Liveblogging the conference: Julia Clarke and Todd Grantham

This is a session on paleontology that I missed the start of because I had to go get my power supply.

Julia, a paleontologist, is discussing the evolution of birds, and how paleontology was misled by hypotheses that used the wrong taxa and characters. I'd love to blog it more extensively, but I missed the start.

She is noting the telic nature of some hypotheses, and how she and her colleagues worked in a different way, phylogenetically. Adopting a method used in molecular phylogeny, they supposed that character states might be decomposeable into smaller subregions, anatomically. They allowed variable evolution in subregions of pectoral and pelvic locomotor systems. Using a Bayesian analysis they looked for sets of models of evolution and came up with two different sets - one was multimodal and is significantly preferred.

Are these multimodal systems preferred merely because they are parameter rich? A test I don't understand showed that the unimodal models in which all characters varied as a single partition were preferred. A mosaic evolution of partitions that are anatomical is preferred [don't know why].

They tested early forelimb evolution and late hindlimb modification models. It appeared to be unsupported - hindlimb and forelimb evolution are independent. This approach might be used to test independent evolution of traits.

Todd Grantham is a philosopher of biology, a student of David Hull's. He's talking about taxon duration in the fossil record. The duration from arisal to extinction, it's fuzzy, and the stratigraphic record underestimates the actual duration. How to estimate the true duration?

Contexts in which it matters: fossil record completeness, phylogenetic reconstruction, and Sepkowski-stye rates of extinction. Controversial topic.

Two sticky issues: range extension and conflicts between fossil record and molecular estimates of age.

Range extension: "ghost lineage" problems - missing record of lineage. Cites Systematics and the Fossil Record: Documenting Evolutionary Patterns by A. B. Smith. Set up the phylogeny and then work out the ghost lineages. Is Smith's method the only one? He allows for ancestors, and anagenetic lineages. If the phylogeny is adapted to the stratigraphy, it can be shorter than if you hold the phylogeny fixed and fit the stratigraphy.

Not prescribing a solution - merely identifying the difficulties and differences of estimates [this seems to me to be about reconstructing histories from phylogenies]. An epistemological question, not metaphysics.

Some estimates are disparate based on rocks and molecular clocks. Is this a theoretical issue? E.g., do crown group mammal lineages extend far back into the Cretaceous? Three models: explosive after K-T, short fuse (deep roots early on, basically accepting the molecular data) and long fuse, accepting early evolution but late diversification. Fossils from 65mya, molecular clock up to 150mya. Relies on assumption of constant preservation potential, obviously wrong. But we're getting better.

Molecular clock stuff using variable rate of changes allows better estimates. Conflicts may be apparent only - fossils show latest derived morphology while molecular clocks start ticking as soon as the lineages diverge.