Foraging in a variable world: species interactions across ecological gradients

Abstract

This dissertation explores one broad question; how consistent are dietary niches? Between species? Under resource limitation? In different ecological contexts? In each chapter, this question is asked of a different aspect of trophic ecology and at a different biological scale. A broad and deep literature documents population-level diets; can synthesizing this literature identify generalities in diet composition? Seasonality defines most ecosystems; do resource-availability fluctuations alter guild-level niche organization? Trophic webs are typically studied with a taxonomically or functionally narrow lens; does a broader scope alter our understanding of network structure and its implications? Chapter 1 is a global synthesis of population-level vertebrate diets that asks whether defining those diets as abundance distributions reveals general patterns in their composition. After compiling a dataset of almost 1200 population-level diets from more than 500 species, I found that dietary-abundance distributions (DADs) were overwhelming (92.5%) best described by a hollow curve function implying that few dietary items are eaten commonly and many are eaten occasionally. Moreover, flatter DADs representing more generalized diets were associated with sampling across seasons, observational sampling methods, herbivory, and omnivory. Chapter 2 explores how seasonality in a Mozambican savanna modulates dietary niche overlap among large mammalian herbivores. Using almost 2000 dung samples from 13 large herbivore species in three seasons across four years, I found that population-level diet breadth was greater in the dry season. However, dietary niche overlap between species decreased concurrently (following Pianka's niche overlap hypothesis). This pair of results suggests that Rosenzweig and Abramsky's theory of centrifugal niche organization, developed to explain habitat use, may structure herbivore dietary niche organization. Chapter 3 documents a clear link, in a Kenyan savanna, between the presence of large mammalian herbivores and the structure of plant-pollinator networks. When large herbivores were excluded from 1-ha vegetation plots, flowers were threefold more abundant and 50% more diverse. As a result, pollinators were 20% more active and 50% more diverse. In combination, networks of interactions between plants and pollinators were larger, more functionally redundant, and less susceptible to pollinator loss when large herbivores were absent.