Predicting invertebrate assemblage composition from harvesting pressure and environmental characteristics on tropical reef flats

Invertebrates represent an essential component of coral reef ecosystems; they are ecologically important and a major resource, but their assemblages remain largely unknown, particularly on Pacific islands. Understanding their distribution and building predictive models of community composition as a function of environmental variables therefore constitutes a key issue for resource management. The goal of this study was to define and classify the main environmental factors influencing tropical invertebrate distributions in New Caledonian reef flats and to test the resulting predictive model. Invertebrate assemblages were sampled by visual counting during 2 years and 2 seasons, then coupled to different environmental conditions (habitat composition, hydrodynamics and sediment characteristics) and harvesting status (MPA vs. non-MPA and islets vs. coastal flats). Environmental conditions were described by a principal component analysis (PCA), and contributing variables were selected. Permutational analysis of variance (PERMANOVA) was used to test the effects of different factors (status, flat, year and season) on the invertebrate assemblage composition. Multivariate regression trees (MRT) were then used to hierarchically classify the effects of environmental and harvesting variables. MRT model explained at least 60% of the variation in structure of invertebrate communities. Results highlighted the influence of status (MPA vs. non-MPA) and location (islet vs. coastal flat), followed by habitat composition, organic matter content, hydrodynamics and sampling year. Predicted assemblages defined by indicator families were very different for each environment-exploitation scenario and correctly matched a calibration data matrix. Predictions from MRT including both environmental variables and harvesting pressure can be useful for management of invertebrates in coral reef environments.