Estimating the Magnitude of Primary Production Loss in the End-Permian Mass Extinction

A number of lines of evidence suggest a large-scale reduction of primary productivity at the end-Permian mass extinction. These include the plant and pollen fossil record, a worldwide hiatus in coal formation, a spike in fungal spores, changes in sedimentation, and shifts in carbon isotope ratios. However, the exact magnitude of this primary production die-off is not well constrained, and its relationship to vertebrate extinctions is not firmly established. Here, using a trophic network model, we infer the minimum level of primary production loss that could result in observed extinction levels of terrestrial fauna in the Late Permian Dicynodon Zone, Karoo Basin, South Africa. We use the CEG trophic network model (Roopnarine 2006) to simulate the results of primary production loss on species at higher trophic levels. We then use Markov Chain Monte Carlo methods, such as the Metropolis-Hastings algorithm, to solve the inverse problem: estimating the magnitude of primary production loss responsible for observed extinction levels among terrestrial herbivores and carnivores. We find that high levels of extinction among primary producers (with posterior probabilities in the 80-100% range) were necessary, consistent with a bottom-up mode of extinction.