Discussion

Considering expected future developments of the energy demand and the concurrent progression of the implementation of climate change goals, the need for energy storage capabilities such as hydropower reservoirs will become more and more important in the near future (Boes et al., 2021). As a result, an increase in HP operation thus poses a further threat to physical habitats and biodiversity within riverine ecosystems, highlighting the need for a prudent approach. Even though the mitigation of HP impacts is a legal requirement in many countries (e.g., Swiss Water Protection Act, European Water Framework Directive), the complex interplay of HP-related impacts on MIV still remains poorly understood, which complicates the implementation of efficient mitigation strategies.
Several studies suggest that river reaches affected by HP are characterized by reduced benthic abundance, biomass and altered community composition (Céréghino et al., 2002; Céréghino & Lavandier, 1998;; Elgueta et al., 2021; Leitner et al., 2017; Moog, 1993), which in consequence may have negative effects on metacommunity-wide functional diversity (Kjærstad et al., 2018; Ruhi et al., 2018). Within this field experimental study – conducted in three HP-regulated Swiss rivers – we could prove short-term (drift and stranding) and long-term (benthic density, established community composition) effects of HP operation within the investigated river reaches. Further, we could identify some HP-related parameters most likely responsible for triggering active and/or passive drift and stranding, as well as show that MIV response to HP is highly taxa-, trait- and river-context dependent. In conclusion, our study adds knowledge about effective implementation of mitigation measures and more sustainable operation of HP power plants (Bruder et al., 2016; Moreira et al., 2019; Smokorowski, 2021).
General response of macroinvertebrate to hydropeaking
Consistent with our hypotheses and in accordance to other published literature (Aksamit, Carolli, Vanzo, Weber, & Schmid, 2021; Bruno et al., 2010, 2016; Imbert & Perry, 2000; Miller & Judson, 2014; Schülting et al., 2018a, 2021; Timusk et al., 2016), HP generally led to increased MIV drift intensities compared to base flow and to the upstream-located RF reaches with highest drift during the up-ramping phase and the first part of the peak flow. Furthermore, the analyzed data partially supports the result of Tanno et al. (2021) that an increased drift of MIV also increases their risk of stranding (especially considering the UR phase). Comparison could be biased by the low number of individuals found in our stranding samples as well as the different equations used for drift intensity (our study) and drift density (Tanno et al., 2021). The use of drift density as response variable artificially decreases the effects of HP scenarios with higher flow magnitude and corresponding flow velocities because MIV drift mostly does not increase proportionally to the increase in water volume, leading to a dilution effect (Naman, Rosenfeld, Richardson, & Way, 2017; Pegel, 1980). This effect was already highlighted in the study of Naman et al. (2017) who showed that the parameter ‘drift density’ negatively correlates with the investigated water volume, whereas ‘drift flux’ (i.e., total number of drifting invertebrates) showed opposite patterns. We therefore decided to use the ‘drift intensity’ (because not affected by dilution effects) as a parameter for analyzing the effect of field HP experiments with non-comparable flow conditions.
We found taxa- and river-specific long-term effects on the benthic community under HP (discussed below). However, in contrast to our hypothesis, considering all taxa and all study reaches together, we did not detect a significant reduction in benthic density between RF and HP reaches, even though drift and/or stranding were increased in all HP reaches. We conclude that drift and stranding can contribute to a reduction in benthic density of some taxa, as reported in several studies (e.g., Céréghino & Lavandier, 1998; Leitner et al., 2017; Moog, 1993), but most likely HP further induces organism stress and hampers important processes such as feeding and reproduction. The combination of the HP-induced effects on biota may in consequence lead to a thinning of the benthic community and to a long-term change in species composition.
Effect of HP intensity and associated hydraulic forces on drift and stranding
Flow velocity near the surface (v100 ) and mean up-ramping rate (URmean ) were identified as major determinants for macroinvertebrate drift propensity, indicating that the flow magnitude and amplitude (leading to different flow velocities) as well as the up-ramping rate are important HP-related parameters. A strong relationship between drift and flow-related hydraulic forces such as flow velocity is widely known (e.g., Gibbins, 2016). Recently, Schülting et al. (2021) found that discharge-related parameters, such as mean-column flow velocity at peak flow, primarily affect MIV drift and the importance of the up-ramping rate increases only once certain discharge-related thresholds are exceeded. Following, a reduction of the up-ramping rate may lead to a lower drift since the MIV have more time to seek shelter in the interstices during flow increase (Imbert & Perry, 2000; Schülting et al., 2021; Timusk et al., 2016).
We also found a positive link between down-ramping rate (DRmax ) and stranding propensity, corresponding to findings by Kroger (1973) and Perry & Perry (1986). In line with Tanno et al. (2016), a larger flow ratio (Qpeak/Qbase ) also led to higher stranding, possibly due to increased drift and a larger wetted area. Additionally, we showed that turbidity (NTU ) and water temperature (T ) most likely influence drift and stranding risk. However, this could also be an indirect effect as variations in water temperature and turbidity are influenced by HP operation (e.g., flow ratio). We conclude that the magnitude of drift and stranding propensity is most likely affected by a combination of hydrological and hydraulic factors as well as by their interaction (v100 ,URmean, Qpeak/Qbase, DRmax ) and possibly by behavioral (NTU , T ) responses. Indeed, thermopeaking has been shown to induce behavioral MIV drift of many taxa and often has a synergic magnifying effect with the HP wave (e.g., Bruno et al., 2013). Similarly, water temperature has been shown to influence stranding risk of fish, with greatest effect in winter due to low temperatures (e.g., Halleraker et al., 2003). HP-induced changes of turbidity and their effects on MIV are poorly understood. Though, re-mobilized and re-suspended fine sediments during peak flows are mainly responsible for an increase in the river’s turbidity and can induce fine sediment infiltration and potential clogging (Hauer et al., 2019), thereby exposing MIV to increased abrasion and reduced potential interstitial habitats (Bo et al., 2007; Crosa et al., 2010; Jones et al., 2012).
Taxa- and traits-specific responses to hydropeaking
Long-term HP effects were found only for a few taxa such as Limnephilidae, Nemouridae and Leuctridae, which were abundant in the RF reaches yet strongly reduced in the HP reaches. Limnephilidae further showed short-term response reflected in their high propensity to drift and strand. In contrast, Heptageniidae seemed to be more resistant in respect to short and long-term HP effects.
Limnephilidae are adapted to slow flowing lentic habitats on the substrate surface and, due to their size and shape, they experience high drag (Rader, 1997). Moreover, for case-building Limnephilidae it has been shown that an increased flow velocity decreases their ability to return to the stream bottom from the drift (de Brouwer, Besse-Lototskaya, ter Braak, Kraak, & Verdonschot, 2017). Therefore, this taxon is easily detached passively from its habitat and has poor settling efficiency which might explain its particularly high sensitivity towards short- and long-term HP effects. In line with these observations, other studies suggest that HP often contributes to a decrease in the density of limnophilic taxa associated to lentic habitats and to strong drift (Graf et al., 2013; Leitner et al., 2017; Schülting et al., 2021). Correspondingly, in our study, lentic taxa were more abundant in the RF reaches compared to the HP reaches, confirming the sensitivity of this trait to HP (Schülting et al., 2021).
Heptageniidae showed higher abundances in the benthic samples of the HP reaches compared to the RF reaches (Appendix F in Data S1) and lowest propensity to drift and strand. This emphasized the tolerance towards HP of this family. Similar observations were reported by other authors (e.g., Bruno et al., 2013; Graf et al., 2013; Moog, 1993, Schülting et al., 2018a, 2021). Although different taxa within this family may show varying responses towards HP, they generally prefer lotic habitats on the substrate surface and are characterized by high agility and active settling efficiency (Elliott, 1971) as well as by various morphological adaptations that enhance attachment (streamlined body shape, sharp tarsal crawls, sucker-like ventral gills) and reduce drag (Rader, 1997). Nemouridae and Leuctridae did not show distinctively high drift and stranding following HP, possibly due to relatively thin source populations. Schülting et al. (2021) reported high drift induced by HP for Nemouridae (Nemoura/Nemurella sp.), whereas Leuctridae (Leuctra sp., a taxon associated with the interstices) showed comparably low response, and this family is often not affected by HP (e.g., Graf et al., 2013; Moog, 1993).
Simuliidae and Rhyacophilidae also showed high propensity to drift – Rhyacophilidae also to strand – but both families were more abundant in the benthic samples of the HP reaches compared to the RF reaches. Both taxa are known to prefer lotic habitats and they own adaptations (claws and silk) to prevent passive drift (Rader, 1997). Nevertheless, Simuliidae have been reported to often drift (Bruno et al., 2013, 2016; Imbert & Perry, 2000) and strand (Perry & Perry, 1986; Tanno et al., 2021) following HP. As surface-dwelling taxon, they are naturally exposed to high flow, which can likely explain their high drift propensity, even under natural conditions (Elliot, 1967). Rhyacophilidae are generally associated to the interstices, but as predators, they may also actively move on the surface seeking prey and are thus exposed to abruptly increased hydraulic forces. The cold thermopeaking in the Hasliaare (Figure 2c) might also explain the high (probably behavioral) drift and stranding of Rhyacophilidae (Bruno et al., 2013) in this reach. However, our experimental setup does not allow a separation of active and passive drift. In conclusion, our data suggests that high drift and/or stranding induced by HP flow fluctuations do not lead to distinct density reductions of Simuliidae and Rhyacophilidae. Thus, these taxa are less sensitive towards HP compared to Limnephilidae.
Taxa associated to lentic areas showed reduced benthic density compared to lotic taxa. However, on the long-term, HP did not considerably reduce benthic densities of most taxa, especially of Chironomidae and Baetidae. Even though these taxa showed comparably high drift and stranding responses (Appendix E and F in Data S1), they were also dominant in the respective benthic samples. Matching our results, these two families are often reported as dominant components of the benthic assemblages and drift following HP (Bruno et al., 2010, 2013, 2016; Gibbins et al., 2016; Imbert & Perry, 2000; Moog, 1993; Schülting et al., 2018a, 2021; Tanno et al., 2016, 2021; Timusk et al., 2016) as well to frequently strand (Perry & Perry, 1986; Tanno et al., 2016, 2021). Chironomidae and Baetidae are likely to be represented strongly in the drift and stranding due to the high density and density dependence of drift (Waters, 1972). Accordingly, they showed comparably low propensity to drift and strand and did not show reduced benthic densities under HP, probably due to their morphological and behavioral adaptations to high currents (Chironomidae: rapid colonizers with rapid growth rates and short life cycles; Baetidae: good swimmers with a streamlined body, rapid settling capabilities, frequent intentional drift; Elliott, 1971; Milner, 1994; Naman et al., 2016; Rader, 1997; Wilzbach, Cummins, & Knapp, 1988). Even though, results regarding Chironimidae cannot be generalized since this family consists of many different taxa with different adaptations. Our study suggests that some Chironomid species and Baetidae are resistant and resilient to long-term HP effects and most likely have flexible habitat needs. We therefore conclude that high passive drift and/or stranding, especially of individual-rich taxa, does not necessarily indicate strong HP sensitivity. Yet it must be stated that the benthic community composition in all investigated river reaches has already been significantly affected by the long history and huge variety of anthropogenic impacts on river ecosystems. The presented results should therefore only be interpreted in consideration of the limitations related to the altered source populations.
Some studies suggest that functional descriptions are maintained, even with coarser taxonomic resolution, and it barely influences the discrimination of impact levels (e.g., Dolédec, Olivier, & Statzner, 2000; Gayraud et al., 2003). However, since organisms in our study were identified only to family-level, taxa-specific results should not be over-interpreted. This level of taxonomic resolution inevitably leads to a loss of information, especially for families that have heterogeneous adaptations among the constituent genera, such as Chironomidae. Thus, diverse species/genera within a family can react quite differently to HP or other flow modifications (e.g., RF). Lastly, some taxa were found only in low numbers (e.g., Limnephilidae), limiting the statistical power of some of our results. Moreover, the comparably low number of drifting and stranding individuals can probably be explained by the peak magnitude of our experimental setup, which were lower as normally found in Swiss HP rivers (e.g., Tanno et al. 2016, 2021).
River- and reach-specific responses
MIV responses to HP were not only taxa- and trait-specific but varied widely among the investigated rivers and reaches. Disparities in the direction and magnitude of responses among rivers likely resulted from differences in community composition but also from variability in the HP intensity and river morphological characteristics. Indeed, it is likely that the lower peak flow in the Sitter in combination with its higher morphological heterogeneity compared to the Hasliaare and the Linth lead to lower hydraulic forces acting on the MIVs favoring benthic abundance in concert with a reduced drift and stranding risk. The high drift in the RF reach of the Linth can be explained by the hydraulic forces measured, such as flow velocities and Froude number, which were similar to the ones measured in the HP reach during peak flow. In contrast, in the RF reaches of the Sitter and Hasliaare, they were comparable to the ones measured during base flow (Appendix B in Data S1). This finding underlines that discharge-related hydraulic forces (e.g., flow velocity) acting on the riverbed are most probably the major determinant for MIV drift. Further, the HP-induced removal and mobilization of organic matter from the substrate, which is often associated to passive drift of MIV (Aksamit et al, 2021; Bruno et al., 2016; Miller & Judson, 2014; Timusk et al., 2016), is a another determining factor. The RF reach of the Linth was characterized by high algal cover on the substrate (>50%; Appendix A in Data S1) and showed the highest drifting biomass (ash free dry mass) of FPOM and CPOM compared to the other two rivers (Tonolla, Kastenhofer, Vögeli Kummert, & Gufler, 2020).
In general, these river- and reach-specific responses confirm the importance of structural complexity (such as in the Sitter) for providing lentic habitats and hydraulic refugia at the locale scale as well as the retention of organic matter during high flow disturbance (e.g., Bruno et al., 2016; Hauer et al., 2017; Lancaster, 2000). In structurally complex rivers with moderately sloping bank, velocity and bottom shear stress remain relatively constant (Naman et al., 2017), thereby reducing passive drift. In contrast, fluctuation of the dewatering area is enhanced if the river-bank slope is low, forcing organisms to shift habitats which subsequently increase their risk of stranding.
In addition to river widening and a general increase in structural complexity, morphological measures should aim to reduce substrate-deficit and clogging due to fine substrate infiltration in the dewatering area (often occurring in combination; Hauer et al., 2019), thereby increasing interstitial space as potential hyporheic refugial habitat during HP (Bruno et al., 2010). Further, the reconnection of tributaries with a natural flow and sediment regime, as well as improving the morphology upstream of HP impacted reaches (e.g., in the RF reach), may provide refugial habitats and a connection with benthic source populations for faster recolonization downstream of HP power plants (Aksamit et al., 2021; Bruno et al., 2016; Hauer et al., 2017; Kennedy et al., 2016; Milner, Yarnell, & Peek, 2019).
Overall, our results highlight that different MIV taxa (and traits) vary in their vulnerability and response to flow alteration (HP and RF) and that the river-specific physical context (HP intensity, morphology, distance to potential, source populations) mediate the magnitude and direction of this response. Therefore, next to structural and/or operational measures to reduce HP discharge (and related hydraulic forces) and the ramping-rates, morphological measures should be considered as an essential component of HP mitigation.