Sampling of environmental variables
At each drift net, flow velocity and water depth were measured three times during base flow and three times each during the P1 and P2 phase. The values of the three measurements were then averaged. Flow velocity was recorded using a micro propeller (Flowatch Flowmeter) 2 cm above the substrate surface (v0 ), at the net-center approximately 13 cm from the substrate surface (vnet ) and 6 cm below the water surface (v100 ). Water depth was measured in front of each net with a rigid meter. Concurrently, turbidity as NTU (Nephelometric Turbidity Unit) was measured at the drift net of the central sampling site using a portable turbidity meter (Hach Lange 2100Q). Additionally, the extension of the dewatering area (DAex ) as distance between the waterline at base flow (Qbase ) and at peak flow (Qpeak ; i.e., flow magnitude) was measured nearby the stranding nets.
Water temperature (T ) and water level were continuously recorded (1 min interval) using one logger (Vemco Minilog II-T, AMIRIX Systems Inc.) and one pressure probe (DCX-22 SG/VG CTD, Keller), respectively, placed at the upstream sampling site of each HP reach. Based on the water level measurements, the mean and maximum up- (URmean , UR max) and down-ramping rates (DRmean ,DR max) were determined. Discharge data were obtained from federal and cantonal gauging stations as well as from the operators of the hydropower plants and were used to determineQbase , Qpeak , flow amplitude (Qpeak - Qbase ) and ratio (Qpeak/Qbase ). The grain size distribution (GSD) along gravel bars of each HP reach was determined by a representative number of line-by-number analyses (n = 1 – 3) using a gravelometer. GSD curves were determined and analyzed, taking particle values as ‘percent finer’. Froude number (Fr ) and bed shear stress (τb ) according to Whiting and Dietrich (1990) were calculated for each HP sampling site.
Data analysis
Data analysis were either based on the total community or at taxa-level. To ensure consistency and to down-weight the influence of rare taxa, only taxa with a relative abundance > 1% over all drift, stranding or benthic samples, were selected to express taxa-specific responses (Appendix C in Data S1). Additionally, to assess trait-specific responses, we classified taxa found in the benthic samples based on two relevant ecological trait categories related to HP sensitivity (RHEOPHYLAX, 2021; Schülting et al., 2021): ‘hydraulic habitat preference’ (classes: lentic/lotic; indicating taxa adaptation to low or high current) and ‘vertical habitat preference’ (classes: interstitial/surface; indicating taxa flow exposure; Appendix C in Data S1).
Benthic community composition and stranding were analyzed in terms of density, expressed as the number of individuals standardized by the sampled area (ind./m2). Single samples were pooled per sampling site (Figure 2a) and chronological sampling (Figure 2b), resulting in 24 and 18 composite samples respectively for each HP reach as well as 12 benthic composite samples for each RF reach.
MIV drift intensity was calculated following Pegel (1980) as the number of individuals in the drift standardized by the drift-net area and the exposure time (ind./m2min). To consider the effect of an entire HP scenario, drift samples collected during the UR, P1, P2 and DR phase were pooled per sampling site and HP scenario. For each HP and RF reach this resulted in 18 composite samples each for base flow and HP scenario. Further, since MIV drift patterns, and probably also stranding patterns, are strongly related to the benthic source population, we calculated the drift and stranding propensity (e.g., Bruno et al., 2013; Timusk et al., 2016) as the drift intensity and stranding density, respectively, standardized by the benthic density. All benthic samples were pooled per HP reach to generate a single estimate of benthic density at the reach scale.