Presentation Detail

Macroevolution

Simpson, Andrew G. [1], Fenster, Charles [1].

Response of BiSSE (Binary State Speciation and Extinction) to known paleontological violation of its assumptions.

Binary state speciation and extinction (BiSSE; Maddison et al), along with its multi-state and quantitative-state cousins, provides a new alternative for studying past rates of speciation, extinction, and character evolution in groups with limited to no fossil record. Moreover, BiSSE is unusual in being able to distinguish past selection at the individual level from selection at the clade level. Of these modes, the former results in high rates of state transition toward favored character states, and the latter in increased net diversification for lineages possessing favored character states. However, BiSSE assumes that intrinsic rates of speciation and extinction and relative fitness of different character states are time-invariant, assumptions that are known to be false. In particular, fossil data tells us that nearly all clades exhibit a decline of net diversification rates with time, eventually leading to the ultimate extinction of a clade. This may not be a serious shortcoming, as the algorithm may or may not be robust to its assumptions being violated. What insights can paleontologists share with neontologists whose study groups lack good fossil records? To address this, I simulate evolution of hypothetical clades featuring state-dependent diversification rates using more paleontologically realistic assumptions. In addition, I also compare speciation and extinction rates inferred from the fossil record of the conifers (Plantae: Coniferophyta) against those inferred by BiSSE using a conifer phylogeny developed from molecular data. Conifers are a good study group for this purpose, being a living group with a resolved phylogeny, an excellent fossil record, and a living diversity featuring both modern, diversifying clades (e.g. Pinaceae) as well as ancient, declining lineages (e.g. Araucariaceae). Results with simulated data suggest that BiSSE performs well for clades whose net diversification rates have been positive through their entire histories, but becomes unreliable and prone to detect non-existent selective effects in groups that have been declining in diversity for part of their history. The bias becomes more severe the larger the difference in net diversification rate at the present-daytime slice in which the clade is analyzed relative to the time of the clade's origination, being most noticeable in clades that are currently declining. Fortunately, most diverse lineages in the present day are diverse because they are actively diversifying, meaning that most studies in the literature that use BiSSE are unlikely to run afoul of the bias introduced by its time-invariant assumptions.

1 - University of Maryland, Behavior, Ecology, Evolution, and Systematics, College Park, MD, 29742, USA

Keywords:
Speciation
extinction
BiSSE
Mathematical modelling
Assumption testing
fossil calibration
Conifers.

Presentation Type: Regular Oral Presentation
Session: 15
Location: Cotton B/Snowbird Center
Date: Saturday, June 22nd, 2013
Time: 11:45 AM
Number: 15006
Abstract ID:459
Candidate for Awards:W.D. Hamilton Award for Outstanding Student Presentation