The 81st Annual Meeting of the American Association of Physical Anthropologists (2012)

An assessment of Bayesian methods for ancestral state estimation


1Interdepartmental Program in Anthropological Sciences, Stony Brook University, 2Department of Anthropology, Brooklyn College (CUNY), 3Department of Anthropology, Hunter College (CUNY), 4Program in Anthropology, The Graduate Center (CUNY), 5Program in Biology, The Graduate Center (CUNY), 6New York Consortium in Evolutionary Primatology, NYCEP

Saturday All day, Plaza Level Add to calendar

Recent studies have attempted to investigate evolutionary trends by utilizing ancestral state reconstruction (ASR) to reconstruct ancestral morphology from phylogenetic variation of extant descendents. However, simulation studies question the precision of ASR because the models often specify a component of theoretical error resulting in 95% confidence intervals that exceed the range of values in the extant distribution. We evaluate Bayesian methods for ASR using the relief index of the second lower molar of 57 euarchontans including 26 strepsirrhines, 18 platyrrhines, two tarsiers, two dermopterans, two scandentians, five extinct euprimates, and two plesiadapiforms. We compared ASR for relief index in the ancestral euprimate using a phylogeny based on 1) extant taxa only, and 2) the entire sample. Analysis 1 reconstructed the euprimate node as having an index of 0.5348, a similar value to basal fossil euprimates, Teilhardina asiatica (0.5467) and Donrussellia (0.4994 - 0.5461). Although the 95% highest probability density (HPD) interval of the estimate is large (0.4683 - 0.5969) and prohibits unambiguous dietary reconstruction, it is still smaller than the observed range of the extant sample (0.3453 - 0.6383). Analysis 2 reconstructed the euprimate node as 0.5772, which is closest to the extant Tarsius bancanus (0.5755); the 95% HPD interval ranges from 0.5454 to 0.6088. This analysis narrows the HPD interval enough to unambiguously reconstruct the ancestral euprimate as an insectivore, indicating the importance of including extinct taxa in ASR studies when possible. Future work will address how changing phylogenetic topologies and branch lengths of fossil species affect nodal reconstruction.

NSF-DDIG-BCS-0622544 (DMB), Evolving Earth Foundatoin Grant (DMB), Professional Development Grant of the American Association of Physical Anthropologists (DMB), Leakey Foundation Grant (DMB and EMS)

comments powered by Disqus