ORGANIZATION IN NATURAL ASSEMBLAGES OF DESERT LIZARDS AND TROPICAL FISHES
Additional Document Info
Techniques are developed for the analysis of community organization and bench tested on a set of simple model systems with known structure (i.e., with and without guild structure, with varying degrees of resource partitioning, with and without "core" resources). Proportional utilization coefficients, ph are positively correlated with the abundance of resources, whereas electivities, e i , correlate negatively with resource abundance. The geometric mean of p i and e i , termed g i , is a superior measure of utilization, more nearly independent of biases associated with resource availability than either of its components, and performs better in bench tests. Organization in observed patterns of resource utilization by four desert lizard and four tropical freshwater fish assemblages is critically evaluated via comparisons with results from two randomization algorithms. Randomizations follow a Monte Carlo technique whereby the dimensions of the original m x n resource matrix are preserved during hundreds of independent runs. The first "scrambled zeros" algorithm rearranges observed values for resource utilization by each consumer and retains consumer dietary niche breadths, but destroys guild structure (zero structure) of observed matrices. The second algorithm ("conserved zeros") also retains observed consumer diet breadths, but only rearranges observed resource utilization coefficients among the particular resources actually used by consumers, thus retaining observed guild structure (i.e., matrix zero structure). By plotting average dietary overlap against ordered niche neighbors, we evaluate (1) relative guild structure using the randomization algorithm that scrambles observed matrix zero structure, and (2) consumer resource segregation within guilds using the algorithm that conserves zero structure. Statistically significant guild structure is evident to varying degrees in all but one low-diversity fish assemblage. All four tropical fish assemblages reveal significant partitioning of food resources during both the wet and dry seasons, particularly among intermediate to distant neighbors. The most species-rich assemblage exhibits extremely high levels of resource segregation during the period of desiccation of aquatic habitat and increased fish densities. Diverse Australian desert lizard assemblages show significant partitioning of microhabitats among ecologically similar species, even though only 15 microhabitat categories are recognized. Analysis of dietary resource matrices based on only 19 prey resources shows that Australian lizards are piled up on certain prey types, forming functional dietary guilds of lizards that eat termites, ants, other lizards, etc. However, no dietary segregation is evident when only 19 prey resource states are recognized. In contrast, when the analysis is redone on more refined resource matrices based on more than 200 prey types, guild structure essentially disappears but niche segregation becomes evident. In the less diverse Kalahari desert, lizard assemblages are more variable and do not appear to be as tightly organized. Patterns of resource utilization in several of these natural vertebrate assemblages are, however, sufficiently organized to suggest broad effects of internal biotic factors, such as ecological constraints of functional morphology and physiology (leading to guild structure), and possibly interspecific competition.