jolly - papionin monkeys & human evolution - 01

A Proper Study for Mankind: Analogies From thePapionin Monkeys and Their Implicationsfor Human EvolutionClifford J. Jolly

Department of Anthropology, New York University, New York, New York 10003

KEY WORDS human evolution; evolutionary analogies; baboons; Papionini; hominins

ABSTRACT This paper’s theme is that analogiesdrawn from the cercopithecine tribe Papionini, especiallythe African subtribe Papionina (baboons, mangabeys, andmandrills), can be a valuable source of insights about theevolution of the human tribe, Hominini, to complementhomologies found in extant humans and/or African apes.Analogies, involving a “likeness of relations” of the form“A is to B, as X is to Y,” can be usefully derived fromnonhomologous (homoplastic) resemblances in morphol-ogy, behavior, ecology, or population structure. Pragmat-ically, the papionins are a fruitful source of analogies forhominins because they are phylogenetically close enoughto share many basic attributes by homology, yet farenough that homoplastic modifications of these featuresare easily recognized as such. In “The Seedeaters,” ananalogy between Theropithecus among baboons and Aus-tralopithecus africanus among hominines was the sourceof a widely discussed (and often misrepresented) diet-based scenario of hominin origins that explained previ-ously unassociated hominin apomorphies, interpretedbasal hominins as nonhuman rather than prehuman pri-mates, and accommodated a basal hominin adaptive radi-ation of at least two lines.

Current usage recognizes an even more extensive evo-lutionary radiation among the basal hominins, originat-ing no earlier than about 7 ma, with multiple lineagesdocumented or inferred by 2.5 ma. Although multilin-eage clades (especially the Paranthropus clade) withinthis complex are widely recognized, and emerge fromsophisticated, parsimony-based analyses, it is sus-pected that in many cases, developmental or functionalhomoplasies are overwhelming the phylogenetic signalin the data. The papionin analogy (specifically the split-ting of the traditional, morphology-based genera Cerco-cebus and Papio mandated by molecular evidence) illus-trates the power of these factors to produce erroneouscladograms. Moreover, the rapid deployment of basalhominins across varied African habitats was an idealscenario for producing morphologically undetectable ho-moplasy. There seems to be no foolproof way to distin-guish, a priori, homologous from homoplastic resem-blances in morphology, but one pragmatic strategy is toseverely censor the datset, retaining only resemblancesor differences (often apparently trivial ones) that cannotbe reasonably explained on the basis of functional re-semblance or difference, respectively. This strategy mayeliminate most morpological data, and leave many fossiltaxa incertae sedis, but this is preferable to unwar-ranted phylogenetic confidence.

Another source of phylogenetic uncertainty is the pos-sibility of gene-flow by occasional hybridization betweenhominins belonging to ecologically and adaptively distinctspecies or even genera. Although the evidence is unsatis-factorily sparse, it suggests that among catarrhines gen-erally, regardless of major chromosomal rearrangements,intersterility is roughly proportional to time since clado-genetic separation. On a papionin analogy, especially thecrossability of Papio hamadryas with Macaca mulatta andTheropithecus gelada, crossing between extant homininegenera is unlikely to produce viable and fertile offspring,but any hominine species whose ancestries diverged lessthan 4 ma previously may well have been able to producehybrid offspring that could, by backcrossing, introducealien genes with the potential of spreading if advanta-geous. Selection against maladaptive traits would main-tain adaptive complexes against occasional genetic infil-tration, and the latter does not justify reducing thehybridizing forms to a conspecific or congeneric rank.Whether reticulation could explain apparent parallels inhominin dentition and brain size is uncertain, pendinggenetic investigation of these apparently complex traits.

Widespread papionin taxa (such as Papio baboons andspecies-groups of the genus Macaca), like many such or-ganisms, are distributed as a “patchwork” of nonoverlap-ping but often parapatric forms (allotaxa). Morphologi-cally diagnosable, yet not reproductively isolated, mostallotaxa would be designated species by the phylogeneticspecies concept, but subspecies by the biological speciesconcept, and use of the term “allotaxa” avoids this incon-sistency. A line of contact between allotaxa typically coin-cides with an ecotone, with neighboring allotaxa occupy-ing similar econiches in slightly different habitats, andoften exhibiting subtle, adaptive, morphological differ-ences as well as their defining differences of pelage. “Hy-brid zones,” with a wide variety of internal genetic struc-tures and dynamics, typically separate parapatricallotaxa. Current models attribute the formation andmaintenance of allotaxa to rapid pulses of population ex-pansion and contraction to and from refugia, driven bylate Neogene climatic fluctuations. An overall similarity indepth of genetic diversity suggests that papionin taxasuch as Papio baboons, rather than extant humans, maypresent the better analogy for human population struc-ture of the “prereplacement” era. Neandertals and Afro-Arabian “premodern” populations may have been analo-gous to extant baboon (and macaque) allotaxa:“phylogenetic” species, but “biological” subspecies. “Re-placement,” in Europe, probably involved a rapidly sweep-

YEARBOOK OF PHYSICAL ANTHROPOLOGY 44:177–204 (2001)

© 2001 WILEY-LISS, INC.DOI 10.1002/ajpa.10021

ing hybrid zone, driven by differential population pressurefrom the “modern” side. Since the genetic outcome of hy-bridization at allotaxon boundaries is so variable, theproblem of whether any Neandertal genes survived thesweep, and subsequent genetic upheavals, is a purely em-pirical one; if any genes passed “upstream” across themoving zone, they are likely to be those conferring localadaptive advantage, and markers linked to these.

In general, extant papionin analogies suggest that thedynamics and interrelationships among hominin popula-tions now known only from fossils are likely to have beenmore complex than we are likely to be able to discernfrrom the evidence available, and also more complex thancan be easily expressed in conventional taxonomic termi-nology. Yrbk Phys Anthropol 44:177–204, 2001.© 2001 Wiley-Liss, Inc.

TABLE OF CONTENTS

Homology, Homoplasy, and Analogy ..................................................................................................................... 179Hominin Origins, Theropithecus, and “Seed-Eaters” ........................................................................................... 181Homoplasy, Cryptic Symplesiomorphy, and the Diversity of Early Hominins .................................................. 183Rheboons, Geboons, and Early Hominin Reticulations ........................................................................................ 188Hybrid Zones and Population Replacements in Baboons and Humans ............................................................. 193Discussion and Conclusions ................................................................................................................................... 200Acknowledgments .................................................................................................................................................... 201Literature Cited ...................................................................................................................................................... 201

To know thyself, look not to apes alone;A proper study for mankind’s—baboon . . .Alexander Papyoe (with apologies to Alexander Pope)

This paper is expanded from a luncheon talk pre-sented to the American Association of Physical An-thropologists in San Antonio, April, 2000. Its titleand theme recognize the meetings’ host: the South-west Foundation for Biomedical Research, and theoutstanding contribution that the Foundation’s ba-boons and scientists have made to anthropological,primatological, and medical knowledge for over 30years. The title of course parodies Alexander Pope’sdictum, and its conceit was that a hitherto unrecog-nized papionin contemporary—“Alexander Pa-pyoe”—claimed equal relevance for his taxon. It isused here as a peg on which to hang some thoughtsabout the use of analogies in understanding humanevolution.

It was also during the San Antonio meetings thatwe learned of the death of Professor SherwoodWashburn, who, among many other contributions toour field, was among the first to appreciate the rel-evance of the large terrestrial monkeys to under-standing human evolution’s earlier stages. This pa-per is therefore dedicated to Sherry Washburn—with respect, affection, and sadness that we cannotenjoy his response to its contents.

Its theme is that the members of the tribe Pap-ionini (baboons, macaques, mangabeys, drills, andtheir extinct relatives), and particularly its Africansubtribe, Papionina (all but macaques), can bringfresh insights to the interpretation of the diversity,adaptations, ecology, and population structure ofspecies within our own lineage, the tribe Hominini(� family Hominidae in the older convention). Be-cause a comparable argument presented just over 30years ago (Jolly, 1970a, 1972a,b) stirred the wrath of

some troglodytophiles, it should be emphasized atthe outset that nothing in this paper is intended todetract from the unique insights that can be drawnfrom studies of the living hominoids, and especiallythe extant nonhuman Homininae (“hominine apes”),Pan and Gorilla. Its thesis is simply that there isinformation that is distinct from and ancillary tothese—insights that arise not from homology butfrom analogy, using resemblances and differencesthat have evolved in a different clade, rather thanthose resulting from common ancestry.

Alexander Papyoe suggests that papionins, espe-cially the African papionins (subtribe Papionina),are likely to be a particularly fruitful source of use-ful analogies for hominin evolution: not simplyhominin origins, but also diversity, paleoecology,and the population structure of all hominins. Unlikethe hominine apes, which are just three species oftwo genera, the African papionins present a complexphylogenetic picture, and include adaptive arrays ofwidely different ages, so that a variety of evolution-ary phenomena is exemplified by extant as well asextinct taxa. Also unlike the African apes, which asfar as is known have always been primarily ever-green forest dwellers, baboons of several generahave shared nonrainforest habitats in sub-SaharanAfrica with the hominins ever since their respectivelineages emerged. Late Neogene climatic and bioticfluctuations that affected hominin distribution, di-versity, and adaptations impacted the baboons inparallel ways, rather than inversely, as was presum-ably the case for forest-dwelling apes. We can there-fore expect the papionins to provide bio-historical aswell as phylogenetic and functional analogues.

Finally, most extant papionin lineages are nowfirmly anchored phylogenetically by multiple lines ofmolecular evidence (Disotell, 2000), and a reason-able, if approximate, timescale for the whole radia-

178 YEARBOOK OF PHYSICAL ANTHROPOLOGY [Vol. 44, 2001

tion is provided by calibrating molecular diver-gences (Tosi et al., 2000; Disotell, personalcommunication) against the paleontologically docu-mented first appearance of Theropithecus apomor-phies at about 4.5 ma (Kalb et al., 1982; Delson,1993; Gundling and Hill, 2000). The overall pap-ionin chronology is comparable to that of the homi-nines (Pan-Gorilla-Homo), while the African sub-clade (Papionina) seems to have diversified withinmuch the same time frame as the hominin subclade(Ruvolo et al., 1991; Disotell, personal communica-tion).

After a short examination of the concept of anal-ogy, the paper focuses on four areas of recurrentcontention where a papionin analogy may make acontribution: 1) the origins of the hominin ancestrallineage, and specifically, the nature of the first hu-man ancestor not shared with any other living pri-mate; 2) homoplasy and the cladistics of “basal” Plio-Pleistocene hominins; 3) the probability ofoccasional hybridization and gene flow betweenbasal hominin lineages; and 4) the status of thevarious populations of “archaic” and “premodern”Homo sapiens, i.e., the Neandertals and their con-temporaries. Each of these topics is of course muchtoo extensive to be comprehensively reviewed in asingle paper. The most I can hope to do is simplyindicate some areas within them where AlexanderPapyoe—as spokesmonkey for the baboons—cansuggest fresh viewpoint.

HOMOLOGY, HOMOPLASY, AND ANALOGY

Although the distinction between homology andanalogy is commonly regarded as one of the funda-mentals of evolutionary theory, both concepts can beslippery, and have stimulated a considerable body ofliterature whose analysis is far beyond this essay’sscope. The following brief discussion of usages isscarcely original, but will serve to define this paper’sstance, which reflects a strong nominalist bias and adistrust of reifications.

In an evolutionary context, homology describes anattribute of individuals (either actual individual or-ganisms, or abstracted representations of a taxon).Structures are often described as homologous (as in“the wing of bats is homologous to the human arm,”or “the wing of bees is not homologous to the wing ofbats”), but this usage can be confusing (is a bat’swing homologous to a pterodactyl’s?). Ambiguity isreduced if the term homology is applied not to struc-tures, but to states of a variable character (“‘posses-sion of a trunk with fore- and hindlimbs’ is a homol-ogous character-state of chiropterans and humans”).A homologous character-state is one whose sharedstatus derives from inheritance from a common an-cestor. (Inthiscontext, thedistinctionbetweensynap-omorphies and symplesiomorphies is irrelevant;both are equally homologous, differing only withrespect to the ancestor from which the shared stateis derived.) This definition of homology, of course,presupposes a known cladogram, and this means

that homology cannot be used a priori to sort rele-vant from irrelevant characters for cladistic analy-sis, although we can distinguish characters that arelikely to be homologous from those that are not. Thedefinition also ignores some fundamental issues,such as the weaselish attributes of verbal character-state descriptions, and the difficulty of defining “in-heritance” in any but a genetic context (and notunambiguous, even there) (Cartmill, 1994).

If “homology” is used in this way, then its ant-onym is “homoplasy,” a shared character-state thatis not derived from common ancestry (Lockwood andFleagle, 1999). A homologous resemblance resultsfrom a common ancestry in which the shared char-acter-state appeared by a single evolutionary tran-sition; in the ancestry of a homoplastic resemblancethere are at least two such transitions.

Analogy is not identical to homoplasy, although italso involves nonhomologous resemblances. Accord-ing to the Oxford English Dictionary, analogy im-plies an “equivalency or likeness of relations,” whichis classically stated in the form “A is to B, as X is toY” (thus, A is the analogue of X). For example, in hisdiscussion of the relationship between “higher men-tal powers” and cerebral development in human an-cestry, Darwin (1871, p. 54) points out that:

“We meet with closely analogous facts in the insects, for in antsthe cerebral ganglia are of extraordinary dimensions, and in allthe Hymenoptera these ganglia are many times larger than in theless intelligent orders, such as beetles.”

Formally stated, Darwin’s analogy is: “with re-spect to both central nervous system complexityand intelligence, ants are to insects, as humanbeings are to primates,” and its implication is thatthere is a functional relationship between the twoattributes. Darwin is not, of course, suggestingthat an ant’s cerebral ganglia (“not so large as thequarter of a pin’s head”) resemble a human brainin size, structure, function, or any other respectexcept their unusual size for an insect. This illus-trates the important difference between “analogy”and “resemblance,” concepts that are often con-fused.

Not all analogical propositions are expressed ex-actly in the “A is to B . . .” format, but they alwaysinclude the “likeness of relations.” For example, pa-leontologists often take dental dimensions from twoextant populations (A and B) known to representgood species, and use them as a standard by whichto determine whether two species are present in amixed bag of fossil teeth. If X and Y are heaps offossils sorted by whatever criteria the paleontologistchooses, he or she is testing a hypothesis in the formof an analogy something like: “when they were alive,the source population of X had the same sort ofrelation to the source population of Y, as A presentlyhas to B” (e.g., they were reproductively isolated,ecologically distinct, fully diagnosable, or conformedto whatever other species criterion is adopted). Sim-ilarly, a paleoecologist might call upon information

PAPIONIN ANALOGIES FOR HOMININ EVOLUTION 179Jolly]

about sympatric chimpanzees and gorillas to under-stand the interactions of large-bodied hominids co-existing at a late Miocene site, proposing an analogyin the form: “N-pithecus related ecologically to M-pithecus, as chimpanzees today relate ecologically togorillas.” Whereas “homoplasy” connotes any resem-blance acquired independently, an analogy, in thesense used here, is a particular kind of logical trian-gulation that uses significant patterning of indepen-dently acquired behavioral, morphological, or phys-iological traits to help understand and interpretevolutionary events in adaptational and functionalterms. Not all such traits may be homoplasies in theusual sense. In the example of analogous speciesdistinctness, for instance, diagnosticity and/or ge-netic isolation is a condition shared by the ana-logues, but it would be stretching the usage to call ita homoplasy. Conversely, many homoplasies (espe-cially, but not exclusively, in the primary structureof molecules) are certainly the product of purechance. Lacking any functional significance, suchrandom resemblances are unlikely to be incorpo-rated into a useful analogy.

Although some useful analogies can be drawnfrom close phylogenetic relatives, a close relation-ship of (A � B) to (X � Y) does not necessarily makefor a more illuminating comparison. There seems tobe little justification, for example, for the practice ofunquestioningly using Pan troglodytes and Gorillagorilla to represent “good species” when attemptingto deduce alpha taxonomy from a sample of fossilhominin skulls or teeth. Chimpanzees and gorillasare, of course, good species, but they are not sister-taxa (Ruvolo et al., 1991); the geometry of the homi-nine cladogram as currently understood means thatno pair of hominin species can be as distantly relatedas are chimpanzees to gorillas. The chimpanzee-go-rilla comparison is likely to provide a poor analogyfor sorting out hominin lineages, underestimatingthe number of newly emerged species present in afossil sample. For hominins, or any case where re-cent speciation is possible, a more appropriate com-parison is with other clusters of good, but minimallydifferentiated species: lemurs of the genus Eulemur,perhaps, or Cercopithecus monkeys, or dogs of thegenus Canis.

Another drawback with drawing analogies fromclose relatives is that it can be hard to make theimportant distinction between analogous and ho-mologous resemblances. Chimpanzee carnivory is acase in point. In the early 1960s, Goodall reportedthe first cases of meat-eating by chimpanzees (Good-all, 1986), at Gombe. Most commentators, notingthat Gombe is a “savanna” (actually, woodland-mo-saic) habitat, interpreted hunting as an adaptationthat appeared separately in early hominins and sa-vanna chimpanzees as each moved “out of the for-est.” This made it a homoplastic behavior, suggest-ing a significant analogy between savannachimpanzees and supposed savanna-dwelling earlyhominins, with the strong implication that savanna-

dwelling stimulated carnivory. It was quite quicklyshown, however, that Pan troglodytes practiced op-portunistic carnivory in other (probably all) habi-tats. This discovery suggested that carnivory, or atleast the ability to adopt it when advantageous to doso, was a shared ancestral (homologous) trait ofhominins and chimpanzees. It was a component ofthe behavioral repertoire of the last common ances-tor, but no more likely to be significant to hominindivergence than any other such trait (unless, ofcourse, there had been archaeological or anatomicalevidence for elaboration of carnivorous behavior inthe stem hominin). The two interpretations thushave quite different implications for the context inwhich simple, chimpanzee-style hunting appeared,and its possible role as a prime mover in homininevolution, but they are easily confounded, and quiteoften are. By contrast, when Schaller and Lowther(1969) compared and contrasted the hunting-scav-enging strategies of African carnivores with those ofearly hominins, there was no doubt that they wereusing an analogy.

Darwin’s ant-brain example is unusual in thegreat phylogenetic distance between the analogues.Useful analogies, especially for anatomical function,are often more readily found among relatives thatare less extremely distant. This fact may bias oursearch for analogies toward closer relatives, but it isimportant to note that this bias is purely pragmatic;the logic of analogy does not require that structureswhose function is compared should be “homologous”(in any sense). For example, flightlessness hasevolved many times on small oceanic islands, amongboth insects and birds; for the cause to be analogous,the wings of birds and beetles do not have to behomologous structures. One could, in fact, arguethat an analogy is all the more powerful if the pairsof analogues are not closely related. As with Dar-win’s ants, the more phylogenetically distant theanalogues, the more striking the coincidence, andthe more obvious the fact that a parallel adaptationhas occurred, and demands an explanation.

The papionins are ideally situated, phylogeneti-cally, to provide analogies for hominin evolution. Asfellow catarrhines, they share (by homology) manyattributes of hominin structure and function. Theyare basically arboreal (but often secondarily terres-trial) animals that live in permanent, bisexual asso-ciations, experience their world predominantly viatheir sense of vision, and use their hands to feed aswell as to locomote. They use 8 incisor teeth forinitial food preparation, 20 cheek-teeth for chewing,and 4 canines for agonistic interactions. All thesehomologies comprise a similar groundplan, liable tobe modified in similar ways. Yet the papionins arealso far enough removed from the hominins thatwhen analogous forces do modify one of the sharedcatarrhine attributes, producing a parallelism, it isrecognized as such, and is not mistaken for homol-ogy. A gelada baboon’s precision-gripping hand, forinstance, resembles a human hand in some of its


functionally related proportions, but it could neverbe mistaken for anything other than a cercopithe-coid’s.

HOMININ ORIGINS, THEROPITHECUS, AND“SEED-EATERS”

The “seed-eater” idea (Jolly, 1970a, 1972a,b) wasbased on an analogy that can be stated as: “[in somerespects] Australopithecus is to Pan, as Theropithe-cus is to Papio” (it might better have been stated,“[in some respects] Australopithecus is to the homi-nine morphotype, as Theropithecus is to the pap-ionin morphotype,” but New York in the late 1960swas still in the terminal Prehennigian). My excusefor summarizing this ancient story here is twofold:first, it illustrates the use of analogy, and, second, itaffords an opportunity to correct a few of the misin-terpretations that, barnacle-like, encrust its aginghulk. A more complete review of the current statusof “Seed-Eaters” will be presented elsewhere.

This analogy was used to construct a novel sce-nario for the origin of the hominin clade, by explain-ing, in functional and ecological terms, the fact thata particular suite of new traits had appeared in theearliest undoubted hominin species (at the time,Australopithecus africanus). It is important to notethat the analogy, like any such exercise, could not,and did not attempt to, address three other, equallyimportant but quite different questions: “Wheredoes Australopithecus fit within the primate phylo-gram?”; “What living species most closely resembleAustralopithecus?”; and “What anatomical changesseparated Australopithecus from its chimpanzee-like ancestor, and what are their functional impli-cations?” These problems require identification andfunctional analysis of Australopithecus apomorphiesin the context of the homologous traits shared withits closest known relatives, i.e., hominine apes andhuman beings. These are, of course, major, ongoingconcerns of paleoanthropology, that continue to pro-duce more refined descriptions and interpretationsin response to new fossil discoveries, work on livingapes, and the application of more sophisticated com-parative and functional anatomical methods. Theycan be tackled only from a base of knowledge of thebehavior, physiology, and anatomy of extant homi-nine apes and humans, and they fully justify thefocus of paleoanthropology upon these species. Butthe focus of “Seed-Eaters” was on a different prob-lem. At the time, the phyletic position of Australo-pithecus, and the major features of its anatomy, hadbeen satisfactorily interpreted (e.g., by Le GrosClark, 1955), to the extent allowed by the currentstate of knowledge. Australopithecus was indeed abasal hominin. In brain size and overall character-istics, it resembled a hominine ape, but it was abiped (of a sort), with small canine teeth in bothsexes, and incisor teeth that were small relative tocheek-teeth. What was not satisfactorily explainedwas why these apomorphies had appeared.

Contemporary interpretations of the hominin apo-morphies, essentially unchanged since Darwin(1871), emphasized not so much their origins astheir functional interdependence. Bipedal stance,for example, was favored because it freed the handsfor tool-use, tool-use led to reduction in the relativesize of the incisor and canine teeth, and to special-ization of the hand for tool use and manufacture,which in turn favored more efficient bipedal adap-tations, and so on. As Darwin (1871, p. 51) put it:

“Man alone has become a biped; and we can, I think, partly seehow he has come to assume his erect attitude. . . Man could nothave attained his present dominant position in the world withoutthe use of his hands, which are so admirably adapted to act inobedience to his Will. . . But the hands and arms could hardlyhave become perfect enough to have manufactured weapons, or tohave hurled stones and spears with a true aim, as long as theywere habitually used for locomotion. . . From these causes alone itwould have been an advantage to man to become a biped. . .”

Darwin’s statement borders on teleology, but it issaved by his cautious insertion of “partly,” and byhis passing reference, earlier on the page, to anantecedent cause, a “change in its manner of procur-ing subsistence, or to some change in the surround-ing conditions” (Darwin, 1871, p. 51)—what todaywe would describe, less elegantly, as an ecologicalprime mover, responsible for assembling the ele-ments of the positive feedback system.

Paleoanthropologists of the mid-1960s enjoyed aconsiderable advantage over Darwin in havingmuch more direct evidence about this prime mover:in naturalistic studies of wild great apes, in paleo-ecological information about the context of homininorigins, and especially in the derived anatomy ofhominins very close to the stem itself. Very littleattention, however, was directed to stem homininanatomy itself as evidence for the nature of hominincladogenesis. If the anatomy of basal hominins wasevaluated functionally (as opposed to phylogeneti-cally), it was from the viewpoint of the “path fromape to man,” most commentators apparently assum-ing that the nature of that path was self-evident,and that a vaguely imagined move from “forest” to“savanna” was sufficient to set the basal homininson their way along it. Basal hominin features wereevaluated according to their resemblance to humansor to apes, and human-like conditions were unques-tioningly interpreted as evidence for the beginningsof human-like behavior, e.g., a hand with relativelyshort phalanges and a long, powerful, and mobilepollex was attributed to tool-making, even in theabsence of artifactual evidence.

Besides their tendency to wander into teleology,such explanations suffer from a lack of generality.As a reviewer of this paper’s first draft put it, “Allanthropologists are bedeviled by the fact that thereis only one living primate (indeed mammal) that ishabitually bipedal and that manufactures stonetools regularly.” It is indeed the case that if we insistthat the only relevant species are those that share


both the adaptations we are trying to interpret, andthe complete suite of homologous, heritage traits inwhich they are embedded, the only relevant speciesfor explaining the evolution of hominin traits suchas bipedalism are hominins themselves. And sinceall of them (so far as we know) owe their bipedalismto the same, ancestral event, any explanation forthis event, no matter how ingenious, is a “Just SoStory,” applicable only to the case from which it isderived. This problem is not anthropology’s alone;every evolutionary transition in any species occursagainst the background of a unique heritage result-ing from an idiosyncratic evolutionary history.

Analogy offers a way out of the dilemma by allow-ing, in fact insisting, that the feature to be inter-preted (habitual truncal erectness supported frombelow by the hindlimb, for example) be decoupledfrom its context of heritage features. The basalhominin transition to bipedalism, for instance, canthen be linked to a wider universe of similar events,some of which may prove illuminating. We can asknot only “What caused hominins to become bipeds?”but rather “What common factors, if any, are asso-ciated with adoption of hands-free, upright postureor gait in other vertebrates?” Posed this way, thequestion yields a slew of potentially analogous tran-sitions in the ancestries of clades as varied as thero-pod dinosaurs, kangaroos, and gerenuks. Not all willprove illuminating, but some may stimulate freshinsights.

“Seed-Eaters” found an explanatory analogy inthe many parallels between trends seen in Thero-pithecus, especially large extinct forms assigned toT. oswaldi (Jolly, 1972b), and canonical descriptionsof Australopithecus africanus. Since A. africanuswas at the time the earliest and least derived knownhominin, its apomorphies were presumed to beadaptive reflections of the origin of the homininclade itself. Yet the relatively small incisors, espe-cially of the extinct Theropithecus oswaldi, paral-leled those of Australopithecus, and artifact-useseemed an unlikely explanation.1 Moreover, manyother distinctive, derived features of Australopithe-cus jaws and teeth were also paralleled in Thero-pithecus, especially the large, extinct species. Someof the Theropithecus-Australopithecus parallelismsinvolved features generally assumed to be functionalcorrelates of tool use or other aspects of culture,while others did not. A reevaluation of the functionalimplications of all the recognized apomorphies ofAustralopithecus concluded that few, if any, of thesederived traits demanded an explanation as adapta-

tions to culture and tool-using. This was true even ofthose, like a habitual bipedal stance and long, pow-erful pollex, that were not paralleled in Theropithe-cus. Instead, it was suggested that all the dental-gnathic homoplasies comprised a single functionalcomplex related to mastication, summarized as“back-tooth dominance,” and that this, together withlower-crowned canine teeth that permitted molarcusps to wear more evenly, postural changes associ-ated with australopithecine-style bipedalism, and ahand with a precise thumb-index pincer and a pal-mar pocket, could all be related functionally to“small-hard-object feeding.” Moreover, “small-hard-object feeding” also described the most salient de-rived feature of the extant Theropithecus gelada’secology, though in its case the food objects requiringthorough mastication were grass blades, rhizomes,and corms, rather than seeds, and the gelada is ahabitual, upright-trunked squatter, rather than abiped. The stem hominin was presumed to be aboutas culture-bearing, artifact-using, and carnivorousas extant chimpanzees, but the staple of its mainlyvegetarian diet was tough savanna-woodland seedsrather than forest fruits. In the original formulation,grass seeds gathered in edaphic grasslands wereimagined to be the staple. Later (Jolly and Plog,1987), in response to work by Walker (1981) andPeters (1982), as well as my own observations inEthiopia, I identified the tough but nutritious fruits,seeds, and pods of thornbush shrubs such as acaciasand Grewia as a more likely dietary focus than grassseed.

“Seed-Eaters” also pointed out that Theropithe-cus and Australopithecus each have apomorphiesthat are not paralleled in the other, and these tooare a significant component of the analogy, be-cause they differentiate the respective econichesof the analogs. For example, Theropithecus hashigh-crowned, complex, molar teeth functionallysimilar to those of other graminivores (grass-eat-ers). Hominins have low-crowned, thick-enameledmolars suitable for milling and crushing. Anothercontrast in apomorphic traits, not apparent in1970, also underscores a difference between gra-minivory and granivory (seed-eating). Theropithe-cus gelada appears to have a smaller brain thansimilarly sized Papio baboons (Martin, 1993),while Australopithecus, by some interpretations,may show modest apomorphic brain enlargement.If confirmed, the difference might reflect a nutri-tional contrast between leguminous seeds eatenby basal hominins (high in calories, protein, lipids,and essential fatty acids) and grassy herbageeaten by geladas (comparatively low in nutrientsper unit bulk). It might also reflect a difference inthe distribution of the foods: relatively dispersedand patchy in space and time for savanna-wood-land seeds, more continuous for grasses and herbsin the gelada’s montane grassland.

I thought that a diet-based explanation for homi-nin origins was appealingly parsimonious. It ex-

1“Seed-Eaters”, following the conventional wisdom, assumed thatAustralopithecus dental proportions differed from those of apes, espe-cially chimpanzees, mainly because incisors had become smaller. Infact, as Wolpoff (1973) soon showed, cheek-tooth enlargement as muchas incisal reduction was the cause of the changed proportions, espe-cially in the newly described, and even more basal, A. afarensis. This,of course, fitted the dietary scenario even better, and the artifact-driven model much more poorly.


plained the evolution of the small-hard-object feed-ing adaptations in hominins, the independentevolution of functionally equivalent features inTheropithecus, and the parallelism between them.Compared to previous explanations, it covered agreater variety of derived basal hominid features,and linked features not previously considered part ofthe same functional complex (e.g., large molars andadept hands). It was also economical in proposing thatthe initial ape-hominin divergence involved a rela-tively small and simple change in dietary emphasis: aminimal ecological change of a kind thoroughly famil-iar from mammalian paleontology, rather than a one-of-a-kind leap into a multifaceted protoculture.

Not all colleagues were convinced. In some cases,the problem seems to have been that the critic mis-interpreted “analogy” as “identity,” mentally con-nected “model” and “baboon,” and jumped to theconclusion that a living baboon was being used as anavatar of an early hominin. (This is equivalent tointerpreting Darwin’s analogy to mean that humanshave brains like those of ants.) Others seemed tofeel, mistakenly, that the importance and relevanceof chimpanzees were being slighted. In yet othercases, the objection seems to have been based on afeeling that an event as momentous as the origin of thehuman lineage must surely have been caused by some-thing more complex and unique than a not-very-pro-found change of dietary emphasis—a case, I think, ofthe “humans are special” bias that tends to afflictbiological anthropology, and also of confusing a simpleprime mover with its multifarious consequences.

Pending the more extensive review, it bears stat-ing, for the record, that “Seed-Eaters” advancednone of the following propositions. Each has been setup as a straw man and gleefully demolished by onecolleague or another:

1) that Australopithecus (or hominins in general)were more closely related, phylogenetically, tobaboons than to chimpanzees.

2) that Australopithecus ate grass like a gelada.3) that Theropithecus is a “seed-eating baboon.”4) that basal hominins inhabited “vast, dry sa-

vanna grasslands” (Spenser, 1997, p. 201).5) that the “Seed-Eater” idea stood or fell with the

hominin status of Ramapithecus.6) that Australopithecus ate only seeds.7) that all the derived features of Australopithecus

were paralleled in Theropithecus, or vice versa.8) that Theropithecus resembled early hominins

more closely than it did Papio baboons or othermonkeys.

9) that male T. gelada have small canine teeth.10) that Australopithecus (or any other early homi-

nin) resembled a Theropithecus baboon moreclosely overall than it did African apes, specifi-cally, the chimpanzee.

By explicitly decoupling the origins and early evo-lution of hominins from the concept of an expansive,

culture-driven econiche, and instead invoking di-etary and habitat shifts as causal factors, “Seed-Eaters” represented basal hominins not as prehu-mans whose importance was measured purely byhow far they had traveled down the road from “ape”to “man,” but as nonhuman primates, subject to thesame forces of diversification, speciation, and adap-tive radiation as other contemporary mammaliantaxa. Since all hominins did not have to lie on thesame adaptive path, the large-jawed, large-toothed“robust” forms, seen as direct derivatives of the firstphase of hominin evolution, were able to live along-side the “human” branch, which by this time hadbuilt a new, more carnivorous, artefact-basedeconiche on the seed-eater base.

In the last few years, new discoveries have begunto document just how extensive a radiation actuallyoccurred among the basal hominins. Two, possiblythree, genera (Ardipithecus, Kenyanthropus, andthe still mysterious Ororin), and three species ofAustralopithecus (A. bahrelghazali, A. anamensis,and A. garhi), have joined the roster of hominins ofthe 6–1.5-mya period, and the moribund genusPraeanthropus has been revived. (Leakey et al.,1995, 2001; Asfaw et al., 1999; White et al., 1994;Haile-Selassie, 2001; Brunet et al., 1995). Some au-thorities remove one or more lineages of “earlyHomo” from the genus and assign them to distinctramifications (Wood and Collard, 1999; Wolpoff,1999). In part, this multiplication of taxa is a prod-uct of shifts in evolutionary and taxonomic philoso-phy and practice. There is less antipathy to propos-ing new names in cases where the evidence isinconclusive, and wider recognition that the shape oforganic evolution in general is “bushier” than previ-ously appreciated. The phylogenetic species concept,and acceptance of clade-based taxonomic usage thatabhors paraphyletic and “wastebasket” taxa, havetended to multiply recognized species. As always,some newly named hominin forms may ultimatelydisappear into synonymy. Nevertheless, even allow-ing for shifting taxonomic fashion and uncertaintiesin alpha taxonomy, it is clear that much of the newlydescribed diversity represents biological reality.Moreover, most of this “new” diversity has been rec-ognized in hominins from a single broad eco-geo-graphic zone: the northern savanna-woodland belt,and especially its extension in eastern equatorialAfrica. Much of the rich basal hominin materialfrom South Africa, which derives from very differentecological and geographical settings (Bromage andSchrenk, 1999), has yet to be fully described andinterpreted within the current, more schizophilictaxonomic climate. It can safely be predicted thateven more basal hominin genera and species willultimately be recognized.

HOMOPLASY, CRYPTIC SYMPLESIOMORPHY,AND THE DIVERSITY OF EARLY HOMININS

The recognition of multiple lineages early in homi-nin history raises the methodological question of


how their cladistic and phylogenetic relationshipscan be retrieved from the distribution of morpholog-ical character-states. The extant papionins, forwhich we have a reliable molecular phylogeny, pro-vide an informative, if not altogether optimistic,analogy.

Opinions differ as to whether molecular data arebest considered independently, or combined withmorphology, in phylogenetic reconstruction (Kluge,1989). As molecular data become more comprehen-sive, and methods of analysis more sophisticated,the case for building molecular trees independent ofother evidence becomes compelling, because theyprovide a powerful heuristic tool: a map on which toplot the evolution of morphological character-states(Collard and Wood, 2000). This method assumesthat the molecular tree accurately represents phy-logeny, an assumption adopted in the following dis-cussion, without implying that present moleculardatasets are adequate, or that current analyticalmethods are infallible.

The accepted molecular phylogeny of the papioninradiation (e.g., Harris and Disotell, 1998; Disotell,1996; summarized in Disotell, 2000) reproducesmany of the features of morphology-based trees (e.g.,Strasser and Delson, 1987), but differs in linkingPapio with Lophocebus, and Mandrillus with Cerco-cebus. This particular discrepancy is especially sig-nificant because the molecular phylogeny disman-tles two groups traditionally considered singlegenera: large, long-faced, terrestrial or semiterres-trial “baboons” (Papio, s.l., becomes Papio � Man-drillus), and smaller, short-faced, arboreal “manga-beys” (Cercocebus, s.l., becomes Cercocebus �Lophocebus). We should note in passing that al-though it is the papionin case, involving higher pri-mates, that has caught the attention of anthropolo-gists, its message—that homoplasies are morefrequent, and harder to detect morphologically, thanwe ever believed—has become almost a cliche ingeneral evolutionary biology. Every issue of the rel-evant journals seems to include a paper or two(many by morphologists, physiologists, or ecologists)that not only revises a cladogram on the basis ofmolecular data, but also uses the resulting homopla-sies to construct informative analogies based on thehomoplastic acquistion of functionally related traits.

As a test case for morphological cladistics, themangabey-mandrill-baboon case nicely exemplifiestwo major kinds of misleading, nonapomorphic re-semblance. Whether the baboon or the mangabeycharacter-states are considered ancestral, the re-vised phylogeny implies major homoplasies in bothskull form and postcranial features. Some of thesewere evidently produced by conventional ho-moplasy: parallel adaptation to similar function.These traits include the degree of posterior angula-tion of the ulnar olecranon process, which is relatedto the proportion of climbing and terrestrial locomo-tion in the animal’s locomotor profile (Jolly, 1967),and unites semiterrestrial mandrills and baboons on

the one hand, and the two groups of mangabeys onthe other.

Other homoplasies, such as those affecting faciallength (and correlated characters) in the Africanpapionins, can be attributed to a less obvious butprobably even more powerful phenomenon: ances-tral patterns of relative growth that independentlyproduce similar phenotypes in animals of compara-ble overall size. The relationship between facial andcranial length in Papionini is an extreme case ofpositive allometric growth, and was one of the ear-liest to be expressed mathematically (Huxley, 1932).In overall skull proportions, the adults of small pa-pionin species resemble the juveniles of larger ones,while the adults of large-skulled forms (mandrills,geladas, and Papio baboons) all have relatively longfaces. If the developmental relationship itself (rath-er than its size-determined phenotypic expression inadult skulls) is considered the character-state, it is aplesiomorphy of papionins, and the homoplasy thatlinks baboons and mandrills is not facial length butabsolute skull (and body) size. Whether long facesare labeled a phenotypic homoplasy, or a develop-mental plesiomorphy, the trait is equally misleadingand phylogenetically uninformative. It is interestingto note that a truly derived state of the morphoge-netic character can be recognized among papionins,but not in the long-faced baboons. It occurs in someSulawesi macaques (Macaca nigra and close rela-tives). Their relative facial length is baboon-like, butbecause it is combined with a much smaller, ma-caque-sized skull, it implies that a shift in relativegrowth trajectory occurred in the ancestry of thisparticular subclade of macaques (Jolly, 1965; Al-brecht, 1977).

Another lesson to be drawn from the papionins ishow completely the combined effects of true ho-moplasy and unrecognized morphogenetic symplesi-omorphy can obscure real phylogenetic relation-ships, and powerfully support false ones. Wesometimes tend to assume, I think, that detailedresemblance in complex anatomical features is suf-ficient evidence for true synapomorphy, becausesuch complexity is unlikely to be duplicated in evo-lution. The papionin case illustrates this argument’sflaw: as morphological genetic work is documentingmore and more clearly, structural and genetic com-plexity are not closely correlated. Though changes inshape may be complex in the sense that many struc-tures (and an almost infinite number of possiblemetrics) are affected, they nevertheless can be du-plicated in parallel evolution, presumably becausetheir genetic basis is comparatively simple. In fact,it is not hard to envision an array of related forms inwhich great diversity in phenotypic characters isdetermined by the differential expression of a fewsimple developmental symplesiomorphies. Here, thereal apomorphies would be the factors underlyingthe differential expression of the developmental pat-terns; they themselves would be excellent candi-


dates for cladistic analysis, but their detection isstill in its infancy.

As Collard and Wood (2000) recently demon-strated, the formal, computer-aided application ofcladistic logic to large suites of morphological char-acters is no more likely to give the correct (i.e.,molecule-concordant) answer than is the kind of in-tuitive, seat-of-the-pants analysis favored by manypaleontologists and morphologists. Like some tradi-tional, premolecular classifications, the PAUP anal-yses of Collard and Wood (2000) suggest that man-drills and Papio baboons are sister taxa, and unitethe great apes to the exclusion of Homo. Both mor-phologists and tree-building computer programs aremisled because the basic logic of parsimony cannotdetect cases where an erroneous tree is overwhelm-ingly supported by suites of unrecognized, correlatedhomoplasies, and by developmental symplesiomor-phies concealed by allometry. Parsimony-based sys-tems are designed to recognize clades by concordant,shared, derived character-states. Suites of pheno-typic similarities resulting from functional ho-moplasy and cryptic symplesiomorphy are not syna-pomorphic, but they are both derived and shared,and they are often distributed concordantly, as “bun-dles” or complexes.

Having shown convincingly that PAUP producesstatistically robust but spectacularly inaccuratetrees when applied to an extensive set of standardcraniodental measures, Collard and Wood (2000)suggested that craniodental data may be intrinsi-cally inadequate: a gloomy prospect for the paleon-tologists who must rely on such data for phyloge-netic reconstruction. Collard and Wood (2000) do notsort their characters by utility (as measured by con-cordance with the molecular phylogeny), but itseems likely those (such as facial length) that areaffected by cryptic symplesiomorphy were highly in-fluential. (It should be noted that the analyticmethod of Collard and Wood (2000) for eliminatingthe influence of overall size as a factor in the anal-ysis also effectively concealed size-related effects.)

The situation may, however, be less hopeless thanCollard and Wood (2000) imply. We know, after all,that the morphological data do contain the neces-sary information, because an informed eye can dis-cern the correct story. Primed by strong molecularhints of the Cercocebus-Lophocebus dichotomy (Cro-nin and Sarich, 1976; Hewitt-Emmett et al., 1976),Groves (1978) was able to find cranial, dental, andsoft-part characters that supported it. Similarly,dental and postcranial character states linking Cer-cocebus with Mandrillus, and Papio with Lophoce-bus, could be discerned (Fleagle and McGraw, 1999),once morphologists knew to look for them. Moreover,once identified, the phylogeny-concordant traitscould then be plausibly interpreted in terms of ecol-ogy and behavior.

If, as this suggests, phylogenetically meaningfulmorphological traits are concealed in a mass of un-informative or misleading information, simply add-

ing characters uncritically to a mechanical cladisticanalysis is not the solution. The new characters areunlikely to be functionally and developmentally in-dependent of the existing list, resulting in stilllarger bundles of covariant homoplasies, and evenstronger statistical support for spurious phylog-enies. The problem is not too little information, buttoo much that is misleading, so pruning the datasetmay be as productive as augmenting it.

Is it even possible to identify the informativetraits and cut away the misleading ones, without amolecular crib-sheet? In the mandrill-mangabeycase, “good” characters were comparatively few, sub-tle, and not concentrated in any one tissue or struc-ture. Apart from the dental features noted byFleagle and McGraw (1999), many of them wereirrelevant with respect to function. Obviously, amore accurate tree would result if homoplasies wereexcluded at the outset, but there is no foolproof wayof doing this. The correct cladogram can be built(even by PAUP) only if homoplasies are identifiedand excluded, but homoplasies are defined by theirdiscordance with the correct cladogram.

While we cannot reliably identify homoplasies bythe definitional criterion, we can spot and excludedata that are most likely to be affected by homoplasy(Haszpruner, 1998). Such data-pruning has becomeunfashionable with the ascendancy of parsimony-based computerized methods, but is implicit in muchintuitive phylogeny building. One strategy devaluescharacters that are likely to covary because they arecomponents of a single functional system, identify-ing them intuitively (Skelton and McHenry, 1992),or by using statistical measures of association(Strait, 2001), and collapsing them into a single vari-able to be entered into the analysis. The whole“small-hard-object-feeding” complex might, for ex-ample, be collapsed to a single entity. This will notremove all homoplasies, but it will ensure that eachfunctionally or morphogenetically determined deter-mined homoplastic “bundle” that is identified con-tributes no more than one data point to the analysis.(It will, of course, also limit the contribution of truesynapomorphies, such as those identified by Fleagleand McGraw (1999), that comprise, or might com-prise, functional complexes.)

A complementary, somewhat more radical proce-dure is to produce a matrix of intertaxon concor-dances, using all available data without regard todescriptive level, from the physiological to the grossmorphological, splitting and rewording comparisonsto make the analysis as comprehensive as possible.This dataset can then be rigorously censored, ex-cluding any resemblances or differences that can beplausibly explained in terms of documented resem-blances or differences in behavior, natural history,or morphogenetic processes (Jolly, 1970b). For ex-ample, geladas are close to humans in their “oppos-ability index,” an expression of the relative lengthsof thumb and index finger, presumably related func-tionally to an efficient precision grip (Napier, 1980).


As a trait, the simple resemblance in proportionsobviously tells the wrong phylogenetic story (geladasand humans are not sister-taxa), but would be ex-cluded because it is easily explained by functionalparallelism (parallel adaptation to a particular kindof manipulation). It can, however, be broken downby considering the relative lengths of the digits asseparate characters. The gelada achieves its effi-cient opposability by shortening its index finger,while humans have a relatively long thumb. Thisdifference does carry phylogenetic weight (in thiscase, against relationship), because the function ofthe two mechanisms is closer than their structuralresemblance. It therefore points to two separatepathways for achieving the same functional end,and, hence, to separate ancestries: one knuckle-walking, the other digitigrade quadrupedal.

Phylogenetically weighty differences and resem-blances are also exemplified by the very close anddetailed similarity in shape of female sexual swell-ing that unites Cercocebus with Mandrillus, andLophocebus with Papio (Hill, 1970). Resemblancesbased on the character “sexual swelling occurs/isabsent” are of low phyletic weight, because the fea-ture is related functionally to social structure andmating strategy. But, absent evidence to the con-trary, we can presume that any of the swellingshapes seen among African papionin species wouldperform as signals. To the extent that the resem-blance of Cercocebus to Mandrillus, and Papio toLophocebus, is closer than function demands, theshape and position of sexual swellings may be phy-logenetically significant. By similar logic, the gela-da’s pectoral sexual skin, though a radical departurefrom the papionin norm, loses its phyletic weightwhen it is interpreted as a functional correlate ofsquat-feeding.

Yet another example is provided by cercopithecoidbilophodonty (as opposed to “hominoid” molar struc-ture) (Jolly, 1970b). Suppose that all we knew of theanatomy of (say) Hylobates lar, Cercopithecus asca-nius, and Nasalis larvatus was their dentition. As-suming that character-states could be unambigu-ously defined, a matrix of resemblances wouldproduce some (Hylobates � Cercopithecus) charac-ter-states, others uniting Cercopithecus and Nasalis,and possibly some (Nasalis � Hylobates). Ratherthan proceeding directly to cladistic analysis, the“pruning” method would then evaluate each resem-blance or difference against resemblances or dif-fernces in natural history. We would ask, for in-stance, whether documented dietary differencesbetween Hylobates and Cercopithecus are sufficientto explain the four-cusp/five-cusp contrast in lowermolars. Having decided that they are not, we deducethat this character-state difference probably hasphylogenetic weight: it reflects “ancient” evolution-ary events. The same conclusion would be drawn bycomparing cusp number and diet in Cercopithecusand Nasalis: in this character, their molars are moresimilar than their contemporary diets demand;

there is a component in their degree of resemblancethat is “nonadaptive.” In fact, across all extant ca-tarrhines, no dietary variable explains the dichoto-mous distribution of the two molar types, so itsrelevance must be “historical”—it can be presumedto date to the initial divergence of hominoid andcercopithecoid stocks. There is nothing originalabout this way of looking at morphology: it is whatDarwin implied when he wrote (1871, p. 153):

“it appears more correct to pay great attention to the many smallresemblances, in giving a truly natural [i.e., phylogeny-based, incurrent usage] classification.”

Unfortunately, Darwin’s insight in this respecthas tended to be overshadowed, for morphologists ingeneral, and physical anthropologists in particular,by an equally important, but quite distinct, evolu-tionary generalization: that major adaptive radia-tions have often been founded on major, functionallyimportant changes, the transition from a “monkey-like” to an “apelike” forelimb morphology being afrequently cited example.

The same process (of attempting to spot the “non-adaptive,” and therefore phylogenetically weighty,aspects of resemblance and difference) should alsobe applied to the analysis of fossil forms, but herethe logic is necessarily even less direct. We cannotevaluate structural resemblance directly againstfunctional similarity, so we have once again to callupon analogy to ask whether, from our knowledge ofextant cases, such structural resemblances and dif-ferences seem to arise frequently, in similar ecolog-ical circumstances, and in comparable combina-tions.

All this is obviously tortuous, subjective, and plau-sibility-based; but it is surely a process that anyexperienced evolutionary morphologist routinely ap-plies to judgments about what an organism “reallyis,” usually without making it explicit. No contem-porary paleontologist, given only a posterior man-dibular fragment and an auditory bulla, would in-terpret Archaeolemur as a monkey with a lemuroidauditory region, rather than a lemur with cerco-pithecoid-like molars. Some such mental process asI have described must underlie this interpretation,unless we resort to pre-Darwinian notions of “arche-type,” or the idea that some structures or featuresare in some way more “fundamental” than others,and therefore more reliable indicators of phylogeny.

The logic of pruning or censoring the data in thisway is similar to that of molecular systematistswhen they collect their data base by base, and thennarrow their analysis to third bases and introns, butthere is an important methodological difference. Themolecular biologist excludes or devalues first andsecond nucleotides, or whole coding regions of genes,generically, and can justify this procedure both em-pirically (it gives results that are internally consis-tent) and theoretically (introns and third bases havefewer epigenetic effects, and their variation is there-


fore less likely to be driven by natural selection).Precensoring morphological data, on the other hand,cannot, in the present state of knowledge, be simi-larly generic. A moment’s reflection tells us that itwill not work if we try to generalize about the phy-logenetic information content of structures (“teethare better indicators of phylogeny than postcrania”),of characters (“molar cusp number is a good phylo-genetic indicator”), or even of character-states (“bilo-phodonty reliably identifies primate clades”). Theevaluation has to be applied to particular concor-dances and discordances in character-states (“theshared bilophodonty of Nasalis and Cercopithecus isa good indicator that they are more closely related toeach other than either is to Hylobates”).

This “pruning” of the data matrix does not replacecladistic analysis. Phylogenetically weighty resem-blances and differences can be either apomorphic(bilophodont molars shared by cercopithecoids) orplesiomorphic (“hominoid” molars among eucata-rrhines) and require the usual Hennigian analysis,by whatever logical scheme is preferred: eitherbrute-force, parsimony-based methods (where thedata are numerous, as in molecular phylogenies), ormethods that use inferred character-state polarity(preferred by many morphologists).

The process of identifying and eliminating poten-tially misleading characters, and identifying usefulones, is the mirror-image of the analogy-buildingprocess discussed above. To discern functional com-plexes, analogues are analyzed by subtracting theirrespective heritages, leaving the informative paral-lelisms. To obtain phylogenetic information, thecommon features that might have functional signif-icance are subtracted, hopefully leaving a residue ofinformative heritage characters. Because the basisof censoring is the functional and developmentalinterpretation of characters, judgments may bechanged radically by new information on behavior,function, and morphogenesis. For example, beforenaturalistic studies of gelada baboons (Crook andAldrich-Blake, 1968) showed them to be specialist,bottom-shuffling grazers, their many anatomical pe-culiarities seemed to betoken a long, independentevolutionary history and great phylogenetic dis-tance from other baboons (Leakey and Whitworth,1958; Jolly, 1966), but this interpretation was im-mediately invalidated when it was shown that most,and perhaps all, gelada autapomorphies formed afunctional complex that could be related to uniquefeatures of present-day gelada ecology (Jolly,1970b).

As a formal procedure, however, precensoringfaces some formidable difficulties. It depends en-tirely on reliable information about functional anat-omy and morphogenetic patterns in the group con-cerned, which are then used to make ad hoc,qualitative estimates of phylogenetic informationcontent. Functional and morphogenetic interpreta-tions are generally supported by arguments fromplausibility rather than statistics, and will probably

be debated (Strait et al., 1997). In the present stateof knowledge, these judgments can only be madeindividually, case by case. We can only hope that, asmorphological datasets are tested against molecularphylogenies in more vertebrate taxa, and functionalcomplexes are identified by the recognition of signif-icant homoplasies, and as the genetics of morpho-genesis become better understood, regularities willemerge that will make the weighting process lesshaphazard.

As the gelada example shows, censoring will fre-quently remove many hard-won morphological datafrom the analysis as phylogenetically uninforma-tive. Worse yet, some real synapomorphies will bediscarded, because they are function-based. For ex-ample, the dental features (e.g., large, broad, P3)linking Cercocebus and Mandrillus are actually con-cordant with molecular phylogeny, and are probablysynapomorphies, but they would be explicable be-cause they are plausibly explicable as dietary spe-cializations (Fleagle and McGraw, 1999), and thuscould have arisen independently. Losing the “real”phylogeny by excluding some true synapomorphiesis, I think, preferable to supporting an erroneous oneby including homoplasies. Uncertainty, which stim-ulates further investigation, is always preferable tounwarranted confidence.

This uncertainty will be greatest for fossil taxa,especially those known only from fragmentary re-mains. It must be recognized that many such taxawill inevitably be left in a cladistic limbo, because alltheir scorable characters have been rejected. Somemay consider this a weakness; others will prefer it tocladograms that are statistically well-supported byproblematic data.

In fact, it seems realistic to suppose that there willbe frequent cases in nature where internodal dis-tances are so short, and adaptive radiation so fastand frequent, that homoplasy in adaptive features isvirtually complete, leaving little or no morphologicalinformation to indicate the true phylogeny. Whilethe “real” clades in such groups might be revealed byanalysis of fast-evolving, selectively neutral molec-ular markers, recognizably nonhomoplastic, mor-phological indications of the correct relationshipsmay be so few, obscure, and trivial, that they willstand little chance of being found, especially in fossilmaterial.2

The basal hominin radiation as a whole may bejust such a “difficult” group. Internode intervalsmust be relatively short, because successive clado-

2Note that this scenario does not imply that speciation in thesecases is caused by the accumulation of the trivial, nonadaptive mor-phological features whose distribution would be good phylogeneticevidence. Reproductive isolation presumably would have resultedfrom any combination of the usually postulated factors: differentialecological adaptation, chance or adaptive fixation of behavioral, phys-iological, or genetic barriers, and so on. Of these, the first would bediscounted because of possible homoplasy, and the others would beinvisible in fossile evidence.


genetic events are constrained to a total of onlyabout 3–4 mya between the well-established Pan-Hominini divergence at 6–8 mya, which must ante-date any intra-Hominini divergence, as well as thepaleontologically documented time (say, 3.5 ma) atwhich multiple hominin lineages are documented orinferred. Though a “young” radiation, basal homi-nins were geographically widespread, with rangeseventually extending across much of tropical andsubtropical Africa. Since each new, clade-foundingspecies presumably originated in a limited home-land, rapid population expansion across suitablehabitats and resultant genetic structuring of theexpanding population (Templeton, 1998) must havebeen influential. Furthermore, judging by the distri-bution of contemporary African mammals (Kingdon,1997), suitable hominin habitats were discontinu-ously distributed, with the edaphic flood-plains andother broad, habitable, nonrainforest areas con-nected by narrow corridors through less hospitableforests and highlands. The East African transequa-torial savanna-woodland zone was surely a criticalpathway between northern and southern homininhabitats (Bromage and Schrenk, 1999), but as King-don (personal communication) has pointed out, ma-jor rivers flowing from forested highlands eastwardsto the Indian Ocean presented hurdles for expand-ing hominin populations. The first population toreach a pristine, extensive, and ecologically variedstretch of nonforest habitat (such as the southernAfrican temperate and subtropical grasslands)would undergo population expansion, and wouldalso be faced with varied ecological opportunities.Adaptive radiation might well occur within eacharea, with different morphs in the array duplicatingthose that had appeared in other regions. This is aphenomenon that is well-documented among extantorganisms, with most cases being unsuspected untildocumented by molecular phylogeny (Schluter andNagel, 1995; Schluter, 2001; Taylor and McPhail,1999). The resultant patterning of morphologicaltraits is the least propitious for reconstructing phy-logeny, because internodes representing commonancestry, during which informative, synapomorphictrivia can accumulate, are relatively short, andadaptive radiations, which spawn homoplasies, arefrequent.

Among basal hominins, the problem is mostclearly exemplified by the taxa usually assigned toParanthropus (the South African P. robustus, in-cluding P. crassidens; and the East African P. boiseiand P. aethiopicus). They share relatively large mo-lars, relatively small incisors, very large masticatorymuscles, a deep mandibular ramus, and a stronglybuttressed facial structure. The generic grouping,implying a shared and exclusive ancestry, is re-garded as one of the most firmly established withinHominini (Grine, 1988), and was supported consis-tently in an exhaustive, PAUP-driven cladistic anal-ysis of hominin phylogeny (Strait et al., 1997). Infact, the Paranthropus clade was supported even by

an analytic run (Strait et al., 1997) explicitly omit-ting characters considered (Skelton and McHenry,1992) to be related to the “heavy-chewing” complex.Yet in spite of this impressive support, the evidencefor Paranthropus monophyly has been cogently chal-lenged (McCollum, 1999), on the grounds that all thecranial resemblances, some of which are only indi-rectly related to the large-back-tooth, heavy-chew-ing complex, boil down to a single functional pack-age with a relatively simple morphogenetic base. Ifthis interpretation is correct, the shared features inParanthropus species might be true synapomor-phies, but because they can also be explained by acombination of functional parallelism and allomor-phosis, they carry little or no phylogenetic weight. Itremains to be seen how Paranthropus monophylywill fare against the alternative hypothesis (parallelderivation of “robust” morphs from separate “nonro-bust” ancestries) if all morphogenetically relatedcharacters as well as functional complexes are rig-orously excluded from the analysis. The work ofMcCollum (1999) hints that a few nonfunctionallycorrelated dental traits might remain to tell the truestory.

The papionin analogy, then, suggests the generalprinciple that we should not rely too heavily on eventhe most formally rigorous and apparently well-sup-ported early hominin cladograms (Strait et al.,1997), since, unless the data are prescreened andcensored, much of the apparent cladistic structuremay be an artifact of functional and developmentalhomoplasy. But there is yet another source of phy-logenetic noise to be considered.

RHEBOONS, GEBOONS, AND EARLYHOMININ RETICULATIONS

A major assumption of most studies of the homi-nin fossil evidence is that relationships among diag-nosable species- or genus-level taxa are accuratelyrepresented by dendrograms rather than reticula-tions. Cladogenesis is a clean break, in which com-plete genetic isolation between sister lineages is es-tablished either before paleontologically recog-nizable ecological and phenotypic divergence, or sosoon afterwards that the interval between is incon-sequential on a geological timescale. This assump-tion is almost essential for conventional cladisticanalysis, which cannot easily accommodate reticu-lation.

Various lines of evidence, however, suggest thatthis assumption may not be altogether valid for thepapionins, and perhaps by implication, in homininsalso. Papionins show a remarkable ability to hybrid-ize (“crossability”). Hybridization, in this context, isdefined as the production of offspring by interbreed-ing of members of genetically differentiated popula-tions (Barton and Hewitt, 1985). The definition de-liberately does not specify the level of taxonomicseparation of the interbreeding populations. Amongpapionins, as in many extant taxa, hybridizationoccurs across a wide spectrum of taxonomic levels.


At one extreme, we have rare, unusually artificial,inter-subtribe crossing, and at the other extreme,the formation of natural hybrid zones between dif-ferentiated, parapatric populations of the same, orvery closely related, species. Although we might betempted to divide this spectrum at the conventional“species level” (assuming, of course, that we couldagree upon what that is), this would be an artificialand unhelpful division of a continuum. The presentdiscussion concerns crossing between taxa at theupper end of the continuum, between taxa that arenormally allocated to different subtribes, genera,subgenera, or species-groups.

The most phylogenetically distant pair of pap-ionins to have produced a well-attested, viable off-spring are rhesus monkey (Macaca mulatta) andbaboon (Papio) (phylogenetic separation � 10 ma).Twenty-six offspring (“rheboons”) were produced atthe Southwest Foundation, of which one male, whenadult, was subjected to detailed karyotypic and re-productive evaluation (Moore et al., 1999). This an-imal was considered behaviorally abnormal (J. Rog-ers, personal communication), and histologicalexamination showed him to be sterile, although, sig-nificantly, no mismatching of his parental haploidchromosome sets could be detected. We can presumethat if rhesus monkeys (or other macaques) andPapio baboons were sympatric in the wild, most ifnot all hybrids produced are most unlikely to breed,and probably would not survive.

The best-documented recent case of hybridizationbetween papionin genera is that between Papio andTheropithecus (Jolly et al., 1997) (phylogenetic sep-aration, � 5 ma). When caged together, they readilyhybridize (Markarjan et al., 1974; Jolly et al., 1997;J. Rogers and T. Newman, personal communica-tion), producing F1 offspring, i.e., “geboons.” In cap-tivity, female F1 geboons are certainly able to func-tion socially, and to attract Papio males, are bothviable and fertile, and produce viable backcross off-spring with Papio males. In fact, an F1 female wasmated by a hamadryas male who also had access tohamadryas (and anubis and hybrid) females in thecage. The resulting female backcross (3/4 Papiograndparents) (Jolly et al., 1997), when last ob-served, was still alive and healthy at age 5, but therewas no indication that she had bred. The fertility ofF1 males is also unproven. In parts of Ethiopia,geladas coexist with anubis or hamadryas baboons(Papio anubis and Papio hamadryas), and occa-sional natural hybridization is suspected (Dunbarand Dunbar, 1974). The case reported by Dunbarand Dunbar (1974) occurred in an area that sup-ported balanced populations of both species, andhuman disturbance was no more severe than inmost of the gelada’s range. Here, “bachelor” males ofT. gelada, a harem-holding species, were seen mat-ing with young female Papio anubis, a species inwhich females are philopatric and polyandrous, andfully adult males are only weakly attracted tosubadult females. The probable hybrids belonged,

however, to the gelada herd, which suggests thatthey may have been backcrosses to that species(Jolly et al., 1997).

The genus Macaca is commonly divided into spe-cies-groups, which are sometimes given subgenericrank (Fooden, 1976). These groups appear to be phy-logenetically valid, and are ecologically distinct tothe extent that species of different groups often haveoverlapping ranges, while species of the same groupdo not. The time of divergence between macaquespecies-groups is not well-documented paleontologi-cally, but from molecular evidence (Tosi et al., 2000),is probably in the 3–4-ma range. Bernstein (1966,1968) documented natural hybrids between mem-bers of sympatric species belonging to two different,well-defined species groups, Macaca fascicularis(mulatta-group) and M. nemestrina (silenus-group).It was suggested that interbreeding occurred be-cause female M. nemestrina joined a M. fasicularistroop after local extermination of male M. nemest-rina (Bernstein, 1966). The hybrids apparently func-tioned well as normal, long-term members of a M.fascicularis social group (Bernstein, 1968).

Assuming that in these cases some interspecies-group or intergeneric hybridization, with survivaland backcrossing by at least some of the hybrids,occurs in the wild, what evolutionary impact can ithave? Can significant effects be ruled out, given thatthe two parental populations overlap widely inrange, remain completely distinct phenotypicallyand ecologically, and, in the case of Papio and Thero-pithecus, are known to have done so for severalmillion years?

Presuming that some of the hybrids produced areat least partially viable and fertile, the effect ofsporadic hybridization is to introduce a trickle ofgenes from one population to the other, the directionof flow guided largely by sex-specific, inter-speciesinteractions. In the Papio � Theropithecus geladacase, for example, F1 females would probably re-main in the baboon troop where they were born, andpresumably would then mate with resident malePapio baboons to produce backcross offspring. MaleF2s would presumably emigrate, and females wouldremain in their mothers’ group. With each genera-tion of backcrossing, the admixture would becomeless phenotypically visible. The backcross Papio �(Theropithecus � Papio) female produced in BihereTsige Zoo in Addis Ababa, when observed as a youngadult, was phenotypically very close to a normalhamadryas female. Her 0.25 gelada inheritancewould not be readily apparent in a wild hamadryasgroup. Similarly, in the Macaca nemestrina � M.fascicularis case, hybrid females would presumablyremain in their mothers’ troop, while males woulddisperse to spread their “foreign” genes to othergroups.

All this is, of course, quite speculative. The criticalfield genetic studies to detect introgression in thesympatric populations of baboons and geladas inEthiopia, and macaques of different species-groups


in overlap areas of east Asia, have yet to be carriedout. Parallel cases in other vertebrates, however(e.g., Goodman et al., 1999; Lehman et al., 1991;Arntzen and Wallis, 1991), suggest that cryptic in-trogression of genetic markers is likely to be muchmore widespread than suggested by the rate of pri-mary hybridization, or observable phenotypic hy-bridity. For example, where native red deer (Cervuselaphus) are sympatric with introduced sika (Cervusnippon) in Britain, only 0.1–0.2% of matings werecross-specific, yet 66% of phenotypic sika, and 33%of phenotypic red deer, carried allelic evidence ofmixed ancestry (Goodman et al., 1999). Presumably,this indicates that prezygotic barriers become lesseffective as backcrossing progresses, and maladap-tive genetic combinations are weeded out by selec-tion.

As a source of new variation for the recipientpopulation, genes entering by hybridization aresomewhat comparable to mutations. Unlike muta-tions, however, immigrant genes have been tested ina donor population. They may be disadvantageous intheir new genetic and ecological setting, but areunlikely to be lethal, and in general the chance oftheir being advantageous must be better than forrandom mutations. Moreover, whereas advanta-geous, random mutations are most unlikely to recurin the same population and the same form, hybrid-ization will present a steady, even if slow, supply ofidentical “immigrant” alleles for selection in the re-cipient population.

Most immigrant genes that are neutral in effectwill disappear by drift in a few generations, thougha random minority will persist or even increase.Natural selection will presumably remove genes de-termining species-specific adaptive characteristicsof the donor species, those that are directly relatedto mate preference, and any that are functionallyincompatible with the host’s genome at a molecularlevel, together with neutral markers closely linkedto any of these. Any alleles that are universallyadvantageous (and hitch-hiking markers geneticallylinked to these) could become rapidly established bypositive selection in the recipient species. In evolu-tionary perspective, the effect would be homoplasy,with similar adaptive features appearing in sepa-rate species lineages.

Is hybridization of this kind (minimal but possiblyinfluential gene flow between fully differentiatedtaxa, i.e., entities that any paleontologist would behappy to call different species, if not genera) likely tohave occurred among basal hominins? We cannothope to answer this question directly, since even if asteady trickle of genes passed between populations,the chances of finding a recognizable hybrid individ-ual as a fossil (a true morphological intermediatelike the F1 geboons; Jolly et al., 1997) must be van-ishingly small. Testing the proposition that basalhominins were “crossable” thus depends on the anal-ogy of the papionins and other vertebrates, espe-cially mammals (e.g., cervids, camelids, bovines,

equids, or canids), among which hybrids are com-monly produced between parental species of equiv-alent degrees of evolutionary divergence.

One potential objection is that some of these casesof hybridization, especially those between relativelydistant parental species (Gee, 1999), occurred insituations that were entirely artificial (such as in acage), or where human influence was strongly sus-pected. As has been noted for many other taxa (e.g.,Carr et al., 1986; Lehman et al., 1991; Struhsaker etal., 1988), interspecific hybridization is most likelywhen sex ratios in one or both species are locallybiased. This situation may be an effect of humandisturbance or hunting, but it can also occur natu-rally, especially on the edge of the range of onespecies. In most papionin societies, wandering bach-elor males are a perennial feature. An emigrantmale in a marginal population might find few con-specific mates, and would therefore try to mate withany sympatric females with which he shared enoughof his mate-recognition system. Cross-taxon matingis presumably rare even when conditions favor it,and favorable conditions are themselves relativelyunusual. Nevertheless, they are not unrealistic, andcan be expected to occur even without human inter-vention, especially when environments are unsta-ble. Even situations mimicking the extreme situa-tion of a zoo colony, bringing together a fewindividuals of species that can interbreed, but rarelyor never meet in the wild, are quite conceivablewhere species ranges are fragmenting or adjustingrapidly to climatically or tectonically driven environ-mental shifts. If it were not for the barriers createdby people, for example, a relatively slight climatic-vegetational shift in the Arabian peninsula couldallow hamadryas baboons to spread from theirpresent range in the southwest to the Persian Gulfcoast, and hence into Iraq, Iran, and the Indiansubcontinent, and eventually “bachelor” hamadryaswould meet a similarly expanding population of rhe-sus monkeys. While most of the resultant rheboonswould undoubtedly perish without issue, repeatednatural expimentation might produce a few fertilesurvivors.

It might also be objected that papionin hybridiza-tion is a poor analogy for hominins, first becausesuch hybridization has never been demonstratedamong extant hominines, and second because thepropensity to hybridize successfully across wide tax-onomic gulfs is probably a papionin peculiarity, de-pendent on their unusual karyotypic uniformity (notto mention their notorious sexual promiscuity). Thisargument is worth examining more closely. It is thecase that gorillas, chimpanzees, and bonobos havenever been reliably shown to hybridize in the wild orin captivity. Rumored wild chimpanzee-gorilla hy-brids have so far always turned out to be one or theother, and extreme differences in genital anatomy,at least, might be expected to limit the possibilitiesof unaided cross-fertilization. Chimpanzee-humanhybrids are occasionally rumored, but the one sup-


posed hybrid individual I have seen (“Oliver”) wascertainly a mutilated Pan troglodytes. Pan troglo-dytes and P. paniscus do not presently meet in thewild, and if opposite-sexed adults of the two formshave been housed together in captivity, the fact, andits outcome, have not been reported as far as I know.In general, successful captive breeding of hominineapes postdates the era in which primates werehoused haphazardly in mixed-species groups, whenmost reported monkey hybrids were produced (e.g.,Gray, 1954).

Assuming, however, that papionins are indeedmuch more crossable than extant hominine apes,can the difference be attributed to gross karyotypicfactors? The papionins share a 2N � 42 karyotypethat is very similar in overall morphology, thoughthe Cercocebus-Mandrillus clade appears to carry aminor chromosomal rearrangement as an autapo-morphy (Dutrillaux et al., 1982), and Macaca fas-cicularis may also have its own minor rearrange-ment. The papionin karyotype has also retained ahigh level of structural homology, even in the mostdistantly related branches (Papio vs. Macaca, Mooreet al., 1999). By contrast, the extant hominines arekaryotypically diverse, with humans (2N � 46) dif-fering from both Pan and Gorilla (2N � 48) (Nick-erson and Nelson, 1998; Jauch et al., 1992).

The role of gross chromosomal rearrangements inthe evolution of reproductive isolation, especially inanimals, is debatable, however. A recent review con-cluded that while speciation driven by gross karyo-typic rearrangement may occur, the “more widelyheld view is that the accumulation of chromosomaldifferences between populations is largely incidentalto speciation” (Rieseberg, 2001, p. 351). Even if chro-mosomal rearrangements are implicated in the evo-lution of hybrid sterility or inviability, changes largeenough to be seen in a metaphase karyotype may notbe more influential than minor changes invisible atthat scale. The better-supported alternative view isthat genetic changes, both small and large, and in-cluding a few large enough to be visible as karyo-typic mutations, accumulate stochastically, as a sideeffect of overall divergence. Some of these changesmay increase genetic isolation, at times dramati-cally (O’Neill et al., 2001), by introducing prezygoticbarriers to hybrid formation, by reducing hybridfitness, or by decreasing recombination rates(Searle, 1998), but such effects are as likely to followfrom physically small chromosomal rearrangementsas large ones. This incremental model seems to ac-count better for the facts of hybridization among thecatarrhine primates, both hominoid and cercopithe-coid. Thus, successful hybridization can occur be-tween some species within Cercopithecus (Dutril-laux et al., 1982) and Hylobates (Van Tuinen et al.,1999) that differ in gross chromosomal morphologyand number, as well as karyotypically uniform pap-ionins, while karyotypically similar chimpanzeesand gorillas do not interbreed. Among the papionins,with their highly conserved, 42-chromosome karyo-

type, hybrid viability and fertility range from thesterile and dysfunctional rheboon, to geboons, inwhich F1s function and breed, at least in captivity,and Macaca fascicularis/nemestrina hybrids, whichwere functional and probably fertile in the wild.Although the sample of papionin cases is small, andthe data are incomplete, they do not support thenotion that members of this tribe are exceptional intheir ability to form viable and fertile hybrids. Theyare, however, consistent with the hypothesis that inall catarrhines, hybrid fitness is more or less in-versely proportional to the time since divergence ofthe parental species. Chimpanzee and gorilla stocksseparated shortly after rhesus and baboons, twice asearly as baboons and geladas, and about three timesas long ago as macaque species-groups. If we assumea common catarrhine scale of stochastic divergenceleading to reproductive isolation, their lack of natu-ral hybridization is not unexpected.

Of course, we have no idea when in the course ofhominin evolution the gross karyotypic autapomor-phies of the human species were acquired, how theywere distributed across basal hominin lineages, andwhat, if any, effect they had on the ability to hybrid-ize. If the present interpretation is correct, however,we do not need to reconstruct the karyotypes ofhominin lineages coexisting 3–4 million years ago topredict that most of them retained an ability toexchange genes occasionally. Such lineages cannothave been separate from each other for more than 3ma, and from the common chimpanzee stem perhaps1 ma more. The vicariance events, population splits,and genetic divergences that resulted in hominincladogenesis were, at the time, much more recentthan the Papio-Theropithecus split is today. If weassume a common catarrhine scale of crossability,basal hominins were, on average, somewhere be-tween macaque species-groups and baboon-geladain their ability to hybridize. This suggests thatwhenever extrinsic, premating barriers were low-ered (e.g., if colonizing individuals of opposite sexand different species entered a habitat isolate simul-taneously), the basal hominins could have producedviable and fertile hybrids, both among themselvesand perhaps also with species of the chimpanzeeclade. Such hybrids could serve as a conduit for theflow of advantageous genes between the parentalpopulations, or could even be, themselves, thefounders of new species.

It should be emphasized that such shenanigans inno way represent an argument for reducing the hy-bridizing forms to congeneric status, let alone con-specificity. Occasional, opportunistic hybridizationis consistent with separate species status under anycurrent definition. It is also completely compatiblewith maintenance, and even reinforcement, of diver-gent adaptive trends in the parental species. Papioand Theropithecus populations, for instance, haveprobably been sympatric, and occasionally exchang-ing genes, for several million years, but have re-mained separate entities with their own distinct and


diverging evolutionary trajectories, and have alsothemselves speciated internally (Delson, 1993; Jab-lonski, 1993), attributes that systematists generallyrecognize by separation at the genus or subgenuslevel.

The analogy of hybridizing papionins does not, ofcourse, prove that hominins did hybridize, but itdoes suggest that crossing might have been possible,even between forms as distinct morphologically (andpresumably ecologically) as Paranthropus boisei andHomo (now Kenyanthropus) rudolfensis, or Austra-lopithecus afarensis and K. platyops, or Ardipithecusramidus kadabba and the proto-chimpanzee. Aslong as they remained ecologically distinct, occa-sional gene flow would not necessarily underminetheir distinct adaptations, because selection amongbackcrosses would effectively prevent maladaptivegenes moving between them. But by the same token,advantageous genes could be rapidly incorporated,without disrupting the balanced, divergently adap-tive genotypes of the parental species. Contrary tothe standard model, the fuzzy zone in which evolu-tion is somewhat reticulate as well as divergent canbe prolonged well beyond the point of adaptive, andpaleontologically documented, morphological diver-gence.

But could limited gene-flow of this kind causepaleontologically recognizable anomalies in the dis-tribution of morphological character states? To an-swer this question, we need to know much moreabout the genetic basis for such traits (Weiss, 1994;Weiss and Buchanan, 2000). Those dependent on anintegrated complex of genes at unlinked loci may notsurvive the recombination inherent in backcrossinglong enough to be selected, whereas traits with asimple genetic base should be more visible, bothpaleontologically and to natural selection. The pos-sibility of significant intertaxon gene-flow shouldcertainly be entertained. Increase in relative brainsize, highly homoplastic in hominins by any cladisticpermutation (Strait et al., 1997), would be an obvi-ous candidate.

From the cladist’s perspective, opportunistic in-terlineage gene-flow represents yet another waythat species might acquire similarities that do notreflect their phylogeny, at least as that term is gen-erally used. If the assumption of speciation by cleanbreaks without subsequent reticulation is relaxed toallow some hybridization-driven reticulation, onecan often reduce significantly the number of steps inthe most parsimonious cladogram (Haszprunar,1998), but presumably the probability of the hybrid-ization itself should be factored into any parsimonycomparison, and it is difficult to see how this mightbe done.

Considering the possibility of occasional reticula-tion, along with the adaptive homoplasy and crypticsymplesiomorphy that were probably rife in basalhominin evolution, Alexander Papyoe feels com-pelled to ask whether a reliable cladogram of thegroup may be a practical impossibility. And, if so,

does this fact significantly affect our understandingof early hominin evolution? Certainly, as Tattersalland Eldredge (1977) have long argued, establishingthe cladistic relationships among a cluster of relatedtaxa is a logical and highly desirable first step to-wards understanding their evolutionary history: notonly their phylogeny, but also causal factors such asdispersal, vicariance, and adaptation. The precedingdiscussion, for example, includes several interpreta-tions that would not have suggested themselves be-fore (((Homo, Pan) Gorilla) Pongo) replaced ((Homo(Pan, Gorilla)) Pongo) or even (Homo, ((Pan, Gorilla)Pongo)) as the accepted cladogram of extant Homi-nidae. However, it could also be argued that theimportance of determining the correct cladogram, aswell as the practical possibility of doing so, de-creases in proportion to the lengths of its internodes.For example, two “robust” hominin species derivedin quick succession from the same, or closely similar,“gracile” stocks might well have been very similar asliving animals, as well as unrecognizable as sepa-rate clades by any methods we can bring to bear ontheir fossil remains. Moreover, for a considerabletime after they acquired their distinctive, derived,adaptive traits, they would probably have been in-terfertile both with each other and with their respec-tive “gracile” sister taxa. While a reliable cladogramof early hominins remains a worthwhile goal, someof its details may be inherently insoluble. (Thisdoubt seems to be increasingly shared; recent com-mentaries (Wood and Brooks, 1999; Collard and Ai-ello, 2000) tend to omit lines of descent from theirdepictions of hominin phylogeny).

Fortunately, the early hominins present plenty ofother important questions: numerous problems ofalpha taxonomy, for example, and paleobiology atthe local deme level (e.g., Lee-Thorp and van deMerwe, 1993; Teaford and Ungar, 2000; Grine andKay, 1988). which (again on a papionin analogy;Benefit, 2000) was probably more diverse than wepresently perceive.

The papionin analogy (supported, in this case, byanalogies from other, similarly divergent mamma-lian clades) suggests that the hominin “morphs” thatwe recognize as synchronous, often sympatric, andecologically distinct species and genera, were prob-ably capable of limited interbreeding, even after sev-eral million years of divergent adaptation.

By extension, the papionins suggest that homininlineages were even more capable of interbreeding atan earlier stage of differentiation, after newly emer-gent stocks had become widespread, and when pop-ulations were diverging by drift, adaptation to localconditions, and econiche specialization. At thisstage, gene flow between them would depend less ontheir inherent reproductive and genetic compatibil-ity (their “crossability”) than on extrinsic factors:accidents of biogeography that created or removedphysical barriers to contact between them, and theintensity of disruptive selection that maintained thedistinctness of their gene-pools.


Several papionin taxa (especially species withinthe genus Papio and within Macaca species-groups)have presently attained this stage of differentiationand exemplify in their contemporary populationstructure the complexities that occur among partialisolates (e.g., Hoelzer et al., 1994). There seems noreason to doubt that the history of hominins fromthe late Miocene onwards was at least as complex,and equally intractable to adequate description interms of a simple, ramifying phylogeny and taxon-omy. We can hope that the paleogeographic pictureand early hominin fossil record will eventually be-come sufficiently continuous and fine-grained intime and space to allow some specific comparisonsand predictions. This remains a distant goal forearly hominin history, and meanwhile, AlexanderPapyoe suggests that early hominin phylogenies andtaxonomies should be regarded as disposable ap-proximations.

The same principle applies to the most recentradiations within Homo and Papio, but here thelevel of documentation has reached the point wherethe complexity of the problems, if not their solution,can be discerned.

HYBRID ZONES AND POPULATIONREPLACEMENTS IN BABOONS AND HUMANS

The “Neandertal problem” has had different con-notations for every generation of human evolution-ists. One constant theme, however, has been thedegree of distinctness between Neandertals and an-atomically modern people. In contemporary discus-sions, this question tends to be couched in taxonomicterms: whether or not the Neandertals should beclassified within Homo sapiens, or, more generally,how many species of the genus Homo existed con-temporaneously during the later Pleistocene, say,from ca. 150–30 ka. Answers to the second questionfrom contemporary paleoanthropologists range froma firm “one” to an equally positive “unknown, butcertainly several.” Even a cursory review of the “Ne-andertal problem,” which has produced several largevolumes and innumerable papers over the past fewyears, is far beyond the present scope. However,Alexander Papyoe suggests that a new angle or twomay emerge by comparing it to an analogous prob-lem in Papionini: the question of how many speciesshould be recognized among contemporary forms ofthe genus Papio, or at a more particular level,whether the olive baboon, for example, should beconsidered a different species from the hamadryas,or the yellow baboon, or both.

In both cases, real biological issues are often ob-scured by the perennial but inevitably unproductivediscussions about what a species “really is.” A recentreview (Hey, 2001) counted 24 active species defini-tions. Those seemingly most widely used among zo-ologists and paleontologists (the “biological” and the“phylogenetic” species concepts; BSC and PSC, re-spectively) are both population-based. The disagree-ment between them seems to hinge on how much

emphasis to put on the distribution of probabilitiesof zygote formation (“zygostructure;” Jolly, 1993), orthe distribution of traits within and among popula-tions (“phenostructure”). The BSC emphasizes zygo-structure (species as reproductive isolates), whilethe PSC emphasizes phenostructure (species as con-sistently diagnosable clades). Justification for theBSC tends to look forward in time, emphasizing thatif genes can flow across populational boundaries, allpopulations so linked can at least in theory evolve asa single unit, and that gene-flow might even homog-enize them in the future. Adherents of the PSC tendto look backwards in time, emphasizing species asterminal taxa in a cladogram defined by their his-torical relationships. The PSC and the BSC areequally valid descriptions of aspects of populationstructure resulting from evolutionary processes.Both are compatible with the traditional, ontologi-cally appealing but epistemologically weak conceptof a species as a cluster of populations with a com-mon and distinct evolutionary trajectory. Neitherkind of species is more “real” than the other; bothare abstractions from the observable attributes oforganisms. My nominalist bias suggests that, sinceboth species definitions use the same basic at-tributes of populations (pheno- and zygostructure),merely differing in how to weight them, we shouldfocus on describing these attributes, and shelve in-definitely the largely bogus “species problem” (Jolly,1993).

In taxonomic practice, the BSC-PSC distinction ismost significant where phenotypically distinct,parapatric populations interbreed at their bound-aries. The PSC calls such populations “species” (inspite of the fact that hybrid individuals can be as-signed only arbitrarily, and thus all individuals willnot be strictly assignable). The BSC calls them semi-species of a superspecies, or subspecies of a Rassen-kreis or polytypic species, since they are potentiallylinked by gene-flow (even though in most cases nogene-flow has been demonstrated). Because such sit-uations are very common in nature, the seeminglyminor difference in species definitions greatly affectsthe number, and size, of recognized species. More-over, within the BSC, many closely related “species”are very similar in diversity and scale to “subspe-cies,” differing from the latter only in that marginalgene-flow has not been shown to occur. If such gene-flow is later demonstrated, the discovery mandates awholesale taxonomic revision (for the baboon case,cf. Thorington and Groves, 1970; Groves 2001).

The population structure and dynamics of taxa atthis level exemplify crucial evolutionary processesgathered under the rubric of “speciation” (Barton,2001; Turelli et al., 2001): the evolution of pheno-typic distinctiveness or reproductive isolation, orboth. Investigation of these processes is too biologi-cally important to be sidetracked by the empty se-mantics of the “species question,” and to avoid thispitfall, Grubb (1999) introduced the concept of theallotaxon. Allotaxa are phylogenetically close, but


well-differentiated and diagnosable, geographicallyreplacing forms whose ranges do not overlap, but areeither disjunct, adjoining, or separated by compara-tively narrow zones in which characters are clinallydistributed. Related allotaxa typically exhibit dis-tinct adaptations to their respective habitats, buttheir defining allo- or parapatry suggests that theyare close enough ecologically and behaviorally topreclude actual coexistence within the same ecosys-tem. Where the ranges of allotaxa meet, often at anecotone, a zone of intermediate phenotypes ( a “hy-brid zone”) frequently occurs. Such zones have beenaptly called “natural laboratories of the evolutionaryprocess” (Harrison, 1993), and a lively biological spe-cialty has grown up around their theoretical andempirical study (Barton and Hewitt, 1985, 1989;Harrison, 1993), the latter greatly facilitated by thegrowing availability of genetic markers (serological,allozymic, and most recently genomic). The use ofgenetic markers, combined with advances in paleo-climatology, has added a fourth dimension to thepicture by documenting genetic patterns that can berelated to the late Neogene history of repeated, oftenvery rapid, environmental change. The dynamics ofsuch processes have been most thoroughly investi-gated in Europe (Hewitt, 1996, 2001) and NorthAmerica (Avise, 1994), but there is ample evidencethat environmental fluctuations were equally influ-ential in the tropics and subtropics, and in thesouthern hemisphere, though the nature and distri-bution of barrier zones in these areas is less obvious.

The model developed by Hewitt (1996, 1999, 2001)describes a plausible process by which late Neogenepaleoclimatic oscillations converted local popula-tions within species into fully diagnosable allotaxa.Responding to rapid habitat shifts, the ranges ofgeographical populations would have expanded andcontracted, repeatedly dividing and reuniting, forc-ing equally rapid fluctuations in population size.The core area of each population, habitable duringadverse climatic periods, tends to preserve ancientgenetic variation, both adaptive and random. Colo-nizing subpopulations at the fringe individually losevariation, but local adaptation, drift, and founder-flush effects may cause them to become diverse andgenetically idiosyncratic. The descendants of “colo-nists” dominate numerically during expansions, andas populations expand it is the demes at theirfringes, which are the most genetically derived, thatmeet their neighbors. Repeated contraction and ex-pansion from separate refugia thus tend to generateand accentuate the kind of genetic differences thatdefine adjacent allotaxa, and sometimes push theprocess to the point of partial or complete reproduc-tive incompatibility (Hewitt, 2001).

When such differentiated allotaxa make second-ary contact, initially at a sharp boundary, interac-tions between them may fall anywhere betweencomplete interfertility and total reproductive isola-tion. Secondary hybrid zones generated by marginalinterbreeding are often quite restricted in breadth.

This spatial restriction may indicate that contact istoo recent for the interpopulational clines to havereached equilibrium, or that habitat-specific or in-trinsic hybrid disadvantage is balancing outwardand inward gene-flow. Field studies of particularhybrid zones have revealed a great variety of struc-tures and dynamics (Barton and Hewitt, 1985; Har-rison et al., 1987).

Like many other terrestrial organisms, extant Pa-pio baboons (and Macaca species-groups) show atypical “patchwork quilt” internal phenostructure ofgeographically replacing, parapatric allotaxa (Fig.1). Though many details and some crucial areasremain to be investigated, the distribution of baboonallotaxa (Jolly, 1993; Groves, 2001) is better knownthan implied by their chronically confused formaltaxonomy. Five “forms” are customarily recognized,either as PSC species (Groves, 2001) or BSC subspe-cies (Williams-Blangero et al., 1990; Jolly, 1993), oras a species of a “superspecies.” At least 2 of the 5include two or more allotaxa, which, strictly, shoulddisqualify them as single PSC species. The geo-graphically circumscribed allotaxa are distin-guished most readily by characters of pelage textureand color. Dental and cranial features also differamong allotaxa (Jolly, 1965, 1970b; Phillips-Conroy,1978), but are less diagnostic. The precise number ofallotaxa that can usefully be recognized is still un-certain; some “forms” that have been described andnamed, especially in the east African corridor fromTanzania to Sudan, may turn out to be hybrids orsegments of a continuously varying cline (Groves,2001). The number of allotaxa recognized also de-pends upon how broad an intergradation zone thesystematist is prepared to tolerate between them.For example, the small yellow baboon (Papio kindaeor P. cynocephaus kindae) has a wide range thatstretches from Angola to the Luangwa Valley inZambia. It is quite distinct from the large, “typical”yellow baboon of Tanzania and Malawi (P. cynoceph-alus, sensu stricto), but the cline that links them isrelatively broad. Conservatively, 8 or 9 allotaxa arerecognized here.

In the cohesion species concept of Templeton(1989), a crucial criterion of conspecificity is thatconspecific organisms should be ecologically inter-changeable. Like other species criteria, this one sit-uates the allotaxa of Papio on the cusp of speciesstatus. On the one hand, the ranges of some Papioallotaxa roughly correspond to broad vegetationalzones, and the lines of contact of parapatric allotaxaoften fall close to an ecotone (Jolly, 1993; Kingdon,1997). In at least one case (the hamadryas), we candemonstrate a pervasive pattern of subtle, physio-logical (Kaplan et al., 1999), behavioral (Kummer,1968), and anatomical (Jolly and Phillips-Conroy,2001 and unpublished data) adaptations to thesemidesert habitat in which most populations arefound. Other habitat-specific adaptations (e.g., yel-low baboons to the woodland vegetation of the Afri-


can southern seasonal tropics) have been suggested(Kingdon, 1997), but have yet to be demonstrated.

On the other hand, ecological similarity is indi-cated by the fact that Papio allotaxa are parapatric,with ranges that meet but do not overlap, presum-ably because groups from either side of the allotaxonboundary interact ecologically and socially as com-petitors when they meet. At allotaxon interfaces,individuals are in fact literally interchangeable, be-cause they migrate across allotaxon lines and suc-cessfully take up residence in a “foreign” group (Al-berts and Altmann, 2001; Samuels and Altmann,1986; Phillips-Conroy et al., 1992), where their eco-logical behavior and preferences are indistinguish-able from those of their hosts (Nystrom, 1992).Moreover, a closer examination of the range of thevarious allotaxa reveals many exceptions to thebroad ecological associations. For example, hamadr-yas baboons in Eritrea occupy comparatively moistmontane habitats that immediately to the south, in

the Ethiopian highlands, are occupied by anubisbaboons (Zinner and Hapke, 2001). Conversely, it isanubis, not hamadryas, that occur in the semidesertSaharan massifs of Tibesti and Air (Jolly, 1965). AllPapio baboons are ecological generalists, a trait thatperhaps makes them especially relevant to humanevolution. Their modest, habitat-specific ecologicaladaptations have not become specializations pre-cluding expansion into neighboring habitats. Conse-quently, the distribution of each allotaxon seems toresult not so much from ecological determinants orpreferences, as from population history, some of itquite recent. For instance, the ecologically anoma-lous distribution of anubis and hamadryas in Eri-trea may be due to the fact that hamadryas reachedand colonized the moist highlands first, and have yetto be displaced by anubis, which are still confined, inlow numbers, to the western savanna lowlands (Zin-ner and Hapke, 2001). This hypothetical scenario issupported by observations from western Tanzania

Fig. 1. Geographical distribution of some of the recognizable allotaxa of Papio baboons.


(J. Moore, personal communication), Ethiopia (Phil-lips-Conroy et al., 1992), and southeastern Kenya(Maples and McKern, 1967; Alberts and Altmann,2001). In each of these areas, anubis baboons seemto be expanding their range, at the expense of neigh-boring yellow baboon and hamadryas populations,partly by displacement, and partly by genetic infil-tration.

In captivity, all allotaxa appear to hybridize indis-criminately (Jolly, unpublished data), and there isno evidence for hybrid breakdown, behavioral in-compatibility, or intrinsic sterility. Similarly, thereis no evidence that Papio baboon allotaxa ever avoidinterbreeding when they meet in the wild, thoughmany boundary areas have yet to be investigated.The fact that documented baboon hybrid zones arenarrow, in spite of the lack of obvious, intrinsicbarriers to gene-flow, strongly suggests that theyare the result of secondary contact following rangeoscillations (Barton and Hewitt, 1985; Hewitt, 2001;Harrison, 1993).

In East Africa, where genetic sampling has beenmore dense than elsewhere, the distribution ofmtDNA haplotypes across baboon allotaxa stronglyhints at previous cycles of hybridization. In partic-ular, all haplotypes so far found in anubis baboonsin Ethiopia form a clade that is distinct from, butrelated to, haplotypes of Ethiopian hamadryas ba-boons, while haplotypes of anubis baboons from Ke-nya are closer to those of Kenyan and north Tanza-nian yellow baboons (Wildman, 2000 and personalcommunication; Newman et al., 2001). In externalphenotype, all anubis baboons are (by definition)quite different from either yellow or hamadryas, andmoreover, Kenyan and Ethiopian anubis baboonsare externally indistinguishable. This combinationof phenotypes and haplotypes is difficult to explainexcept by a scenario involving a cycle of hybridiza-tion previous to the present one, with sex-specificintrogression, and probably also radical fluctuationsin allotaxon ranges (Wildman, 2000 and personalcommunication; Newman et al., 2001). With benefitof hindsight, we can see vestiges of this earlier hy-bridization cycle in the skull structure of Ethiopianand Kenyan anubis baboons, which in some aspectsof size and shape resemble those of neighboringhamadryas and yellow baboons, respectively (Jolly,1965 and unpublished data). If this scenario is ap-proximately correct (and much more genetic work isneeded to test it fully), it suggests that the pelagephenotype defining anubis baboons, which is verystable not only in Ethiopia and Kenya but across thewhole northern savanna belt as far as Sierra Leone,was already characteristic of the baboons contribut-ing the anubis genes to the mix in Kenya and Ethi-opia. If this is so, it must have originated well beforethis round of hybridization.

The anubis phenotype itself may have originatedin a still earlier round of hybridization. MtDNA(Wildman, 2000; Newman et al., 2001), nuclear ge-netic (Williams-Blangero et al., 1990), and pheno-

typic (Jolly, 1965, 1993) evidence all suggests thatthe primary division among stocks leading to extantPapio was between a “southern” branch, ancestral toextant chacma baboons, and a “northern” one, an-cestral to all the others, and that this split occurredabout 1.7 ma. Genetically and phenotypically (aswell as geographically), anubis baboons are “north-ern,” but have some “southern” traits such as tailshape (Jolly, 1965). As Kingdon (1997) suggested,the anubis phenotype may have originated as a sta-bilized hybrid, in a small, isolated “northern” popu-lation, resembling Guinea baboons, that receivedand incorporated “southern” immigrants, resem-bling chacmas, via a glacial-period corridor throughthe central African rainforest. This scenario willrequire testing against extensive genetic informa-tion from anubis baboons of west-central Africa,which at present are totally unknown.3

The complex populational history of Papio is notunique, or probably even unusual, among taxa ofcomparable time-depth. Among the papionin mon-keys, observational and genetic investigation of spe-cies-groups of macaques has revealed a phylogeog-raphy that is at least equally complex, involving alldegrees of hybridization, sex-specific gene-flow, andsecondary fusion between differentiated taxa(Fooden, 1963; Bynum et al., 1997; Tosi et al., 2000;Hoelzer et al., 1993; Melnick and Hoelzer, 1992;Evans et al., 1999).

How can these speculative scenarios about popu-lation structure and phylogeographic history in Pa-pio baboons be translated into useful analogies forunderstanding evolution within the genus Homo? Acomparison of chronologies indicates that the initialdiversification of stocks leading to extant Papioforms (Wildman, 2000; Newman et al., 2001) is com-parable in age (at about 1.7 ma) to the origin andrapid deployment of Homo, sensu stricto (i.e., theclade stemming from African Homo erectus a.k.a. H.ergaster, but excluding “H.” habilis and “H.” ru-dolfensis) (Wood and Collard, 1999). If we can in-deed assume a common timescale of average, intrin-sic reproductive isolation for all catarrhines, asargued above, this suggests that all human lineagesstemming from the H. ergaster stock were probablyas fully interfertile as are extant Papio populations.On these grounds, they could be regarded as mem-bers of a single, polytypic (BSC) species (cf. Wolpoffet al., 1993; Hawks et al., 2000). Several caveatsshould be observed, however.

First, the postulated “common catarrhine cross-ability scale” is at best a stochastically driven ap-proximation that is necessarily less predictive forparticular cases and over shorter timescales. Muta-tions contributing to reproductive isolation, perhaps

3The phyletic position of yellow baboons is ambiguous. Their genet-ics seem to ally them closely with anubis baboons (Newman et al.,2001; Williams-Blangero et al., 1990), but they have been sampledonly from East Africa, close to a zone of hybridization with anubis.Morphologically they are certainly “southern.”


initiating a rapid “cascade” to full intersterility withneighboring populations (Barton, 2001; Rieseberg,2001), could arise at any time in any geographicallyisolated lineage. Such an event in a human lineagewould be difficult, if not impossible, to detect fromfossil evidence.

Second, it is important to stress that if we reducerecognizable “forms” of Homo (currently, oftennamed as PSC species such as H. neanderthalensisand H. erectus) to (BSC) subspecific status becauseof the possibility of some marginal gene flow be-tween them, this would not imply that they were“ephemeral” or “evolutionarily unimportant,” anymore than these terms could be applied to, say,anubis baboons. This point is worth emphasizing,because a suggestion to the contrary seems to havecrept into the debate over the implications of theLagar Velho “hybrid child” (Duarte et al., 1999). Itseems to represent a semantic confusion. If indeedthe Lagar Velho child was the result of marginalgene-flow between Neandertals and “moderns,” itwould prove them to be conspecific by the BSC, butnot by the PSC. But it is only PSC taxa that aredefined so closely that intraspecific variation is, al-most by definition, unimportant. Large, polytypicBSC species can and often do include persistent,diagnosable, “important” allotaxa, named as subspe-cies.

Nor would the assumption of universal interfertil-ity within the genus Homo (strictu senso) conflictwith evidence pointing to long-term, consistently di-agnosable human lineages, such as a pre-Neander-tal-to-Neandertal lineage persisting for �300,000years in Europe (Trinkaus, 1991, 1993; Rosas, 2001),or Homo erectus populations locally surviving theorigins and spread of “modern” humans (Swisher etal., 1996). Even if they were not surrounded by in-trinsic barriers to interbreeding with their neigh-bors, the genetic integrity of such populations couldbe effectively maintained by a combination of peri-odic geographical barriers, selection for habitat-spe-cific ecological adaptations, behavioral differences(which themselves could have a combined geneticand environmental basis), and simple, outward pop-ulation-pressure. A PSC advocate could happilyname each one a full species, while a BSC enthusiastcould describe the same phenomenon as polytypyand regional continuity within a single species.From the point of view of hybrid-zone theory, suchcases just represent some of the many possible in-teractions between neighboring allotaxa.

As for the “Neandertal question,” it is a reason-able working hypothesis that, like baboons today,the genus Homo by the beginning of the Late Pleis-tocene presented a patchwork of allotaxa, each withits own history of successive glacial (cold-dry period)retreats and interglacial (warm-humid period) ex-pansions. As with the baboons, we can assume thatall human allotaxa alive at that time were notequally related to each other: that if a time-travel-ling geneticist could sample them in depth, genetic

markers would reveal a comparable, underlying cla-distic history, probably greatly complicated by suc-cessive events of marginal gene-flow and hybridiza-tion. Some human allotaxa (e.g., the one representedby the Ngandong specimens), analogous to Guineaor chacma baboons, might have been more-or-lessisolated and little changed since the initial deploy-ment of the genus Homo (strictu senso). Others,analogous to contemporary hamadryas, anubis, andyellow baboons of the East African corridor, proba-bly had a more recent and complex history of inter-mittent genetic interaction.

We can also assume that where they met, theboundaries between human allotaxa showed thekind of complex dynamics seen in other vertebrateand invertebrate contact zones. Unfortunately, inmost cases the meager fossil evidence limits us toobserving that there was appreciable regional vari-ation in cranial form among humans of this period(Wolpoff, 1999). The analogy of Papio (and othercatarrhines) suggests that early human allotaxawere probably much more distinct in the flesh thanfrom the kind of evidence available in the fossilrecord. There are significant average skeletal anddental distinctions among the baboon allotaxa, butthey are not easily seen without large samples andprior allotaxon assignment, based on soft-part (pel-age) characteristics. On the other hand, Papio doesnot altogether fit the predictions of “Tattersall’s law”(Tattersall, 1993), which suggests that any taxa di-agnosable on hard-part criteria can be assumed tobe even more distinct in the flesh. For example, the“kindae” and “typical” yellow baboon allotaxa havevery similar coloration (and are classified togetherin 1 of the traditional 5 species), but are easilydistinguishable on cranial and dental size, and areconnected by a phenocline in the wild. The onlygeneralization that we can make on the analogy ofthe baboons is that we should expect our diagnosis ofindividual fossil human specimens to be unreliable,even if the existence of significant phenostructuringcan be inferred from divergent central tendencies ingeographically defined clusters of specimens. Atbest, a much denser record will be required to dis-cern what the human allotaxa were, where theirgeographical limits lay, and how they interactedwith each other.

In fact, in the densest fossil record that we have(of the Neandertals and their neighbors to thesouth), the pattern on one corner of the Late Pleis-tocene human patchwork is dimly discernible. Leav-ing aside the largely semantic question of Neander-tal species status, and the issue of their geneticcontribution to later human populations, we findsubstantial agreement. It is generally accepted thatNeandertals were a localized, recognizable humanpopulation of Europe and southwestern Asia, com-prising an allotaxon distinct from pene-contempo-rary Afro-Arabian populations. The latter are repre-sented by specimens such as Skhul V, Kafzeh, andOmo II, which are widely believed either to repre-


sent the ancestral modern human stock itself, or atleast to be more closely related to it than the Nean-dertals were. If Neandertals and Africans came intoperiodic contact, this probably occurred at an eco-tone in the eastern Mediterranean, where an alter-nation of faunas and human allotaxa seems to haveoccurred in synchrony with glacial-driven oscilla-tions (Tchernov, 1989). This extended, marginal in-teraction between Neandertal and Afro-Arabian al-lotaxa, apparently similar in culture, technology,and ecological role in their respective habitats,would have been quite distinct from subsequent “re-placement” events.

During a period that was recently narrowed to36–30 ka, Neandertal morphology was replaced inEurope by a more “modern” type of human(Churchill and Smith, 2001), a local manifestation ofthe general replacement of “archaic” by “modern”humans that had begun some time before in thetropics. The interface between them is believed tohave moved quite rapidly from the southeast west-wards, reaching northwest Europe and Iberia last.The disagreement concerns dynamics at the inter-face.

When considering Neandertals’ interactions withneighboring populations, most paleoanthropologistshave drawn analogies from the behavior of extantHomo sapiens, assuming for instance that culturalfactors such as language (rather than “hard-wired”differences in mate recognition systems, believed tobe characteristic of other mammals) were the majordeterminant. An early, extreme form of the multire-gional hypothesis saw only cultural elements flow-ing between populations, and physical changes asindigenous adaptations to the resultant technologi-cal shifts. Certainly, only modern humans approachthe cultural sophistication inferred for both Nean-dertals and their contemporaries, but they present apoor analogy in that no two modern human popula-tions are as distantly related as were Neandertalsand the precursors of modern humans, or have hadas long a period of separation in which to acquiredistinct mate-recognition systems. Whatever thereason, it is now clear that the genetic structure ofthe extant human species is very unusual amongwidespread primate taxa. It is distinguished by apaucity of ancient genetic lineages, except in lociwhere diversity is likely to be maintained by selec-tion (Disotell, 1999), and in general is dominated bythe effects of massive population expansions, most ofwhich occurred in the past 20 ka (e.g., Harpendinget al., 1998). At the very least, this suggests thatadditional analogies should be sought outside thegenus Homo. Another reason for doing this is thatthe uniquely human, culture-driven, in situ conver-sion of Neandertals to “moderns” (Hawks and Wol-poff, 2001) without any appreciable populationmovement or “gene-flow” is now hard to reconcilewith the rather short timescale of replacement(Churchill and Smith, 2001), and has been aban-doned by its original formulators. As a result, dis-

cussion now focuses on whether replacement was amatter of gene-flow between static populations, oran expanding population replacing a shrinking one,or something in between. The problem is thusbrought into the realm of general hybrid-zone the-ory, in which newts (Arntzen and Wallis, 1991),crows (Saino et al., 1992), and baboons are at leastas relevant as human beings.

The fragments of Neandertal mtDNA sequence(Krings et al., 1997; Hoss, 2000; Ovchinnikov et al.,2000) suggest the point at which the Neandertalstory can be linked to the analogous history of ba-boons. Discussion of the Neandertal mtDNA se-quence has focused mainly on its relatively ancientseparation from the root of all extant human se-quences, and its implications for a Neandertal ge-netic contribution to modern human populations.From the baboon (or chimpanzee, or gorilla) perspec-tive, however, the separation is not very ancient. Itis comparable to �600 ka divergences between oliveand hamadryas baboon mtDNA haplotypes, andmuch more recent than, e.g., the Guinea-hamadryassplit. Mitochondrial diversity in Papio may be anal-ogous to the condition in Homo before the “event”(generally interpreted as an “out-of-Africa” expan-sion of a relatively small subpopulation) that elimi-nated most of the diversity from its collective mito-chondrial (and Y-linked, and autosomal) gene-pool.Unfortunately, investigation of continent-wide ge-netic phenostructure in Papio is still in its earlieststages, so we cannot pursue the analogy further inthis direction. We can, however, make some sugges-tions based on work in contemporary zones of hy-bridization, especially the Awash anubis-hamadryashybrid zone. For example, we can conclude that un-less an undocumented, radical genetic event oc-curred in the 600 ka since they shared mtDNA an-cestry with the Neandertals, premodern humanswere certainly able to interbreed with them andproduce viable, fertile, offspring, as hamadryas andanubis baboons do.

The baboon analogy does not, however, inspireconfidence that the detailed dynamics of “archaic”-“modern” interactions will ever be determined. Cer-tainly, finding any fossil documentation is quite un-likely, as obvious evidence of intergradation (or itsabsence) is likely to be seen only in a spatially (or,with a moving zone, chronologically) very restrictedzone at the allotaxon interface. In Ethiopian andKenyan baboon hybrid zones, as in many zones in-volving other taxa, there is a very narrow region inwhich most individuals are phenotypically obvioushybrids, yet such hybrids are found only within oneor two dispersal distance units (about 30 km) of thezone’s center, although genetic evidence of hybrid-ization is much more widespread. For the humancase, this has an important implication: demonstrat-ing phenotypic distinctness (lack of overlap) of Ne-andertal and “modern” samples drawn from areasremote in time and space from the zone of contactdoes not disprove the occurrence of interbreeding at


the interface. It also means that the Lagar Velhochild, if indeed it is a hybrid, is a rare and valuablefind, even though it is irrelevant to the Neandertal“species question,” and does not tell us whether Ne-andertals (or other “archaic” humans) contributedgenes to the Upper Paleolithic, or the extant, humangene-pool. Not that these are equivalent, as is oftenimplied; there was ample opportunity for the loss ofa few stray Neandertal genes from European UpperPaleolithic populations when the latter shrank andwere replaced by food-producing peoples.

If interbreeding did occur at the Neandertal-mod-ern interface, the number of different possible sce-narios of hybrid zone dynamics is enormous. Anynumber of factors besides intrinsic hybrid disadvan-tage could have restricted gene-flow out of the con-tact zone, or filtered it, or directed it asymmetricallytoward one population or the other. Only one suchzone in primates (between anubis and hamadryasbaboons in the Awash National Park, Ethiopia) hasbeen the subject of both detailed behavioral andgenetic work, and this has revealed a complex situ-ation in which gene-flow into and out of the hybridzone is directed and limited largely by idiosyncraticpatterns of social behavior distinctive of the twoparental taxa and their hybrids (Sugawara,1979;Nystrom, 1992; Bergman, 2000; Beyene, 1998).

In the Neandertal case, the fact that the interfacemoved historically from east to west indicates thatthe pressure of gene-flow was greater in that direc-tion; if a hybrid zone existed, the genes in it werecontributed disproportionately by “moderns.” “Ne-andertal morphological genes” may have been re-moved by natural selection from a narrow zone ofhybridization, or been swamped by differential ge-netic inflow, or perhaps they simply died out withtheir carriers without any hybridization at all. Anycombination of these factors could have contributedto their disappearance. A much more fine-grainedtemporal record of the transition would be necessaryto decide between these alternatives, and the precisescenario is immaterial both for the eventual out-come, and for the so-called “species question.”

What is important, and hotly contested, iswhether Neandertals (and other archaics) contrib-uted any genes to the gene-pool of the human pop-ulation who succeeded them. This would imply aflow of genes from the marginal hybrid zone into theexpanding modern population: swimming, as itwere, against the tide. The important question is notwhether Neandertals could have passed some genesby hybridization to incoming Afro-Arabians; theyalmost certainly could. It is certainly not the neoes-sentialist (Cartmill, personal communication) redherring of whether or not they were “really” differ-ent species. The important questions are purely em-pirical: first, whether they actually did contributeany distinctive alleles to the incoming population,and second, whether any of these have survivedpost-Pleistocene upheavals in the human gene-pool.The first question can only be answered by genetic

investigation of the DNA of post-Neandertal fossilhumans (cf. Hawks and Wolpoff, 2001); the secondby trawling the extant human gene-pool itself.

So far, almost all genetic systems investigated inextant humans show no signs of a Neandertal inher-itance, but perhaps we need to be more selective inour search. A moving hybrid zone may leave in itswake a few neutral markers derived from the re-treating population (Arntzen and Wallis, 1991), butthese are likely later to be eliminated by drift. Mostlikely to survive and be incorporated are genes fortraits strongly favored by local conditions (and“hitch-hiking” markers linked to these). Some yearsago, a popular work (Kurten, 1971) plausibly sug-gested that Neandertals were blond and blue-eyedin adaptation to cloudy, periglacial Europe, whileincoming “moderns” had the darker pigmentation ofa subtropical people. Perhaps we should survey nor-dic Europeans for unusually “deep” diversity in non-coding genetic elements closely linked to loci deter-mining pigmentation. Less fancifully, Parham et al.(1994; and Parham, personal communication) spec-ulatively identified a possible Neandertal legacy: anallele of the human MHC system that is found at lowfrequency in the old Neandertal range. It is remark-able for its inferred ancient separation from otheralleles, which themselves form a tight, young clade.MHC alleles are among the likeliest genes to passthrough a semipermeable hybrid zone, since selec-tion favors immunological diversity per se, so if theinterpretation is confirmed it would set a likely up-per limit on the Neandertal genetic contribution toextant Europeans.

The message from A. Papyoe is, once again, toconcentrate on biology, avoid semantic traps, andrealize that any species-level taxonomy based onfossil material is going to be only an approximatereflection of real-world complexities. With extanttaxa such as Papio baboons, we can document soft-part as well as dental and skeletal anatomy, andsample enough individuals to document inter- andintrapopulational diversity (Fig. 2). Even then, wecannot reach a consensus about the “number of ba-boon species.” At a less global level, our researchgroup has studied gene-flow on the ground in thebest-known of the intergrade zones, and has typedmitochondrial haplotypes and 10 microsatellite lociin nearly 1,000 individuals, as well as examiningtheir external phenotype and features of their den-tition, and observing and quantifying mating pref-erences, social barriers to interbreeding, and thebehavior of hybrids. Yet we still have only a verygeneral notion of the amount of gene-flow betweenthese populations, the breadth of the genetic hybridzone, and the factors that determine its structure.This suggests that it is hardly worth getting tooexercised about similar problems with fossil Homo,or with any other fossil taxon for that matter. Wecan, of course, define convenient paleospecies bycarving the continuum of spatial and temporal vari-ation into chunks that roughly match the diversity


of whatever living taxa one considers to provide thebest analogy. There is no harm in this, provided thatwe recognize that we have not thereby answeredmany biological questions (about degree of pheneticoverlap or separation; about the position, breadth,and permeability of hybrid zones; and about theexistence of population structure defined by matingpreferences and probabilities) that complicate thedefinition and diagnosis of extant species.

DISCUSSION AND CONCLUSIONS

I have argued that paleoanthropology could takegreater advantage of analogies drawn from acrossbiological science, to supplement the insights that ithas always drawn from the anthropocentric sci-ences. I have also tried to illustrate with a few ex-amples that the papionin monkeys (especially thebaboons) are an unusually valuable source of suchanalogies, especially for the earlier, less “human”periods in hominin evolution. Apart from particularanalogies, the major lesson to be drawn from thepapionins (in fact, much of it could be derived fromany widespread, diverse group of actively speciating,terrestrial vertebrates) is that some areas of inquiryare likely to be difficult to resolve for hominins,simply because most hominin lineages have no liv-ing representative. The areas that will challengepaleontologists most severely include the detailedcladistics of closely related extinct forms, the popu-lation structure of the genus Homo for much of itshistory, the dynamics of most interpopulational hy-brid zones and replacement scenarios, and, ofcourse, to the extent that it depends on such infor-mation, the species-level taxonomy of the hominins.

The use of analogy in the ways suggested here isroutine in zoology. The value of phylogeneticallydistant analogues seems to be much less appreciatedby paleoanthropologists working on early hominins,who seem to reach almost reflexively for the nearestcollection of Pan and Gorilla material, often withoutconsidering whether alternative analogues might bemore appropriate. If they justify the ape analogue, itis usually in terms of phylogenetic proximity: a suresign, as I have attempted to show, that the truevalue of analogy has been misunderstood.

The practice of drawing analogies from whereverthey can be found in nature is an application of avery broad scientific principle, that particular casesare to be explained wherever possible in terms ofgeneral “laws,” and that such explanations are to be

Fig. 2. Distribution of ecologically related and presumablyadaptive traits in neighboring baboon allotaxa, chosen to be anal-ogous to cranial and postcranial features believed to be diagnosticof Neandertal and “modern” Homo sapiens. The pattern commonto both is the highly significant difference between mean valuesfor comparable age and sex categories in the two allotaxa, butoverlapping ranges of variation in which bimodality is hardlyobservable, and where individuals would not be readily assignedto the correct taxon. All data are from Awash National Park.A: Relative toe length in male hamadryas and anubis baboons.Mean and 95% confidence intervals for age categories. Withineach taxon cluster, age categories are (left to right): young juve-niles, older juveniles and subadults, and adults. Note that thereare clearly significant age and intertaxon differences, concordantwith the ecology of the two forms: anubis, more arboreal, hamadr-yas, more of a rock-climber. B: Same variables as a bivariate plotof individuals. The mean difference is apparent, but only whenindividuals are identified by taxon. C: Relative molar row lengthin adult anubis and hamadryas baboons, both sexes. The index isan approximation of the relative importance of molars in thedentition. D: Same variables as a bivariate plot; sex-asociatedsize clusters are apparent, but taxon clusters are not separate.


preferred over particularistic, ad hoc explanations.It is not surprising, then, that analogies and com-parative studies that range widely across taxa,searching for broadly applicable generalizations, arecommonplace in vertebrate socioecology and paleon-tology, as in biological science in general. What issurprising is that many paleoanthropologists, espe-cially those whose subjects are human and prehu-man hominins, seem suspicious of them. Perhapsthe reason is historical rather than scientific or ra-tional. Paleoanthropology, whose major roots lie inthe anthropocentric disciplines of anthropology, ar-chaeology, medicine, and human anatomy, ratherthan in general biology (Spencer, 1982), sometimesseems to treat human (and by extension, prehuman)biological evolution as a phenomenon to which otherorganisms have relevance only in proportion to theirphylogenetic proximity to humans. Chimpanzeesand gorillas are not human, but they are the nextbest thing, and other primates come a distant third.

Using more analogies drawn from nonhominidsources would, I believe, open up the anthropologicalimagination, suggesting new and alternative direc-tions to take, and new hypothetical scenarios andexplanations to test. But analogies are not them-selves hypotheses; they are not “refuted” by showingthat the correspondence between analogues is notexact—which, of course, it can never be. In fact,divergences between analogues may be as informa-tive as resemblances, especially when viewedagainst the backdrop of the parallel resemblances:the molar crowns and brain size of geladas are a casein point. One of the most difficult aspects of thestudy of human evolution has always been to recon-cile the fact that early hominins were more likeliving humans than are any living species, with theconcept that they were not themselves human.Moreover, the “mix” of human and nonhuman itselfevolved in the lineage from which H. sapiens is de-rived: by the beginning of the late Pleistocene, hu-mans can be considered mentally “like us,” yet closerto many nonhuman species in terms of populationstructure and diversity. Though no living speciescan provide an exact analogy for these intermediatestages, the baboons can, I think, often provide in-sights into the nonhuman element.

ACKNOWLEDGMENTS

To be given the opportunity to parade one’s hobbyhorses before a captive audience is a rare treat. I ammost grateful to Mark Teaford for inviting me to givethe AAPA lunchtime talk, to the colleagues who satthrough it, to Chris Ruff who invited this versionand endured the editorial process, to three anony-mous reviewers who provided invaluable sugges-tions, and to the many colleagues from whose work,over the years, the ideas contained in this paperwere derived. Particular thanks are due to ToddDisotell, Tim Newman, Jane Phillips-Conroy, RyanRaaum, Jeff Rogers, and Derek Wildman for gener-ously sharing unpublished data.

LITERATURE CITED

Alberts SC, Altmann J. 2001. Immigration and hybridizationpatterns of yellow and anubis baboons in and around Amboseli,Kenya. Am J Primatol 53:139–154.

Albrecht GH. 1977. Craniofacial morphology of the Sulawesi ma-caques. Basel: S. Karger, AG.

Arntzen JW, Wallis GP. 1991. Restricted gene-flow in a movinghybrid zone of the newts Triturus cristatus and T. marmoratusin western France. Evolution 45:805–826.

Asfaw B, White TD, Lovejoy O, Latimer B, Simpson S, Suwa G.1999. Australopithecus garhi: a new species of early hominidfrom Ethiopia. Science 284:629–635.

Avise JC. 1994. Molecular markers, natural history and evolu-tion. London: Chapman and Hall.

Barton NH. 2001. Speciation. Trends Ecol Evol 16:325.Barton NH, Hewitt GM. 1985. Analysis of hybrid zones. Annu

Rev Ecol Syst 16:113–148.Barton NH, Hewitt GM. 1989. Adaptation, speciation and hybrid

zones. Nature 341:497–503.Benefit BR. 2000. Old World monkey origins and diversification:

an evolutionary study of diet and dentition. In: Whitehead PF,Jolly CJ, editors. Old World monkeys. Cambridge: CambridgeUniversity Press. p 133–179.

Bernstein IS. 1966. Naturally occurring primate hybrid. Science154:1559–1560.

Bernstein IS. 1968. Social status of two hybrids in a wild troop ofMacaca irus. Folia Primatol (Basel) 8:121–131.

Bromage TG, Schrenk F, editors. 1999. African biogeography,climate change, and human evolution. New York: Oxford Uni-versity Press.

Brunet M, Beauvilain A, Coppens Y, Heintz E, Moutaye AHE,Pilbeam D. 1995. The first australopithecine 2,500 kilometerswest of the Rift Valley (Chad). Nature 378:273–275.

Bynum EL, Bynum DZ, Froehlich JW, Supriatna J. 1997. Revisedgeographic ranges and hybridization in Macaca tonkeana andMacaca hecki. Trop Biodivers 4:275–283.

Carr SM, Ballinger SW, Derr JN, Blankenship LH, Bickham JW.1986. Mitochondrial DNA analysis of hybridization betweensympatric whitetailed and mule deer in West Texas. Proc NatlAcad Sci USA 83:9576–9580.

Cartmill M. 1994. A critique of homology as a morphologicalconcept. Am J Phys Anthropol 94:115–123.

Churchill SE, Smith FH. 2000. Makers of the Early Aurignacianof Europe. Yrbk Phys Anthropol 43:61–115.

Collard M, Aiello LC. 2000. From forelimbs to two legs. Nature404:339–340.

Collard M, Wood B. 2000. How relaible are human phylogenetichypotheses? Proc Natl Acad Sci USA 97:5003–5006.

Cronin JE, Sarich VM. 1976. Molecular evidence for dual origin ofmangabeys among Old World monkeys. Nature 260:700–702.

Crook JH, Aldrich-Blake P. 1968 Ecological and behavioural con-trast between sympatric ground dwelling primates in Ethiopia.Folia primatol. 8:192–227.

Darwin CD. 1871. The descent of man. London: Charles Murray(3rd ed, 1877).

De Heinzelin J, Clark JD, White T, Hart W, Renne P, WoldeGa-briel G, Beyene Y, Vrba E. 1999. Environment and behavior of2.5-million-year-old Bouri hominids. Science 284:625–629.

Delson E. 1993. The fossils from Africa and India and the taxon-omy of the genus. In: Jablonski NG, editor. Theropithecus: riseand fall of a primate genus. Cambridge: Cambridge UniversityPress. p 157–190.

Disotell TR. 1996. The phylogeny of Old World monkeys. EvolAnthropol 5:18–24.

Disotell TR. 1999. Origins of modern humans still look recent.Curr Biol 9:647–650.

Disotell TR. 2000. Molecular systematics of the Cercopithecidae.In: Whitehead PF, Jolly CJ, editors. Old World monkeys. Cam-bridge: Cambridge University Press. p 29–56.

Duarte C, Mauricio J, Pettitt PB, Souto P, Trinkaus E, van derPlicht H, Zilhao J. 1999. The early Upper Paleolithic humanskeleton from the Abrigo do Lagar Velho (Portugal) and mod-


ern human emergence in Iberia. Proc Natl Acad Sci USA 96:7604–7609.

Dunbar RIM, Dunbar P. 1974. On hybridization between Thero-pithecus gelada and Papio anubis in the wild. J Hum Evol3:187–192.

Dutrillaux B. 1980. Chromosomal evolution of the great apes andman. J Reprod Fertil [Suppl] 2.

Dutrillaux B, Couturier J, Muleris M, Lombard M, Chauvier G.1982. Chromosomal phylogeny of forty-two species or subspe-cies of Cercopithecoids (Primates, Catarrhini). Ann Genet 25:96–109.

Evans BJ, Morales JC, Supriatna J, Melnick DJ. 1999. Origin ofthe Sulawesi macaques (Cercopithecidae: Macaca) as sug-gested by mitochondrial DNA phylogeny. Biol J Linn Soc 66:539–560.

Fleagle JG, McGraw WS. 1999. Skeletal and dental morphologysupports diphyletic origin of baboons and mandrills. Proc NatlAcad Sci USA 96:1157–1161.

Fooden J. 1963. Rhesus and crab-eating macaques: intergrada-tion in Thailand. Science 143:363–365.

Fooden J. 1976. Provisional classification and key to living spe-cies of macaques (Primates: Macaca). Folia Primatol (Basel)25:225–236.

Fooden J. 1980. Classification and distribution of living macaques(Macaca lacepede, 1799). In: Lindburg DG, editor. The ma-caques: studies in ecology, behavior and evolution. New York:Van Nostrand Reinhold Co. p 1–9.

Gee H. 1999. Hybrid camel bridges 11 million years. NatureScience Update, 990506:1.

Goodall J. 1986. The chimpanzees of Gombe: patterns of behavior.Cambridge, MA: Harvard University Press.

Goodman SJ, Barton NH, Swanson G, Abernethy K, PembertonJM. 1999. Introgression through rare hybridization: a geneticstudy of a hybrid zone between red and sika deer (genus Cer-vus) in Argyll, Scotland. Genetics 152:355–371.

Gray AP. 1954 Mammalian hybrids: a check-list with bibliogra-phy. Farnham Royal: Commonwealth Agricultural Bureaux.

Grine FE, editor. 1988. Evolutionary history of the “robust” aus-tralopithecines. New York: Aldine de Gruyter.

Grine FE, Kay RF. 1988. Early hominid diets from quantitativeimage analysis of dental microwear. Nature 333:765–768.

Groves CP. 1978. Phylogenetic and population systematics of themangabeys (Primates: Cercopithecoidea). Primates 19:l–34.

Groves CP. 2001. Primate taxonomy.Grubb P. 1999. Evolutionary processes implicit in distribution

patterns of modern African mammals. In: Bromage TG,Schrenk F, editors. African biogeography, climate change, andhuman evolution. New York: Oxford University Press. p 150–164.

Gundling T, Hill A. 2000. Geological context of fossil Cercopithe-coidea from eastern Africa. In: Whitehead PF, Jolly CJ, editors.Old World monkeys. Cambridge: Cambridge University Press.p 180–213.

Haile-Selassie Y. 2001. Late Miocene hominids from the MiddleAwash, Ethiopia. Nature 412:178–181.

Harpending HC, Batzer MA, Gurven M, Jorde LB, Rogers AR,Sherry ST. 1998. Genetic traces of ancient demography. ProcNatl Acad Sci USA 95:1161–1167.

Harris EE, Disotell TR. 1998. Nuclear gene trees and the phylog-eny of mangabeys (Primates: Papionini). Mol Biol Evol 15:892–900.

Harrison RG. 1993. Hybrid zones and the evolutionary process.Oxford: Oxford University Press.

Haszpruner G. 1998. Parsimony analysis as a specific kind ofhomology estimation and the implications for character weight-ing. Mol Phylogenet Evol 9:333–339.

Hawks J, Wolpoff MH. 2001. Paleoanthropology and the popula-tion genetics of ancient genes. Am J Phys Anthropol 114:269–272.

Hawks J, Hunley K, Lee S-H, Wolpoff M. 2000. Population bot-tlenecks and Pleistocene human evolution. Mol Biol Evol 17:2–22.

Hewitt GM. 1996. Some genetic consequences of ice ages, andtheir role in divergence and speciation. Biol J Linn Soc 58:247–276.

Hewitt GM. 1999. Post-glacial recolonization of European biota.Biol J Linn Soc 68:87–112.

Hewitt GM. 2001. Speciation, hybrid zones and phylogeogra-phy—or seeing genes in space and time. Mol Ecol 10:537–549.

Hewitt-Emmett D, Cook CN, Barnicot NA. 1976. Old World mon-key hemoglobins: deciphering phylogeny from complex pat-terns of molecular evolution. In: Goodman M, Tashian RE,editors. Molecular anthropology. New York: Plenum. p 257–275.

Hey J. 2001. The mind of the species problem. Trends Ecol Evol16:326–329.

Hill WCO. 1970. Primates: comparative anatomy and taxonomy:a monograph. Volume VIII: Cynopithecinae: Papio, Mandrillus,Theropithecus. Edinburgh: University of Edinburgh Press.982 p.

Hoelzer GA, Hoelzer MA, Melnick DJ. 1993. The evolutionaryhistory of the sinica group of macaque monkeys as revealed bymtDNA restriction site analysis. Mol Phylogenet Evol 1:215–222.

Hoss M. 2000. Neanderthal population genetics. Nature 404:453–454.

Hunt KD. 1994. The evolution of human bipedality: ecology andfunctional morphology. J Hum Evol 26:183–202.

Huxley JS. 1932. Problems of relative growth. New York: DialPress.

Jablonski NG, editor. 1993. Theropithecus: rise and fall of aprimate genus. Cambridge: Cambridge University Press.

Jauch A, Wienberg J, Stanyon R, Arnold N, Tofanelli S, Ishida T,Cremer T. 1992 Reconstruction of genomic rearrangements ingreat apes and gibbons by chromosome painting. Proc NatlAcad Sci USA 89:8611–8615.

Jolly CJ. 1966. Introduction to the Cercopithecoidea, with noteson their use as laboratory animals. Symp Zool Soc Lond 17:427–457.

Jolly CJ. 1965. The origins and specialisations of the long-facedCercopithecoidea. Ph.D. thesis, University of London.

Jolly CJ. 1967. The evolution of the baboons. In: Vagtborg H,editor. The baboon in medical research, volume 2. Austin: Uni-versity of Texas Press. p 23–50

Jolly CJ. 1970a. The seed-eaters: a new model of hominid differ-entiation based on a baboon analogy. Man 5:1–26.

Jolly CJ. 1970b. The large African monkeys as an adaptive array.In: Napier J, Napier P, editors. Old World monkeys—evolution,systematics, and behavior. New York: Academic Press. p 227–262.

Jolly CJ. 1972a. The classification and natural history of Thero-pithecus (Simopithecus) (Andrews, 1916), baboons of the Afri-can Plio-Pleistocene. Bull Br Mus Nat Hist (Geol) 22:1–123.

Jolly CJ. 1972b. Changing views of hominid origins Yrbk PhysAnthropol 16:1–17.

Jolly CJ. 1993. Species, subspecies, and baboon systematics. In:Kimbel WH, Martin LB, editors. Species, species concepts, andprimate evolution. New York: Plenum Press. p 67–107.

Jolly CJ, Phillips-Conroy JE. 2001. Testicular size, mating sys-tem, and maturation schedules in wild anubis and hamadryasbaboons. Int J Primatol (in press).

Jolly CJ, Plog F. 1987. Physial anthropology and archeology, 3rded. New York: Random House.

Jolly CJ, Woolley-Barker T, Beyene S, Disotell TR Phillips-Con-roy JE. 1997. Intergeneric hybrid baboons. Int J Primatol 18:597–627.

Kalb JE, Jolly CJ, Tebedge S, Mebrate A, Smart C, Oswald EB,Whitehead PF, Wood CB, Adefris T, Rawn-Schatzinger V. 1982.Vertebrate faunas from the Awash Group, Middle Awash Val-ley, Afar, Ethiopia. J Vertebr Paleontol 2:237–258.

Kaplan JR, Phillips-Conroy J, Fontenot MB, Jolly CJ, FairbanksLA, Mann JJ. 1999. Cerebrospinal fluid monoaminergic metab-olites differ in wild anubis and hybrid (anubis/hamadryas) ba-boons: possible relationships to life history and behavior. Neu-ropsychopharmacology 20:517–524.


Kingdon J. 1997. The Kingdon field guide to African mammals.San Diego: Academic Press.

Kluge A. 1989. A concern for evidence and a phylogenetic hypoth-esis of relationships among Epicrates (Boidae, Serpentes). SystZool 38:17–25.

Krings M, Stone A., Schmitz RW, Krainitzki H, Stoneking M,Paabo S. 1997. Neandertal DNA sequences and the origin ofmodern humans. Cell 90:19–30.

Kummer H. 1968. Social organization of hamadryas baboons.Chicago: Aldine.

Kurten B. 1978. Dance of the tiger: a novel of the Ice Age.Leakey LSB, Whitworth T. 1958. Notes on the genus Simopithe-

cus with description of a new species from Olduvai Occ. PapCoryndon Mem Mus 6:1–14.

Leakey MG, Feibel CS, McDougall I, Walker AC. 1995. Newfour-million-year-old hominid species from Kanapoi and AlliaBay, Kenya. Nature 376:565–571.

Leakey MG, Spoor F, Brown FH, Gathogo PN, Kiarie C, LeakeyLN, McDougall I. 2001. New hominin genus from eastern Af-rica shows diverse middle Pliocene lineages. Nature 410:433–440.

Lee-Thorp JA, van der Merwe NJ. 1993. Stable carbon isotopestudies of Swartkrans fossils. In: Brain CK, editor. Swartkrans.A cave’s chronicle of early man. Pretoria: Transvaal Museum. p251–256.

Le Gros Clark WE. 1955. The fossil evidence for human evolution.Chicago.

Lehman N, Eisenhawer A, Hansen K, Mech LD, Peterson RO,Gogan PJP, Wayne RK. 1991. Introgression of coyote mitochon-drial DNA into sympatric North American gray wolf popula-tions. Evolution 45:104–119.

Lockwood CA, Fleagle JG. 1999. The recognition and evaluationof homoplasy in primate and human evolution Yrbk Phys An-thropol 42:189–232.

Maples WR, McKern TW. 1967. A preliminary report on classifi-cation of the Kenya baboon. In: Vagtborg H, editor. The baboonin medical research, volume 2. San Antonio: University ofTexas Press. p 1322.

Markarjan DS, Isakov EP, Kondakov GI. 1974. Intergeneric hy-brids of the lower (42chromosomal) monkey species of theSukhumi Monkey Colony. J Hum Evol 3:247–255.

Martin RD. 1993. Allometric aspects of skull morphology inTheropithecus. In: Jablonski NG, editor. Theropithecus: riseand fall of a primate genus. Cambridge: Cambridge UniversityPress. p 273–298.

McCollum MA. 1999. The robust australopithecine face: a mor-phogenetic perspective. Science 284:301–305.

Melnick DJ, Hoelzer GA. 1992. Differences in male and femalemacaque dispersal lead to contrasting distributions of nuclearand mitochondrial DNA variation. Int J Primatol 13:379–393.

Moore CM, Jaquish C, Eddy CA, Hubbard GB, Leland MM,Rogers J. 1999. Cytogenetic and fertility studies of a rheboon,rhesus macaque (Macaca mulatta) � baboon (Papio hamadr-yas) cross: further support for a single karyotype nomenclature.Am J Phys Anthropol 110:119–127.

Napier JR. 1980. Hands. London: Allen & Unwin.Newman TK, Jolly CJ, Rogers J. 2001. Mitochondrial DNA se-

quence variation among baboons (Papio hamadryas): phyloge-netic and phylogeographic analysis (submitted).

Nickerson E, Nelson DL. 1998. Molecular definition of pericentricinversion breakpoints occurring during the evolution of hu-mans and chimpanzees. Genomics 50:368–372.

Nystrom PD. 1992. Mating success of hamadryas, anubis andhybrid male baboons in a “mixed” social group in the AwashNational Park, Ethiopia. Ph.D. dissertation, Washington Uni-versity.

O’Neill RJW, Eldridge MDB, Graves JAM. 2001. Chromosomeheterozygosity and de novo chromosome rearrangements inmammalain interspecies hybrids. Mammalian Genome12:256–259.

Ovchinnikov IV, Gotherstrom A, Romanova GP, Kharitonov VM,Linden K, Goodwin W. 2000. Molecular analysis of Neander-thal DNA from the northern Caucasus. Nature 404:490–493.

Parham P, Arnett KL, Adams EJ, Barber LD, Domena JD, Stew-art D, Hildebrand WH, Little AM. 1994. The HLA-B73 antigenhas a most unusual structure that defines a second lineage ofHLA-B alleles. Tissue Antigens 43:302–313.

Peters CH. 1982. Nut-like oil seeds: food for monkeys, apes, andprobably ape-men. Am J Phys Anthropol 57:283–301.

Phillips-Conroy JE. 1978. Dental variability in Ethiopian ba-boons: an examination of the anubis hamadryas hybrid zone inthe Awash National Park, Ethiopia. Ph.D. thesis, New YorkUniversity.

Phillips-Conroy JE, Jolly CJ, Nystrom P, Hemmalin HA. 1992.Migration of male hamadryas baboons into anubis groups inthe Awash National Park, Ethiopia. Int J Primatol 13:455–476.

Rieseberg LH. 2001. Chromosomal rearrangements and specia-tion Trends Ecol Evol 16:351–358.

Rosas A. 2001. Occurrence of Neanderthal features in mandiblesfrom the Atapuerca-SH site. Am J Phys Anthropol 114:74–91.

Ruvolo M, Disotell TR, Allard MW, Brown WM, Honeycutt RL.1991. Resolution of the African hominoid trichotomy by use ofa mitochondrial gene sequence. Proc Natl Acad Sci USA 88:1570–1574.

Saino N, Lorenzini R, Fusco G, Randi E. 1992. Genetic variabilityin a hybrid zone between carrion and hooded crows (Corvuscorone corone and C. c. cornix Passeriformes, Aves) in north-western Italy. Biochem Syst Ecol 20:605–613.

Samuels A, Altmann J. 1986. Immigration of a Papio anubis maleinto a group of cynocephalus baboons and evidence for ananubis-cynocephalus hybrid zone in Amboseli, Kenya. Int JPrimatol 7:131–138.

Schaller GB, Lowther GR. 1969. The relevance of social carnivorebehavior to the study of early hominids. Southwest J Anthropol25:307–341.

Schluter D. 2001. Ecology and the origin of species. Trends EcolEvol 16:372–380.

Schluter D, Nagel LM. 1995. Parallel speciation by natural selec-tion. Am Nat 146:292–301.

Searle JB. 1998. Speciation, chromosomes and genomes GenomeRes 8:1–3.

Skelton RR, McHenry HM. 1992. Evolutionary relationshipsamong early hominids. J Hum Evol 23:309–349.

Spencer F, editor. 1982. A history of American physical anthro-pology, 1930–1980. New York: Academic Press.

Spenser LM. 1997. Dietary adaptations of Plio-Pleistocene Bovi-dae: implications for hominid habitat use. J Hum Evol 32:201–228.

Strait DS. 2001. Integration, phylogeny, and the hominid cranialbase. Am J Phys Anthropol 114:273–297.

Strait DS, Grine FE, Moniz MA. 1997. A reappraisal of earlyhominid phylogeny. J Hum Evol 32:17–82.

Strasser E, Delson E. 1987. Cladistic analysis of cercopithecidrelationships. J Hum Evol 16:8l–99.

Struhsaker TT, Butynski TM, Lwanga JS. 1988. Hybridizationbetween redtail (Cercopithecus ascanius schmidti) and blue (C.mitis stuhlmanni) monkeys in the Kibale Forest, Uganda. In:Gautier Hion A, Bourliere F, Gautier JP, editors. A primateradiation: evolutionary biology of the African guenons. Cam-bridge: Cambridge University Press.

Sugawara K. 1979. Sociological study of a wild group of hybridbaboons between Papio anubis and Papio hamadryas in theAwash Valley, Ethiopia. Primates 20:21–56.

Swisher CC III, Rink WJ, Anton SC, Schwarcz HP, Curtis GH,Suprijo H, Widiasmoro. 1996. Latest Homo erectus of Java:potential contemporaneity with Homo sapiens in SoutheastAsia. Science 274:1870–1874.

Tattersall I. 1993. Speciation and morphological differentiation inthe genus Lemur. In: Kimbel W, Martin L, editors. Species,species concepts, and primate evolution. New York: Wiley.

Tattersall I, Eldredge N. 1977. Fact, theory and fantasy in humanevolution Am Sci 65:204–211.

Taylor EB, McPhail JD. 1999. Evolutionary history of an adaptiveradiation in species pairs of threespine sticklebacks (Gasteros-teus): insights from mitochondrial DNA. Biol J Linn Soc 66:271–291.


Tchernov E. 1989. The Middle Paleolithic mammalian sequenceand its bearing on the origin of Homo sapiens in the SouthernLevant. In: BarYosef O, Vandermeersch B, editors. Investiga-tions in South Levantine prehistory. Prehistoire du Sud-Le-vant. BAR Int Ser 497:25–42.

Teaford MF, Ungar PS. 2000. Diet and the evolution of the ear-liest human ancestors. Proc Natl Acad Sci USA 97:13506–13511.

Templeton AR. 1989. The meaning of species and speciation: agenetic perspective. In: Otte D, Endler J, editors. Speciationand its consequences. Sunderland, MA: Sinauer Associates.

Templeton AR. 1998. Nested clade analyses of phylogeographicdata and testing hypotheses about gene flow and populationhistory. Mol Ecol 7:381–397.

Thorington RW, Groves CP. 1970. An annotated classification ofthe Cercopithecoidea. In: Napier JR, Napier P, editors. The OldWorld monkeys. New York: Academic Press. p 629–647.

Tosi AJ, Morales JC, Melnick DJ. 2000. Comparison of Y-chromo-some and mtDNA phylogenies leads to unque inferences of ma-caque evolutionary history. Mol Phylogenet Evol 17:133–144.

Trinkaus E. 1991. Les hommes fossiles de la Grotte de Shanidar,Irak: evolution et continuite parmi les hommes archaiquestardifs du Proche-Orient. Anthropologie 95:535–572.

Trinkaus E. 1993. Variability in the position of the mandibularmental foramen and the identification of Neandertal apomor-phies. Riv Antropol 71:259–274.

Turelli M, Barton NH, Coyne JA. 2001. Theory and speciation.Trends Ecol Evol 16:330–343.

Van Tuinen P, Mootnick AR, Kingswood SC, Hale DW, KuamotoAT. 1999. Complex, compound inversion/translocation poly-morphism in an ape: presumptive intermediate stage in thekaryotypic evolution of the agile gibbon Hylobates agilis. Am JPhys Anthropol 110:129–142.

Walker AC. 1981. Dietary hypotheses and human evolution. Phi-los Trans R Soc Lond [Biol] 292:57–64.

Weiss KM, Buchanan A. 2000. Rediscovering Darwin after aDarwinian century. Evol Anthropol.

Weiss KM. 1994. A toe, a tooth, and a vertebra: the geneticdimensions of complex morphological traits Evol Anthropol2:121–137.

White TD, Suwa G, Asfaw B. 1994. Australopithecus ramidus, anew species of early hominid from Aramis, Ethiopia. Nature371:306–312.

Wildman D. 2000. Ph.D. dissertation, New York University.Williams-Blangero S, Vanderberg J, Blangero J, Konigsberg L,

Dyke B. 1990. Genetic differentiation between baboon subspe-cies: relevance for biomedical research. Am J Primatol 20:67–81.

Wolpoff MH. 1973. Posterior tooth size, body size, and diet inSouth African gracile australopithecines. Am J Phys An-thropol. 39:375–394.

Wolpoff MH. 1999. Paleoanthropology, 2nd ed. Boston: McGraw-Hill.

Wolpoff MH, Thorne AG, Jelinek J, Zhang Y. 1993. The case forsinking Homo erectus: 100 years of Pithecanthropus is enough!Courier Forschungsinstitut Senckberg 171:772–773.

Wolpoff MH, Hawks J, Caspari R. 2000. Multiregional, not mul-tiple origins. Am J Phys Anthropol 112:129–136.

Wood BA, Brooks A. 1999. We are what we ate. Nature 400:219–220.

Wood BA, Collard M. 1999. The genus Homo. Science 284:65–71.Wrangham RW, Peterson D. 1996. Demonic males: apes and the

origins of human violence. New York: Mariner Books.Zinner D, Hapke A. 2000. Ecological comparisons of primates in

Eritrea. Poster, at www.biozentrum.uni-wuerzburg.de.


jolly - papionin monkeys & human evolution - 01

Documents