SCALE-DEPENDENCE OF FOOD-WEB STRUCTURES - TROPICAL PONDS AS PARADIGM

Two sets of time-specific food webs were constructed using all known details of trophic biology of the 87 component species found in two tropical freshwater pond systems. Analysis of the webs showed that several food web statistics were scale-dependent. To examine the robustness of this finding, 11 500 computerized analog webs were built upon the specified rules of non-random trophic links (L) between the organisms found in the ponds. The structural properties of the analog webs, like the recently published species-rich webs, disagreed with several generalizations made by food web theorists over the last two decades. The magnitude of L and the frequency of long food chains increased with S. Proportions of top, intermediate and basal species showed strong nonlinear variation with S. The mean number of links between different trophic levels seemed to be scale-dependent, though apparently scale-invariant for the pond webs. However, the relative proportions of link fractions appeared to be scale-invariant, with the intermediate-basal links most numerous. The mean predator-prey ratio appeared to be the most scale-invariant of all the web properties, but the scale invariance of the ratio may not be directly derived from species fractions. The study represents the largest tropical aquatic food web database, and indicates that (a) as many details of inter-specific interactions as possible within a web are necessary for unraveling the actual complexity of real ecosystems, and (b) computerized analogs of real webs, incorporating all necessary details of interactions, may be fruitfully used to ensure robustness of conclusions about the web features.