As above, so below: A perspective into the application of land‐forest monitoring methods for the assessment of marine animal forests

Abstract Marine animal forests (MAFs) are benthic ecosystems dominated by vertically structuring filter‐ and suspension‐feeders. As terrestrial forests, they are considered biodiversity hotspots, forming canopies, serving as a refuge, nursery, reproduction and feeding shelters for many species. Until just recently, these habitats have represented a challenge in terms of field work and monitoring activities, especially when approaching medium to large scales. However, thanks to the coupling of emerging techniques, such as machine learning and photogrammetry, the development of cost‐effective approaches to assess MAFs might be closer than ever. The main aim of this study is, therefore, to provide an innovative perspective to monitor canopy‐forming organisms (CFOs). In the first place, Structure from Motion (SfM) photogrammetry was applied to capture the three‐dimensional (3D) features from the surveyed benthic assemblage. Subsequently, similarly to the procedures used in airborne LiDAR surveys of terrestrial forests, the resulting 3D point clouds were further processed in a five‐step workflow: (i) point cloud denoising; (ii) classification and filtering of ground points; (iii) Z ‐value normalization to correct substrate irregularity during the height assessment; (iv) individual CFO detection and height assessment; and (v) canopy volume and occupancy quantification. The results obtained with the semi‐automated pipeline were compared to manual quantifications of CFO abundance, density and height frequencies derived from point clouds produced for five Mediterranean MAFs. Broad comparability was found for CFO detection, with an average F 1 ‐score of 0.77 ± 0.12. However, limitations were identified in the estimation of CFO heights, which showed an average value R 2 value of 0.55 ± 0.29. This limitation stemmed from the overall height overestimation in areas with highly irregular substrates. Despite these limitations, the approach presented here marks the first semi‐automatic quantitative analysis of MAF canopy structure and represents a crucial step forward in monitoring and managing these endangered habitats, shifting from species/population‐level assessment to a more ecologically comprehensive approach for evaluating MAF's integrity.