AMERICAN MUSEUM Novttates PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET NEW YORK, N.Y. 10024 U.S.A. NUMBER 2627 JULY 7, 1977

NORMAN I. PLATNICK The Hypochiloid : A Cladistic Analysis, With Notes on the Atypoidea (Arachnida, Araneae)

AMERICAN MUSEUM Novitates PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, N.Y. 10024 Number 2627, pp. 1-23, figs. 1-31 July 7, 1977

The Hypochiloid Spiders: A Cladistic Analysis, With Notes on the Atypoidea (Arachnida, Araneae)

NORMAN I. PLATNICK1

"The value of any classification rests on the soundness of the princi- ples underlying it." -A. Petrunkevitch, 1933, p. 303 "But I shall certainly admit a system as empirical or scientific only if it is capable of being tested by experience. These considerations suggest that not the verifiability but the falsifiability of a system is to be taken as a criterion of demarcation. In other words: . . . it must be possible for an empirical scientific system to be refuted by experience." -K. R. Popper, 1968, p. 40 ABSTRACT A hypothesis of interrelationships of the prim- ordinated classifications derived from this clado- itive araneomorph spiders based on shared de- gram are presented for purposes of comparison rived characters is presented. The genera Hy- and evaluation. The family Ectatostictidae pochilus and Ectatosticta are regarded as sister Lehtinen is newly synonymized with the Hypo- groups and as constituting the sister group of all chilidae. Serrula morphology suggests that the remaining araneomorphs. Three other genera superfamily Atypoidea is not monophyletic and (Hickmania, Gradungula, and Thaida) are sequen- that the Mecicobothriidae are more closely re- tially considered plesiomorphic sister groups of lated to the Dipluridae than to the Antrodiaeti- the remaining araneomorphs. Sequenced and sub- dae or Atypidae. INTRODUCTION The present paper represents an attempt to (1965, 1966), and (2) with the establishment of examine the higher classification of araneomorph a classification of these spiders that reflects the spiders from the viewpoint of phylogenetic sys- hypotheses of relationship thus generated. In tematics. As such, it is primarily concerned (1) addition, some data collected on mygalomorph with the placement of the "hypochiloids," the spiders for purposes of out-group comparison, most primitive of the true spiders, in monophy- and relevant to a phylogenetic analysis of that letic groups on the basis of shared derived charac- group, are also provided. ters with the methods developed by Hennig One opinion occasionally expressed is that the

'Associate Curator, Department of Entomology, the American Museum of Natural History.

Copyright © The American Museum of Natural History 1977 ISSN 0003-0082 / Price $1.65 2 AMERICAN MUSEUM NOVITATES NO. 2627

higher classification of many or most groups rep- 31 represents a hypothesis that liphistiids (Meso- resents a consensus view based on consideration thelae) are more closely related to opisthothe- of many characters by many workers, and that lines than either of those groups are to amblypy- no substantial improvements can be made in gids. Brief inspection will show that phylogenies those classifications unless newly available or pre- are merely internested series of such three-taxon viously ignored characters are brought to bear on statements that proceed from greater to lesser them. If previous classifications of the hypo- levels of universality in terms of the number of chiloid spiders are representative, and there taxa they include. seems to be no reason to suspect that they are As stressed by Hennig, the only evidence we not, this argument is indefensible. Although can have that taxa A and B are each other's clos- some new observations are added below, this in- est relatives is that they share a uniquely derived formation only corroborates hypotheses for (synapomorphic) character state that they do not which sufficient indicatory data have been avail- also share with C. For any three-taxon statement able in the literature at least since the work of there are three possible explicit (dichotomous) Marples (1968). That a satisfactory hypothesis of hypotheses of relationship (A and B are closest hypochiloid interrelationships has not yet been relatives, A and C are closest relatives, or B and C presented seems to indicate that arachnologists are closest relatives) and one possible general (tri- have lacked a satisfactory method of analyzing chotomous) hypothesis of relationship (A, B, and their data, not that they have lacked a sufficient C are closest relatives, as compared with D). If all data base on which to reach a consensus. Cladis- we have available is a synapomorphic character tic analysis provides that sorely needed metho- state shared by A, B, and C, we must choose the dology. general hypothesis. If we also have available a Taking to heart the statement Petrunkevitch synapomorphic character state shared by only made when he began his famous study on the two of the three taxa, we can choose one of the internal anatomy of spiders, and which is quoted three explicit hypotheses, not because we have above, a brief discussion of the principles under- demonstrated it to be true, but because we have lying this study is in order. The point of depar- falsified the two alternate explicit hypotheses ture is the view expressed by Popper (1968) that (such falsifications are apparent and not abso- the line of demarcation between science and lute, of course; absolute falsification seems as non-science is falsifiability, that only a statement theoretically unobtainable as absolute truth). that can potentially be shown to be wrong by Given a set of purportedly synapomorphic char- some possible observation of the real world quali- acter states that do not have perfectly congruent fies as a scientific hypothesis. Thus if classifica- distributions among the taxa considered, par- tions are to be scientific they must be falsifiable simony dictates that we must always choose the hypotheses. Since the hypothesis involved in hypothesis that appears to have been falsified the grouping a set of taxa together is one of relation- least number of times. Thus if A and B share six ship (namely that those taxa so grouped are more purportedly synapomorphic character states, and closely related to each other than to taxa ex- A and C share two purportedly synapomorphic cluded from the group), the process of construct- character states, we must choose the hypothesis ing a scientific classification (as opposed to a that A and B are closest relatives, the implication classification designed only as a technique for being that the two incongruent character states information storage and retrieval) necessarily are not actually synapomorphic. They may have involves an attempt at reconstructing the inter- been derived more than once by parallelism, they relationships of the organisms concerned. If life may be primitive rather than derived, or they on earth had a common origin, all organisms are may indeed not be homologous in all the taxa related, and hypotheses of relationship must be considered. The crucial point is that such a deci- comparative. The most explicit possible compara- sion, once made, is always open to falsification tive hypothesis is a three-taxon statement: taxon by simply finding a larger number of purportedly A is more closely related to B than either of them synapomorphic character states in favor of one are to C. For example, the circled area of figure of the alternate hypotheses (Gaffney, MS). 1 977 PLATNICK: HYPOCHILOID SPIDERS 3

Polarity judgments, statements that one given there is no possible combination of shared de- character state, such as paraxial , is rived characters that can falsify a hypothesized primitive (plesiomorphic) at a given level of uni- ancestor-descendant relationship (autapomorphic versality and that an alternate character state, character states found in a purported ancestor such as diaxial chelicerae, is derived at that level but not in a purported descendant could be the of universality, are thus themselves subsidiary result of character reversal within the lineage). hypotheses that can be falsified by incongruence Thus a cladogram uniting A and B may mean with an otherwise corroborated hypothesis of that they had a common ancestor (X) that under- relationship. Primitive character states shared by went a speciation event, or it may mean that A is two out of three taxa cannot be used to choose the ancestor of B, or that B is the ancestor of A. between alternate three-taxon statements be- A cladogram containing fossil taxa may include cause they do not falsify any of the alternate branching points that represent no speciation hypotheses, but all shared primitive character events at all, but only ancestor-descendant rela- states do represent synapomorphies at higher lev- tionships. Similarly, in a cladogram containing els of universality. Shared character states de- only Recent taxa a common ancestor may be rived separately in different lineages (i.e., paral- identical with one of its descendants (if, for ex- lelisms) cannot be used to choose between alter- ample, one parasitic gives rise sympatri- nate three-taxon statements and do not represent cally to another by a host change without itself synapomorphies at higher levels of universality; undergoing any genetic change) or not (if one they can, however, be detected by incongruence species gives rise allopatrically to two or more with more numerous synapomorphy patterns. others). It is for this reason that common ances- Methods of arriving at a polarity judgment tors in cladograms are always hypothetical and include, in roughly the order of their usefulness, that no taxa are placed at the nodes of such dia- out-group comparison, study of ontogenetic grams. Certainly we may choose to regard all transformations, character correlation, and com- branching points in a cladogram as allopatric parison of character-state frequency, but as with speciation events for the purposes of a biogeo- any hypothesis the criterion of its scientific graphic analysis, just as we may choose to regard acceptability is not the method by which it is all terminal taxa in a cladogram as biological reached, but its testability and potential for falsi- species for the purposes of a reproductive analy- fication (Popper, 1968). Like any science, phylo- sis. Applications of such assumptions as axioms genetics does not now and never will provide us for some particular purpose involve converting with the truth in any final sense, but only with a the cladogram into one (and only one) of the set of hypotheses that have not yet been falsi- numerous possible phylogenetic trees it repre- fied. sents, and such assumptions are not involved in It is worth noting, primarily to forestall irrele- the construction of cladograms or of classifica- vant criticism, that speciation has not been men- tions from them. Criticisms of cladistics as neces- tioned in this discussion, for the simple reason sitating dichotomous speciation or equal (or that although classical phylogenetic trees may unequal) evolutionary rates, and the like, are purport to show speciation events, cladograms, in based on confusion between cladograms and and of themselves, do not. Cladograms operate at phylogenetic trees. a level of generality above that of phylogenetic What follows, then, is an attempt to develop a trees; they function as sets of trees (Nelson, MS). series of three-taxon statements reflecting the In other words, a branch of a cladogram uniting interrelationships of the hypochiloid spiders, five two taxa says only that those two taxa are, com- genera (Hypochilus Marx, Ectatosticta Simon, paratively, closest relatives. An ancestor and its Hickmania Gertsch, Thaida Karsch, and Gradun- descendant are more closely related to each other gula Forster) that have been placed in from two than either is to a third taxon not also a descend- to five families and ranked at categorical levels as ant of the same ancestor. Cladograms do not dis- high as the suborder. The only consensus is that tinguish between ancestor-descendant relation- they do represent the most primitive of the ships and sister-group relationships, because known araneomorph spiders, as evidenced by 4 AMERICAN MUSEUM NOVITATES NO. 2627 their retention of four pairs of heart ostia (in all groups, the "Araneae theraphosae" for those five genera) and two pairs of book lungs (in all with paraxial chelicerae and the "Araneae verae" except Thaida), both primitive character states as for those with diaxial chelicerae. The former determined by out-group comparison with all group was divided into three families: the liphis- other spiders. The term "hypochiloid" is there- tiidae, Atypidae (corresponding to the present- fore used here to indicate a grade, not a clade. day superfamily Atypoidea), and Aviculariidae The out-group comparisons made below are (corresponding to the present-day superfamily based on the subordinal classification argued by Ctenizoidea), whereas the latter group was di- Platnick and Gertsch (1976). vided into two sections, the Cribellatae for those Hypochilus contains four described and one true spiders with a cribellum and undescribed (R. L. Hoffman, personal commun.) (including the Hypochilidae) and the Ecribellatae species from the southeastern United States, Col- for those true spiders lacking a cribellum and orado, and California, Ectatosticta a single calamistrum (fig. 2). This basic division of arane- species from , Hickmania a single species omorph spiders into cribellate and ecribellate from Tasmania, Thaida (formerly Austrochilus) a groups was maintained by Pocock (1900), Com- single species from Chile and western Argentina, stock (1912), and Bristowe (1938). A further re- and Gradungula two described and several un- finement of this viewpoint was added by Capori- described (R. R. Forster, personal commun.) spe- acco (1938), who divided the Cribellatae into cies from Australia and New Zealand. Gradun- two cohorts, the Palaeocribellatae, containing the gula has been treated by Forster (1955) and four-lunged hypochiloids, and the Neocribellatae, Davies (1969), Thaida by Zapfe (1955), and the containing all the other (two-lunged) cribellates other genera by Gertsch (1958, 1964). (fig. 4). Gerhardt and Kastner (1938) and Bonnet I thank Mr. Robert J. Koestler for his assis- (1959) followed Caporiacco in recognizing the tance with the scanning electron microscope. Palaeocribellatae; Marples (1968) transferred Drs. J. A. Beatty, R. R. Forster, W. J. Gertsch, B. Hickmania and Thaida from the Palaeocribellatae J. Kaston, 0. Kraus, H. W. Levi, R. T. Schuh, W. to the Neocribellatae (fig. 6). A. Shear, and P. Wygodzinsky read and com- A third point of view is represented in an mented on a draft of the manuscript. early proposal by Petrunkevitch (1923) in which the hypochiloids are considered more closely HISTORY related to the Filistatidae and some haplogyne ecribellates than to the other cribellates; Petrun- The phylogenetic significance of the hypo- kevitch published a diagram in which the hypo- chiloids was recognized immediately when Hypo- chiloids are considered the most plesiomorphic chilus thorelli was discovered by Marx (1888); as branch of a lineage including also (approximately Marx (1889, p. 166) said, the is "so anom- in order of ascending apomorphy) the Filista- alous that it appears like the representative of a tidae, Sicariidae, Ammoxenidae, Dysderidae, prototype." As a result, we need be concerned Oonopidae, , Urocteidae, Leptoneti- here only with classifications proposed since dae, Prodidomidae, , and Telemidae. 1888. With respect to the placement of the hypo- Among more recent authors, Lehtinen (1967) chiloids, four basically different views can be has presented similar views placing the hypo- recognized. chiloid genera in two separate plesiomorphic Thorell (1891) divided spiders into two lineages of . The first lineage, groups, Tetrapneumones for those with two pairs called Filistatides, is broken into three superfami- of lungs and Dipneumones for those with a single lies: the Hypochiloidea (containing Hypochilus pair of lungs (and presumably those few forms and the , Ochyroceratidae, Pholci- now known to be lungless); as a result he was dae, and Scytodidae), the Gradunguloidea (con- forced to consider the hypochiloids more closely taining Hickmania, Gradungula, and probably related to mygalomorphs and liphistiids than to Ectatosticta), and the Filistatoidea (containing the araneomorphs (fig. 1). Dahl (1904) excluded the Filistatidae, , Caponiidae, Oono- the liphistiids from Tetrapneumones but still pidae, and Dysderidae); three families (Digueti- placed the hypochiloids with the mygalomorphs. dae, Sicariidae, and Segestriidae) are incertae Simon (1892) also divided spiders into two sedis within Filistatides (fig. 5). Lehtinen's sec- q 03 0 0b 0 & 0 .Q A,,ood at.. r k, 00Yi .0qI 0 //c)

,raneae verae

0 o C .0- .0

If Ecribellatae

- Araneomorphae

Opisthothelae

0.

Neocribellatae

Cribellatae - Ecribellatae

Araneomorphae

6

FIGS. 1-6. Cladograms derived from previous classifications of spiders; names have been standard- ized wherever possible. 1. Thorell (1891). 2. Simon (1892). 3. Petrunkevitch (1933). 4. Caporiacco (1938). 5. Lehtinen (1967). 6. Marples (1968). 6 AMERICAN MUSEUM NOVITATES NO. 2627 ond plesiomorphic lineage, called Thaidides, con- liphistiids; these homologies have been well tains only the genera Thaida and Megadictyna, established by embryological studies (Dahl, each placed in a separate superfamily. 1901; Montgomery, 1909). Thus the only ques- Finally, a fourth point of view was first ex- tion is whether the cribellum and colulus are also pressed by Petrunkevitch (1933), who regarded homologous to each other; in other words, are the hypochiloids as the sister group of all other there two separate transformation series (anterior araneomorphs; the cladogram representing his median to cribellum and anterior view (fig. 3) is taken not from his classification median spinnerets to colulus) or only one (ante- (which recognizes five separate lineages at the rior median spinnerets to cribellum to colulus, or suborder level) but from a branching diagram alternately, anterior median spinnerets to colulus provided in the same paper (1933, table 2). to cribellum)? If the first model holds, then Forster (1955) followed Petrunkevitch's group- cribellate and colulate spiders could both repre- ing when describing Gradungula, which he placed sent monophyletic groups; if the second model as the sister group of the other hypochiloid holds, one or both groups are not monophyletic. genera, an arrangement subsequently followed by In attempting to answer this question, the Gertsch (1958) and Davies (1969). first difficulty encountered is that of clearly de- limiting the cribellum and colulus (Lehtinen, 1967, pp. 398403). Since some , THE CRIBELLATAE , and Hersiliidae have a large and even The following two sections are devoted to platelike colulus, the only defining character of analysis of the classifications of hypochiloids the cribellum seems to be the possession of func- that have been offered in the past. The question tional spigots. However, a number of species of of the validity of the division of araneomorphs various groups are known in which the cribellum into cribellate and ecribellate groups is singled is functional in females, but not in males (which out first because there is certainly no other char- are thus colulate; Lehtinen, 1967). In other acter that has caused more controversy in spider words, a colulus is simply a non-functional cribel- classification. Indeed, it is not uncommon even lum, and the two structures are therefore homol- today to hear arachnologists speak of "the cribel- ogous. late problem." Given that , is the presence of a One of the insights provided by phylogenetic functional cribellum in some araneomorphs a systematics is that when examining the status of symplesiomorphy, a synapomorphy, or a paral- any given character state at any given level of lelism? No one has argued seriously in favor of universality, there are only three possibilities: the last hypothesis, presumably because the cri- either the character state is symplesiomorphic in bellum is a complex structure and is associated those taxa that share it or it is synapomorphic in with the calamistrum, one or two rows of setae them or it represents a parallelism (Hennig, 1965, on the fourth metatarsus used to comb silk from fig. 1). Of course, Hennig's insight is not original; the numerous spigots on the cribellum. As Pocock (1892) made this point quite clearly in pointed out in the last section, however, numer- his analysis of the Cribellatae, but many modern ous authors over the last century have treated the arachnologists (and other systematists) have ig- presence of a functional cribellum as a synapo- nored this crucial axiom of character analysis. morphy. What is the nature of the evidence in The first task of character analysis is identifi- favor of such a decision? Kaestner (1968), fol- cation of homologous character states. There is lowing Crome (1955), distinguished cribellates no problem here as all arachnologists since from ecribellates on the basis of the number of Thorell's time have agreed that both the cribel- dorsoventral abdominal muscles (four pairs in cri- lum (a broad platelike structure sometimes found bellates, three or fewer in ecribellates). This char- in front of the anterior [lateral] spinnerets of acter (1) does not seem to work, since Millot araneomorphs) and the colulus (a smaller fleshy (1933) demonstrated that the abdomen of Ecta- lobe found between or in front of the same spin- tosticta has only three pairs of dorsoventral nerets of other araneomorphs, and of which muscles, and Millot (1936) indicated that at least several degenerative forms are known) are homol- some , , and Filistatidae ogous with the anterior median spinnerets of resemble ecribellates in having three pairs of dor- 1 977 PLATNICK: HYPOCHILOID SPIDERS 7 soventral muscles (plus anterior "latero-cardiac" the functional cribellum as synapomorphic, a muscles) rather than other cribellates such as the decision that ruined an otherwise impeccable Eresidae, which have four pairs of dorsoventral analysis ofhypochiloid interrelationships. The dis- (plus the latero-cardiac) muscles; (2) would not covery of as yet undescribed cribellate gradungu- argue for the monophyly of the Cribellatae even lids (R. R. Forster, personal commun.) should pro- if it did work, since out-group comparison with vide the final nail for the coffin of the Cribellatae. both (Millot, in Bristowe, 1933) and Thus it appears that there is a single transfor- Heptathela (Marples, 1968, fig. 4) indicates that mation series (anterior median spinnerets to cri- the presence of four pairs of dorsoventral mus- bellum to fleshy colulus to reduced colulus bear- cles is plesiomorphic; and (3) cannot, even dis- ing setae to colular setae only to colular setae regarding the parallelisms in some mygalomorphs absent). If this view is correct, the cribellum is a and cribellates, be regarded as synapomorphic for synapomorphy, but a synapomorphy for all ara- ecribellates, because it is not the same pair of neomorphs rather than just for those with a func- muscles that is lost in all the ecribellate groups tional cribellum. That the homologue of the (Millot, 1936). Thus, although the abdominal anterior median spinnerets is functional must musculature may provide useful data at lower obviously be primitive for spiders. Since the cala- levels of universality, it does not aid in analysis mistrum disappears whenever the homologue of of the Cribellatae. It would appear that, to date, the anterior median spinnerets loses its function the advocates of this classification have not been (even in males of species whose females have able to discover any derived character shared calamistra and a functional cribellum; Shear, uniquely by all the ecribellates; for that matter, 1970), the only derived character of the Cribel- other than the functionally associated calamis- latae (fused anterior median spinnerets) is the trum, no additional derived characters shared same as one of the derived characters of the Ara- uniquely by all the cribellates seem to have been neomorphae. To put it baldly, all true spiders are discovered either. cribellates; in some the cribellum has simply lost There is, moreover, voluminous evidence argu- the function it primitively had. Moreover, the ing against regarding the functional cribellum as transformation to a colulus is clearly not a synap- synapomorphic for the Cribellatae. Lehtinen omorphy for all colulates; it has happened at (1967), Forster (1970), Baum (1972), Forster least twice (independently in the and Wilton (1973), and Davies (1976) have pre- and in other araneomorphs) and in all probability sented numerous cases of families and genera a great number of times. The cribellate nature of that contain both cribellate and ecribellate (i.e., Araneomorphae is reflected below in the enlarge- colulate) species and that are nonetheless each ment of the Neocribellatae to include all araneo- united by obvious synapomorphies. To view the morphs other than Hypochilus and Ectatosticta. cribellum as derived from the colulus is to re- quire that within each of these groups a non- A CRITIQUE functional colulus has been converted into a functional cribellum, with the silk gland connec- Evolutionary systematists often maintain that tions, the associated calamistrum, and the ap- the cladistic emphasis on shared derived charac- propriate behavior patterns being re-evolved in ters is merely an explicit statement of what they each lineage. Since the number of incongruent have always done. To some extent this is true, characters provided by all these groups is vastly since random chance will dictate that of any two greater than the single character of the cribellum, groups distinguished by alternate states of the parsimony dictates rejection of the functional same character, one will be monophyletic and cribellum as a synapomorphy. Further, character the other might be. But one has merely to look correlation also falsifies that hypothesis. Of the at past spider classifications to discern how fre- five genera of araneomorphs with four pairs of quently groups have been delineated by shared cardiac ostia, Hypochilus, Ectatosticta, Hick- primitive characters only. wmnia, and Thaida are cribellate. Gradungula was For example, Thorell and Dahl placed the described as ecribellate, and the fact that Marples hypochiloids with the mygalomorphs since both (1968) knew only ecribellate gradungulids was groups have two pairs of lungs. Thorell's Dip- certainly instrumental in his decision to accept neumones is a group based on a purported synap- 8 AMERICAN MUSEUM NOVITATES NO. 2627 omorphy, but out-group comparison with am- hoped that he will present the shared derived blypygids shows clearly that the presence of characters needed to provide his groups with a four lungs is primitive for spiders and therefore basis that can be discussed and evaluated scien- cannot be used to demonstrate relationship be- tifically. It is unfortunate that Lehtinen's superb tween any two subgroups of spiders. analysis of the evolution of the cfibellum and the The division of araneomorphs into hypochi- inadequacy of many classical groups was pre- loids and non-hypochiloids (Dipneumono- sented together with undocumented proposals of morphae plus Apneumonomorphae) by Petrun- new groupings. Such undocumented groupings kevitch (1933) is a parallel case; each of the char- may be heuristic when offered as preliminary acter states listed by Petrunkevitch as character- suggestions, but when used as the basis for entire izing the Hypochilomorphae (abdomen without classifications (as they frequently are by evolu- tergites, cribellum present, four pairs of ostia, tionary systematists), they represent authoritari- two pairs of lungs, diaxial chelicerae, endoche- anism and not science. liceral poison glands, coxal glands with one out- let, endites present, maxillary glands unicellular, and three tarsal claws) that is still known to be CHARACTERS accurate (the cribellum is lost in some gradungu- The distributions of the states of the follow- lids; Hickmania, Gradungula, and Thaida have ing numbered characters are used to support a endocephalic poison glands; the maxillary glands cladogram (fig. 7) in which a lineage containing of all five genera are multicellular [see Marples, only Hypochilus and Ectatosticta is shown as the 1968]) is indicated on his own table as being sister group of all other Araneomorphae. plesiomorphic for araneomorphs. The mono- 1. In Ectatosticta (Millot, in Bristowe, 1933) phyly of the Hypochilomorphae has been effec- and Hypochilus (Marples, 1968) the thoracen- tively disputed by both Lehtinen (1967) and teron (the prosomal portion of the midgut) has Marples (1968). diverticula that extend into the base of the chelic- The third view of hypochilomorph interrela- erae. According to Millot (1933) who had ear- tionships, placing them with other haplogyne lier (Millot, 1931) studied the thoracenteron of families, is more difficult to deal with. This view numerous araneomorphs, Liphistius is the only was presented by Petrunkevitch (1923) and re- other in which the diverticula are known peated, with minor differences, by Lehtinen to extend into the chelicerae. Hickmania, Thaida, (1967). Although Lehtinen claimed to be pre- and Gradungula lack the anteriorly extended senting a phylogenetic classification, both his and diverticula (Marples, 1968). Immediate out-group the early Petrunkevitch arrangement are typical comparison with mygalomorphs indicates that evolutionary classifications and share the funda- the absence of cheliceral diverticula is plesio- mental flaw of such systems. Since neither morphic. The presence of cheliceral diverticula in author has presented a list of shared derived char- Liphistius might tend to contradict this polarity acter states uniting such groups, it is not possible hypothesis, but both Heptathela (Millot, in Bris- to falsify their proposals. Since we do not know towe, 1933) and amblypygids (Millot, 1949, fig. how many (if any) synapomorphies support such 337) lack them. To consider the presence of che- groupings, we can never know whether we have liceral diverticula synapomorphic for Araneae found enough conflicting characters to have falsi- (and therefore plesiomorphic for any spider) re- fied those groupings. Any proposal that cannot quires three separate cases of character reversal be potentially falsified is unscientific. Both those (in Heptathela, , and Araneo- authors have united Hypochilus with haplogyne morphae other than Hypochilus and Ectato- ecribellates; I am unaware of any shared derived sticta) and is therefore a less parsimonious hy- characters supporting that grouping, and there pothesis than the polarity proposed here, which are several characters, discussed below, that sup- requires one case of parallelism in the acquisition port a closer relationship of those haplogyne of cheliceral diverticula (between Liphistius and families to other araneomorphs than to Hypochi- Hypochilus plus Ectatosticta). lus. Petrunkevitch (1933) eventually abandoned 2. Hypochilus and Ectatosticta have distinct this view; if Lehtinen still accepts it, it is to be concavities on the median surface of the chelic- q 39l4z. 0b q.41 0N,. M5G lb.

FIG. 7. Cladogram of hypochiloid interrelationships. Numbers refer to characters discussed in text. Dark squares denote apomorphic character states; partially darkened squares denote transformation series of uncertain polarity; subscript letters denote different, independently derived apomorphic states of a single character. 10 AMERICAN MUSEUM NOVITATES NO. 2627

......

FIGS. 8-13. Scanning electron micrographs. 8. Hypochilus sp., chelicerae, posterior view, 60X. 9-13. Serrulae of females, anterior views. 9. Gradungula sp., 1300X. 10, 11. Thaida sp., 120X, 600X. 12, 13. Hickmania sp., 130X, 1300X. erae (fig. 8) into which the tips of the long fangs regarded as autapomorphic for the lineage con- fit when closed. These concavities are not found taining Hypochilus and Ectatosticta. in the other hypochiloid genera nor, to my 3. Marples (1968) noted that two types of knowledge, in other spiders, and their presence is serrula (a group of toothlike structures situated 1 977 PLATNICK: HYPOCHILOID SPIDERS I1I

J-6

t5

FIGS. 14-17. Scanning electron micrographs of female serrulae, anterior views. 14, 15. Hypochilus sp., 120x, 1200X. 16, 17. Ectatosticta sp., 130X, 650X. at the ventral tip of the anterior surface of the below. Since neither type of araneomorph serrula palpal endites) are found in araneomorph spiders. is found in mygalomorphs or liphistiids, both the Most have a single row of closely spaced teeth; multiple row type of Hypochilus and Ectatos- Thaida (figs. 10, 11), Gradungula (fig. 9), Hick- ticta (character 3a) and the single row type of all nania (figs. 12, 13), and the non-hypochiloid other araneomorphs (character 3b) are regarded araneomorphs (including the other cribellates as apomorphic. and the haplogyne ecribellates) have this type of 4. Marples (1968) examined the musculature serrula. InHypochilus (figs. 14, 15) and Ectatos- associated with the cuticle-lined foregut of a ticta (figs. 16, 17), however, the serrula consists wide variety of spiders (although he did not pro- of a plate bearing several parallel rows of teeth. vide a complete list of examined taxa) and noted To judge which state is derived, a wide variety of that in most mygalomorphs and araneomorphs mygalomorphs (listed below) were examined the dorsal dilator muscles of the pharynx origi- with light and scanning electron microscopy. nate dorsally on the carapace and insert on the Most mygalomorphs lack serrulae, and those that pharyngeal lobes (a pair of dorsally directed have been found generally consist of a patch of cuticular folds situated at the juncture of the scattered teeth not aligned in distinct rows (as in pharynx and esophagus); the anterior dilator fig. 18). Further comments on mygalomorph ser- muscles of the pharynx also originate dorsally on rulae will be found in the section on Atypoidea the carapace but insert on a groove of the phar- 12 AMERICAN MUSEUM NOVITATES NO. 2627 ynx anterior of the pharyngeal lobes. Among the morphs that have such glands (Petrunkevitch, hypochiloids, Hickmania, Gradungula, and Tha- 1933) do extend into the cephalothorax. Mygalo- ida have muscular systems of this type. In Hy- morphs and liphistiids have endocheliceral poison pochilus and Ectatosticta, however, the phar- glands, and the endocephalic glands of araneo- yngeal dilators have the same insertions but morphs other than Hypochilus and Ectatosticta originate anteriorly on an apodeme of the ros- are regarded as synapomorphic. trum. Immediate out-group comparison with the 6. In liphistiids and most mygalomorphs the Mygalomorphae suggests that a carapace origin is endosternite has ventral extensions that reach the the plesiomorphic state of the character, but cuticle of the labium and sternum and are at- both the liphistiid genus Heptathela (Marples, tached there at rounded sigilla. Gertsch (1958) 1968) and amblypygids (Millot, 1949, fig. 334) noted that Ectatosticta have sternal sigilla, and have a rostral origin for at least the anterior dila- Marples (1968) demonstrated that both Ectatos- tors. As a result, the polarity of this transforma- ticta and Hypochilus have a pair of labial sigilla. tion series seems indeterminable; if the rostral In the other hypochiloids and araneomorphs, the origin is plesiomorphic for opisthotheline spiders, ventral extensions of the endosternite do not one case of parallelism is required (between reach the cuticle (Marples, 1968), and the short- Mygalomorphae and araneomorphs other than ening of both the sternal and labial endosternite Hypochilus and Ectatosticta), whereas if the extensions is regarded as synapomorphic for ara- carapace origin is plesiomorphic for opistho- neomorphs other than Hypochilus and Ectatos- theline spiders, one case of character reversal is ticta (with a parallel shortening being found in required (in the lineage containing Hypochilus some mygalomorphs). Lehtinen (1967) indicated and Ectatosticta). Although parallelism may in that some Filistatidae appear (externally) to have fact be more common in evolutionary sequences sigilla; this needs to be confirmed by internal than character reversal, both polarity hypotheses examination of the endosternite. are equally parsimonious and there is no compel- 7. Marples (1968) examined the coxal glands ling reason to prefer one over the other. Thus the of the hypochiloids and confirmed that all five distribution of the states of this character sup- genera resemble araneomorphs rather than liphis- ports the dichotomy between Hypochilus plus tiids and mygalomorphs in having a single outlet Ectatosticta and all other Araneomorphae, and at the base of the first coxae and lacking an out- the monophyly of one group or the other, but let at the base of the third coxae. However, he we do not know which. also noted that although Hickmania, Gradungula, This character could also be used to support and Thaida have the simple inverted U-shaped an alternative cladogram in which Hypochilus ducts of other araneomorphs, Hypochilus and and Ectatosticta constitute the sister group of all Ectatosticta have highly convoluted ducts like other rather than of all other Ara- those of mygalomorphs (Buxton, 1913, diagram neomorphae, but a much larger number of synap- B3) and liphistiids (Millot, in Bristowe, 1933). omorphies (the presence of modified anterior Thus out-group comparison indicates that acqui- median spinnerets, fewer than three articles in sition of the inverted U-shaped duct is apomor- the posterior lateral spinnerets, diaxial chelicerae, phic. coxal glands with a single opening, and maxillary 8. Millot (1933) indicated that all spiders glands opening on a sieve; Platnick and Gertsch, have four ventral abdominal invaginations of the 1976) unite those genera with the other araneo- cuticle forming short endosternites and repre- morphs. senting the posterior margins of the anterior 5. The venom glands of Hypochilus (Petrun- abdominal segments (bearing the anterior and kevitch, 1933) and Ectatosticta (Millot, 1933) posterior respiratory organs and spinnerets), and are endocheliceral and do not extend into the that Liphistius retain in addition five extra pos- cephalothorax; the venom glands of Hickmania terior endosternites reflecting the primitive ab- (Marples, 1968), Gradungula (Forster, 1955), dominal segmentation. Marples (1968) indicated Thaida (Zapfe, 1955), and the other araneo- that Heptathela also retain the five extra abdomi- 1 977 PLATNICK: HYPOCHILOID SPIDERS 13 nal endosternites, and that Hypochilus and Ecta- morphs rather than the other araneomorphs in tosticta (but not the other hypochiloids) retain having an M-shaped intestine. Consideration of the most anterior of the extra invaginations. the straight intestine as a synapomorphy for the Since the endosternites reflect abdominal seg- non-hypochiloid Araneomorphae is also sup- mentation that is primitive by both ontogenetic ported by the presence of an M-shaped intestine evidence and out-group comparison, the loss of in both Liphistius '(Millot, in Bristowe, 1933) and the fifth endosternite is regarded as a synapo- Heptathela (Marples, 1968). morphy for Araneomorphae other than Hypo- 12. Most araneomorphs have only two or chilus and Ectatosticta (paralleled in the Mygalo- three pairs of heart ostia. Among the hypochiloid morphae). genera, however, Thaida (Zapfe, 1955), Gradun- 9. It has been argued elsewhere (Platnick and gula (Forster, 1955), Hickmania (Marples, 1968), Gertsch, 1976) that the presence of two pairs of Ectatosticta (Millot, 1933), and Hypochilus (Pe- spermathecae is primitive for spiders; among the trunkevitch, 1933) all have four pairs of ostia. Araneomorphae, only Hypochilus, Ectatosticta, Mygalomorphs have three or four pairs of ostia, and Hickmania retain two pairs of spermathecae. but the presence of five pairs in Liphistius (Pe- The loss of the second pair of spermathecae is trunkevitch, 1933) and six pairs in Amblypygi regarded as a synapomorphy for Gradungula, (Millot, 1949) provides ample evidence that the Thaida, and all other araneomorphs (with paral- direction of the transformation series is toward lelism in some mygalomorphs). Some non-hypo- reduction in number. The presence of three or chiloid araneomorphs have two longitudinally fewer pairs of ostia is regarded as synapomorphic interconnected receptacles on each side of the for the non-hypochiloid Araneomorphae (with internal female genitalia, but these are clearly dif- parallelism in some mygalomorphs). ferent from the side-by-side, non-interconnected H. W. Levi has argued (in litt.) that the pres- spermathecae of liphistiids, atypoids, and some ence of four pairs of cardiac ostia and two pairs hypochiloids, and are presumably specializations of book lungs are necessarily correlated with of single spermathecae. The single median recep- each other for physiological reasons, and that the tacles of some other araneomorphs are presuma- two features therefore cannot be considered sep- bly specializations acquired by fusion. Kraus and arate characters. That Thaida has four pairs of Baur's (1974, figs. 42-44) observations of what cardiac ostia but only a single pair of lungs seems may be a strongly reduced single median recep- to be a sufficient refutation of that contention. tacle in some A typus may provide evidence There are a few other characters not included against the polarity adopted here; ontogenetic in figure 7 that may serve to support the dichot- studies might help to resolve the question. omy between Hypochilus plus Ectatosticta and 10. The posterior respiratory organs ofHypo- all other Araneomorphae. First, the male palpi of chilus, Ectatostictd, Hickmania, and Gradungula Hypochilus and Ectatosticta are peculiar in that are book lungs, whereas those of Thaida and all the alveolus and bulb are situated at the apex of other Araneomorphae are tracheae. Out-group a long cymbium (Gertsch, 1958, figs. 17, 27). In comparison with Mygalomorphae, , the other hypochiloids the bulb has its normal and Amblypygi indicates that the presence of (for most araneomorphs) placement. Out- posterior book lungs is plesiomorphic, and re- group comparison with mygalomorphs and liphis- placement of the posterior pair of lungs with tiids is not particularly useful in this case since tracheae is regarded as a synapomorphy for the cymbium in those groups is generally so short T7haida plus the non-hypochiloid araneomorphs. that the bulb cannot be judged to be either basal 11. Millot (1933) pointed out that Ectatos- or apical. Second, Hypochilus and Ectatosticta ticta resembles mygalomorphs rather than ara- have calamistra composed of two rows of hairs; neomorphs in having the opisthosomal portion of the other hypochiloid genera and, to my knowl- the midgut M-shaped in lateral view rather than edge, all cribellates other than a few Amaurobi- straight, and Marples (1968, fig. 4) has shown idae have calamistra consisting of a single row of that all five hypochiloid genera resemble mygalo- hairs. It is tempting to regard the biseriate cala- 14 AMERICAN MUSEUM NOVITATES NO. 2627 mistrum as derived on the basis of frequency zoids; (3) the abdomen of atypoids bears tergites alone, but since out-group comparison is inopera- lacking in ctenizoids; (4) the penultimate male tive (because mygalomorphs and liphistiids have has a swollen palp in atypoids but not in cten- no calamistrum) and ontogenetic evidence is izoids; (5) atypoid females have two pairs of lacking, no well defended polarity judgment is spermathecae, ctenizoid females usually a single possible. Third, Marples (1968) reported that the pair; and (6) the male palp of atypoids has a anal tubercles of Hypochilus and Ectatosticta conductor lacking in most ctenizoids. (but not the other hypochiloids) have multicellu- Out-group comparison with liphistiids indi- lar glands known elsewhere in araneomorphs cates that the larger number of spinnerets, the only in the Oecobiidae (including Urocteinae); a separated anal tubercle, the presence of abdomi- polarity hypothesis on this character must await nal tergites, a swollen in the penulti- study of the anal tubercles of mygalomorphs and mate male, and two pairs of spermathecae are the liphistiids. plesiomorphic character states. It has been ar- Finally, there are derived characters defining gued elsewhere (Platnick and Gertsch, 1976) that each of the hypochiloid genera. Hypochilus has a homologue of the palpal conductor is probably false articulations in the tarsi and a spine-bearing present in liphistiids and that the presence of a cymbial apophysis on the male palp (Gertsch, conductor is historically associated with the pres- 1958, fig. 17) not found in the other genera, ence of two pairs of spermathecae in females and Ectatosticta a uniquely serrate cymbial tip therefore a necessary step in the evolution of the (Gertsch, 1958, figs. 26, 27), Hickmania an api- ctenizoid palp. If this argument is correct, and cally twisted male embolus (Gertsch, 1958, figs. the presence of a palpal conductor and doubled 34, 37), Gradungula elongated proclaws on legs I spermathecae in the most primitive ctenizoids and II (Forster, 1955, fig. 2), and Thaida a and araneomorphs argues strongly for it, the Aty- raised, semicircular, flangelike embolus (Gertsch, poidea are not united by any known shared de- 1958, fig. 41). rived character. For purposes of out-group comparison in con- THE ATYPOIDEA nection with character 3 above, the endites of a variety of mygalomorph genera were examined: Recent workers, following the lead of Simon and Calomnmata (Atypidae); Antrodi- (1892), have often divided the Mygalomorphae aetus, , and (Antrodiaeti- into two groups, the Atypoidea (containing the dae); Hexura and Megahexura (Mecicobothri- families Atypidae, , and Mecico- idae); Ummidia, Galeosoma, Bothriocyrtum, and bothriidae) and the Ctenizoidea (containing the Nemesia (); Actinopus and Missulena remaining families). The monophyly of these (Actinopodidae); Migas, Micromesomma, and superfamilies has been defended by Chamberlin Calathotarsus (Migidae); Euagrus, A trax, Accola, and Ivie (1945) and Coyle (1971, 1975), even and Hexathele (Dipluridae); Psalistops (Bary- though this hypothesis requires parallelism in the chelidae); Ischnocolus, Eurypelma, and Gram- acquisition of a rastellum and trapdoor burrow- mostola (Theraphosidae); Paratropis (Paratropidi- ing habits between the Antrodiaetidae and Cteni- dae); and an unidentified genus of Pycnothelidae. zidae, and in the acquisition of elongated pos- Of these, only the Mecicobothriidae and Dipluri- terior lateral spinnerets and sheet-web building dae have a serrula. The typical diplurid serrula is behavior between the Mecicobothriidae and found in Hexathele and Euagrus (figs. 20-22; see Dipluridae. also Marples, 1968, fig. 5), and consists of a Six characters have been used to support this broad patch of more or less scattered teeth; this dichotomy: (1) atypoids generally have six spin- patch is greatly reduced in size in Atrax (fig. 23), nerets, rarely four, and ctenizoids generally have and modified into a row of sharply pointed teeth four spinnerets, rarely six; (2) the anal tubercle is that resembles that of araneomorphs (but is moderately separated from the spinnerets in sinuous and subapical rather than evenly curved atypoids, approximate to the spinnerets in cteni- and apical) in Accola (figs. 24, 25). Both Hexura 1 977 PLATNICK: HYPOCHILOID SPIDERS 15

(figs. 18, 19) and Megahexura have serrulae like is, will also be a family (Yidae). In other words, those of typical diplurids. given the following classification: To determine whether or not the presence of a serrula in the Dipluridae and Mecicobothriidae (1) Order Araneae should be regarded as a synapomorphy, Liphisti- Suborder Mesothelae us and Heptathela were examined. No structure Suborder Opisthothelae resembling a serrula was found in Heptathela, Infraorder Mygalomorphae even under scanning electron microscopy. Liphis- Infraorder Araneomorphae tius has a structure that might be regarded as a Section Cribellatae precursor of a serrula; the cuticle of the anterior Cohort Palaeocribellatae surface of the endite is scalelike (fig. 26), and Cohort Neocribellatae toward the lateral side of the endite the tips of Section Ecribellatae the cuticular scales become more elevated and Cohort Dipneumones toothlike (fig. 27). Even if this structure is re- Cohort Apneumones garded as a serrula, it is clearly quite different in position and morphology from that found in the only evolutionary information we can obtain Hexura and Hexathele, and the latter structure is the system of interrelationships portrayed in can be considered derived. figure 4. Thus, what the taxonomic hierarchy Thus on the basis of two characters, serrula "mirrors" is a hypothesized phylogeny. morphology and the length of the posterior lat- Evolutionary systematists argue that addi- eral spinnerets, the Mecicobothriidae appear to tional information can be included in a classifica- be more closely related to the Dipluridae than to tion that reflects the amount of change that has the Antrodiaetidae or Atypidae (and might well occurred in different branches. Disregarding the be regarded as merely the most plesiomorphic empirical difficulty of measuring the amount of members of the Dipluridae), and the superfamily change, such a systematist might argue that "Yes, Atypoidea, at least as currently delimited, ap- the branching sequence is like that in figure 4, pears not to be monophyletic. but the Mygalomorphae are very similar to the Mesothelae in appearance and biology. The Ara- *neomorphae, however, are very different; they CLASSIFICATION have been able to fill a greater variety of niches and According to Petrunkevitch (1923, p. 145), because of their increased reliance on silk, " is the mirror of evolution." What is have therefore radiated and become tremen- the meaning of this poetic phrase? In what way dously diverse. This diversity and shift into a new can the taxonomic hierarchy of organisms "mir- adaptive zone should be recognized at a higher ror" evolution? Numerous authors, including taxonomic level, as in the following classifica- Hennig (1966), have argued at length that a hier- tion: archical classification can include all the informa- (2) Order Araneae tion found in a cladogram, that is, that it can Suborder Mesothelae accurately "reflect" both the content of Suborder Mygalomorphae branches and their sequence. It has not been suf- Suborder Araneomorphae ficiently stressed, however, that a hierarchical Section Cribellatae, etc." classification can reflect only that information (Cracraft, 1974). But what evolutionary information is contained What evolutionary information is conveyed in this hierarchy? If we use the hierarchy to con- when a reference is made to, say, the family struct a scheme of interrelationships like that of Xidae? All we can say is that (1) there is a pur- figure 4, we find that we get a different tree, portedly monophyletic group containing the resembling that of figure 3 in having three basal genus Xus and perhaps other genera as well, and branches. In other words, where at first the hier- (2) that the sister-group of the Xidae, whatever it archy did convey some limited evolutionary in- 16 AMERICAN MUSEUM NOVITATES NO. 2627

FIGS. 18-23. Scanning electron micrographs of female serrulae, anterior views. 18, 19. Hexura sp., 240X, 1200X. 20, 21. Hexathele sp., 125x, 625x. 22. Euagrus sp., 240X. 23. Atrax sp., 260X.

formation, it no longer does, because that small morphs are more closely related to araneomorphs amount of information that it did include has than to liphistiids. Had the argument been car- now been distorted. In this case, the distortion is ried further and the Araneomorphae treated as an omission of the information that mygalo- equal in importance (in diversity, adaptive zone, 1977 PLATNICK: HYPOCHILOID SPIDERS 17

l I

..I" s

FIGS. 24-27. Scanning electron micrographs of female serrulae, anterior views. 24, 25. Accola sp., 240x, 2400x. 26, 27. Liphistius sp., 1 30X, 1 300X. or the like) to all other spiders, as in the follow- How can we construct classifications that do ing classification: reflect exactly the information put into them? (3) Order Araneae Two methods have been suggested. One (subordi- Suborder Orthognatha nation) requires that we name every monophy- Infraorder Mesothelae letic group (i.e., every inclusive taxon); the other Infraorder Mygalomorphae (sequencing) does not. Take, for example, the Suborder Labidognatha classification proposed by Petrunkevitch (1933): Infraorder Araneomorphae (4) Order Araneae Suborder Liphistiomorphae the distortion would be replacement of the in- Suborder Mygalomorphae tended hypothesis of relationships with an incor- Suborder Hypochilomorphae rect one. The question then is whether we want Suborder Dipneumonomorphae the taxonomic mirror to be flat, and reflect ex- Suborder actly the information put into it, or curved, and Apneumonomorphae distort the information put into it. Since sys- In a subordinated classification, the number of tematists have gone to considerable effort to immediate subtaxa within a group reflects the compile the information in the first place, a flat number of branches originating at that point, so mirror seems to be required. if Petrunkevitch's classification was a subordi- 18 AMERICAN MUSEUM NOVITATES NO. 2627

28 30

FIGS. 28-30. Cladograms and classifications, or, what did Petrunkevitch mean? 28. Petrunkevitch (1933) spider classification if subordinated. 29. Petrunkevitch (1933) branching diagram. 30. Petrun- kevitch (1933) spider classification if sequenced. See text for explanation. nated one, his cladogram would look like figure mania, all Neocribellatae except Hickmania and 28. In a sequenced classification, equally ranked Thaida, and all Ecribellatae except Gradungula, taxa represent branches that arise in the order but these inclusive taxa were not named by Mar- they are listed, so if Petrunkevitch's classification ples. was a sequenced one, his cladogram would look Given, then, that there are three ways to con- like figure 30. Petrunkevitch provided his own struct a classification (subordination, sequencing, branching diagram, like that of figure 29 (the or a system combining the two), which is best? dichotomy between Dipneumonomorphae and To answer the question, of course, one must Apneumonomorphae represents his taxonomic know: best for what? If classifications are to be grouping of what his diagram shows as four sep- used by comparative biologists in general (includ- arate lineages). Clearly, Petrunkevitch subordi- ing those who may be unfamiliar with the tech- nated the Liphistiomorphae and Mygalomorphae niques of constructing them), we must presuma- but sequenced the remaining three suborders, bly use either subordination or sequencing con- thereby eliminating two inclusive taxa (Araneo- sistently throughout our classifications, as it is morphae, and an unnamed taxon for Dipneu- unlikely that the subtleties of a combined system monomorphae plus Apneumonomorphae). Millot will be understood by non-systematists. The use (1933) criticized the Petrunkevitch system be- of sequencing eliminates some inclusive taxa, so cause it excluded the inclusive taxon Araneo- the question is really, do we need inclusive taxa morphae (and also because he doubted, with (i.e., names) for every monophyletic group? good reason, the monophyly of the Apneumono- Given an infinite expansion of biological knowl- morphae). edge, every monophyletic group will eventually Marples (1968) presented a classification be found to have important generalizations true based on a cladogram like that in figure 6 and in only of its members, and names for each such which some of the groups (Cribellatae and Ecri- group will presumably be needed. The time when bellatae; Palaeocribellatae and Neocribellatae) are most genera of spiders will be so well known that subordinated and some (the families of Neocri- every monophyletic subgrouping of species has bellatae and Ecribellatae) are sequenced. The important biological meaning is clearly very far cladogram shows that there are monophyletic off, yet at some level inclusive taxa are necessary groups containing all Neocribellatae except Hick- for purposes of communication. For example, 1 977 PLATNICK: HYPOCHILOID SPIDERS 19

Palaeocribellatae q) ~0 C, 'S 'l '9 CN,§ N q) Co0 C,

Araneoclada

FIG. 31. Cladogram of hypochiloid spiders with inclusive taxa indicated. Circled area at lower left represents a single three-taxon statement. given the cladogram shown in figure 31 and de- quently, and (2) as the interrelationships of the fended earlier, and the fact that the rank of non-hypochiloid araneomorphs are worked out, Araneae has been, for our purposes, preset at the we may end up with numerous additional sub- ordinal level, we could construct the following orders. The choice of when to stop sequencing sequenced classification: and shift to a categorical level lower than the suborder appears to be an arbitrary one, and it (5) Order Araneae seems better to have (temporarily) useless names Suborder 1 Mesothelae than not to have names for which there is a need. Suborder 2 Mygalomorphae Those workers who do prefer a sequenced classi- Suborder 3 Hypochilomorphae fication could relimit the Araneomorphae and Suborder 4 Hickmaniomorphae use that name for suborder 7, without great dam- Suborder 5 Gradungulomorphae age to the present concept; they would have to Suborder 6 Thaidomorphae be prepared, howe.ver, to relimit the group again Suborder 7 [all other spiders] when further divisions of the cladogram are es- Note that we must include the number of the tablished. suborder to indicate both that it has been se- Since sequenced classifications omit impor- quenced and that the sister group of any one sub- tant and unimportant inclusive taxa (at least by order is represented by all other suborders with a the standard of current usage) and combined higher number taken together. Two points are systems are unlikely to be understood by the clear: (1) we have lost some inclusive taxa (like non-initiate, subordinated classifications seem Araneomorphae) whose names we now use fre- best for use by comparative biologists in general. 20 AMERICAN MUSEUM NOVITATES NO. 2627

In subordinated classifications, categories (rather ably become increasingly useful if the cladogram than equally ranked taxa) are proliferated. Some is corroborated as our knowledge grows. The cor- authors, notably McKenna (1975), have provided responding new names for individual branches new category names to deal with this problem, (Hickmanithecae, Gradungulospira, and Thaido- but such names are a source of confusion in that clada) will certainly be less used, because they one must always turn back to the classification are redundant with available generic and family to determine which of two ranks (say, sublegion names. Farris (1976) has proposed deleting such and magnorder) is higher. Farris (1976) has pro- redundant names from classifications; they are vided an ingenious mechanism of generating eas- included here because not to do so eliminates ily understood categories, and his system is fol- half of the information content of any taxo- lowed in the classification presented below. The nomic name, i.e., the knowledge that the sister Farris method allows an infinite increase in the group of a given taxon will have the same rank as number of categories by proliferating their pre- that taxon. fixes, of which only eight easily memorizable Finally, to those who may find both the se- forms are used. To avoid nomenclatural compli- quenced and subordinated classifications pre- cations from the International Code, it seems sented here intolerably radical (and certainly best to retain all the new names proposed below both have their drawbacks), I can only reply (1) at the ordinal (rather than familial) level, so that although no previously proposed classifica- single prefix names are used down to the lowest tion is consistent with the cladogram shown in ordinal rank (the picoorder), where the use of figure 31, numerous classifications that are con- double prefix names is initiated. sistent with it are possible; (2) that whether we Some comments on the classification pro- recognize all monophyletic groups is of less im- posed below are in order. The family Ectatostic- portance than whether all the groups that we do tidae Lehtinen is placed as a junior synonym of recognize are monophyletic; and (3) that those the Hypochilidae (NEW SYNONYMY); the sis- workers who prefer to combine subordinated and ter-group relationship between Hypochilus and sequenced systems are therefore free to do so. Ectatosticta can be expressed at the generic level The two classifications presented here for pur- without requiring redundant monotypic family poses of comparison and evaluation can be ig- names; a similar case is the purported sister-group nored with impunity, but the cladogram pro- relationship between Liphistius and Heptathela posed here cannot, because it is only by the use within the Liphistiidae. New names are intro- of such hypotheses of relationship that classifica- duced for three inclusive taxa (Bispermathecae, tions can be falsifiable and that systematics can Tracheospira, and Araneoclada) that will prob- be a science. 1 977 PLATNICK: HYPOCHILOID SPIDERS 21

PROPOSED CLASSIFICATION Superorder Labellata Petrunkevitch, 1949 Order Amblypygi Thorell, 1900 Order Araneae Clerck, 1757 Suborder Mesothelae Pocock, 1892 Family Liphistiidae Thorell, 1869 Suborder Opisthothelae Pocock, 1892 Infraorder Mygalomorphae Pocock, 1892 [other families] Infraorder Araneomorphae Smith, 1902 Microorder Palaeocribellatae Caporiacco, 1938, new rank Family Hypochilidae Marx, 1888 Microorder Neocribellatae Caporiacco, 1938, new rank Gigapicoorder Hickmanithecae, new Family Hickmaniidae Lehtinen, 1967 Gigapicoorder Bispermathecae, new Megapicoorder Gradungulospira, new Family Gradungulidae Forster, 1955 Megapicoorder Tracheospira, new Hyperpicoorder Thaidoclada, new Family Thaididae Lehtinen, 1967 Hyperpicoorder Araneoclada, new [other families]

LITERATURE CITED ders. Ann. Ent. Soc. Amer., vol. 38, pp. 549-5 58, figs. 1-20. Baum, Steffi Comstock, John H. 1972. Zum "Cribellaten-Problem": Die Geni- 1912. The spider book. Garden City, New talstrukturen der Oecobiinae und Uroc- York, 721 pp., 770 figs. teinae (Arach. :Aran. :Oecobiidae). Ab- Coyle, Frederick A. handl. Verhandl. Naturwiss. Ver. Ham- 1971. Systematics and natural history of the burg, vol. 16, pp. 101-1 5 3, figs. 1-66. mygalomorph spider genus A ntrodi- Bonnet, Pierre aetus and related genera (Araneae: An- 1959. Bibliographia araneorum. Toulouse, trodiaetidae). Bull. Mus. Comp. Zool., vol. 2, pt. 5, pp. 4231-5058. vol. 141, pp. 269-402, figs. 1-330. Bristowe, William S. 1 975. Systematics of the trapdoor spider 1933. The liphistiid spiders. With an appendix genus A liatypus (Araneae: Antrodiaeti- on their internal anatomy by J. Millot. dae). Psyche, vol. 81, pp. 431-500, figs. Proc. Zool. Soc. London, 1932, pp. 1-194. 10 1 5-1057, figs. I -1 1, pls. 1-6. Cracraft, Joel 1938. The classification of spiders. Ibid., ser. 1974. Phylogenetic models and classification. B, vol. 108, pp. 285-322, figs. 1-13. Syst. Zool., vol. 23, pp. 71-90, figs. 1-5. Buxton, B. H. Crome, Wolfgang 1913. Coxal glands of the . Zool. 1 95 5. Die Beziehungen zwischen dem dor- Jahrb., suppl. 14, pp. 231-282, figs. salen Zeichnungsmuster und der Meta- 1-7, pls. 1-43. merie des Spinnen abdomens. II. Zool. Caporiacco, Lodovico di Jahrb. Syst., vol. 83, pp. 541-638, figs. 1938. I1 sistema degli Araneidi. Arch. Zool. 1-175. Italiano, vol. 25, pp. 35-155, 1 fig. Dahl, Friedrich Chamberlin, Ralph V., and Wilton Ivie 1901. Uber den Werth des Cribellums und 1945. On some Nearctic mygalomorph spi- Calamistrums fur das System der Spin- 22 AMERICAN MUSEUM NOVITATES NO. 2627

nen. Sitzber. Gesell. Naturf. Fr. Berlin, Abhandl. Verhandl. Naturwiss. Ver. pp. 177-199. Hamburg, vol. 17, pp. 85-116, figs. 1904. Ueber das System der Spinnen (Ara- 1-46. neae). Ibid., pp. 93-120. Lehtinen, Pekka T. Davies, Valerie Todd 1967. Classification of the cribellate spiders 1969. The mature female and male of Gra- and some allied families. Ann. Zool. dungula woodwardi Forster (Arane- Fennici, vol.4, pp. 199-468, figs. 1-524, ae: Hypochilomorphae: Gradungulidae). tables 1-52. Jour. Australian Ent. Soc., vol. 8, pp. McKenna, Malcolm C. 95-97, figs. 1-14. 1975. Toward a phylogenetic classification of 1976. Dardarus, a new genus of amaurobiid the Mammalia. In Luckett, W. P., and spider from eastern Australia. Mem. F. S. Szalay, Phylogeny of the Primates. Queensland Mus., vol. 17, pp. 399-411, New York, pp. 21-46, figs. 1-3. figs. 1-21, pls. 59, 60. Marples, B. J. Farris, James S. 1968. The hypochilomorph spiders. Proc. 1976. Phylogenetic classification of fossils Linn. Soc. London, vol. 179, pp. 11-31, with recent species. Syst. Zool., vol. 25, figs. 1-7. pp. 271-282, figs. 1, 2. Marx, George Forster, Raymond R. 1888. On a new and interesting spider. Ent. 1955. A new family of spiders of the sub- Amer., vol. 4, pp. 160-162, figs. 1-13. order Hypochilomorphae. Pacific Sci., 1889. On a new and interesting spider from vol. 9, pp. 277-285, figs. 1, 2. the United States. Proc. Washington 1970. The spiders of New Zealand. Part III. Ent. Soc.,vol. 1, pp. 166-167. Otago Mus. Bull., no. 3, pp. 1-184, figs. Millot, Jacques 1-530. 1931. Les diverticules intestinaux du cephalo- Forster, Raymond R., and C. L. Wilton thorax chez les Araignees vraies. 1973. The spiders of New Zealand. Part IV. Zeitschr. Morph. Okologie Tiere, vol. Otago Mus. Bull., no. 4, pp. 1-309, figs. 21, pp. 740-764, figs. 1-27. 1-1088. 1933. Notes complementaires sur l'anatomie Gaffney, Eugene S. des liphistiides et des hypochilides, a MS.] An introduction to the logic of phy- propos d'un travail recent de A. Petrun- logenetic reconstruction. kevitch. Bull. Soc. Zool. France, vol. Gerhardt, Ulrich, and A. Kastner 58, pp. 217-235, figs. 1-9. 1938. Araneae. In Kukenthal, Willy, and T. 1936. Metamerisation et musculature abdomi- Krumbach, Handbuch der Zoologie. nale chez les araneomorphes. Ibid., vol. Berlin, vol. 3, pt. 2, pp. 394-656, figs. 61, pp. 181-204, figs. 1-14. 484-854. 1949. Ordre des Amblypyges. In Grasse, Gertsch, Willis J. Pierre-P., Traite de Zoologie. Paris, vol. 1958. The spider family Hypochilidae. Amer. 6, pp. 563-588, figs. 318-348. Mus. Novitates, no. 1912, pp. 1-28, Montgomery, Thomas H., Jr. figs. 1-50. 1909. On the spinnerets, cribellum, colulus, 1964. A review of the genus Hypochilus and a tracheae and lung books of araneads. description of a new species from Colo- Proc. Acad. Nat. Sci. Philadelphia, vol. rado (Araneae, Hypochilidae). Ibid., 61, pp. 299-320, figs. 1-36. no. 2203,pp. 1-14,figs. 1-11. Nelson, Gareth J. Hennig, Willi [MS. I Trees and cladograms. 1965. Phylogenetic systematics. Ann. Rev. Petrunkevitch, Alexander Ent., vol. 10, pp. 97-116, figs. 1-4. 1923. On families of spiders. Ann. New York 1966. Phylogenetic systematics. Urbana, 263 Acad. Sci., vol. 29, pp. 145-180, pls. 1, pp., 69 figs. 2. Kaestner, Alfred 1933. An inquiry into the natural classifica- 1968. Invertebrate Zoology. New York, vol. tion of spiders, based on a study of 2, 472 pp. their internal anatomy. Trans. Con- Kraus, Otto, and Helga Baur necticut Acad. Arts Sci., vol. 31, pp. 1974. Die Atypidae der West-Palaarktis. 299-389, figs. 1-59, tables 1, 2. 1977 PLATNICK: HYPOCHILOID SPIDERS 23

Platnick, Norman I., and Willis J. Gertsch America, Mexico, and the West Indies. 1976. The suborder of spiders: A cladistic Bull. Mus. Comp. Zool., vol. 140, pp. analysis. Amer. Mus. Novitates, no. 129-1 64, figs. 1-77. 2607, pp. 1-15, figs. 1-18. Simon, Eugene Pocock, Reginald I. 1892. Histoire naturelle des Araignees. Paris, 1892. Liphistius and its bearing upon the clas- vol. 1, pt. 1, pp. 1-256, figs. 1-215. sification of spiders. Ann. Mag. Nat. Thorell, Tamerlan Hist., ser. 6, vol. 10, pp. 306-314, figs. 1891. Spindlar fran Nikobarerna och andra 1, 2. delar af Sbdra Asien, etc. K. Svenska 1900. The fauna of British India, including Vetensk.-Akad. Handl., vol. 24, pp. Ceylon and Burma. Arachnida. Lon- 1-149. don, 279 pp. Zapfe C., Hildegard Popper, Karl R. 1955. Filogenia y funcion de Austrochilus 1968. The logic of scientific discovery. New manni Gertsch y Zapfe (Araneae-Hypo- York, 480 pp. chilidae). Trab. Lab. Zool. Univ. Chile, Shear, William A. no. 2, pp. 1-53, figs. 1-60. 1970. The spider family Oecobiidae in North