Résumé
Coral reefs around the world are facing increasing disturbance; however, the ability to monitor them is constrained by cost and experience factors. The Line Intercept Transect (LIT) method is usually used with an expert level of identification (30 benthic categories and sometimes coral genera), but these are time-consuming, require a high level of expertise and are therefore expensive. Over the last 20 years, surveys performed by non-specialist volunteers have provided data for increasing numbers of reefs, using simplified sampling methods and more basic levels of identification. In between these extremes, new stakeholders (e.g. environmental assessment agencies, consultancy firms, marine nature reserves, etc.) requiring an intermediate level of expertise and resolution are now carrying out coral reef surveys. Across all levels, it is desirable to identify monitoring methods that optimize the data provided and the ease of implementation according to the users’ needs. In this study on surveys from Réunion Island collected between 2003 and 2013, we used multivariate analyses to compare four increasing levels of identification using the LIT method. For each level, we studied the structure of the benthic community and its spatial (reef flat vs. outer slope) and temporal (before vs. after a cyclone) variation. The most basic level of identification showed differences between the reef flat and outer slope, but did not show a significant effect of the cyclone on the benthic community. The two highest levels of identification, both of which we rated ‘expert’, showed the highest differences in variation. The intermediate level of identification supplied the same information as the expert levels, but required less effort and was therefore less expensive. Next, we compared the LIT with the Point Intercept Transect (PIT) method. At a level of identification appropriate to the interval used, PIT supplied an almost equivalent level of information as LIT, while reducing the time spent underwater. Thus, when managing coastal areas or carrying out impact studies to monitor disturbance, it may be useful to (i) perform surveys with an intermediate level of identification, and (ii) using PIT rather LIT.