Lets say I have a phylogenetic tree and two traits A and B mapped to the tips of the tree, and I want to test the hypothesis that evolution of trait B is contingent on trait A having certain states. For example, if A and B have two possible states, 0 and 1, then, how would be the best way to test that the transition B:0->1 is contingent (or more likely to occur) on A being on state 1?
Stochastic character mapping, as implemented in programs like SIMMAP (http://www.simmap.com), allows you to test for the correlation (specifically the co-occurence along a phylogeny) of two discrete characters. Technically this is not indicating contingence in an absolute sense (i.e., no causality is inferred), but if the correlation is strong enough, it would likely be consistent with such a hypothesis.
Check out WP Maddison 2000 (Journal of Theoretical Biology) for a discussion of character correlation using pairwise comparisons on a phylogeny.
- Badges
- Science topic
- Similar topics
- Data Mining
- Classification