AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 601103-107 (1983)

Tale of Tails: Parallelism and Prehensility

ALFRED L. ROSENBERGER Department of Anthropology, University of Illinois at Chicago, Chicago, Illinois 60680 KEY WORDS Tails, Platyrrhines, Prehensility, Parallelism, Phylogeny, Homology

ABSTRACT The occurrence of prehensile tails among only five platyrrhine genera-Cebus, Alouatta, Lagothrix, Ateles, and Bmchyteles-might be inter- preted as evidence that these are a closely related, possibly monophyletic group. In the absence of behavioral data, it is impossible to test whether all possess equivalent biological roles; such would lend credence to the idea that their tails evolved from an homologous, derived character complex. Contrariwise, the ten- dency for species of Cebus to have “averagely” proportioned or relatively short tails, in contrast to the relatively elongate tails of howlers and other atelines; osteological differences in caudal and sacral morphology; and a lack of ateline- like tailheocortex correlates in Cebus, all imply that prehensility has evolved twice in parallel: once (homologously) in atelines and again in capuchins. Despite the vague claims of yesterday’s genera present tails that are functionally and texts and taxonomies, which often stated or equally prehensile. There also seems to be implied that the tails of the nonclawed New some confusion in the literature concerning World monkeys are (‘nearly always of consid- the degree of morphological similarity shared erable length and frequently prehensile” by Cebus and atelines, especially regarding (Dollman, 1933), we now know that the grasp- functionally relevant features such as relative ing tail is limited to a handful of genera. Only tail length. Ankel (19721, for example, empha- in Cebus, Alouattu, Lagothrix, Ateles, and sized that both ateline and capuchin tails are Bmchyteles are tails capable of a finger-like long, whereas Napier (1976) distinguished Ce- clasping that one could call “prehension.” This bus monkeys on account of their relatively excludes several other candidates, like Sai- short tails. Inferential inconsistencies and miri and Callicebus, which commonly engage clear-cut anatomical differences notwithstand- their tails in behaviors such as body wrapping ing, the restricted taxonomic distribution of or tail twinning, respectively, but apparently semi- and fully prehensile tails among the pla- without the degree of coordination-and cer- tyrrhines poses one of the classic problems in tainly sensation-manifest in the ateline forms. New World monkey systematics (Rosenber- (Following Rosenberger (1981), ceboids are re- ger, 1981). Of course, tail prehensility must be classified into Families Cebidae, including Ce- a derived characteristic among primates, and binae and Callitrichinae, and Atelidae, with it seems that many early workers drew the Pitheciinae and Atelinae, including Alouatta.) obvious conclusion that this demonstrated a Although some modern authors have consis- close relationship between Cebus and the Ate- tently distinguished the tail of Cebus from linae. Although other interpretations of Cebus that of atelines on account of the latter’s ven- affinities (e.g., Simpson, 1945; Hill, 1960; Ro- tral skin-covered sensory pad, occasionally la- senberger, 1979) cast doubt upon such a con- beling the capuchin condition “semiprehen- clusion, the issue has not been adequately sile” and the ateline pattern (“fully”) “prehen- resolved on morphological grounds, and it con- sile” (e.g., Napier, 1976; Hershkovitz, 1977), tinues to erroneously influence scenarios of we still lack a comprehensive behavioral study platyrrhine evolution (eg., Hershkovitz, 1977), of tail use in any of these taxa. Thus the evi- dence does not justify the frequent claim that apart from contrastive tactile abilities all five Received April 12,1982accepted August 3,1982

0002-9483/83/6001-0103$02.000 1983 ALAN R. LISS, INC. 104 A.L. ROSENBERGER even when couched in gradistic terms (Rosen- There is also some evidence that the relative berger, 1980). The purpose of this brief note is external length of ateline tails is generally to bring together some observations that sug- quite long and includes a comparatively large gest that the grasping abilities of the tails of number of caudal vertebrae, whereas that of Cebus and the atelines have evolved indepen- Cebus is reduced in both respects. In a bivar- dently and in parallel. I attempt to demon- iate plot (Fig. 1) of log head and body length strate this by showing that significant versus external tail length for a spectrum of morphological differences exist and combine platyrrhines, the cluster of capuchin species to form two different patterns of organization, tends to exhibit tails that are of average or one in Cebus and the other shared in common short length by comparison with ceboids of by Alouatta, Lagotrhix, Ateles, and Brachy- roughly similar body size or of the entire ar- teles. Like all arguments designed to refute ray. This appears to contrast the proportions hypotheses of shared homology, this one is of atelines-especially Lagothrix, Ateles, and largely negative and, when taken alone, far Brachyteles. These data are applied to a num- from foolproof. However, in conjunction with ber of models of interspecific scaling to con-o- positive evidence on the phylogenetic relation- borate this observation, although I emphasize ships among the animals in question (e.g., Ro- that such manipulations can support only the senberger, 1979)) it seems to be the most most general conclusions. Using logarithmic reasonable and robust interpretation of many regressions of sex-pooled means for 26 platyr- homologous character distributions and nu- rhine species, specified in Figure 1, I have merous contrasting morphologies. generated predicted lengths for capuchin and To my knowledge, Ankel’s (e.g., 1972) work ateline tails. For species of Cebus, the ob- represents the most complete discussion of served mean lengths tend to be somewhat less the osseous anatomy of platyrrhine tails. She than the predicted values, whereas the con- distinguished the ateline caudal region by a verse is true of atelines, particularly Ateles, number of features: eight proximal elements, Brachyteles, and Lagothrix, by a wide margin. each of which is relatively short and bears For all the included taxa, the pattern and lumbar-like articulations and elevated neural magnitude of deviations from the least-squares arches; a distally positioned longest-tail ver- regression lines are consistent across five of tebra; short and flattened terminal elements; the six models (Table l), whether devised or differentially enlarged, and rearranged, dorsal along taxonomic lines to reflect phylogeny or and ventral muscle masses. Together with a upon functional criteria to eliminate biases relative increase in the size of the sacral canal that might arise when comparing forms hav- and the external tactile pad, these traits are ing widely divergent caudal adaptations. Two thought to enhance the flexibility, proximal additional models, based exclusively on 1) atel- articular integrity, power, and nervous coor- ids and 2) nonprehensile-tailed cebids plus dination of the tail. Ankel (see also German, atelids, are not shown because their low cor- 1981) stated and implied that the Cebus mor- relation coefficients (0.15; -0.06) would make phology is nearer that of atelines, and “inter- similar inferences moot at best. Those regres- mediate” relative to the nonprehensile tails of sion models with the highest correlation coef- other platyrrhines in almost all ways. Still, she ficients, models 11, 111, and IV, which avoid pointed to the following discrete characters as the necessity of accounting for the great vari- non-ateline: six proximal elements, all with ance encountered among the heterogeneous standard caudal zygopophyses and without atelines, are of particular interest. Models I1 evidence of an abbreviated length; and ab- and IV suggest that by comparison with un- sence of flattened terminal elements. Addi- specialized cebids or ceboids, for an animal of tionally, the index of sacral canal size is lower its “size”-i.e., head-and-body length-Cebus in Cebus (see below) and the appendage lacks has a tail of “average” or expected length. a friction skin. Thus, added to the well-known Model IV, which provides a specifically cebid hirsute versus glabrous difference are osseous phylogenetic standard of interspecific allome- features that probably relate to caudal dexter- try, emphatically indicates that Cebus tails ity, curling potential, and articular strength- are relatively shorter than any of its pre- i.e., a combination of continuous and discrete sumed closest relatives (e.g., Rosenberger, features of probable functional and behavioral 1979, 1981). This same regression framework significance. It seems reasonable to presume also gives the impression of atelines having that these involve both acrobatic and strictly normally proportioned or even reduced (e.g., tactile biological roles. Alouattu) tails. However, the discrepancy be- PLATYRFWINE PREHENSILE TAILS 105

- Atelel gwffroyi 6.60 Brschyulel arachnoides A Lagothrix lagothricha

bslzebu’ Atsles belzehuth A t A Alouatta rsniculus Abuam palluu 6.30- Alouana fusca A

0 Cdliabut torquaus CJllicObus moloch 0 Cebur slbifronr 0 0 caZEZmonhut - SIimiri xiureus 0 Cebw wells 6.00 s.oUinus mdas 0 Pithecia pithecia Chiropotsssatanur 0 Aotur trivirgatus

0 Lantopithscus roula &~IMI5fUXlWlli5 0 5.70 - 0 Callimico paoldti callithrix pcchus 0

0 Pitheciines 1 OCebids 1 5.40 - Callithrix pymmu

Cacajaa rubicundus 5.10 0 0 Cacalaocalvus

I 5.10 5.40 5.70 6.00 6.30 Log Head and Body Length

Fig. 1. Scatter diagram showing the relationship be- ters, and relative measures of tail length based upon tween adult head and body length, a measure of body size, regression predictions. The visual impression that the Ce- and external tail length in a sample including all living bus species have “averagely” proportioned or relatively ceboid genera. Points represent sex-pooled species means. short tails, in contmst to the atelines, is confiied for both See Table I for data sources, regression models, parame- functional and phyletic subsamples.

tween the predictions generated by model IV least elongate in Alouatta. Finally, to stretch versus all others is so great in this particular this approach to its limits, extrapolating a log- cross-phyletic comparison, which also involves arithmic regression of head and body length a huge body size disjunction essentially lack- versus average number of caudal vertebrae in ing any range overlap, that the extrapolation nonspecialized platyrrhine genera (Callithrix, to atelines has dubious value if it is at all Callimico, Saimiri, Callicebus, Pithecia, justifiable. All in all, it appears that atelines Aotus; data from Schultz (1961); r=0.63, do have relatively long tails, although it is exp. = -0.115, int. =3.91), in analogy to model 106 A.L. ROSENBERGER

TABLE 1. Ratio between observed mean length and predicted tail length for semiprehensile and prehensile-tailed adult ceboids, sex-pooled’ Models SDecies I I1 I11 IV V VI Cebus capucinus 0.98 0.95 0.77 0.99 0.87 Cebus ulbifiuns 1.07 1.04 0.85 1.07 0.92 Cebus upella 0.98 0.95 0.78 0.98 0.85 Alouatta belzehl 1.22 1.18 1.13 0.89 1.25 Alouatta &sea 1.13 1.09 1.05 0.84 1.15 Alou&tta saieulus 1.19 1.14 1.10 0.86 1.22 Alouatta palliata 1.19 1.14 1.10 0.86 1.21 Ateles belzehth 1.30 1.25 1.21 0.98 1.32 1.22 Ateles geoffroyi 1.43 1.45 1.40 1.10 1.51 1.51 Brachyteles urachnoides 1.40 1.35 1.29 1.00 1.45 1.45 Laoothrix laaothricha 1.38 1.33 1.29 1.03 1.33 1.33

’Model I-All eeboids. N = 26 int = 2.66 exp = 0.57; r = 0.51. Model 11-Unspeciali ceboids (excludes atelines, Cebus, Cacajao). N = 13int = 2.46; exp = 0.60; r = 0.92. Model 111- All cebids. N = 10; int = 2.26; exp = 0.W r = 0.95. Model IV-Unspecialized cebids (excludes Cebus). N = 7; int = 0.93; exp = 0.89; r = 0.97. Model V-Unspecialized atelids (excludes atelines, Cacujuo). N = 6 int = 3.05; exp = 0.60; r = 0.59. Model VI-Unspecialid atelids plus Alouattu. N = 10; int = 5.86 exp = 0.06; r = 0.56. Data from Kellogg and Goldman, 1944; Napier, 1976 Hershkovitz, 1977; Rosenberger, pew. obs.

11, similarly overestimates the count for Ce- region, which again alleges intermediacy be- bus (observedlexpected = 0.95) but underes- tween nonspecialized and fully prehensile tails. timates the values for ALouatta (l.l2), Lu- Morphological and neurological studies of gothrix (1.08), and Ateles (1.28). Thus the same platyrrhine brains and endocasts supply still contrast between Cebus and atelines obtains another basis for evaluating the evolution of when the number of caudal segments is prehensile tails, and lend support to the prop- considered. ositions developed here. Cortical mapping In spite of the limitations of the data used in studies reviewed by Radinsky (1972) have this examination, it does seem that, on the shown that, in Ateles, caudal sensorimotor whole, the tail of Cebus is not relatively long, representation has shifted from its primitive and may be slightly short for a platyrrhine or position within the great sagittal sulcus to a cebid of its linear dimensions, confirming Na- more lateral location on the dorsolateral sur- pier’s (1976) observation. In contrast, the tail face of the brain. Following this work, Radin- of ALouatta is at least moderately elongate, sky suggested that several topographic and those of Lagotrhix, Ateles, and Brachy- features of the sulcul pattern of Ateles, such teles are relatively very long. Thus the evi- as the confluence of the Sylvian and the intra- dence of tail proportions argues against the parietal sulci, which are also clearly shown in notion that the semiprehensile tail of Cebus is Brachyteles and Lagothrix (e.g., Falk, 19Sl), intermediate, phyletically and functionally, be- are correlated with this shift and expansion. tween ateline and nonateline platyrrhines. A similar condition occasionally seen in Sai- Moreover, these data imply that in at least miri is demonstrably not functionally related some species of Cebus tails are perhaps auta- to homologous expansion of the tail’s sensori- pomorphically reduced in length (assuming motor area (see Falk, 1981). It is noteworthy that Cacajao is monophyletically related else- that the poorly folded ALouatta brain, which where, which seems doubtless true). Further- is quite small relative to its body size (e.g., more, the anomalously short capuchin tail may Gould, 19751, nonetheless shares this as well partially account for its rather low rank on as other cortical features that are uniquely Ankel’s (1972) scale of sacral canal indices: its derived and possibly related to tail prehensil- value of 80.7 is hardly “intermediate” be- ity with other members of the Atelinae (see tween that of Saimiri (77.1)and atelines (99.4- Hershkovitz, 1970; Falk, 1981). Thus the pre- 121.0). The low index is perhaps better ex- hensile complex common to all atelines may plained as a correlate of tail shortening (andl be seen to extend consistently from the os- or lesser body size?) rather than by an implicit teology and integument of the tail itself to its appeal to the absence of an ateline-like hyper- neurological manifestation. The important innervation and vascularization of the caudal point here is that in the relatively enlarged PLATYRRHINE PREHENSILE TAUS 107 (e.g., Clutton-Brock and Harvey, 1980) and pare this article. NSF grant BNS 8108359 par- elaborate brain of Cebus, there is no conflu- tially supported the research. ence of Sylvian and intraparietal fissures; even in Alouatta, where the cortex is poorly folded, LITERATURE CITED the converse is true. Ankel, F (1972) Vertebral morphology of fossil and extant In conclusion, it appears that Cebus differs primates. In R H Tuttle (ed): The Functional and Evolu- from Alouattu, Lagothrix, Ateles, and Brach- tionary Biology of Primates. Chicago: Aldine, pp. 223- yteles in a host of features relating to the 240. Clutton-Brock, TH, and Harvey, PH (1980) Primates, brain morphology of the tail, its proportions, and and ecology. J. Zool. Lond. 190:309-323. cortical correlates. This implies that the semi- Dollman, G (1933) Primates, ser. 3. London: British Mu- prehensile faculties of Cebus are not homolo- seum (Natural History) Booklet, Trustees of the British gous, and may have evolved independently Museum. Falk, D (1981) Comparative study of the endocranial casts from that of atelines. The confirmation of this of New and Old World monkeys. In R.L. Ciochon and hypothesis requires another data set and the A.B. Chiarelli (eds): Evolutionary Biology of New World demonstration that Cebus is phyleticdly re- Monkeys and Continental Drift. New York: Plenum lated to a separate clade; the evidence strongly Press, pp. 275-292. favors a sister-group relationship between Ce- German, R (1981) Functional morphology of New World bus and Saimiri (e.g., Rosenberger, 1979). In monkey tails. Am. J. Phys. Anthropol. 54:224. Gould, SJ (1975) Allometry in primates, with emphasis on terms of relative tail length, which marks a scaling and the evolution of the brain. Contrib. Primatol. shared specialization of the ateline tail com- 5t244-292. plex, Cebus may have evolved in the opposite Hershkovitz, P (1970) Cerebral fissural patterns in platyr- direction, toward autapomorphic abbrevia- rhine monkeys. Folk Primatol. 13t213-240. tion. The latter point, in particular, diminishes Hershkovitz, P (1977) Living World Monkeys (Platyrrhini) With an Introduction to Primates, Vol. 1. Chicago: Uni- the argument against parallelism-Le., that versity of Chicago Press. the tail complex of Cebus is merely at a more Hill, WCO (1960) Primates. Comparative Anatomy and primitive “stage” than an ateline such as Al- Taxonomy. Vol. IV. Cebidae. Part A. Edinburgh: Univer- wttu-because howlers share with the other sity Press. atelines all of the other morphological pecul- Kellogg, R, and Goldman, EA (1944) Review of the spider monkeys. Proc. U.S. Nat. Mus. 96:l-45. iarities of prehensility even without being Napier, PH (1976) Catalogue of Primates in the Rritifih comparably elongate. One would thus be Museum (Natural History). Part 1: Families Callitrichi- forced to argue that Cebus presages atelines dae and Cebidae. London: British Museum (Natural His- behaviorally but not morphologically-a pat- tory). ently unacceptable line of phyletic reasoning. Radinsky, L (1972) Endocasts and studies of primate brain evolution. In R Tuttle (ed): The Functional and Evolu- It is hoped that studies of tail use in locomotor tionary Biology of Primates. Chicago: Aldine-Atherton, contexts will clarify more aspects of this prob- pp. 175-1&2. lem and explain why platyrrhines may be Rosenberger, AL (1977) Xenothrir and ceboid phylogeny. preadapted to the parallel evolution of prehen- J. Hum. Evol. 5t461-481. sility, as I have elsewhere suggested (Rosen- Rosenbeger, AL (1979) Phylogeny, evolution and classifi- cation of New World monkeys (Platyrrhini, Primates). berger, 1977). Nevertheless, now that the PhD dissertation. Ann Arbor: University Microfilms. question of nonhomology has been specifically Rosenberger, AL (1980) Gradistic views and adaptive ra- addressed, amplifying the concern of many diation of the platyrrhine primates. Z. Morphol. Anthro- twentieth-century systematists (Rosenberger, pol. 71:157-163. 1981), it behooves us to consider once again Rosenberger, AL (1981) Systematics: The higher taxa. In AF Coimbra-Filho and RA Mittermeier (eds): Ecology the evidence of Cebus affinities and the ration- and Behavior of Neotropical Primates, Vol. 1. Rio de ale for taxonomically separating Alouatta from Janeiro: Academia Brasileira de Ciencias, pp. 9-27. the other “fully” prehensile-tailed monkeys. Schultz, AH (1961) Vertebral column and thorax. Primato- logia 4(5):1-66. ACKNOWLEDGMENTS Simpson, GG (1945) The principles of classification and a classification of mammals. Bull. Am. Mus. Nat. Hist. I thank Eric Delson for his comments, and 85t1-350. A. Muskin and L. Perlini for helping me pre-