Pa V. I

PALEO. LIBRARY

[ANNALS N. Y. ACAD. SCI., Vol. XXIX, pp. 145-180, Pls. I-II]

EARTH SCIENCES LIBRARY

ON FAMILIES OF SPIDERS

BY ALEXANDER PETRUNKEVITCH, PH.D.,

PROFESSOR OF ZOOLOGY IN YALE UNIVERSITY

Taxonomy is the mirror of Evolution. In it the relationship of forms is reflected, bringing us face to face with results of past processes, just as Physiology and Genetics reveal to us inner activities of living organisms. It is the supreme effort of the human mind to bring into a natural system knowledge derived from the study of Morphology in its widest sense. Gross and microscopic anatomy, embryology, behavior and instinct, paleontology, geographical and geological distribution-nay, even physiology and chemistry-contribute to the building of the system wherein the breadth and depth of our knowledge find their expression. Only limitation of knowledge makes the system imperfect, only limitation of mind makes some present-day zoologists relegate taxonomy to the dust-heap of accumulated facts not worthy of further examination. As in a poor mirror the image may be distorted, so in a system built up by a taxonomer the relationships may be wrongly represented. The smaller the knowledge, the greater the distortion. Yet the study of taxonomy has this advantage, that even a wrong step is a step forward, for it reveals to us something that until then had escaped our observation, something that may possess special significance. Thus parallelism and convergence in development for a while were mistaken for true relationship, but are now of help in the understanding of directions in evolution.

Long before evolution became the object of discussion, taxonomy was trying to establish relationships of forms. Viewed from a distance of several decades of study, not only the system of Linnæus, but those of Cuvier, of Latreille, and even of Huxley and Haeckel, appear to us crude and unnatural. Yet, if regarded in the light of contemporary knowledge, they embodied the latest discoveries and represented a great advance over what was known previous to them. It is not my object here either to discuss these systems, or to give a system of my own which would embody our present knowledge and cover the entire field of animals. Such a task is now beyond human power, for firsthand knowledge of taxonomy requires so much time to acquire that one can possess it only in the case of a smaller group. In another paper I have touched upon the relationships of the various orders of Arachnida

(145)

M647697

among themselves and with other classes of Arthropoda. Here I intend to discuss the order of spiders, which I have been studying from many points of view for many years past.

Needless to say, all old systems are quite out of date Even such an illustrious student of spiders as Walckenaer did not have a clear conception of families and did not know the value of characters. He was the first, however, to apply to his system characters derived from the study of habits, locomotion, nest-making and snare-construction. This system, somewhat modified, was in vogue for many years afterwards. It found its application even in the system of Thorell and has resulted in the approximation of such different forms as Uloboridae and Epeiridae, Dictynidae and Theridiidae, Amaurobiidae and Agelenidae. An earlier attempt by Latreille to divide spiders into Tetrapneumones and Dipneumones, based on the number of book-lungs, later followed by Bertkau and many other araneologists, led Dahl to the separation of the Hypochilidae from the sub-order to which it naturally belongs, on the ground that spiders of that family have four book-lungs. Many attempts have been made during the past fifty years to create a comprehensive system, and invariably the limits of established groups were obliterated by the widening of our acquaintance with exotic and especially with tropical faunas. Our present system is almost entirely due to the genius and unsurpassable knowledge of Eugène Simon. It is embodied in the second edition of his Natural History of Spiders and represents the results of an analysis of a vast number of species and of patient labor extending over forty years. With slight modifications, it is generally accepted and all attempts at change are directed chiefly against minor points in the system. Yet it is far from perfect. Simon makes the persistent attempt to widen the limits of genera and of families, because he finds so many intermediary forms that all lines between groups become effaced. We see this in the recognition of a single family, Aviculariidae, instead of three or four, as is done by Pocock and others; in the inclusion of the Deinopinae in the Uloboridae; in the fusion of the Linyphiidae with the Epeiridae and Tetragnathidae into one immense family, Argiopidae; in a similar fusion of groups in Thomisidae, and perhaps quite especially in the establishment of a single family, Clubionidae, for such diverse forms as Selenopidae, Ctenidae, Heteropodidae, Clubionidae. We see the attempt to widen the limits of genera in the merging of a number of genera in the single genus Araneus which I myself saw was necessary to treat as a cohors in my Catalogue; in a similar merging of several genera into one genus Dendryphantes, and again in the case of Lycosa. Frederick P. Cam-.

bridge, who did so much during his brief span of life for the elucidation of natural relationships in spiders, made a praiseworthy attempt at the delineation of more limited genera in several instances, but without much success. Such genera or cohortes as Araneus, Dendryphantes, etc., require extensive study and a natural division into genera may be accomplished only when material from all over the world is taken into account. All attempts at classification based on limited faunas are invariably unsuccessful, applicable only to the particular fauna and therefore misleading when it comes to comparisons. Several studies of limited groups were made within recent years by various araneologists, Simon, Cambridge, Dahl, Peckham, Comte de Dalmas, Fage, Chamberlin, Mello-Leitão, but the field is still open to extensive investigations which would be most welcome.

The great difficulty in delimiting families lies in the value which one accords to each character. Simon, for example, entirely omits to mention the taxonomic value of trichobothria, perhaps because he does not attribute any particular value to them, or perhaps because we still lack sufficient comparative knowledge of these sense organs. On the other hand, Dahl goes so far as to base a great deal of his classification on the distribution of trichobothria which he calls "auditory hairs." Following this clue, he separates Simon's Thomisidae into several families, and in fact establishes the new sub-orders Chalinurae, Oligotrichiae, Stichotrichiae, and Polytrichiae, each embracing several of Simon's families, but not in any way coinciding with Simon's sub-orders. I believe that the value of this character is greatly overestimated by Dahl, but I also believe that we are indebted to him for having pointed out to us a character which may be used to advantage in special cases.

And here, then, naturally arises the question, What is a family-character? Can any structure that distinguishes one group of animals from another group be reasonably considered a family-character? Evidently not, for otherwise we should immediately commit the error of uniting convergent forms in one family, though their entire pasts were quite different. The reader may dissent on the ground that the object accomplished, not the process by which that object was accomplished, is of primary importance; that water is water, whether we get it by condensation of vapor or melting of ice, by distillation from a chemical solution. or by the combination of hydrogen and oxygen atoms in the presence of an electric spark. But an organism is not a simple chemical and it carries within it the marks of its origin. Even where such marks are almost entirely obliterated, as in some parasitic Crustacea, embryology reveals traces of them. But should even the last marks be lost, the

species would, nevertheless, be distinct, though we could not grasp the distinction with the aid of our existing methods, and would cease to be a distinct species only when and if it should become absolutely identical with some species of another group. This, of course, to our knowledge

has never happened.

Have, then, the trichobothria such value that the various families composing Simon's division Cribellatae, may be split into four groups and, together with ecribellated spiders, placed under the above-mentioned suborders, as Dahl has done? If we remember that the trichobothria are sense-organs and, moreover, sense-organs of the lowest type, for they are nothing but modified tactile hairs, we must expect to find an inclination in that character to great variations. And so it is and it would be entirely out of harmony with all other sense-organs if trichobothria were permanent or even, generally speaking, dependable characters. Take, for instance, the eyes; they belong to two different somites, two neuromeres of the brain supplying them and them only with nerves. And yet eyes vary in number in closely related species, the variation being sometimes directly traceable to cave-dwelling habits. The auditory funetion of the trichobothria is far from having been demonstrated in a way precluding all controversy; but without going into the discussion, treating them broadly as a type of tactile organs, why should they be exempt from variation when they receive their nerve supply from five different neuromeres and when they share that nerve supply with a number of other organs in the same limb?

If we pause for a moment to think, we will come to the conclusion that in the process of evolution some characters must have been subject to greater variation than others, since otherwise all resemblance between the forms would have been lost at the start. These more stable characters will be present, therefore, in a greater number of forms, and in the sequence of time will have the value of class-characters, order-characters, family-characters, and so on. Examine the mutation-characters observed or produced in breeding experiments during the last decade. Not one of them has a greater value than that of a sub-species or at best of a species. Even generic characters remain unaltered, not to mention family-characters or order-characters. This shows that most of the changes occur, so to say, on the surface, leaving the substratum intact. It follows that if we could find the exact extent of variability for every character of a number of species, we could almost infallibly decide the taxonomic value of each character. Unfortunately, instead of welcoming variability and making it a subject of particularly painstaking study, the systematist usually either disregards it as something of no particular