Predictive distribution modeling of the sandy-beach supralittoral amphipod Atlantorchestoidea brasiliensis along a macroscale estuarine gradient

Abstract

Exposed marine beaches are physically rigorous habitats in which macrofaunal patterns have been well correlated with physical factors. In this context, the habitat safety hypothesis (HSH) predicts an increase in abundance of supralittoral species from dissipative to reflective conditions in microtidal oceanic beaches. However, the HSH has not been adequately tested in estuarine sandy beaches. Here, we build a predictive model based on the supralittoral talitrid amphipod Atlantorchestoidea brasiliensis that allowed us to test the HSH along a macroscale estuarine gradient. Fifteen Uruguayan sandy beaches along a salinity gradient (from 0.1 to 34.3) generated by the Río de la Plata estuary (SW Atlantic Ocean) were sampled over a two-year period. A conditional two-step procedure using a General Additive Model (GAM) was performed in order to model A. brasiliensis occurrence (1st-step) and abundance (2nd-step), based on a comprehensive set of environmental variables [salinity, water temperature, beach face slope (BFS), mean grain size, wave height and sand compaction, moisture and organic matter]. Each GAM was parameterized using generalized linear models (GLMs). An external validation procedure was used. Data were divided randomly into training (75%) and test (25%) sets. The 1st-step GAM/GLM retained 5 physical descriptors in the model (decreasing order of importance): wave height, salinity, BFS, organic matter and sand compaction. The 2nd-step GAM/GLM retained 6 physical descriptors: sand compaction, water content, salinity, wave height, water temperature and organic matter. The predictive ability of these models and the resulted combined model, as well as their external validity, was highly significant and supported the predictions of the HSH. However, the highest and lowest abundances were observed, respectively, on reflective and dissipative beaches in the outer estuary; i.e., the predictions based on the HSH did not account for the high levels of abundance observed on these beaches compared with oceanic ones. Therefore, the HSH was expanded to include microtidal estuarine beaches.