Vocal behaviors are related to nonrandom structure of anuran breeding assemblages in Guyana

An important problem in community ecology is that of identifying patterns of structure in data by employing null models. Null models use reiterative methods to generate large numbers of virtual communities that are compared with empirical assemblages. These allow detection of nonrandom structure of organization attributable to biotic and abiotic factors. We tested the hypothesis that vocal behaviors are related to nonrandom structure in frog breeding assemblages. Male frogs of many species gather at ephemeral bodies of water, and produce advertisement calls that primarily attract conspecific females who are ready to mate. We determined whether null models and multivariate techniques could test for nonrandom patterns in pond breeding assemblages comprised of 16 and 14 species in Guyana, South America. Comparisons of empirical data against two null models indicated that observed patterns of temporal, spatial, and spectral acoustic variables, and calling site use by heterospecific frogs were significantly nonrandom patterns. Observed patterns of spatial and calling site usage appeared to contribute to high species richness of these Neotropical frog breeding assemblages. Principal component analysis of advertisement calling variables of sympatric treefrogs (Hylidae) suggested that species differed in dominant frequency, fundamental frequency, pulse rate, note duration, call duration, and call rate. Discriminant function analysis of these same call variables in the numerically three most common treefrog species of the genus Scinax indicated that fundamental frequency and pulse rate alone separated the three species. Overall, the null models and multivariate statistical techniques suggested that nonrandom patterns of spatial and calling site usage were implicated in the apparent nonrandom structure of these Neotropical pond breeding anuran assemblages.