Medium- and long-term temporal trends in the fish assemblage inhabiting a surf zone, analyzed by Bayesian generalized additive models

The present study characterizes the fish assemblage in the surf zone of Cassino Beach, Rio Grande, Brazil, and analyzes temporal fluctuations in richness and abundance of these species in medium (months) and long terms (years), associating them also with abiotic covariates. Data were collected monthly between 1996 and 2012 at two locations. Bayesian generalized additive models (GAMs) were used as statistical tool, placing this study among few that have used Bayesian GAMs in Ecology. Our results show a decrease in both species richness and abundance of the most representative species, over the last 16 years, but no significant distinction between locations. Water temperature and salinity along with seasonality were the statistically most influential explanatory covariates to describe fluctuations in richness and abundance. Higher discharge rates of the three main rivers that flow into Patos Lagoon (Jacui, Taquari, Camaqu) were associated with increased richness and abundance of some species in the assemblage. Hence, our findings show that medium- and long-term fluctuations in richness and species abundance are controlled by abiotic factors related to seasonal cycles (temperature) and productivity of the ecosystem. Long-term changes seem to be also related to man-induced factors and climate change; but further research is needed.