Résumé
Acoustic surveys represent the standard fishery-independent method worldwide for evaluating the biomass and spatial distribution of small pelagic fish populations. Considering the peculiarities of the spatial behaviour of pelagic fishes, the efficiency of the survey design in determining their biomass and spatial distribution is related to the ability to capture the portion of the patches accounting for a larger part of the total biomass. However, the spatial structure of the patches could be strongly influenced by ecosystem characteristics as well as by changes in total biomass related to a density-dependent mechanism. This is of particular interest for anchovies and sardines, which are known for their wide fluctuations and high sensitivity to the environment. In this study, we analysed the efficiency of acoustic surveys targeting European anchovies (Engraulis encrasicolus) and European sardines (Sardina pilchardus) in 10 different areas of the Mediterranean Sea spanning three years of different biomass levels. Using the geostatistical coefficient of variation (CVgeo) of the average occurrence probability of high/medium density values, we showed different patterns in terms of survey design efficiency among areas and species. Anchovies usually showed a lower CVgeo than sardines in the Alboran Sea. In 4 out of 20 cases, CVgeo values showed a consistent decrease with increasing biomass, while in the remaining cases, the CVgeo did not follow any clear pattern, suggesting the presence of important environmental effects. Higher survey design efficiency was found in highly productive sectors influenced by river run-off, allowing us to hypothesize that higher productivity along with the presence of well-localized enrichment mechanisms could favour a spatially consistent distribution and coherent organization of fish populations, leading to higher precision estimates with a given transect design. While most surveys displayed CVgeo close to 10% or less even at low biomass, indicating generally good performances of the survey design, a few areas exhibited higher CVgeo, yielding a potential need to decrease the intertransect distance, always keeping in mind that any survey should be as synoptic as possible.