全球变化知识资源中心

Ongoing global changes are causing major ecological shifts worldwide. Biological trends need to be assessed over long periods of time to better understand past and current community responses. The present study developed a methodological framework for meta-analyses to be conducted that account for the temporal and spatial autocorrelation of observational data. We provided the R code for processing this framework, which enables temporal trends to be tested through the analysis of long-term, spatially structured datasets. Taking both types of autocorrelation into account resulted in more conservative but arguably more reliable statistical outcomes. This meta-analysis framework was then applied to investigate long-term trends in environmental and fish-community time series in multiple stations in large French rivers over the past 4 decades. General significant upward and downward trends were highlighted in water temperature and flow discharge, respectively, over the study period. Concomitantly, the density of numerous species increased, resulting in large increases in both species richness (about + 50%) and total fish abundance (approximately four-fold), but with no significant trend in species evenness. Strong changes in species composition were observed during the study period, with an overall upward trend in the relative abundance of newcomers (i.e. species not sampled during the first years of the survey), while the trend in relative abundance of non-native species was non-significant. Moreover, the strongest signal underlying community changes was replacement of northern by southern species. This study showed major changes in fish density and community structure in large rivers over the past 40years and represents, to our knowledge, one of the first large-scale actual demonstrations (i.e. based on observations rather than predictions) of an overall poleward shift of freshwater fish communities in response to ongoing global changes.