INFERRING ANCESTOR-DESCENDANT RELATIONSHIPS IN THE FOSSIL RECORD (WITH STATISTICS)

The idea that some of the taxa we recognize in the fossil record may actually be ancestral to other fossil or living lineages is wide-spread in popular thought. Historically, many paleontologists qualitatively implied ancestor-descendant relationships among sampled taxa. However, typical approaches for inferring phylogenies don’t allow taxon units to be ancestors, leading to the development of ‘parsimonious’ rules for diagnosing ancestry (e.g. lack of autapomorphies). One method, stratocladistics, inferred ancestors but that feature was overshadowed by larger debates, and did not lend itself to quantifying statistical support for particular ancestor-descendant relationships. Recent development of phylogenetic dating methods, such as cal3 and Bayesian tip-dating, allow for taxa to be inferred as sampled-ancestors a in a model-based framework. We present two empirical case examples of inferring ancestor-descendant relationships. First, with Paleozoic pterocephaliid trilobites, we contrast previously qualitative assessments of ancestor-descendant species-pairs with those inferred by cal3, and find high agreement except for anagenetic pairs, which were overwhelmingly inferred by cal3 as budding cladogenesis instead. Overall, this analysis suggests that budding is the dominant mode of morphological speciation. Second, we utilize a dataset of Mesozoic theropods to compare ancestral inference from tip-dating (via BEAST2 and MrBayes) and cal3. These analyses, despite similar model assumptions, infer very different frequencies of sampled ancestry, with some incongruence about which species are considered to be sampled-ancestors. Interestingly, although Archaeopteryx is popularly thought of as an ‘ancestral bird’, it is rarely placed as a ancestral taxon with this dataset, and then only to its close relative Wellnhoferia, not to the lineage leading to crown birds. Overall, while there is some dependency on the approaches used, there is a bright future for quantitative inference of ancestors, particularly as we expand methods to account for persistent chronospecies, and to account for different patterns in how lineages morphologically differentiate from each other.