Assessing sandy beach macrofaunal patterns along large-scale environmental gradients: A Fuzzy Naïve Bayes approach

• We model macrofaunal distribution patterns on a large-scale estuarine gradient.
• We use a Fuzzy Naïve Bayes (FNB) classifier adapted to numerical data.
• The FNB considers simultaneously all environmental components.
• Our modeling approach reproduced sandy beach macrofauna descriptors.
• Salinity range was a critical variable governing distributional patterns.

A Fuzzy Naïve Bayes (FNB) classifier was developed to assess large-scale variations in abundance, species richness and diversity of the macrofauna inhabiting fifteen Uruguayan sandy beaches affected by the effects of beach morphodynamics and the estuarine gradient generated by Rio de la Plata. Information from six beaches was used to estimate FNB parameters, while abiotic data of the remaining nine beaches were used to forecast abundance, species richness and diversity. FNB simulations reproduced the general increasing trend of target variables from inner estuarine reflective beaches to marine dissipative ones. The FNB model also identified a threshold value of salinity range beyond which diversity markedly increased towards marine beaches. Salinity range is suggested as an ecological master factor governing distributional patterns in sandy beach macrofauna. However, the model: 1) underestimated abundance and species richness at the innermost estuarine beach, with the lowest salinity, and 2) overestimated species richness in marine beaches with a reflective morphodynamic state, which is strongly linked to low abundance, species richness and diversity. Therefore, future modeling efforts should be refined by giving a dissimilar weigh to the gradients defined by estuarine (estuarine beaches) and morphodynamic (marine beaches) variables, which could improve predictions of target variables. Our modeling approach could be applied to a wide spectrum of issues, ranging from basic ecology to social-ecological systems. This approach seems relevant, given the current challenge to develop predictive methodologies to assess the simultaneous and nonlinear effects of anthropogenic and natural impacts in coastal ecosystems.