Dedication. Acknowledgments

Abstract

xvii Dedication xixAcknowledgments xx . See panel C. for a scenario where one prey is preferred over another. C. A tri-trophic system, where a top-predator consumes a meso-predator, which in turn consumes two prey. The sizes of the nodes denote relative abundance, and the thickness of the arrows denote the magnitude of the biomass flow. The top-predator in the leftmost network is relatively abundant, depressing the meso-predator population, thereby limiting the meso-predator's e↵ect on the two prey populations. The two rightmost networks (boxed) have reduced top-predator populations. In both cases, the meso-predator has greater abundance due to decreased predation pressure. However, in one scenario, both prey are preferred equally, resulting in apparent competition dynamics. In the second scenario, one prey is preferred over another, resulting in an asymmetric cascade. Such preferences could result from specific anatomical constraints of the meso-predator, a scenario explored in Chapter 4. C 1 and C 2 , and 4 prey: r 1 r 4 ) is represented by both pairwise interaction distributions for each link, and a network-level interaction distribution for the system. Link thickness represents the median link-strength, shown with the vertical bar in each pairwise interaction distribution. B-C. Both model food-webs X and Y have link-strengths drawn from the NLID. In model food-web X, the weights are distributed similar to the median weights of the empirical web (i.e. equivalent structure), but they are distributed di↵erently among nodes (i.e. alternative interactions). In model Y , the weight structure is di↵erent than that of the empirical web (i.e. alternative structure), and necessarily, the distribution of weights among nodes di↵ers as well (i.e. alternative interactions). Dotted nodes indicate those with link-strengths that di↵er from the median link-strengths of the empirical food-web. D-F. Given a single niche axis based on a continuous resource trait (e.g. body size) where prey are ranked from r 1 to r 4 , distributions of resource use are shown for the two consumers. Representations of dietary overlap illustrate higher similarity in diet between predators of the empirical system and model X, whereas the consumers in model Y have little dietary overlap, despite sharing the same NLID. 5 A-C. Sensitivity analyses of the Saskatchewan, Amboseli, and Lake Naivasha food-web ensembles, respectively. Similarity values for each cuto↵ (s i ) for Model 2 are on the y-axis, and the x-axis ranges from a lower standard deviation (SD) than the empirical SD to a higher SD (such that the empirical SD is central to the range; stippled vertical line). The colored data points represent the median similarity indices for each cuto↵ value (legend), while the top and bottom whiskers denote the 25 th and 75 th percentiles, respectively. The underlined values mark the standard deviation corresponding to the highest average similarity across cuto↵ values. For all systems, the highest average similarity is close to or exactly matches ...