NEWS

Paleoprimatology at the Pyramid of the SunCHRISTOPHER P. HEESY AND ROBERT V. HILL

Evolutionary anthropologists seek tounderstand aspects of the biology, be-havior, ecology, and evolution of pri-mates and humans. One valuablesource of data for addressing such is-sues is the fossil record, which for pri-mates extends at least as far back as theearliest Eocene. At the 60th annualmeeting of the Society of Vertebrate Pa-leontology in Mexico City last October,more than 500 paleontologists con-vened to discuss the latest discoveries inpaleontology, touching on such diversetaxa as dinosaurs, horses, rodents, croc-odilians, and, of course, primates.Herein we discuss new fossils relevantto the study of primate evolution, aswell as new advances in the phyloge-netic analysis of fossil data.

THE FOSSILS

The late middle Eocene PondaungFormation of Myanmar (formerlyBurma) has yielded several fossils,such as Pondaungia, Bahinia, and Am-phipithecus, which may offer clues tothe origin and early evolution of an-thropoids. Gregg Gunnell (Universityof Michigan Museum of Paleontology)and colleagues presented new post-cranial fragments of a primate, in-cluding a complete left humerus, frag-ments of the right humerus, andfragments of both ulnae. Althoughthese postcrania were not associatedwith dental remains, these research-ers tentatively assigned the specimensto Amphipithecus, based mainly ontheir size. Numerous aspects of themorphology suggest strong strepsir-rhine affinities. Focusing on the mor-phology of the left humerus, Gunnellinterprets the locomotor adaptationsas being consistent with a slowclimber, not dissimilar to Loris. Ifthese postcranial elements do repre-sent Amphipithecus, then they further

strengthen the argument that Am-phipithecus is an adapiform.1

Gary Schwartz (George WashingtonUniversity) and coworkers reportedon the dental development rates of theMalagasy subfossil Palaeopropithecusingens. Extant indriids are known tohave a highly accelerated dental erup-tion rate relative to that of all otherprimates, which is likely to correlatewith rapid life-history strategies. Theanalysis of daily enamel accretion rateas collected from histological sectionsof Palaeopropithecus molars indicatesthat this taxon is similar to extant in-driids in having accelerated dentalprecocity. However, the rate ofenamel accretion is far faster thanthat in any extant primate, implyingthat although these animals had pre-cocial dental (and likely overall) devel-opment, the life-history strategies ofsuch a large-bodied animal differgreatly from anything seen among liv-ing primates.

Jonathan Bloch and colleagues(University of Michigan Museum ofPaleontology) showed tantalizingviews, including crania and much ofthe postcranium, of a new paromo-myid plesiadapiform that was ex-tracted from a limestone block in theClarks Fork Basin of Wyoming. Paro-momyids have been important to Ar-chontan relationships (the Grand-order that includes primates, treeshrews, bats, dermopterans, andplesiadapiformes). During the last de-cade and a half, it has been suggestedthat some paromomyids, based on thepossible gliding morphology of thepostcranium, are sister taxa to mod-ern dermopterans or “flying lemurs.”Bloch and colleagues, based on theiranalysis, dispute this claim, suggest-ing that this very early and primitiveparomomyid shares no recognizablesynapomorphies with dermopterans.However, a more extensive analysis of

the taxonomic status of this fossil andthe implications for Archontan sys-tematics must await next year’s pre-sentation by this group.

Since the molecular revolution ofcladistics, systematists have soughtways to integrate fossil and molecularevidence in phylogenetic analyses.The total-evidence approach was dis-cussed in two different papers as themost amenable method of incorporat-ing these different lines of evidence.Marcia Delanty and Callum Ross(Stony Brook) presented the results oftheir analysis of tarsiid relationships.Tarsius has long been thought to rep-resent the sister taxon to anthropoidsbased on morphological data (derivedrhinarium, postorbital septum, retinalfovea, anterior accessory cavity of themiddle ear, and hemochorial placen-tation). On the other hand, moleculardatasets and analyses often fail toresolve the phylogenetic position ofTarsius. In addition, the relationshipsof Tarsius and anthropoids to the fos-sil superfamilies Omomyoidea andAdapoidea are a continuing source ofdiscussion and controversy. Delantyand Ross addressed these issues bycombining mitochondrial and rRNAdata with a large morphological data-set coding not only extant taxa buteach relevant fossil group, and con-ducted a total-evidence analysis. Al-though their results differed based onassumption sets and data partitions,the most surprising and controversialresult supported a monophyletic Pro-simii (Tarsius 1 Strepsirrhini), as wellas other traditional groupings such asAnthropoidea 1 Omomyoidea. Amonophyletic Prosimii is very likelydriven by the molecular component ofthe dataset and will undoubtedly leadto more work on this topic.

In contrast, the total-evidence anal-ysis of callitrichid relationships pre-sented by Felicia Brenoe (StonyEvolutionary Anthropology 10:79–80 (2001)

Evolutionary Anthropology 79

Brook) was, in some respects, moresuccessful. Brenoe applied total-evi-dence methodology to the analysis ofextant callitrichids as well as the fos-sils Lagonimico and Patasola, using amorphological dataset compiled fromthe literature and both mitochondrialgenes (12S, 16S, cytochrome b) andnuclear genes (epsilon globin andbeta2 microglobin). The combinedmorphological and molecular resultswere strongly supported with a single,completely resolved tree. Surprising, ahighly nested Callimico 1 Patasola 1Lagonimico clade was found. This im-plies that the taxa may have re-evolved third molars and hypocones,the loss which are putative synapo-morphies of the callitrichid clade.

MISSING DATA SYMPOSIUM

The potential for large amounts ofmissing data has long been a con-founding factor in evolutionary stud-ies that include fossils. This issue wasthe subject of a symposium entitled“Missing Data—Practical Problemsand Theoretical Issues,” in which sev-eral speakers addressed the problemsposed by missing data and some pos-sible solutions.

An abundance of missing data incertain taxa or characters can oftenadd to the number of most parsimo-nious trees produced in a phyloge-netic analysis, increasing instances ofambiguous character optimizationand leading to reduced resolution inconsensus trees. For this reason, frag-mentary fossil taxa or characters thatcannot be scored in such taxa aresometimes entirely excluded fromdata matrices either before or afteranalysis. Are these practices justified?

Maureen Kearney (George Wash-ington University, now at Field Mu-seum of Natural History) argued thatmethods that exclude data a priori areill-conceived because the informationcontent of incomplete taxa cannot beeasily judged before analysis. Meth-ods that involve identification and re-moval of problematic taxa a posteriorido not fare much better, as they areliable to obscure the true reason forambiguous optimizations, which maybe due to missing data or characterconflict. An alternative is to include allavailable data and examine consensustrees to determine the source of am-

biguity. Using a measure called theClade Concordance Index, the amountof ambiguity in consensus trees de-rived from character conflict can bedetermined and ambiguity that arisesas an artifact of missing data can besuppressed. This method allows theamount of information gleaned fromfragmentary taxa to be maximizedwhile simultaneously minimizing theundesirable effects of missing data.

Kearney’s method employs a total ev-idence approach, followed by carefulexamination of resulting consensustrees. The total evidence approach,while not universally accepted, is oper-ationally sound and, under ideal condi-tions (complete data for all taxa) wouldencounter few obstacles. However, 99%of all life is extinct, and thus exists onlyas fossilized remains, which are alwaysmissing some data. Maureen O’Leary(Stony Brook), an advocate of total ev-idence analyses, addressed some of theoperational difficulties encounteredwhen using this approach.

Hypotheses of relationships that arebased on non-fossilizable data cannotbe tested for the majority of taxa, andare therefore subject to a weakenedparsimony criterion. Non-fossilizabledata, however, often comprise 80% ormore of data matrices. Thus, the totalevidence approach is operationallyconfounded by the fact that most taxaare extinct. To avoid this pitfall, someauthors advocate partitioned analy-ses, segregating morphology and mol-ecules or cranial and postcranial char-acters, and then comparing resultingphylogenies with one another. Theproblem with this approach, calledtaxonomic congruence, is that thepartitions themselves are, in a bio-logical sense, completely arbitrary.O’Leary suggests, however, that twobiological data partitions actually doexist: fossilizable and non-fossilizabledata. A thorough phylogenetic analy-sis, therefore, should consider notonly total evidence, but also taxo-nomic congruence between these twopartitions.

In addition to the discussions on theproper handling of missing data,some authors gave valuable insightinto the effects of missing data on thesupport measures used to evaluateconfidence in phylogenetic trees. Pe-ter Makovicky (American Museum ofNatural History) examined two com-

mon methods of measuring support:bootstrap and Bremer support. By in-crementally adding missing values toexisting data matrices, it was shownthat Bremer support predictably de-creases. Bootstrap values showed asimilar trend, but exhibited fargreater variability with regard to theoverall size of the data set. Althoughboth methods may show tendencies tooverestimate support, Bremer supportis less likely to do so, and thereforemore appropriate for dealing with pa-leontological data sets and the uniqueproblems posed by missing data. Theuse of current methods for assessingconfidence may, however, lead to theacceptance of false phylogenetic hy-potheses.

Mark Wilkinson (Natural HistoryMuseum, London) echoed this con-cern, calling for better support mea-sures that are not excessively sensitiveto unstable taxa. Instability that leadsto poorly resolved consensus treesmay often be caused by a concentra-tion of missing data in a small subsetof taxa, rather than missing entriesscattered throughout a data matrix.The detection of such unstable taxacan be achieved using double decayanalysis. This method improves ontraditional decay analyses by reveal-ing strong phylogenetic signals thatare otherwise obscured by largeamounts of missing data.

REFERENCES

1 Szalay FS. 1970. Late Eocene Amphipithecusand the origins of catarrhine primates. Nature227:355–357.2 Wilkinson M, Thorley JL, Upchurch P. 2000. Achain is no stronger than its weakest link: doubledecay analysis of phylogenetic hypotheses. SystBiol 49:754–776.

Christopher P. HeesyInterdepartmental Doctoral Program in

Anthropological SciencesDepartment of Anatomical Sciences

Health Sciences Center, T-8Stony Brook University

Stony Brook, NY 11794-8081Email: cheesy@ic.sunysb.edu

Robert V. HillDepartment of Anatomical Sciences

Health Sciences Center, T-8Stony Brook University

Stony Brook, NY 11794-8081Email: rvhill@ic.sunysb.edu

© 2001 Wiley-Liss, Inc.

80 Evolutionary Anthropology NEWS