PARENT SESSION
Poster Session #24: Modeling.
Tuesday, August 6. Presentation from 5:00 PM to 6:30 PM. Exhibit Hall B & C, TCC

Sensitivity and success of the Niche Model at predicting the network structure of extremely large food webs.

VACCARO, ERIN¹, PULESTON, CEDRIC¹, MARTINEZ, NEO^*,1, WILLIAMS, RICHARD¹, ¹ San Francisco State University, San Francisco, CA

ABSTRACT- A recent refinement of the cascade model of food web structure, called the niche model, previously demonstrated powerful predictive ability for a variety of aquatic and terrestrial webs within a limited parameter space. However, the sensitivity of the niche model to input parameters and its ability to predict food-web structure well beyond parameters of the earlier analyses had yet to be explored. Therefore, we systematically examined the sensitivity of the niche model to its two input parameters: number of functionally distinct species and amount of trophic connectance. We also tested the model against two highly resolved webs larger than any in the literature. The two webs, Lake Tahoe, California, and Mirror Lake, New Hampshire, each contain 172 trophic species. Our sensitivity analyses predict patterns of food-web properties expected to be seen among webs with 10-200 trophic species and connectance from 0.05 and 0.30. The percentage of cannibals, the variability of the degree distribution and trophic generality appear insensitive and therefore scale-invariant. Most other properties appeared sensitive and therefore scale-dependent. These include percentages of top, intermediate and basal species, vulnerability variability, maximum similarity and measures of food chain lengths. Our tests confirm the accuracy of the model's predictions of several key properties of the large lake webs. Of those properties that the model predicts with less accuracy, most are ameliorated by aggregating the webs to define trophic species as a group of taxonomic species sharing 90% or more of their prey and predators. Such aggregation may improve model fit by reducing the effect of suspected methodological artifacts.

KEY WORDS: ecological modeling, food chain, statistical ecology, network theory