View metadata, citation and similar papers at core.ac.uk brought to you by CORE providedMagazine by ElsevierR745 - Publisher Connector

Correspondence Figure 1

(a) Most-parsimonious tree % Scored 446 steps characters 6 Africa-to-Eurasia dispersals Australopithecus evolution — in (99.6) 68 (including humans) and out of Africa 80 Chimpanzees (100.0) 58 Gorillas (100.0) 68 (61.2) Comments from 31 Ouranopithecus (36.7) Michael M. Miyamoto* 80 Orangutans (100.0) 56 † (70.0) 80 and Timothy S. Young Lufengpithecus (34.6) 71 (65.8) 59 Stewart and Disotell [1] have contro- (99.6) 56 versially proposed that the hominoid (42.1) lineage dispersed from Africa to (58.8) Eurasia about 20 million years ago, (97.5) Hypothetical and that the common ancestor of outgroup human, chimpanzee and gorilla ≥ ≤ returned to Africa within the past 10 (b) Retained (p 0.126) Australopithecus (c) Rejected (p 0.012) 452 steps 463 steps (including humans) million years. Their model relied on 3 Africa-to-Eurasia Chimpanzees 2 Africa-to-Eurasia dispersals dispersals the phylogeny shown in Figure 1a to Gorillas estimate the minimum number of Dryopithecus migrations under two different sce- Orangutans narios: dispersals only from Africa to Sivapithecus Eurasia (scenario A); or dispersals in Lufengpithecus both directions (scenario B). Ouranopithecus Scenario B is strongly supported by Oreopithecus the phylogeny because it requires Gibbons only two dispersals to explain the Kenyapithecus distributions of extant and fossil Afropithecus hominoids. Scenario A invokes a Proconsul minimum of six dispersals to explain Hypothetical outgroup hominoid distributions. Current Biology This phylogeny was derived by synthesis of molecular data for living Parsimony analyses of 240 morphological have not been considered. Extinct genera characters for extant and extinct hominoids are in italics; arrows track dispersals from hominoids and morphological data [2]. (a) The single most-parsimonious phy- Africa to Eurasia; Eurasian groups are black, for both extant and fossil groups. logeny for these data; this is the same tree African taxa are red, and the same colours The hominoid phylogeny of Begun found by Begun et al. [2] and emphasized by apply to their ancestors under scenario A. In et al. [2] was emphasized, as it incor- Stewart and Disotell [1]. It requires six (a), numbers next to internal branches corre- Africa-to-Eurasia dispersals under scenario spond to bootstrap scores with 2,000 pseu- porated 240 cranial and post-cranial A. (b,c) Shortest phylogenies that require dosamples; parenthetical values indicate the characters for the same extant and three (b) and two (c) dispersals under sce- percentages of morphological characters, fossil genera as considered by nario A (with each of these trees invoking out of 240, scored for each genus. The boot- Stewart and Disotell. Relationships only two dispersals under scenario B). There strap scores provide another measure of are one and seven additional trees of 452 character support for each group [5]. Of among extant hominoids are well and 463 steps, with three and two ‘out of greatest importance with regard to the % established and not in dispute [3], Africa’ dispersals, respectively (the other scored is that all extinct genera, except but many questions persist about the alternative of 452 steps is in Figure 3 of [2]). Australopithecus and Proconsul, are ≤70% positions of the extinct genera, All parsimony analyses were completed with complete, with Kenyapithecus, version 4.0d64 of PAUP*. Post- Lufengpithecus, and Ouranopithecus known because of the fragmentary nature of Australopithecus movements by humans for only 35–42% of their characters. their morphological data (Figure 1). To what extent is the controversial model of Stewart and Disotell based on an unreliable estimate of homi- characters of Begun et al. [2] and parsimony analyses show that these noid phylogeny [4]? have identified all of the shortest are 448, 452, 463, and 492 steps long, To address this question, we have trees with four, three, two or one dis- respectively (2, 6, 17 and 46 steps re-analyzed the 240 morphological persal(s) under scenario A (Figure 1): longer than the most-parsimonious R746 Current Biology, Vol 8 No 21 phylogeny, respectively). The two Comments from the African and human clade, phylogenies of 452 steps, each and perhaps again in the early requiring three dispersals out of Carol Ward ape Morotopithecus. Like the Africa, are not significantly worse apparent parallelisms seen in than the most-parsimonious tree, The model of catarrhine evolution colobines, this supports the hypothe- according to their Templeton tests of presented by Stewart and Disotell [1] sis that evolution produces similar character support (p = 0.126 and highlights the roles of parallelism and morphologies under similar selective 0.227 for the two comparisons, competitive exclusion in shaping regimes in closely related . α = 0.05, two-tailed testing [5]). The adaptations in various lineages. It is same conclusion applies to the four interesting to note that for most of Address: Department of Anthropology and phylogenies of 448 steps, with four the time that monkeys were predomi- Department of Pathology and Anatomical Sciences, 107 Swallow Hall, University of dispersals. The eight trees of 463 nantly terrestrial, were abundant Missouri, Columbia, MO 65211, USA. changes, with two dispersals, are sig- and arboreal. In Africa, about the time E-mail: [email protected] nificantly worse than the most-parsi- colobines became increasingly arbore- monious phylogeny (p = al, the gorilla, chimp and human lin- 0.002–0.012), as is the one tree of eages appeared, and hominids Comments from 492 steps with one dispersal. became terrestrial bipeds. In Asia, Adrienne L. Zihlman* Although these results re-empha- arboreal monkeys appeared about the size that the Stewart and Disotell time fossil apes disappeared. and Jerold M. model is most-parsimonious and Arboreality in its various forms Lowenstein† requires the fewest dispersals, the appears to have evolved at different evidence for this model cannot be times and rates in different lineages. What is missing from the synthesis regarded as conclusive, as there are Old World monkeys were largely ter- reported by Stewart and Disotell [1] trees of insignificantly longer length restrial for most of their evolutionary is a study of the comparative anato- which invoke only one or two extra history, with many returning to the my and locomotor behavior of living migrations under scenario A (e.g. see trees only recently. Re-invasion of an apes, evidence that may further Figure 1b). Retaining these trees arboreal niche seems to have hap- strengthen their hypothesis. forces one to consider whether a sin- pened after the divergence of Asian The paradox in interpreting ape gle inferred migration is enough to and African colobine lineages. Shared evolution is that the Asian apes, convincingly reject scenario A in arboreal traits among colobines, then, belonging to the older hominoid lin- favor of scenario B. must represent independent evolu- eage, are dependent upon continu- The failure of the morphological tionary acquisitions, suggesting that ous forest canopy. Their younger data to distinguish between the mechanical requirements of an arbo- African cousins, chimpanzees and most-parsimonious phylogeny and real lifestyle select for predictable gorillas, are equally at home in the these alternatives can be most easily morphological changes given a simi- trees and on the ground. Knuckle- attributed to the large numbers of lar ancestral form. walking accommodates weight-bear- missing data for most extinct genera Hominoids have always been ing on their long forelimbs and their [3,4]. There remains a critical need predominantly arboreal, even in the massive hindlimbs approach those of for more complete specimens of earliest parts of the lineage. The quadrupedal, terrestrial Old World known fossil genera. With more only fossil hominoid that has been monkeys [7]. complete material, gaps in the mor- interpreted as displaying any appar- As global cooling occurred in the phological matrix can be filled, ent adaptations to part-time terrestri- Miocene, the Asian forests contract- allowing the assignment of these fos- ality is Kenyapithecus [6]. It would ed, the Himalayas rose, and the sil genera to the hominoid phyloge- also be surprising if Proconsul major Tethys Sea shrank. These environ- ny with greater certainty and permit- and Gigantopithecus sp. were not at mental changes probably contributed ting more precise estimates of the least partly terrestrial, due to their to the extinction of Miocene ape number and timing of hominoid dis- gorilla-like body sizes. The rest were genera, such as Sivapithecus in Asia persals ‘out of and into Africa’. generalized, above-branch arboreal and Europe [8]. One of the survivors creatures. Below-branch arboreal may well have been the ancestor of Acknowledgements specializations probably originated the gorilla and the chimp, anatomi- We thank Michele R. Tennant for her sugges- tions about our study and David L. Swofford for with the modern ape clade. Fossil cally equipped to travel permission to use his latest version of PAUP*. evidence suggests, however, that the quadrupedally between discontinu- extreme adaptations seen in extant ous forests and to radiate into new Address: Department of Zoology, University apes were only ‘completed’ indepen- habitats. Presumably the ancestor, of Florida, Gainesville, Florida 32611, USA. E-mail: *[email protected]; dently in the /siamang ances- like the living gorilla and chimp, was †[email protected] tor, in orangutans, at least once in able to survive outside of primary Magazine R747

rain forest; gorillas in fact seem to group turns out to be, the more sup- of detail approaching that with prefer secondary forests, and chim- port there is for the Stewart and which they treat the morphological panzees survive in deciduous forests. Disotell model. On the other hand and molecular data. The more specialized anatomy of David Pilbeam has queried the once gibbons and orangutans, the small seemingly unassailable orangutan Address: Department of Anthropology, C.B. size of the former and the extreme affinities of Sivapithecus [10], and such 233, University of Colorado, Boulder, Colorado 80309-0233, USA. joint flexibility of the latter, has evi- a proposal would argue against the dently evolved during the millions of model. Among the new research that years since they diverged from their may be stimulated by the hypothesis, Caro-Beth Stewart* and common ancestor with each other I would like to see work to determine Todd R. Disotell† respond: and with their earlier and later the real affinities of Asian fossils African relatives. Further discussion including Sivapithecus, Indopithecus is needed on the timing, as a gibbon and Gigantopithecus, and new investi- We thank the authors of the above branch at 13–14 million years ago, gations of the late Miocene fossils of letters for their thoughtful com- rather than 18 million years ago, southern Europe, Graecopithecus and ments, and wish to make a few com- would seem to better accommodate Ankarapithecus, which could probably ments in response. Foremost, we the anatomical data. fit into either scenario. encourage the authors to perform the The weakest link in Stewart and various analyses that they suggest Disotell’s scenario is the fossil Address: Department of Archaeology and here. Given that all of the data record. Although Miocene fossil apes Anthropology, The Australian National matrices used in our analyses have University, Canberra 0200, Australia. exist in Eurasia and Africa, limited been explicitly presented in the lit- evidence of their locomotor anatomy erature, anyone can test various evo- shows little affinity with living apes. Comments from lutionary hypotheses by reanalyzing For the time of the supposed re- these data, recoding them or adding migration into Africa during the past Herbert Covert new characters as they see fit. 8–10 million years, the fossil cup- For example, Miyamoto and board is virtually bare. We neverthe- There are some potential problems Young present an elegant illustration less congratulate Stewart and with Stewart and Disotell’s [1] analy- of how Swofford’s computer pro- Disotell for presenting a well- sis that need to considered before gram, PAUP* [11], can be used to defined and testable model. the results are considered to be com- statistically test various evolutionary pelling. The most vexing is the over- hypotheses about hominoid evolu- Addresses: *Social Sciences 1, University of ly simplistic biogeographic approach. tionary relationships and dispersal California Santa Cruz, Santa Cruz, CA In their Figure 2, such Eurasian patterns. They present a reanalysis 95064, USA. †Department of Nuclear Medicine, California Pacific Medical Center, forms as Oreopithecus, Dryopithecus, of the dataset of Begun et al. [2], the San Francisco, CA 94115, USA. Lufengpithecus, Ouranopithecus and main one we used for our placement E-mail: *[email protected]; Sivapithecus are included. While each of the hominoid fossils in the pri- † [email protected] is known from Eurasian localities, it mate evolutionary tree. We applaud would be more accurate to note that Miyamoto and Young for taking our Comments from some are from Western Europe, oth- study one step further, and showing ers central Europe, one from the how rigorous phylogenetic analysis Colin Groves Indian subcontinent, and one from can be performed to rule out alterna- eastern Asia. In fact, only tive evolutionary hypotheses. They The new model proposed by Stewart Lufengpithecus is known from geo- have set a new standard in the field and Disotell [1] is controversial, but it graphic localities within 1,500 km of of paleoanthropology. is an ideal scientific hypothesis: it is the range of modern Asian apes. Covert rightly suggests that more testable, and it is supported by the To make this analysis more sophisticated biogeographical analy- discovery of new specimens of a meaningful, the authors should ses could be performed to better long-known East African fossil, now divide Eurasia and Africa into three understand the details of primate named Morotopithecus, and its redating or four regions and repeat the analy- dispersal patterns. We chose to focus at more than 20 million years ago. sis. For example, Eurasia might sim- on the simple, traditional dichotomy The new material shows clear affini- ply be divided into Europe, Western between Africa and Eurasia for sev- ties with modern hominoids, in a way Asia and Eastern Asia, and Africa eral reasons. Importantly, the water which Proconsul and Afropithecus do divided into Saharan Africa, East and desert that have separated Africa not, confirming that these two genera Africa, West Africa and southern and Eurasia during most of evolu- form separate lineages [9]. The more Africa. The authors would then be tionary history have presented major diverse the early African hominoid treating biogeographic data at a level barriers to primate dispersal between R748 Current Biology, Vol 8 No 21

these land masses. Thus, the major 8.6 million years ago, as proposed by and computer technologies will allow dispersal events between Africa and Kumar and Hedges [13] based on the acquisition and storage of three- Eurasia are the first that need to be analysis of a few unidentified pro- dimensional images of fossils, making studied before more sophisticated teins. Thus, this key divergence date the primary data widely available. analyses are performed. does not rely solely on the phyloge- Whether multiple dispersals out netic position of Sivapithecus. Addresses: *Department of Biological Sciences, University at Albany, State of Africa should be considered more Groves points out the controversy University of New York, 1400 Washington or less likely than one dispersal out surrounding Sivapithecus, which Avenue, Albany, New York 12222, USA. of and another back into Africa derives from relatively recent discov- †Department of Anthropology, New York depends, in part, on the locomotor eries of a few postcranial bones that University, 25 Waverly Place, New York, New York 10003, USA. behavior of the migrant species, as are thought to belong to this genus. E-mail: *[email protected]; discussed originally by Sarich and Few paleoanthropologists contest the †[email protected] Cronin [12] and here by Ward, detailed likeness of the Sivapithecus Zihlman and Lowenstein. We note face and palate to that of the orang- References that the common ancestor of the utan, but the arm bones display a 1. Stewart C-B, Disotell TR: Primate evolu- tion — in and out of Africa. Curr Biol modern African apes was most likely curious mix of ape and -like 1998, 8:R582-R588. a semi-terrestrial quadruped, as are features that suggest a quadrupedal 2. Begun DR, Ward CV, Rose MD: Events in gorillas and chimpanzees (and young locomotor pattern unlike that seen in hominoid evolution. In Function, Phylogeny, and Fossils: Miocene Hominoid humans). Similarly, the fossil the arboreally-committed orangutans. Evolution and Adaptations. Edited by colobine Mesopithecus is believed to Some authors, notably Pilbeam [10], Begun DR, Ward CV, Rose MD. New York: Plenum Press; 1997:389-415. have been semi-terrestrial. So both have taken these postcranial features 3. Goodman M, Porter CA, Czelusniak J, Page of the ancestral primate groups that to imply that Sivapithecus may not be SL, Schneider H, Shoshani J, et al.: Toward we propose dispersed about 10 mil- uniquely related to the orangutan lin- a phylogenetic classification of primates based on DNA evidence complemented lion years ago were probably capable eage. Importantly, however, parsimo- by fossil evidence. Mol Phylogenet Evol of quadrupedal, terrestrial locomo- ny analysis of the Begun et al. [2] data 1998, 9:585-598. 4. Gibbons A: New study points to Eurasian tion through suboptimal, thinly- matrix indicates that Sivapithecus does ape as great ape ancestor. Science forested environments. branch with orangutans, and with a 1998, 281:622-623. The longer trees that Miyamoto respectively high bootstrap support of 5. Swofford DL, Olsen GJ, Waddell PJ, Hillis DM: Phylogenetic inference. In Molecular and Young present have implications 80% (Figure 1a, above). This place- Systematics, edn 2. Edited by Hillis DM, for the divergence date of the orang- ment suggests that the Sivapithecus- Moritz C, Mable BK. Sunderland, MA: utan and African ape lineages, one of like ancestor was more terrestrial than Sinauer; 1996:407-514. 6. McCrossin ML, Benefit BR: On the rela- the most important and hotly-con- are orangutans, a locomotor behavior tionships and adaptations of tested dates in primate evolution. It that would have more easily allowed Kenyapithecus, a large-bodied hominoid from the Middle Miocene of eastern is often claimed that calibration of their dispersal into southeast Asia. Africa. In Function, Phylogeny and Fossils: this key divergence date relies solely We also agree with these authors Miocene Hominoid Evolution and on the placement of Sivapithecus on that more — and more complete — Adaptations. Edited by Begun DR, Ward CV, Rose MD. New York: Plenum Press; the orangutan lineage, but this is not fossil primates need to be included 1997:241-268. true. Any hominoid fossils that clus- in computer-aided parsimony analy- 7. McFarland RK, Zihlman AL: Body compo- sition in male chimpanzees, P. paniscus ter with either the African ape or ses. Sadly, more complete fossils do and P. troglodytes: a preliminary report. orangutan clade are relevant to this exist for some of the genera shown in Amer J Phys. Anthrop suppl. 1996, 22:177. divergence date, as we previously these trees, but they have not yet 8. Laporte L, Zihlman AL: Plates, climates, and hominoid evolution. So Af J Sci explained. In Figure 1b above, for been made available for such analy- 1983, 79:96-110. example, Dryopithecus and ses. Due in part to logistical difficul- 9. Gebo D, MacLatchy L, Kityo R, Deino A, Ouranopithecus are found to cluster ties, some paleontologists do not Kingston J, Pilbeam D: A Hominoid genus from the Early Miocene of Uganda. with the orangutan clade, rather than allow access to ‘their’ fossils by many Science 1997, 276:401-404. with the African apes, as they do in other researchers in the field; nor do 10. Pilbeam D: Genetic and morphological records of the Hominoidea and hominid the most parsimonious tree (Figure all present explicit data matrices that origins: a synthesis. Mol Phylogenet Evol 1a above); either of these placements can be reanalyzed, challenged or aug- 1996, 5:155-168. suggests that the African ape and mented. These situations are unac- 11. Swofford D: PAUP* (Phylogenetic Methods Using Parsimony and Other orangutan lineages diverged before ceptable in modern science. We call Methods). Sunderland, MA: Sinauer; 1998. the earliest appearance of these fossil for a change in the culture of pale- 12. Sarich VM, Cronin JE: Molecular system- hominoids. Indeed, the placement of oanthropology, such that all relevant atics of the primates. In Molecular Anthropology. Edited by Goodman M, any of these Eurasian hominoid fos- specimens can be examined by other Tashian RE, Tashian JH. New York: Plenum sils on either the orangutan or African researchers and all data used to infer Press; 1976, 141-170. 13. Kumar S, Hedges SB: A molecular ape lineages precludes this key diver- evolutionary hypotheses must be timescale for vertebrate evolution. Nature gence date from being as recent as explicitly presented. Modern imaging 1998, 392:917-920.