Precision and accuracy of common coral reef sampling protocols revisited with photogrammetry

The rapid decline of coral reefs calls for cost-effective benthic cover data to improve reef health forecasts, policy building, management responses and evaluation. Reef monitoring has been largely based on divers' observations along transects, and secondarily on quadrat-based protocols, video and photographic records. However, the accuracy and precision of the most common sampling approaches are not yet fully understood. Here, we compared benthic cover estimates from three common sampling protocols: Reef Check (RC), Atlantic and Gulf Rapid Reef Assessment (AGRRA) and photoquadrats (PQ). The reef cover of two contrasting sites was reconstructed with ∼450 m² orthomosaics built with high resolution Structure-from-Motion (SfM) photogrammetry, which were used as references for comparisons among protocols. In addition, we explored sample size requirements for each protocol and provided cost-effectiveness comparisons. Our results evidenced between-reef differences in the accuracy and precision of estimates with the different protocols. The three protocols performed similarly in the reef with low macroalgal cover (<0.5%), but PQ were more accurate and precise in the reef with relatively high (∼20%) macroalgal cover. The sample size for estimating coral cover with a 20% error margin and a 0.05 significance level was lower for PQ, followed by AGRRA and RC. Considering performance, cost surrogates and equipment needs, cost-effectiveness was higher for PQ. We also discuss costs, limitations and advantages/disadvantages of SfM photogrammetry as a sampling approach for coral reef monitoring.