Predicting spatial patterns in intertidal seaweed habitats using remote sensing and hierarchical species community modelling

Remote sensing of intertidal seaweed habitats is challenged by overlapping spectral signals from multiple co-occurring species groups, making it difficult to predict their distributions. This study applies Hierarchical Modelling of Species Communities (HMSC), a joint species distribution framework that accounts for statistical dependencies and co-occurrence patterns among species groups, thereby improving prediction accuracy, particularly for groups that are otherwise difficult to model individually. We tested this approach using Sentinel-2 (S2) satellite imagery to estimate the coverage of green, brown, and red algae along the Veraval coast, Gujarat, India. To assess whether the findings could be reproduced beyond the original study area, we repeated the joint-versus-single-group comparison using a larger, spatially independent regional dataset from the Saurashtra and Gulf of Kachchh coasts of Gujarat. The HMSC framework effectively modelled the simultaneous distribution of green and red seaweeds based on their shared and individual responses to covariates derived from remote sensing data. Brown algae were weakly predicted at Veraval, where they were sparsely represented, and remained less predictable in the Saurashtra and Gulf of Kachchh dataset despite their higher occurrence and cover, suggesting limited spectral separability, within-group optical heterogeneity and mixed-pixel effects rather than insufficient representation alone. Still, the results show that this method is a promising tool for monitoring complex shallow coastal habitats at local scales and may support ecosystem assessment and management when applied within its demonstrated domain.