Common census methods can lead to fundamentally different ecological interpretations of coral reef fish communities

Abstract In an era of large-scale stressors, scientists are racing to collate equally large datasets to answer macroecological questions. Coral reefs lie at the forefront of system-wide transformations and are thus the focus of many macroecological studies. A lack of standardised, global data across time, however, has resulted in the coalition of data from disparate sources, often comprising different survey methods. However, can we make the same ecological inferences from data collected using different survey protocols? Here, we employed two of the most common reef fish census techniques—point counts and belt transects—of varying survey effort (time and area) to assess whether we can draw similar ecological conclusions from different methods. We found that point counts produced the most extreme estimates of abundance, biomass, and species richness across all methods, with instant point counts recording the lowest and 10-min point counts recording the highest values; underscoring the potential for total survey time to influence results. Likely owing to the near-maximum diver effect, point counts also consistently missed the largest fishes on the reef and were characterised by common species. Consequently, point counts produced biomass-size distributions that were vastly different from belt transects, which could lead to markedly different interpretations of the energetic profiles of fish communities. Combining reef fish data collected using different survey methods can bias inferences, as survey time can produce abundance-based biases, while survey methods can produce size-based biases. Regardless of survey method, we found that each site would need at least 30 individual transects to precisely quantify community dynamics. Using uniform survey methods with enough replication would ensure biases and errors are more consistent, which can lead to more robust inferences.