Comprehensive knowledge of the effects of disturbances on biodiversity is crucial for conservation and management, not least because ecosystems with low biodiversity may be the most vulnerable. In rivers, the role of disturbance in shaping aquatic biodiversity has mainly focused on floods. Perennial rivers (PRs) often flood, whereas intermittent rivers (IRs) flood, stop flowing and dry. Despite the recent and significant increase in research on IRs, controversy remains about whether they are more or less biodiverse than PRs. Our aim was to determine (Q1) if PRs and IRs differ in biodiversity and (Q2) if the direction and magnitude of the differences (effect sizes) are related to environmental (climate, season, habitat, longitudinal zonation and anthropogenic disturbance) and/or biological factors (taxonomic group). We conducted a meta-analysis on 44 published studies of PR and IR biodiversity that had replicated data. We applied random effects models to the data to obtain weighted mean effect sizes for differences between PRs and IRs, and their confidence intervals, by first considering all studies and then by splitting studies into groups on the basis of the above factors. We found that biodiversity was significantly higher in PRs than in IRs (Q1). We also detected significant differences (PRs>IRs) in studies of macroinvertebrates, in those conducted within arid and temperate climates, dry and wet sampling seasons, headwaters, and regions subject to different levels of anthropogenic disturbance (Q2). Our meta-analysis suggests that the expected increase in the prevalence of IRs in certain regions of the world due to global change could result in a decrease in freshwater biodiversity. To better manage and preserve aquatic biodiversity under future global change scenarios and to avoid potential ecosystem consequences of biodiversity loss, conservation efforts should be targeted towards those environmental conditions or taxonomic groups with significant differences (PRs>IRs).

Meta-analysis_PRs_IRsPublished studies that recorded biodiversity of PRs and IRs. For each replicated study (i.e. with n > 1): • We extracted means (i.e. P_M and Int_M), SDs (i.e. P_SD and Int_SD), and number of sites (i.e. P_N and Int_N). • We obtained information on site. In particular, we considered the following 6 factors, each with several levels within: climate, sampling season, habitat, longitudinal zonation, level of anthropogenic disturbance, and taxonomic group (see Table 1 and Appendix S3).