Image-based ecological assessment of deep-sea sponge, coral and other cnidarian assemblages through a morpho-functional approach

Porifera and Cnidaria are key phyla that contribute to the structural complexity and functioning of marine ecosystems and are commonly considered in the identification of Vulnerable Marine Ecosystems. Their diversity and distribution largely rely on seafloor images, but low-resolution taxonomic identification is hindered by morphological plasticity and convergence, and by the need for microscopic diagnostic characters or molecular markers that are not accessible from images. Higher-rank taxonomy also poorly reflects their morphological and functional diversity. To address this issue, we developed an image-based classification combining multiple macro-morphological traits to define statistically derived morpho-functional groups (MFGs) which were then used as response variables in ecological analyses instead of taxa. This framework provides a reproducible and transferable approach for comparing patterns of habitat-forming taxa across regions, regardless of the level of taxonomic knowledge available, by relying on observable morphological traits rather than uncertain taxonomic IDs. We applied this framework to sponge and cnidarian assemblages from seamounts and island slopes in the Mozambique Channel, a poorly sampled region where taxonomic limitations are acute. We assessed the relevance of an MFG approach to provide additional information on sponge and cnidarian assemblage diversity, spatial variability and relationships to environmental gradient, as well as their structuring roles on the surrounding megabenthic community, compared with the available taxonomic resolution. We identified 37 sponge and 34 cnidarian MFGs revealing substantial morphological variation within the coarse taxonomic categories available from imagery. MFGs captured greater spatial differentiation and local variability than the taxonomic classification, particularly for sponges, for which only two taxonomic classes were available. Several MFGs were also associated with particular environmental conditions and with other megabenthic taxa, although the proportion of community variation explained by these relationships was low. Overall, the MFG approach, especially effective for sponges, allowed to retain and quantify morphological and structural information of habitat-forming taxa that can be lost using coarse image-based taxonomy. Its multi-trait classification can be readily translated into existing hierarchical image-classification systems such as CATAMI, allowing finer morphological information to be retained while maintaining consistency with broader standardised categories. Finally, the framework proposes metrics to detect and quantify the heterogeneity and spatial variability of potentially vulnerable biogenic habitats in imagery-based studies. Keywords: Morpho-functional traits, Porifera, Cnidaria, Habitat-forming species, Vulnerable marine ecosystems (VMEs), Trait-based classification, Deep-sea ecology