Aim

To determine the ecoregions (spatial marine areas with similar environmental and physical conditions associated with relatively homogeneous fish assemblages) for shallow reef fish assemblages based on predictive models of beta diversity (β‐diversity) that account for both large‐scale environmental factors and local habitat characteristics. We assessed the influence of a spatial scale to rank the importance of these factors.

Location

New Caledonian (south‐west Pacific Ocean, 17–24° S, 158–172° W) Exclusive Economic Zone, Coral Sea Marine Park.

Taxon

Fish.

Methods

Fish and habitat data that were collected at 13 sites around New Caledonia using unbaited rotating underwater video (285 sampling stations) were analysed. Gradient forest modelling was used to predict the fish β‐diversity along the gradients of environmental factors. Ecoregions were obtained by applying clustering methods to gradient forest predictions.

Results

The gradient forest models of β‐diversity retained 59 species (total: 206 fish species) with R² > 0, including 19 fish species with R² from 0.03% to 69%. For these 19 species, the models explained up to 26% of the variance. At a large scale, β‐diversity was significantly explained by nutrient concentrations, sea surface salinity and temperature. Among the eight ecoregions that were delineated based on the β‐diversity predictions, three regions corresponded to remote sites under oceanic influence where human pressures are low and the surface nutrient concentrations are high. On the local scale, the benthic habitat explained β‐diversity better than the physical and chemical parameters, particularly in the areas subject to anthropogenic pressures.

Main conclusions

On the local scale, the respective importance of environmental factors (physical and chemical parameters versus benthic habitat) differed according to ecosystem health. Our findings suggest that nutrient enrichment due to avifauna may have a positive effect on fish β‐diversity when an ecosystem is healthy. The ecoregions reflect fish species composition in relation to a large set of environmental parameters.