2006 Philadelphia Annual Meeting (22–25 October 2006)

Paper No. 11
Presentation Time: 10:45 AM

SENSITIVITY ANALYSIS OF FOOD WEB MODELS FOR PERMIAN AND TRIASSIC TERRESTRIAL COMMUNITIES OF THE KAROO BASIN, SOUTH AFRICA


ANGIELCZYK, Kenneth D.1, ROOPNARINE, Peter D.2, WANG, Steve C.3 and HERTOG, Rachel2, (1)Earth Sciences, University of Bristol, Wills Memorial Building, Queens Road, Bristol, BS8 1RJ, United Kingdom, (2)Department of Invertebrate Zoology & Geology, California Academy of Sciences, 875 Howard St, San Francisco, CA 94103, (3)Mathematics and Statistics, Swarthmore College, 500 College Ave, Swarthmore, PA 19081, k.angielczyk@bris.ac.uk

Probabilistic trophic network models are a powerful means for gaining insight into how paleontological communities responded to biotic and abiotic perturbations, as well as broader questions such as whether the extinction resistance of communities increases over time. A study of eight Middle Permian to Middle Triassic terrestrial faunal assemblages from the Karoo basin, of South Africa shows that although extinction resistance differs significantly among communities, there is no secular trend for increasing resistance. Likewise, observed end-Permian animal extinctions in the Karoo could have been caused by a disruption of primary producers that precipitated a trophic network collapse. However, these results depend on network parameters that include varying levels of uncertainty, such as the presence or absence trophic links between guilds and the taxonomic diversity of individual guilds.

Here we examine the robustness of our model results for the Karoo by altering the underlying trophic networks in two ways. First, we address the uncertainty surrounding links between guilds by allowing small tetrapods to have omnivorous diets, and by assigning more similar sets of links to aquatic and terrestrial tetrapod carnivores. To examine the effects of guild diversity, we tracked the extinction susceptibility of our eight communities when each included guild is assigned the minimum, mean, or maximum diversity observed for that guild among the communities. Our results suggest that changing these network parameters does have some effect. For example, differences in extinction susceptibility among the communities are less pronounced when guild diversity is controlled. However, many of the patterns from the observed data are still apparent, indicating that the results based on those data are not artifacts. This study also emphasizes one of the major advantages of a model-based approach, the ability to perform experiments on ancient communities to test hypotheses about extinction.