Manila clam Ruditapes philippinarum is one of the most important bivalve species, with high commercial value, cultured in coastal areas around the world. However, few studies have attempted to model the suitable habitats of this species. For farming, knowledge of suitable habitat is critical information. Here, we use an ensemble-modelling approach implemented in the biomod2 platform to examine potential suitable habitats in Moon Lake. Initially, nine single-algorithm species distribution models (SDMs) were constructed by combining 96 Manila clam actual presence-and-absence data with oceanographic and sediment data. A weighted ensemble model was then derived from the different individual models. Results showed that the contribution of the six predictor variables differed among modelling techniques, in general, the substrate-related variables contributed the best at predicting habitat suitability. The performance of single-algorithm models ranged from poor to excellent, and the weighted ensemble model performed better than any individual modelling technique according to performance metrics. The use of an ensemble of predictions allowed us to identify potentially suitable areas for clam farming. The predicted suitable, secondary suitable and general suitable habitat for R. philippinarum accounted for 7.69%, 9.03% and 11.80% of the total area of Moon Lake, respectively. And both the average shell length and biomass of Manila clam displayed positive relationships with HSI value. This ensemble-modelling approach could be a useful tool to identify suitable aquaculture sites for Manila clam farming.
Read full abstract