[Statistical methods for analysis of communities' species structure (with riverine macrozoobenthos as an example)]

Species turnover or coherence in species co-occurrence as well as boundary clumping and nestedness in structural composition of ecological communities reflect the extent of determinancy in their organization (Leibold, Mikkelson, 2002). These phenomena may be a consequence of either interactions between species or heterogeneity in spatial distribution of populations density. We have examined statistical patterns of species structure variability using benthic communities of riverine ecosystems as an example. The ecosystems studied are characterized by strongly pronounced linear gradient of landscape features and environmental factors. The results of a long-term hydrobiological survey being conducted at 22 observational stations on the Sok River along with its tributary, the Baytugan River (Lower Volga basin, total watercourse length is 375 km) are involved into the analysis. A spreadsheet for statistical processing of the data included 375 macrozoobenthic taxa contained in 147 samples. An assessment of species structure nestedness in benthic communities at separate sites and along the watercourse as a whole has been carried out using various metrics such as nestedness "temperature" (Patterson, Atmar, 2000), discrepancy measure (Brualdi, Sanderson, 1999), nestedness based on overlap and decreasing fill (NODE--Almeida-Neto et al., 2008) and others. Statistical significance of ecosystems structural determinancy has been tested by means of randomization procedures and standard null models (Gotelli, 2000). The conclusions seem to be ambiguous and dependent on a level and scale of an ecosystem resolution into separate blocks, also on configuration and completion of initial bio-geographical tables. A searching for reliable and representative criteria of nestedness, invariant to various non-ecological modifications of the matrices but sensitive to estimation of analyzed ecological processes and suitable for comparisons of communities, is clearly needed. A quantitative estimation of species turnover and coherence in species cooccurrence has been performed using different indices of unique combinations and checkerboard score (Stone, Roberts, 1992) as well as Schluter's variance test. By means of empirical Bayesian approach (Gotelli, Ulrich, 2010) records of species pairwise combinations are formed where the frequency of species co-occurrence cannot be interpreted as a random value. Positive and negative relationships between taxa in macrozoobenthic communities, which are found out to be statistically significant, in most cases can be explained as being not the consequence of competition for resources but of spatial heterogeneity of biotopical conditions along the whole length of the watercourse.