Probabilistic assessment of the species composition of benthic communities using Bayesian models

The results of using a unified Bayesian model to estimate components of species diversity on the example of bottom communities of small and medium-sized plain rivers are considered. The model was built using data on the occurrence of 147 macrozoobenthos taxa in 132 sections of watercourses in the Middle and Lower Volga basin. Estimates of ecological affinity for the phenomenon of "dark diversity" obtained by two logistic regression models were calculated and interpreted: for each species according to their 4 main biological properties and for each site according to 7 abiotic factors. The suitability of each species for each site was also estimated, it was calculated as the standardised deviation of the frequencies of co-occurring species from the expected probabilities according to a hypergeometric distribution. Based on these model estimates, unified (pooled) estimates of the probability Pij of detecting the i-th species at the j-th site were calculated. The accuracy and efficiency of the constructed model are reviewed, as well as suggestions for improving its quality are made