2.5| Data analyses
Changes in taxa composition between each subsequent sampling period was measured as temporal beta-diversity index (TBI). We calculated two different TBI indices, using the percentage difference form abundance data and the Sorensen index based on presence-absence data. The percentage difference (D%diff ) between two sampling periods (ti and ti+1 ) was defined as D%diff= (B+C)/(2A+B+C), where A is the sum of the abundance of all taxa found in both sampling periodti and ti+1 . This represents the unscaled similarity between the two sampling periods. B is the part of abundance of the common taxa that is higher in samples atti than at ti+1 and is the unscaled sum of taxa losses between sampling periods. C is the part of the abundance that is higher ti+1 than atti and is the unscaled sum of taxa gainsbetween the two sampling periods. The equation for the Sorensen index, using presence-absence data, is the same as for the D%diff outlined above (see Legendre (2019) for more details). Differences in abundances and presence-absence data among subsequent sampling occasions were tested for significance with a paired t-test. The p‐values (based upon 9999 random permutations) were controlled for the family-wise error rate using Holm correction. The calculations of the TBI indices and significance tests were done in the R package “adespatial” (Dray et al., 2019).
We analyzed data from riverine and lake habitat separately. Responses to rotenone treatment is expected to vary among habitats. The abundance of each invertebrate taxon is expressed as average number of specimens per kick-sample for each sampling period at each site.