COMMUNITIES FALL APART: MEASURING EXTINCTION POTENTIAL IN PALEOCOMMUNITIES

The extinction of multiple species within a community is the result of ecological crisis. Declines or abrupt changes in productivity are often cited as proximal causes of such extinctions in the fossil record. The impact of an interruption of productivity is transmitted through a community via its newtork of trophic interactions. We argue that the extent to which such crises emanate through community networks, resulting in secondary and cascading extinctions, is a function of the magnitude of the productivity crisis, taxonomic diversity, and ecological diversity. An understanding of the interactions of these factors may lead to deeper comprehension of community responses to ecological crisis and extinction during the Phanerozoic.

The problem is approached using a probabilistic model of community trophic structure. Species within a community are divided into guilds, where members of a guild share the same set(s) of potential prey or food sources. Trophic links between members of different guilds are assigned randomly, based on theoretical or empirical link distributions. Network complexity is varied according to the total number of taxa, the number of taxa assigned to specific guilds, the number of guilds, and guild trophic ecologies. Secondary extinction and extinction cascades are triggered in the network by mimicking productivity crises, removing primary producers or primary consumers. Complexity of the model's dynamics is increased by: 1) Considering a model where a taxon's links to food sources are of uniform interaction strength and extinction results only from a loss of all links. 2) Allowing extinction to occur prior to the loss of all links by considering survivorship to be a function of maximum sustainable population size, which in turn is a function of the number of surviving links. This consequently generates feedback between consumers and prey. 3) Varying link strength according to empirically observed and theoretically satisfying distributions.