A MODEL-BASED APPROACH TO EVALUATE THE EFFECT OF SAMPLING AND COMMUNITY STRUCTURE ON THE INFERRED PREDATION ESTIMATES (Invited Presentation)

Predation estimates based on preserved records of predation marks in shelled marine invertebrates are important in evaluating the evolutionary effect of ecological interactions. Comparing predation estimates across time and space is built on the assumption that the estimates are not differentially influenced by abiological factors. Although the biological reliability of such interpretation is tested when taphonomic forces are affecting such record, the effect of sampling and community composition is largely ignored as a potential contributor in controlling the inferred predation estimates at the community level. Aspects of a specific community such as evenness, intensity of predation may influence the estimates of predation intensity and in recognizing prey taxa.

Using a resampling technique, we tried to develop a methodological approach to understand the effect of community structure on the inferred predation estimates. We theoretically simulated a number of cases representing communities with different levels of evenness (low to theoretical maximum), predation intensity (PI: low, medium and high) and predatory behavior (selective, non-selective). Simulated communities with 3000 individual and 30 species representing each combination of evenness, intensity and behavior were resampled without replacement and the process was iterated 1000 times. We noted the variation in the overall PI and the number of recognized prey taxa in each iteration. Our results demonstrate that evenness of a community does not influence the inferred PI for non-selective predation. However, the overall inferred PI decreases with increasing evenness when a predation event is selective, preferentially targeting the common species. We find a positive relationship between evenness and inferred PI when rarer species are preyed upon. Our results also show that the inferred PI is sensitive to the sample size: it converges to the original PI with increasing sample size. Inferred number of prey taxa follows rarefaction curve and can have large variability depending on the sample size and evenness.

This approach provides a post-collection normalization framework to recognize communities that can be compared and provide critical insight regarding the validity of biological inferences about predator-prey interaction.