A review of seascape complexity indices and their performance in coral and rocky reefs

Seascape complexity is an important driver of ecological processes in marine systems. Today, high-resolution, multiscale bathymetric data are being collected due to rapid advances in marine technologies and image processing, drastically improving the detailed mapping of the physical structure of the seascape. However, these data are rarely synthesized to create comprehensive complexity estimates, and complexity is still mostly measured using a small set of simple indices. The aims of this study are to: (a) review existing seascape complexity indices and propose innovative indices designed to capture the marine organism perspective, (b) quantify the interrelationships among these complexity indices; (c) test the performance of these indices in explaining fish assemblage structure; and (d) provide R code to easily reproduce the indices. Seascape bottom topography for this study was quantified using digital depth recordings along transects in Mediterranean sub-tropical rocky reefs and Red Sea tropical coral reefs. These were used to generate complexity indices that were then compared to visually surveyed fish assemblages. We found that several common indices captured similar complexity facets, while an innovative family of structural diversity indices, representing the diversity of physical elements, captured distinct complexity facets not represented by existing indices. No single index was consistently superior; however, vertical relief was consistently included as a top predictor of fish assemblage structure. Interestingly, the most commonly used index, rugosity, was a poor predictor. We suggest a new, distinct set of structural diversity indices that may explain considerable variation in fish assemblages. The results suggest that several indices may need to be combined to capture the full influence of complexity on marine diversity and caution against the use of a single ‘universal’ index. While bathymetric data are increasingly being collected at high resolutions and increasing scales, synthesizing these data requires the use of appropriate complexity indices. The guidelines and recommendations presented here, along with the supplemental R code, will facilitate progress towards a more complete representation of different complexity facets.