SELECTION BY DIFFERENTIAL PERSISTENCE AMONG ANIMAL GENERA QUANTIFIED USING THE PALEOBIOLOGY DATABASE

Selection by differential persistence has been proposed as the evolutionary basis for the Gaia hypothesis, potentially explaining how a biosphere might be selected for the property of promoting its own survival (Doolittle, 2013). However, this process cannot practically be observed on such a scale. To test whether selection by differential persistence might be seen operating on smaller clades than life as a whole, we analysed the lifespans of extinct animal genera using fossil occurrence data from the Paleobiology Database (PBDB). If this process is in operation, we would predict that the extinction probability of a clade decreases as the clade gets older and accumulates adaptations favourable for its survival. This is in opposition to a constant extinction probability per unit time – a common assumption for each lineage in evolutionary modelling – which would predict that the histogram of extinct clade lifespans follows an exponential distribution.

We observe that old clades are over-represented in the histogram compared to the exponential distribution, and the degree of over-representation (ratio of observed number of taxa in each bin vs. that predicted by the exponential curve) increases with clade lifespan. This pattern is robust within major phyla and classes tested, as well as through time. Age selectivity of extinction has been observed before for particular clades or extinction events (Finnegan et al., 2008, for example), but it has not been considered as a general process potentially deriving from selection by differential persistence. Observing this process at lower taxonomic levels also hints at its possibility for life as a whole – indeed, the possibility of Gaia.