanical Sciences, of which Edmund W. Sinnott is Consulting Ediwr, and in the A.gricultural Sciences, of which R. A. Brink is Consulting Editor.

\'v\

NEW YORK

TORONT O

LONDON

McGRAW-HILL BOOK COMPANY, INC. 1953

3cp_!J,JL

-

am 43.0

PREFACE

METHODS AND PRINCIPLES OF SYSTEMATIC ZOOLOGY Co~yright, 1953, by the McGraw-Hill Book Company, Inc.

Printed in the United States of America. All rights reserved. This book, or parts thereof may not be reproduced in any form without permission of the publishers'. Library of Congress Catalog Card Number: 52-10335 II

}'

/

,)'

~

6HBAHOTE~

ioc111cxoro YnKJmn•;

,______.,

Hll.U. JlOMOIOIOla, /

?/f9!3698

THE MAPLE PRESS COMPANY, YORK, PA.

The authors have long felt the need for a treatise on the principles and methods of taxonomy. Such a work should be useful not only as an adjunct to teaching but also as a reference work for the practicing taxonomist and as a source of information to the general biologist. An analysis and full statement of the often disputed principles on which the taxonomic method is based are urgently needed. We share the view of O. W. Richards (1947) that "it is less the findings of taxonomy than its principles and methods which need to be taught" and understood. We believe that taxonomy is an important branch of biology which deals not only with the identification and classification of natural populations but with objectives that go well beyond these fundamental activities. The teaching of taxonomic theory and method has been a seriously neglected phase of biology. Most formal courses in systematics have concentrated upon the end products of taxonomic research and have not provided the student with a means for critically evaluating these end products or for tracing the steps by which, they were attained. An understanding of taxonomic theory and practice is essential not only to the beginning and the practicing taxonomist but to all those who draw upon the results of his studies. This is true to a greater or lesser extent for all biological sciences, but in particular for such fields as ecology, population genetics, comparative morphology, anthropology, comparative physiology, and applied biology. Sound taxonomy is a prerequisite to intelligent conclusions in all these fields. At the present time there is no book available that deals comprehensively with the principles and methods of taxonomy. Available works are merely commentaries on the International Rules, or they deal with selected phases of taxonomic theory with occasional reference to taxonomic practice. The treatment in this book of certain phases of systematic zoology has necessarily been restricted because of limitations of space. Collecting techniques, for example, are so diversely specialized in each group of animals and so completely covered in separate works that they are not discussed in detail. A full discussion of the phyla and classes of animals is considered beyond the scope of this work, although a listing is presented (Table 2). Nomenclature, although strictly a means to an end, has occupied a disproportionate part of the time and energy of taxonomists. One reason v

PREFACE

J>Rb'FACR

for this is that the subject is inherently complex and that revisions of the Rules become necessary from time to time, since the practicability of the Rules, like that of any other code of law, can be tested only by application. A more fundamental reason is that a basic philosophy or theory underlies the Rules of Nomenclature. This theory has not only tended to change in the course of years (as, for instance, with respect to the significance of types), but some of it could not be fully understood until the principles of taxonomy themselves were more clearly understood (e.g., treatment of infraspecific names). We feel, therefore, that a presentation of the Rules of Nomenclature would be incomplete which does not deal with the history of the field, or which omits a discussion of the basic principles. We have attempted to present both these aspects. On the other hand, it is not the purpose of this book to enter into nomenclatural eontroversies. Since at this writing there is no edition of the International Rules of Zoological Nomenclature which is accurate or up to date, we hope that the simplified review of the Rules in Part 3 of our book may prove to be especially useful. At the same time, the treatment is open to the criticism that it is an unofficial version of a highly technical and, at the moment, controversial subject. It has been our aim to make nomenclature comprehensible to the practicing taxonomist, leaving it to nomenclaturists to analyze the voluminous proceedings of the International Commission and to debate the various issues of the moment. In attempting to bring together the more important elements of modern taxonomic theory and practice, we have, of necessity, selected our materials primarily from the point of view of the student of living animals and have chosen illustrative examples with preference from our mvn work. The problems of the paleontologist, microbiologist, and botanist have been taken into consideration as far as practicable, but the materials of these groups are often sufficiently different to require different approaches to the solution of taxonomic problems. Nevertheless, there is much common ground of theory and method shared by the workers in these diverse fields, and it is to be hoped that at some time in the not too distant future all biological taxonomy may be viewed· as a single cohesive field. If this book, by focusing attention on the problems of the systematic zoologist, serves as a step in that direction, one of its goals will have been achieved. If it also assists in stimulating a more critical evaluation of taxonomic theory and methods and in a wider dissemination of knowledge concerning them, the authors will feel that their labors have been justified. It is well-nigh impossible to acknowledge sources in a book of this kind, which has grown out of the accumulated contacts and experiences of the three authors throughout their lives. Suffice it to say that our early teachers in Germany and at the University of California and om

colleagues at The American Museum of Natural Hit:1tory and in the Biosvstematists Discussion Group at the University of California and at Stanford University have done much to shape our thinking along the Jines expressed in this book. We also acknowledge the role of several generations of students at the University of California, who have unwittingly provided opportunity to test the clarity and effectiveness of portions of the manuscript during its formative stages. Their response has been most helpful. Formal acknowledgment of quoted material is made through literature citations. Special thanks are due to several colleagues who generously gave of their time to read portions of the manuscript. Their detailed suggestions and criticisms were carefully considered and were in most cases adopted. To these readers should go a large share of credit for accuracy of statements. On the other hand, the authors individually and rollectively assume the responsibility for the errors which undoubtedly will be discovered. The following persons read the chapters indicated: H. E. Blackwelder (1 to 17); E. Dougherty (10 to 16); Alden H. Miller (l to 9); C. F. W. Muesebeck (1to17); C. W. Sabrosky (1to17); M.A. Cazier (4, 5, 8, and 9); G. G. Simpson (7); L. M. Klauber (7); H. LPvenr (7), and R. F. Smith (7). Finally, we wish to express our sincere thanks to the secretaries, who meticulously typed the various drafts of the manuscript and helped in checking the bibliography and in various other tasks conneded with the preparation of this "·ork. Ernst Mayr E. Gorton Linsley Robert L. Usinger

vi

I

)

New York, N. Y. Berkeley, Calif. .January, 1953

VH

CONTENTS

Preface PART 1. TAXONOMIC CATEGORIES AND CONCEPTS 1. Taxonomy, its History and Functions

2. The Species and the Infraspecific Categories 3. Classification and the Higher Categories

3 23 40

PART 2. TAXONOMIC PROCEDURE

4. Collecting and Collections 5. Identification and Taxonomic Discrimination 6. Taxonomic Characters 7. Quantitative Methods of Analysis 8. Presentation of Findings 9. Preparation of Taxonomic Papers

63 72 105 125 155 178

PART 3. ZOOLOGICAL NOMENCLATURE

l

I, I

10. Historical and Philosophical Basis of Nomenclature 11. The Principle of Priority 12. The Type Method and Its Signifieance 13. Specific and Infraspecific Kam es 11. Generic Names L5. Family Nam es 16. Names of Orders, Classes, and Phyla 17. Ethics in Taxonomy

201 212 236 246 261 271 276 279

I

Bibliography Glossary Index

285 301 317

'"

PART l TAXONOMIC CATEGORIES AND CONCEPTS

l

I, I

CHAPTER 1

TAXONOMY, ITS HISTORY AND FUNCTIONS

l 1.

Taxonomy, or systematics, is the science of classification of organisms. The term taxonomy is derived from the Greek ra~is, arrangement, and vbµo~, law, and was proposed by de Candolle (1813) for the theory of plant classification. Systematics stems from the Latinized Greek word 8 ystema, as applied to the systems of classification developed by the early naturalists, notably Linnaeus (Systema naturae, 1735). In modern usage both terms are used interchangeably in the fields of plant and animal classification. 1 Taxonomy is built upon the basic fields of morphology, physiology, ecology, and genetics. Like other scientific disciplines it is a synthesis of many kinds of knowledge, theory, and method, applied in this case to the particular field of classification. Its potentialities and its limitations are largely those of the basic fields whose raw materials it utilizes. The first step in the resolution of any kind of biological knowledge is the classification of phenomena in an orderly system. This means ultimately the naming, description, and classification of all plants and animals. Something of the diversity of organic nature and the magnitude of this task may be indicated by the following figures. There are now known more than one-third of a million species of plants, sixty times as many as at the time of Linnaeus (Merrill, 1943). Every year about 4,750 new species of plants are described. Including synonyms and subspecies, more than 1 million names were proposed for phanerogams and cryptogams between 1753 and 1942. The number of known species of animals is much greater than that of plants and has been estimated at about 1 million (Table 1). Including subspecies, there are probably more than 2 million named forms of animals, and new ones are being described at the rate of about 10,000 per year. For the insects alone, Metcalf (1940) calculates that l}~ million names are already applied. Accepting an estimate of 3 million probable insect species (Silvestri, 1929), and assuming that each species has on the average five distinct developmental or morphological phases, 15 million descriptions will eventually be required to characterize the stages of all insect species! When we superimpose the necessity for arranging 3 million species in a framework of higher categories expre8s1

For different. usage see '.\iason (1950).

3

TAXONOMIC CATEGORIES AND CONCEPTS

TAXONOMY, l1'S lllSTORY AND FUNCTIONS

ing inferred natural relationships, and analyzing the population structure of the species concerned, something of the magnitude of the task facing just one group of taxonomists may be seen. The objectives of taxonomy can only be achieved by sustained cooperative effort. Furthermore, the ability of the individual taxonomist to contribute to this effort depends on the breadth of his training as •vell as on his native talent. The complexities of modern systematics, its dependence on related fields, the refinement of modern techniques, and the magnitude of the literature have made it inevitable that the days of

Greater specialization has been the inevitable consequence of the tremendous growth of our knowledge of living things.

TABLE 1. E8TIMATED 1\UMBER OF KNOWN SPECIES OF RECENT ANIMAJ,S (Mayr) Protozoa. . . . . . . . . . . . . . 30 , 000 Linguatula . . . . . . . . 70 .Mesozoa. . . ........... .. 50 Chelicerata. . . . 35, 000 Porifera. . . . . . 4, 500 Crustacea. . . . . . . . 25, 000 Coelenterata. . . . . \), 000 Other arthropods 13,000 Ctcnophora . . . . 90 (excl. insects) ..... . 850,000 Platyhelminthe~. 6,000 Insecta ... . 80,000 Acanthocephala . 300 Mollusca .................. . 1 Rotifera. . . . . . . . . . . 1 , 500 Pogonophora . ............. . 3,300 Gastrotricha. . . . . . . . . . . . . . . . 175 Bryozoa .................. . 250 Kinorhyneha. . . . . . . . . . . . . . . . 100 Brachiopoda .............. . 4,000 Nematomorpha.............. 100 Echinodermata ............ . 4 Nematoda ..................... 10,000 Phoronidca ................ . 30 Priapulida. . . . . . . . . . . . . . . . . . . ,5 Chaetognatha ............. . 80 N emertina. . . . . . . . . . . 750 Hemichordata ............. . 1,600 Entoprocta. . . . 60 Tunicata. . . . . . . . . . ...... . 20,000 Annelida . . . . . . . . . . . . . 7 , 000 Fishes ............. . 6,000 Echiuroida. . . . . . . . . . . . . . . . . . 60 Reptiles and amphibians .... . 8,590 Sipunculoidea......... . . . . . . . . . 250 Birds......... . ......... . 3,200 Tardigrada. . . . . . . . . . . . . . . . . . . 180 Mammals ................. . 1,120,310 Onychophora. . . . 65 Total ......... . One of the objects of this tahulation is to indicate the relative size of the various groups of animals. Even the smallest phyla have therefore been included, because they are quite important from the points of view of phylogeny and comparative anatomy. The number of species of birds is based on an accurate count. All other figures are estimates, subject to two sources of error. Only 60, 50, or 40 per cent (or even less) of the existing species have as yet been described in many animal groups. On the other hand, in the less-known groups of animals, many populations have been described as full species which appear to be merely subspecies of widespread polytypic species. The two sources of inaccuracy thus cancel each other to some extent.

the untrained taxonomist are limited. The amateur will always play a most important role in assembling much of the raw material with which the taxonomist works, but he needs a broad background and special training if he is to make direct taxonomic contributions of the quality which will be required in the future. Even the trained taxonomist can no longer cover the entire field in any major group of plants or animals.

5

HISTORY OF SYSTEMATIC ZOOLOGY

The history of taxonomy may be divided into a number of periods. These in turn correspond loosely to the various levels of taxonomy (alpha taxonomy, beta taxonomy, gamma taxonomy, see below). Definitions of these periods facilitate the understanding of the progress that has been made in the field. The complexity of taxonomy must be kept in mind when studying its history. Progress in the taxonomy of various animal groups (and in the study of animals from different regions) has been very uneven. Taxonomy is most advanced in the most popular groups (birds, butterflies, mammals, some genera of beetles), while in others it may still be on an elementary level. It is most advanced in the North Temperate Zone and lagging behind in the tropics and other distant places. Consequently the three historical periods here outlined are not strictly consecutive but largely overlapping. First Period-the Study of Local Faunas. The history of taxonomy is almost as old as man himself. Natives of even the most primitive tribes may be excellent naturalists, with specific names for local trees, flowers, mammals, birds, fishes, and the more conspicuous (or most edible) invertebrates. A tribe of Papuans in the mountains of New Guinea was found to have 137 specific names for 138 species of birds. Only one ;.;pecies was confused with another. Often the nomenclature of such tribes is clearly binominal, with a generic and a specific name (Bartlett, 1940).

Several early Greek scholars, notably Hippocrates (460--377 B.c.) and Democritus (465-370 B.c.) included animals in their studies. However, only fragments of the works of these earlier authors are in existence. Apparently it was Aristotle (384-322 B.c.) who brought together the knowledge of his time and formulated it into the beginnings of a science. A.ristotle did not propose a formal classification of animals, but he provided the basis for such a classification in his statement that "animals ma~ be characterized according to their way of living, their actions, their habits, and their bodily parts." He referred to such major groups of animals as birds, fishes, whales, and insects, distinguishing among the last both mandibulate and haustellate types and winged and wingless conditions, and utilizing certain terms for lesser groups, such as Coleoptera a~1d Diptera, which persist today. Aside from these larger groupings, his categories, according to Nordenskiold (1928) were but two in number, the genos and the eidos, "the latter corresponding to the individual animal form-horse, dog, lion-the former to all combinations of a higher degree." The Aristotelian philosophy-it can scarcely be called a sys-

6

TAXONOMIC CATEGORIE S AND CONCEPTS TAXONOMY , ITS HISTORY AND PUNGl'IONS

tern-suffice d for the students of animals for nearly two thousand years. It is only in the works of the immediate predecess?rs o~ Linnaeus that we find more than probing attempts at ammal classificat10n. . The botanists were far ahead of the zoologists during this period, srnce they were the first to break away from the Aristotelian tradition a~d describe and classify local plants. From Brunfels (1530) and B~uhm (1623) there has been a continuous refinement of conce~t~ and techmq~es (e.g., Tournefort and Plumier). The contempora ry wntrngs of zoologists

I

I

FIG. 1. Carolus Linnaeus (1707-1778) and title page from the foundation work in systematic zoology.

(e.g., Gesner, Aldrovandi, and Belon) were, on the whole, still domi~ated by Aristotelian concepts and showed onl.y ru~iments of a consist~nt nomenclature and of principles of classification. Of all the earlier authors, the one who had the greatest influence on Linnaeus was John Ray (1627-1705), who recognized the difference bet':e~n t.h~ genus a~d the species and who, through evaluation of both si~mlanties ~nd ~is­ similarities in animals, arrived at a more natural higher classificat10n than did those who had gone before him (Raven, 1942). The type of taxonomy that is based on the study of local faunas reached its peak in the great Swedish naturalist Linnaeus (1707-177~), whose contribution s were so influential on subsequent students that, with

7

much justification, he has been called the father of taxonomy. In the tenth edition of his great work Systema naturae (1758) (Fig. 1), the binominal system of nomenclature was for the first time consistently applied to animals, and this work became the foundation of systematic zoology. In addition to his new system of nomenclature, the work of Linnaeus was characterize d by clear-cut species diagnoses and by the adoption of a hierarchy of higher categories: genus, order, class. The methods of Linnaeus were by no means wholly original, but his eminently practical system was quickly adopted, expanded, and elaborated because of his great personal prestige and the influence of his students. It dominated taxonomy for the next century, and most of the essentials of the Linnaean method are still components of modern taxonomy. It is generally assumed that Linnaeus accepted the doctrine of fixity of species, species tot sunt, quot f ormae ab initio creatae sunt. Indeed, despite certain evidence to the contrary (Ramsbotto m, 1938), systematic concepts of the Linnaean period were static concepts. Higher classification was largely mechanical and showed what we now recognize to be natural relationships only in cases where fundamenta l characters happened to be selected. The thinking of this period was characterized by the concepts of classical typological taxonomy. The species was the nondimensional species of the local naturalist. The particular importance of this period for the history of taxonomy is that at that time biology consisted almost entirely of taxonomy, and nearly all the eminent biologists of that day were taxonomists. Linnaeus was not only the classical representati ve of this first period of taxonomy, his work also heralded the coming of the second period. Although Linnaeus in his earlier writings (e.g., Fauna suecica, 1746) exemplified the local naturalist, he became more and more cosmopolitan in his later publications, utilizing the discoveries of naturalists in faraway countries. Still, his philosophy remained that of the student of local faunas, except that the Systema naturae was the product of the joint labors of many local naturalists. Second Period-the Acceptance of Evolution. Evolutionar y thought was already widespread in the eighteenth century (Maupertuis , Buffon, Lamarck, and many others), but it owes its firm foundation to the second period in the history of taxonomy, the period of exploration. This movement started modestly during the previous period and reached a grand climax during the middle of the nineteenth century. It was characterized by an intense interest in the faunas of faraway places, in magnificent world voyages and expeditions, and in the accumulation of vast numbers of specimens from all over the world, which permitted the monographic treatment of genera and families. Charles Darwin (1809-1882) Was the naturalist on one of these expeditions (Voyage of the Beagle)

I'

TAXONOMY, ITS HISTORY AND PUNC1'/0NS

8

TAXONOMIC CA1'EGORIES A.VD CONCEPTS

and worked up some of its results. He became the world's leadit~~ · l' t f the Ci'rripedia (barnacles) and wrote a monograph of this specia is o · f h' · · roup that was authoritative. It was largely on the ba~1s o i~ expen;nces as a traveling naturalist and taxonomist that Darwm conceived the theory of evolution. Combined with the reading of Malthus's Essay on Po ulation it also gave him an answer to the problem of the_ ca~se of ev~lution the theory of natural selection. It is more than a comcidenc)e that another traveling field naturalist, Alfred R. Wallace (1823-1913 ,



FIG. 2. Char1es Rob er t D am m

(1809-1882) and title page from the foundation work '

in evolutionary theory.

\

l

'i

came simultaneously to the same conclusions. T~e v~ews of both me~ were jointly presented in 1858 to the Linnaean Society m ~n~ of the most dramatic episodes in the history of science. That Darwm1sm was toha large extent based on taxonomic work is perhaps one of the. reasons w y it did not actually alter taxonomic arrangements very basically, as has been pointed out by Dobzhansky (1951). . . The publication of Darwin's On the Orig~n of. Species (1859) (Fig. 2) resulted in a tremendous stimulation of b10logic~l ~bought and work. The decades immediately following 1859 were prmc1pally taken up ~ the question, Is evolution a fact? Or, stated different!!, Are tl~ t~: livin or anisms descendants of common ancestors? The mteres o perio~ w!s preeminently phylogenetic. The chief effect of the acceptance

of the theory of evolution on taxonomy has consequently been a greater preoccupation with phylogeny. Ernst Haeckel (1866), more daring and speculative than Darwin, introduced (Fig. 3) the method of representing phylogeny by means of trees or branching diagrams (see Chap. 8). Although his formalized diagrams resemble but little those that are in use today, the method itself ,ms useful and stimulating, and it provided the taxonomist with a graphic means for expressing supposed relationships. The search for facts to improve the designs of phylogenetic trees dominated biology during the sPcond half of the nineteenth century and led to a boom in the fields of comparative systematics, comparative morphology, and comparative 0mbryology. In taxonomy, in particular, it spurred the search for "missing links" and "primitive ancestors." These efforts were not "asted but led to a far-reaching understanding of the animal forms and to the establishment of a natural system that is still considered essentially ntlid. This was an exciting period in the history of taxonomy. Not only were new species and genera discovered daily, but with reasonable frequency eYen new families or orders. The reward of such exciting discoveries attracted the keenest minds to the field of taxonomy. Alas, the wealth of nature is not inexhaustible, and the period of major new discoveries in the higher animals was over well before the end of the nineteenth century. Those who were anxious to describe new orders, families, and genera had difficulty discovering them. As an alternative choice they resorted to the splitting of the existing categories. Some splitting was justified and led to an elucidation of classification by doing away with heterogeneous, polyphyletic groups. In other cases, however, it led to a disintegration of natural categories. It appears, in retrospect, as the most retrogressive period in the history of taxonomy. Few of the splitters were good biologists, nor did they understand the proper function of the taxonomic categories. Part of the disrepute into which taxonomy fell during the latter part of the nineteenth and early twentieth century was caused by the activities of those who unnecessarily split well-known and wellfounded taxonomic categories, thereby hopelessly concealing natural affinities. Third Period-the Study of Populations. While the preceding period was dominated by the study of evolution of the higher categories, with a great interest in ancestral forms or missing links (such as Amphioxus or Peripatus), the most recent phase in the history of taxonomy is characterized by a study of the evolution within species. The typological concept of the species, which was already shaky in the preceding period, was abandoned and replaced by a dynamic, polytypic concept. Interest reverted to the fauna of local areas and to the study of variation within

JO

p

'.l'AXON OMIC CATEG ORIES AND CONCE PTS

a

TAXON OMY, ITS HISTO RY AND FUNCT IONS

c

11

popula tions and the slight differences betwee n adjace nt popula tions. The taxono mist is no longer satisfie d to possess types and duplica tes; .he collects series and analyze s them quantit atively . This type of study was commenced almost simulta neously by ornitho logists , entomo logists, and malacologists in the second half of the ninetee nth century . The detaile d history of this phase of taxono my has not yet been written , but it would be well worth the attenti on of historia ns of biology. Althou gh the study of popula tions reache d its domina nt positio n in systematic s only within recent genera tions, its roots go back to the preDarwin ian period. In ornitho logy, after the pionee r efforts of Schlegel, the system atic collect ing of series was particu larly in vogue among the Americ an school, following the leaders hip of Baird (1854): As the object of the [Smithsonian] Institution in making its collections is not merely to possess the different species, but also to determine their geograp hical distribution, it becomes important to have as full series as practicable from each locality. . . . The number of specimens to be secured will, of course, depend upon their size, and the variety of form or condition caused by the different features of age, sex, or season. In gathering specimens of any kind, it is important to fix with the utmost precision the localities where found.

m

I, Frl1l: Monoph yletisch er

Moue1·es jStammbaum der Organismen ;tU(O~onu.Dl~

tllfUTJ/feJt

wtt(qezeidut el

vtJlt.

},'mst l!aeckel. Jena, 1866.

s

b

....... t

~~~~~~~~

d

FIG. 3. The phyloge ny ofliving things as conceive d by Haeckel (1866) and expresse d in a formal tree-like diagram .

Among the malaco logists are to be mentio ned particu larly Kobelt (1881), Gulick (1905), the Sarasin s (1899), as well as Cramp ton, whose biometrical studies in the local geogra phical variati on in the genus Partula (1916, 1932) have become classical. The results of this work caused the abando nment of the typolog ical species concep t. Species were no longer conside red as someth ing fixed and uniform , but rather as polytyp ic, consist ing of many subspe cies and local popula tions, each differing from the others and each showin g considerable variabi lity within itself. Two facts, in particu lar, were outstandin g. First, that the differences betwee n subspe cies and species were compo unded of very numero us small variati ons; and second , that much of the local and geogra phical variati on was closely correla ted with the environ ment. The workin g and thinkin g of the leading taxono mists of this period was thorou ghly modern and biologi cally correct , except in one respect. Most of them interpr eted the close correla tion betwee n variatio n and the enviro nment as indicat ing a direct effect of the environ ~ent. They were Lamarc kians. In spite of this error they were essentially much closer to the truth than the early Mende lians. . It was during this period that the Mende lian rules were redisco vered (in 1900), an event which eventu ally led to the spectac ular rise of the field of genetics. . Howev er, the early Mende lians emphas ized the role of large mutati ons (De Vries and Bateso n) and though t that they produc ed new species by a single step. They minimi zed the role of the environ -

12

T11XONOMIC CA1'EGORIES AND CONCEPTS

ment even as a selective agent. This attitude was in part due to the unforlunate choice of some of the most popular genetic material of that period, namely, the aberrant plant Oenothera (by De Vries) and the normally self-fertilizing, nearly homozygous garden plant, the common bean (by Johannsen). The early conclusions of the geneticists, ~hich differed so drastically from those of the students of natural populations, explain the fact that the first two and one-half decades of the twenti~th century were the period of the greatest cleavage between the naturahsttaxonomist and the laboratory biologist. Further discoveries helped to resolve the conflict, and it was eventually realized that both groups were in part right, in part wrong. The , geneticists began to appreciate the extreme importance of small and very small genetic changes, and the concept of mutation was extended to include these. The choice of more suitable material for genetic studies, Drosophila by Morgan and Antirrhinum by Baur, was also helpful. Fisher's (1930) demonstration that even a very small selective advanta~e · of a new gene or gene combination would cause in due time a genetic ; transformation of populations was a tremendously important contribution. Perhaps the most important factor in bringing taxonomists and · geneticists together was the work of three animal geneticists who had been trained as taxonomists and who studied material from natural populations, Goldschmidt, Sumner, and Dobzhansky. They introduced the population concept of the taxonomists into genetics and prepared the. foundation for the establishment of the new prosperous science of population genetics, which is, so to speak, an offspring of the harmonious union of taxonomy and genetics. In turn, the genetic results, together. with their mathematical interpretation by Haldane, Fisher, and Wright, forced taxonomists to give up their Lamarckian thinking and made them· realize that the small variations which they had known so long were actually small mutations.

I

THE NEW SYSTEMATICS

I

The taxonomic work of the twentieth century is characterized by a. continuous refinement of the methods and concepts developed in the nineteenth century. Current taxonomy is customarily referred to as the new systematics (Huxley, 1940), but it must not be forgotten that its roots go back to the first half of the nineteenth century, and that even the concept of geographical speciation was expressed in an almost modern form as early as 1825 by Leopold von Buch (translated, from/ Mayr, 1942): The individuals of a genus spread out over the cpntinents, lll-Ove to far-distan places, form varieties (on account of differences of the lo.calit_ies, o_f the food, and the soil), which owing to their segregation [geographical isolation] cannot

TAXONOMY, ITS HISTORY ,!ND FUNCTIONS

13

interbreed with other varieties and thus be returned to the original main type. Finally these varieties become constant and tum into separate species. Later they may reach again the range of other varieties which have changed in a like manner, and the two will now no longer cross and thus behave as "two very different species."

Huxley (1940) comiiders the new systematics as a synthesis of such modern approaches as the geographic, ecologic, cytologic, and physiologic, and of population genetics. He adds that To hope for the new systematics is to imply no disrespect for the old. . . . Even a quarter of a century ago it was possible to think of systematics as a specialized, rather narrow branch of biology, on the whole empirical and lacking in unifying principles, indispensable as a basis for all biological workers, but without much general interest or application to other branches of their science. Today, on the other hand, systematics has become one of the focal points of biology. Here we can check our theories . . . , find material for innumerable experiments, build up new inductions: the world is our laboratory, evolution itself our guinea-pig.

To bring out more clearly the change of concepts that has occurred in the field of taxonomy, the old and the new systematics may be contrasted as follows: The old systematics is characterized by the central position of a specfes, typologically conceived, morphologically defined, and essentially nondimensional. Very little significance is attached to geographic variation. Many species are known from single, or at best a few, specimens; the individual is therefore the basic taxonomic unit. There is great preoccupation with technical questions of nomenclature and the identification and description of "types." The new systematics may be characterized as follows: The purely morphological species definition has been replaced by a biological definition which takes ecological, geographical, genetic, and other factors into consideration. The population, represented by an adequate sample, the "series" of the museum worker, has become the basic taxonomic unit. Most taxonomic work is done with subdivisions of the species. N omenclatural problems occupy a subordinate position in systematic work. The interests of the taxonomist are those of a biologist. This seems like a far cry from the simple taxonomy of a Linnaeus or Fabricius, and new terms have been suggested for the new science. Although modern taxonomy may be referred to as new systematics or (so far as it applies) as experimental taxonomy, it would be misleading to Use these terms in contradistinction to taxonomy. There has been such a gradual change from classical taxonomy to the new systematics, and the change has been so uneven in the various groups of animals

14

'l'AXONOMIO CATEGORIES AND CONCEPTS

(and in various geographical regions), that it would be ~isleading to refer to parts of it as taxonomy and to other parts under a diff~rent name .. We use the term new systematics descriptively rat~er than m.a formal sense. There are various grades of new systematics, depending upon the degree to which a group is known. Some traces of the new systematics are found in the writings of taxonomists as far back as one hun~red twenty-five years ago. No one can foresee wha~ refinements of techmque· and what changes in point of view may occur m th~ future. What we: consider as new systematics in the year 1950 may, mdeed, be very old• systematics fifty years hence. There is an unmistakable ~rend. among taxonomists to approach their material more and more as b1~lo~1sts and, less and less as museum cataloguers. The modern sy~tei:iat1st is sh~w­ ing an increasing interest in the formulation of generahzat10ns, for whic the naming and describing of species is only the first step. Many groups of animals are still so poorly known that the newe, principles and techniques of taxonomy cannot be ap~lied to any grea tent In such cases it is inevitable that the attent10n of the taxo~o f .b. d . ex mist is still almost entirely taken up with the escn mg o new species, the construction of keys, and similar preliminary tasks. o~ the othe hand the taxonomy of a few groups is far advanced. ~here is no g~ou . in which the new principles are being appli~d as ex~ens1vely as to birds· Less than 200 new species have been descnbed durmg the ~ast twenty five years; the last North American species was.di~covered m 1889, th. 1as t Aus t ra1.ian, I·n 1911 · Thus the bird taxonomist is able to concentrat f b" · his efforts primarily on intraspecific analysis, although studies o ir . . phylogeny lag far behind other groups. Some nontaxonomists have formed the erroneous impress10n that al animal groups are rather well known taxonomically. Actually, the s.t~d. of many groups of animals has hardly begun (Mayr, 1942). A stn~m illustration of this is presented by Remane's (1933) w~rk on the i_mcr · man· ne fauna of the Kieler Bucht, an area prev10usly f r t•cons1dere scop1c to be well known. By thorough search and with the ap~ ica I?n o ne methods, Remane found 300 new species in ten years, mclud1~g repr sentatives of 15 new families. Sabrosky (1950) has re~ently pomted ou. how poorly much of the North American insect fauna is known. Man so-called "common species" actually represent whole complexes of go . species not previously discriminated. He writes, A few examples will suffice for illustration. Ross (19~7) i~ the ~europter t' genus Sialis (alder flies) recognized a number of new species with this commen. "Critical study of the genitalia has revealed no less than ten eastern and . Oman (1933).' who studi · :fu m ata" . · grouped un d er the name in western species the important economic group of agallian leafhoppers, recogn.ized m one part 0 the genus Aceratagallia a total of 26 species that had preVIously been plac

TAXONOMY, ITS HISTORY AND FUNCTIONS

15

under five names, besides considerable confusion among these five in their proper application. Shewell (1939) revised the dipterous genus Camptoprosopella, a genus of small yellowish flies commonly collected on grasses and other low vegetation, and described twelve new species, most of which had been recorded for years under the name C. vulgaris (Fitch), a species which, incidentally, could not even be recognized with certainty because the type is a female. The present 1Vriter (1949) found twelve new species, with characteristic terminalia in both sexes, in what had been widely determined for many years as a common Holarctic species, Leptocera lutosa (Fallen) (Diptera). In the abundant and persistently annoying eye gnats or Hippelates flies, which have been of some interest as vectors of yaws and various eye diseases, the writer has found no less than nine distinct species of three genera standing in collections under the name Hippe/ates pallipes (Loew), in this case all named species whose correct recognition requires painstaking attention to minute details.

Even less known is the taxonomy of tropical animals. Thus the major portion of the work of the taxonomist still remains to be done. CHANGE IN TAXONOMIC CONCEPTS

Another way of bringing out the revolutionary change in the thinking of the taxonomist is to define the two concepts that are most characteristic for early and for recent taxonomy. The Type Concept. Taxonomy in its early history was completely dominated by the type concept. The type concept goes back to Greek philosophy. The "ideas" of Plato are such "types." 1 Applied to taxonomy, the type concept postulates that all members of a taxonomic category conform to a "type." Whether a taxonomist adhered to the type concept consciously or unconsciously, it inevitably affected his methods and results. In particular, the type concept tended to exaggerate the constancy of the categories and the gaps that separate them and to minimize variability. Typologists have often either denied evolution altogether or explained its operation by macromutations. This Philosophical type concept should not be confused with the type method of modern taxonomy, discussed in Chap. 12. The Population Concept. During the past seventy-five years the Population concept has gradually replaced the type concept, but by no means completely. According to this view, species are composed of variable. populations, and even within the higher categories there may be considerable deviation from the "type" of the category. 1

The early nineteenth century was the heyday of the "typologists," the adherents of the type concept. Cuvier was an outstanding representative of this school and ~~Were the Ger~an Naturphi!osophen (Sc~elling, Oken, Carus, etc.) of that p~riod. 0 udents of the higher categories were particularly strong typologists, but this philosPhy also affected taxonomists who worked at the species level.

16

IUE S AN D CO NC EP TS T1t XO NO MI C CA TE GO

me tho ds and of con cep t on the wo rki ng Th e im pac t of thi s cha nge are var iab le ons ati has bee n enormous. Po pul tio n of var iresults of the tax ono mi st lua eva cri pti on, me asu rem ent , and stu den t of and , con seq uen tly , the des the of ns the pri nci pal pre occ upa tio al" specipic ati on has become one of "ty o gis t needed onl y one or tw cat es. " pli lower categories. A typ olo "du as m more, he disposed of the ma ny of h me ns of a species; if he had eac at ies em pts to collect lar ge ser tes lua eva Th e mo der n tax ono mi st att he y ntl que ran ge of a species. Su bse . ics tist sta localities thr ou gh ou t the and is tho ds of pop ula tio n ana lys otax the of thi s ma ter ial wit h the me t par ds has become a sta nd ard Th e use of sta tist ica l me tho re widely use d all gro ups and is bec om ing mo nomic tec hni que in ma ny qua lity of cer tai n : nt is gre atl y im pro vin g the the time. Th is ref ine me kin ds of tax ono mi c work. the new sys tem -· t wit h all the adv anc es of It mu st be recognized tha com par ati ve mo rl forced to dep end up on a.tics, the tax ono mi st is stil can, inf orm ati on · ta and to fit, as well as he phology for his pri ma ry da e which is pri ma rily . int o a classification sch em der ive d from rel ate d fields e of genetics hav e , of evo lut ion and the scienc morphological. Th e the ory ds for app roa chi ng and hav e pro vid ed me tho giv en me ani ng to his work Fin ally , the new wit h nat ura l pop ula tio ns. the problems asr;ociated e of tax ono my and ' rec ogn itio n of the tru e rol syr;tematics has bro ugh t of mo der n biology. pla ced it at the ver y hea rt XONOMIST TH E TASKS OF TH E TA

I

l

e tax ono mi sts , tai nty in the mi nds of som Th ere is conr;iderable un cer , as to wh at the real of ma ny non tax ono mi sts and eve n mo re in tho se me n and ecologist st are. Some lab ora tor y ati tem sys the of ns ctio fun self wit h ide nti fy, ono mi st sho uld con ten t him tax the t tha nk thi to m see uld kee p his col keys. Be yon d tha t he sho g isin dev and ial ter ma ing e eve ry spe cim e cribe new species, and hav des er, ord d goo in ns tio lec is the me re pigeon g to thi s view, sys tem ati cs din cor Ac d. ele lab ly per pro the se par tic ula r tas tax ono mi st will den y tha t No . ens cim spe of ing hol groups ma y no t y worker in the less-known the and , job his of t par are tax ono mi c wo r the cat alo gui ng pha se of ond bey ch mu go to e abl be r, is no t rer;train ter -kn ow n gro ups , howeve bet the of st ati tem sys e Th tha n an aux ilia him, sys tem ati cs is mo re for n; tio ita lim a h suc by bu t also as to th no t onl y int o the "w ha t" e uir inq can He . nce scie "w hy ." ker of a collectio mo re tha n the me re car eta is st mi ono tax n der mo e Th stu die s int o th own ma ter ial , car rie s his his s her gat he es cas st In mo of view of t the tec hni que and po int y reb the ps elo dev and field, tho rou gh tra ini ng i sys tem ati sts hav e had a r nge you st Mo ist. log eco is experience in bo t y, inc lud ing genetics. Th various bra nch es of biolog

TO TA XO NO MY , JTS RIS RY AN D FU NC TIO NS

17

. the well-tr . field and lab ora tor y givesent al st di a.med sys tem at1 st an ex<',ellent bs,ckdam gro und for .tnore fun u es. Th hr of the sys tem ati st are . ks tas pal nci pri ee ~ e .. Id ge) It . h en cat ion (Analytical Sta t . l" . is t e basic tas k of the. sys al d" rs1ty of the . tem ati st to bre ak up mos un im1ted and co nf usm g ive k . d"iv1"dual s in nat ure int o e .1y recogm.zable gro t m as1 ups, o wordiffou t the "t . . cant cha rac ter s of th t t erences ~1gmfi ese . um s ' hand to find .cons an rth Fu s. one r wi th ila its un sim the se between w"llef1m?I~te, e mu st pro vid e .. "sc ien tifi c" nam es which I am I ate the ir sub se que n t recogmt10n by rld. wor k ers thr ou gh ou t the wo . . of the ta Ev en thi s "lo we st " tas kgeol . I x~om1st is of tre me ndo us scientific importance. Th e ent ire ges on the cor rec t k ogrna. c ronology hin identification of the fossil ific ecological sur vey h e[ t~ec1es. N.o scient wit t ou d rie car be uld sho g ide nti fic atio n of all s·ou'fi e mo st pam sta kin I . the species of ecological igm canoe Ev en the exp enm ent a biologist . . th atio n has I ear ned to app rec iat e for sound, solid ide nti fic · of' e neces~1thy two s ber num at gre e thr are ere Th gen era wit :°1ore ver y sim ila r or e, ' f . d'f y oft y in the ir :>pec1es. Su ch r;pecies ver :or 1 ;rl m_ore conspicuousl has fre ire the in n tha its tra l ica physiolog hp o og1cal cha racf ter s. It k ns que ntl y hap pen ed tha t tw o wor ers ave come t o d"f I ere nt con clu sio . t" l · ,, ta· " e, aus (·oncernmg the physiologic a pro per ies of a cer bec s cie m spe . · f t was wo rki n wit h h wit er oth the and ~ iu act , one stu den dsp eci es A fgA . k . st 1 xed b. mi a e h Ev wit or sB an B · ~ ~pecie oc o ry 10 og1st will recall h t ..: n field. ~ sue cases in his ow .. eti c St ) Th t10n and . acc ura te ~ Classification (Sy nthcie ogm rec e fir e is t:!" sys tem ati st. Bu t -~ description of the spe ws Id the bst tas k of . he h re the p "th sto d ~ should he ou sooTn e con fro nte w1 a c aot ic acc u' I .0n of species des pt· cri mu at1 the sys tem ati st mu st a:;~:ent ~t:~vent t~is: he try to find an ord erl y arr mu st cha rac ter ize e specrns' h . . . · and arr ang e hig her cat ego nes , m ot er wo rds ' h ~ mu st d ev1se a classith f fi . k t ond sec g of a cat10n. Th is is the e ~axonom1st. Th e dev isin o ~s t e e som to is fic classification a tas k as the ide nti ati on en ' as pra cti cal x l . i"t t bu' ens cim · t of spe re specul a wn an d theorizing. Th e tax mv o ves mo ' . ·1 t e eth ·d wh ide dec onomist mu st 1. ar for ms sho uId be cons1 ered one s1m wo r "P ec1. es or two. He mu st 1 d t ities of two erm1.ne wh eth er the similar e ver con to due species are ene tic rel ahab1:us or to close phy log o~· ce t~e to ds lea is Th tionship. hig her cat es 1011 o wh eth er or no t the que et' hy1 nop mo ent res gories rep ic groups. Th stio ns th t que the wh o is of e som . ese are con fro nt the sys tem ati st . ild bew the fy y lea d the and s, ~rYlng to classi sm ani mu ft1 tud e ?f org g ;n~ th of dy stu a to ine vit abl y 0n. e ac ors o evolut1 . F Study of S th F of and rk on ati orm 1es pec act ors of Evolution. Wo . ses the th. d t k f e sys tem ati st. It is her e tha t 111 thi s field com pri as o the Ir

:1:0

t

18

T AXUNO MIC CATEGO RIES AND CONCE PTS

he comes into closer contact with the other branche s of biology, with genetics and cytology , with biogeog raphy and ecology, with compar ative anatom y and paleonto logy. All these sciences pursue the study of evolution in their own way, with their own question s, and with their own method s. One of the principa l differences, for example , between the systema tist and the genetici st is that the genetici st can test many of his conclus ions by experim ent, whereas the systema tist can rarely do this and usually has to rely on the implica tions of observe d data. He can therefor e say very little as to either the origin of taxonom ic charact ers or their mode of inherita nce. On the other hand, the genetici st has • difficulty in duplica ting in the laborato ry the conditio ns under which the at only ents experim speciati on proceed s in nature and can conduct level where some interbre eding is possible. Many animals cannot be kept in a laborato ry, and others will not reprodu ce in captivit y. Further , a even of analysis genetic more, the enormo us time which the thoroug h best-. two the in e single species requires (it is still very far from complet studied organism s, the fruit fly, Drosophila melanogaster Meigen, and the corn plant, Zea mays Linnaeu s), makes it impossi ble for the genetici st to study more than a very small proport ion of the known organism s. Up to the present time only about Hoo per cent of the known species of animals have been studied with any degree of thoroug hness by genetici sts. It is therefor e obvious that the systema tist can and will have to fill man very large gaps. But there is a more basic difference between the · approac h of the genetici st and the taxonom ist to the problem s of evo th atoms," ical lution. The genetici st, in his analysis , seeks the "biolog genes, and other basic units. The taxonom ist, on the other hand, work i, with much more compre hensive entities : with the carriers of taxonom course of is, characte rs, with individu als, populat ions, species. There some recent overlap ping of the two fields owing to the develop ment populat ion genetics, but the difference is striking enough to lead to conside rable difference in outlook and somet~mes even in conclusions. The systematist who studies the factors of evolution wants to find out ho species originate, how they are related, and what this relationship means. H studies species not only as they are, but also their origin and changes. He tri to find his answers by observing the variability of natural populations und. different external conditions and he attempts to find out which factors enhan and which retard evolutionary changes. He is helped in this endeavor by knowledge of the habits and the ecology of the studied species (Mayr, 1942). LEVELS OF TAXONOMY

The three tasks of taxonom y are rarely underta ken simulta neously Evoluti onary studies cannot be pursued unless a satisfac tory classific tion is availabl e, and this in turn is based on the prior identific ation an

1'AXON01vJY, l'i'S HIS'l'OR Y AND FUNC'l' IONS

.

.

l9

. f descript ion of species. The tax group, therefor e passes through several stages The ohnomy o a given ' · f se ave sometim es b een m · ormally referred to as a 1pha, beta and gamma t level at which the species are c~xono:iY_· dAlpha taxonom y refers to the named; beta taxonom y to the arrangin g of these spec. . atrac erize and ies m ot a natural syst em of 1esser and higher · categories; and gamma tax onomy o the analysis f . t . d . ic varia t wn an to evolutio nary stu di es (see above) A to m raspecif . . 11 'bl imquite is it y c ua possi e to delimit alpha ' b et a, an d gamma ta. h from one arply s xonomy d . anot h er, since they overla r, the trend is unmista kable It is the enf an ifnthergrade. Howeve ' · d d eavor o t e biologicall · y mm e taxonom ist to. pass f rom the alpha level throu h th level to the gamma level. Still, even in the taxonom ically bg t ke beta groups . th nown es ere is need for ' more re fi ne d work on the al Ph a an d b eta levels. THE RELATION OF TAXONOMY TO

OTHER BRANCH ES OF BIOLOGY Th' . is is an age of speciali zation E very aspect of life is studied by a different branch of biologi'c 1 . · d a science Biochem . t I is ry a~ much of physio ogy deal with life at the molecular. level cytolog y with cells and their compon ents histolog y with th t'ISsues that' are db fo e ' rme y cells, anatom y . (an d parts of physiolo gy) with behavior and the interact ion org:s,.~ociology and psychol ogy with the !ems of growth and develop men:n ~ uals, a~d embryo logy with prob. e next higher levels of integrat ion · . of life are dealt with by th used to be called natural history whicdh e :asxcienc as but is now referred to onomy an ecology · T axonom y deals with . na t ural populat ions subsp · No other science takes care ~f this ~ciesl, spfe?ies, and higherthcategori es. . . eve o mtegrat ion rn t d A world. orgamc e . s u y of systema tics is theref of und backgro the of part rntegral an ore every well-tra ined biologist. . . As long as a science is strictly descri tive . .P . ' it usually has little contact with other sciences. When th d r and the compar ative and fu et' escripti ve stage has passed, howeve reached , c?ntact a.nd ~~rlap with neighboring scien:::~~:~t:~f:~e~r. e The two sciences with t 1 th has y taxonom modern ch ion genetics populat are and ecology. Familia rity with ~he ose: c~ntact sable part indispen an is elds of the training of the taxonom ' t ese wo is . Contrib . . ob. . utions of System atics to Biolo . d gy. Some idea of the scope and .Jectives of systema tics may b tr1butions of systema tics to oth e irme hfrom a conside ration of the coner ranc es of science and to mankin d as a whole. Simpso n (1945) h as commen ted that Taxonom . t th . e same time the most element y is a zooJo ary and most mclusive part of s . gy, most elementarv because . ls a cahn~ot be discussed or treated in a t~lentific way until some taxonomy h~imbeen . · 1usive ac 1eved ' and m Ost me . · 1'ts various Xonom Y m because h guises and b ranc es gathers together , ut'l' 1 1zes, sum-

!

20

TAXONOMIC

CATEGORIJ~S

AND CONCHP'l'S

marize;;, and implements everything that is known about auimah;, morphological, physiological, psychological, or ecological.

Or, according to Pearl (1922), It is the systematist who has furnished the bricks with which the whole str c ture of biological knowledge has been reared. Without his labors the fact organic evolution could scarcely have been perceived and it is he who to really sets the basic problems for the geneticist and the student of experimen, evolution.

Aeeording to Elton (1947), The extent to which progress in ecology depends upon accurate identificati and upon the existence of a sound systematic groundwork for all groups animals, cannot be too much impressed upon the beginner in ecology. This, the essential basis of the whole thing; without it the ecologist is helpless, and whole of his work may be rendered useless.

Systematics has contributed to the applied sciences both directly a, indirectly. This is true of medicine, public health, agriculture, con · vation, management of natural resources, etc. A few examples may mentioned from the field of applied entomology. Here insect identi cation has provided a filing system for economic entomologists, a co venient method for assembling the vast amount of detail accumula over the years. It has also provided a useful tool, the natural classi cation, by means of which generalizations may be made as to the tribution and habits of economic insects. This is especially true of t new pests which appear from time to time and about which little nothing was previously known. Generalizations drawn from near rel tives may provide valuable clues as to probable habits, future importan · and means of control of an insect whose economic importance has ju been recognized. Systematics has proved to be the key to the solution of some of them perplexing problems in economic entomology. For example, malaria distributed unevenly over Europe. The supposed vector, the mala mosquito, Anopheles maculipennis Meigen (1818), was reported throug out the continent, and large amounts of money were spent to control in certain areas with no corresponding decrease in the incidence of t disease. At the same time, there was no malaria in some parts of t range of the malaria mosquito. Careful systematic studies, summariz by Hackett (1937) and Bates (1940) finally provided the key to t situation. The maculipennis complex was found to consist of sever sibling species, distinguishable at first only in the egg stage, each wi its own peculiar breeding habits, each with decided host preferences, an usually only one species actually responsible for the transmission

'1'.1XONOJJ1l', 11'8 HJS'l'OHi' AND FUNCTIONS

21

malaria in a given area. Armed with this information, it was possible to direct control measures to the exact spots where they would be most effective. Applied entomological taxonomy is the basis of quarantine procedure. Here the taxonomist finds himself working under intense pressure with the fate of carload shipments of produce hanging in the balance, dependent upon his identifications. An example (Keifer, 1944) is "the identifit'ation of a single caterpillar from Orange County (California) in October, 19-±2, as the larva of the oriental fruit moth. [Before a second specimen \\'aS discovered) many meetings of fruit growers and entomologists (were held, and] the State Legislature appropriated over $850,000 for various phases of investigation and control." Accurate identification is essential in connection with the biological control of plant and animal pests. As Clausen (1942) has remarked, A mistake in the identity of the host may result in the complete loss of years of work and the useless expenditure of large amounts of money. If, for instance, a pest is of oriental origin but is mistakenly identified as a closely related European species, the search for natural enemies in Europe, and their collection, rearing and colonization for biological control, might well prove utterly futile. Pemberton (1941) cites an outstanding instance of the value of insect collections, assembled for taxonomic study, in the solution of a biological control problem. Some 20 years ago the fern weevil, Syagrius fulvitarsis Pascoe, became very destructive to Sadleria ferns in a forest reserve on the island of Hawaii, and control measures became necessary. Entomological literature failed to reveal its occurrence anywhere outside Hawaii except in greenhouses in Australia and Ireland. These records, of course, gave no clue as to the country of origin. However, while engaged on other problems in Australia in 1921, Pemberton had the opportunity of examining an old private insect collection at Sydney, and among the beetle specimens was a single Syagrius fulvitarsis bearing the date of collection, 1857, and the name of the locality in Australia from which it was obtained. This provided the key to the solution of the problem, for a search of the forest areas indicated on the label revealed a small population of the beetles and, better still, a braconid parasite attacking the larvae. Collections were made immediately for shipment to Hawaii, and the establishment of the parasite was quickly followed by satisfactory control of the pest. The data borne on a label attached to a single insect specimen in 1857, in Australia, thus contributed directly to the successful biological control of the pest in Hawaii 65 years later.

Systematics as a Profession. What opportunities exist for the student who contemplates a career as a professional systematist? As Ferris (1942) has put it, How is he going to get an opportunity to work? Research of the quality that is needed now . . . cannot well be done by some hard-worked doctor, or policeman, or janitor or gardener, or even by every college professor, who in his few

22

TA.XONOMJC C1i'l'EGOJUES A.ND CONCEPTS

spare moments of relaxation after having earned his daily wage whiles away time with his collection . . . . The days when that could be done are passing. Systematic work of the kind that we need now most of all is a full-time job for a trained professional who has the technical facilities that are needed and who, through the help of assistants, is freed from the pressure of routine so that he can concentrate on his essential task.

There are not many positions of this kind in the world. Probably less than 1,000 professional taxonomists are employed in the world today. Many of these serve their governments in a capacity where their first duty is identification for economic purposes. At the state level, Illinois, California, and the Territory of Hawaii employ systematists to identify animals in connection with quarantine work and state surveys. Curators of the zoological collections of the world are full-time workers in the field of systematics, and a few of the larger universities have one or more staff members employed as taxonomists. It should be noted, however, that the above-mentioned positions do not fulfill the need expressed by Ferris. Although they fall within the definition of professional systematics, the salaries are paid for identification, curatorial, or teaching activities. Pure systematic work thus becomes a side issue, accomplished hy zealous workers who are willing to burn the midnight oil. Adding to the hulk of systematic work are a great many amateur taxonomists, a few hundred applied zoologists who maintain an interest in special groups as a hobby or who produce systematic work as a by-product of their applied studies, and several hundred students in each college generation who select and carry out a taxonomic problem in order to fulfill the requirements of a thesis for a higher degree. It has been stated that the applied >1cience of the future will lean even more heavily than now upon the diagnostic work of the taxonomist and the systems he devises. Taxonomy, of itself, is not spectacular and has little appeal in legislative halls or budgetary hearings. Applied science has a real and serious responsibility from the standpoint of its own selfish needs in guaranteeing that its reciprocal partner, taxonomy, has the opportunity and funds to function as it should. Officials in responsible Federal and state administrative positions must insist that funds and personnel for more comprehensive taxonomic studies be secured.

CHAPTER 2

THE SPECIES AND THE INFRASPECIFIC CATEGORIES The following sections will be devoted to a discussion of taxonomic categories. A taxonomist cannot work with these categories unless he understands their meaning and how each category differs from the others. Some of the purely practical difficulties of assigning individual specimens to the right category will be treated in the section on taxonomic diserimination (Chap. 5). KINDS OF TAXONOMIC CATEGORIES

The 20 or more categories that are used by the taxonomist in his elassification (Chap. 3) are of unequal value and of different significance. Essentially there are three kinds of categories: 1. 2. e.g., 3.

The species Groups of populations within species ( = infraspecific categories, subspecies) Groupings of species (collective categories = higher categories)

From evidence to be presented later it will appear that the species occupies a unique position in the taxonomic hierarchy. The discussion of taxonomic categories will begin, therefore, with the species. As stated above, the alpha level of taxonomy is an essentially analytical stage. It consists of the distinguishing, identifying, describing, and naming of the species. At this stage no building up of a classification is undertaken, only fabrication and accumulation of the bricks of which the natural system is composed. A study of species and other lower categories is therefore essentially an analytical study. Classification, on the other hand, is synthetic, and the principles of classification will be discussed together with the higher categories in Chap. 3. THE SPECIES

The species is the most important taxonomic category, not only for the taxonomist but also for the general biologist. An understanding of the nature of species is indispensable for taxonomic work and has to precede any attempts at giving a formal definition of the term. The word species, meaning originally kind, is older than the current biological eoncept. Even today the term is sometimes used for inanimate objects, for instance, species of minerals. The Greeks, in particular Plato and 23

24

I

L

TAXONOMIC CATEGORIES AND CONCEPTS

'l'HE SPb'CIES AND 'l'HB INFllASPECIFIC CATBGOIUES

his followers, used the word eidos in much the same sense. The species concept of the biologist goes back to J. Ray, who in his H istoria plantarum (168G) used the term species much as it was used later by Linnaeus and the nineteenth-century taxonomists. The modification of the static and morphological species concept of Ray and Linnaeus during the twentieth century will be discussed presently. The Species in Nature. A species definition is merely the verbalization of a species concept. Species concepts are derived from a study of species in nature. A student of any local fauna finds that it is composed of well-defined "kinds" of animals and plants. Around New York City, for instance, there are about 125 "kinds" of breeding birds. These are the species. The individuals within a local population of such a species are freely interbreeding but are separated by a distinct gap from individuals of all other species. In eastern North America there are five species of thrushes of the genus Hylocichla: vvood thrush (Hylocichla mustelina), hermit thrush (H. guttata), olive-backed thrush (H. ustulata), gray-cheeked thrush (H. minima), and veery (H. fuscescens). These five kinds of thrush are quite similar to one another, and some can be told apart with certainty only by the more experienced bird student. In spite of the morphological similarity of some of these five species, each one is separated from every other one by a definite gap. No intermediate or hybrid has ever been found. They do not interbreed; they are reproductively isolated. Taking any pair of species of a local fauna, no matter how closely related, they will be found to be separated from each other by a definite gap. It is this discontinuity between natural populations that impressed the early naturalists from Ray and Linnaeus on, and which remains the cornerstone of the species concept of the modern systematist. Species Definitions. Taxonomists have always been faced with the problem of making their working concept of species conform to the species in nature. The pioneers were impressed by the fact that the species in nature usually differ by clear-cut characters, the "species characters." They concluded, therefore, that species should be defined on the basis of the degree of morphological difference. Morphological species definitions, however, sooner or later ran into difficulties. First of all, many kinds of individuals were found that were clearly conspecific, in spite of striking differences in structure owing to sexual dimorphism, age differences, polymorphism, and other forms of individual variation. Such forms were often described as species, but as soon as they were found to be members of a single interbreeding population, they were deprived of their specie::> status, regardless of the degree of morphological difference. At the other extreme, sympatric (that is, occurring in the same area) natural populations have been found which are almost indistin-

g;uishable on the basis of structure, but which fail to interbreed (sibling species, see below). These are recognized as distinct species in spite of the lack of morphological differences. It is for these reasons that a species definition based solely on morphological criteria has proved to be unsatisfactory. Taxonomists have therefore been obliged to seek another philosophical basis for their definition of species. In spite of practical difficulties in its application, reproductive isolation has proved to be the soundest theoretical criterion. Species therefore may be defined as follows: Species are groups of actually (or potentially) interbreeding natural populations which are reproductively isolated from other such groups. Such a definition is called a biological species definition, because it is based on the biological criterion of reproductive isolation (Mayr, 1942). How this species definition cau be applied to the specimens of the taxonomist will be discussed in Chap. 5. Sibling Species. Among the species is one kind which deserves to be singled out for purely practical reasons, the sibling species (Mayr, 1942). This name is applied to pairs or groups of very similar and closely related species. It has been found that such groups occur commonly from protozoa to mammals. For a more detailed discussion see Chap. 5. Sibling species are not a separate taxonomic category. They do not differ from other species in any respect except for the minuteness of their structural differences. The Subjective Element in Classification. Simpson (1943, 1945), in particular, has called attention to the subjective element in much of classification, particularly in paleontology. The student in many cases does not classify species but samples from natural populations:

25

From a series of concrete specimens in hand an inference is made as to the nature of a morphological group from which the sample came, and an endeavor is made to frame the morphological concept in such a way that the inferred morphological group will approximate a genetic group. The thing that is actually classified is an inference, a purely subjective concept, which approximates a real, hnt unobservable, morphological unit, which in turn approximates an equally real hut e\·en less observable genetic unit. It may seem that Simp::>on'::> views on the observability of genetic species are unduly pessimistic when applied to easily observable species like the monarch butterfly or the wood thrush. In these cases, and they are actually very common, it appears that the classifier has much more information available than a mere inference. In rare and localized species, like the whooping crane (Grus americana Linnaeus) all the living representatives of the species may be observable simultaneously. The philosophical basis of Simpson's argument is, however, correct.

l

26

THE SPECIES AND THE INFRASPECJFIC CATEGORIES

TAXONOMIC CATEGORIES AND CONCEPTS

The Species without Dimension. The original concept of a species, the species of the local naturali:sts Ray and Linnaeus, was a species without the dimensions of space and time. Such a species is always separated by a complete gap from other sympatric species. In its purest form it is clear-cut and has objective criteria, because it is defined by the gap that separates it from other sympatric species. This local species is the yardstick by which all other situations are measured. Lacking the dimensions of space and time, such a species is not evolving, it is static. It is for this reason that the nondimensional species has a great deal of objectivity and can be defined unequivocally (Mayr, 1949). The modem concept of a species is multidimensional and often lacks the preciseness of the "ideal" nondimensional species. Polytypic Species. In the period of exploration which followed Linnaeus and which continues into our own day, it was found that some species are wideflpread and consist of many local populations. If these local populations are sufficiently distinct from each other, they are called silbspecics (see below for details). Species which have no subspecies or which, to be more precise, consist of only a single subspecies are called rnonotypic species. Species that consist of two or more subspecies are called polytypic species. Recognition of the significance of polytypic species was one of the most important developments of the new systematics. It is therefore appropriate to discuss the polytypic species in more detail. Populations that are mutually exclusive geographically are called allopatric. About one hundred years after Linnaeus it was found that certain local species that had been described from various parts of the world could be combined into groups of allopatric "species" that were obviously more closely related to one another than to any other species. Finally, when the gaps between the ranges of such species were explored, it was often found that they were occupied by intermediate populations. In other words, it was found that these allopatric species intergraded with one another. Whenever this was the case, these allopatric "species" were united into a single polytypic species (Fig. 4). The reclassification of all related forms originally described as monotypic species into polytypic species has led to a tremendous clarification of the system. This reorganization of species classification is virtually completed in birds. It is in full swing for mammals and under way for butterflies, beetles, and land mollusks but has hardly begun in most groups of animals. Only after it has been completed will it be possible to say how many species of animals exist. In 1870 about 11,000 and in 1910 about 19,000 species of birds were listed, but in spite of numerous subsequent discoveries, only 8,600 species of birds are now recognized (Mayr, 1946). A similar reduction of numbers is to be expected in many

27

Frn. 4. The distribution of 35 subspecies of the kangaroo rat, Dipodomys ordii Woodhouse, as an example of a range map of a polytypic species (Setzer, 1949). 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12.

D. D. D. D. D. D. D. D. D. D. D. D.

ordii richardsoni ordii oklahomae ordii compactus ordii sennetti ordii evexus ordii medius ordii obscurus ordii terrosus ordii panguitchensis ordii uintensis ordii sanrafaeli ordii fremonti

13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24.

ordii monoensis ordii ordii ordii luteolus ordii extractus ordii chapmani ordii montanus ordii cinderensis ordii fetosus ordii utahensis D. ordii columbianu.«r

D. D. D. D. D. D. D. D. D.

D. ordii idoneus D. ordii priscus

25. 26. 27. 28. 29. 30.

D. D. D. D. D. D. D. D. D. D.

ordii celeripes ordii cineraceus ordii marshalli ordii inaquosus ordii attenuatus ordii fuscus ordii longipes ordii pallidus ordii nexilis ordii cupidineus

31. 32. 33. 34. 35. D. ordii pal meri

29

'l'AXONOMIC CA'l'EGUIUES AND CONCEPTS

THE SPECIES AND 'l'Hb' lNFllASPEClFJC CA'l'JSGOJUES

groups of animals, especially terrestrial mollusks and fresh-water fish, as soon as a biologically defin.ed polytypic species concept is applied. The consistent application of the polytypic species concept to all groups of animals is one of the chief tasks of the taxonomy of the future. Dualistic Taxonomic Terminologies. The deficiencies of the simple Linnaean terminology of species and variety became more and more apparent as the knowledge of species increased during the latter half of the nineteenth century and during the first two or three decades of the present one. The polytypic species of the new systematics differs from the Linnaean species particularly by being a collective category. It is in many instances a compound of several "species" originally proposed as monotypic. What scientific name should one give to this new collective category, and who should be the author? Two solutions have been proposed. Although trinominals had been used since 1844, Kleinschmidt (1900) was the first zoologist to recognize this problem clearly (Mayr, 1942). He proposed that a new category be established between the (Linnaean) species and the genus. He gave to it the term Formenkreis and proposed that the first taxonomist who gathered the various allopatric Linnaean species into this new collective category ( = polytypic species of modern authors) should provide a new name for it of which he would be the responsible author (Stresemann, 1936). Although we consider this proposal as inconvenient (and contrary to the International Rules), there is nothing wrong with Kleinschmidt's logic. When Linnaeus named the white wagtail M otacilla alba, to cite one example, he had in mind the European population with the specific characters described by him. The M. alba of Linnaeus is what is now referred to as the nominate subspecies M. alba alba Linnaeus. The collective category formed recently by uniting M. alba of Linnaeus with M. lugubris Temminck, M. dukhunensis Sykes, M. baicalensis Swinhoe, M. leucopsis Gould, M. personata Gould, M. hodgsoni Blyth, M. ocularis Swinhoe, M. lugens Kittlitz, and other species is a far cry from the M. alba of Linnaeus. 1 Rensch (1929), dissatisfied with the term Formenkreis, with Kleinschmidt's evolutionary philosophy, and with his failure to distinguish between superspecies and polytypic species, coined the term Rassenkreis for the latter, distinguishing it sharply from the ordinary monotypic species, to which alone he restricted the term species. These were the foremost attempts in zoology to solve the change in

the species concept by introducing a dual system of terminology, for the Linnaean species on the one hand and the modern, polytypic species on the other. These attempts were not successful, because the vast majority of zoologists preferred a different procedure (A.O.U. Code, 1885). Instead of using two terms, one for the Linnaean species and the other for the species of the new systematics, they broadened the species concept so as to include both kinds. They no longer restricted the term species to the nondimensional species of the old naturalist but included also species which are variable in space and time. The qualifying adjectives, monotypic and polytypic, were proposed by Huxley to distinguish between the two kinds of species in conformance with the use of these terms for higher categories (monotypic and polytypic genera, etc.). Kleinschmidt had no followers, while Coues, Allen, Hartert, K. Jordan, and others were so energetic in the consistent application of the polytypic species concept that no lasting terminological dualism developed in zoology. The Superspecies. Closely related allopatric forms are usually subspecies of a polytypic species. Occasionally, however, the evidence indieates that these allopatric forms have attained species rank (particularly if effectively isolated for a long time). It is frequently important in evolutionary and zoogeographical studies to single out such groups of entirely or largely allopatric species and apply to them a unit term. The term superspecies was proposed for these (Mayr, 1931) as a substitute for the earlier term Artenkreis proposed by Rensch (1929). .A superspecies is a monophyletic group of very closely related and largely or entirely allopatric species. When the ranges of its component species are plotted on a map, the superspecies usually presents the picture of a polytypic species. However, there is evidence that the component species have attained reproductive isolation. This evidence is threefold. Either the species, although completely isolated from each other, are morphologically as different as are normally sympatric species, or they are in geographical contact without interbreeding, or there is actually a slight distributional overlap. Superspecies are not distinguished by a special nomenclature. They are, however, listed as such in monographs and catalogues. They are chiefly important in zoogeographical and speciation studies. The unique position of the species has been pointed out by many recent authors (Dobzhansky, 1951; Mayr, 1942; Huxley, 1942; Simpson, ~945; and others). It is the only taxonomic category which, at least in its nondimensional expression, can be objectively defined and delimited. Occupying a definite ecological niche at a given locality, it has a precise ecological meaning. The infraspecific categories are groupings of populations within species. The supraspecific categories are groupings of

28

i Several recent authors refrain from associating an author's name with the binominal of polytypic species, thereby meaning to indicate that the author of the name of the nominate subspecies i8 not the responsible author of the collective species to which the nominate subspecies belongs. This practice is not sanctioned by th<' International Rules.

L

30

31

TAXONOMIC CATEGORIES AND CONCEPTS

THE SPECIES AND TH1!l INFRASPECIFIG G11TEGORIES

species into "higher" categories. 'fhe species is the "base line" in either case. Practical difficulties in delimiting species may arise from two causes. Either they are due to insufficient information (such difficulties are discussed in Chap. 5), or they are due to the multidimensional character of species in nature. Species are evolving systems, and the paleontologist should in theory be entirely unable to delimit species vertically. This has been pointed out by several recent paleontologists (Arkell and MoyThomas, 1940; Burma, 1949; Dunbar, 1950; etc.). Actually, the fossil record is in most instances sufficiently incomplete to provide artificial breaks in the sequences that can be utilized by the taxonomist as species borders. Some of the unbroken vertical sequences of species recorded by paleontologists show such slight degrees of difference that they could equally well be considered as subspecies and be combined into vertical polytypic species. However, this still leaves some instances in which the paleontologist will have to break continuous sequences arbitrarily into separate species.

This is important, because the work of the population geneticist has proved that in sexually reproducing species no two natural populations are genetically identical. If large enough samples are available and a sufficiently accurate analysis is made, it can usually be proved that a statistically significant difference exists not only genetically but also in morphological characters. Many species will have several hundred, and some widespread species many thousand, populations that differ from each other significantly (in the statistical sense). The naming of all these slightly different populations was proposed by some authors in order to establish uniform categories that conform to the type concept as we have described it earlier. However, this endeavor was doomed to failure from the beginning, because completely uniform population groups do not exist in sexually reproducing species. The typological approach has led the splitter astray. It is now realized that all taxonomic categories are somewhat heterogeneous. Not only the species but the subspecies also is an assemblage of populations, except in the rare cases of exceedingly localized relict forms or insular populations. To qualify as a subspecies, such an assemblage of populations must be taxonomically different from other subspecies. What is taxonomically different can be determined only by agreement among taxonomists. The difference must be sufficiently great so that it is possible to identify the great majority of specimens without knowledge of their provenience. For that purpose, many taxonomists adhere to the 75 per cent rule (see Chaps. 5 and 7). Subspecies as Geographical and Ecological Races. The word race is not used consistently by taxonomists. The majority usage is that which is current among the taxonomists of mammals, birds, and insects, namely, to use the terms subspecies and geographical race synonymously and interchangeably. Ichthyologists and anthropologists sometimes apply the word race to local populations within subspecies. Other taxonomists refer to them simply as local populations. A subspecies is geographically localized and consequently a geographical race. However, since no two localities are exactly identical with respect to their environment, every subspecies is, at least theoretically, also an ecological race. Warm-blooded and highly mobile animals, such as birds, are rather independent of local environmental factors, and subspecies in such groups are primarily geographical races. Plants and many sedentary cold-blooded animals are broadly exposed to the effects of local environmental conditions, and subspecies in such groups impress the observer as ecological races. The case of host races of parasites on plants and animals is particularly suitable to demonstrate the dual aspect of races. Host races are

THE SUBSPECIES

The subspecies is the only infraspecific taxonomic category. The status of other infraspecific forms is discussed in Chap. 5, their nomenclature in Chap. 13. The subspecies may be defined as follows: Subspecies are geographically defined aggregates of local populations which differ taxonomically rrom other such subdivisions of a species. Not more than one subspecies of any one polytypic species can exist in breeding condition in any one area. Adjacent subspecies interbreed or are potentially capable of doing so if separated by extrinsic barriers. It may be helpful to make the following comments on the above-stated specifications of subspecies: "Differ taxonomically": The subspecies concept has an old philosophical tradition, rooted in typological philosophy. When the species concept was developed by Ray and Linnaeus, the species was first thought to be something stable and uniform, composed of individuals that conform to the type. Individuals that did not agree with the type were segregated as "varieties." Subsequently it was found that the "variety" (see below) was a composite concept, including both variant individuals and variant populations. The name variety for the latter category was eventually replaced by the term subspecies. At first only the most distinct subspecies were described, but after taxonomists in certain groups, particularly birds, had nearly completed the describing of conspicuously different species and subspecies, some authors began to name as subspecies every population they could prove distinguishable. Although such populations may be statistically different, they are not necessarily taxonomically different, for the two terms do not coincide.

32

TAXONOMIC CATEGORIES AND CONCEPTS

ecological races because they occur in different niches. However, they may also be regarded as geographical" races, because they are spatially separated, with gene flow severely inhibited. When parasites have the ability to get from host to host, they generally fail to produce host races. The spatial isolation of races of parasites is thus often one between hosts rather than a geographical one in the strict sense of the word. It must be emphasized that there is no geographical race that is not also an ecological race, nor an ecological race that is not also a geographical, or at least a microgeographical, race. The geographical and the ecological aspects are two facets of the same phenomenon, the subspecies. Polytopic Subspecies. If subspecies of a species differ in only a single diagnostic character, such as color, size, or growth form, it may happen that several unrelated and widely separated populations acquire independently the same character. Dice (1941) has described this for Peromyscus populations in the mountains of Arizona and New Mexico; Cazier (ms.) for tiger beetles (Cicindela) on alkali flats; Vaurie (1949) for East Indian drongos; and so forth. Although such populations may be morphologically indistinguishable, they are not necessarily more closely related to each other than to other populations, and they are unquestionably genetically different in various cryptic characters. However, if such populations do not differ from each other in any taxonomic character, they must be united under a single subspecific name (Mayr). A subspecies is a composite, heterogeneous . category, even where it consists of contiguous populations. The criteria to be utilized in considering whether a given isolated popu- .· lation should be classified as a species or a subspecies will be discussed in .· Chap. 5. Intermediate Populations. Intermediate populations are usually found in the area of contact of two well-defined subspecies a and b. Such intermediate populations may have an extensive range, or they · may be restricted to a narrow belt. Individuals of such intermediate populations may either be more or less uniform in character, namely, · intermediate between the topotypical populations of the two adjacent·. subspecies a and b; or else this intermediate population may be composed : of a mixture of individuals, some of which resemble a, others b, while still . others are intermediate. What should be the taxonomic treatment of the individuals of the · intermediate populations? First of all, let us state what should not be done: They should not be described as a separate subspecies if they do not satisfy the requirements of the 75 per cent rule (see Chap. 7). The best solution is to find the halfway point between the "most typical" population of subspecies a and that of subspecies b, and nse {I

THE SPECIES AND THE INFRASPECIFIC CATEGORIES

33

this halfway point (in phenotype, not in distance) as the dividing line between subspecies a and b; or else specimens of the intermediate populations are labeled as X-us albus subsp. or X-us albus albus ~ secundus. If there is a definite "step" in the character change between subspecies a and b, that step should be chosen as the subspecies border, even if it is not exactly at the halfway point. If, on the other hand, there is a perfect gradation between two extremes, then the term cline (Huxley) may be applied to this series of populations. Cline. Huxley (1939, 1940) has proposed the useful term cline for a <~haracter gradient. A series of adjacent populations in which the gradual change of a character occurs forms a cline. At right angles to the cline are the points of equal expression of the character (of equal phenotype), ~md therefore these lines are referred to as isophenes. For instance, if in the range of a species of butterfly, the percentage of white specimens varies from north to south, the corresponding isophenes may be indicated on a map (Fig. 5). There are clines of morphological, physiological, ecological, and other dmracters and also of the percentage frequencies of polymorphic characters. Clines may be smooth, or they may be stepped clines with rather ~udden changes of values (Huxley, 1939). Clines do not receive nomenclatural recognition. In fact, it is advisable not to obscure the presence of clines by the recognition of too many only slightly differentiated subspecies on a single cline. Clines are usually produced by selection and are therefore usually parallel to the environmental factors that are reponsible for them. Close analysis may reveal several clines within a group of populations. These may be quite independent and may run parallel to different environmental factors. The Local Population. The subspecies is the lowest taxonomic category which it is advisable to distinguish nomenclaturally. However, the 1mbspecies is by no means the lowest subdivision of the species. Subspecies are not homogeneous but are composed of numerous local populations, all of them differing slightly in gene frequencies and the mean values of various quantitative characters. Repeated attempts have been made during the past fifty years to give uomenclatural recognition to such slight populations. Semenov-TianShansky (1910) proposed the term natio for them. Fish taxonomists re('ognize races within subspecies, generally without naming them. This usage is not recommended as a formal terminology, since most animal taxonomists consider the terms subspecies and (geographical) race as synonymous. Now that it is being realized that every local population is different from every other one, even if they live only a few miles apart or less, and that these populations are ~ot sharply separated from one another

35

TAXONOMIC CATEGORIES AND CONCEPTS

THE SPECIES AND 'l'HE INFRASPECIFIC CATEGORIES

(except where separated by unsuitable areas), there is no longer a valid excuse for the formal nomenclatural recognition of innumerable local subdivisions of subspecies. It is legitimate and even desirable to describe the trends of variation within a subspecies-particularly if material from numerous localities is available-but it serves no useful purpose and

the females of that species but are not a different taxonomic category, nor are the immature forms, nor individual variants such as albinos, redheads, or blue-eyed individuals. This was not always so clearly recognized as it is now. In the beginnings of taxonomy no distinction was made between infraspecific variants based on populations and such variants based on individuals. Either kind was recorded under a single heading, namely, as variety. The variety is one of the oldest taxonomic categories. It was used by preLinnaean authors and also by Linnaeus. The history of the term variety is closely correlated with that of the type concept and of the morphological species definition. It was originally defined as an individual which somewhat differs from the type of the species but not sufficiently to require recognition as a separate Hpecies. Later analysis showed that many different phenomena were included under the heading variety. In addition to similar species, it included principally subspecies and individual variants of various sorts. Subspecies are a taxonomic category; individual variants are not. When the subspecies received formal recognition in the Rules as a taxonomic category, it was substituted for the "variety" of the earlier version of the Rules. Although not taxonomic categories, many types of individual variants ( = intrapopulation variants) have been named, especially by amateur insect collectors. No reference was made .to the status of such names in the original Rules of Zoological Nomenclature. However, in order to avoid confusion, the International Commission at the Thirteenth International Zoological Congress in Paris (1948) ruled on how to deal with :mch names. This will be discussed in Chap. 13. Intrapopulation variants are more important from the taxonomic point of view than from the nomenclatural one. Many of these variants are sufficiently different from each other to simulate different species. A thorough understanding of the various kinds of individual variation and of the different types that can occur within a single population is of the greatest importance in taxonomic discrimination. It will therefore be discussed in detail in Chap. 5.

34

N .l-74

1'-~?

30·4?

?50


60

. ,, r

T

Frn. 5. Percentage at different Fennoscandian localities of white (napi-Iike) specimens of Pieris napi 9, first generation (Petersen, 1949).

merely complicates nomenclature to give names to these localized, slightly different populations or colonies. Variants within Populations. It is one of the most characteristic attributes of taxonomic categories that they are based on natural populations. They may, and usually do, include many populations, but they can never include only part of a population. Selected individuals from a population are not a population or a taxonomic category. The males of a given species may be a different biological or genetical category from

NEUTRAL TERMS FOR CATEGORIES

It is very convenient in systematic work to have some terms that can be given informally to taxonomic units, particularly in incompletely analyzed cases. These are the so-called "neutral terms." The ones that are most frequently used in taxonomy are form, for a single unit, and group or complex, for a number of units. We often speak of a form when we do not know whether the systematic unit in question is, for example, ~ full species or merely a subspecies of a polytypic species, or whether it is a subspecies or an individual variant. Seasonal and polymorphic vari-

TAXONOMIC CA1'EGORIES AND CONCEPTS

'l'Hl!J SPKCIES 11ND 'l'HE INFRASPECIFIC CATEGORIES

ants are often referred to as forms. The term is also used in the plural when two unequal units are combined. For example, when describing attributes common to a species and a subspecies of another species, one refers to the species and the subspecies as "these two forms." The term group is most commonly applied to an assemblage of closely related species within a genus. In Drosophila, for example, one speaks of the melanogaster group, the virilis group, the obscura group, the willistoni group, and so forth. The usage of the term group for such cases is popular among taxonomists, since it eliminates the need for subgenera. Occasionally the term group is also applied to subspecies in species with very many subspecies. The common Palearctic jay, Garrulus glandarius, has a total of 41 subspecies, but they can be arranged in eight groups, the garrulus group, the bispecularis group, and others. The term group is also used, though more rarely, to denote a number of closely related units in the higher categories-genera, for example. The word complex is frequently used synonymously with the term group. Terms like section, series, and division are sometimes used for groups of higher categories. Their use is, however, not standardized, and they are sometimes used above and sometimes below the family, the order, the class. They are essentially still neutral terms, corresponding to the term group. Botanists use a recently coined word, the convenient taxon (plural, taxa), in place of taxonomic category and have recently modified their rules of nomenclature accordingly.

The paleontologist deals with the additional dimension of time, and his subspecies are not all synchronous. In fact, most of the paleontological subspecies are presumably ancestors and descendants. It ha8 therefore been proposed to use two sets of terminology, one for the synchronous allopatric subspecies of the neontologist and one for allochronic subspecies. However, such a dual terminology would be impractical, for in most paleontological situations it would be impossible to determine the exact chronological relationship of allopatric populations. It is therefore most practical to follow Simpson's suggestion that the term subspecies be used in paleontology for any subspecific category, whether contemporary or not. However, a paleontologist must never forget that it is impossible by definition that two subspecies coexist at the same locality. If he finds two "subspecies" in one sample, he can be sure that they arc either intrapopulation variants (the usual situation) or different species (see discrimination grid, Chap. 5).

36

LOWER CATEGORIES IN PALEONTOLOGY

The categories of paleontology are those in general use elsewhere in zoology. However, the application of these categories is often different. The samples in paleontology are often small, frequently consisting of a single specimen. When large samples are available, they are often from a single horizon or a single exposure. It is therefore often impossible to sample the represented species adequately. This uncertainty about the species leads the paleontologist to place greater reliance on the genus and to consider it "the basic unit of practical and morphological taxonomy" (Simpson, 1945). Past paleontological practice has been to describe as a full species every sample that appeared reasonably distinct. Infraspecific categories hardly figure in the paleontological literature, except an occasional "variety," which is usually an intrapopulation variant. It is only within the last decade or so that the term subspecies has been used more freely in paleontology (Simpson, 1943; Newell, 1947; Sylvester-Bradley, 1951). It is applied, as in neontology, to populations (or samples of populations) that are sufficiently similar morphologically to make it probable that full reproductive isolation between them had not yet developed.

37

THE TAXONOMIC TREATMENT OF HYBRIDS

Hybrid specimens require special nomenclatural treatment. Hybrids are often named before their hybrid nature has become apparent. Such names become invalid as soon as the hybridity of the bearer has been established. There are three kinds of hybridism in nature: 1. Sympatric hybridization. The occasional production of hybrid individuals in the region of overlap of otherwise well-defined species is sympatric hybridization. The many hybrids in the birds of paradise and in the hummingbirds belong in this group. Such hybrids are listed as a cross of the two parental species, Tetrao urogallus X Lyrurus tetrix. 2. Allopatric hybridization. If hybridization is defined as the crossing of unlike parents, it is very difficult to make a sharp distinction between the hybridization of subspecies and of allopatric species. Until such a case is fully investigated, it is often convenient to consider as allopatric hybridization any case of interbreeding between distinct allopatric populations. No taxonomic difficulty is produced where the zone of hybridization is narrow. However, if it is wide and if a well-defined, stabilized hybrid population with intermediate characters develops, it is "ometimes convenient and justified to recognize the "hybrid" population taxonomically. The fact that such complete interbreeding prevails indicates that the two parental "species" are actually conspecific. The "hybrid population" may be named as a subspecies if it satisfies the requirements of the 75 per cent rule. The taxonomic recognition of a hybrid population is not justified if it is very variable and includes a high percentage of parental types in addition to intermediates.

38

TAX0},'0.lilC CATEGOHlb'S AND CONCEPTS

3. Amphiploidy. Hybridism in plants may lead to the instantarteous production of a polyploid that combines the chromosome s~ts of t~'9 parental species. Such a hybrid may give rise to a new populat10n wluch is reproductively isolated from the parents and which may become a n~w species, provided that it is sufficiently fertile and able to compete with . other species (including the parents). There are a few hybrid situations which do not fit mto the above scheme and for which no practicable taxonomic solution seems possible .. One of these occurs when the reproductive isolation between two otherwise well
THE SPECIES A1VD THE INFRASPECJFJC CATEGORIES

39

return to sexual reproduction under certain environmental conditions. It is unjustifiable to give nomenclatural recognition as "races" or "microspecies" to such temporary clones. Such parthenogenetically reproducing lines will terminate sooner or later either by extinction or by returning through a sexual process to the joint gene pool of the parental sexually reproducing species. Some exceptional cases are known in which a form that is as distinct as a good species reproduces strictly parthenogenetically, and no biparental species is known from which it might have branched off. N omenclatural recognition is justified in such cases. Whenever several reproductively isolated chromosome types occur within such a "species,'' as in various austaceans (Artemia salina Linnaeus) (White, 1945), it may be convenient to distinguish them nomenclaturally. Although conventionally referred to as races, reproductively isolated chromosomal populations are more logically designated (micro)species.

· , · .

CLASSIFICATION AND THE HIGHER CATEGORIES

CHAPTER 3 CLASSIFICATION AND THE HIGHER CATEGORIES The definition, description, and naming of more than 1 million species of living animals are the analytical tasks of taxonomy, as stated in Chap. 2. The other task of taxonomy is synthetic. It consists of organizing the otherwise chaotic mass of species into a classification. Such a classification is an indispensable prerequisite to the identification, cataloguing, and arrangement in collections of these species. Many systems of classification are possible. 1 Classifications of books in a library are often cited as an analogue to the classification of organisms. Such books may be classified according to subject matter or according to the initials of the author, or chronologically according to the date of publication or accession, or according to size, or on the basis of a combination of several of these classifying criteria. Species of animals, likewise, might be arranged according to the alphabetical sequence of the scientific names, or according to size, or to habitat, or to the climatic zone or geographical region in which they live. All these classifications have actually been proposed; they are more or less logical and sometimes useful if we are interested merely in certain practical aspects of classifi- ·, cation. Classifications based on similar adaptations or modes of living have been particularly popular. Pliny, for instance, classified animals into those of land, water, and air. Other early authors classified bats· with birds, whales with fishes, and all linear invertebrates as "worms." . :Many of the early classifications of the Linnaean period were also based on such adaptive features. Birds with webbed feet were classed together, . and so were those with long legs. The rodents and lagomorphs were , placed in a single order, owing to an adaptive similarity of the incisors; the artiodactyls and perissodactyls were classified as ungulates, owing to similarities in feeding habits, foot structure, and general body build. ; Actually, even some of the currently accepted categories are apparently based on convergent adaptive characters, such as the Old World warblers, ' babblers, shrikes, and perhaps titmice, among the birds (Mayr and Amadon, 1951). Aristotle, almost twenty-three hundred years ago, was the first to rea1ize that the most practical system of classification of animals is based on the degree of similarity of their morphology or anatomy. The great. advantage of this system is that it is based on the sum total of many 1

For an illuminating discussion of the principles of classification see Simpson, 1945.

40

41

rnorphological similarities or differences. Such a system is apt to indicate" natural affinities," as Linnaeus and even some pre-Linnaean authors contended, and was therefore called the natural system, in contradistinction to artificial systems, based on single characters. Even in a strictly morphological classification, the assignment of a :;pccies to a definite category characterizes it usually as possessing a very definite combination of structures and biological attributes. So perfect indeed was the agreement of taxonomic position and structural characteristics that it became a source of considerable amazement and speculation among the naturalists in the post-Linnaean period. Although they ;;poke of natural systems and natural affinities, they did not understand by these terms what we do today. To explain the orderliness of the Hatural system, some of the natural philosophers in the first half of the nineteenth century attempted to construct systems on the basis of logical eategories, similar to the periodical table of the chemical elements (Stresemann, 1950). Why these systems of horizontal and vertical columns or of concentric circles were so sterile and unsuccessful did not Lecome apparent until it was realized what it was that made the natural · ~~·stem natural. The theory of evolution solved this puzzle in a manner that was as ~imple as it was satisfactory: the organisms of a "natural" systematic category agree with one another in so many characteristics because they are descendants of one common ancestor. The natural system became :t phylogenetic system. The natural system is based on similarity; the phylogenetic system on the degree of relationship. One would expect a priori that such a complete change of the philosophical basis of classification would result in a radical change of classification, but this was by no means the case (Dobzhansky, 1951). "From their classifications alone, it is practically impossible to tell whether zoologists of the middle decades of the nineteenth century were evolutionists or not" (Simpson, 1945). The reason for the congruence is principally the fact that similarity i,, usually caused by relationship. The more closely two animals are related, the more morphological characters they will usually have in eommon. There was hardly any change even in method before and after Darwin, except that the "archetype" was replaced by the common ancestor. "The common ancestor was at first, and in most cases, just as hypothetical as the archetype, and the methods of inference were much the ':lame for both, so that classification continued to develop with no immediate evidence of the revolution in principles" (Simpson, 1945). PHYLOGENY AND CLASSIFICATION

One of the objects of taxonomists in the post-Darwinian period was to construct a classification of animals composed of monophyletic groups.

TAXONOMIC CA'l'EGORIES AND CONCEPTS

CLASSIPICA1'10N AND THE HIGHER CATHGORIES

As we have seen, this aim had already in part been anticipated by the "natural systems" of the pre~Darwinian period. However, much of this classification was still based on conspicuous adaptive characters (e.g., open incisors of rodents and lagomorphs; thickened forewings of beetles, roaches, and earwigs) rather than on true relationship. The first step, then toward the achievement of a phylogenetic classification is an analysis of the taxonomic characters to determine which of them are derived from common ancestors (homologies) and which are spurious similarities (analogies), usually convergent adaptations correlated with similar habits. This second task of the taxonomist is by no means completed in most groups of animals. It is farthest advanced in groups with an abundant fossil record, as mammals, reptiles, mollusks; it is most backward in essentially uniform groups, the subdivisions of which are largely based on adaptive specializations (e.g., birds and many parasitic groups). The development of a sound classification which is not in conflict with phylogeny can be accomplished in these difficult cases only by utilizing every conceivable taxonomic character, not only gross morphology but also cytological characters (e.g., chromosome numbers and patterns), serology, and other chemical characters, habits, ecology, and· others. For a discussion of taxonomic characters, see Chap. 6. Every one of the sources of information on phylogeny that has been used in the past has its limitations and pitfalls. This is true for genetics, physiology (including serology), embryology, and zoogeography (Simpson, 1945). It is even true for paleontology, because there are several interpretations possible for many fossil remains, particularly if they are incomplete. Still, paleontology (when fossils are available) and comparative morphology are on the whole the most productive sources of phylogenetic information. Since it is the avowed aim of a modern classification to reflect phylogeny, one might assume that classifications could not be attempted until phylogenies are clearly and unequivocally established. This is not the case. Many of our existing classifications are actually pragmatic and based on the degree of similarity, regardless of whether they reflect blood relationship or not. Such a system may occasionally be more useful than a strictly phylogenetic system. In fact, ever since the theory of evolution was accepted, there has been a conflict among taxonomists as to whether to strive for a purely practical classification or for a classification "that expresses phylogeny." Some authors compromise; they strive for the most practical classification that is still based on monophyletic groups. But the breaking up of polyphyletic groups does not necessarily lead to a more practical system, if we do not know where the fragments belong. For instance, the large flightless birds such as ostriches, rheas, cassowaries, moas, and elephant birds were long classed

in the subclass Ratites. Eventually it became evident that these forms were unrelated, and that they have become secondarily similar only in connection with the increase in size permitted by the loss of flight. Classifying these "Ratites" in five independent orders has not simplified the classification of birds, particularly since it has not yet been established to which of the remaining orders of flying birds each of the five orders of flightless birds is most closely related. A similar example has been related by Richards (1938):

42

43

Among the bees there is a number of cuckoo genera of which the larvae live as parasites in the nests of industrious species. These cuckoo bees have evolyed from industrious species and in favorable examples the resemblance is still so close that the ancestral genus is pretty certain. Yet some of the genera no

.. .. . .. 0

-~

"O 0

0

c

u

.c

a. 0 (fl u

. . .. " -. 0

~

c:

~

~ <{

0

c

~

0

c

c: 0

a..

::1'

s"0. 0

0

u

.9

x"'

Solitary bees

.. 0

~ 0

.c a. 0 .c c: <{

.:?

;:. u

0

"' "' ::1'

"c: "cx 0

0

.. u

::;:

>,

0

Q; 0 (j

. 0

c: Cl

~ (fl

"c: ,.. .c 0

"' a..

Parasitic bees

. . 0

~

.0

E

0

c:

0

a.

Cl)

<{

Social bees

Frn. 6. A practical classification of bees (after Friese, 1926), in which the parasitic bees are treated as a group independent of their close relatives among the solitary and social bees.

longer closely resemble any industrious genus. Moreover, there is a very definite "parasitic facies" dependent not only on the loss of pollen-collecting apparatus, but on the presence of bright, sometimes wasp-like colors, etc.; most parasitic bees can be recognized as such without observation of their habits. For these reasons two different classifications have grown up. One endeavors to place each parasitic genus next to its supposed ancestor. This is the phylogenetic scheme and, in general, I believe the best one, but it has the disadvantage that a number of genera are hard to place. The other scheme places all the parasitic bees together in one group which, at least in the female sex, is easily defined by the absence of pollen-collecting apparatus. Sub-groups within this assemblage roughly correspond to the various lines of ancestry. Although artificial this scheme has certain adyantages in classifying the bees of, say, Africa which are very imperfectly known.

In order to illustrate this point, we may compare the strictly practica! classification of Friese (1926) with that of Michener (1944). Friese recognizes four polyphyletic categories of parasitic bees (Fig. 6). Michener

CLASSIPICA'l'ION AND THE HIGHER CATEGOIUBS

T.lXONOMIC CATEGORIES AND CONCEPTS

44

places a number of parasitic genera in categories with their nearest nonparasitic relatives (in the Halictinae, Anthidiini, Megachilini, Euglossini, and Bombini) but in order to do so is forced to recognize fourteen "monophyletic" categories composed wholly of parasitic bees (Fig. 7).

Ceratinini

w

i3

~ 1--

GJ :!'

Mocropidinoe

8:J J: u

m :!'

Fw. 7. A phylogenetic classification of bees (after Michener, 1944) in which the parasitic bees are placed according to their supposed relationships among the solitary and social bees. Wholly parasitic groups are circled, partially parasitic groups underlined.

It is obvious from these examples that a compromise must often be made between the practical aims of classification and its phylogenetic basis. Another kind of difficulty is often encountered by the paleontologist when he finds that similar levels of morphological specialization are reached independently in unrelated or only distantly related lines. Often

45

a real dilenima develops between the convenience of a horizontal classification and the consistency of a vertical ( = phylogenetic) classification (Bl.lther, 1927; Arkell and Moy-Thomas, 1940). For instance, in the aromonite family Cadoceratidae, four well-known genera, Cadoceras, Qiwnstedtoceras, Cardioceras, and Amoeboceras, follow each other in a clearly defined time sequence. Each genus is characteristic for a given geological period. However, each genus embraces many different forms which differ in some respects (but not those of the generic diagnosis) more widely from one another than from corresponding species in other genera. These forms are classed in many subgenera. "As knowledge advances, it becomes increasingly probable that each subgenus of Cardior,eras evolved from a different subgenus of Quenstedoceras." There is thus often greater true affinity between subgenus a of Genus A and subgenus a' of genus B than between subgenus a and b of genus A. However, since the "true affinities" of many of these subgenera are obscure, and ::;ince the horizontal genera are so characteristic of well-defined geological periods, paleontologists are justifiably reluctant to replace their eminently practical horizontal classification by a more nearly "correct" phylogenetic, vertical arrangement. Still another problem is presented by unequal rates of evolution. For instance, the class of reptiles consists of many branches, some of which branched off the main stem at an early geological date. The turtles and mammal-like reptiles are such branches. One well-defined group of reptiles (pseudosuchians, archosaurians) gave rise to the birds and the following reptilian groups: pterodactyls, saurischians, ornithischians, and crocodilians. Phylogenetically, the birds are thus more closely related to the crocodilians than the latter are to the lizards or the turtles. It is evident from the last-mentioned difficulties that it would not always be easy to translate phylogeny into classification, even if all the facts of phylogeny were fully known. Where the phylogeny is still obscure, it would be only confusing to have anything but an openly practical classification. In spite of these practical difficulties it should . ' r.em~m the ultimate aim of the taxonomist to devise a phylogenetic classification, that is, a classification in which the categories are monophyletic. Such a phylogenetic system has two advantages: (1) it is the only known .~ystem that has a sound theoretical basis (something the natural philosophers of. the early nineteenth century looked for in vain); and (2) it has the practical advantage of combining forms (and there are only a few exceptions to this rule) that have the greatest number of characters in common. The difficulties of translating phylogeny into classification are twofold. (1) There is the need for expressing different degrees of relationship through a system of categories (see below, hierarchy of categories).

46

TAXONOMIC CATEGORIES AND CONCEPTS

(2) There is the need for presenting a multidimensional phylogenetic . tree graphically and, more particularly, in a linear sequence (see Chap. 8). SUBJECTIVITY OF HIGHER CATEGORIES

The assignment of rank to a given taxonomic group is a subjective• matter. This may be demonstrated by three types of evidence. Historical Differences. A genus, a family, or other category has a different value in different historical periods of taxonomy. While the· species recognized by Linnaeus are still, in most cases, listed as species,. his genera are usually completely changed. Most Linnaean genera of animals have been raised to the rank of families or even higher. Fur-, thermore at the end of a splitting period in some branch of taxonomy,· ' . each category has a very narrow definition. In the subsequent 1umpmg• . period, the limits of these categories are again expam~ed. Synchronous Differences. Even at the same penod, categones are treated differently by different authors. For imitance, in Parker and. Haswell's textbook of zoology (1940), the insects are classified as a class' of the phylum Arthropoda, and the Orthoptera as an order with four· suborders. In Handlirsch's treatment of the insects in Kiikenthal' , H andbuch der Zoologie ( l 926-1936), the Insect a are listed as a subphylum, and the Orthoptera are arranged in two superordets and four orders. GrouJl Differences. The recent birds are classified in from 20 to 5 . orders (by various authon;). There is less difference between the orders than there is between the currently recognized orders of insect or of mollusks. Likewise the families in the order Passeres (songbirds are much less distinct than the families in most other groups of animals Obviously the categories order and family do not have the same meanin for an ornithologist as for most other taxonomists. Likewise, th physical anthropologists are in the habit of giving generic ran~ to muc finer subdivisions than those so ranked by most other taxonomists. . The three types of evidence illustrate that the ranking awarded to. taxonomic group above the species is strictly subjective. However, i, does not mean that in this regard any one group of taxonomists is eith "right" or "wrong." Nor does it mean that the categories themselv. have no objectivity (see below). It is important to keep these facts I mind. THE TAXONOMIC HIERARCHY

The actual method of establishing a classification consists in definin groups or categories on a hierarchic scale. Each of th.ese. categori includes one or more groups from the next lower level, which is.the ~ex lower category. The result is that all the animals ~an be classi~ed m taxonomic hierarchy consisting of a series of categories of ascendmg rank from the species to the kingdom, each successive category embracing on

CLASSIFICATION AND 'l'HE HIGHER CATEGORIES

47

or more of the next lower category. As will be presently discussed, all these categories (except the species) are both objective and subjective. They are objective because they consist of objectively definable entities; they are subjective because the categorical level, as well as the delimitation of the categories against one another, is subjective. The function of taxonomic categories is to reduce the diversity of nature to a CQmprehensible system. Groups may be understood and remembered more readily than the countless units of which they are composed. Every classification involves two steps: (1) the arranging of lower units into groups and (2) the joining of these groups in an ascending hierarchy of more and more unlike groups. Both steps involve numerous practical, a,; well as scientific, questions. The most essential point to be kept in mind is that primarily a classification should be practical: it should create order out of chaos. As we have already emphasized, taxonomic classification existed before the theory of evolution was accepted by biologists, and even today it may be pursued without regard to phylogeny. Linnaeus, who founded the hierarchy of taxonomic categories, recognized within the animal kingdom only five, classis, ordo, genus, species, varietas. As the knowledge of animals grew (and with it their number), it became necessary to make finer divisions, of which two are now universally accepted, the family (between genus and order) and the phylum (between class and kingdom). The varietas, as used by Linnaeus, was an optional category under which were placed various types of variation (geographical and individual). The remaining categories form the basic taxonomic hierarchy of animals, any given species belonging thus to seven obligatory categories, as follows: Wolf Kingdom Phylum Class Order Family Genus Species

Animalia Chordata Mammalia Carnivorn Canidae Canis lupus

Honey Bee Animalia Arthropoda Insect a Hymenoptera Apidac A pis mellifera

The respective position of two animals in the zoological system can be expressed by this hierarchy with a fair degree of accuracy. However, in most groups of animals the need has arisen for an even more precise definition of the taxonomic position of a species. This has been accom?lished historically by a splitting of the original categories and by inserting additional categories between the seven basic ones. Most of these are formed by combining the original names with the prefixes super or sub. Thus there are superorders and suborders, superclasses and sub-

TAXONOMIC CATEGORIES AND CONCEPTS

CLASSIFICATION AND THE HIGHER CATEGORIES

classes, etc. Many other names have. been proposed for higher ca~e­ gories, but none of them is in general use, except perhaps the ter~ tribe between genus and family. Some authors, as for example Simpson (1945), use a category cohort between order and class. The generally accepted categories are the following:

the size of the unit. The latter qualification will prevent the recognition of unjustified monotypic genera. Genera are tied down by type species, and although no one species can be "typical" of a group of species assigned to a genus, the generic type serves as a fixed point for the generic concept (Chap. 14). This situation has been likened to a fiat piece of rubber nailed to a table at a single point on its surface. The rubber (generic contents) may be stre~ched in o?e direction or another by adding or subtracting species, but it always mcludes the nail (type species). The species which serves a;; the type of a genus is, in turn, tied to type specimens so that the genus is fir~ly anchored. It is only the extent or limits 'of the genus 1hat are arbitrary. The type system provides another aid in delimiting genera, i.e., all the species in a genus must resemble the type of that genus more closely than they resemble the types of other genera. The genus as a taxonomic category is based on the fact that species are not evenly distinct from one another but are arranged in smaller or larger groups, separated by smaller or larger gaps. Recognition of the genus is therefore based upon recognition of a natural phenomenon. How many s~e~ies should be included in one genus and how a genus shou.ld ?e. dehm1ted from other genera are matters for the judgment of the . md1v1dual systematist. Taxonomic characters that prove generic drntmctness do not exist. Taxonomic literature could have been spared n~any unnecessary generic names if taxonomists had kept in mind Lrnnae~s'.s (1737) warning: "The characters do not make the genus, ru.ther it is the genus that gives the characters." Generic Characters. After a group of species has been evaluated bv the taxonomist and judged to comprise a genus , it is found that such . 8 Peeies have certain morphological characters in common. These chara<·~ers are. the generic characters of the taxonomist. Although they are, P~1lmmphically speaking, an a posteriori phenomenon, nevertheless they

48

Kingdom Phylum Subphylum Superclass Class Subclass Cohort Superorder Order Sn border Superfamily (-o·idea) Family (-idae) Subfamily (-inae) Tribe (-ini) Genus Subgenus Species Subspecies

In current practice the names for tribes, subfamilies, families, and, superfamilies have standardized endings, which are added to the stem of the name of the type genus (Chap. 15). No standardized ending exist for the categories above the family (Chap. 16). THE GENUS

The genus is a collective taxonomic unit consisting of a number similar or related species. It is distinguished from all other higher cat gories by being recognized in the scientific name. The nomenclat~ proposed by Linnaeus is binominal, consisting of two names, each wit its own function. The functions which Linnaeus visualized for the tw components of the scientific name are diametrically opposite. Th specific trivial name signifies singularity and distinctness; the generi name calls attention to the existence of a group of similar or relate species-it relieves the memory. An objective criterion for generic rank does not exist equivalent, l~ us say, to reproductive isolation as a species criterion. It is there~o impossible to give an objective definition of the genus. A convemen . definition is as follows: A genus is a systematic category including species or a group of species of presumably common phylogenetic origi which is separated from other similar units by a decided gap. It is sug-i gested for practical reasons that the size of the gap be in inverse ratio to

49

8X1St.

There are few practical hints that can be given as to the choice of gPneric characters, since as Linnaeus said, "they are given by the genus." However, a genus that has no "diagnostic character," that is, no charac·ter or combination of characters that separates it clearly from other rnlated genera, is of doubtful validity. . The Meaning of the Genus. The genus, as seen by the evolutionist, 1 It i8 " a group of species that has descended from a common ancestor. a Y?ylogenetic unit. The characters of the genus are thus either the ;:r~twal characters of the ancestral species or such characters as have been Jou1tly acquired by all the species. 'l:lic genus, however, has a deeper significance. Upon clo8er '8Xami11ati on, I't is · usua 11y f oun d t h at all the species of a genus occupy a more

TAXONOMIC CXI'EG01UE8 AND CONCEPTS

GLASSIFICA1'1 0N AND THE HIGHER CA1'EGOJUES

or less well-defined ecological niche. The genus is thus a group of species adapted for a particular mode of life. The "g~nus niche". is obvio~sly broader than the "species niche," but both exist. On this theoretical basis, it is probable that all generic characters are either adaptiv~ or correlated with adaptive characters. Lack (1947) has made a particularly convincing analysis of the adaptive significance of the genera of

these extremes, making possible the recognition of genera most of which arc neither monotypic nor too large.

.50

~~p~~.

.

.

An apparent conflict between the phylogenetic and the fu~ct10nal .concept of the genus is that in certain groups unr~lated sp~c1.es acqu~re a superficial similarity owing to parallel adaptations to s1m1l~r .environments. This is particularly obvious when a loss of characters is m~olved and is often very confusing when reduction affects a whole sene~ of characters which are modified concurrently (for instance, loss of wmgs in insects is frequently accompanied by profound changes in thoracic structure, which may involve several characters to which considerable taxonomic importance is normally attributed). . . In such cases either the independent specializations (or despeciahza- . tions) of such species are not recognized generically and each of the species is included with the genus from which it originated, or a separate genus is recognized for each of the aberrant species. However.' the latter alternative leads to excessive splitting and is therefore undesirable both on practical grounds and because it obscures relationships. It is not always easy to unmask such cases of convergence. The problems relating to the Ratites (large flightless birds) and the ~nquiline .bees· have already been discussed, and many other examples might be given.:. The flattened bill in insect-catching songbirds has resulted in the recog . nition of the presumably polyphyletic family of Old World flycatchers' The loss of eyes in cave animals has led to the recognition of unnatura categories based on the character of blindness. Convergence of ~hes ' types occurs at all levels of higher classification, and the genus is n.; exception. However, a more detailed analysis usually reveals the art1 ficiality of classifications based upon convergence. . Different Significance of Species and Genus. The essential propert. , of species is reproductive isolation, the essential. property of ge~era 1 morphological distinctness (usually correlated with the o~cupat1~n distinctly different ecological niches). These two propert1~s .are. md pendent. In one group of species there may be great mult1phcat10n o, species without the acquisition of striking morphological differences resulting in "large" genera, that is, genera with many species (as in th . solitary bees). In another group, species once formed may diverge s much from other species that they require recognition as separate genera This leads to the recognition of many monotypic genera (as in the longhorned beetles). In the majority of cases there will be a balanre between'

51

THE FAMILY

Under the International Rules the family level is the highest at which <'ategories are tied nomenclaturall y to actual genera, species, and, in turn, ~pecimens. The type of a family is a particular genus, and this genus serves as a fixed point for the family concept. The family concept is elusive because of varying rates of evolution and because of different levels of knowledge in the various groups of animals. Furthermore~ as with other higher categories, whether or not a particular group is to be ranked at the family level is entirely subjective. In spite of these difficulties, it seems desirable to attempt a definition of the family, if for no other reason than to be consistent in treatment of the various categories. A family may be defined as a systematic category including one genus or a group of genera of common phylogenetic origin, which is separated from other families by a decided gap. As for the genus, it is suggested that the size of the gap be in inverse ratio to the size of the family. Like the genus, the family is usually distinguished by certain obviously adaptive characters which fit it for a particular, though somewhat broader, niche, e.g., the woodpeckers of the family Picidae, the leaf beetles of the family Chrysomelidae , etc. Unlike the genus, which is mmally confined to one or several adjacent continents, the family is commonly world-wide in distribution. An entomologist who knows the 414 families of British insects can go to Africa or even Australia and rerngnize nearly all the same families occupying similar habitats. Thus the family is a very useful category, the British entomologist having to learn only 414 names to place a total of 4,767 genera and 20,244 species. It is especially useful to the general zoologist, because each family usually presents a general facies which is recognizable at a glance, and all its species occupy a similar niche in their particular community, as, for instance, most of the thousands of species of Cerambycidae (long-horned beetles) in the world. In any given locality the various families, like the various species, are generally distinct. Decided gaps between families are the rule rather than the exception, and little or no difficulty is encountered in "keying out" families in local faunal \Yorks. Unfortunately, the situation becomes much more complicated when a world-wide study is undertaken. Families are often found to break up into different distinctive groups on each continent, and types annectent with other families are sometimes found. Relict groups may exist at the family level and defy efforts to attain a clear-cut classification. Thus in many insect groups

52

'l'AXONOM LC C.1 TEGOlUi'JS AN 1J CONCEPTS

(scale insects, aphids, water striders, etc.) a choice h~s .had to be made between enlarging the family concept beyond the limits of lo.c~l convenience or recognizing exotic annectent types as separate families and using a superfamily category for the group as ~ whol~. In e~to~ology there appears to be a trend, not necessarily desirable, m the direct10~ of the second of these choices. In ornithology a knowledge of the tropic.al . relatives of the Temperate Zone forms has led to a reduction in the number of families. For instance, a study of tropical genera has induced. many authors to consider the Old World flycatchers (Mus?~capidae),, warblers (Sylviidae), and thrushes (Turdidae) as only subfamih~s.. . .· Linnaeus did not recognize the family as a category, but it is significant that most of his genera have since been elevated to the rank of families. From this we may infer that his generic concept was not.· incompatible with our modern family concept, the difference between the genus and family being merely one of degree. With only 312 genera of animals in 1758 Linnaeus had no need for an intermediate category between order and genus. However, the number of newly discovered. animal types increased so rapidly that the early-nineteenth-century naturalists gradually evolved and universally applied the family concept . . (Chap. 15). The number of families continued to grow because of the advance m knowledge of existing animals and the discovery of new types. Thus b . the end of the nineteenth century approximately 1,700 families of ani mals were recognized (Perrier, 1893-1932, Traite de Zoologie). That th trend is continuing is indicated by the fact that Brues and Melande (1932) recognized nearly 1,000 families in the insects alone. The age of specialization has resulted in a general pushing upward the categories, subfamilies becoming families, and f~mi~es becomin .· superfamilies. Such a procedure is justifiable if, as mdicated above sufficiently distinctive types exist in various parts of the world. 0 the other hand, there is a tendency for specialists to exaggerate ~ importance of their groups and to split categories to an extent whic is inconsistent with the practice in related groups. Dividing such grou as the Chrysomelidae or the Cerambyeidae, one a group of .leaf fe~de. and the other wood borers, destroys the biological homogeneity which an important part of the family concept. Another family of Coleopter . the Scarabaeidae has been divided into 20 families, although a morph . logical homogeneity is broken up in the process. It is we~l ~o reme~~ . that the taxonomic hierarchy is indefinitely expandable withm the limi of the family, e.g., subfamily, tribe, subtribe, and even ~ivision, seri etc. The essential point, as stressed by McAtee (1926) is to preserve sense of proportion, for in dealing with higher categories, judgment mus he exercised, and com;ervatism should he the rule.

CL,1SS1FJCA.TJON AND THE HJGHEll CATEGORIES

ORDERS, CLASSES, AND PHYLA

Taxonomic categories above the family level are not based upon type genera and species. And yet, in general, the orders, classes, and phyla have proved to be the most stable categories in our taxonomic hierarchy. It is true that a few names have been changed (these categories are not rlearly subject to the International Rules of Zoological Nomenclature) and a few new phyla have been proposed within the last generation. But most of the higher categories are well known and well defined, and there is seldom any doubt as to the limits of the groups involved. As with genera and families, gaps between higher categories should be in inverse ratio to the size of the group. The higher categories, representing the main branches of the phylogenetic tree, are characterized by a basic structural pattern which was laid down early and within which the seemingly endless adaptive modifi<'ations have taken place. In general, then, the higher categories are definable in terms of a basic structural pattern, and in most cases they are widely distributed over the earth's surface and each higher category exhibits a variety of adaptations so that each of its component families, for example, occupies a particular and usually distinctive ecological niche. Except for certain highly specialized groups such as the order Siphonaptera (fleas), the order Chiroptera (bats), etc., the higher categories are not obviously or even predominantly distinguished by adaptive rharacters. As might be expected, the numbers of higher categories follow the trend noted previously for genera and families. According to recent tabulations, there are approximately 350 orders, 68 classes, and 30 phyla of recent animals (Table 2). THE VALUE OF A CONSERVATIVE TREATMENT OF THE HIGHER CATEGORIES

The higher categories are collective categories. Their function is not only to group together related species (genera, etc.) but also to serve as convenient labels for such groups. The terms Coleoptera and Lepidoptera must mean the same to all zoologists in order to have a maximum of usefulness. The same is true for families and even for genera. To reach complete unanimity is quite impossible, since the limits of higher categories are subjective. However, an effort should be made to ~aintain some standards. Wetmore (1940) recognizes 27 orders of recent birds; Stresemann (1927-1934) places the same families in 48 orders. Rensch (1934) gives a particularly lucid discussion of such problems, supported by many instructive examples.

TAXONOMIC C1iTEGOIUES AND CONCEPTS 2. THE PHYLA AND CLABBEB OF ANIMALS (Extinct groups are indicated by an asterisk)

TABLE

Kingdom Animalia Subkingdom Protozoa t Phylum Protozoa 8uhphylum Plasmodroma (Cytomorpha) Class Flagellata (Mastigophorn) Class Rhizopoda (Sarcodina) Class Ciliatoidea (Protociliata, Opalinidea) Class Sporozoa Subphylum Ciliophora (Cytoidea) Class Ciliata Class Suctoria Subphylum Amoebosporidia ( Cnidosporidia, Haplosporidia) Snbkingdom Mesozoa Snbkingdom Parazoa Phylum Porifera Class Calcispongiac Class Hyalospongiac Class Dcsmospongiae Phylum *Plcospongi
l

CLASSIFICATION AND 'l'HE HIGH!i:R CA1'l!JGORIES Phylum N emertina Class Anopla Class Enopla Phylum Entoprocta Phylum Annelida Class Polychaeta Class Oligochaeta Class Hirudinea Phylum Echiuroidea Phylum Sipunculoidea Phylum Tardigrada Class Heterotardigrada Class Eutardigrada Phylum Onychophora Phylum Linguatula Phylum Arthropoda Subphylum *Trilobita Class *Opisthoparia Class *Proparia Class *Agnostia Subphylum Chelicerata Class Merostomata Class Pycnogonida Class Eurypterida Class Arachnida ::lubphylum Man
55

50

TAXONOMIC CATEGORIES AND CONCEPTS Phylum Echinodermata Subphylum Pelmatozoa Class *Edrioasteroidea Class *Cystoidea Class *Blastoidea Class Crinoidea Subphylum Asterozoa Class Asteroidea Class Ophiuroidea Subphylum Echinozoa Class Echinoidea Class Holothurioidea Phylum Hemichordata Class Ptcrobranchia Class Entcropneusta Class *Graptolithoidca (Grnµtozoa) Phylum Chordata Subphylum Tunicata Class Ascidiacea Class Larvacea Class Thaliacea Subphylum Acrania Class Cephalochordata Subphylurn Vertehratit Class Agnatha Class *Placodcrrni Class Cho~drichthycs Class Osteichthyes Class Amphibia Class Reptilia Class Aves Class Mammalia

Nothing is gained by splitting well-established natural categories too fine. Arkell and Moy-Thomas (1940) cite some particularly extreme examples for fossil invertebrates: Examples of flagrant disregard of this rule [a uniform scale of values] are\ Buckman's innumerable genera made by splitting up contemporary species of . the single good Liassic ammonite genus Dactylioceras, and by Heinz's pulverization of the Cretaceomi lamellibranch genus Inoceramus. Out of what was originally a single genus Inoceramus Heinz created a whole systematic hierarchy, comprising 2 families, 24 subfamilies, 63 genera, and 27 subgenera; and even so he did not take the Jurassic forms into account. [Even if this minute subdivision were taxonomically justified] it should be carried out downwards in the scale, starting with the genus I noceramus and proceeding through subgenera to groups and sections. All the advantages of minute subdiYisions can thus be achieYed without disturbance to the whole classification of Mollusca.

CLASSIFICATION AND THE HIGHER CATEGORIES

57

Handlirsch (1929) also warned against the chaos produced. by excessive splitting and cited the following examples: In the family of the praying mantids Giglio-Tos (1927) arranges the known 1500 species in 30 subfamilies and 500 genera. There is thus an average of only 3 species per genus. Among the genera no less than 150 were made by Giglio-Tos himself; 175 are monotypic, 74 have two species. Roewer (1923) puts the 1700 species of opiliones in 500 genera, of which more than half (300) contain only 1 or 2 species. The order Strepsiptera with 150 species has been split by Pierce into 5 super-families, 20 families, subfamilies or tribes, and 45 genera or subgenera, of which 29 are monotypical.

Smith (1947) also points out how much is lost through too fine a splitting of genera. THE MEANING OF THE HIGHER CATEGORIES

There has been much dispute as to whether the higher categories have objective reality or not. Paleontologists are almost unanimous in considering the genus a better defined, more objective category than the species. Most botanists agree with this view. Students of birds, however, and of other well-known groups of living animals insist that the 8peries is the category that is most clearly defined in nature. Who is right? The genus of Linnaeus was very broadly defined. As has been pointed out above, it included genera that are now considered to comprise a family or an order. With some recent authors, to mention the other extreme, the genus has become so narrow that it is monotypical in the majority of cases. Linnaeus arranged all the 554 species of birds known to him into 63 genera. Nearly all his species are still recognized as such today; most of his genera have been raised to the rank of families. Some recent ornithologists allow no less than 7,000 to 8,000 genera for the 8,600 species of birds, others about one-fifth of this number. There is thus about a 500 per cent difference between the genus of the extreme lumper and that of the extreme splitter. On the other hand, the extreme 13plitter recognizes less than twice as many species of birds as the extreme lumper. It is obvious from these figures that-at least as far as birds are concerned-the delimitation of species is less a matter of subjective opinion than that of genera. To obtain a deeper understanding of the nature of the higher categories, it may be useful to consider a concrete case. Figure 8 represents a diagrammatic family tree of the 37 living species of river ducks of the genus Anas (sensu lato). This is not a phylogenetic tree, because it is not based on any information on fossil forms which might be ancestral connections of the various branches. Such a diagram-

L

59

TAXONOMIC CATEGORIES AND COl•iCEPTS

CLASSIFICATION AND THE HIGHER CATEGORIES

matic illustration of degree of relationship based on degree of similarity (morphological and otherwise) may be called a dendrogram. There is comparatively little difference of opinion among ornithologists as to the grouping of these species. It is agreed that there are several more closely bunched clusters of species, each of which forms a natural group, such as species 32-37 (mallard group) and species 7-13 (blue-winged teal-shoveler

The "lumpers," on the other hand, point out that these 37 species have numerous characters in common, and that the group as a whole is well defined and separated by a decided gap from other groups of ducks. They therefore propose to include all 37 species in a single genus. With four other genera, they make up the tribe Anatini of the subfamily Anatinae. The important point is that both antagonists agree on the essential facts of the dendrogram. The major and minor groups of species of Fig. 8 may be objectively defined, but the higher categories based on these groupings are subjective, because it is a matter of opinion how many species groups to include in a genus. The delimitation of the genus is subjective. If we accept the definiteness of delimitation as a criterion, we come to the conclusion that the genus lacks the objective basis of the species, because most genera are not separated from other genera by big, clear-cut gaps. Most genera (particularly the monotypic ones) could be united with other genera, and most polytypic genera could be subdivided equally well into smaller genera. Where the limits of the genus :-;hould be drawn in any given case is left to the subjective judgment of the individual worker. The same is, of course, true for the tribe, the family, and the other higher categories. Summarizing the essential conclusions of this discussion, one may make the following statements on the higher taxonomic categories: The genus is a group of related species, the family a group of related genera, and the order a group of related families. 2 Each of these groups i::; composed of units of the next lower categorical level, which share a number of biological and structural characters that distinguish them from members of other groups. Each group either occupies a well-defined ecological niche or, particularly the family and the order, shares at least a common adaptive pattern. The exact delimitation and ranking of the groups is often subjective. Even though the higher categories, as such, are more subjective than the species category, they appear nevertheless to have a biological and structural basis with some objective criteria.

58

37

13

22

19

FIG. 8. Dendrogram of the genus Anas of the tribe Anatini (river ducks) (after Delacour and Mayr). 1 = leucophrys, 2 = waigiuensis, 3 = angustirostris, 4 = capensis, 5 = punctata, 6 = versicolor, 7 = querquedula, 8 = cyanoptera, 9 = platalea, 10 = discors, 11 = rhynchotis, 12 = clypeata, 13 = smithi, 14 = erythrorhynchos, 15 = bahamensis, 16 = georgica, 17 = acuta, 18 = falcata, 19 = strepera, 20 = americana, 21 = penelope, 22 = sibilatrix, 23 = flavirostris, 24 = crecca, 25 = f ormosa, 26 = aucklandica, 27 = castanea, 28 = bernieri, 29 = gibberifrons, 30 = specularis, 31 = sparsa, 32 = undulata, 33 = melleri, 34 = fulvigula, 35 = platyrhyncha, 36 = poecilorhyncha, 37 = luzonica.

group). It is also agreed that certain species such as 1 2 18 19 and 30 ' ' ' ' groups ' are rather isolated, and that some of the species and species are' somewhat intermediate between others. Furthermore, it is agreed among ornithologists that the whole group of 37 species is fairly isolated from the ' other genera of the tribe Anatini. The "splitters" interpret this evidence as indicating that this group of species should be divided into at least 12 genera, many of them monotypic. They insist that this is the only way of indicating the existence of the various groups of species.

2

Monotypic higher categories are recognized by the gaps which separate them, and their level is determined by evidence extrapolated from related polytypic ea tegories.

PART 2 TAXONOMIC PROCEDURE

The selection of a suitable problem is especially important to the beginner in taxonomy. Much time and effort are wasted on groups which are too difficult or on groups which present no real problem. Some important considerations in selecting a taxonomic problem are as follows: (1) Its scope should be such as to permit completion in a reasonable period of time. (2) In little-known groups the types should be aecessible. (3) The group selected should be such that the taxonomist can collect and study it in the field. After a problem has been selected, the steps in taxonomic procedure are assembling and care of material, identification, analysis of material, preparation of descriptions, keys, bibliographies, and illustrations, synthesis of a classificat;_on, and, finally, publication. Each of these steps will be discussed in detail in the following chapters, with examples >'elected to illustrate pertinent points. A certain amount of imagination will be necessary iI\ order to apply the methods illustrated to the endless variety of animal groups. Special problems demand special answers, and even routine work suffers if it becomes stereotyped. However, although originality is an asset to the scientist, new methods or novel treatment of data must be over and above the acceptable minimum requirements of standard taxonomic procedure. In the following pages, then, are described the taxonomic practices which, through actual use, are approved by a majority of systematists. In addition to commonly used procedures, newer methods are described which, in our opinion, point the way to standard practices of the future. The methods that are of particular importance in a given group depend on the state of taxonomic maturity of the group. In birds the methods of alpha taxonomy play a rather minor role, while in some poorly known groups of acari the methods of gamma taxonomy are inapplicable. It will be evident from the discussions in the subsequent chapters which methods are of special importance to a particular worker. 61

CHAPTER 4 COLLECTING AND COLLECTIONS

Students will find material of the commoner types of animals in most large museums. If they plan a monographic treatment, they may also borrow material from other institutions, which will usually send it willingly unless it is being currently studied. Such material consists of the products of expeditions and the random collections of all-round naturalists. Material of this sort has many shortcomings: the total range of a species is rarely adequately sampled, important stages in the life cycle are often missing, and there may be no data on ecological differences in the various parts of the range, to mention only a few of its many deficiencies. It is obviously quite insufficient for the kind of analysis required by systematics on the gamma level. The modern taxonomist attempts, therefore, to round out such material with his own collections. The late .\dmiral H. Lynes, for instance, who was especially interested in Cisticola, a genus of African warblers with some 40 species, made a whole series of collecting trips to nearly every corner of Africa. He combined his collecting with a detailed study of the ecology, habits, songs, and nest construction of these birds. The result is that the genus Cisticola, formerly the despair of the bird taxonomists, is now well understood (Lynes, 1930). :\font A. Cazier, who is now monographing the 90 North American species of the tiger beetle genus Cicindela, has not only devoted several seasons to collecting them but has also engaged several other collectors. His collections now total some 80,000 specimens and are being enlarged by .),000 to 10,000 specimens annually from selected critical localities found during preliminary working out of the material. In other cases the systematic collecting has concentrated not on certair1 genera but on a definite geographic region. The most ambitious ,;ingle undertaking along these lines was probably the Whitney South Sea Expedition, operated under the auspices of the American Museum of ~ atural History in New York with the support of the Harry Payne Whitney family. This expedition visited practically every island in the South Seas, obtaining nearly complete bird collections and fair collections of other material. It operated continuously from 1921 to 1934, and its work was continued by single collectors into 1940. The student of such 1·ollections has the gratifying feeling that it is unlikely that his findings will he upset hy future discoveries (Fig. 9). 63

64

'/'AXON OM IC PROCEDURE

COLLECT I NG AND COLLECTWN S

WHERE AND HOW TO COLLECT

If the specie8 wears ur bleaches, collections should Le made at the time when specimens are in their best condition-in birds, for example, after they have completed their molt. In many invertebrates the season during which sexual maturity occurs is relatively short, so collecting should be done during that time. If various stages of the life cycle are of taxonomic importance, an effort must be made to have them all represented. If a species shows noticeable geographical variation, the collections should cover the ranges of all the subspecies and should attempt to delimit the areas of intergradation. That this is not an unattainable objective has been proved by several modern collectors such as Kinsey for cynipid wasps (1936). The Whitney South Sea Expedition collected birds so systematically that up to 1950 not a single new species or sub,;pecies had been discovered on the islands visited by that expedition. There are almost endless techniques for collecting, different for every group of animals. These are described in standard collecting manuals, a few of the better known and most recent ones of which are listed in the bibliography of this chapter. Labeling. A specimen that is not accurately labeled is worthless to the student of the new systematics. So important, in fact, is the label that it is sometimes stated jocularly that the label is more important than the specimen. Many kinds of information are desirable, but by far the most important single·piece of infqrmation is the exact locality. In forms like certain land snails that may have racially distinct populations as little as 7-2 mi. apart, the locality must be stated with great prec1s10n. If the locality is a small community, farm, hill, creek, or other geographical feature which cannot easily be found on commercial or geodetic (e.g., United States topographic) maps, its position relative to a well-known place should be added on the label ("15 mi. N.W. of Ann Arbor, Mich."). The county or district should be given with all less well-known localities. If the specimen was collected in mountains, the altitude should always be given. Additional ecological information is valuable. It is essential in forms like plant-feeding insects or host;;pecific parasites. Whenever possible, the label should be written in the field at the time the specimen is prepared. Any replacing of temporary labels by later permanent ones is a potential source of error. However, it cannot be avoided with insects when labels are printed for entire lots of specimens. .\.11 essential data should be recorded on the original labels. Data recorded in a field book are frequently overlooked and may be unavailable if the collection is divided. The original label should never be replaced by a museum label. A certain number of mistakes are always made in the transfer. If a museum label is desired, it should be added to the original label.

The modern taxonomist is a student of populations. He is therefore intere:;ted in adequate samples of these populations, samples that give a sufficient picture of the variability of these populations and permit, when necessary, their statistical analysis. Formerly a museum. retained only a few "typical" representatives of every species and considered t~e re~t of the material as duplicates. Nowadays a museum takes pnde m

FIG. 9. Camp of a field expedition in New Guinea (American Museum of Natur ; History photo).

possessing large series of specimens of each :;pccies, originating from alr parts of the range of the species. The replacement of the type concept by the population concept made such a shift of emphasis inevitable. The choice of collecting stations should be planned carefully. Ideally the working out of the geographic variation of a species should be done in two steps: (1) All the material already available in collections should · be assembled and analyzed. (2) The filling of the crucial gaps thus shown will be the goal of all subsequent collecting. Collecting sta~ions, . whenever possible, should be spaced sufficiently closely to permit. an .. accurate mapping of the range of the species. Aberrant populati?ns · occur most frequently along the periphery of the range of a spec1~s. This area should be collected particularly thoroughly. If the specie:; · 8 hows seasonal variation, collections should be made in various seasons.

U5

66

67

'l'AXONOMIC PROCEDURE

COLLECTING AND COLLECTIONS

What data in addition to locality are needed depends on the given group. Most good bird collectors, for instance, record on the label not only locality, date, and collector's name but also the sex (based on autopsy), the actual size of the gonads, the degree of ossification of the skull (important for age determination), the weight (in grams), and the colors of the soft parts. The little extra time required to take these records is more than compensated for by the added value of the specimens.

100,000 specimens per year ~re not uncommon, it is customary to cata-' Iogue accessions by lots, each lot consisting of a set of specimens from a given locality or region. Lot numbers, in turn, may refer to collectors' diaries or to other sources of information on each collection. It is also customary to note whether a lot was received as a gift or by purchase or exchange, and the name of the collector and donor is always given. In cataloguing, entries of vertebrates usually contain the following items:

THE CARE OF TAXONOMIC COLLECTIONS

The value of much taxonomic work depends on the quality of the · collections on which it is based. It is therefore necessary to say a few words on the methods of curating taxonomic collections. Preparation of the Material for Study. Bird and mammal skins are ready for study as sent in from the field by the collector. Mammal skulls have to be cleaned. Some insects should never be placed in alcohol or other liquid preservatives; others are useless when dried. Invertebrates that are preserved in alcohol or formalin are usually ready for study as preserved. Microscopic slide mounts or slides of parts of the organs may have to be prepared for the smaller forms. Instructions may be found in textbooks of microscopic technique (Guyer, 1936; Lee,; 1937). Most insects are pinned (Oman and Cushman, 1946), and the: wings are spread if they are taxonomically important (or beautiful), as in .. butterflies and moths and some grasshoppers. Species can be identified' in many groups of insects only by a study of their genitalia. Micro-I scopic slides or dry or liquid mounts of the genitalia may have to be. prepared. CATALOGUING

The methods of cataloguing depend on the group of animals. In the ·. higher vertebrates, in which collections consist of a limited number of specimens, each specimen receives a separate number and is usually catalogued separately. The cataloguing is done geographically, that is, all the specimens collected at a given locality or district in a given period of time or by one expedition are entered in the catalogue together. This greatly facilitates the subsequent accumulation of distributional data or of faunistic analyses if the collection is subsequently broken up and the specimens entered into the collection according to their systematic relationships. The cataloguing is usually done after the specimens have been identified, at least as far as the genus. This permits a permanent reference to the contents of the collection long after it has been broken up and distributed according to the system or even to other institutions. In groups in which the collections consist of large numbers of specimem;, as, for example, in most insect collections, where additions of

1. 2. 3. 4. 5. li. 7. 8.

Consecutive museum number Original field number Scientific name (or, at least, generic name) Sex Exact locality Date Collector Remarks

The collections of a museum are arranged according to a system, that is, following some generally adopted classification. The sequence of orders and families is fairly standardized in many classes of animals. Unidentified material (if not to be worked out as a collection) is placed with the family or genus to which it belongs. A properly organized and well-curated collection is a catalogue in itself, and most large museums do not maintain card files of individual specimens filed in taxonomic sequence. The maintenance of such card files, useful as they are, is too time-consuming to be practicable with the small staffs of most museums. The bird collection of the American Museum of Natural History lists on the tray labels the scientific names not only of all the available species and subspecies but also of those lacking in the collection (which is specifically stated) (Fig. 10). The names on the collection cases and trays thus constitute a check list of the known species and subspecies of birds. Such a system might be unwieldy in very incomplete collections. The maintenance of collection catalogues and card files is a timeconsuming task and should never be carried to the point where it interferes with work on the collections. A list of the accessions, however, i:,i important, particularly since it often allows the recording of additional information on localities where the specimens were obtained, which eannot be entered in full on the labels of all the specimens. TYPES OF MUSEUM COLLECTIONS

The modern taxonomist is expected to operate efficiently, supply pre('ifie information, and produce the very best in modern research. To do

68

TAXONOMIC PIWCEDURE

this his most important single asset is a large, well-curated, up-to-date collection. Museum collections are of various types, depending on the objectives of each museum and the purpose for which the collections are intended. For example, one museum may have all its material displayed as public exhibits, another may be connected with a university and be

Fm. 10. Method of storing study specimens of birds in open trays (collection case in : American Museum of Natural History).

primarily concerned with the teaching of students, and a third museum , may have as its objective the accumulation of a world-wide research col-·· lection. It is important to differentiate between these types of collections · and to adhere closely to specific objectives in each case. No deviation · or overlapping should be allowed without full realization of the additional demands on funds, personnel, space, and specimens already on hand. · Survey Collections. Some collections are devoted exclusively to surveying a particular geographic area, e.g., the former U.S. Biological

COLLECTING AND COLLECTIONS

69

Survey, the Illinois Natural History Survey, and the Zoological Survey of India. Such surveys may cover the entire biota, as in Illinois, where an effort is made to ascertain the total number of species occurring in the area specified. On the other hand, a survey may be designed to cover a particular group, e.g., the California Insect Survey, or certain economic pests, e.g., wartime airport, port, and maritime pest surveys. Survey collections may involve large numbers of specimens-often far more of a given species than is practicable in a general research collection. In an exhaustive survey a large proportion of the material may consist of immature specimens and obscure specimens of small size. This type of material is of great value for the study of variation, distribution, seasonal abundance, and density. Data should be very fully recorded and, hceause of the mass of material, should be kept with the specimens. The problem of identification is the most serious obstacle in conducting a survey an
l

70

TAXONOMIC PROCEDURE

prehensive material suitable for original taxonomic analysis (Fig. 10). They are the finest and highest type of scientific collection and should be prepared and preserved with great care. They may be private in the hands of .an .ind~vidual; semipublic, in the laboratories of a privately endowed rnst1tut10n such as an academy of science or private museum· or public, in a city, state, or national museum or in public schools o; universities. The methods employed in these various collections are essentially the same, differing only as the size of the collection may introduce special problems. Sooner or later most private research collections are given or sold to the large, privately endowed or publicly supported museums. Large museums, like libraries, are the repositories of the accumulated scientific materials of our civilization. In biology they are a priceless heritage. But far from being merely a storehouse for specimens of former years ~he several .great collections of the world today are actively develop~ mg, expandrng centers of taxonomic research. New material is being g.athered on expeditions, and specimens are being curated as rapidly as time and funds will permit. These large collections serve as the focal points for ~a~onomic research and publication. Volunteer workers aug-' ment the hm1ted staffs and carry on a share of the actual research. ' T~e inherent weakness of these research centers and, as a consequence, . the rnherent weakness of systematic zoology is the type of financial sup- ' port of museums. Money is usually available for the accumulation of material, especi~lly by spectacular expeditions, but additional support is , rarely forthcommg to care for the material or to study it. Few museums can afford to build up equally good collections in all' branches of zoology. It is advisable, in fact, that each museum should ( lay special emphasis on the accumulation of as complete a collection as ?ossible fro~ particular areas or of certain groups that are of special· mterest to its curators. In view of the cordial relationships now cus- · ~omary among museums, such specialization will lead to greater efficiency ~­ m the working up of collections. Type Collections. Original descriptions and all subsequent descriptions and illustrations must obviously be based on actual specimens. With advancement in technique and knowledge it almost inevitably happens that such descriptions are deficient. This may be no reflection on the original worker who failed to see in 1840, for example, that the entire classification of the Trichoptera (caddisfiies) would eventually be based on the structure of the male genitalia. The use of new characters often leads to the discovery that a so-called "species" is actually composed of several similar species. In such cases it is essential to refer to the type specimen to determine which of these species should bear the name originally given. This function of the type

COLLECTING AND COLLECTIONS

71

:;pecimen, discussed in greater detail in Chap. 12, explains why types are 110wadays considered with such respect, and why every effort should be made to ensure the safety of these irreplaceable specimens. It has been argued that a specimen used as a basis for description or illustration becomes common property, belonging to science rather than to an individual. For this reason as well as for safety and to make them available to as large a group of zoologists as possible, types are customarily deposited in large collections of public or private institutions which have come to be recognized as standard type repositories. Although some zoologists advocate centralization of types in a single world clearinghouse (to avoid the necessity of traveling thousands of miles to obscure corners of the globe to see types), and others advocate the deposition of types in collections as near as possible to the place where they were collected, the present haphazard system is likely to continue for many years. The vast majority of types today are preserved in the collections of large and presumably permanent institutions with full-time curators to care for them. The vexing question of the lending of types is dealt with by each institution in its own way. Many institutions have entrusted types to the mails in order to lend them to competent individuals for monographic work. Although fraught with risk, such a practice has resulted in the correct placement of many species when the original types have since been destroyed. . Ideally, types should be housed in a separate collection to facilitate removal in case of emergency and to avoid the constant handling inherent in a general study collection. They should be clearly labeled with the distinctive colors described (see Chap. 12). If not previously catalogued, they should be numbered individually for convenience in referring to them in the literature and to facilitate finding the specimen in the collection. They may be arranged either systematically or chronologically as received or alphabetically according to the originally given scientific name. A card index by genera and another by species will save much time in locating the desired type.

IDENTIFICATION AND TAXONOMIC DISCRIMINATION

CHAPTER 5

IDENTIFICATION AND TAXONOMIC DISCRIMINATION After the collection and preparation of material, the next step in taxonomic procedure is identification. It is an integral part of all taxonomic work. Identification is the utilitarian side of taxonomy. The routine task of identification is in one sense a stumbling block to progress, because it ; occupies much time and effort that might otherwise be devoted to mono- 1 graphic studies. Ironically, it is only through such monographs that routine identification is made possible. Yet, in another sense, identification is the groundwork upon which all progress in taxonomy is based. , The great collections of the world are accumulations of the identifications of past generations and, as such, are the storehouses of the raw material . of our science. Identification to the species level may be a difficult task ·n groups with large numbers of species and scattered taxonomic literature. It is now: impossible for one person to make authoritative or even reasonably · satisfactory identifications in all groups of animals. Recognizing this · fact, Federal and state agencies charged with the responsibility of identifying economically important animals have resorted to employing special- : ists, each of whom is assigned a particular group. It is the duty of these specialists to make available to others their knowledge and experience,• but it is also the responsibility of the beginner, nonspecialist, quarantine inspector, or economic zoologist to show consideration to the specialist and identify his own material in so far as the available ,literature and collections will permit. Only in this way can the specialist render the, real service for which he alone is fitted: making determinations of difficult forms, confirming identifications to be used in publications, and : preparing original monographs in order to facilitate future identification. 1 There is no better way to learn taxonomic procedure than to try to . identify material with the help of a good monograph. The student is referred to Roger Smith's (1942) Guide to the Literature in the ZoologicoJ, Science$ or to the volumes of the Zoological Record for references to more recent monographs. If necessary, a specialist may be consulted to recommend a suitable publication. Segregation of Material. Some animals are segregated roughly in the field. A careful collector of small animals, such as insects, usually keeps 72

73

1:>pecimens of different species from different hosts in separate containers. But this is only a crude segregation and may not even follow taxonomic Jines. After the specimens are mounted and labeled, however, they are :oegregated taxonomically. This process is carried as far as the knowledge of the collector permits. Beginners may have trouble placing some specimens in the correct order, whereas specialists may carry their rough segregation to the generic level or ~ven down to species. In museums the usual practice is to segregate unstudied material according to orders and, in the case of easily recognized groups of some size, to families or even genera. This procedure varies, of course, with the animal groups involved. One further step in segregation is very useful. Most species have a distinctive facies, and a taxonomist with a good "eye" can save a great deal of time by assembling all specimens of a given species in one place. The eye in this case may be assisted by reference to the labels bearing localities and dates of collection. Steps in Identification. Every species name is based on a published description or figure and usually also on a type specimen. Identification, or determination, is the association of other specimens with the appropriate description or type specimen. This may prove difficult for any of Hcveral reasons: the general classification of the group may be so poor or so neglected that it is difficult to determine the genus or higher catc12;ory involved; the description may be inadequate or inaccessible; the type may not be "typical" (in the zoological sense), or it may be inaccessible or loHt; and finally, the specimen may represent an undescribed species. In spite of the inherent difficulties, correct identifications are possible for a majority of specimens in most groups of animals. The steps in identification are as follows: (1) preliminary key to orders and families, (2) key to genera and species if recent monographs or faunal works are available, (3) reference to the most recent catalogues, (4) reference to eurrent bibliographies for literature published subsequent to the most recent catalogues, (5) reference to original descriptions, (6) comparison with authentically determined specimens or with the type. . Preliminary Key to Orders and Families. This step is very necessary for the beginner and is best done with the simple keys given in general textbooks or handbooks. Even the advanced student may encounter unusual species or immature or exotic forms which cannot be placed in the proper family or order on sight. However, modern works are gener~lly available which provide family and subfamily keys and greatly assist Ill this stage of identification. .Dr~ver's (1950) Name That Animal is a good elementary guide to the prmc1pal groups of animals. In addition to the general keys, a bibliography to the most important works on each group of animals is given.

74

'l'AXONOMIC PROCEDURE

For example, Brues and Melander (1932) is cited among the _g~neral works on insects, and this, in turn, gives keys to the orders, fam1hes, and subfamilies and cites the more important monographs under each group. There are many works on the animals of specific regions (e.g., Park, Allee, and Shelford, 1939, for the Chi~ago regio~): There is an excellent survey to taxonomic works dealing :vith the Br1t1sh Isles (Smart, 1942). Other regional works are Die Tierwelt D~utschlands ~Da?l, 1925 et ~eq.), Faune de France (1921-1950 and continumg), and Die Tierwelt der Jl.iordund Ostsee (Grimpe and Wagler, 1925 et seq.). There is, unfortun~tely, ; no comprehensive bibliography of regional taxonomic treatises available for the United States. Key to Genera and Species. Identification is relatively easy if a thorough monograph of recent date is available. In this case, _the ' specimen is run through the keys; the description ~f the. appropriate species is checked, character by character; the specimen is compa_red _ with any illustrations that may be given; and the record~d ge_ogra~h1c~l distribution is checked. If all these points agree, the ident1ficat10n is: considered as tentatively made, subject to comparison with authenti0'1 specimens, and provided that no additional species have subsequentlY: been described. Reference to Recent Catalogues. In the absence of a mon?graph revision, or for the period since such a monograph was published, th _ most recent catalogue of the group should be consulted. The catalogu will give literature citations to the descriptions of all specie~ known up t , the time of completion of the catalogue. Some catalogues give more tha this, e.g., complete bibliographies under each genus and _species, lists . ., synonyms, and geographical distribution. Identification is grea_tly_fac1li tated by a good catalogue, because it brings together the _most s1gmfic~n,, published references in the group and guides the taxonomist to the spec1 :, most likely to occur in the territory in which his specimens were collected~:: Reference to Current Bibliographies. Catalogues are inevitably o~t of, date soon after they are published. This difficulty may be partially. compensated for by the issuance of supplements. Nevertheless it is not at all unusual to find the most recent catalogue dated twenty years previously. In some of the major insect orders there is no general catalogue: :oince 1900, and some groups have never been catalogued from a world :otandpoint. . Fortunately, there exists an annual bibliography of the literature_ in_ systematic zoology. This great reference work is _call_ed the Zoologic'!1 _ Record. It is the most indispensable reference publication for taxonomic work. Starting in 1864, the Zoological Record has appeared each y~ari llp to the present time. Each new scientific name is given, together with ' a reference to the place of publication and the type locality. The names

IDEN'l'IFICA'l'ION AND TAXONOMIC DISCRIMINATION

75

are arranged alphabetically under families, but a systematic arrangement is followed for families and higher groups. Current numbers are available separately by purchase or subscription. In the United States it can be ordered through the Society of Systematic Zoology. The usual method of using the Zoological Record is to start with the most recent volume and work back to the date of completion of the most recent catalogue or revision. The particular genus or other group in question may be located in the table of contents of the section devoted to the particular class of animals involved. New names, synonymy, distribution, and in some instances, even biological references are given. If the citation is not clear because of its abbreviated form, or if the exact title of the publication is important, reference may be made to the bibliography of papers arranged according to authors at the beginning of the section. For special needs there is an elaborate subject index covering various phases of morphology, physiology, ecology, and biology. The Zoological Record is published by the Zoological Society of London in cooperation with the British Museum (Natural History) and the Commomvealth Institute of Entomology. The following 19 sections of the Zoological Record are published separately and may be obtained singly or as an entire volume each year: (1) Comprehensive Zoology, (2) Protozoa, (3) Porifera, (4) Coelenterata, (5) Echinoderma, (6) Vermes, (7) Brachiopoda, (8) Bryozoa, (9)_j\follusca, (10) Crustacea, (11) Trilobita, (12) Arachnida, (13) Insecta, (14) Protochordata, (15) Pisces, (16) Amphibia and Reptilia, (17) Aves, (18) Mammalia, and (19) List of X ew Generic and Subgeneric Names. Some groups of animals have never been catalogued or monographed. This is especially true of insects. In such cases it is necessary to work back through the entire Zoological Record (Vol. 1, 1864). Prior to 1864 the best annual review of the taxonomic literature is found in the Berichte uber die wissenschaftlichen Leistungen in different branches of zoology, including entomology and helminthology, published in Wiegmann's Archiv fur Nat11rgeschichte (Berlin, 1835 et seq.). Additional important bibliographical helps covering this early period of zoology are Engelmann (1846), Agassiz and Strickland (1848), and the C::talogue of scientific papers published by the Royal Society (18001863). Sherborn's Index Animalium (1758-1800, 1801-1850) gives a complete list of generic and trivial names proposed up to 1850. The Zoological Record is always one or two years behind, so other bibliographies must be consulted for the most recent literature. Biological Abstracts (1926 to date) is an important source of recent literature. Its section, Systematic Zoology, contains abstracts of taxonomic papers and hence is a valuable source of information for papers which are not immediately available elRewhere. However, Biological Abstracts rovers the taxo-

l

76

TAXONOMIC PROCEDURE

nomic field very incompletely, so it is no substitute for Record. There are numerous zoological bibliographies dealing with zoology as a whole, or with vertebrate zoology (Wood, 1931), or with special groups,, such as birds, fishes, or other taxonomic subdivisions. The student of , taxonomy is advised to familiarize himself with the bibliographical aids · that are available in his special field. aid greatest the are keys Although Descriptions. Reference to Original to identification, reference should always be made to original or more. r~~ent authoritative descriptions. Unless this is done, there is a possi~ bility that the specimen in question represents a species not included in; the key. Original descriptions are located by means of catalogues, mon<>"' graphs, the Zoological Record, or other bibliographical sources as described: above. Copies of original descriptions may be difficult to find. Even the largest libraries are not complete, and the average university library' will be found wanting from 5 to 25 per cent of the time. This is not so much a reflection on the caliber of libraries as it is evidence of the extent·; and diversity of scientific publications throughout the world. Althougl{ largely confined to a half-dozen languages, taxonomic papers are pub;;. lished in practically every country in the world. This poses a very rea : problem for libraries with limited budgets. The situation is further com plicated because the law of priority places a premium on the earli works. No taxonomic work since 1758 becomes "out of date" if it con tains new names, and as a result of limited editions, losses through th. years, and other factors, there are not enough copies available to suppl : .; all biological libraries. The search for original descriptions in connection with identification. involves full use and familiarity with all available scientific libraries, reference to the Union List of Serials to locate publications in other libraries for interlibrary loan, extensive use of microfilm services, and'1 accumulation of reprints by purchase or by exchange with other workers. t Descriptions are the foundation of taxonomy, since only the printed~ word is indestructible. Types may be lost, and the original author is available to pass on "his" species for only a brief span of years. Descriptions should be read several times, first to obtain a general impression or mental picture of the actual specimen which the original i author had before him. Then particular characters which the original , or subsequent authors considered important should be extracted and checked against the specimens in question. Finally, any comparative notes given by the original author should be checked. In many cases · such comparative characters are the most useful clues to identification. i Original descriptions are, normally, the court of last appeal for pur-

IDENTIF'ICATI ON AND TAXONOMIC DISCRIMINAT ION

77

poses of general identification. However, many original descriptions a.re totally inadequate. This is particularly true of the descriptions published prior to 1800. The value of a description is in direct proportion to the judgment of the author and his ability to select significant characters and describe them in words, and the extent and nature of the material available to him at the time of description. For this reason descriptions given in a thorough and authoritative monograph of recent elate are usually much more usable than original descriptions. Illustrations are often equally as valuable as, or more valuable than, original descriptions. In popular groups, such as birds or butterflies, there are many works with colored plates. Such works are often a great help in the rapid identification of specimens. It is, however, advisable to check such tentative identifications hy comparison with previously identified material or to take other precautions. Colored plates are not always well reproduced, and there are many opportunities for error if too much dependence is placed on them. If the original description is accompanied by an illustration, the diffieulty occasionally arises that characters of illustration and description are in conflict. It can sometimes be proved in such cases that the artist did not have access to the type specimen and utilized another specimen that was believed to agree with the type.yuch discrepancies occur not infrequently in the work of early authors. Comparison with Types and Other Authentically Determined Specimens. It is sometimes impoS'Sible to make a satisfactory determination from the literature alone. Such a situation exists if the group has been neglected, if the keys are inadequate, or if the descriptions are poor. Even under ideal conditions, identification is greatly facilitated if types or other authentic material are available for comparison. Comparison of specimens is a highly technical job and requires a contiiderable background of knowledge and preparation in the particular group in question. For this reason preliminary identifications made by direct comparison with authentic collections, without first studying the literature and the significant characters in the group, are often valueless. Reference collections are often accumulated for the express purpose of facilitating identification. In such cases comparison is made with whatever series of specimens is available, and it is necessary to judge whether the specimen in hand falls within the possible range of variation of a given species. Care should be taken not to rely exclusively on comparison with supposedly authentic specimens. Even "authoritative " collections may contain wrong identifications or may be incomplete. In such cases hasty comparison without the other steps in identification may lead to erroneOUR eondusions.

l

78

TAXONOMIC PROCEDURE

Type specimens are the.most authentic of all but should not be used for routine identifications. Ideally, in the course Of a monographic study• of a group, all type specimens should be reexamined. At this time the.: significant characters are usually known and can be checked, using the'. same technique and the same interpretation of the characters as are.'. applied to the rest of the material. In work with subspecies it is not always necessary to have type specimens for comparison (if there is no question as to the identity of the/ species). On the other hand, a series of specimens from the type locality (" topo-typical specimens") is desirable to provide information on the. characters and variability of the subspecies. Determination Labels. Each specimen or each series should be labeled at the time an identification is made. The determination label should give the scientific (generic and specific trivial) name and author and in: addition should give the name of the determiner and the year in whichi the identification was made. With this information on every specimen,: the authenticity of the determination is established, and its dependability~ may be readily evaluated at any subsequent date on the basis of progre which may have been made in the study of the group during the inter..:\. vening yearn. In bird and mammal collections these names are usually written in pencil so that they can be changed easily if there is a chang · of nomenclature. It sometimes happens that existing knowledge and available materia . are not suffieient to place a specimen positively. In such cases the specf men should be set aside to await further material or evidence. Tentativ or doubtful identifications should always be clearly indicated as such b means of a question mark. Unfortunately, most large collections con tain numerous examples of doubtful or hasty determinations. In suchl cases the original specimen in a series may be quite authentic, but speci-· mens may have been added subsequently with little or no regard for th; critical characters of the particular species and without individual deter' mination labels. The result may prove to be a hodgepodge includin · several species or subspecies. Obviously, this procedure is not only con fusing but delays and complicates the task of identification. TAXONOMIC DISCRIMINATION

When identifying material, particularly in less-known groups or from' less-known regions, the taxonomist frequently comes across specimens that defy classification. They do not key out properly, or they do not agree with the specimens of the species with which they key out (or ii: close analysis reveals that specimens identified as a single species appea. to belong to two or three distinct species). The question inevitably, arises, Do these specimens belong to an undescribed species? In the

lDENTIFICATION AND TAX01V01iJIC DISCIUMINATION

7fl

past, this question has often been too hastily answered in the affirmative, with the result that we now have thousands of synonyms in zoological nomenclature. A great many of these might have been avoided if the describer had observed a few elementary precautions. A worker may greatly minimize the possibility of error if he (1) asks himself the basie questions of the taxonomist and (2) studies the discrimination grid (Table 3). This procedure is equally useful when applied to the classifiTABLE

3.

DISCRIMINATION

GmD

Not reproductively isolated Morphologically identical: Sympatric.... . .. (1) Same population Allopatric ............ · (2) Same subspecies :\Iorphologically different: I Sympatric..... (3) Individualvariantsofthesame population Allopatric ............. (4) Different subspecies

Reproductively isolated

(5) Sibling species (6) Sibling species

1

(7) Different species

I(8) Different species 1

cation of previously described but still dubious entities. In the preparation of any taxonomic revision the student is apt to come across caseR in which he has to disagree with previous treatments. The Basic Questions of the Taxonomist. If a taxonomist hasMore him two samples of specimens which may or may not belong to the same taxonomic category, he should ask himself the following three fundamental questions: 1. Is it likely that the two compared samples (when they are sympatric) were drawn from the same population or not? If it is not likely, he should ask, 2. Do the two populations (from which the samples were collected) he long to the same species or not? If the two populations appear to be conspecific, he should ask, 3. Is the difference between the two populations sufficient to merit ~ubspecific separation or not? Well-defined species are usually characterized by (1) morphological and physiological differences, (2) reproductive isolation, and (3) ecologir·al differences. If a taxonomist has accurate information on these three points, he is rarely in doubt whether or not to classify a population as a species. For subspecies the additional information of geographical relationship is highly important, subspecies being allopatric. The Discrimination Grid. In order to determine whether two given samples are intrapopulation variants, subspecies, or species, the working taxonomist relies chiefly on three sets of data: reproductive isolation,

80

TAXONOMIC l'JWCEDURE

presence ur absence of morphological differences, and geographical re tionship. On the basis of this information a table with eight square· can be constructed which we shall call the discrimination grid (Table 3).· All three sets of information are not usually available to the taxonomist; The determination of the presence or absence of morphological difference . is no major problem to the experienced taxonomist (except in the case o sibling species). Geographical relationship can nearly always be deter mined in properly labeled material. It is the lack of information o reproductive isolation which causes most of the difficulties. If we loo at the discrimination grid, we see at once that without information o reproductive isolation we may have difficulty in choosing between th following alternatives: 1 or .5, 2 or 6, 3 or 7, 4 or 8. A high percentage .· taxonomic errors is due to the wrong choice in one of these four alte . natives. What can we do-~in the absence of direct information o reproductive isolation~to avoid error? Fortunately, there is a grea/ deal of indirect information available which may help us to reach t right decision. Sibling Species (Alternative 1 versus 5, Alternative 2 versus 6). It one of the most interesting findings of painstaking modern taxono work-often coupled with ecological analysis, chromosome deter nations, and breeding tests--that in many taxonomic groups, exceedin similar populations may coexist side by side without interbreedi They satisfy every species criterion except that of morphological tinctness. Mayr (1942) has coined the term sibling species for su: pairs or groups of morphologically nearly or completely identical speci translating into English the previously existing terms Geschwister-A (Ramme, 1930) and especes jumelles (Cuenot, 1936). They have been referred to as biological or physiological species (or races), cryp species, and phenons. Mayr (1948), in a recent survey, has shown how widespread and rer'" tively common sibling species are in the animal kingdom. The discove of sibling species is possible only in groups which are either very we . known taxonomically or to which particularly refined methods of analysi · have been applied. Such methods attempt to prove reproductive is lation, either directly or by establishing discontinuities between the nearly identical populations. Suitable methods are either biometric (applied particularly in fish taxonomy), cytogenetic (e.g., in Sciara an Drosophila), or combined taxonomic-ecological (e.g., in Anopheles); Differences in parasite faunas may also provide clues on sibling species Experimental evidence, useful as it is, is not necessarily conclusive, sin the degree of reproductive isolation between. two species may be differen in the laboratory from what it is in nature, owing to changed ecologica conditions. The absence of morphological differences is a negativ .

IDEN'l'IFICATION AND TAXONOMIC DISCRIMINA'l'lON

81

character which alone is nondefinitive. Distributional data are not decisive, since sibling species may be sympatric or allopatric. :For this reason elaborate studies of sibling species have thus far been carried out only in groups that are medically (e.g., Anopheles), genetically (e.g., Drosophila, Paramecium), or otherwise of special significance. It is therefore impossible to give the approximate percentage of sibling species in various orders. However, they appear to be particularly common among Diptera (Drosophila, Anopheles), Hymenoptera (ants), Lepidoptcra (especially moths), and Protozoa (Paramecium). Sibling species are not a different type of species; they are merely those species that are near the invisible end of the spectrum of morphological species differences. They grade imperceptibly into species that are morphologically more and more distinct from one another. Morphological differences are often eventually found after a particularly painstaking scouting of previously unstudied structures. Sibling species are obviously inconvenient to the museum taxonomist. Often specimens of sibling species cannot be recognized in preserved material. However, since species are not a creation of the museum taxonomist but phenomena of nature, it is impossible to ignore the existence of sibling species. The museum worker will be unable in many cases to do better than to label museum specimens from a group of sihling species by the group name, e.g., Anopheles maculipennis group. Individual Variants or Different Species (Alternative 3 versus.;71? Different individuals that belong to the same interbreeding population may be very different. This intrapopulation variation, also termed individual variation, has been the source of much confusion to taxonomists. It is estimated that more than half of all synonyms owe their origin to an underestimation of individual variation. A careful study of the phen~mena .of individual variation in general, and specifically in the group with which the taxonomist is concerned, is an indispensable prerequisite of all sound taxonomic work. A thorough knowledge of all the possible forms of individual variation ic neeessary whenever the taxonomist is forced to make a decision as to wh~ther re1taiu speeimens represent a different species or individual ,.~l'l~nts. The tabulation and discussion of the major types of variation withm a single population, as shown on page 82, may be helpful. I. Extrinsic (Noninherited) Variation. Whether a certain variant within a population has a genetic basis or is merely a noninherited modification is difficult to determine, especially in museum specimens. Never~he~es~, it is_ import~nt for the taxonomist to know that various types of arrnt10n exist, and m better known groups, field observations and experiniental evidence are often sufficient to decide the precise status of a given variant.

l

82

TAXONOMIC PlWCEDURE

A. Progressive Individual Variation. 1. AGE VARIATION. Few species of animals are born in such an advanced condition that they resemble the:' . adults of the species. Animals, whether they are born more or less developed or whether they hatch from an egg, in general go through a series of juvenile or larval stages in which they may be quite different' from adults. The catalogues of any group of animals list numerous synonyms that are due to the failure of taxonomists to recognize the: relationship between various age classes of the same species. MAJOR TYPES 01' V ARIATIO:\ I. Extrinsic (noninherited) variation A. Progressive individual variation 1. Age variation

WITHIN A SINGLE POPULATION

2. Seasonal variation B. Social variation (social polymorphism) C. Ecological variation 1. Habitat variation (ecophenotypic) 2. Host-determined variation 3. Density-dependent variation 4. Climatically induced variation 5. Heterogonic variation 6. Neurogenic color variation D. Traumatic variation 1. Parasite-induced variation 2. Accidental and teratological variation 11. Intrinsic (inherited) variation _-1. Sex-associated variation 1. Primary sex differences 2. Secondary sex differences 3. Alternating generations 4. Gynandromorphs 5. lntersexes B. Non-sex-a.ssociated variation 1. Continuous variation 2. Discontinuous variation (genetic polymorphism) 3. Sex-limited polymorphism

In reptile8, birds, and mammals there are no larval stages, but imm ture individuals may be rather different from the adults, particularly in birds. For example, Linnaeus described the striped immature goshawk' (Accipiter gentilis Linnaeus) as a different species (gentilis) from th. adult (palumbarius) with its crossbarred underparts. Several hundre · bird synonym" are based on juvenal plumages. By finding specimen. that molt from the immature into the adult plumage, it is usually easy to; clear up this difficulty. · In many fishes the immature forms are so different that they have be described in different genera or even families. The immature stages the eel (Anguilla) were originally described as Leptocephalus brevirostris.

IDENTIFICATION AND TAXONOMIC DISCRILlflNATION

83

Kaup. The unmasking may be difficult in neotenic animals, that i8, animals that may become sexually mature in a larval stage . The difficulties for the taxonomist are even greater in groups with larval stages which are so different that they have not even the faintest resemblance to the adult (e.g., caterpillar and butterfly). The floating or free-swimming larvae of sessile coelenterates, of echinoderms, mollusks, aud crustaceans are often extremely different from the adults. The taxonomic status of such larval stages can be settled either by establishing the chain of intermediate stages or by rearing them. The taxonomic identification of larval stages of parasites of which the different stages occur on different hosts is particularly difficult. It is customary in helminthology to assign formal taxonomic names to the larval (cercaria) stage of flukes (trematodes), in order to facilitate their ready identification. Such dual nomenclature is, of course, dropped as 800 n as it becomes known to what trematode species a given cercaria belongs. This can be established only through rearing. Age variation is not restricted to differences between larval stages and adults but occurs also among adults. In various species of deer (Cervus, etc.), for example, it is known that older stags have antlers with more points than younger ones. The shape of the antlers may also change. This age variation must be considered when the antlers of different species are compared. There is probably no more addition of points (or only an irregular one) after a certain age has been reached. It would theref()(e be as risky to try to determine the exact age of a stag by the number of points of its antlers as to try to determine the age of a rattlesnake (Crotalus) by the number of rings in the rattle, or the age of a hornbill (Aceros plicatus Forster) by the number of folds in the casque on the bill. It is the aim of the taxonomist to work with samples that are as homogeneous as possible. It is much easier to achieve this in animals that have a definite adult stage (after the larval one) than in those that show continuous growth, such as snakes or fishes, which may reach maturity after having attained only half or less of their potential size. In such forms as the latter it is better to work with ratios of absolute measurements (proportions) than with the measurements themselves. }lany meristic characters (e.g., number of scales or fin rays) are not inereased aft.er they are formed, in spite of the enormous subsequent i!:rowth. Hence the importance of these characters in herpetology and ichthyology. In birds it is generally assumed that final size is reached with the first eompletely adult plumage. There is some evidence that this is not always so. In the hornbill, Aceros plicatus, from the Papuan region, it is well established that "adults" with only two or three folds on the bill are younger than those with five or more folds. Birds in adult plumage

'l'AXONOMLC PROCEDURE

with five to eight folds have a bill length of 198 to 227 mm. (average 212.3 mm.), while birds in equally adult plumage with two or three fold, have a bill length of 185 to 199 mm. In some passerine birds it has been possible through banding to show that the average size of known adul. · individuals increases slightly over the years (Lang, 1946). 2. SEASONAL VARIATION OF THE SAME INDIVIDUAL. In animals live as adults through several breeding seasons, it happens not infre.' quently that the same individual has a very differ.ent appe~rance i . different parts of the year. Many birds have a bright nuptial dres which they exchange for a dull plumage at the end of the breeding season, Among North American birds this is true, for example, for many ducks shore birds, warblers, tanagers, and others. In many cases such a chang. ·~ . of plumage is restricted to the males. In arctic and subarctic birds and mammals, such as ptarmigan (Lagopus) and weasels (Mustela), there may be a change from a crypti white winter dress to a "normally" colored summer dress. In oth birds the colors of the soft parts change with the seasons. In the comm egret (Egretta alba Linnaeus) and in the European starling (S~urn vulgaris Linnaeus), the bill may change from yellow to black; m t black-crowned night heron (Nycticorax nycticorax Linnaeus), the le· may change from salmon-colored to olive, etc. The plumage chan~. in birds are usually effected by molt, but wear alone may produce stri ing changes. In the European starling (Sturnus vulgaris), for exampl' the freshly molted bird of October is covered with white spots, and all t feathers show whitish or huffy margins. During the winter the ed of the feathers wear off, and in the spring, at the beginning of the bree, ing season, the whole bird is a beautiful glossy black without the m , of a single feather. A similar process of wear brings out the full col of the nuptial plumage in the males of the linnet (Acanthis. cannabi · Linnaeus), the snow bunting (Plectrophenax nivalis Linnae~s), the hou: sparrow (Passer domesticus Linnaeus), and many other birds. In an regions, particularly in real deserts, the sun bleaches the pigments. bird before the molt will look much paler than one in freshly molt · plumage. In all these cases it is the same individual which in different parts o the year looks very different. Such seasonal variation is particular! · common among vertebrates, with their elaborate endocrine systems Many such seasonal variants were described as distinct species befor their true nature was realized. B. Social Variation (Social Polymorphism). In the social insects such as some bees and wasps, but particularly among ants and termit . "castes" have developed. These are definite groups of individuals withi · a colony, such as females (queens), workers (sometimes of different

llJ.liN'l'lFLC.tl'l'LON ANJJ TAXONOl'dlC JJLSCRLMLNA'l'LON

85

types), and soldiers (also sometimes of different types). In the Hymenoptera, these are most commonly modified females and genetically identical (except for the workers in some social bees; Kerr, 1950), but in the Isoptera (termites) both sexes may be involved. The structural types observed may result from different larval food or may be due to hormonal or other controls. Obviously taxonomic names should not be applied to these intracolonial variants; but invalid species have sometimes been described because it was not realized that there were different types of soldiers or workers in the same colony. C. Ecological Variation. 1. HABITAT VARIATION (ECOPHENOTYPIC). Populations of a single species that occur in different habitats in the same region are often visibly different. The taxonomic treatment of such local variants has fluctuated between two extremes: some authors have described them as different species; others consider them all as nongenetic variants. Actually they may be (a) microsubspecies (or ecological races) or (b) nongenetic ecophenotypes. The latter are particularly common in plastic species, such as some mollusks. Dall (1898) gave a very instructive account of all the variations he observed in a study of the oyster (Ostrea virginica Gmelin): The characteristics due to situs may be partially summarized: When a specimen grows in still water, it tends to assume a more rounded or broader form, like a solitary tree compared with its relatives in a crowded grove. When it g;rows in a tideway or strong current the valves become narrow and elongated, usually also quite straight. Specimens which have been removed from one situs to another will immediately alter their mode of growth, so that these facts may be taken as established. When specimens are crowded together on a reef, the elongated form is necessitated by the struggle for existence, but, instead of the shells being straight they will be irregular, and more or less compressed laterally. When the reef is dry at low stages of the tide, the lower shell tends to become rleeper, probably from the need of retaining more water during the dry period .... When an oyster grows in clean water on a pebble or shell, which raises it slightly above the bottom level, the lower valve is usually deep and more or less sharply rndially ribbed, acquiring thus a strength which is not needed when the attachment is to a perfectly flat surface which acts as a shield on that side of the shell. Perhap
In fresh-water snails and mussels such habitat forms are particularly common. The upper parts of rivers, with cooler temperatures and a

~

86

more rapid flow of water, have different forms from the lower reaches, with warmer and more stagnant waters. In limestone districts the shells are heavy and of a different shape from those which grow in waters poor in lime. This dependence of certain taxonomic characters on environmental factors was, curiously enough, entirely overlooked by some earlier'., workers, a fact which resulted in completely absurd systematics. Schnit-: ter (1922), who largely cleared up the situation, described these absurdi""'. ties as follows: The last step in the splitting of the fresh-water mussels of Europe was doru' by the malacozoologists Bourguignat and Locard. According to the shape and: the outline of the shell, they split up the few well-known species into countles new ones. Locard lists from France alone no less than 251 species of Anodonta j

87

IDENTIFICATION AND TAXONOMIC DISCRI1v1INA'l'ION

'l'AXONOMIC PJWCEDURE

phenomenon is most commonly expressed in size differences but may involve other morphological or physiological characters. Ebeling (1938) studied variation in the scale insect Lecanium corni Bouche, grown on different hosts (Fig. 11). Those from Prunus had large 15 14 13 12 c

;g

I

I

0.

8. 8 0 7 ~ 6

\

'\ I

-s ·9

l\

I

I

I

10

I II

\

I/

\

'

)

Ii

I\.

)

!'... L_I/

0 5

I\

.

~/

i;. 4

\

I/

~ 3

0

,;t--.. I

I

I

"'~ 2 a. I

['\

I/

I/

\

.v

1'. ~

_i....-

190

200

210

220

230

240

250

260

270

280

290 300

310

320

Length of antennae in microns

A

.....

330

-

340

A Fw. I 2A. Bimodal curve resulting when individuals of Lecanium corni from two host species, apricot (Prunus) and Christmasberry (Photinia), are combined into one frequency distribution for antenna! length (Ebeling, 1938).

B Frc. 11. Difference in structure of the scales Lecanium corni Bouche on different ho species: A, on apricot (Prunus); B, on alder (Alnus) (X 2). (Ebeling, 1938.)

On the other hand, two mussels were given the same name, if they had the sam~ outline of the shell, even though one may have come from Spain and the othefi· from Brittany. It seems incredible to us that it never occurred to these authorsi to collect a large series at one locality, to examine the specimens, to compare all, the individuals and to record the intermediates between all these forms. It isi equally incomprehensible that these people did not see the correlation between:, environment and shape of shell, even though they spent their entire lives in; collerting mussels.

l

All these "speries" of ilnodonta are now considered to be habitat forms. of two species, and the other names have been sunk into the synonymy•. of the two valid species. Whether a given habitat form is an ecophenotype or a microgeographic race is not always evident. It is sometimes necessary to transplant it or to raise it in the laboratory in order to solve this question. MuchIVork of this sort still remains to be done. 2. HOST-DETERMINEDVARIA'rION. Host-determined variations in para~· sites of plants and animals provide a source of taxonomic error and permit <'onfusion with microgeographical races or with sympatric species. This

15 14 13 c 12 211 0 :;a. 10 8. 9

E

,,...._ I/

\

'

J

J

I

8

.Ec

0

I

190

1..... •

200

..... 210.

'

J

I\

~

"7

/,

,_ 1.....-

!\ \

'~

/

2 I

I

.

i\

r1

~ 4 &' 3

\

\

'

''/

6 5

--

J

\

~ 7

~

II'

220

230

240

250

....... "' 260

"·· ..... 270

280 290

I\ [\... 300

310

320

_

......

330 340

Length of cntennae in microns

B Fw. 12B. Slightly overlapping "normal curves" resulting when individuals of Lecanium corni from apricot (left) and Christmasberry (right) are placed in their respective frequency distributions for antennal length (Ebeling, 1938).

bodies and short appendages; those from Photinia, small bodies and long appendages. When individuals from the two hosts were combined in one frequency distribution for antennal length, a bimodal curve rel:lulted; when plotted separately, two "normal curves" were produced (Fig. 12). Transfer of adults from Prnr111s to Photinia provided offspring

88

TAXONOMIC PROCEDURE

IDl!JNTJFJCATION AND 'l'AXONOMJC DJSClllilllNA'l'lON

of the Photinia type. A third typ~ on Alnu8 when transferred to Prun gave progeny of the Prunus type. . Gerould (1921) has reported that the braconid wasp, Apanteles fiav( conchae Riley, spins white cocoons when reared from blue-green cater pillars of Golias philodice Godart, golden cocoons when reared fro yellow-green caterpillars from the same species. With regard to anothe parasitoid, Salt (1941) found that males of the parasitic wasps, Trich . gramma semblidis Aurivillius, tend to be wingless and otherwise modifi when they develop in the eggs of the alderfly, Sialis lutaria (Fabriciu • (Megaloptera), but not when reared from lepidopterous hosts.

89

phase; and when isolated and reared separately, into the solitary phase. Similar phases have been reported by Faure (1943a, 1943b) in two species of armyworms (Lepidoptera), Laphygma exigua (Hilbner) and L. exempta (Walker). 4. CLIMATICALLY INDUCED (SEASONAL) GENERATIONS. Many species of short-lived invertebrates, particularly insects, produce several generations in the course of the seasons of a single year. In such species it is uot uncommon that the individuals which hatch in the cool spring are quite different from those produced in the summer, or that the dry-season individuals are different (e.g., paler) from the wet-season population.

Transiens ( dissocions)

Solitorio

Gregorio

Fw. 14. Locusta migratoria Linnaeus. Pronotum of \;? in dorsal view of A, phase danica (solitary), and B, phase migratoria (gregarious) (Uvarov, 1921).

Tronsiens (congregons)

Frn. 13. Diagram illustrating the phases of locusts from solitary through transiti to the gregarious phase (hnms, 1937). .

Another kind of case is cited by H. S. Smith (1942). the encyrtid wasp, H abrolepis rouxi Compere, readily parasitizes r sca:e on citrus but is unable to do so when the red scale is reared on Cyc .' This apparent immunization by the plant host might well confuse inte pretations which utilize parasites as a taxonomic index. 3. DENSITY-DEPENDENT VARIATION. The effects of crowding are som times reflected in morphological variation. This is not uncommon wher crowding produces a shortage of food materials. However, density' dependent variation need not be related to food supply. Uvarov (192 et seq.) has shown that gregarious species of locusts exist in three unstabl. biological phases, solitary, gregarious, and transitional (Fig. 13). The phases differ in anatomy (Fig. 14), color, and behavior characteristi and have often been described as distinct species. When newly hatch nymphs are reared under crowded conditions, they mostly develop into the gregarious phase; under Jess crowded conditions, into the transitiona

Such seasonal forms can usually be recognized not only by the occurrence of intermediates in the intervening season, but also through identity of wing venation, genitalia, etc. Cyclomorphosis. A special kind of seasonal variation is found in certain fresh-water organisms, particularly rotifers and cladocerans. The populations of a species undergo quite regular morphological change8 through the seasons, in connection with changes in the temperature, turbulence, and other properties of the water (Coker, 1939). Many "species," particularly in the genus Daphnia, have been named that are nothing but seasonal variants. A study of the causes of cyclomorphosis has been begun (Brooks, 1946). 5. HETEROGONIC VARIATION. Heterogony (allometry) may result in the disproportionate size development of some structure in relation to the re~t of the body. If individuals of a population show allometric growth, ammals of different size will show heterogonic variability. This is particularly well marked among insects. It involves such features as the heads of ants (Fig. 15), the mandibles of stag beetles (Lucanidae), the fro~tal horns and thoraces in scarabs, antenna! segments of thrips, etc. Failure to recognize the nature of such variations has resulted in much synonymy. .The exact causes of much of this variation are unknown. In species with continuous growth it is actually a form of age variation (see Al

!)(}

IDENTIFICA TION AND TAXONOMIC DISCIUMIN A'l'lON

TL1XONOMIC PROCEDURE

above). Some of it has a genetic basis and is properly classified under, intrinsic variation (see IIB below). In holometabol ic insects, however, where this phenomenon is particularly common, it is closely correlated \\·ith size, and this in turn is thought to be the result of variation in food supply which causes the larva or nymph to metamorpho se at different. growth stages. 6. NEUROGENIC OR NEUROHUMO RAL VARIATION. Neurogenic or neuro humoral variation is color change in individual animals in response to th environmen t. Such changes are accomplishe d through the concentratio · or dispersal of color-bearin g bodies known as chromatophores. This typ .

Fm. 1.5. Hetcrogony ns a cause of variability. Neuters of Pheidole instabilis, showin ' increase in the relative size of the head with absolute size of the body (after Wheeler,. 1\HO).

··

of variation was first thoroughly studied in the chameleon (Brucke, 1852)) It occurs sporadically in the lower animals but is best developed among)

the cephalopods , crustaceans, and cold-blooded vertebrates (cyclostomes, 1 elasmobranc hs, teleost fishes, amphibians, reptiles). Space will not per-: mit a discussion of this specialized type of variation. For details the/ reader is referred to the recent review and bibliography by Parker (1948). i D. 1'raumatic Variation. Traumatic variation occurs with varying' frequency in different groups of animals. The abnormal nature of this type of variation is usually obvious, but in some cases it is subtle and : may be misleading. 1. PARASITE-IN DUCED VARIATION. Aside from such familiar effects of parasitism as swelling, distortion, and mechanical injury, parasites may produce conspicuous structural modification s. In the bee genus Andrena, for instance, parasitism by Stylops frequently results in reduction in the size of the head, enlargement of the abdomen, and changes in punctura-

91

tion, pubescence, and wing venation. It commonly results also in intersexes. Since Andrena is markedly sexually dimorphic, these intersexes have been a source of taxonomic confusion and synonymy. However, in one case (Linsley, 1937), a stylopized intersex proved of value in associating the sexes of a bee which had been described as two different species. Salt (1927) has made the most comprehens ive study of the morphological effects of stylopizatio n in Andrena. In females he found reduction of the pollen-collec ting organs, loss of anal fimbriae, changes in relative length of antenna! segments, reduction of facial foveae, reduction of the sting and accessory organs, paling of ventral abdominal pubescence, acquisition of angular cheeks, and yellow on the normally dark clypeus. In males he reports the developmen t of long hairs resembling the female f!oceuli, broadening of the posterior basitarsus, changes in proportions of antennal segments, loss of cheek angles and some yellow from the clypeus, indications of facial foveae, and reduction in size of genitalia. Holmgren (1913) described some strikingly different termite soldiers from the Orient and assigned them to a new genus and species, Gnathotrrmes aurivillii Holmgren. Kemner (1925) showed that these modified l:loldiers were nothing but parasitized individuals from colonies of .Macrotermes malaccensis (Haviland). 2. ACCIDENTAL AND TERATOLOGI CAL VARIATION. Accidental variation is usually externally induced, although it may work internally through some developmen tal or hormonal system. The external stimuli may be mechanical, physical, or chemical. Such variation is extremely diverse aud in mo::-;t animals may be readily identified, because the individuali:i involved either deviate so markedly from type as to be recognized as freaks, or because the injuries or abnormalitie s involved are asymmetrica l. However, in those forms which undergo metamorpho sis, injuries to an earlier stage may produce later abnormalitie s which are not so easily recognized as such. This is especially true when the anomalies involve eharacters which are normally of taxonomic value in the group concerned. For instance, certain types of pupal injury in beetles may produce symmetrical abnormalitie s in punctation, surface sculpturing, or segmental iou of appendages; in butterflies, symmetrica l modification of wing patterns. In most cases, however, even with such subtle differences, the abnormal nature of the variation may be detected by the specialist without much difficulty. Teratologica l variation has been elaborately studied and classified by Cappe de Baillon (1927) and Balazuc (1948). The student interested in Pursuing this subject further is referred to these works for details and for further references. II. Intrinsic (Inherited) Individual Variation. In all the cases of Yariation mentioned in the preceding section, the same individual is

92

TAXONOMIC PROCEDURE

IDEN1'1FLCA1'10N AND TAXONOMIC DISCRIMINA'l'ION

actually or potentially subject to a change in appearance. In addition to this noninherited variation, there is much intrapop.ulation variation that is due primarily to differences in genetic constitution. This genet- ; ically based individual variation can-somewhat arbitrarily-be divided into two classes. A. Sex-associated Variation. Among genetically determined varia- · tions within a population, there are many which are sex-associated in, that they may be sex-limited (express themselves in one sex only) orl otherwise associated with one or the other sex, or which involve sex characters or modes of reproduction. Some of these are as follows: 1. PRIMARY SEX DIFFERENCES. These are differences involving the primary sex organs utilized in reproduction (gonads, genitalia, etc.).\ Where the two sexes are otherwise quite similar, primary sex differences will rarely provide a basis for taxonomic confusion. 2. SECONDARY SEX DIFFERENCES. There is more or less pronounced' sexual dimorphism in most groups of animals. The differences between. male and female are often very striking, as for instance, in the birds of paradise, hummingbirds, and ducks. In many cases the different sexes' were originally described as different species and retl}ined this status until: painstaking work by naturalists established their true relationship. A celebrated case is that of the king parrot [Larius (Eclectus) roratus Muller]' of the Papuan region, in which the male is green with an orange bill,· the female red and blue with a black bill. The two sexes were considered different species for nearly one hundred years (1776-1873) until naturalists proved conclusively that they belonged together. Striking sexual dimorphism is particularly frequent in the Hymenoptera. The males of the African ant Dorylus are so unlike other ants . that they were not recognized as such and were for a long time considered · to belong to a different family. In the tiphiid wasps (Tiphiidae), the . small wingless female and the large winged male are so different that some taxonomists use a different nomenclature for the two sexes. Whole "genera" consist entirely of males, others of females. The best way of · determining with which female of "genus" B a given male in "genus" A belongs is to find a pair in copula or to watch a female in the field and catch the males as they are attracted to her. Once it has been established that B is the female of A, it is usually possible to associate many "species pairs" in the same genus by utilizing additional information on distribution, frequency, color characters, etc. 3. ALTERNATING GENERATIONS. In many insects there is an alternation of generations that is very confusing to the taxonomist. In the ' genus Cynips (gall wasps), the agamic generation is so different from the bisexual one that it is quite customary to apply different scientific names 1

93

to the two (Kinsey, 1930). In the aphids (plant lice) the parthenogenetic wingless females are usually different from the winged females of the sexual generations (Fig. 16). 4. GYNANDROMORPHS. Gynandromorphs are individuals that show mnJe characters in one part of the body and female characters in another part. Thus the two halves of the body may be of opposite sexes, or the division may be transverse, or the sex characters may be scattered in a mosaic. In the latter case symmetrical variants may be produced. Usually gynandromorphs may be recognized as such, and therefore they rarely provide a source of taxonomic confusion. Gynandromorphism is produced by an unequal distribution of chromosomes, particularly the ~f'x chromosomes . .). INTERSEXES. Unlike gynandromorphs, intersexes are likely to exhibit a blending of male and female characters. They are generally thought to result from an upset in the balance between male-tendency and female-tendency genes. This upset may result from irregularities in fertilization or mitosis or from physiological disturbances associated with parasitism (see 1 under ID). Intersexes are particularly apt to appear in populations of interspecific or intersubspecific hybrids. Inter~exes have been studied in greatest detail in Lymantria (Goldschmidt, l 933) but are well known in many other animals. B. Non-sex-associated Individual Variation. This term is simply one of convenience applied to intrapopulation variation which is not sexlimited or does not primarily involve sex characters. l. CONTINUOUS VARIATION. The most common type of individual variation is that which is due to the slight genetic differences which exist between individuals. No two individuals (except monozygotic twins) are exactly alike, genetically or morphologically, in a population of sexually reproducing animals. One of the outstanding contributions of population genetics has been the establishment of this fact. The differences are in general slight and are often not appreciable unless special techniques are employed. The study of this variation is one of the foremost tasks of the taxonomist. It is now evident that no one individual is "typical" of the characters of a population. Only the statistics of the whole population can give a true picture of the population. This explains why it is necessary to procure adequate samples of each population. The collection and evaluation of the statistics of populations will be discussed in more detail in Chap. 7. The variability of different characters of the same population is often very different. Likewise there are different degrees of variability among related species. Just why one species should be variable and another

94

95

IDEN'l'IF'ICATION AND TAXONOMIC DISCRIMINATION

TAXONOMIC PROCEDURE

. ~

~

i'.

~

~

~

g

e.

"

]

~

~~

..j

l !l,

·~

~ !l,

E

j

"'8

.i

e .,.,!l,

"' .~

t

~

~a

g

r

~

+



;

.g

E

~

~

j ., .,.,l ,·~ ~ e. ~ e -s .ii ~ -'"1 *" ~ ~ -~ 1E g'"3

·;

"'

t-s ]

~

0

!.ll

xx."

0

{iei:

:2~

.., IX

vl

2~ ~

n

~ I

:Or

a.

-~

*

X)t-

] -s~

(J

!l,

[ E

c

~ ·~

~

:~ ;

~

§

lg

-to

~

~

)@r

i

.,c ~

~

1

I

.:i

g

" "'8

j

~

!l,

:~

ll .~

'

1t !l,

~

"+ ~

~-;;-

8 =a ~ !'.!

.,.,

!l,

~ 3

~

e.

]

~

~

0

c

~

~

_g

1

~ ] 0

E

+

~

,;:

] .,,0c

! j" ":;- l 0

E

]

~

~

.;'

~

~

83

E

·~0 ~



~

~ ~

]

[

1

~

" .,,~

:~

3 ·~

1

~

!l,

·~

~

-5

+-

g~

~-g

c

,; .g

~

'§~·

i

]'

ol

" llE ., ~" " i·i j iJ *11lb -! ~~ ~ ., B e ~ 'Ee ] " -~ ~ * ~ .§ " X)+J. l4)+- ()l- "Q » :Qi'O ~

*"

'O

~

~~

,

~

'O

0

~

I

0

·~

X)+-

;:

o•

:::

~

:::

:2

~ ;

l

"o ~

! it T

~ ~~))!<

~ ~~:o;.

3 Frn. I6A. Periphyllus californiensis (Shinji). 1, Fu!ldatrix or stemmother; 2, normal· apterous parthenogenetic viviparous female; 3, alateOf same; 4, smallest spring alat ~ viviparous female; and .5, smallest spring apterous female (see data on facing page).

hi

96

TAXONOMIC PROCEDURE

one not is not always clear. A taxonomist who has adequate materi of one species should not hastily decide that this will permit him to certain of the variability of related species. Early taxonomists vastly underrated individual variation in man genera of animals. The species of the snail genus Melania (fresh an brackish water) have been described largely on the basis of shell chara ' ters, such as the presence or absence of spikes and of diagonal and spir ribs. However, spined and spineless specimens occur in the species scabra, M. rudis, and M. costaba, sculptured and smooth specimens ·· M. granifera, and so forth. In a revision of this genus no less than 1 "species" were found to be nothing but individual variants and had to added to the synonymy of other species (Riech, 1937). 2. DISCONTINUOUS VARIATION (GENETIC POLYMORPHISM). ences between individuals of a population are, in general, slight an' intergrading. In certain species, however, the members of a populati , can be grouped into very definite classes, determined by the presence certain conspicuous characters. Such discontinuous individual variati is called polymorphism. Frequently such polymorphism is controlL by a single gene transmitted by simple Mendelian inheritance. Polymorphism is more pronounced in some groups of animals than others. The spotting in lady beetles (Coccinellidae) is a well-kno example of genetic polymorphism, as is industrial melanism in mot . Polymorphism has great biological importance, since it proves selecti' differences between apparently neutral characters. For a more detai · discussion of polymorphism, see Ford (1940, 1945), Mayr (1942), a' Mayr and Stresemann (1950). The practical importance of polym. phism to the taxonomist is that it has led to the description of ma~ so-called "species" that are nothing but polymorphic variants. ·: ornithology alone about 100 species names were given to polymorp The establishment of their true nature has led to a considerable simplific tion of taxonomy. Genetically different seasonal forms. In exceptional cases of rapidl reproducing forms it happens that selection is so strong that the summ · generation is genetically different from the spring and fall generatio This has been shown for Drosophila (Dobzhansky, 1951) and for sever polymorphic species (Adalia bipunctata Linnaeus, and the hamster, Cri r:etus cricetus Linnaeus). These changes can only be demonstrated b' special techniques and are not likely to confuse the taxonomist. 3. SEX-LIMITED POLYMORPHISM. Perhaps the most spectacular case of polymorphism are to be seen in the Lepidoptera and more particular in certain species of butterflies. The common alfalfa butterfly, Coli eurytheme (Boisduval), for example, has two strikingly different fema forms, one resembling the male in ground color and the other, var. alb~

IDENTIFICATION AND TAXONOMIC DISCJUMJNA'PION

Strecker, being largely white. The most complicated cases of sex-limited polymorphism which have been studied genetically are the examples of mimetic polymorphism in African swallowtail butterflies of the genus Papilio. Quite apart from the fact that allopatric populations throughont Africa show distinct subspecific differences which are correlated with differences in the species of butterflies which they mimic, we find that ~everal distinct female forms exist within a single population. Thus in \Vest Africa one finds, in the same population of Papilio dardanus

c Fw. 17. Mimetic polymorphism in the Papilio dardanus complex. A, male of ceneaalso basic type of nonmimetic female, ground color yellow; B, dionysus, nonmimetic female, ground color of forewings white, hind wings yellowish; C, trophonissa, mimetic female, ground color of forewings white, hind wings brownish; D, hipocoon, mimetic fe'llale, ground color white (redrawn from Eltringham, 1910, by Goldschmidt, 1945).

Brown, one male form and five female forms, three of the latter mimicking different models which belong to the families Danaidae and Nymphalidae (Table 4 and Fig. 17). The most remarkable feature of this polymorphism is that, although the various forms are so distinct as to resemble representatives of three different families of Lepidoptera, breeding experiments have shown (Goldschmidt, 1945) · .. that this type of heredity is nothing but ordinary Mendelian heredity under the control of the respective sex and its specific physiology of development. . . .

!)8

TA

\(()J\'()J[

JDENTIFICA'l.'ION AND 'l'AXONOMIC DISCRIMINATION

re l'JWCEJ>U HE

[In a related case, Pap,ilio polytes Linnaeus,] one type of female (cyrns Fabricius resembles the male, another, polytes, mimics P. aristolochiac Fabricius, and a third' romulus Cramer, mimics P. hector Linnaeus. The breeding results were explaine satisfactorily by a dominant factor A which converts cyrus into polytes, and one B which only in the presence of A converts polytes into romulus" [Table 5].

Another celebrated case is that of Pseudacraea eurytus Linnaeus (Carpen( ter, 1949). TABLE

MIMETIC PoLY:VCORPHIS~i IN \VEST AFRICAN

4.

Papilio dardanns

(From Goldschmidt, 1945) Male -

Nonmimetic females -

---

Typical dardanns

,_ _ _ _ _ _ _ _

---~

Mimetic females ---·

---~-----

Models

--

Basic type ':;! similar hippocoon Fabricius , A maitris niavins Lin to ci' · I nae us dionysus Doubleday I trophon-issa Aurivil- Danaus chrysippus and Hewitson lius Linnaeus niobe Anrivillius Bematistes tellns I , Aurivillius I

TABLIC

5.

Papilio polytes (From Goldschmidt, 1945)

GicNOTYPES OF FoRMS OF

I

I

I '\!ale-like Female I Mimetic Female

Males All Alike cyr-us aaBB aaBb aabb AaBB AaBb Aabb AABB AABb AAbb

aaBB aaBb aabb

polytes

LINNAEUS

I ,i

Mimetic Female -----romnlns

IAaBB AaBb Aabb AABB AABb AAbb

The Recognition of Individual Variants. How can individual variant be recognized? There is no simple answer to this question. If a larg' sample of a population is available, intermediate forms between th various more extreme variants are usually found. Also, there are certain characters in every group that are less subject to individual variation than others. The genitalic armature in insects, the palpus in spiders, the radula in snails, the structure of the hinge in bivalves are sue ' characters. If several sympatric forms agree in their genitalic armature (or one of the other mentioned characters) it is very probable that they· are conspecific.

99

However, even here one has to apply a balanced judgment. Although in most genera of Lepidoptera there are characteristic differences between the genitalia of related species, there are cases known in which forms have identical genitalia, even though they are different species by every other criterion. The establishment of correlations is often very helpful. If two forms which differ in character a can be shown to differ also in the less conspicuous characters b, c, and d, it becomes very probable that they are different species. Some yearn ago Mayr (1940) found that among birds identified as the southeast Asiatic minivet (Pericrocotus brevirostris Yigors), some had the innermost secondaries all black; others had a narrow red margin on these feathers. A detailed study revealed that those birds with red on the innermost secondaries had seven additional characters: a more yellowish red of the underparts, a different distribution of black and red on the second innermost tail feather, a narrmv whitish margin along the outer web of the first primary, and four other minor characters. Slight though they were, these characters were well correlated with each other and with geographical and vertical distribution. The conclusion that two full species were involved has since been confirmed by several authors. Post-mortem Changes. The taxonomist must guard against one further type of individual variation. In many groups of animals it is impossible to prevent post-mortem changes of preserved specimens. Some extreme cases are known in birds. The deep orange-yellow plumes of the twelvewire birds of paradise (Scleilcides ignotus Forster) fade in collections to white. Skins of the Chinese jay (Kitta chinensis Boddaert), whose plumage is green in life, turn blue in collections, owing to the loss of the volatile yellow component in the pigment. Many birds that are clear gray or olive-gray when freshly collected become more and more rufous through oxidation of the black pigment ("foxing"). Many synonyms have been created in ornithology owing to the comparison of freshly r·ollected material with old museum specimens. Other post-mortem changes result from chemical reaction of preservatives or from killing agents. A common color change of this nature takes place when certain yellow insects, especially wasps, are overexposed to cyanide. The specimens turn bright red, and thus far no method hat' been found for reversing this reaction without injury to the specimens. When preserving specimens with evanescent colors (corals, marine slugs, etc.) it is essential to take full notes and preferably color photographs or water-color sketches. This will make possible an accurate description of the living animal. Different Subspecies or Not (Alternative 2 versus 4)? The discrimi,, nation grid indicates that two allopatric conspecific populations are to be

l_

100

TAXONOMIC PROCEDURE

considered as belonging to different subspecies if they are morphologicall different and as belonging to the same subspecies if they are morph logically identical. If there is a clear-cut difference between populatio or none at all, there is no difficulty in making this decision. But ther ' are many borderline cases. The term identical is not meant to be take too literally. No two natural populations are ever exactly identical, a least in sexually reproducing species. The proper question then is Ho ' ' . does a population have to be, in comparison with another one, i different , order to qualify as a subspecies? Complete unanimity on this point has not yet been reached b taxonomists. There are two schools, the Jumpers and the splitters. T extreme splitter recognizes as a subspecies every population which ca be shown by statistical tests to be different. The extreme lumpe recognizes as subspecies only populations of which every specimen c '. be identified. Very few taxonomists hold with either extreme. Th· two most common standards are that either more than 50 per cent of th specimens of the examined population must be identifiable with certainty or more than 75 per cent. These are sometimes referred to as the 50 p cent rule and the 75 per cent rule. For practical reasons we prefer th 75 per cent standard, and it will be the basis of the subsequent discussion· The expression 75 per cent nonoverlap, or 75 per cent of the specimen must be identifiable means that 75 per cent of the individuals have to · different from all other individuals of other subspecies (of the species). O the other hand, if 75 per cent of the individuals of population A different from only 75 per cent of the individuals of population B, the it is possible for only 5.05 per cent of the individuals of A to be differen from 99.865 per cent of the individuals of B. This means that only abou 5 per cent of the individuals can be identified with certainty. Clearl ·. this is not sufficient. A fuller discussion will be given in Chap. 7. Two further points must be mentioned here, although they also ar fully discussed in Chap. 7. One is that the overlap rule applies to popu lations, while the taxonomist, of course, studies only samples that ar drawn from these populations. Obviously, the larger the sample, th greater the chance for overlap. The second point is that the range o variation of a sample or of a population is not linear but two-dimensional., If the wing length of a subspecies of birds varies from 70 to 80 mm. on the basis of an adequate sample, we can predict that about two-thirds of the specimens have a wing length of 73 to 77 mm. The values near · the mean are much more frequent than those near the extremes. If the coefficient of difference (see Chap. 7) is greater than 1.5, two different subspecies are usually involved. If the coefficient is less than 1, it is not advisable to separate the two forms. If the coefficient of difference is between 1 and 1.5, it is necessary not :

lDEN'l'IFlCATION AND 'J'AXONOMIC DISCRIMINA'l'ION

101

only to use more refined methods of calculation, but also to take into consideration additional evidence. If the population in question has some distinctive biological characteristics, or numerous lesser characters in addition to the clearly diagnostic main character, it may be named even if the coefficient of difference is somewhat lower than average. Such biological characteristics occur when the population in question has a well-defined, isolated range, or when it is situated on the periphery of the species range. In contradistinction, the coefficient of difference has to be well above average if the population in question is intermediate between two other subspecies, or if it is situated in the middle of a cline. Subspecies or Allopatric Species? (Alternative 4 versus 8). The word allopatric is essentially an antonym of sympatric and means therefore geographical distribution without geographical overlap. There are five kinds of allopatry that may be encountered by the taxonomist: 1. Allopatric populations A and B are in contact and intergrade in the (usually fairly wide) zone of contact. 2. Allopatric populations A and B are in contact and interbreed completely in the (usually fairly narrow) zone of contact. 3. Allopatric populations A and B are in contact but do not interbreed freely in zone of contact. Occasional hybrids occur. -±. Allopatric populations A and B do not interbreed at all, even though meeting in a zone of contact. 5. Allopatric populations A and B are separated by a distributional gap which prevents contact. Populations that qualify under (1) and (2) are nearly always to be considered subspecies; under (3) and (4), species; and under (5), species or subspecies. The following comments, numbered to correspond to the list above, may be helpful: (1) Allopatric populations that intergrade with each other belong to the same species. It depends on the degree of difference whether or not they are to be considered subspecifically different. (2) There is no clear-cut distinction between intergradation and allopatric hybridization. In general, we speak of intergradation when a :;eries of intermediate populations is intercalated between two sub:;pecies, each with approximately the same amount of variability as any population of either subspecies. We speak of allopatric hybridization when the two subspecies meet in a well-defined zone and form there a hybrid population with greatly increased variability, often containing the entire spectrum of character combinations from subspecies a to subspecies b. There must be evidence for random interbreeding in this zone. Allopatric hybridization is :;ometime:; al:;o referred to as secondary intergradation, because it is a secondary event, following a breakdown of a previous extrinsic isolation of the population. Among North

102

TAXONOMIC PROCEDURE

American birds the flickers (Colaptes), juncos (Junco), and Canada jays (Perisoreus) furnish good examples of hybridization between widely divergent subspecies. For further details and additional examples see Mayr, 1942, pp. 263-270. (3) Allopatric forms that hybridize only occasionally in the zone of contact are full species. There are a few cases where it is difficult to decide whether the hybridization is occasional or complete. Much recent evidence indicates that hybridization has to be fairly complete in order to restore secondary intergradation. Two North American warblers, the blue-winged warbler (Vermivora pinus Linnaeus) and the golden-winged warbler (V. chrysoptera Linnaeus), meet along a front extending from the New York region to the Middle West. They regularly form a few hybrids in the zone of contact (or narrow overlap), which have been named Brewster's and Lawrence's warblers (based on different combinations of the parental character). Howeyer, the frequency of the hybrids seems not to have increased over the years, nor has the gap between the species been narrowed. More difficult to evaluate are cases where two species remain as dis~ tinct species over most of their range but form complete hybrid populations in a few areas. This happens particularly in regions in which the natural ecological balance has been badly disturbed in recent years by human interference. It is recommended that such forms be treated as full species in spite of the occasional free hybridization under the stated conditions. (4) Allopatric populations that are in contact but fail to interbreed are full species. The failure of interbreeding indicates reproductive isolation and attainment of species rank. The failure of overlapping may be due to either one of two opposite reasons. The zone of contact may connect two very different ecological areas (e.g., savanna and forest). If one of the two neighboring species is specialized for one of these habitats and the other species for the other, the two species cannot invade each other's ranges because their ecological requirements are too different. The other possible reason for nonoverlap of full species is that their ecological requirements are so similar in every respect that they compete with each other. On one side of the zone of contact one species is slightly superior, on the other side the other. A full understanding of this 8ituation is very important, because allopatry has often been taken as an automatic criterion of conspecificity. Mayr (1951) lists numerous cases of birds which have been restored from subspecies to species rank after the nature of their allopatry had been more closely investigated. (5) Allopatric populations that are separated by a distributional gap may be either species or subspecies. The most important of the species

JJJEN'l'JFIC,A'l'iON AND 'l'AXONOMJC DISCJU:\11NL1TION

1

103

criteria the presence or absence of reproductive isolation, cannot be used (except' experimentally, and even then with r~servation.s) .to determine the status of geographically isolated populat10ns. This is the reason why the classification of allopatric populations is so often subject to a considerable amount of disagreement among taxonomists. Many solutions for this dilemma have been proposed, but all of them are beset \vi th difficulties. Some taxonomists insist that all morphologically distinct, isolated populations be treated as full species "until. it is pr?ven that the~ a~e subspecies." This solution is, of course, impractical, because it is impossible in most of these cases ever to obtain clear-cut proof, one :va.y or the other. Furthermore, this solution overlooks the fact that it is just as serious an error to call a population a species if it is really only a subspecies as to call it a subspecies if it is a species. The second solution is to treat as full species all populations that are not connected by intergradation. This procedure is founded on the correct observation that populations which are connected by intergradation are conspecific, and it jumps from this observation ~o the reve_rse eonclusion that populations which are not connected by mtergradat10n are not conspecific. This conclusion is correct only so far as sympatric or contiguous populations are concerned, because with them the lac~ of intergradation proves the absence of interbreeding and thus cons.t1tutes de facto proof of specific· distinctness. It does not necessarily apply to isolated, allopatric populations. Geographical isolation is not an intrinsic isolating mechanism, and there is no guarantee that the morphological hiatus caused by the temporary stop in the gene fl.ow is proof of the evolution of isolating mechanisms. Others have proposed the opposite extreme, namely, to consider all related allopatric forms as conspecific. . A complete experimental analysis, including studies on matmg preference and a cytological examination of hybrids, is usually impossible and mav not be conclusive when it is possible. Ecological preferences are pa;t of the isolating mechanisms between species, and these cannot ~e p~·operly evaluated in the laboratory. For example, the sym~at:1.c :,;ibling species Drosophila pseudoobscura Frolova and D. ~ersimilis Dobzhansky and Epling always hybridize in laboratory populat10ns, but no hybrid has ever been found in nature. Since direct proof is unavailable, it becomes necessary to decide the status of isolated populations by inference. Several kinds of evider:ce are available. All these are based on the observation that reproductive isolation is correlated with a certain amount of morphological difference, which is fairly constant within a given taxonomic group. The taxonomist ran uRe this evidence to work out a yardstick which can be applied to

104

'l'AXONOMIC PIWCEDURE

isolated populations. There are three sets of morphological differences that can be utilized to calibrate such a scale. 1. Degree of difference between sympatric species. Within a given genus or within a group of closely related genera, there is usually a fairly well defined amount of morphological difference between valid sympatric species. This difference may be great, as in the case of birds of paradise, or it may be very slight, as in the case of sibling species. This amount of difference between good species can be used to determine the status of isolated populations in these same genera. 2. Degree of difference between intergrading subspecies within widespread species. The amount of morphological difference between the . most divergent subspecies in species of the same genus indicates how much morphological difference may evolve without acquisition of reproductive isolation. 3. Degree of difference between hybridizing populations. Subspecies or groups of subspecies within a species sometimes become temporarily separated from one another through the development of a geographical barrier but merge again after the breakdown of the barrier. Free interbreeding, which often occurs even after morphological difference of considerable magnitude has developed, proves conspecificity. Good examples of such free interbreeding of morphologically strongly differentiated populations are to be found in North American birds among some of the juncos (Junco) and flickers (Colaptes). Even after all the8e criteria have been applied, some doubtful cases : remain. It is preferable for various reasons to treat doub(ful allopatric populations as subspecies. The fact that a population has been unable to invade the range of its nearest relative implies that it has been unable so far to develop isolating mechanisms that would permit coexistence. There is no zoogeographical barrier more formidable for a subspecies than the range of another subspecies (Mayr). Furthermore, the use of trinominals conveys two important pieces of information: (1) closest relationship and (2) allopatry. Such information is very valuable, particularly in large genera. To treat su<'h allopatric forms as separate species has few practical advantages.

CHAPTER 6

TAXONOMIC CHARACTERS The essence of original taxonomic research is the analysis of material and the synthesis of the results into a classification. Although in practice these steps are often combined, they are really two separate operations. The first consists of finding and evaluating differences, the second of discovering points of resemblance. In both cases we are dealing with certain attributes of organisms which are k~own as taxonom~c characters. The present discussion of the various kmds of taxonomic characters is given as a prelude to the discussions on analysis and synthesi8 of material. Organisms differ from one another in many ways. Differe~ces ma~ be in8ignificant, as in identical twins, clones, and parthenoge~1~tic offsprmg, but more often they are extensive and numerous. Ind1v1duals of the human species differ in innumerable points, some well marked and easily described such as size and hair color, and some elusive and difficult to describe. ' Even greater is the number of differences between individuals representing two different species. Such individuals differ in an infinite number of characters and yet may retain certain features in common. A taxonomic character may be defined as any attribute of an organism or of a group of organisms by which it differs from an organism belonging to a different taxonomic category or resembles an organism bel?nging to .the 8ame category (Mayr). Taxonomic characters are thus attnbutes which permit placement of an organism in the formal classification. Taxonomic characters thus have a double function: (1) they have a diagnostic aspect as indicators of difference (emphasis on differentiating properties is particularly strong in the lower taxonomic categories); and (2) they function as indicators of relationship (this property makes them especially useful in the study of the higher categories). Differences between organisms belonging to the same taxonomic category (male vs. female, immature vs. adult form, etc.) are not taxon~mic ('haracters, Most of the differences between individual variants classified in Chap. 5 are of this sort. THE DIAGNOSTIC VALUE OF TAXONOMIC CHARACTERS

Diagnostic Characters. lf we aecept a re<·ent estimate, we might assume that a hi~hcr animal may have in the neighborhood of 10,000 105

,I !1

1,:

1

i

106

TAXONOMIC PROCEDURE

genes, while the number of characters is limited only by the patience of the investigator. Even two related species of the same genus may differ in from 400 to 600 characters. Zarapkin (1934) in a study of two races of the beetle, Carabus cancellatus Illiger, studied 166 characters, but these were only features relating to the sculpture and the proportions of the sclerotized exoskeleton. In addition to such morphological characters, there exist all sorts of physiological differences, particularly those involving thresholds and rates of growth and development, inherited ecological adaptations, and psychological reaction norms, such as differences in instinctive behavior. It would require more than a lifetime to prepare an exhaustive species description with references to all these characters. Furthermore, the results would be so voluminous that no one would be willing to publish them. However, not only is such a complete species description impractical, it is also unnecessary, since even a small fraction of the morphological differences is sufficient, in most cases, to ensure a correct diagnosis. In fact, the inclusion of physiological differences in a formal diagnosis is actually undesirable, since most identifications have to be made from dead specimens. The most practical diagnostic characters are those that relate to some easily visible structure with but slight variability. Such characters may be of no particular importance to the species, but they serve as markers for the taxonomist. To illustrate this point we might make the following comparison: If we want to direct a person to one of two houses on a street, we need not go into a detailed description of all of its features; we merely say: It is the white, not the red one! Color is a superficial attribute of a house. Actually the white house may be built of wood, the red one of stone or brick; the white one may have six rooms, the red one ten rooms and so forth. Even if the red house is painted white, it will remain basically different from the other white house. The relationship of the diagnostic characters of an animal to its other species characters is frequently of an equally superficial nature. It is necessary to emphasize the biological insignificance of many of the key characters or diagnostic characters, because this is frequently insufficiently realized by both taxonomists and non-taxonomists. For instance, if one of two related genera of insects is diagnosed as having two extra bristles on the thorax, this by no means implies that this is the basic difference between these genera. Actually it may be the least important difference, but it may also be the one which can be recognized most quickly by the taxonomist (Mayr, 1942). It has often been stated that taxonomy is an art rather than a science, and there is a half-truth in this statement. It is as true as saying that a doctor who is a good diagnostician makes his diagnosis by intuition. Actually the good doctor and the good taxonomic;t make their diagnoses

107

1

by a skillful evaluation of symptoms in the one car-;e and of taxonomic eharacters in the other. · To be truly diagnostic, a taxonomic character must be constant for all members of a given category and for no others. In the case of variable characters it must be constant for a stated percentage of the population of a given category; hence the importance of the study of variability of characters. Consequently the use of taxonomic characters involves a study of their reliability. The same character does not always indicate the same amount of taxonomic difference. For instance, depending on the group of birds involved, the presence of a crest may be a generic, specific, subspecific, age, or sex character. An intimate knowledge of the group is necessary in order to evaluate properly a particular taxonomic character. By way of further example, a prominent, much-enlarged canine tooth is an important character for distinguishing species and o-enera in some families of mammals and quite unimportant in others. J,ikewise the number of premolars, whether two or three, is important in the primates, distinguishing the catarhines and platyrhines, but in other groups of mammals the mean number may vary among individuals of the same species. Furthermore, the value of a taxonomic character may change within a single phyletic series. For instance, in one section of a genus a character may be constant and useful in separating species. Elsewhere it may break down and be subject to individual variation. This fact, however, by no means invalidates the use of the character in that section of the genus where it is constant. KINDS OF TAXONOMIC CHARACTERS

Early taxonomists used morphological characters almost exclusively to distinguish taxonomic categories and as a basis for classifications. Although morphological characters are still more useful than others, they are being supplemented to an ever increasing extent by other kinds of characters as listed and discussed below. This is not only necessary ' application of the biological species concept, but the great to permit the broadening of the use of new kinds of taxonomic characters has increased the reliability of classifications. A single character is not as reliable as a character complex. However, we must never lose sight of the fact that characters are of unequal importance. It is here that the art of the taxonomist comes in, for he has to decide what weight must be given to each character. The bases on which classifications are built may be greatly increased by using all stages in the life cycle of a species. In addition to the taxonomic characters presented by adult males, the taxonomist should also use those of adult females, of the various immature or larval stages. and of the eggs.

109

'l'AXONUMIC PJWCEDURE

'l'AXONOMIC CHARAC'l'ERS

Frequently characters of immature stages are more helpful than those of the mature animal. Thus the various entities of the Anopheles maculipennis complex may be more readily distinguished by characters of the egg than by those of the adults, and the classification of the Aleyrodidae (whiteflies) is based primarily on the pupa, which is the stage most commonly used for purposes of identification. Furthermore, phylogenetic classifications may frequently be developed more readily from larval characters than from adults in many groups of arthropods in which the adults are degenerate or subject to convergence. In groups with complete metamorphosis, evolution often proceeds independently in larvae and adults, and conclusions drawn from characters of one stage form a very useful check on conclusions drawn from those of another. The taxonomist in his practical work selects from the hundreds of taxonomic characters those that are most significant as being diagnostic or as indicating relationship. The ability to select these significant characters distinguishes the superior taxonomist. The kinds of available taxonomic characters may be somewhat arbitrarily classified under five headings: (1) morphological, (2) physiological, (3) ecological, (4) ethological, (5) geographical. Within these five classes we can distinguish additional subdivisions.

of taxonomic characters, such characters need little discussion here. They range from such superfidal features as plumage and pelage characters of birds and mammals, through linear scale counts of fish and reptiles, to the highly conservative and phylogenetically significant sutures and sclerites of the arthropod body. Animals with an external skeleton (arthropods, mollusks, etc.) present in general the greatest array and most useful range of external structural characters. Genitalic Structures. Because of the fact that reproductive isolation is a sine qua non at the species level, differences in genitalia have been employed in many groups as the last court of appeal in delimiting specie8. It has even been suggested by Dufour and others that a lock-and-key relationship exists as regards the copulatory structures of the males and females of those species with sclerotized genitalia. Such appears to be the case in certain groups of insects, e.g., the Fulgoridae. On the other hand, genitalic characters have been found to vary in the 8ame manner as other characters (Jordan, 1905). In general, it may be said that genitalic differences must be evaluated just like other characters. In groups where their significance has been proved they are usually very useful, because genitalic structures appear to be among the first to change in the course of speciation. · Internal Morphology. Anatomy provides an abuuuant source of taxonomic characters in practically all groups of higher animals. However, the extent to which such characters have been used varies greatly from group to group, generally in inverse ratio to the abundance and usefulness of the external morphological characters. In many groups of vertebrates selected portions of the internal skeleton (e.g., the skull) are routinely preserved and used in identification, but in general both the hard and soft parts of the internal anatomy of most animal groups are m;ed primarily as a source of characters for the elucidation of higher categories. Paleontologists, of course, must deal almost exclusively with hard parts, and as a result they have focused attention on many useful skeletal characters in groups of animals with an internal skeleton. Embryology. Comparative embryology offers taxonomic characters of great phylogenetic significance. Thus cleavage patterns, gastrulation, and other embryological phenomena may be characteristic for whole phyla or for series of phyla and thus assist greatly in the understanding of our highest categories. On the other hand, in such groups as insects, the total (holoblastic) cleavage of the Collembola (springtails) emphasizes the wide gap which separates this group from the other Apterygota (primitively wingless insects) and the Pterygota, in spite of the secondary reappearance of this cleavage type in a few highly specialized parasitic Hymenoptera near the top of the insect series (8ee also de Beer, 1940, 1951).

l08

KINDS OF TAXONOMIC CHARACTERS

1. Morphological characters

2.

;3.

4.

5.

a. General external morphology b. Special structures (e.g., genitalia) c. Internal morphology ( = anatomy) d. Embryology e. Karyology (and other cytological differences) Physiological characters a. Metabolic factors b. Serological, protein, and other biochemical differences c. Body secretions d. Genie sterility factors Ecological characters a. Habitats and hosts b. Food c. Seasonal variations d. Parasites e.' Host reactions Ethological characters a. Courtship and other ethological isolating mechanisms b. Other behavior patterns Geographical characters a. General biogeographical distribution patterns b. Sympatric-allopatric relationship of populations

Morphological Characters. General External Morphology. Since external morphology has traditionally provided a primary and evident source

110

'l'AXONOMIC PROCEDURE

'l'AXONOMIC CHARACTERS

Karyology. Karyological and other cytological characters may be us ful to the taxonomist, though the degree of differentiation and limits o variation in chromosomal structure must be tested in each group before: the significance of such characters can be determined. The simplest cytological character is chromosome number. This is determined by relatively simple technic involving the crushing or smearing of the test on a slide. Chromosome numbers have been recorded for thousands animals, and the results of such studies have been used as evidence phylogenetic relationship by White (1945, 1949), and others. Chromosome morphology is being used by the plant taxonomist to a ever increasing extent. Karyology seems to be equally promising i many genera and families of animals. Dobzhansky, Patterson, an, Sturtevant, as well as several other authors, have made substantiw contributions in recent years to our knowledge of chromosomal variatio . in Drosophila. Such closely related species as Drosophila pseudoobscur, and D. persimilis are diagnosed more easily by their chromosome con figuration than by any other feature. In the genus Sciara also th chromosomes have excellent diagnostic value. In a study of the Finnis bugs of the family Lygaeidae, all the genera and nearly all the 56 cyt , logically investigated species could be identified by their chromosom alone. An exhaustive summary of the field is given by White (1945) Some of these cytological differences interfere with chromosome pairin and thus serve as isolating mechanisms. Gene arrangements on chrom somes have been used to analyze populations of Drosophila, Anophel and Tendipes ( = Chironomus), and the presence of supernumera chromosomes to study populations of grasshoppers ('i'rimerotropis). Such studies are useful only if the student has a thorough knowledgd; of cytology. The number of chromosomes may be different in clos~ relatives (owing to the joining of two chromosomes after the loss of a kinetochore); genetically inert chromosome sections also are easily lost, Two species with superficially identical chromosomes may be much mor~: different genetically than others with various gross chromosomal differ..l ences. The chromosomal polymorphism in species of Drosophila, Tri-. merotropis, and others supplies excellent evidence for this. There is a, very useful recent summary of our knowledge of the chromosomes of vertebrates (Matthey, 1949). Physiological Characters. Physiological characters have been very; unevenly exploited for taxonomic purposes. Yet in constancy, diversity,·, and significance they probably far exceed morphological characters. ' They have, however, the disadvantage that in most cases their study requires living organisms. Thus the most suitable subjects for this· approach have been forms with a short life cycle, small body size, or other features which make for ease of laboratory experimentation or

observation. However, the array of physiological characters in general increases with the complexity of the organism. We can never hope for a complete comparative physiology for taxonomic purposes (any rnore than we can hope for a complete comparative morphology). Nevertheless, physiological characters are coming into greater use, not o~ly as a supplement to morphological characters, but as a means for checkmg conclusions based on other kinds of data and as an aid in the development of sound classifications. Few detailed comparisons of the physiological constants of closely related species have so far been undertaken. Growth rates and egghatching periods in mosquitoes and growth rates and temperature reactions in various species of frogs of the genus Rana differ significantly. A combination of two or three hemolysiR constants is absolutely diagnostic for each examined species of the mouse genus Peromyscus. The amount of difference between the studied species leucopus, gossypinns, truei, and eremicus corresponds approximately with the de.gree ~f morphological distinctness. The various species of Daphnia differ. m the spectroscopy of their hemoglobins. All this work confirms the view that many, if not most, proteins are species-specific. For a summary of this field see Landsteiner (1945). Metab~lic Factors. Up to the present time, the microbiologists, especially the bacteriologists, who have had little morphology to rely on, have made the greatest use of physiological characters both in the development of a classification and for purposes of identification. Thus enzymatic activity is an important taxonomic character, and both anabolic and catabolic reactions are used. Cell chemistry is important in the differential ability of certain bacteria to react to certain stains (as Gram-positive or Gram-negative). Metabolic requirements are of great importance, as, for instance whether the bacteria are aerobic or anaerobic, and how cultural growth patterns and coloration develop on standardized media. .\!though students of higher plants and animals have rarely used characters such as these for purposes of identification, nevertheless many broader physiological processes, especially differential growth rates and other developmental phenomena, have proved very helpful in the separation of closely related species. Serological Protein and Other Biochemical Differences. These have ' increasing ' attention as taxonomic tools. Serology is . conbeen receiving cerned with the nature and interactions of antigens and antibodies. Antigens are substances capable of inducing the formation of antibodies when introduced into the blood stream of other animals. Antibodies obtained from the blood sera of immunized animals are serum globulins which are produced in response to the introduction of a foreign antigen. These are the principal substances concerned in serological reactions.

111

112

TAXONOMIC CHARACTERS

'l'AXONOMIC PROCEDURE

Underlying .these reactions is the principle of quantitative specificity, i.e., a given kind of antibody will react more strongly, under comparable conditions, with the particular kind of antigen used in its formation than with any other substance. TABLE

6.

A

COMPARISON OF THE SEROLOGICAi· REACTIONS

m'

TmJ SERA

m'

COMMON:

CRUSTACEA*

Test antigens '--

Antiserum

Homologous antigen

·r .... ;g ~-:i 1l" " " . ,__ ,_".a" .. "" "' " ..... . . t> ~ "" "" "" """ ". ..""" "t "~-5g. """ "t "" "-"'" I!"' "" ;E]. ·a. "t1 "" I! I> ";; ",_ "" "" ",_ .... "" "!> ""' ~2 ~I! c' ~a ~ ~"' "° ~" ti::"' "" -- -~

~

~-~

142(1

+ 2)

140(1

+ 0)

147(1

+ 0)

+ 1) 152(1 + 0) 148(1 + 1) 151(1 + 1) 154(1

160(1

+ 1)

161(1

+ 1)

+ O) 149(1 + 0) 163(1 + 2) 150(1 + 1) 164(1 + 1) 162(1

Homarus americanus Callinectes sapidus Callinectes sapidus Callinectes sapidw~

-

L3: 100 L4

maenas Cancer borealis Cancer borealis Ganeer borealis Cancet

26

I ·..

44

HC38-1 I ...

100

17

3b

...

16 141 17 I

7

34 100

22

8

...

100

58

100

59

41

3d

'100

29

19

39-1

41

33

paourus

3!!-2

55 \ 51

.ll,fenippe mercenaria 1lfenippe

36-A

mercenaria

3

I

Cancer

Gery on quinquedens Gery on quinquedens

I

36-A, .

6

12

28

16

%'-1: 39-2

THE Ih~LATIONSHIP OF THE SERA OF Dn'FERE1'i'l' 8P1Dc1Et5, GENEHA, A1'D

(Summarized from Table 6) I. The relationships of the sera of species of the same genus A. Cancer 1. Cancer borealis vs. C. pagurus 41, 41, 19, 55, av. 39 58, 59, 29, av. 49 2. Cancer borealis vs. C. irroratus 33, 51, av. 42 3. C. pagnrus vs. C. irroratus B. Homarus l. H. americamts vs. H. vulgaris 54 .~4 Grand average.. 46 I I. The relationship of the sera of genera of the same famil)· A. Callinectes vs. Carcinns 26, 44, 17, ::\4, a\'. :~o lll. The relationship of the sera of different families of Brachyura A. Portunidae vs. Cancridac 16, 14, 22, 13, 17, a1·. 16 B. Portunidae vs. Xanthidae 4, 6, 8, 7, 6 (28), average of 5 values 6 C. Portunidae vs. Goneplacidae 3, 3, av. 3 D. Portunidae vs. Maiidae 1, 1, av. 1 E. Cancridae vs. Xanthidae 12, 16, av. 14 F. Cancridae vs. Goneplacidae 24, 4, av. 14 G. Xanthidae vs. Goneplacidae 10, 25, 24, 6, av. 16 H. Xanthidae vs. Ocypodidae 6 l. Xanthidae vs. Maiidae 5

I

aj

7.

FAMILIES OF BRACHYURA

HCl

pagurus

TABLE

54

100

Carcinus

1. The antigenic composition of animals is an important part of their essential natures and must be considered in any sound natural system of classification. 2. Protein antigens are conservative hereditary traits. 3. Good precipitin techniques are well adapted to reveal the relative degrees of biochemical similarity of protein antigens.

;:..O

-

100

371A

He lists the basic premises

Numerous examples of the application of serologic methodH to taxonomy are now available, one of the moHt thorough of recent date being Boyden (1943) on Crustacea. Boyden summarizes the data from all his <'rustacean comparisons in Tables 6 and 7.

.t

;

refine the techniques of systematic serology. of systematic serology as follows:

113

i

!.

I

100

i

10 I

: 100

I

251

... i

19 3

::1

6

2J 100 I

i

6 100

I * The homologous area represents 100 per cent, and the heterologous per cent values indicate t ratio of heterologo11s area to the homologous area. ~'.'

The precipitin reaction was discovered by Kraus in 1897 and has been widely used in the taxonomy of microorganisms. It consists in the forma.-tion of a visible precipitate at the interface when one brings together an antigen and the corresponding antisernm. The precipitin test was first applied to broad taxonomic problems by Nuttall (1901), who showed that the relative intensities of precipitin reactions paralleled the systematiOJ positions of the species whose antigens were tested. Boyden (1943 et seq.) has done much to elaborate the concepts and

The data in Tables [6] and [7] give a quantitative approximation to the serological relationship of the species tested and they appear generally to agree with their systematic positions. The data may be presented phylogenetically as in the case of Geryon, representing the family Goneplacidae, in relation to the families Xanthidae, Cancridae and Portunidae. According to Rathbun . . . the family Goneplacidae is most closely allied to the family Xanthidae and our data confirm this conclusion but show in addition that the Goneplacidae are almost as close to the Cancridae as to the Xanthidae . . . . The position of Geryon and the Goneplacidae in relation to the other families as indicated by the present data is ~hown in Fig. [18]. Actually three dimensions would be needed to express this relationship properly, but it can be done on a plane surface as shown here.

114

TAXONOMIJ PROCEDURE

Still another use of serological tests is for identification. Brooke and. Proske (1946) used the precipitin test for determining insect predators of immature mosquitoes. They tested the stomach contents of Belostoma, Ranatra fusca Palisot-Beauvois, a hydrophilid larva, a damselfly nymph and a dytiscid larva and concluded that "it is possible to demonstrate, by precipitin tests, the presence of mosquito larvae and pupae in th ' c digestive tracts of aquatic insec · 1 I predators." 1 Body Secretions. Another kind o to/ . physiological character which ha ~ \\ o>,.. w/ · '. I ' proved useful in taxonomy involve' / \ 6 body secretions which form consisten. • ~}> ....-.-, ................. • bi / / patterns. The waxy secretions o. ', ~,,." scale insects and mealybugs fall iii ' , ,' _,/ 94 / The wax pattern!li P this category. x Fm. 18. A diagram to show the rela- produced reflect, of course, the under..;, , tive distances of four families of Crustacea from one another as indi- lying morphology, especially the num'.' ber, size, and arrangement of the wax· cated by serological tests. The data are tentative, inasmuch as the families glands, and therefore tend to be about;· have not yet had really adequate testas constant as the morphology of:, ing. The smaller the figure, the closer structures. these the assumed relationship. The famiSterility Factors. Even thei Genie lies concerned are Cancridae (C), Portunidae (P), Xanthidae (X), and earliest taxonomists knew that ther~ Goneplacidae (G). The latter family, is much sterility between members of represented by Geryon, is apparently different species. In fact, sterility: closest to Xanthidae, represented by has often been cited as the specie NI enippe; a little less close to the It is now known that criterion. Cancridae; and considerably more distant from the Portunidae. Three sterility is only one of the many exist.:, dimensions would be required to ing isolating mechanisms, and that it; express these relationships properly, is gradually built up. In some case and the plane figure is really a projecdistinct species may be completel~7 tion of such three-dimensional disinterfertile (as for instance Ana, tance onto a plane surface (Boyden, platyrhynchos Linnaeus and A. acu . 1943). Linnaeus among birds), while in othe ~ ca1:1es there may be a considerable amount of sterility even betweeri, subspecies of a single species (as, for instance, between the 1:1ubspecies of', Drosophila pallidipennis Dobzhansky and Pavan). Sterility is thus a. taxonomic character that must be used with discrimination. (For treatment of the genetic basis of sterility see Dobzhansky, 1951.) Ecological Characters. Through the work of field naturalists an < experimental ecologists during recent decades, it has been well established' that each species of animal has its own range of tolerance of habitat, food,,

115

'l'AXONOMIC CHARACTERS

breeding season, and other ecological factors. No two species with identical ecological requirements can coexist in the same place (Gause's rule) (Lack, 1949). Likewise it has been shown fo~ genera and the still hi.gher categories that each occupies a separate adaptive plateau (Sewall Wright, In view of :
8.

Species

Anopheles maculipennis (After Hackett and Missiroli, 1935, and Bates, 1940)

ECOLOGICAL CHARACTERS IN 'l'HE

Habitat

Water type

COMPLEX

I Hibernation I Mah1.ria carrier

------- -----

Fresh water -\ melanoon . ...... Rice fields Fresh water messeae ........ . Cool standing ! water rnaculipennis. . . Cool running water Fresh water Brackish atroparvus. . . . . . Cool waters Brackish labranchiae. . . . . Mostly warm waters .sacharo••i . ..... . Shallow standing , Often brackish water

I

No Yes

I No AlmoRt never

Yes No No

No Slightly Very dangerous

No

IVery dangerous

1

Nearly every issue of the ecological journals describes cases of pronounced ecological differences between closely related and morphologically very similar species. As a matter of fact, a species description which .! 11

'I

,I

TAXONOMIC PROdl!JDUHE

TAXONOivJJC CHARACTERS

does not give some account of the ecology, where such data are availabl · should be considered incomplete. Specific food preferences are important taxonomic characters in mon phagous or parasitic organisms. Numerous new species of insects hav been discovered when those that appeared to live on the "wrong" fo plant were more closely studied. At one time such bark beetle gener as Dendroctonus, Ips, and Phloeosinus were thought to contain but a fe · species, each highly variable in structure, host, and boring patter · When these were segregated according to host plants, it was discover that many species were involved, each with relatively constant characte of structure, host preference, and boring pattern. The host-parasite relationship can be worked both ways by the taxon mist. A knowledge of the parasites can be used to determine ferences within the host group (host discrimination), and a knowled of the hosts can be used to discover differences in the parasites (parasi discrimination). Host Discrimination. The method of using differences in parasites ' commensals to distinguish between exceedingly similar sibling species • very useful. This method has been used advantageously by botanis For example, Pinus jeffreyi Murray is now recognized as distinct fro .r P. ponderosa Lawson; but entomologists were never in doubt about thi .· because each species of pine is attacked by a different bark beetl . Dendroctonus jeffreyi Hopkins on Jeffrey pine and D. brevicomis LeCon on Ponderosa pine. Emerson (1935) found morphological differen between two highly similar species of termites only after he had receiv a clue as to their distinctness by differences in the termitophile beetl ' found in their respective nests. Very similar species of fresh-wat . turbellarians may differ in the ciliates that live on their surface. Parasites may also be used in the classification of the higher categorie · The use of host-parasite relations for the elucidation of taxonomic pro lems has been desc·ribed in detail by Metcalf (1929). Kellogg (189' 1913) utilized this method extensively in his study of the Mallophag: For more rerent discussions of this subject see Clay (1949) and Hopkin ;, (1949). sho. birds, of group isolated rather a (Phoenicopteri), The flamingos anatomieal similarities to both the storks and the geese. The bird lie'. (Mallophaga) of the flamingos are clearly related to those of the geese thus (with the proper safeguards) indicating closer relationship to th geese than to the storks. Parasite Discrimination. W~ have already mentioned the case of bar beetle genera, the species of which were not properly discriminated un sorted out according to host. The tapeworms Hymenolepis nana in man and H. fraterna in rodents do not differ significantly in morphology, no .

do Ascaria of man and of pigs, although host specificity indicates the existence of differences in both cases. Many cases of this kind are recorded in the parasitological literature. On the other hand occurren~ of a parasite on a different host does not prove specific disti~ctness. Ecological characters are also useful at the infraspecific level. Nearly every geographical, and often microgeographical, race differs in its ecological requirements from other races. In botany such local ecological races have been called ecotypes (Turesson, 1922) and have been studied particularly by Clausen, Keck, and Hiesey (1948). These ecotypes are discussed in Chap. 2. In birds much geographical variation in ecological requirements has been described (Mayr, 1951). It affects particularly vertical distribut.ion, habitat preference, food preference, and nest-site preference. The continental race of the European cormorant (Phalacrocorax cormoranus sincnsis Shaw) nests on trees, the North Atlantic race (P. cormoranus cormoranus Linnaeus) nests on rocks. Instances of such geographical variation of ecological characters have also been described for other groups of animals, although the tendency is for the ecology of a species to remain fairly constant within its entire geographical range. It is this property which usually permits the diagnosis of species on the basis of ecological ('haracters. A special case of infraspecific variation of ecological characters is given by the host races of insects, often formerly referred to as "biological races" (Thorpe, 1930, 1940). The term biological race is not descriptive and has been applied to many different phenomena, particularly to sibling ~pecies. The term host race is more precise. Host races are common in insects and have attained various levels of 1 cfo tit1ctness. Thus the codling moth, Carpocapsa pomonella (Linnaeus), has developed a distinct walnut-preferring race. This is inferred in the al1~ence of morphological differences from the fact that "walnuts, pome 1nuts, and the codling moth have existed together in California since 1873, and although the moth has been a severe pest on pome fruits for many decades, only since 1918 has it become a major pest of walnuts" (Smith, 1941). ::Vlany other host races without distinctive morphological characters have been recorded. Perhaps the best known examples in forest entomology are the races of Dendroctonus monticolac Hopkins and other woodbori~g beetles upon which Hopkins based his "host-selection principle" (Craighead, 1921). Another case concerns the small ermine moths ~lyponomeuta '?adella Linnaeus, which, when transferred from apple t~ a':thorn or VIce versa, develop a significant preference for the plant on which they were reared as caterpillars and proceed to oviposit on the plant of their choice as adults (Thorpe, 1930).

l Ul

117

ii

~ II;

I

118

T11XONOJJJIC PROCEDURE

These examples from the field of economic entomology illustrate t importance of the ecologic approach in taxonomic research. The t ctnomist of the present and future is working in a highly specialized an interrelated world. New methods must be adopted to meet changin conditions. The greatest pressure will come from applied fields such plant quarantine, biological control, and insect vectors of plant an, animal diseases, and it is here that the ecologic approach may be expect to yield the best results. Another type of ecological character, especially useful at higher lev of classification, is the presence and kind of intracellular symbion · According to Steinhaus (1949), Buchner (1940) describes the situatio in the Homoptera as follows: All the aleyrodids, for instance, have the same type of symbiosis and the sa . manner of symbiote transmission, in which a number of intact mycetocytes a carried over in the ovum. A similar uniformity exists in the psyllids. In t superfamily Aphidoidea, on the other hand, similarity of type is limited families. Thus Aphidae and Eriosomatidae ( = Pemphigidae) have round . symbiotes, Adelgidae ( = Chermesidae) have rod-shaped symbiotes, and Ph loxeridae are apparently free of intracellular micro-organisms. In the cocci • there is no uniformity of type except in subfamilies. All the Lecaniinae ha " similar yeast-like symbiotes in the hemolymph and in the fat cells; the orthezii contain bacteria in the fat bodies; the diaspids harbor degenerate round bacteroids. The mode of generation-to-generation transmission in these ca is also specific for the subfamily. Among the monophlebines all the genera ha. paired, elongate rnycetomes, although M archalina appears to be an exceptio .· In this genus the symbiotes are carried in greatly enlarged cells in the gut epith ium. This discrepancy is clarified if one accepts the rearrangement presen by Morrison in 1928, which removes Marchalina from the monophlebines an places it as a tribe in the new subfamily Coelostomidiinae. One wonders if man similar changes would not be made if the taxonomist had the advantage of kno ing the symbiotic arrangement of the insects with which he worked.

Finally, as a further example of ecological characters, we may menti . host reactions. Where host reactions are specific and conspicuous, th · have frequently been used for taxonomic purposes. Symptoms, althou largely being replaced by other characters in microorganisms, still a important in the classifieation of plant viruses. Some of the most usef plant reactions, however, are the galls produced in response to attack b various gall wasps (Cynipidae) and gall flies (Cecidomyiidae or Itonidi. dae). Some of these are more readily distinguished than the adul insects and in some cases new species have been described from the g . alone. ' Although this practice is to be frowned on, nevertheless it ' • indicative of the importance that various workers have attached to such characters.

TAXONOMIC CHARAC'l'ERS

119

Ethological Characters. Just as morphological characteristics change from species to species and from genus to genus, thus supplying material for a taxonomic analysis, so behavior patterns change from group to group. It may be too early to speak of a science of comparative ethology, but beginnings have been made in the analysis of the unit elements of which behavior patterns of some animal groups are composed and the comparison of their evolutionary modification from species to species. This was done by Lorenz (1941) for most species of river ducks (Anatini), by Spieth (1947) for the species of the Drosophila willistoni group, and by .Jacobs (1950) for grasshoppers. It has been found that the behavior pattern is on the whole composed of homologous elements within a given taxonomic group, but that there is great variety in the manifestations of these elements, and that many of the modifications are species-specific. Courtship and Other Isolating Mechanisms. Differences in ma ting habits are especially important behavior characters, since they are more likely to result in reproductive isolation and consequent speciation. For example (Mayr, 1942), The slugs are a group of animals which, although morphologically very similar, tend to have color phases and varieties, most of which had originally been described as good species. No two taxonomists could agree as to which of these forms were good species and which were not. Iii a study of the pairing behavior of these slugs, Gerhardt . . . showed that the displays that led up to copulation :uc exceedingly complicated and consequently highly specific. In the genus Limax six definite pairing types could be determined, which apparently correspond to six good species. Many of the other described 'species' of this genus :tre probably nothing but color varieties.

Other Behavior Patterns. In addition to behavior patterns which serve as isolating mechanisms, an infinite variety of behavior characters is available to the taxonomist. For example, the nature of the webbing constructed by various spiders, mites, and caterpillars may be used at various levels in the classification. The two bee genera Anthidium and Dianthidium were slow to be recognized on morphological grounds, yet all known species of the former construct their nests of cottony plant fibers, those of the latter from resinous plant exudations and sand or small ~~~.

'

The use of extraneous materials in the construction of nests or larval or pupal cases provides characters at various levels in the classifieation of caddisworms and bagworms, and the manner in which such materials are attached to the shell is a useful taxonomic character for distinguishing species of the molluscan genus Xenophora. Finally, certain behavior patterns are more conservative than stmctural (•haracters. Examples of such are the drinking habits of pigeons and

l

120

TAXONOMIC PROCEDURE

sandgrouse (Pterocletidae); dust bathing of Ploceidae and Passer (bu: not of finches); use of mud in nest building of barn swallows (Hirund and crag martins [formerly erroneously associated with bank swallo (Riparia) ]. Geographical Characters. Geographical characters are among the mos useful of tools for clarifying confused taxonomic pictures and for testin · taxonomic hypotheses. Most sound classifications show some carrel tion with geographic or associated ecologic features. Essentially t taxonomist is interested in two kinds of geographical characters, ( ' general biogeographic patterns, which are especially useful in the arrang ment and interpretation of higher categories, and (2) the allopatri ' sympatric relationship, which is most helpful in determining whether not two populations are conspecific. General Biogeographic Patterns. The broad geographic patterns wi · which we are concerned have been determined by the study of distrib' tional patterns of large numbers of groups of plants and animals. Bi · geographers have divided the world into various realms, regions, pro inces, subprovinces, etc., based upon generalized comparisons of fau and floras. These are not rigidly defined, but in general they represen. distributional centers which exist today or have existed in the pas · Depending on the group, they may be expanding or retreating, and thus find it more useful to refer to them as faunas or floras or biotas rath than zones or areas. A taxonomjst should have an understanding of t · geological history of the regions in which such biotas center, as well as· knowledge of the past relationships of the faunas and floras concern Armed with this information, the interpretation of various higher ca gories can be made on a much sounder basis. For instance, the mammals of South America are either not related those of Africa or, if of common ancestry, have presumably reached Sout America by way of North America. The hystricomorph rodents, seem ingly close to the African p~rcupines, appeared to be an exception, th history of which was inexplicable in view of the absence of such forms · • the early Tertiary period of North America. A reexamination of thes porcupines and their relatives, prompted by this zoogeographical puzzl revealed, indeed, that the porcupines of South America and of Africa a' of independent origin (Wood, 1950). Distributional difficulties hav shed light on taxonomic relationships in many other instances. The suggest, for instance, that the New Zealand "thrushes" (Turnagra) a not thrushes but Pachycephalinae, and the New Zealand "tits" no Paridae but Malurinae, both reassignments leading to a considerabl t.uogeographical simplification. Distribution is, thus, an import.ant to in taxo11un1ie analysis. Sympatric-Allopatric Relationship of Populations.

TAXONOMIC CHJ1llACTERS

121

must useful methods of approaching the question of whether or :b.ot two populations represent distinct species (se_e Cha?. 5). If a series of for~s shows geographic replacement, e.g., a cham or nng of forms, each of which differs from its neighbors, the forms are said to be allopatric. Such a distributional pattern in closely related forms is now generally considered to be indicative of a polytypic species consisting of several subspecies. On the other hand, if the ranges of two or more forms partly or entirely overlap and these forms do not intergrade, they are said to be sympatric. Such a distributional pattern is considered to indicate that the forms involved are full species, since sympatric coexistence without interbreeding is one of the basic tenets of the species concept. THE EVALUATION OF TAXONOMIC CHARACTERS

The determination of the characters that distinguish closely related rategories or are shared by taxonomic groups (higher categories) is only one step in taxonomic research. The evaluation and analysis of these characters must go further. For instance, it must be determined how t'onstant a given character is. Most taxonomic characters are variable, trnd a study of this variability is part of the taxonomic procedure. It is ohvious that taxonomic characters should not be drawn from single representatives of populations, but rather from adequate samples, as described in Chap. 7. The relation between taxonomic characters and taxonomic categories it> not always clearly understood. E~en though a species has specific characters and a genus generic characters, these characters have no absolute values. The taxonomic categories are not a consequence of the characters, but rather the opposite-the taxonomic characters are a consequence of the categories (Chap. 3). As stated above, the value of a given character may change from category to category. The experiences of the taxonomist in this respect may be summarized in a few simple rules. 1. The degree of difference between characters is often an indication of the degree of relationship, at least within a given taxonomic group. However, the degree of morphological difference may indicate different categorical rank in different groups. In the genus Drosophila, for instance, many good species (sibling species) are hardly different morphologically, while the species of birds of paradise are always strikingly different, and even subspecies may differ conspicuously. The rates of phyletic evolution of the phenotype and of speciation are only very loosely correlated. 2. A character may be of great taxonomic significance in one case and of none in another. For instance, the number of tail feathers in birds, whether eight, ten, or twelve, is sometimes a generic character, sometimes a subspecific character, and sometimes varies individually within a single population.

122

TAXONOMIC PROCEDURE

3. Reduced or degenerating characters are particularly unreliabl Such characters are, for instance, the presence or absence of a fourth t . in certain genera of birds, the number of teeth in certain genera of mam. mals (as, for instance, Talpa), the presence or absence of tarsal spurs o wing veins in certain insects or of wings in pterygotan groups, etc. Clas ifications should not be based on characters that are in the process of bei lost within a taxonomic category. 4. So-called "primitive" characters are often an indication of t absence of specializations. Pseudoprimitiveness may be acquired secon arily in phyletic lines by a loss of specializations. Contrary to a wide held opinion, specializations may indeed be lost in the course of evolutio resulting in the return to what appears to be a more primitive conditioq THE BIOLOGICAL MEANING OF TAXONOMIC CHARACTERS

Up to this point we have discussed taxonomic characters primaril from the practical point of view as indicators of taxonomic difference or relationship. Such a concentration on the practical aspect of taxonomi characters is one-sided and has been responsible for many taxonomi: difficulties. It will help our understanding of taxonomic characters to point ou their biological significance. It was formerly held by many biologist t~at the majority of taxonomic characters were without biological sig'.' mficance. The modern trend is to assume that no character can becom 1 established in a population unless.it has superior selective qualities. Th: adaptive significance of many taxonomic characters is obvious such " those that have to do with food getting or protection against ene~ies an
A satisfactory system or classification must be based on properly , evaluated taxonomic cha:facters. The more characters two animals have, in common, the closer we generally group them in the system. The higher~li systematic categories are formed by uniting lower categories that share . certain characters. We have discussed the philosophical basis of the .

7'AXONOMIC CHARACTERS

123

principles of classification in Chap. 3, and we shall deal here only with some of the practical difficulties with which we are confronted in the evaluation of taxonomic characters (see also Rensch, 1934). Linnaeus and most of his followers for nearly a century classified birds by purely adaptational characters. Birds with webbed feet were put into one category; birds with a hooked bill were considered another group; etc. Eventually it was realized that characters that are adaptations to a specific mode of living are not only subject to rapid changes by selective forces, but may also be acquired in different unrelated lines. Such characters have only limited value in establishing taxonomic categories. They are most useful in separating species and genera. When dealing with the classification of higher categories we must search for characters that tend to remain stable, characters that are phylogenetically conservative. Without entering into the controversy on homology, it should be pointed out that classifications are based on homologous characters. TAXONOMIC CHARACTERS AND EVOLUTION

The use of taxonomic characters in classification is based on the simple fact that some characters change very rapidly in evolution, while others only change slowly. The rapidly changing characters are used to distinguish subspecies and species; the slowly changing ones are used to l'haracterize the higher categories. If we look at a group of fossil animals, we find that they usually start with primitive forms and eventually die out with extremely specialized forms. Dollo's rule of irreversible evolution was established on this observation. As Simpson and others have pointed out, evolutionary lines do not move undeviatingly toward specialization. In fact, a character can be lost again in a phyletic line, and a similar or equivalent character can be reacquired. Specialization and despecialization often alternate in evolution. Also, each taxonomic l'haracter may evolve to a large extent independently of other characters. For this reason it is often misleading to consider the mere number of differences between two categories as indicating degree of difference. Too often several characters are partially or completely correlated. For instance, the arboreal mode of living in a group of mammals will inevitably lead to changes in the locomotor apparatus that may affect nearly every bone and muscle in the whole body. A change of feeding habits in birds may result eventually in structural modifications of the bill, the tongue, the palate, the jaw muscles, the stomach, and perhaps other features. All these characters are a single adaptive complex and should not be treated and considered as a series of independent characters. A shift into a new adaptive zone may lead to a comparatively rapid structural reorganization in order to acquire the needed specializations. Such specializations should not be overrated when making classifications.

l_

124

'l'AXONOMIC PROCEDURE

In conclusion, a taxonomic character is any attribute by which a organism or group of organisms resembles, or differs from, another; t ·• attribute may involve any comparative feature of the dead or livin organism. Taxonomic characters which are conservative (i.e., whic. evolve slowly) are most useful in the recognition of higher categori those which change most rapidly, of the lower categories; taxonomi characters are subject to parallelism, especially those involving loss reduction, and such characters should be avoided or used only with t greatest of care; character complexes which vary as a unit should treated as a unit and weighted as though they were a single character· the same characters vary in value and constancy from group to group an even within a single phyletic series, but this fact does not invalidate the'· use in those parts of the series where they are constant; taxonomic chara ·. ters provide our most useful tool for the recognition of taxono · · categories and thus ultimately for the interpretation of the course evolution; the proper evaluation of taxonomic characters is thus on of the most important, as well as one of the most difficult, tasks of th . taxonomist.

CHAPTER 7

QUANTITATIVE METHODS OF ANALYSIS THE IMPORTANCE OF QUANTITATIVE METHODS IN TAXONOMY

The use of quantitative data in taxonomy is important for several reasons. First of all, they add to the preciseness of a description. The actual measurements of a series of specimens are infinitely more useful than the meaningless statement, "of medium size." "Sex comb with seven teeth" is more precise than merely "sex comb present." Such precision is important, since related species and subspecies often differ not by the presence or absence of a structure, but rather by its size, or proportions, or number. Such relative differences can be stated precisely by the use of quantitative data (figures). This has been recognized by La.xonomists from the earliest times. Even in the days of Linnaeus it was the custom of many authors to record the total length of the type and similar quantitative data. The second reason for the importance of quantitative dataisthatspecies and the other taxonomic categories are not fixed "types" but consist of variable populations. Such variability cannot be described adequately except in quantitative terms. This is particularly true when the characters of two variable species or subspecies overlap. A third and essential reason for using quantitative data is the possibility of deriving from them (with the help of statistics) estimates of the characteristics of the natural populations from which the samples were drawn. Descriptive methods have now become standardized (see Chap. 9) to Ruch an extent that any description is incomplete which does not include at least a minimum of quantitative data. The systematic presentation of quantitative data and their evaluation is the function of statistics. A detailed presentation of the principles of statistics and of the application of the various statistical methods is beyond the scope of this manual of taxonomy. Fortunately, there are a number of good texts available, among which we recommend especially Simpson and Roe (1939), Quantitative Zoology. A short survey of those statistical methods that are particularly useful to the taxonomist is given by Cazier and Bacon (1949). The textbooks of Snedecor (1946) and ~father (1947) present more detailed treatments, with emphasis on the analysis of variance. 125

12()

TAXONOlYI IC PlWCEDUUE

Of particular interest to all taxonomists is the series of papers by; Klauber on the application of statistics to the taxonomy of reptiles (e.g.,i Klauber, 1936-1940, 1941, 1943a, 1943b, 1945), as well as Burma's (1948, 1949) discussion of the application of statistical methods in inverte-. brate paleontology. Our discussion will concentrate on some of the principles of elementary' statistics and their application to taxonomy. Some taxonomists have expressed their reluctance ''to become involved· in statistics." Actually, they are using statistics not only when giving; means and size ranges, but also when expressing the results of comparingi entities. When we state that species minuta differs from species grandis: in its smaller size, we are, in effect, stating that the mean length of minuta'· is less than that of grandis, and that the range of variation in an adequate sample of the population of species minuta does not overlap the range of. variation in an adequate sample of species grandis. If we say "averages: smaller," we imply that there is a difference between the means but a · overlap of the population ranges. The statistics recommended in modern taxonomic research are merel an extension of the simple statistics employed unconsciously by ever', taxonomist. More elaborate methods are not only made possible by th great increase in the size of the available collections (samples) but hav become actually necessary in order to settle many problems of taxonomi research at the infraspecific level. Statistics are employed most frequently in the taxonomy of contem~ porary species with respect to two kinds of problems: 1. The study of the consistency of expression of a taxonomic charactel!i' within a population (variability). In these problems one attempts t~ answer the question, How constant or variable is a given character?, The answer to this question is found by calculating standard deviatio and coefficient of variability (see below). Knowledge of the variabilit " of a character is indispensable to the solution of the second type of, problem. 2. The study of the degree of difference between two populations in regard to one or many characters. The methods of analysis of differences between populations are described in the second half of this chapter. Some additional problems may occur in paleontology (Simpson, 1941; ~'. Burma, 1948). For a more extensive discussion of the meaning of statistics in taxonomic work, see the last section of this chapter. SAMPLES AND SAMPLING METHODS

The taxonomist attempts to study the properties of natural populations. : However, an entire population cannot be brought into the laboratory or · studied in the field. The specimens actually available to the taxonomist

QUANTITATIVE METHODS OF ANALYSIS

127

are usually only a small fraction of the individuals of which the population in nature is composed. Such a portion of the true population is called a sample. From the study of such a relatively small sample, we attempt to reconstruct the properties of the population from which it was drawn. We also seek to determine how different other samples drawn from the 8ame population might be. What qualities should a sample have in order to be a sound basis for taxonomic work? "The ideal representative of a population is a sample that is homogeneous, adequate and unbiased" (Simpson and Roe, 1939; see this work for a detailed discussion of these three properties of samples). The taxonomist is only rarely in the position to obtain collections that are perfect in respect to these three qualifications. This is inevitable and not too serious, but the worker should be aware of the possible shortcomings of his material. Homogeneity. A large heterogeneous sample can often be segregated into several smaller homogeneous samples by separating the specimens according to age, sex, and locality. Season and habitat differences are additional factors which often introduce heterogeneity. If, for example, in a species of birds, samples from various localities are compared, con8picuous differences may exist when adult males are compared with adult males but may not be apparent when females and immatures are grouped in a single sample with the adult males. Homogeneity may occasionally be ignored in a qualitative analysis, as, for instance, in the selection of a diagnostic character that is equally valid for males and females, immature forms and adults. Great care must be taken when segregating a homogeneous sample to avoid bias (see below). Sometimes, as, for instance, in a growth series, heterogeneity cannot be eliminated. In other cases, e.g., in fossil material, sufficient information may not be available to segregate the material according to sex or age. Indeed, it i8 sometimes the very object of the statistical analysis to facilitate such a tiegregation of the material into homogeneous components. A number of techniques have been described to achieve thi8, such as analyses of bimodal curves, plotting of regression lines of tentative segregates, etc. Adequacy. The question of adequacy will have a different answer accotding to whether we are dealing with a variable character (such as :-;iie) which is present in all members of the population or a polymorphic d1aracter which has a given frequency within a population. Different Htatistical techniques must be employed for the two kinds of characters. It was believed at one time that statistical analysis was possible only with large samples. It is now known that valuable information can also be obtained from small samples, in fact, even from single specimens (Simpson and Roe, 1939). Small size of samples is no excuse for failing to treat them Rtati:-;tically. On the other hand, it i>i axiomatic that the

(

TAXONOMIC PROCEDURE

128

QU1lNTl1'ATIVE METHODS Off' 11N,,1LYSJS

larger the sample, the more precise the inference that can be made from it (in regard to the characters of the population). With polymorphic characters, which are present only in part of the. population, the question is, How large must a sample be to include a specimen with a character that occurs in the population at a known frequency? Or, reciprocally, between what limits can the frequency of a given character fluctuate in a sample of a given size? We refer to Simpson and Roe (1939) for a discussion of the sampling limits and reproduce here a tabl01 (Table 9). In a sample of 30 specimens, for example, a character that occurs in 60 per cent of the population can be expected to be found in at' least 10 but not more than 26 specimens. If a character occurs in 20' per cent of the population, a sample of 50 specimens is necessary to be· virtually certain that it will include one with this character. Most likely: it will be represented in a sample of 5 specimens. Cazier and Bacon (1949) state that in taxonomic statistics, "for all practical purposes: samples of at least 1.5 to 25 specimens may be used with good results, but samples of 50 to 100 specimens are more desirable." TABL~J

\cl.

SAMPLING LIMI'l'S

(From Simpson and Roe, 193!)) Size of sample

-~N-5 10 15 20 25 30 40 50 75 100

Percentage of oer·urrcnce in population

I 103__:03__ o 0 0 o o i 0 i 0 0 0 1

3 o 4 0 5 0 6 o 1 o 8 0 10 0 12 I 1 16 I 4 19 I s

4 6 8

---.-

~-I---1--~,

303

o 0 0 g o 11 I o 13 I 1 16 I 3 19 5 26 10 32 16

5 8 10 12 is 18 21 25 35 44

I~~ -~~_j

o 5 o 51 0 9 0 10 i 0 12 1 14 i 15 : 3 11 3 18 ! 5 20 4 20 6 23 6 25 10 30 g 31 14 36 17 43 24 51 25 55 35 65

603 I 703 o l 3 5 1 10 14 19 32 ·*5

I

I

.

803 ! 90%

.•

51o511 10 2 10 4 15 5 15 7 19 I 8 20 10 23 : io 25 14 26 13 29 17 34 19 37 24 41 I 25 45 31 58 I 4;0 65 49 15 156 84 68 II

5 10 15 20 25 30 40 49 71 92

2 6 10 14 18 22 30 38 60 81

5 10 15 20 25 30 40 50 75 99 i

Bias. The sample should be unbiased, that is, the method of getting, the sample should be such that the variations of the pertinent characters occur in the sample at the same frequency as in the population. A sample is unbiased when every member of the population has an equal chance of : being drawn. In order to approach this goal the specimens should be collected completely at random. In taxonomic work this ideal is rarely achieved. Collections are usually made at particular seasons and times of the day. Most of the early locality records of wild species of western X orth American Drosophila show them to be from national parks. A

120

certain amount of bias is unavoidable, but it should be recognized an
Only quantitative data can be subjected to a statistical analysis. In fact lies the importance of charaeters that can be counted or meas11 n•d. Meristic ( = countable) characters permit greater aceuracy than measurements and are therefore favored wherever possible as, for instance, by students of echinoderms, fishes, and/ reptiles. Simpson and Roe (1939) give nine criteria of good numerical observations. Most important for measurements is that they he standardized (applying to a specified distance) and accurate. For instance, the length of the bill in birds may be measured in several ways: (1) from the nostril to the tip, (2) from the beginning of the feathering to the tip, or (3) from the beginning of the hony forehead to the tip. Observations have shown that the first can be measured very accurately but does not give the full length of the bill; the third can be measured fairly accurately in all birds with a steep forehead; a11d the second can rarely be measured with any accuracy. Consequently, in some genera of bird:;; the third is the preferred measurement, in others the first. In this, as well as in all similar cases, the record should show which of several possible measurements was actually taken. It is only rarely possible to predict which of a set of possible measurements will be most important in the comparison of several samples. It is therefore advisable to measure all variates that may possibly be of importance. Subsequent analysis will show that many of these measurements either fail to show significant differences or are merely duplications of other data. In vievv of the high costs of printing, such superfluous dat'l should not be published. They may be placed in the archive:;; of a public institution (museum or library) where they are available to other ;;tudents. When one is measuring an important lot of specimens or measuring specimens before one's method has been completely standardized, it is strongly advisable to measure each variate repeatedly. The duplicate sets of measurements should be taken on different days and on new record sheets. When completed, the various sets of measurements should be compared and averaged. Particularly deviating measurements should be checked for possible errors in the measuring technique. 1his

130

131

TAXONOllI IC PROCEDURE

QUAN'l'l'l'ATIVE METHODS OF ANALYSIS

Measurable Characteristics. Total length is usually a very important measurement, particularly when it is used as the yardstick for ratios and proportional measurements. In each case it should be specified what is '. meant by total length. Is it taken before or after preservation? Does it;. include or exclude appendages on the head and the tail? Total length is most satisfactory in beetles and other rather rigid, hard-shelled animals. ' In birds the wing length ( = actually the length of the longest primary) is. a much less variable quantity than total length measured in the flesh. The cube root of the weight may under certain conditions replace the total length in calculations of allometric ratios. Body length, i.e., total, length minus tail length, is usually a more accurate measure of size than· total length. Different measurements are used for nearly every category of animal. .· In mammals, for instance, body and tail length are measured, as well as · length of the hind foot and ear and the various dimensions of the skull. ' In birds, wing, tail, bill, and tarsus are the most commonly measured '. variates. In most groups of insects not only length should be given but·, also width and antenna! and tarsal formulas. These data should be given .. as a routine matter regardless of their immediate diagnostic value. Spe- · cial measurements are traditionally given in particular taxonomic groups,.· such as the length of the rostrum in Hemiptera, length of the wings in·'· some Diptera, etc. It is important for comparative purposes to give;\ measurements that conform with the system which is customary in the'_ group under study. Technical Aspects of Measuring. Zoological measurements are now universally given in terms of the metric system. However, many descriptions written in the nineteenth century use inches and lines (1 line = 71 2 in.) (Table 10).

jected outline, and the various parts can then be measured, enlarged on a table. This method is particularly useful when relative sizes and angles are to be measured. As far as refinement is concerned, it is important to carry measurements out to whatever decimal point may be necessary, but not to waste

TABLE

10.

A, M. N, H. Dept, oOlrds

~i'Mculata

.,...

Cllllogotl~

CoU«tor

71ines = 8 lines= 9 lines = 10 lines = 11 lines = 12 lines =

14.81% 16.93!-'3 19. 05 21.167,3 23.28~'3

25 .4

mm. mm. mm. mm. mm. mm.

....

locaffty

........

Sex

Will

Toi

Ill

I

i

I

! 14

!

\

l !oel1

ca. Dec.231 11,Bengal,Darjeeling 1936 District, Tiger Hill 17,000 1 ~ fresh

80.5 58

2 Koel1

Dec.23 1 1936

80

56

13.S

59

14

.





Free~

~

A.M.N.H. ~84801

llov.13, II. Bengal 1903 Darjeeling District

1,000•.

Slight~ l:r 110l"l 80

A.11.11.H,

Dec.1, 1938

11. Buraa, Pyepat Ridge

s,ooo•

~

Fresh

77.5

56

13.5

~

Fresh

79

57

13.5

Slight6,500 1· ~ 17 won 76

58

24.5

~07283 ernav

.

1.11.N.H. ~~~84

Dec.21, 1938

A.M.11.H. 3072!!9

Nov.12 1 H. Burma, nr. Hpo.wte 1939

A.M.N.H.

Nov.28 1 N. Burma, Hpimaw 1939

v---

i~~~



llov.24 N. Burma, Htawgaw 1939

A.II.II.Ho

i~~~

Dec.15 Szechuan, China lu-kua-chi 1893

1,11.11.H. .307794

I

I

~



77

55

13

'?



78

59

14

?8

55,5 14

I '? IFresh

!

' ;

i

I

CONVERSION OF LINES IN'l'O MILLIMETERS

l line = 2.117,3 mm. 2 lines = 4.23!-'3 mm. 3 lines = 6. 35 mm. 4 lines = 8.46.% mm. 5 lines= 10.58~'3 mm. 6 lines = 12. 7 mm.

--

India, Buma, China

'!

I

!

1

:

i I

.,

I

I

I

I

I

FIG. 19. Filled-out data sheet as used in the American Museum of Natural History.

Various measuring tools are used for different groups of animals. A millimeter rule (often with a "zero stop") and dividers (calipers) are used for most larger animals. The eyepiece micrometer is used to measure microscopic objects. It may be divided into small or large units and may be arranged as a linear scale or in squares. The individual units must be ' translated into the metric system by calibration with a stage micrometer. Projection devices, such as microscopic projectors, are sometimes useful. By means of such devices, the specimens can be drawn from a pro-

effort by an unreasonable accuracy. It would be useless to give the height of a person as 176.583 cm. How, then, shall the proper degree of refinement be decided? Simpson and Roe (1939) recommended as the unit of measurement one-twentieth of the difference between the largest and the smallest specimen, if an adequate series is available. Thus if the measurements range from 10 to 12 mm., one should measure to 71 0 mm.; if they range from 40 to 50

132

133

TAXONOMIC PROCEDURE

QUAN1'1TATIVE METHODS OF ANALYSIS

mm., to Y2 mm. If they range from 70 to 90 mm., no decimal places n be recorded. If fractions are rounded up, they should consistently b · rounded to the nearest full number, halves to the nearest even number. When fractions are measured, a bias in favor of integral numbers should be avoided. Recording of Measurements. It is advisable whenever large number of meai:mrements are taken to enter them on special data sheets. adequate samples are available, each sample should be recorded on separate sheet. Each specimen should be entered separately, it museum number, age, and sex recorded, and then the various measur ments recorded in separate columns. If there is room, the calculate ratios between measurements can be entered on the same sheets (Fig. 19),

variation of a normal curve is that they depend on a great number of genetic factors, which have either a positive or a negative effect on the character. Literally hundreds of genes, for instance, tend to increase body size, a similar number to decrease it. Many individuals of a population will have an approximately equal number of size-increasing and size-decreasing factors. Many fewer will have largely plus or largely minus factors. As a result, the population as a whole will show a variation pattern corresponding to the normal curve. The normal curve is taken as the distributional pattern of the total population from which a given sample is drawn. By means of this curve one can determine the chances that a particular observation or measurement will fall within a given range of variation or the chances that a given character will appear in samples of 10 or 100 specimens. The statistics which determine the position, height, and spread of the normal curve are

THE ELEMENTARY STATISTICS

Mean. The moHt commonly used statistic is the average, or arithmetic\ mean (lvl). It is calculated by dividing the sum of the measurements (x1 + X2 + x~ + x1 + · · · ) by the total number of specimens (N). The mean thus obtained is the mean of the sample, not of the total population.· Range. By determining the smallest and the largest specimens of a' sample, we obtain the obseri ed sample range, e.g., wing 72 to 83 mm. It. is evident that with an increase in the size of the sample, very soon speci-1 mens will be found that are smaller (70, 71) or larger (84, 85). The' larger the sample, the larger will be the range between the smallest and the largest specimens. This dependence of the observed range on the size of. the sample is one of the reasons why range is usually not considered a very' useful statistic unless the size of the sample is stated exactly. Normal Curve. By arranging numerically all the measurements within a sample, one finds that not all measurements are equally frequent.· . . Most frequent are specimens close to the arithmetic mean (M) of the. sample, while specimens near the minimum and maximum are rare.;' If a sufficiently large number of measurements is plotted, it is found: that the resulting frequency curve usually corresponds to the so-called' "normal curve." The theory and properties of the normal curve are: described in every textbook of statistics. This curve is based on the laws of probability when the chances of an event occurring or not occurring; are equal. For example, when a coin is tossed once, there is a fifty-fifty'. chance that a head will turn up. When a coin is tossed 10 times, a total of 5 heads and 5 tails will occur most frequently; other combinations, such as 6 heads and 4 tails, 7 heads and 3 tails, or 8 heads and 2 tails, ' with decreasing frequency; and a series of 10 beads or 10 tails will occur only rarely. The normal curve is a graphic representation (plotting) of',· an infinite number of such trials. The reason why most biological characters seem to show the pattern of

34.13% 34.13%

1

-I

mean

+1

+2

+35.D

Frn. 20. Areas of the normal curve (after Cazier and Bacon, 1949).

the mean, which determines the center of the curve; the frequency, which determines its height; and the standard deviation, which shows how rapidly the curve falls off on each side of the mid-point. Figure 20 gives an illustration of the normal curve. It can be seen that the curve is convex near the mean and becomes concave "less than onethird " 1 of the distance away from the mean. The point where the curve changes from convex to concave is one standard deviation (see below) distant from the mean. This convex center portion of the curve includes 08.27 per cent of the area under the curve. In other words, 68.27 per cent of the individuals of the population have values within this range. It is thus evident that the great majority of values occur near the mean, and that increasingly fewer measurements are found the farther away from the mean one moves. We shall return to a discussion of the concave 1 The expression less than one-third is mathematically meaningless, because in theory the normal curve does not meet the base line (zero) until infinity. However, the term is used here because in practice, in biological populations, certain upper and lower extremes do not occur. There is no bee as large as an elephant or as small as a bacterium, no matter how many billions of bees are examined. Even though a finite linear range of a population does not exist mathematically, the part of the curve beyond three and one-half or four standard deviations is of negligible practical importanC'e in work with natural populations.

134

TAXONOMIC PROCEDURE

QUANTITATIVE METHODS OF ANALYSIS

"tails" of the curve in connection with the treatment of the overlap of· two distribution curves. Non-normal Curves. The curves describing biological populations are almost never completely normal. The most frequent d
kinds of animals more directly comparable, it is advisable to calculate the coefficient of variability (C.V.). C.V. = (S.D. X 100)/M, in other words, the S.D. as percentage of the mean. To use the abbreviation V instead of C.V. is undesirable, since V is frequently used for variance. Certain minor objections against this statistic have been raised by biometricians, but no better measure to compare variabilities has been proposed so far. What is a small C.V.? The numerical value of the C.V. depends on the measured character and on the particular taxonomic group. There are different coefficients of variability for meristic quantities, linear measurements, and ratios. The number of eyes (a meristic quantity) in the human species has a C.V. that is virtually zero; the height of the human body (even in a sample as homogeneous as a local population of adult males) ha5 a C.V. exceeding 4. The C.V. is often a sensitive indicator of the homogeneity of samples. If, for instance, the C.V. of a certain statistic fluctuates around 2.2 in ,.;amples of a series of populations, but is 4.5 in one sample, such a sample :-;hould be reinvestigated. It may include an additional sibling species, wrongly sexed specimens, or some other alien component. Zones of secondary intergradation are often characterized by an increased C.V. Variability is the tendency of individuals of a population to differ from one another. It finds its numerical expression in the coefficient of variability. The calculation of C.V. is particularly useful when comparable samples of the same species from different localities are investigated, or· when the variability of different variates is compared. Linear Measurements. Absolute size is extremely variable in most animals that continue to grow throughout life, such as fishes, snakes, and ,;nails, not to mention such forms as corals and Bryozoa. It is, however, fairly constant for certain measurements of adult mammals and even more so in adult birds. A series of 49 adult males and 29 adult females of the kingfisher, Halcyon chloris pealei Finsch and Hartlaub, from Tutuila Island, Samoa, had the measurements and coefficients of variability shown in Table 11, below. In carefully measured homogeneous samples of adult birds, the C.V. of wing length is usually between 1 and 2.5, rarely above 3. In mammals, the C.V. for linear dimensions is usually between 4 and 10, occasionally between 3 and 4. In insects which reach the imago stage through molt or metamorphosis, a small C.V. of linear measurements might be expected, since there is no further growth after the sclerotic exoskeleton has hardened. However, it is actually rather large, since the final size of the imago depends a great deal on the feeding conditions of the larvae or nymphs.

S.D.

=

~ '\}N

The square of the S.D. is called the variance. The calculation of the S.D .•. is not very laborious (Simpson and Roe, 1939; Cazier and Bacon, 1949;~~ or any textbook on statistics). In small samples (less than 15 specimens};-: the sum of squares is divided by (N-1) instead of by N. Some statis-~ ticians prefer this for samples of all sizes. If the exact value is not . required, but only a rough approximation, an estimated S.D. can be read~. from a table (Simpson, 1941) if one knows N, M, and the observed range, This is one additional reason why the size of the sample (N) should always be given. The S.D. here defined is the sample S.D., which constitutes an estimate of the corresponding S.D. of the population. Knowledge of the S.D. of the population permits predictions as to the range because

M M M

± 1 S.D. includes 68.27 per cent of the population ± 2 S.D. includes 95.45 per cent of the population ± 3 S.D. includes 99.73 per cent of the population

For instance, if the mean of the sample is 70 mm. and its S.D. is 2 mm., less than 5 individuals among 100 of the population may be expected to have values outside the range 66 to 74 mm. (M ± 2 S.D.). Coefficient of Variability. The numerical value of the S.D. is closely correlated with the value of the mean. An S.D. of 2 indicates extremely low variability if the mean is 120, but very high variability if the mean is 8. In order to make the variability of different characteristics in different

135

136

QUANTITATIVE METHODS OF ANALYSIS

'l'AXONOM IC PJWCEDURE

The proportions of body parts are usually much more stable than the linear measurements, and ratios (see below) are therefore commonly. studied in variable animals. Meristic Quantities. If the number of discrete, countable characters,. such as the number of segments, scales, or chaetae, varies, we speak of · meristic variation. Some meristic characters may be exceedingly constant, as the number of eyes or legs in man; others may have a characteristic variability, as scales in lizards or fin rays in fishes. TABLE

11.

MEASUREMENTS OF SERIES OF

Halcyon chloris pealei

FINSCH AND HART-.

LAUB FROM TUTUILA, SAMOA I

iI

Bill ....

I

Bill....... . ·)

I Mean

Range

-::-1!~4

Adult mules: Wing. Tail. Adult females: Wing.. Tail...

N

I

S.D.

~·-·

HO!

:-1-

49 49

63.5- 69.5 31.0- 39.0

29

95.5-102 ..5

29 28

64.0- 72.0 33.5- 37.5

97.48 66.44 34.46 \J8.86 67.62 :35.20

!



C.V.

-----

1.71 1.32 1.56

1.75 4.54

0.88 1.56 0.98

0.90 2.29 2.63

I. 98

The 0.V. of most meristic characters is smaller than that of linea .. measurements, and it is not permissible to compare the coefficients o . variability of the two kinds of characters. If one wants to compare th· O.V. of different groups of animals, one should compare relatively equiv ... lent sets of data, such as linear measurements with linear measurements;:· ratios with ratios, etc. · Ratios and Indices. While over-all length may be very variable in population, particularly in species that continue to grow as adults, th proportions of the various body parts to each other may remain rathe constant. For comparisons between populations, taxonomists therefo~~ often use ratios rather than linear measurements. They are usually·: expressed in the formula s x 100 R (ratio) l

where s = the smaller of the two values, l = the larger. This expresses • the size of the smaller value as the percentage of the larger. For example 1 . if we want to determine the relative size of the head in a species of fish,· we calculate (length of head X 100)/length of body (without head). Such a ratio is very quickly calculated with the help of a slide rule. If R is near 100, it may happen that s is larger than l in some samples. It is obvious that the positions of sand l cannot be reversed in such cases, even when R becomes larger than 100.

137

If we want to tell how much larger the large measurement is than the 8mall one, we can express the ratio as R = l/s. Ratios are best demonstrated visually in the form of scatter diagrams in which one value is plotted on the abscissa, the other on the ordinate (Fig. 21). By using different symbols for different populations, the presence or absence of overlaps between populations can be detected quickly. It can also be detected whether or not a straight-line correlation exists between the values. If one wants to determine the relative size of an organ or appendage, it is important that the proper standard of comparison be chosen. For instance, relative head width in insects is calculated against head length (without rostrum). Relative tail length in birds is usually calculated against wing length (as standard of general size). However, the "fog is not an accurate yardstick for general size in migratory and highaltitude birds, nor in some birds in which the wing is used in courtship. The cube root (~) of the weight might be a better measure in such species (Amadon, 1943). If an appendage is calculated against the whole, as tail against body, the appendage should not be included in the whole; the trunk without the tail should be used as standard of the "whole." As mentioned above, ratios are more useful as taxonomic characters than direct measurements, because -the variable factor of size is minimized (see below for change of ratios with size). Thus the head width of a small specimen of Cimex lectularius Linnaeus from Ain Sefra is 5.30, and that of a large specimen from Burkham, 6.30 (Johnson, 1939). Jenyns type of C. columbarius has a head width of 5.80. At first glance it would appear from these data that head width is of no value as a taxonomic character for the separation of lectularius and columbarius. However, the ratio of head width to length of third antenna! segment is 1.43 for both the small and the large specimens of lectularius and 1.81 for columbarius. This difference was borne out by measurements of large numbers of individuals throughout the range of the bugs, the average ratio for 1,723 specimens of lectularius being 1.45 and the standard deviation 0.079, whereas the comparable figures for 409 specimens of columbarius were 1.78 and 0.096. Specimens of ledularius from animal and fowl houses tended more toward columbarius (ratio of head width to length of third antenna! segment, 1.52; S.D., 0.079) but fell completely within the range of lectularius (after Johnson, 1939). Care should be taken to check the rate of increase of each of the measurements used in a ratio, because different parts of the body of an animal commonly grow allometrically. Thus in the genus Cimex, the percentage increase with age is greater for length of the third antennal segment than for head width in lectularius, whereas the opposite is true in columbarius (Fig. 22). In this example allometry is so slight that it does not affect the validity of the conclusions for taxonomic purposes. In some

l

138

r.=====;r======r=====r======r=====T=====i=====r=====;=====;i: 0 0 0

0

0

0

0 0 0

0

0 0

0 Oo

300

0

o

o



0

250

8 oco

o

0

0<6>

0

e

0

.,oo & 0

• •

eases, however, allometry renders a particular pair of characters completely useless for taxonomic purposes. Parr (1949) describes a method of regression analysis dealing with a pair of characters showing allometric growth, a method which he found useful in fish taxonomy. Qualitative Characters. When comparing two samples, one often finds that they differ merely by the degree of expression of a qualitative character. For example, birds from one region may be more brownish, from another more grayish, with some overlap. There are various ways

0

0 0

0

0

.; 200

2nd ANT. SEC. 3rd ANT. SEG. HEAD-WIDTH

X 2nd+ 3rd +4th ANT. SEG.

o

4th ANT. SEC.

COLUMBARIUS

LECTULARIUS

~-cP_'=f--0~-0~~~~ 0

00

..cl.

0

0 b 0

0

o

139

QUANTITATIV E METHODS OF ANALYSIS

TAXONOMIC PROCEDURE

50

..

00

--- ......--

_,/./ .·"' .·••·

/ ..-

______ .o

0

0

0 0

0

'DIS 'J.NY PJWfflQIM-aY!H NY!K 1Y001

ADULT

ADULT

JKSTAllS

INSTARS

FIG. 22. Percentage increase with age in head width and antenna! length for Cimex lectularius Linnaeus and Cimex columbarius Jenyns, showing allometric growth in columbarius (after Johnson, 1939).

of translating this qualitative difference into quantitative terms if it is desirable to determine the zone of overlap more accurately. For instance, all the specimens of the various samples can be arranged in a single series, ranging from one extreme (numbered 1) to the other (numbered n). As an example, the 23 adult specimens of the thrush Bmchypteryx leucophrys Temminck from the Malayan islands, which are in the collections of the American Museum of Natural History, can be arranged in a single series. Listing the most rufous bird first and the most olive bird last, we find the following sequence: S, S, Sb, S, B, Sb, Sb, S, T, T, T, T, J, T, J, Sb, Sb, L, L, J, L, L, L (B = Bali, J = Java, L = Lombok, S = Sumatra, Sb = Sumbawa, and T = Timor). The average rank of these populations is then as follows: Sumatra, 1, 2, 4, 8 (3. 75) ; Bali, 5; Sumbawa, 3, 6, 7, 16, 17 (9.8); Timor, 9, 10, 11, 12, 14, (11.2); Java, 13, 15, 20 (16.0); Lombok, 18, 19, 21, 22, 23 (20.6). There is much

l

140

TAXONOMIC PROCEDURE

overlap in the characters of the populations, even though the Lombok ·. birds are strikingly more olive than the Sumatra birds (Mayr, 1944).' A more accurate method, which is definitely to be preferred whenever large samples are available, is to establish a number of "classes." For. these Brachypteryx specimens one might choose the following classes: class 1 (rufous), class 2 (fairly rufous), class 3 (rufous olive), class 4 (olive), class 5 (very olive); and select as standard of comparison that specimen that is closest to the mid-point of each class. If material is plentiful and greater accuracy is desired, the differences between the mid-points, can again be divided into decimals that can be estimated. Differences in pattern, such as degrees of spotting or banding, can also: often be expressed in quantitative terms. Students of small mammals. measure color differences quantitatively with the aid of a photovolt,'. reflection meter (Blair, 1947). ANALYSIS OF DIFFERENCES BETWEEN POPULATIONS

When two populations are compared (or, more accurately, sample from two populations), the taxonomist usually wants to know one these three things : 1. If they are polymorphic, whether or not the frequencies of the tw ~ forms are the same in the two populations. This is achieved by the chi.i square test (comparison of frequencies). ' 2. If they differ only slightly by a quantitative difference, whether o' not the difference is statistically significant. This is determined by· · statistical comparison of the means. 3. If they are clearly different, how much the population curves overla. ' (determination of overlap). Comparison of Frequencies: Chi-square (x 2 ) Test. A taxonomist i · often confronted with the problem of having to determine whether tw · variants occur in two or more populations at the same frequency. The ·'~ variants may be color forms or even the two sexes. The conventional . way is to express the frequencies in percentages. But the calculation · of a percentage does not tell us whether the populations are actually:different or whether the observed difference between the samples is_ merely due to accidents of sampling. This can be determined by the : chi-square test. Let us consider a specific example. Among 80 specimens from locality A there are 58 of type 1 and 22 of type 2; among 43 specimens from local- 1 ity B there are 24 specimens of type 1 and 19 specimens of type 2. Type 1 is therefore represented at locality A by 72.5 per cent of the specimens and at locality B by 55.8 per cent. Is this difference indicative of a population difference?

QUANTI7'ATIVE METHODS OF ANALYSIS

141

To test it, we set up a four-square table, Type Type 1 2

Type Type 2 1 Locality A

I

a

I

22 58 -----2,i 19 B

A

b

i--:--Locality B I c I d

and solve the equation: 2

X

(ad - bc) 2 (a + b + c + d) = 3 65 + d)(a + c)(b + d) ·

= (a+ b)(c

What does thii; chi-square value of 3.65 mean'? It h~i; It would lead ui; too far here to explam the statistical theory of significance, and we refer to textbooki; of statistics (Simpson and Roe, 1939; Snedecor, 1946). When the difference between samples is "significant,'' it indicates that they were presumably not drawn from the same population. Significance is something relative; it indicates a deviation from expectatipn. Significance is expressed in P ( = ptobability) values. If an event is expected to occur in 1 out of 20 tries, this would indicate a P value of ~2o = 0.05 ( = 5 per cent level of significance). If it is expected to occur in less than 1 in 100 tries, the P value is less than 0.01 (below the 1 per cent level of significance). p tables can be found in all standard statistical texts. By reference to a p table it is found that a chi-square value of 3.84 or larger ii; considered 1,;ignificant (P = 0.05). Thus the above-calculated figure of 3.65 does not quite reach the 5 per cent level of significance. If the samples (N) are small Yates's correction must be made (see statistics textbooks). Some stati 8~icians recommend this as a routine procedure in all cases. Comparison of Means. The si~ples~ solution of_ the problem of whether or not samples from two (b10log1cal) populat10ns are taxonomically identical is to compare their means. If the two sample means do not differ significantly, it indicates that the samplei; could have been drawn from the same (statistical) population. The closeness of the sample mean to the real but unknown mean of the population is indicated by the standard enor (uM or S.E.M), which obviously depends on the size of the sample (N) and is therefore expressed as follows: Significance.

to be checked for "significance."

SE • ·M

S.D.

= .VJJ

(standard deviation divided by the square root of N, or sample size). The standard error is the same kind of probability estimate as thP

l

142

TAXONOivfIC PROCEDURE

QUANTITATIVE METHODS OF ANALYSIS

standard deviation and has the same distribution characteristics (i.e.,1 those of a normal curve): 68.27 per cent of the observed sample means will fall within ± 1 S.E., 95.45 per cent within ± 2 S.E., 99. 73 per cent within ±3 S.E., etc. As a simple rule it can be stated that two samples are probably : different if the difference between the means (M 1 - M 2) is more than-• twice the sum of the standard errors (S.E.M 1 + S.E..lf,) and almost'' certainly different if it is more than three times the sum of the standardi errors. If one wants to know whether the difference between the two sample'., means is statistically significant, it is necessary to calculate the standard error of the difference (S.E.a) between the two means, which is the square·; root of the sums of the squared standard errors:

(while one glance at a pair of overlapping population curves shows that the "hump" of population curves is much more important than the "tails"). The calculation of linear overlap is obviously unsatisfactory. Before presenting a more adequate method, a few words need to be said on the aims of these methods. The object of the taxonomist when comparing allopatric populations is most frequently to determine whether or not they belong to different subspecies. How different do two populations have to be to be recognized as two different subspecies? There is no general agreement on this point. Some splitters recognize populations as subspecies even if only the means differ "significantly" (in the statistical sense). Reasons against a subA 114

99

In this case we assume that the means of the two populations fro which the samples were taken are equal (i.e., we assume that the tw samples were taken from the same population). Then if the differenc between the means is over 3 times the S.E.a, the hypothesis is incorrect' and the two samples .yere drawn from different populations. If the number of specimens (N) is very different in the two samples, more elaborate formula is advisable (Simpson and Roe, 1939):

s E = '\jIN N2 (S .E 1



·d

)2

·Mi.

+

N (S.E.M,) 2 N1 2

However, Hubbs and Perlmutter (1942) have shown that the simpler form' is adequate in most cases, and that it leads only rarely to serious error/ They also give a table oft values for these statistics. .Overlap between Populations. The simplest case is that everyc specimen of population A_ is different from every specimen of population B. More difficult are the cases where there is an overlap in characters. , For instance, in the Polynesian honey-eater, Foulehaio carunculata (Gmelin), adult males from the Manua Islands have wing measurements of 99 to 106 (average 104.7) mm., from Tonga 104 to 114 (average 108.3) mm. How great is the overlap? The coarsest way of testing overlap would be to plot the linear overlap _ of the observed samples (Fig. 23). The figure shows that 2 of the 7 mm. ( = 28.5 per cent) of the 0bserved range of sample A (Manua) are overlapped by sample B (Tonga). This method of linear overlap is misleading in two ways: it gives only the overlap of the samples (which is much smaller than the overlap of the populations); and it exaggerates the importance of the end points of the range

143

I

j 1I I I

i 1I I

I j 1I I I

I

100 105 110 115 8 Fw. 23. Linear overlap of observed samples.

species criterion of such low significance have been stated in Chap. 2. Some lumpers, on the other hand, go to the opposite extreme and insist that populations are not worthy of subspecific separation unless all individuals are different. The most frequently proposed convention is the so-called "7 5 per cent rule." This rule is subject to various interpretations. For instance, some taxonomists are satisfied if 75 per cent of all the specimens before them can be placed as one subspecies or the other. Most taxonomists accept, however, an interpretation of the rule according to which population A can be considered subspecifically distinct from population B if 75 per cent of the individuals of A are different from "all" the individuals of population B. Unfortunately, this approach does not eliminate all the weaknesses of linear overlap, because again the end points of the range of variation are given crucial importance. How many standard deviations on either side of the mean should one include in a curve which theoretically reaches infinity? The tail of the curve attains such a flat slope beyond about 2 S.D. from the mean (Fig. 20) that little is added to the population by extending it. If we cut off the "tail" at 2.06 S. D. from the mean, we lose only 1.97 per cent of the population. By adding 1.18 S.D. (extending it to 3.24 S.D. from mean), we add only 1.91 per cent, which would give us 99.94 per cent of the population. This is the standard accepted by Amadon (1949), who proposes the following interpretation of the 75 per cent rule: A population A is subspecifically distinct if 75 per cent of its individuals differ from a "standard population" (Simpson, 1941) of

l,

N

00

:

70

60

QUANTITAT IVE METHODS OF ANALYSIS

...········........ .: i

90

lo)

B

1,000 individuals ( = 99.94 per cent of the theoretical range of variabil-

:

ity) of B (which corresponds to 3.24 S.D. on either side of the mean). At this standard, more than 97 per cent of the individuals of A are different from more than 97 per cent of B, assuming the two standard

: i.

50 40

..

30

20

·······:

10

24

25

26

27

28

1-!5

29

No of dorsal rays

A lb)

deviations to be roughly alike. This would amount to no overlap at all in most of the relatively small :;amples usually available to taxonomist8 . Such a i:;tandard appears unnecessaril y high. Current taxonomic practice8 permit greater overlap, although there is no agreement as to how much (Rand and Traylor, 1950). We suggest accepting as a 8tandard of subspecific separation that 75 per cent of population A be different from 97 per cent of population B. Then about 90 per cent of the individuals of A are different from about 90 per cent of the individuals of B. How can this be expressed in terms of standard deviations? When calculating overlaps, we are interested only in the part of the t"·o population curves which is between the two means. In order to ealculate the standard deviations that concern us, we must divide each population into that portion which lies between the means and that which lies outside. Of the 75 per cent of A, we find 50 per cent below the mean (M 1 ) and 25 per cent ( = 0.674 S.D.) in the zone of overlap. Of the 97 per cent of B, we find 50 per cent above the mean (M 2) and 47 per cent ( = 1.881 S.D.) irr'the zone of overlap. ·It is then evident, on the basis of the 75 and 97 per cent standard adopted by us, that two populations are 1m h;.;pecifically different if the difference of the means Olf 2 Mi) exceedi; the :;um of 0.674 S.D. + 1.881 S.D. =2.56 S.D. This simplified statement ignores the faCt that the standard deviatiorn of A and B are usually different. Even u:;ing the formula, l

0.674 S.D.A A Id)

S.D.JJ = 2.56 S.D.,

i8 only a slight improvemen t, since the 8olution is asymmetrica l, and we :;hall get different results when determining subspecific difference by cal('Ulating 75 per cent of B differing from 97 per cent of A, that is,

0.674 S.D.JJ

Fm. 24. Various interpretation s of the 75 per centrule for subspecies discrimination . (a) ~o~e :han. 75 per ?ent .of the available sample will be correctly identified by placing the d1VIdmg line arbitrarily at a stated point. Illustrated by the number of dorsal rays in Cyn-0scion regalis, (A) subsp. arenariU8 (Gulf Coast) (usually 26 or less) and (B) subsp. regalis <.Atlantic Coast) (usually 27 or more) (from Ginsburg, 1938). (b) 75 per cent of A differ from 100 per cent of B. (c) 75 per cent of A differ from all ( = 99.94 per cent) of a standard population of B. (d) 75 per cent of A differ from all ( = 1.881 S.D. = 97 per cent) of B. 144

+ 1.881

+ 1.881

S.D ..1 = 2.56 S.D.

It is evident that we should look for a i:;ymmetrical solution. The ideal solution would be to determine the point of intersection (J) of the two curves. Unfortunate ly, the calculation of this point is very laborioui; and not suitable for routine taxonomic work. However, an approximati on method exists, the error of which is small compared with the various other inaccuracies of the determinati on and comparison of subspecies differences. It is possible to calculate very simply a point which is close to the point of intersection, provided that the larger of the two standard devia-

146

147

TAXONOMIC PROCEDURE

QUANTIT11TIVE METHODS OF ANALYSIS

tions is not much greater than one and a half times the smaller. We obtain this approximated point by dividing the difference between the means by the sum of standard deviations. Let us call this figure the ' coefficient of difference (C.D.).

The method will be illustrated by an example. Oliver (1943) attempted to determine whether lizards of the species Uta ornata from northern Sonora (Pilares) and southern Sonora (Guirocoba), Mexico, are subspecifically distinct. The chief differentiating character between the two popubtions is the number of the enlarged dorsals of the primary row of scales. Means and S.D. of the two populations are as follows:

C.D.

Jfs -

S.D.A

1lfA

+ S.D.e

Guirocoha Pilares



The value which corresponds to our standard of subspecific difference (75 1 per cent A from 97 per cent B) = 2.56/2 = 1.28. Then, if the C.D. exceeds 1.28, it seems probable that it will be advisable to separate the two, populations subspecifically. At this value about 90 per cent of A is; different from about 90 per cent of B. ' It is perhaps advisable to express the difference of two populations in. terms of the magnitude of equal nonoverlap, e.g., 90 per cent of A not overlapped by 90 per cent of B. In view of the importance of this stand-' ard, we add a table of the percentages of symmetrical nonoverlap asso-~ ciated with various C.D. values (Table 12). This will permit a quic TABLE

12.

PERCENTAGE

OF NONOVERLAP

OF PARTIALLY

OVERLAPPING

CUR

ASSOCIATED WITH STATED VALUES OF THE COEFFICIENT OF DIFFERENCE (C.D

Values

C.D. -----

Below the level of conventional subspecific distinctness

Conventional level of subspecific difference Above the level of conventional subspecific differen~c

0.675 0.84 0.915 0.995 1.04 1.08 1.13 1.175 1.23 1.28 1.34 1.405 1.48 1.555 1.645 1. 75

Joint nonoverlap, per cent ~-'-

75 80 82 84 85 86 87 88 89 90 91 92 93 94 95 96

check on whether or not a population difference is presumably level of subspecific distinctness. Only values of C.D. near 1.28 are given. Obvious subspecific identity is indicated by values much lower ' than 1.28; obvious subspecific difference is indicated by values that are much higher.

Mean ...

S.D ..

27.76 1.92

34.60 2.07

i11e - MA = 34.60 - 27.76 = 6.84 = 1 71 C.D. =SD +SD 1.92 + 2.07 3.99 . • •A • ·B

The C.D. of 1.71 indicates that more than 95 per cent of the Guirocoba population is different from more than 95 per cent of the Pilares population, in other words, that the two populations deserve to be separated suhspecifically. It must be understood that this method is only a rough approximation. It makes various assumptions that are not necessarily correct, e.g., that the distribution curves are normal and that the sample mean is the population mean. The evaluation of the co~fidence limits is laborious and has not been attempMd here. Nor has any attempt been made to extend this rather coarse method to multiple characters. What is an even greater shortcoming is that such an arbitrary method does not allow for the many biological and biogeographical considerations of subspecies recognition. A yardstick such as the coefficient of difference is a useful guide and a help toward more uniform standards, but ·~11 borderline cases should be evaluated in the light of additional information. A well-isolated population may be worthy of subspecific recognition with a C.D. as low as 1.28, while a population that is part of a eline or of a checkerboard pattern may not be worth naming even though the C.D. exceeds 1.5. In a Polynesian honeyeater, Foulehaio carunculata (Gmelin), for example, adult males from Tonga have a wing length of 104 to 114 (average 108.3) mm., those from Fotuna Island of only 93 to 98 (average 95.9) mm. This would seem like a difference more than sufficient for subspecific recognition. However, not only do additional populations on some thirty other islands bridge the gap between these two extremes, but the populations of large-sized and small-sized forms are distributed in such an irregular manner as to make an intelligible delimitation of subspecies impossible (Mayr, 1932). Multiple Character Analysis. Two populations that cannot be distinguished unequivocally by a single character can often be separated by using in the analysis simultaneously two or more characters. There are several methods available that permit such multivariate analysis. They

148

149

TAXONOMIC PIWCEDURE

(JUANTITATJVE METllUDS OP ANALYSIS

are based on the observation that two characters are usually only incom-, pletely correlated. If, for instance, a bird population differs from another,, one by longer measurements of wing and of bill, and if we arrange all the · specimens in a series from the smallest wing length up to the longest, it is· very unlikely that they will fall into exactly the same series when arranged . according to bill length. If there is a slight overlap in the measurements of the two populations, it is sometimes possible to eliminate the overlap·.· by adding (or multiplying) wing and bill length of each individual. Much more reliable, but also more elaborate, are various other methods.~ Fisher (1938) gave a short review of the subject, and Burma (1949) has demonstrated one of the methods of multivariate analysis on a practical' example. ' Fisher's method of discriminant functions is probably the most usefuf; of these methods; it is explained in detail in Mather's (1947) textbook/ Recent applications of this method relate to differences of populations of fish (Stone, 1947), Drosophila (Carson and Stalker, 1947), and birds' (Storer, 1950).

where the value for one character (read off the ordinate) intersects the value for the other character (read off the abscissa); each population is indicated by a different symbol (circles, squares, triangles, solid or empty, etc.) (Fig. 21). Scatter diagrams have many advantages. They help to visualize allometric relationships and facilitate the plotting of regres::;ion lines. They also sometimes disclose errors of measurement or sexing that might otherwise go undiscovered.

1

Vagrons

Biscutatus Elegans (Lassen County) Elegans (Oregon, Californio West of Sierran Divide) Elegans (San Bernadina Mts.)

VISUAL PRESENTATION OF QUANTITATIVE DATA

It is often desirable to present numerical data visually. Such a visu presentation not only permits a rapid survey of all the data, but actuan · often brings out fine points that are not apparent in the raw data. A fe ' simple methods may be described (see also Anderson, 1949, Chap. 6) ... Histograms. Unreduced samples are best shown as histograms. 'M histogram consists of a set of rectangles in which the class means ar·:· plotted on the abscissa and the frequencies (usually number of specimens.' on the ordinate. This presentation has several advantages. Th··' principal one is that it presents the original data in minimum spac(ff Whatever form of statistical analysis a subsequent author may want tor apply, he will find the actual number of specimens given for each size'. class. A quick comparison of different populations is made possible by arranging a series of histograms above one another (Fig. 25). ., Population-range Diagrams. Even more data can be compressed into minimum space by giving sample range, one or more standard deviations, and two standard errors. This is the method of Hubbs and Perlmutter (1942), who plot one standard deviation (Fig. 26). A better solution•' would probably be to plot one and one-half times the standard deviation. : Nonoverlap of these plotted standard deviations (131 + lYz = 3) would · at once indicate probable subspecific difference. (For a discussion of the · Hignificance of the difference of means, see above.) Scatter Diagrams. The difference between two or more populations in respect to two characters is best illustrated by a scatter diagram. Each individual is indicated by a spot or other symbol which is placed

Hydrophila (Humboldt ond Mendocino Counties) Hydrophilo (Umpqua, Rogue ond Klomoth River Basins)

Couch ii Hommondii At rat us

Ordinoides

30

40

50

60

Frn. 25. Histograms showing head and body length in centimeters of adult males of 7'harnnophis ordinoides. Each square represents a specimen (Fitch, 1940).

If three characters are involved, triangular charts can be employed. In this case the actual values are not plotted, but rather their percentage <'ontribution to the sum of the characters. For example, if character a = 80 mm., b = 32 mm., and c = 48 mm., then

a

+b+c =

160 mm.

= 100 per cent.

Then a = 50 per cent, b = 20 per cent, and c = 30 per cent of the whole. These percentages are plotted on the graph, which thus shows proportions rather than absolute sizes. In each individual case the triangular

150

QUANTITA TIVE METHODS OF ANALYSIS

TL1XOf'-lOJ!IC PIWCEDURR

graph is scaled in such a way as to produce a maximum spread of the points. As an illustration we reproduce such a triangular chart from a recent paper by Burma (1948) (Fig. 27). Mapping of Quantitative Data. It is often desirable to illustrate the geographica l relationship s of various populations with different quantitative characters. In the case of continuous characters (size, etc.), the simplest method is to record the means of the various populations on a base map, and if there is regularity to draw in the isophenes ( = lines connecting points of equal expression of a character). For instance, if the means of a series of populations in a species vary from 142 to 187, it is helpful to draw in the isophenes of 140, 150, 160, 170, 180, and 190. If qualitative or semiqualita tive characters are to be plotted, it is sometimes helpful to choose a different symbol for each class of characters. The relative size of the symbol can be used to indicate sample size (Fig. 5). To present frequencies of polymorph characters on a map, the "pie graph" is the most convenient method. The percentage occurrence within the population is indicated hy the size of the segments (Fig. 28).

44

42

"'~

~

>"'

t A

n

40

8

t f t c

F

0

[

G H

J

38

K

f tf N

L

I;, I

0

M

locol1t1es from North to South

FIG. 26. Population-ra nge diagram. Variation in the number of vertebrae of th~ anchovy, Anchollietla mitchilli. The letters A to 0 refer to 15 population samples' arranged from north (A) to south (0). In each sample the vertical line indicates thei total variation of the sample; the broad portion of the line, one standard deviation 011,, each side of the mean; the hollow rectangle, twice the standard error on each side the mean; and the crossbar, the mean (Hubbs and Perlmutter, 1942).

oi'

D

Frn. 27. Triangular graph of the length (L), height (H), and distance to maximum
APPLICATIO N OF STATISTICA L METHODS

Comparison of populations is an important task of the taxonomist, and it is the principal object of an accurate quantitative description of a population to characterize it in such a way as to facilitate such comparisons . In experimenta l sciences the investigator usually wants to know whether there is a significant difference between two sets of experiments . This significance is usually expressed as the probability that the various samries were drawn from the same "population " (in the special statistical sense of this word). Most experiments are designed to tm;t whether a stated ch::mge in the experimenta l conditions produces a "significant " change in the experimenta l retmlts. The experimente r is interested primarily in knowing whether or not a change has occurred and only secondarily in measuring its quantity. The taxonomist, when comparing (allopatric) populations of the same ,.;pecies, knows before he starts that they are not completely identical. Population geneticists have demonstrate d conclusively that in sexually reproducing animals no two natural populations are ever exactly alike. In fact, even populations from the same locality may be slightly different at different seasons of the year. The mere fact of a (statistically provable) difference between several populations of a species is therefore of no 1
152

-~··.l>):·;.,.. \ ),,.(__c·~~

TAXONOMIC PROCEDURE

QUANTJTA1TVE J.v!ETHODS OF 11NALYSIS

153

wants to know is whether these differences between populations are large enough to justify classification in different taxonomic categories. The animal taxonomist is thus principally interested in the quantity of difference. On the other hand, statistical methods do not usually reveal much about the quality of a difference. They do not permit a decision as to whether or not two allopatric populations belong to the same species, or whether two sympatric variants are individual .variants or sympatric species. Reproductive isolation (the species criterion) and morphological differences are not necessarily closely correlated. Sibling species may be almost identical morphologically, while subspecies are often (e.g., birds of paradise) Htrikingly different. Presence of intergradation between populations prove8 conspecificity; absence of intergradation bet ween spatially isolated populations is ambiguous evidence-it does not prove reproductive isolaT;ion. Even with respect to the recognition uf subspecies, statistical methods provide only one line of evidence. There are thus many limitations to the information that statistical methods can yield. This still leaves a wide scope for the application of ~tatistics to taxonomy. The particular method that needs to be applied depends on the taxonomic group and on the particular problem. For instance, species of birds are well known and well defined, and the most frequent problem with which the student of living birds is faced is whether two allopatric populations are sufficiently distinct to be considered different subspecies. Occasionally he has to analyze a sample of migrants and allocate them to one of the breeding populations. A paleontolo~ist has to deal with many additional problems. If he has a secondary deposit, he wants to know whether the sample is homogeneom; from a single population. A study of the enough to have. been derived I variability of the sample will yield clues useful in answering this question. In the last analysis, statistics is merely an extension of ordinary reasoning as applied especially to numbers. Hence the statistical treatment of a problem is no better than the facts and judgments upon which it is based. Statistics should never be used to impart a false sense of precision. As applied to the comparison of populations by means of the normal curve, stati8tics deals with probabilities, not with certainties. In practice, taxonomists should realize that there is no particular virtue in statistics per se. Statistical analysis is one of several tools which are available and which may or may not be used to advantage in a particular problem. Taxonomic studies at the alpha level utilize in general only the simplest statistics, e.g., size, including range, if several specimens are available, and simple proportions or ratios. Population analyses are impossible because of the small numbers involved, and because there is no uniformity of sampling. Furthermore, comparisons

I'

l/.

15-!

TAXOXOJflC PROCEDURE

between int.raspecific populations to determine overlap cannot be ma as long as the diagnostic differences of the species and their delimitati •· are still uncertain. Gamma taxonomy, on the other hand, focuses on the populati rather than the individual. The polytypic species is of paramount inte ' est, and a study of quantitative characters is the rule. Taxonomis who study groups of organisms whose study has advanced to this lev will find that statistical methods are an indispensable tool.

CHAPTER 8

PRESENTATION OF THE FINDINGS (DESCRIPTIONS, KEYS, PHYLOGENIES) After taxonomic characters have been studied and analyzed, there remains the important task of recording the findings and presenting them in a useful manner. This is achieved in systematic zoology by means of descriptions, keys, and classifications. Depending on the immediate objective of the taxonomist, any or all of these methods of presentation may enter into the final published work (Chap. 9). DESCRIPTIONS

The chief objective of a description is to facilitate subsequent recognition of the category involved. It was realized at an early date that different kinds of descriptions approach this goal in a different manner. Linnaeus distinguished clearly between the general descriptio (character naturalis) on one hand and the polynominal differentia specijica (character essentialis) on the other (Svenson, 1945). The latter contains "the essential characters by which the species is distinguished from its congeners." ,l It corresponds to what is nowadays called a diagnosis. The functions of the two kinds of description, the general description and the diagnosis, are ,by no means identical. The diagnosis serves to distinguish the species (or whatever taxon is involved) from other known similar or closely related ones. The general description has a broader function. It should present a general picture of the described taxon. It should give information not only on characters that are diagnostic with relation to previously described species, but also characters that may distinguish the species from yet unknown species. It should also provide information that may be of interest to others besides taxonomists. Linnaeus and many taxonomists since have stressed the extreme practical importance of a short, unambiguous diagnosis. It can only rarely be combined successfully with the general description. The latter, in turn, no matter how exhaustive it is, cannot always provide a substitute for a type specimen· (see Chap. 12) or, in many cases, for illustrations. There is still considerable confusion in the literature concerning the meaning and usage of the terms description and diagnosis. Simpson (1945) states that in describing animals the taxonomist should achieve 155

156

TAXONOMIC PROCEDURE

two objects, that of diagnosis and that of definition: "Diagnosis* is t art and practice of distinguishing between things. Definition t is th art and practice of setting limits to things. Both enter into taxonom .• and . . . they are essentially different and their complementary role · should be clearly understood." Although the formal diagnosis i taxonomic work sometimes assists in the definition of a category, th' function is mainly performed by the general description. The tw. terms, diagnosis and description, may then be used as follows: Description. A more or less complete statement of the morphologic characters of a taxon without special emphasis on those characters tha distinguish it from coordinate units. Diagnosis. A brief listing of the most important characters or char acter combinations that are peculiar to the given unit and by which i. can be differentiated from other similar or closely related ones. Th ' direct comparison of a species (or other taxon) with other specificall · mentioned species (or other taxa) is usually called a differential diagnosia• Such a comparison with other species is of great practical help t .· students who have no material of the newly described form. It also. forces the author of a new form to review all the evidence for and against, the publication of the description (Rensch, 1934). Such a compariso ensures also that the diagnostic characters of the new form are mentione _ and is therefore recommended by the International Commission · Zoological Nomenclature (Paris, 1948). If the nearest relatives are ra' or poorly known, it is also helpful to make a comparison with a we known, if not so closely related, species. The Original Description. The description given at the time o . proposal of a name for a new species, genus, or other category is calle the original description. It has two primary functions. The first, a~­ stated above, is to facilitate subsequent recognition and identification; the second is to make the new name available by fulfilling the require.1' men ts of Art. 25 of the International Rules of Zoological Nomenclature'. 1 (Chap. 11). 1 The preparation of a proper description is a task the importance of'. which cannot be overemphasized. The describer is forced to rely on< words to convey his meaning. Yet words, no matter how carefully chosen, are rarely adequate to give an accurate mental picture of the , appearance of an organism. Nevertheless, it is the function of the. description to enable a subsequent worker to identify specimens without . reference to the type. This goal can be achieved in most cases by the ' careful worker, particularly when the description is properly coordinated with illustrative material. * Ultimately from the Greek o'"''Y''Y i-w"'"'' to distinguish between two (things). . t Ultimately from the Lntin defi11io, to enclose within limits.

PRESEN'l'ATION OF THE FINDINGS

157

The good description requires on the part of its author (1) a thorough knowledge of the group of organisms concerned, (2) a knowledge of structure and terminology, (3) an ability to evaluate differences and similarities, (4) an ability to select and emphasize the important, (5) a full nnderstanding of the precise meaning of the words and the correct usage of the grammar of the language employed, and (G) a concern for the future worker. Ferris (1928) has stated, "If [the describer's] work of recording the data has been properly done those data are available for re-examination and re-evaluation. His conclusions can be checked, they can be extended or modified or rejected as appears desirable, all without the necessity of recourse to his types." A brief review of the literature is sufficient to reveal that the form and · style of descriptions are as individual as their authors, and that many authors are inconsistent in their choice of form and style. As we have previously emphasized, originality is an asset l.n approaching a problem but becomes a liability when carried over to the recording of data. In the lesser known groups much of the taxonomist's time is spent in comparing and contrasting one description with another. This task is difficult under any circumstances but is easier when the descriptions approximate one another in style, arrangement, and form. This does not mean that a completely standardized description is always possible or even desirable. The factors which influence the order of presentation, form, and style are factors which vary from group to group. Within a particular group, however, much can be done to standardize descriptions and thus increase their effectiveness and utility. Style. The style generally used in descriptions as well as in diagnoses is telegraphiul and concise. It is usually characterized by elimination of articles and verbs and by selection of adjectives and nouns of explicit meaning. It further involves proper use of capitals and punctuation and adherence to a logical sequence of presentation. Thus the telephonicKtyle statement "The head is one-third longer than it is wide, the ' antennae are shorter than the body, and the outer segments are serrate " becomes simply "Head one-third longer than wide, antennae shorter than body outer segments serrate." The descriptive style of the second ' . . statement has lost none of the preciseness or clarity of the first, yet If' only one-half as long and may be both read and understood more quickly. Sequence of Characters. The recommended sequence of characters depends on the form of the description. It is customary in a diagnosis to present characters in the order of their diagnostic importance (or what the author regards as the order of importance). This will facilitate rapid recognition. In the full description the material should be arranged in a standardized natural order, as, for instance, describing the body parts from anterior to the posterior, first on the dorsal and then on the

L

158

1'AXONOMIC PROCEDURE

l'RESEN1'A1'ION OF THE FINDil,iGS

ventral surface. The details may be varied to fit the group, yet stil~ maintain a natural and readily comparable order. For instance, th· sequence of presentation for a dorsoventrally flattened animal groull' would be different from that for either a laterally compressed or a robu ' group because of the different methods of orientation during stud~~ The standardized sequence of characters helps assure that nothing impor.;., taut is being overlooked and that the description is comparative. It is! very frustrating to try to use a taxonomic paper in which half a doze ' species are described independently of one another, details being given;'. for example, of the antennae of one species, the pronotum of a second' and the elytra of a third. Such a procedure makes comparison quite impossible. Authoritative monographs usually adopt a standardized sequence of characters, and subsequent describers should follow it as fa , as possible. The utility of a description may be increased by the use of device " which enable the reader to locate quickly the particular characters fo_ which he may be looking. One such device is the use of paragraphs t'' break up the description according to main body divisions (e.g., in insec head, thorax, abdomen, wings, genitalia, etc.). Where paragraphing· undesirable, the same effect may be gained by italicizing these same ke' words. If the author has followed a natural sequence of presentatio ' either method will permit the reader to orient himself quickly at som particular point in the description without the necessity of reading th: whole description. '! What to Include in a Description. An exhaustive description of organism would fill many volumes, as may readily be seen from a perus of volumes on the morphology (physical anthropology) and anatomy o the human species. It is, therefore, obvious that even the so-calle ' "detailed description" of a taxonomic species is highly selective and i , the nature of an expanded diagnosis. How much subject matter shoul~'=" be included in a description depends on the group-concerned and the stat · of knowledge of that group. Excessively long descriptions obscure the essential points; excessively short descriptions omit pertinent data. While the diagnosis serves to distinguish a species from other known . species, the description should be detailed enough to anticipate possible : differences from as yet undescribed species. The description should therefore be very detailed in poorly known groups, because it is impossible to predict which characters will distinguish a new species from those that are still undiscovered. On the other hand, the subspecies in a well- -known species of birds may differ from one another so little in detail that an extensive description would be a repetition of the species description. , In such a case the description may not differ from a diagnosis as, for

instance, "Like subspecies alba but larger, upper parts blackish gray, not ash gray" (followed by a tabulation of the measurements). Descriptions should include, in so far as practicable, all characters, both positive and negative, which are known to be useful or potentially useful in distinguishing other units in the same category. However, characters of higher categories should be omitted except where they are 11 nomalous or where the assignment of the unit to the higher category is in doubt. For example, the description of a subspecies of song sparrow should not include reference to characters that are typical for all song sparrows (or worse, for all sparrows!). Violation of this rule is not only uneconomical but distracts attention from the essential features of the category concerned. Beyond the above generalizations, there is little to guide the describer other than his own good judgment. The description, more than almost 11 ny other aspect of taxonomy, provides a permanent record of the author's ability to observe accurately, record precisely, select and interpret intelligently, and express clearly and concisely the facts which are before him. The description should include a statement of the differences between the sexes and, if only one sex is available, a frank statement of the fact (e.g., "female unknown"). Likewise the characters of immaturity sho_uld be discussed as well as larval stages. Available biological and ecological data should be presented. Such information is, in the case of sibling species, often more important than morphological chara?ters. Whether or not the description should be based exclusively on the type is a much disputed point. Those who favor this method argue that all too often it has eventually turned out that the original material and consequently also the description were a composite of several spe~ies. This makes it very difficult to disentangle the characters of the various species. They argue that it is much safer to restrict the description to the type and have it followed by a discussion of the variability of the n~st of the material. Others believe that such treatment favors the erroneous typological view that the type has a special significance as far as the characters of the species are concerned. They prefer the description to be a composite drawn from a consideration of the entire material and propose to mention at the end by what characters (if any) the type specimen differs from the rest of the material. Actually both methods agree that (1) the entire variability of the species material should be described and (2) that it is advisable to mention the special features of the type specimen. Different authors may use different methods to achieve these objectives.

c

159

160

TAXONOMIC PROCEDURE

Des?ription of ~olorati?n. Differences in coloration are among· t most important diagnostic characters in many groups of animals. detailed descri~tion of the_ general pattern of coloration and of the preci tones of ~he v~nous color~ is t~erefore essential in many taxonomic groups Subspemfic differences m birds, mammals, and butterflies are ofte largely a matter of coloration. Many attempts have therefore bee made t~ standardize color descriptions, since rujous or tawny do no necess_anly suggest the same shade of color to every taxonomist. It i fo: this reason that color keys are widely used in taxonomy. Those R~dgway (1912), Maerz and Paul (1950), and Villalobos-Dominguez an' Villalobos (1947) are specially recommended. When fine shades · colo_r are involved, a direct comparison with topotypical material advisable. Even here the color keys are useful for standardization terminologies. . Numerical Data. The recording of a set of precise measurements is mtegral part of a well-rounded description. If the new form differs fro its relatives in its proportions, such proportions should be recorded (s Chap. 7). Exact data should be given of numerically variable features structure or pattern, such as numbers of spots, spines, scales, tail feathers and so f?rt?. The reasons for including such data are stated in Chap. Descriptive Treatment. A full descriptive treatment of a species ma · take the following form: Scientific name

Taxonomic references and synonymy (if any) Type (including type locality and repository) Diagnosis and differential diagnosis (brief statement of essential differ; ences from nearest relatives, see above) Description Measurements and other numerical data Discussion Range (geographical) ~abitat (ecological notes) and horizon (in fossils) List of material examined Illustrations are in most instances vastly superior to Anything that can be made clearly and sufficiently' VISible m a picture should be illustrated. The value of illu:;trations is: recogn~zed i~ the I~terna~ional Rules, since a scientific name given to . a published illustrat10n (prior to Jan. 1, 1931) is valid even if not accompanied by a single word of description. Such a naming of illustrations· was quite custo~ary in the days of Linnaeus. In our day, however, sound taxonomists always present a diagnosis and full description Illustrations.

a_v_erba~ descr~ption.

PRESENTATION OF THE FINDINGS

161

together with the illustrations. See Chap. 9 for a discussion of illustrations. Redescriptions. The redescription of hitherto poorly described forms is an extremely important element of revisional and taxonomic work. In the present state of our knowledge of many animal groups, it is of greater importance than the description of new forms. Ferris (1928), in commenting on this phase of systematic entomology, has stated that ... a distressingly large percentage of the named species, in almost every group of the insects, cannot be recognized positively or even at all, on the basis of the existing literature. It is more important, for the advancement of our study, to redescribe such forms than it is to describe new species. The redescription of such forms should be regarded by the student as an essential part of his work upon any group which he may elect to study. The fact that a species has been named should make no essential difference in the way in which it is treated. . . . The proper aim is not to name species but to know them. The writer who contributes to the genuine knowledge of species is accomplishing far more than one who merely names them. The fact that the author's name accompanies the names of the new species which he describes should not be allowed to influence his activities.

With this view the authors heartily concur. On the other hand, if a good description is readily available in the literature, it is wasteful to publish copies of it again and again. The specimen or specimens on which a redescription or illustration are based should be clearly indicated (the term plesiotype may be used) because, in the event that the species has been misidentified, a new species may be propoi;i.yd for X-iis albus Jones, not Smith. In such a case the type specimen of the new species is the specimen, or is selected from the specimens, on which the redeBcription or illustration was based. Summary. Recommendations on the preparation of descriptions may be summarized as follows: 1. The taxonomic characters should be treated in a standardized sequence. 2. The most easily visible characters should be featured. 3. A direct diagnostic comparison with the nearest relative or relatives Hhould supplement the description. 4. Since words alone can seldom give an adequate picture of the diagnostic characters of a form, appropriate illustrations should be provided whenever possible. 5. The description should provide quantitative data, supplemented with information on geographical range, ecology, habits, and similar data. 6. Species in poorly known genera should be very fully described. 7. The formal description should be followed by an informal discussion of the variable characters.

182 8. type 9. gory

TAXO;\fOMIC PROCEDURE

The. description should be accompanied by full information on th specimen (see Chap. 12) and other material before the author. Character13 that are common to all members of the next higher cateshould be omitted from the description. KEYS 1

The object of keys is to separate and segregate characters in such a way as to provide, by means of a series of alternative choices, a safe road to. identification. The ultimate goal of a key is the identification of taxa: (species, genera, etc.). The procedure involved is somewhat analogous: to that of the physician who, by means of a series of questions and exami.:' nations, arrives, by a process of elimination and confirmation, at the, diagnosis of the ills of a patient, or to the elimination method in culture'· identification of bacteria. Keys are also a tool for taxonomic analysis, since in their preparation one must select, evaluate, and arrange taxonomic characters. In this sense keys are an integral part of taxonomic procedure, as well as a means' of presenting findings. The construction of keys is a laborious and time-consuming task '. involving the selection and sifting of the most useful and most clear!;::. diagnostic characters. Ideal key characters apply equally to all indi-). viduals of the population (regardless of age and sex); are absolute (two4 scutellar bristles vs. one scutellar bristle); are external, so that they ca . be observed directly and without special equipment; and are relatively con-':· stant (without excessive individual variation). Unsuitable key charac-i ters include those that require a knowledge of all ages and stages of a': species (e.g., "sexual dimorphism present" vs. "sexual dimorphism:\ absent"; "male larger than female" vs. "male smaller than female"; "fall molt complete" vs. "fall molt partial"; etc.), relative characters without absolute standard (e.g., "darker" vs. "lighter," "larger" vs.' "smaller,'' etc.), and overlapping characters ("larger, wing 152 to 162". vs. "smaller, wing 148 to 158"). In most cases the data will permit the, choice of several characters for the various primary and secondary divisions : of the key. It is here that the writer is called upon to exercise his best · judgment in order to select the most satisfactory characters at the various · levels. Frequently he is torn between a phylogenetic and utilitarian approach to the problem. The primary purpose of a key is utilitarian; diagrams, lists, numbers, or order of subsequent treatment will take care of phylogeny. However, when making a key in a poorly known group (with many undescribed species) it is useful to arrange the key in such a manner that closely related species key out near one another. This facilitates the subsequent insertion of new species, as well as the decision as to whether or not a species is new. The worker is indeed fortunate ~-----

-- --

163

PRESENTATION OF THE FINDINGS

whose material will permit the construction of a key which will permit presentation of a phylogenetic arrangement without interfering with the main function, that of ensuring identification. A good key is strictly dichotomous, not offering more tha:a two alternatives at any point. 1 Alternatives should be precise. Ideally the statements should be sufficiently definite to permit identification of a single o;pecimen without reference to other species. In any event, identification o;hould be possible without reference to the opposite sex or to immature stages. These should be treated in different keys when dimorphism is exhibited. Ordinarily new species should not be designated as such in a key. Also, it is usually customary to omit authorities from specific names in keys unless these are not mentioned elsewhere in the article. The style of keys is telegraphic, like that of descriptions, and the phrases are usually separated by semicolons. Even though the primary contrasting characters of each couplet may be diagnostic and definitive, o;upplemental characters are desirable in the event that the primary eharacter may not be clearly discerned or the specimen may be injured or mounted in an unsatisfactory manner. One of the most satisfaetory methods for assembling data for the construction of a key is shown in an example of the method and the subsequent analysis given in Table 13. TABLE 1:3. Name of species

ARRANGEMEN'l'

Wings Antennae

Antenna! color

OF

KEY CHARACTERS*

Eyes

Tarsal segments

Leg color

s Ill ithi . . . . . . clear

filiform

black

entire

linear

black

completa . ... opaque

serrate

black

entire

linear

black

e111arginata .. opaque

serrate

black

emarginate

linear

black

rufipes . ..... opaque

filiform

black

entire

linear

red

nigripes . .... opaque

iiliform

black

entire

linear

black

,.'lavicornis . .• clear

filiform

yellow

entire

biloberl

'

black -·---------

rii.ficornis . .. clear

filiform

red

entire

linear

-·~------------------------·-----·-----·

rallfornica. . clear

filiform

black

t>ntirP

bilobed

black

-----black

*Characters used in examples are italicized. 1 If it is impossible to work out a key that permits the identification of all species, it is advisable to indicate this clearly and to key out !IR groups any species that cannot

he diagnosed hy key characters.

165

TAXONOMIC PROCEDURE

PRESENTATION OF THE PINDINGS

This example is oversimplified in order to demonstrate the method mo · clearly. Several types of key are used in taxonomic papers, but those mos frequently used fall into two classes, each of which is subject to consider able modification, although all are dichotomous and based on a series choices. One of these is typified by the dichotomous bracket key. Th·· other is the indented key. The latter type of key has the advantage tha. the relationship of the various divisions is apparent to the eye. It the disadvantages, especially in a long key, that the alternatives may widely separated and that it is wasteful of space. For these reasons th' best uses of this type are for short keys, keys to higher categories, or com parative keys (keys which not only serve the purposes of identificatio but also treat the same comparative characters at each level for eac ' group). An indented key based on the hypothetical data given in Tabl' 13 might be as follows: ·

-l (2). Legs red .............................•..••••••••.•.............. runpes

164

A. Wings opaque B. Antennae serrate C. Eyes entire .................................... . CC. Eyes emarginate ................................ . BB. Antennae filiform C. Legs red .................................... . CC. Legs black .......................... . AA. Wings clear B. Tarsal segments linear C. Antennae black .. CC. Antennae red ...... . BB. Tarsal segments bilobed C. Antennae black .... . CC. Antennae yellow ............................... .

The second type of key, and the one in most common use today, is the! bracket key. This key has the advantages that the couplets are com-~ posed of alternatives which are side by side for ready comparison, and} that it is more economical of space because it is unindented. When" properly constructed it may be run forward or backward with equal\ facility by following the numbers, which indicate the path that the· various choices follow. This is the type which best fulfills the diagnostic purpose of a key. Its main disadvantage is that the relationship of the divisions is not apparent to the eye. An example based on the same data previously used is as follows: 1.

Wings opaque ................. .' ..................................... 2 Wings clear ........................................................• 5 2 (1). Antennae serrate ..................................................... 3 Antennae filiform .................................................... .4 3 (2). Eyes entire ................................................... completa Eyes emarginate. . . . ........................................ emarginata

Legs black .................. , ........••.•.•.••.••.•........... nigripes 5 (1). Tarsal segments linear .....................................•....•..... 6 Tarsal segments bilobed .............................................. 7 6 (5). Antennae black. . . . . ............................................ smithi Antennae red. . . . . .............................. ruficornis 7 (5). Antennae black........ . ........................... . californica Antennae yellow. .flavicornis

A third type of key (serial key) combines certain features of the bracket key and the indented key. It shares with the indented key the advantage that the species are arranged according to the criterion of number of key characters in common, but it is more saving of space and therefore more satisfactory for long keys. Its main disadvantage is that the alternatives are widely separated. Thus in the choice of a key for a particular purpose, the advantages and disadvantages must be weighed against the objectives of the moment. No key can serve all purposes simultaneously. The following is an example of the third type of key: (8). 2 (5). 3 (4). 4 (3). 5 (2). 6 (7). 7 (6). 8 (1). \) (12). 10 (11). I 1 (10). 12 (9). 13 (14). 14 (13).

Wings opaque Antennae serrate Eyes entire ................................................ . completa Eyes emarginate .......................................... emarginata Antennae filiform . ........................ . rufipes Legs red ................ . ............................ . nigripes Legs black ........... . Wings clear Tarsal segments linear Antennae black ............................................... . smithi Antennae red. . . . . . . .. , . . . . . . . . . . . . . . . . . . . . . . . . ruficornis Tarsal segments bilobed . . californica Antennae black .... . . . flavicornis Antennae yellow ..

Among the keys designed for special purposes may be mentioned pictorial keys, branching keys, box-type keys, and circular keys. The pictorial key is of value for field identification by nonscientists. During the Second World War, for example, malaria crews based their control operations on the results of field identifications of anopheline mosquito larvae (Fig. 29). The fact that critical characters were illustrated as well as described made the keys usable by such persons as medical corpsmen and engineers as well as by entomologists. Pictorial keys have been employed also in field guides to vertebrates and flowering plants. Other types of key have been devised from time to time in an effort Lo convey a mental picture of the interrelationships of a group of organisms. Failure to impart such a picture at a glance is the chief defect of the traditional dichotomous key. This is not a serious defect for the

167

1'AXONOMIC PJWCEDUUE

PHHSENTATION OF THE FIN DIXG!:'j

specialist who is accustomed to using and interpreting keys, but it is al shortcoming from the viewpoint of the nonspecialist. To overcome this · difficulty, three different types of key have b€en devised: the branching type (Fig. 30), the box type (Fig. 32), and the circular type (Fig. 31).

in the groups listed in Table 13, smithi-ruficornis-flav£cornis-cal1Jurnica may be a group that splits off very early from the other four species (completa, emarginata, rufipes, and nigripes), the visible difference between the groups may be ill defined (wings clear vs. opaque). To use such an

166

COMPLETA EMARGINATA

PICTORIAL KEY TO ANOPHELINE LARVAE - OF UNITED STATES

I

I

outerc~ls

sporselycred _...__ _ _, __ outer ond il'l'lel' ciyp:eals sparsely feathered Ol'I

outer ond inner ~Is USUOlly feathered nearly to

~ ~~

(Salt wott1r of Atlonlic ond Gulf Cottstsl

,.

entire

I

d9ntelyr

emorginate

~

outer c1ypeals

outer clypeal1

eyes

eyes

Antennae

M

RUFIPE5

NIGRIPE5

SMITH!

RUFICORNIS

CAUFORNICA

FLAVICORNIS

legs red

legs block

antennae: block

antennae red

antennae

antennae

v

Antennae fil1form

serrate

'

Wings opaque

~

bl~low Tarsal segments bilobed

Tarsal segments I inear

~ Win gs clear

BARBERI

(Lower Rio Groro
(Tree holes of eastem holf US)

Volley oi Texas)

~/

l

innerclypeolsc1ose. oflen crossing

~

Fm, 30. Example of branching key based on analysis of characters in Table 13.

I

PIJNCTIPENNIS

IAIJ:IE~idR°Nl IWestern U.S l

Modified from freeborn and Gerberg by H.Knutson

MAl..,ARlA CONTROL IN WAR AREAS U.S. PUBLIC HEALTH SERVICE

Fm. 29. Pictorial key to larvae of anopheline mosquitoes of the United States (U.S. Public Health Service).

The data of Table 13 are used in each case, so that the types are comparable. Certain keys such as the indented (also the box-type and circular type) are sometimes referred to as phylogenetic keys, Such an appellation makes the silent assumption that the series of dichotomies chosen parallels the phylogenetic history. The taxonomic record in wellknown groups has shown how easy it is to make mistakes in the interpretation of the phylogenetic value of characters. For instance, although

Fm. 31. Example of circular key based on analysis of characters in Table 13.

unreliable character as the very first bracket might lead to many misidentifications. Furthermore, subsequent discovery of additional characters may show that actually the form of the antennae is a more deepseated character than the wings, and this would lead to a different interpretation of the phylogeny. Finally, even the bracket-type key

168

PRESENTATION OF THE FINDINGS

TAXONOMIC PROCEDURE

can be constructed in such a way as to group together the most closel related forms, except that monotypic dichotomies sometimes have to be placed out of order. The Presentation of Phylogeny. Natural classifications are based on;'., phylogeny, although they can represent phylogeny only imperfectlY; (Chap. 3). The great interest in phylogeny goes back to the early, Darwinian days. Darwin (1859) stated that all animals and plants were" derived from common ancestors, but he made no attempt to reconstruc · the genealogy of species and higher categories. It was Haeckel (1866); who made a first attempt at presenting the relationships of all animals phylogenetically. A phylogeny is traditionally represented by a branch-! ~ 0 c ·~

~

"' Ci E

0

0

E

u

w

Eyes entire

Eyes emorginote

Antennae serrate

"'a.

"' a::"

.."'

·~

.c

z""

E

·c:"' 0

";

If>

0

<.> > 0

a::

u

0

;;:

Antennae

Antennae

Legs

Antennae

Antennae

red

black

block

red

Wings opaque

"'c

0

<.>

Legs

Antennae filiform

0

-~ c

black

Tarsal segments linear

yel law

Tarsal segments bi lobed Wings clear

Frn. 32. Example of box-type key based on analysis of characters in Table 13.

ing tree, somewhat as in human genealogies. Ever since the invention . of the phylogenetic tree by Haeckel, it has been customary among taxon- . omists to express phylogenetic conclusions in the form of diagrams ; (Jepsen, 1944). In spite of their numerous shortcomings, such diagrams · are useful summarizations of taxonomic knowledge and provide a pie- ·. torial representation of the author's concept of the evolutionary history . of a group. Often a simple diagram shows more than many pages of · detailed discussion or description. Some of the more useful kinds of ' diagram are discussed below. Phylogenetic Evidence. Before diagramming can be attempted, an interpretation of the probable phylogeny must be reached on the basis of the taxonomic data. It is here that the systematist must muster all his judgment and experience. Because of the subjective nature of the problem, it is difficult to lay down any hard and fast procedures for attaining satisfactory results. As has been remarked by Simpson (1945), "Phylogeny cannot be observed. It is necessarily an inference from observations that bear on it, sometimes rather distantly, and that can usually be interpreted in more than one way."

169

One of the first steps in a phylogenetic study is usually the tabulation of the characters shared by the groups concerned. Not only structural features should be tabulated, but also biological, embryological, p~ysio­ logical, and geographical data to the extent to which they are available. The second step consists in determining which of the tabulated characters are primitive and which specialized. This often requires reference to characters in related groups which fall outside of the study. Reduction (e.g., loss of wings, fewer segments in appendages, etc.) is norma~ly, but not always, an indication of specialization. Na~rowly adaptive characters which restrict or limit the habits of a species or group are usually specializations. . . Since the more primitive species or groups are hkely to retam t~e ~?st primitive characters, it is important to know where the most pruniti:e forms are apt to be found. Here geographical distribution and habits aid greatly. New Zealand and Australia, and to a le~ser degree South America, are great reservoirs of primitiv~ type~. Outsid~ of.these areas primitive groups may be widely but discontmuously d~stnbuted, frequently with highly localized, only distantly rel~t~d: species. When the primitive groups have been located and the primitive characters recognized, a rough approximation of the relative ages of the _groups co~­ cerned is possible. Fossil evidence, when available, may aid greatly m confirming such conclusions. With many animal groups, however, only . limited help is normally available from this source. Phylogenetic reasoning on the basis of degree _of resemblance ~s confused by several natural consequences of evolut10n. The first i~ _convergence due to adaptations to similar en:ir?nmental _cond1ti?~s. Familiar examples are the distantly related but similar-appearmg f~i:iihes of water beetles which possess a common streamlined form; the stnkmgly similar structu;e of the forelegs in manti'ds (Mantodea) and mantispids (Neuroptera); and the superficially similar ect?parasites of vertebrates, . which belong to at least six different or!fers of msects. Second, phylogeny may be obscured by parallelism. The various 8pecies of Drosophila, for example, show similar mutations, such as orange eye. Consequently, orange-eye variants in Drosophila are not mon?phyletic but cut across phylogenetic lines. They are part of the genetic pattern of the group as a whole. The same phenomenon is evide.nt in the white females of'the various species of Golias. Mayr and Vaune (1948) have given examples of such characters in certain bi~ds, and Mic~~ner (1949) in saturniid moths. Michener concluded that m the _Satur.nndae a hind tibial spur has been independently lost at least 1~ times m ?ne subfamily, the epiphyses of the female at least 1~ times m the fannly, and the articulation of the male genital harpes 7 times. He found that reduction of the labial palpi ha8 occurred at least \l times, and of the eyes

170

171

TAXONOMIC PROCEDURE

PRESJmTA'l'ION OF THE FINDINGS

and structures of the head capsule at least 14 times. Finally, front tibial spines have been acquired independently at least 10 times. It is obvious that any phylogenetic scheme utilizing these as primary characters · indicative of close affinity would provide highly erroneous conclusions. Traditional lines of relationship may be confused still further by the 1 shuffling of characters seen in some closely related species. In these . cases it would appear that all possible combinations of a given set of characters have been tried and are preserved to confuse the evolutionary •· picture. This seems to be true in various genera of bees. ' Still another source of confusion is the reversal of evolutionary trends. It happens not infrequently in evolutionary lines that a specialization is ·•

has evolved at varying speeds at various periods in its history and in various parts of the world. The reasons for this are discussed by Simpson (1944). Suffice it to say here that the possibility of relic forms must always be kept in mind, especially when phylogenetic reasoning is based solely on the present-day fauna. Phylogenetic Diagrams. Phylogenetic diagrams are symbols designed to represent an author's interpretation of the evolutionary history of a group.

B

A

.,

Ancestral Stoel<.

E

,_

s

Fw. 33. Three-dimensional phylogenetic diagram (after Larn, 1936).

lost and that the descendants of specialized forms become secondarily "primitive." It is sometimes very difficult to distinguish between secondarily "primitive" forms and really primitive ones, unless there is supplementary evidence from the fossil record. On the subspecies level, the distribution pattern sometimes yields valuable clues. For instance, the rosy finches (Leucosticte tephrocotis) are descendants of the Asiatic L. brandti-arctoa group and entered North America via Bering Straits. The southernmost American form (L. tephrocotis australis, southern Rocky Mountains) has lost much of the bright coloration and sexual dimorphism of its species and has become secondarily similar to its primitive relatives in Central Asia.' Similar secondarily "primitive" conditions of peripheral subspecies have been described in many species of birds (e.g., in the genera Pachycephala, Lalage, and Junco). The preservation of annectent types leads us to another kind of phylogenetic problem. In phylogeny the survival of relic types is particularly difficult to evaluate. Here we are dealing with the problem of differential rates of evolution. It appears that each group of organisms

1

p a c

e

b a Fm. 34. (a) Diagram illustrating how geographical fragmentation of successive populations (the numbered rectangles) may accompany vertical differentiation of a phylctic line. The populations rarely remain in one locality for long but migrate. 8ome migrants become isolated from the parent stock by barriers, becoming ultimately differentiated into geographical races. The fauna! succession in any locality (A or B) is never absolutely continuous, even though gaps may be obscure. The gaps may be produced by migrations, by depositional hiatus, and by local extermination. (b) A population becomes divided by a barrier causing partial isolation with limited gene flow for a time-the subspecies stage in speciation. After sufficient genetic differentiation has been reached, interbreeding ceases, gene flow is stopped, and the two branches become separate species (Newell, 1947).

The phylogeny of a group may be depicted as a three-dimensional figure (Fig. 33), with time as the ordinate, differentiation as the abscissae, and the angles (slopes of ascending branches from the vertical) indicating rate of divergence or velocity. Thus A represents an ancestral population which diverged from the vertical axis A-A' into the branches B and G. Line B diverged more rapidly (angle B-A-A') and hence shows greater differentiation at the present time (top level). Line C, although less differentiated from the ancestral type, split again into D and E. The actual split of a phyletic line is called speciation and is visualized as in Fig. 34b (Newell, 1947). Here "a population becomes divided by a

172

TAXONOMIC PROCEDURE

barrier causing partial isolation with limited gene flow for a time-th subspecies state in speciation. After sufficient genetic differentiation h been reached interbreeding ceases, gene fl.ow is stopped, and the tw '· branches become separate species." Actually the course of evolution is more involved than this, because subspecies are not always separated by barriers, and horizontal differen; tiation and vertical differentiation are simultaneous processes. Thus· f, Fig. 34a illustrates (Newell, 1947) . . . how geographic fragmentation of successive populations (the numbe rectangles) may accompany vertical differentiation of a phyletic line. The pop lations rarely remain in one locality for long, but migrate. Some migran become isolated from the parent stock by barriers, becoming ultimately diffe entiated into geographic races. The faunal succession in any locality (A or , is never absolutely continuous, even though gaps may be obscure. Th gaps may be produced by migrations, by depositional hiatus, and by loc extermination.

PRESEN1'A'l'JON OF 1'HE FINDINGS

173

. .

:3 0

~

0

. ~

-

a.

2! 0

-'

.. . .. c

:0

0

~

:::;:

:::;:

The foregoing diagrams are theoretical and therefore relatively simple; However, the application of the phylogentic concept to the classificatio of a group of organisms is not a simple matter. One of the best exampl of such application is the phylogeny of the Equidae (Fig. 35). Here th. fossil record is more complete than in most groups, so the pictorial di gram is based on actual specimens at numerous points along each of th · evolutionary lines. Unfortunately in most groups the fossil record is so incomplete thai. · most phylogenetic diagrams cannot be based on historical data at a . In the absence of adequate paleontological evidence, it is necessary t .'. resort to degree of differentiation and to geographical distribution o present-day forms as criteria for phylogenetic classification. In othe•. words, we can arrive at a phylogenetic picture only indirectly fro '!. what amounts to an aerial view of the phylogenetic tree, the position o~ the main limbs or branches being inferred from the arrangement of th. terminal twigs. This results, even under the most favorable circum.i stances, in only a rough approximation of the actual course of evolution;: in a group, because the relative rates of evolution are unknown, conver-:~ gence of various lines is usually obscured, and extinct lines are lost. ' Horizontal classification differs from vertical classification by the empha-. sis placed on direct descent: "Horizontal classification separates ancestral.' from descendent groups and unites contemporaneous groups, or those in'., a similar stage of evolution, if they are derived from a common ancestry. Vertical classification unites ancestral and descendent groups and sepa--; rates contemporaneous groups that are diverging from a common ancestry" (Simpson, 1945). The relationships of these two types of classifica-

<:"'

... 0

..

2!

.s

c

"" .,,.!! 0 .,, CJ>

·-

~

0

"'

<:

s

2!

.s

FIG. 35. Phylog1my of the Equidae after Stirton (original).

tion are shown in the accompanying diagrams (Figs. 36 and 37). In Fig. 36, which is deliberately drawn so as to be comparable to Fig. 33, group B is a large and diverse taxon. It is connected to the smaller group D by two attributes which, on the basis of experience, are con' to' be fundamental or significant characters (i.e., characters wh'ic h , sidered

175

'l'AXONOMIC PROCEDURE

PRESENTATION OF THE FINDINGS

when used in a classification, make possible a large number of inferences)~, E is connected to B by one character and to D by four characters. Th ' resulting diagram indicates, on the level or place of the present day, (1) size and diversity of the groups and (2) degree of differentiation of the groups. If the same data are projected; on a three-dimensional figure (Figi; 37) we arrive at a purely hyp .· thetical but nevertheless usef ' phylogenetic picture. The chie' advantages of the tree-like repre ~ sentation over the simple line dia", FIG. 36. Diagram of horizontal classification (modified after Lam, 1936). gram are that relative sizes ofi groups can be indicated, and the, perspective aids in comprehending the third dimension. As mentioned in the discussion of categories and concepts (Chap. 3),: no design has yet been devised which will fully reflect all the data and conclusions regarding the history of any moderately complex and reason .. ably large group. At best, with a relatively complete fossil record, e.g., the phylogeny of the horse, it is possible to give only a rough approxima-; tion of the probable course of evolution. ·

may or may not be used to indicate the supposed rate of evolution as compared with another angle on the same diagram (Osborn, 1895) (Fig. 38). This type of diagram is applicable to any level in the taxonomic hierarchy. It is usually schematic, i.e., designed to fit the dimensions of a page rather than to indicate spatial interrelationships of the groups.

174

/ /

\

\

\

\

I

I

I

\ \

I

\

\

I

\

\ I

\ \

\ \

I

I I

\ \I

\

)

I \

~--------1

/

I

/ \

\

I

/

I \

/

I

\

\

///

I

\

\

I I I

F31DE

,/

~\

I I I I I I I

L ________ _

A ,______ ;I

I

FIG. 37. Diagram of vertical classification (modified after Lam, 1936).

By far the majority of phylogenetic diagrams are made without any fossil evidence at all. This involves the basic but unproved assumption ; that degree of resemblance of recent organisms is a direct function of ' their respective ages, the most divergent forms having branched off earlier than more closely related groups. Based upon this axiom, various types of phylogenetic diagrams have been prepared and are in •. current use. The simplest of these is a two-dimensional diagram with a hypothetical ancestor and branches indicating the supposed points of ' divergence. The degree of departure from the vertical, i.e., the angle,

FIG. 39. Diagram showing relationships and origin of the Hemiptera (China, 1933).

Dendrograms. A tree-like modification, known as a "dendrogram," embodies the same principles but is slightly more adaptable to large groups. In this type of diagram great liberties are taken in bending "branches" \o fit the space on a page, and angles are of no importance; but point of origin of branches is very important, and the most closely

17ti

TAXONOMIC PROCEDURE

related forms are in proximity to one another (China, 1933) (Fig. 39) An example of a clear three-dimensional dendrogram is the representatio of the evolution of the caddisworm case cunstruction (Trichoptera) by1 Milne and Milne (1939) (Fig. 40). Sericastoma1idae

~---=~~--DIFFERENTIATION-----""'--'

(Morphologie>I similarity)

FIG. 40. A three-dimensional dendrogram representing the evolution of caddisworm case construction (Milne and .Milne, 1939).

The actual steps in preparing a phylogenetic diagram or dendrogram are difficult to pin down, because so much depends on the breadth of vision and judgment of the taxonomist. Lam (1936) has analyzed the various · types of diagram and has suggested a method by which characters can be tabulated for each species and then used quantitatively fat arrange the branches of the phylogenetic tree. This is a laudable attempt to reduce L

PRESENTATION OF THE FINDINGS

177

the procedure of making phylogenetic diagrams to certain mechanical steps, and as a matter of fact, most taxonomists consciously or unconsciously go through these steps. However, the suggested procedure imparts a false sense of precision to a method which is very largely subjective, and which, in the last analysis, depends not on the number or degree of affinities but on the relative importance to be attached to the various resemblances and differences. Here judgment reigns supreme. Consequently no phylogenetic diagram is better than the taxonomic concepts on which it is based. Linear Arrangement. Phylogeny can best be expressed in three dimen:;ions. However, degrees of relationship can be expressed in two dimenKions, and, as stated above, this is often done. The most universally used expression of relationships is the linear arrangement. Because of our system of printing and binding books in a ;;equence of pages, some order of treatment is necessary. One specie:o lrns to be treated first and one last. In common practice, species are arranged as nearly as possible according to degree of relationship with one another. Thus two species appearing on the same page are presumably more closely related to each other than either is to any other species. This scheme is of some value, presuming that the group is reasonably well known and that the classification includes all the species known and not just a restricted fauna. However, it should be remembered that the forms in a given group can almost never be arranged really "naturally" in a single linear sequence, and that in reality no arrangement of contemporaneous forms is directly phylogenetic. But what of the over-all sequence? Here it is necessary to understand enough of the phylogeny of the group to decide which characters arc "primitive" or generalized and which are specialized. Then the sequence may proceed in either direction, though commonly from generalized to specialized. Since, as shown above, the actual phylogenetic history of animal groups is unknown, most linear arrangements of species, genera, and higher categories in catalogues and check lists are very dubious indeed. Some cataloguers have rebelled against the whole system and utilize a strict alphabetical arrangement. This may be more intellectually honest but is less satisfying to the student of evolution, let alone the curators who wish to arrange their collections in as nearly a natural order as our knowledge will permit. A phylogenetic arrangement has numerous obvious advantages. It calls attention to possible synonyms; it prepares the way for the combining of allopatric forms into polytypic species; it often points to zoogeographical conclusions. By listing primitive species first and the more specialized ones later, it permits the working out of evolutionary trends.

l

PREPARATION OF TAXONOMIC PAPERS

CHAPTER9 PREPARATION OF TAXONOMIC PAPERS No taxonomic study is complete until the results have been published .. · Each worker owes to his science a report on the ground that he has cov-~ ered. The fact that his predecessors have published the results of their: investigations has made his own studies possible. Further, the field of: systematic zoology is so great and the workers so few that significant·. progress will be impossible unless each systematist contributes his bit. This is, however, no justification for the publication of hasty or slipshod·:. work or for flooding the literature with trivia. Even when the worker has accomplished something worth while, he · may fail in his duty to science by inadequate attention to the details of preparing his report for publication. The following suggestions, by no . means exhaustive, are designed to provide a guide to some of the more important problems encountered in connection with the preparation of taxonomic articles for publication. The serious student will wish to 1 pursue this subject further and consult authoritative sources (Trelease, , 1951; Hurt, 1949; also A Manual of Style of the University of Chicago· Press, the U.S. Government Printing Office Style Manual, and .various •· style manuals or pamphlets designed for specialized fields or particular journals). There are various kinds of taxonomic paper. Descriptions of new subspecies, species, and genera are forms of taxonomic presentation which may be published separately in short papers but are most useful when incorporated into larger, more comprehensive studies. Except in wellknown groups, the isolated description, divorced from revisional or :' monographic work: is the least important form of taxonomic contribution and often a handicap rather than a help to subsequent workers. In the lesser known groups isolated descriptions are justified when names are needed in connection with biological or economic work, or in faunal works, or when a group has been recently revised and the new species can be readily fitted into the classification. Theoretically, except in the last case, in order to provide an adequate description the author must under- ' take much of the work basic to the preparation of a synopsis or revision .'. anyway and with a little extra effort can carry it to that point. All too · frpquently, however, the isolated description results from !Jilt a cursory familiarity with previous work, and a greater percentage of synonyms is created through isolated descriptions than through revisional works. l

1 l

178

179

Size and Number of Publications. Some authors apparently pride themselves on the number of titles in their bibliography. This leads to the publication of each new species or subspecies description as a separate article. As previously mentioned, there are times when the publication of separate descriptions is justified. As a general rule, however, material that belongs together should be published together, and in the final analysis the author is judged not by the number but by the quality of his publications. Other authors go to the opposite extreme and include the most heterogeneous material in a single publication. This is particularly apt to be true for taxonomic monographs. It is a fact that taxonomic monographs are rarely read by general biologists; in fact, many of them are read by only a few specialists. If the author of a taxonomic revision has made interesting ecological, evolutionary, or zoogeographical discoveries, he should not conceal these findings in the introduction of the monograph (where they may remain forever buried) but should publish them in a general journal like Ecology or Evolution where the articles will, in turn, draw attention to the monograph. Many highly interesting and significant findings of the taxonomist will be overlooked by the general biologist if they are not brought to his attention by publication in a suitable journal. Among the more comprehensive types of taxonomic publication, the following broad classes may be recognized. Many published works will be found which combine features of more than one of these classes. Synopses and Reviews. Synopses and reviews are brief summaries of current knowledge of a group, and the inclusion of new material or new interpretations is not necessarily implied. They serve the utilitarian purpose of bringing scattered information on a group together in one place, perhaps as a basis for some future revisional or monographic study. Examples of taxonomic synopses and reviews are as follows: La Rivers, Ira. 1947. A synopsis of the genus Ji~nrlrodes (Coleoptera: Tenebrionidae). Ann. Ent. Soc. Amer., 40 :318-328. Ross, H. H. 1946. A review of the :\earctic Lepidostomatidae (Trichoptera). Ann. Ent. Soc. Amer., 39 :265-291, 37 figs.

Revisions. Revisions are presentations of new material or new interpretations, integrated with previous knowledge through summary and reevaluation. They vary greatly in completeness of treatment. Some revisions are monographic in approach but fall short of a monograph because of inadequate material. Others are limited to a new arrangement of a group. Most of the important current taxonomic contributions in groups where new species are still constantly being discovered fall in this category. Such revisions may deal with a whole family (or part of one), with a genus, or with a species group. Generic revisions,

180

TAX.UNOAIIC PROCEDL'Hl!-'

illustrated by the following example, are the most common type of sue work. Sommerman, K. M. 1946. A revision of the genus Lachesilla north of Mexicot (Corrodentia: Caeciliidae). Ann. Ent. Soc. Amer., 39:627-657, 4 plates.

Monographs. Monographs are complete systematic publications; They involve full systematic treatment of all species, subspecies, an other taxonomic units and a thorough knowledge on the author's part the comparative anatomy of the group, the biology of the species an, subspecies included, the immature stages in groups exhibiting metamo ' phosis, and detailed distributional data. For the student of evolutio such monographic treatises are the most rewarding type of taxonomi' publication. They permit a detailed treatment of geographic variation of relationships, and of distributional history. Generalizations on t · structure of species, modes of speeiation, nature of taxonomic categories'. and the like are based on such monographs. They have the disadvantag · that they require more complete material than other kinds of taxonomi , papers. However, with the growth of the collections in the museums of the world, it is becoming more frequently possible to prepare monographs:·" Some papers which fully qualify as monographs are published as revision&i or under some other title. Unfortunately, in the present state of our,1 knowledge of many groups, especially among the invertebrates, few taxer:. nomic papers can justify the title, monograph. :VIonographs are more~ frequently possible among vertebrates. Two fairly typical examples of~ monographs are Felt, E. P., and L. H. Joutel. 1904. Monograph of the genus Saperda. N.Y. Sta ' Mus. Bul. 74 (Ent. 20), 81 pp., 14 plates. Hubbell, T. H. 1936. A monographic revision of the genus Ceuthophilus (Ortho~: tera, Gryllacrididae, Rhaphidophorin:w). Fla. Univ. Pub., Biol. Sci. Ser., Vol~i II, No. 1, 551 pp., 38 plates. -

Faunal Works. The fauna! work ii'i a method of presenting taxonomi~) material defined by a geographical area rather than by a taxonomic unit:; Its objects are to make possible identifications in a particular area and ta' report detailed geographical distribution, rather than to clarify problems. in general systematics. If the fauna involved is that of a very limited'. region, the report may consist of a "local list." Such a list is often the work of a local resident who has collected the area intensively. It can be, exceedingly useful if based on adequate and carefully identified collections ' and if accompanied by quantitative data and ecological comments. The reports of expeditions and voyages also belong to the category of faunistie: papers. They offer a convenient opportunity to describe new species and genera an
PREPARATION OP TAX.U.VUJllC F,lf't:HS

181

In order to prepare a faunal work in most of the lesser known groups, t.he taxonomist usually finds it necessary to delve deeply into problems of classification and general systematics. Thus most faunal works of this type make contributions to systematic zoology over and above their immediate objectives. However, although faunistic papers maybe amine of information for the biogeographer and the ecologist, they are usually not designed to provide data for the evolutionist; and in better known groups the preparation of fauna! lists rarely permits the accurate determination of subspecies. Examples of fauna! works are Fauna of British India. Taylor and Francis, London. Many volumes rovering most groups of animals, published over the past half-century. Biologia Centrali-Americana, 1879-1915, Parts 1-215. Dulau and Co., London. Faune de France. 1921-1950 et seq., Vols. 1-53. Office Centrale de Faunistique Paris. Fnune de l'U.R.S.S. Institut Zoologique de l'Acadcmie de U.R.S.S. (some 30 volumes published).

An example of a local list is Brown, H. E. 1939. An annotated list of the species of Jassinae known to occur in Indiana (Homoptera, Cicadellirlae). Arner. Midland Nat., 21 :663-673.

Atlases. In recent times the need has been felt for complete illustrations of the species of various taxonomic groups. This is a reflection of the inadequacy of the printed word as a means of conveying a mental picture of the general facies of an animal. The idea of an atlas grew also ont of the need for taxonomic data which are strictly comparable from one species to another. Since the purpose of an atlas is purely taxonomic, Hemidiagrammatic drawings are commonly used, though full halftones or colored plates have been employed when dealing with 8uch groups as butterflies and birds. Examples of thi1:1 type of treatment are as follow1:1: i'erris, G. F. 1937-1950. Atlas of the seale insects of North America. Stanford University Press, Stanford University, Calif., 5 vols. Ho;;8, E. S., and H. R. Roberts. 1943. Mosquito atlas. American :Entomologieal Society, Philadelphia, pt. I, 44 pp., pt. II, 44 pp.

Handbooks and Manuals. Certain works, although taxonomic, are designed primarily or exclusively for field identification. In such cases new species are expressly excluded, and emphasis is placed on clear-cut key characters or recognition characters. Examples of this type of publication are :\'ecdham, J. G. 1929. A handbook of tlw dragonflies of North Americn. C Thomas, Publisher, Springfii>ld, Tll.

Charli>s

L

182

TAXONOMIC PROCEDURE

PREPARATION OF 'l'AXONOM!C PAPERS

Hoffman, R. 1927. Birds of the Pacific states containing brief biographies a/' descriptions of about 400 species with especial reference to their appearance : ' the field. Houghton Mifilin Company, Boston. Mayr, E. 1945. Birds of the Southwest Pacific. The Macmillan Company, N . York, 316 pp. Bond, J. 1947. Field guide to birds of the West Indies. The ~facmillan Compa New York, 257 pp. Klots, A. n. 1951. A field guide to the butterflies of North America east of Great Plains. Houghton Mifilin Company, Boston, 349 pp.

Catalogues and Check Lists. Catalogues and check lists, althou designed for very different purposes from the above-mentioned types taxonomic publication, are among the most useful aids to the taxonom· Catalogues are essentially indexes to taxonomic papers, arranged in s{i a manner as to provide a complete series of references for both zoologi and nomenclatural purposes, according to taxonomic categories. Th · preparation is a highly technical task requiring infinite patience, metic.· lous care, and an intimate knowledge of bibliographical sources a ; methods. Check lists, on the other hand, are designed to provide skeleton classification of a group and a convenient and quick method f reference and arrangement of collections. They frequently cont · little more than a list of valid names and synonyms, with a broad indic tion of the geographical area occupied by the species included. Chee lists complement, but are not substitutes for, catalogues. They are mo' useful in the better known animal groups. The style of check lis depends on the group. In ornithology usually a complete reference tot valid names and synonyms is given, together with a detailed descripti of the range. Typical examples of each of these are 1

Van Duzee, E. P. 1917. Catalogue of the Hemiptera of America north of Mexic · Calif. Univ. Pub. Ent., no. 2, XIV + 902 pp. McDunnough, J. 1938-1939. Cheek list of the Lcpidoptcra of Canada and t United States of America. Part I. Macrolepidoptera. Part I[. Mier'' lepidoptera. South. Calif. Acad. Sci. Mein., 1, 1-275; 2(1), 1-171. Mayr, E. 1941. List of New Guinea birds. American Museum of Natural HistoJ New York, 260 pp. FORM OF THE TAXONOMIC ARTICLE

Title. The title is the first part of the paper encountered by the reader although it is often the last item to be added in the preparation of th paper. Its bibliographical prominence and significance warrant muc care in its selection. The title should be long enough to be specific as t ·. the contents of the paper but brief enough for easy indexing. Sho words are preferable to polysyllabic terms. The most important nouns: should be near the beginning of each series of words. The title should

183

contain key words which in indexing will classify the article. Punctuation should be avoided unless essential to meaning. Among the essential elements of a title are (1) a clear indication of the field involved (taxonomy, mor~ho~og:-~~' ecology, etc.), (2) the scientific name of the category treated, (3) md1cat10ns of the order and family either by means of scientific names (which may be in parentheses) or, rarely, by a well-known comm?n name, and (4) the geographical area, fauna, or locality. The followmg are examples of good titles: "A Taxonomic Revision of American Leafhoppern (Homoptera, Cicadellidae)" "A Check List of the Birds of Alabama" "Geographical Variation of Hippodamia con vergens in Southern California (Coleoptera, Coccinellidac)" "Two New Species of Wood Rats (Neotoma) from the Rocky Mountain Region" The foll~wing are a few examples of poor titles for taxonomic papers. ( )n the basis of the above-enumerated principles the objections to these arc obvious. ''Xew Bymenoptera'' ''I\ otes on Mammals'' "The ·western Biota" "A Collecting Trip to Texas" "Additions to the Fauna of Nebraska" "Studies in the Mollusca" "A New Acanthiza" Titles need not be as bad as these, however, to cau1:1e difficulties for cataloguers, abstracters, reviewers, and other bibliographers. No author has cause for complaint of his work being overlooked if it masquerades under an incomplete, ambiguous, or misleading title. Author's Name. The author's name follows the title. Bibliographical problems are simplified if an author always uses the same form of his name. The entomologist Laporte sometimes published under the name Laporte, sometimes under le Comte de Castelnau. The bibliographicai confusion which resulted still persists in modern literature. Women t~tx?nomists who begin publication before marriage frequently avoid similar confusion by continuing to publish under their maiden names or by a system of hyphenation, e.g., Dorothy McKey-Fender. It is customary in America .to omit degrees and titles from the author's name, although these are used m many European journals. The author's address should f?llow his name in order to facilitate correspondence and should be precise so that postal authorities can recognize it.

l

184

TAXONOMIC PROCEDURE

When more than one author is involved, the order of names is det mined by the nature of the contribution each has made. When the wo has been rather equally shared, the problem is solved by a coauthorshi in which case the names are usually arranged in alphabetical ord . When the work has been disproportionately divided or there is a mark . discrepancy in age or experience, a senior and junior authorship ma: result. In such cases the name of the senior author appears first. · Introduction. Every taxonomic paper should include an introducto paragraph stating the scope of the paper and, where pertinent, the reaso / for the study, as well as the nature of the studied material. Frequent\ a brief historical review is appropriate. These features serve to orie · the ca1mal reader and the new student of the group, as well as to refr the minds of other workers in the field. Acknowledgments. Acknowledgments may be included in the intr duction when they can be treated as part of the natural sequence ~ exposition. Some authors place them in a footnote appended to t. author's name. This system is in regular use by the Annals of the E mological Society of America and certain other journals which are pri rily taxonomic in content. Sometimes the acknowledgments precede t, summary. Methods Used and Materials Studied. graphic work it is desirable to include a statement on methods utilized collections, specimens, or other materials studied. This enables t · reader to evaluate conclusions and to judge the thoroughness of work. Standard methods for measuring, mounting, staining, spe ; preparations, etc., may be referred to by name and reference. Only n methods need to be described in detail. Body of the Text. The material comprising the body of the text wi of course, depend on the scope and objectives of the particular pa . It is perhaps sufficient to mention that a complete systematic pa includes (1) a definition of the highest category included (family, tri etc.), (2) a key (or keys) to all intermediate categories treated (genera},,, (3) synonymies and descriptions of the intermediate categories (genera· (4) statement of the generic types, (5) comparisons with other genera,, (6) keys to the species of each genus, (7) synonymies and descriptions each species, and (8) statements as to type localities and to location types, general distribution, hosts and other significant biological dat comparisons with other species, etc. (for details on preparation of descri tions and keys see Chap. 8). Synonymy. In monographs, revisions, and catalogues it is customary and advisable to give the complete synonymy of every species. During; the earlier stages of the development of our taxonomic literature it was customary to give not only the synonyms but also a more or less '

PREPARATION OF TAXONOMIC PAPERS

185

complete list of all references to the species, with the names and combinations used in previous publications. In the better-known groups of animals this is both unnecessary and uneconomical, and this function is reserved for bibliographical catalogues. 1 Unfortunately, some groups of animals (e.g., insects) have been so incompletely catalogued, or existing catalogues are so out-of-date, that bibliographical synonymies are still an essential element of the full taxonomic treatment. This is especially true when much of the literature prior to 1900 is more significant nomenclaturally than zoologically, and the later publications are more significant wologically than nomenclaturally. An understanding of both is required by the modern working taxonomist in such groups. X ew synonymy can most usefully be cited with the following sequence of data: (1) scientific name (in its original form), (2) author, (3) date of publication, (4) reference, (5) type locality, (6) present location of type (optional). For example, Oncideres rhodostictus Bates Oncideres rhodosticta Bates, 1885, Biol. Cent.-Amer., Coleopt., 5 :367. [Lerdo, }{ex.; British Mus. (Nat. Hist.)]. Oncideres trinodatus Casey, 1913, Mem. Coleopt., 4:352. [El Paso, Tex.; U.S. Natl. :\1us.]. New synonymy.2

The above form is sufficient for a revision of a well-catalogued group. In groups where the literature has not been summarized adequately and the nomenclature remains confused, a full synonymy (i.e., a list of scientific names, incorrect and correct, specifying the books and authors employing them) may be required. This should include all referehces which have nomenclatural or zoological-Significance, arranged chronologically under the actual name (correct or incorrect) by which the author actually referred to them. Many authors here use the convenient device of a comma inserted between the specific name and the author [X11s albus, Smith (not Brown)] to distinguish between a misidentification which has no nomenclatural status, and a homonym [X-us albus Jones (not Brown)] which has. A widely used form for a full bibliographical ".Vllonymy is as follows: Oncideres rhodostictus Bates Oncideres rhodosticta Bates, 1885, Biol. Cent.-Amer., Coleopt., 5 :367 [type: Lerdo, .\Iex.; British Mus. (Nat. Hist.)]; Linsley, 1940, .Tour. Econ. Ent., 33:562 (synon., 'For example, Peters, in his Check-list of the Bfrds of the World (1931 et seq.), does not list synonyms that ean be found in the previous standard works: Catalogue of Birds of the British Museum (1873-1892) and the Handlist of Birds (1896-1910). :\lore recent checklists of birds do not repeat synonyms correctly cited by Peters. 2 This synonymy was published as new in the Journal of Economic Entomology, 33 :562, 1940. Its use as an example here and elsewhere in the present dis<'ussion is not to be interpreted as a nomenclatural change.

L_

186

TAXONOMIC PROCEDURE

distr.); Linsley, 1942, Proc. Calif. Acad. Sci., (4) 24:76 (distr.); Dillon and Pill 1945, Sci. Pub. Reading Mus., no. 5: xv (key); Dillon and Dillon, 1946, l.c., 6:313, . (revis.). Oncideres putator, Horn (not Thomson, 1868), 1885, Trans. Amer. Ent. So' 12 :195 (key, distr.); Schaeffer, 1906, Gan. Ent., 38 :19 (key). Oncideres cingulatus, Hamilton (in part) (not Say, 1826), 1896, Trans. Amer. E ' Soc., 23 :141 (distr.). Oncideres trinodatus Casey, 1913, Mem. Goleopt., 4:352, [type: El Paso, Te · U.S. Natl. Mus.]. Oncideres sp., Craighead, 1923, Gan. Dept. Agr. Bul. 17 (n.s.), p. 132 (larva, hos . Oncideres pustulatus, Essig (not Le Conte, 1854), 1926, Insects of Western No America, p. 460, Fig. 368 (habits, distr.). · The above synonymy might appear in an abbreviated check list follows: Oncideres Serville, 1835 So. Caljf. to T 1. rlwdostictus Bates, 1885 trinodatus Casey, 1913 No. Mex. L. Calif. When a check list contains a terminal bibliography, the usefulness the check list may be increased by giving page references which may the be located by author, date, and page in the bibliography. Thus, rhodo '1 tictus Bates, 1885 :367, or, more simply, 85 :367. In a complete synonymy it is often desirable to indicate the vario combinations under which each name has appeared. This may be corii veniently accomplished by taking the oldest specific name and followin. it through its various combinations, then the next oldest, etc., as follo M egacyllene antennata (White) Clytus antennatus White, 1855, Cat. Coleopt. Brit. Mus., 8:252 [type: "W. Coast q

America"; British Mus. (Nat. Hist.)]. Cyllene antennatus, Horn, 1880, 1'rans. Amer. Ent. Soc., 8 :135 (descr., syn., distr.) Craighead, 1923, Can. Dept. Agr., Bul. 27, p. 33 (larva, biol.); Hopping, 1937, An Ent. Soc. Amer., 30: 441, pl. 1 (revis.). Megacyllene antennata, Casey, 1912, Mem. Coleopt., 3:348, 351 (descr.). Arhopalus eurystethus LeConte, 1858, Proc. Acad. Nat. Sci. Phila., 1858:82 [type· Sonora, Mex.; Mus. Comp. Zool., Harvard]; LeConte, 1859, in Thomson, Arcan. · Naturae, p. 127, pl. 13, Fig. 9. In the above example, the comma between the specific combination an the author's name has again been used, this time to distinguish between new combination (Cyllene antennat'Uts, Horn, 1880) and an original com-; bination (Clytus antennatus White, 1855). Generic synonymy is handled in much the same way as a specific: synonymy, except that in the case of new synonymy or full bibliographic: treatment, the generic type (and its designator, if any) is cited in place o( the type locality and type location. The synonymy of the genus Dicru~~ rus, as cited in Vaurie's (1949) revision of the Dicruridae, may be listed as an example.

PREPARATION OF TAXONOMIC PAPERS

187

Genus Dicrurus Vieillot Dicrurus Vieillot, April 14, 1816, Analyse d'une nouvelle ornithologie elementaire, p. 41. Type, by subsequent designation, Corvus balicassius Linnaeus (G. R. Gray, 1841, A list of the genera of birds, ed. 2, p. 47). Edolius Cuvier, Dec. 7, 1816, Le regne animal, vol. 1, p. 350. Type, by subsequent designation, Lanius forficatus Linnaeus (G. R. Gray, 1855, Catalogue of the genera and subgenera of birds, p. 58). Drongo Tickell, 1833, Jour. Asiatic Soc. Bengal, vol. 2, p. 573. Type, by monotypy, Drongo caerulescens Tickell = Lanius caerulescens Linnaeus. Chibia Hodgson, 1836, India Rev., vol. 1, p. 324. Type, by subsequent designation, Edolius barbatus J. E. Gray = Corvus hottentottus Linnaeus (G. R. Gray, 1841, A list of the genera of birds, ed. 2, p. 47). Bhringa Hodgson, 1836, India Rev., vol. 1, p. 325. Type, by original desig11ation and monotypy, Bhringa tectirostris Hodgson. Bhuchanga Hodgson, 1836, India Rev., vol. 1, p. 326. Type, by subsequent designation, Bhuchanga albirictus Hodgson (Sharpe, 1877, Catalogue of birds in the British Museum, vol. 3, p. 245). Chaptia Hodgson, 1836, India Rev., vol. 1, p. 326. Type, by monotypy. Chaptia muscipetoides Hodgson = Dicrurus aeneus Vieillot. Dissemurus Gloger, 1841, Gemeinniltziges Hand- und Hilfsbuch der Naturgeschichte, p. 347. Type, by monotypy, Cuculus paradiseus Linnaeus. Musicus Reichenbach, 1850, Avium systema naturale, pl. 88, fig. 9. Figure of generic details, no species included, cf. Bonaparte, 1854, Compt. Rendmi Acad. Sci. Paris, vol. 38, p. 540. Type, by tautonomy, Dicrurus musicus Vieillot = Corvus adsimilis Bechstein. Dicranostreptus Reichenbach, 1850, Avium systema naturale, pl. 88, •fig. 12. Figure of generic details, no species included. Type, by subsequent designation, Edolius megarhynchus Quoy and Gaimard (G. R. Gray, 1855, Catalogue of the genera and subgenera of birds, p. 58).

Summary. A summary is usually unnecessary in a strictly taxonomic paper. When required, it should be brief but should not be in telegraphic style. It should be written as a series of short paragraphs and should be ~pecific, not in broad general terms. References and Bibliography. References are generally treated either in footnotes, or in parentheses in the text, or in a terminal bibliography. Footnotes are useful when but a few references are involved and when repeated reference to the same bibliographical item is unnecessary. Since they are costly to handle in typesetting and printing, however, and may add materially to publishing costs, parenthetical references are to be preferred. When references are numerous, they are most frequently handled in a terminal bibliography. In most cases this bibliography should be as short as is consistent with its purpose, and the items included should be selected. Frequently the value of the terminal bibliography may be greatly increased by including parenthetical comments on the

188

TAXONOMIC PROCEDURE

nature of the subject matter covered. Unverified references may included when necessary for completeness, but they should be mark with an asterisk or some other device to indicate the fact that the autho~r has not seen them. Bibliographical items should receive full citation, including author,; title, publication, volume, pages, date, etc. Text references to the termi nal bibliography may be made by enclosing an author's name and da (sometimes also page) in parentheses. Two or more references to public tions by a single author in the same year may be designated by append letters (Smith, 1940a; Smith, 1940b). The author-date system of bibli graphical reference is far more satisfactory than the straight numberin i:;ystem which is sometimes used. The number system tells nothin' about the reference; moreover, the author-date system permits t addition of references during the preparation of the manuscript withou the necessity of renumbering all references beyond the point of insertio · A formal "Bibliography" implies completeness of coverage of the su '" ject. "Literature Cited" indicates restriction of references and is sef · explanatory. · giv are bibliographies terminal Examples of footnote entries and below. Numerous exceptions will be encountered, especially various go, ernment documents, but a majority of literature citations will fit into o . or another of the simple styles illustrated. It is becoming standard pra ; tice to list the year of the publication immediately after the author's na since this sequence agrees with that of the author-date system of referen Bibliography Wheeler, William Morton 1889a. Homologues in embryo Hemiptera of the appendages to the first abdo nal segment of other insect embryos. American Naturalist, Vol XXIII, pp. 644-645. 1889b. Uber drtisenartige Gebilde im ersten Abdominalsegment der Hemip enembryonen. Zoologisclier Anzeiger, Band XII, pp. 500-504, 2 figs. 1910. Ants, their structure, development and behavior. New York. Columbia University Press, pp. xxvi + 664, front. 286 figs. 8vo (Columbi University Biological Series, IX). Literature Cited Wheeler, W. M. 1889a. Homologues in embryo Hemiptera of the appendages to· the first abdominal segment of other insect embryos. Amer. Nat., 23:644-645.'· 1889b. Uber drtisenartige Gebilde im ersten Abdominalsegment der Hemipterenembryonen. Zool. Anz., 12 :500-504. 1910. Ants, their structure, development and behavior. Press, New York. xxvi + 664 pp. [Footnote] *Wheeler, W. M.

1889.

Amer. Nat., 23:644-645.

PUEP,WA'l'ION OF 'J'.lXONOlll/C PAPERS

l89

In the typescript, footnotes are entered beneath a marginal line in the text (see exam~le above) rather than at the bottom of the page, because in the final publication, pagination is entirely different from the original pagination. If the work cited is by several authors, only the first need be reversed for alphabetical purposes. Thus, McAte~, W. L., a~d J. R. Malloch. 1922. Changes in names of American Rhyu" ho ta, chiefly Emesmae. Proc. Biol. Soc. Wash., 35 :95-96.

The original style of capitalization and italics may be followed. However, many titles are set entirely in capitals; others are set entirely in low~r ~ase, exc:pt that .the first word and scientific and place names are capitalized. The full title should be given in all but the briefest of footnotes, because readers obtain valuable leads in this way. Abbreviations of journals should follow such standard works as the IV orld List of Scientific Periodicals or the list of Abbreviations used in the Department of Agriculture for Titles of Publications (Whitlock, Carolyn, U.S. De?t. :4-gr. Misc. Pub. 337, pp. 1-278, 1939. Price 30 cents). The citat10n: (Wheeler, 1889a) is specific enough because it refers to a two-page paper. On the other hand, it may be necessary to refer to a particular page of a larger work, thus: (Wheeler, 1910, p. 263). In this case the page is indicated in the citation, and the complete work is listed in the bibliography. PREPARATION OF THE MANUSCRIPT FOR PUBLICATION

Aside from matters involved in the actual organization and constructi~n o~ a taxonomic paper, there are some points which should be kept in mmd. m order to facilitate editorial handling after the paper has been submitte? for ~ublication. Editors are much more apt to accept readily and publish qmckly papers which are in good form and require a minimum amount of editing. Most publications have special form requirements, and much editorial time can be saved by careful advance reference to the journal in which the paper is to be published. Typing. All manuscript submitted for publication should be typed. The original drafts may be on yellow paper, but the final copy should be on standard (8Yz by 11 in. or 8 by 10 Yz in.) white paper, entirely double~paced \some publications require triple spacing), and with a wide margin for addmg proof marks and for editing. If approximately the same number of _lines is typed on each page, the editor can conveniently estimate the size of the final printed paper. However, some editors require that pages end with completed paragraphs. Pages should be numbered consecutively i~ the upper right-hand corner. Inserted pages are numbered alphabetically (e.g., 65a). Whole sheets should be used for inser-

TAXONOMIC PIWCEDURE

190

PREPARATION OF TAXONOMIC PAPERS 1

tiom; regardless of length of inserted matter. When it becomes necessary to c~t and rearrange, sheets should be assembled by pasting, not. by· pinning. All tabular material should be typed on separate sheets, smce. it is usually set in a different type from the text. . . _ Underlining. Underlining indicates that the matenal so marked is t be printed in italics. In a taxonomic manuscript submitted for publica- . tion, underlining should be limited to scientific names of genera. and; species which appear in the text. New names should not _be un~erl~ned 1\ because the editor will usually mark these with a wavy lme to md1cate; boldface. Indications of style or sizes of type for titles, headings, sub ; headings, sideheadings, and the like should b_e left to t?e editor: r, . general, marks which the author makes merely mterfere with. the editor work though marginal notes as to the relative rank of headmgs may b helpful. Legends and Text Citations to Illustrations. Titles and legends should be self-explanatory. The manuscript of these titles should typewritten, double-spaced on separate sheets (several titles on a sin. sheet), and assembled in numerical order at the end of the manuscn following the bibliography. A short identifying title may be plac:d o each plate for purposes of identification, but this title will not be prmte Usually in the process of handling, titles and legends go to the typesette with the rest of the manuscript, whereas illustrations are sent to th. engraver. The printer may never see the original drawings. . .' The place of insertion of the illustrations should be marked m th manuscript and also in the galley proof. Illustrations are usually nu bered starting with each article, but some journals number plates co secutively throughout a volume. In any event, a new series of figu numbers or letters should be used on each plate. Many journals desi nate figures with Arabic numbers, plates with Roman numerals. A figures should be referred to in the text by number. . Revision of the Manuscript. Some few authors have sufficient master _ of the English language so that they can write directly in final form for publication. Other equally competent scientists find it necessary to c, revise page after page not once but many times. T. D. A. Cocker:ll was-_ an example of the former type of writer, whereas, by his own testimony, ; Charles Darwin was an inveterate reviser. Trelease (1951) recommends careful reading of the manuscript 10 times, · each time for one of the following: (1) consistency, (2) sentences, (3) clear-'. ness, (4) repetition, (5) connectives, (6) euphony, (7) pu~ctuation, (8)'· style, (9) accuracy, and (10) length. Authors of taxonomic papers seldom follow the details of this recommendation, but most papers would benefit by more revisions than are usually given. It often helps to put .a _ manuscript aside for a while before the final revision is made. It is '

j

1

191

always advisable to have other persons read a manuscript before it is submitted for publication. A fully corrected carbon copy of the manuscript should be kept by the author for use in case the original is lost. Proofreading. Most scientific journals permit the author to read proof on his papers before publication. Some few journals place the entire burden of the proofreading on the author and hold him responsible for typographical or other errors which may pass undetected. In any event, where the author sees the proof, proofreading becomes a very important part of his scientific responsibility. The scientific value of his paper can be greatly lessened by unfortunate typographical errors. Such errors are sometimes obvious to the reader, but they may be insidious and wholly misleading. In general, the submission of proof to the author is to permit the elimination of errors for which the printer is responsible. Author's errors are his own responsibility, and some publications charge authors for corrections other than printers' errors. Changes in proof are costly and therefore should not be made unless necessary or unless the author is willing to assume the cost of the change. Proofreading cannot always be done satisfactorily by one person. It is advisable to supplement the personal reading by having someone else . read slowly from the original manuscript, while the proofreader (preferably the author) carefully reads the proof. Special attention should be given to punctuation, spelling of scientific names, numbers, and dates of all kinds. When corrections are necessary, they should be made 1 according to the standard system of proofreaders' marks (Fig. 41). Most authors see only galley proofs of their papers. These are long :;heets with the text continuous and not broken into pages. For most journals a galley is the equivalent of about three printed pages. Some publications also submit page proofs to the authors. In such cases proofreading cannot be restricted to individual words which were corrected in the galley proof but must include the whole line in which the word appeared. Modern linotype machines cannot change a single letter in a word but must reset the whole line. If a word was inserted, it may have been necessary to reset several lines or perhaps the remainder of the paragraph. The author should carefully check everything which has heen reset. Corrected proof should be returned at once to the editor or printer in order to avoid delay in publication. The printing of an entire issue of a periodical may be held up by a single tardy author. Illustrations. The object of illustrations in taxonomic papers is to present precise, comparative information which cannot be so well expressed in words or which is needed to elucidate the written text. Thus accuracy, simplicity, and intelligibility are prime considerations. In the preparation of illustrations, advance consideration should be given

Loading...

Methods and Principles of Systematic Zoology - VePub

McGRAW-HILL PUBLICATIONS IN THE ZOOLOGICAL SCIENCES E. J. Boell, METHODS AND PRINCIPLES CONSUL TING EDITOR OF SYSTEMATIC ZOOLOGY Baitsell . HUMAN ...

13MB Sizes 1 Downloads 0 Views

Recommend Documents

Department of Zoology and Fisheries
The Department of Zoology and Fisheries was established in 1994 and continues the long tradition of education and resear

Section 1 - PRINCIPLES AND METHODS OF TRAINING
Dedicated trainers make it an essential part of their continuing education as professionals to locate such publications

Extrapolation Methods - Forecasting Principles
It can be used for short-range or long-range forecasting. The quotation from Cairncross, however, suggests that the erro

Course: ZOOLOGY - University of Mumbai
8. Zoology- S. A. Miller and J. B. Harley, Tata McGraw Hill. 9. Modern Textbook of Zoology, Invertebrates, R. L. Kotpal.

SYSTEMATIC METHODS FOR MEMORY - ScholarBank@NUS
Jan 13, 2016 - Many thanks to all my lab-mates, from whom I have learned broadly in research and daily life: Kailas ....

epidemiologic research principles and quantitative methods magazine
distinctive features in sarnami hindustani bound offprint from phonetica 25 · 4th edition practice test edcor bls cpr fi

Cancer epidemiology: principles and methods - iarc
light, air pollution and occupational exposures; lifestyle variables like diet, smoking and physical exercise; ...... he

Communication Design: Principles, Methods, and Practice
pay attention to human factors that are outside the expertise of computer scientists. Beyond recognizing this phenomenon

Social Science Research: Principles, Methods, and Practices
Descriptive research is directed at making careful observations and detailed documentation of a phenomenon of interest.

Part I Basic Principles and Methods
Analytic Epidemiology. FETPAFI. 6. Descriptive Studies. • Populations (Correlational studies). Case Reports. • Indiv