Artificial flood reduces fine sediment clogging enhancing hyporheic zone physicochemistry and accessibility for macroinvertebrates

1 Geography and Environment, Centre for Hydrological and Ecosystem Sciences, Loughborough University, Loughborough, UK 2 Department of SurfaceWaters Research andManagement, Eawag (Swiss Federal Institute of Aquatic Science and Technology), Kastanienbaum, Switzerland 3 Department of Aquatic Ecology, Eawag (Swiss Federal Institute of Aquatic Science and Technology), Dübendorf, Switzerland 4 Institute of Integrative Biology, ETH Zürich, Zurich, Switzerland

to consider flow disturbance and restoration in a holistic manner that encompasses the multiple spatial dimensions of connectivity, including the hyporheic ecotone.

K E Y W O R D S
connectivity, environmental flow, experimental flood, flood disturbance, flow pulse, hyporheos, sedimentation, subsurface INTRODUCTION Hydrological variability is widely acknowledged to play a pivotal role in the structuring and functioning of aquatic ecosystems (Karaouzas et al., 2019;Palmer & Ruhi, 2019). However, many rivers across the globe have been impounded for the purposes of navigation, irrigation, hydropower and water-diversion schemes, and it is estimated that 2.8 million dams (constituting reservoir areas >10 3 m 2 ) regulate over 500,000 km of rivers and canals (Lehner et al., 2011). Despite these large numbers, recent studies indicate this could be a significant underestimate (Belletti et al., 2020;Jones et al., 2019). River fragmentation disrupts the high levels of spatio-temporal heterogeneity that support healthy river systems by reducing a rivers' connectivity longitudinally, laterally and also vertically -the latter being a component that is often overlooked in the management of rivers (Boulton, 2007;Krause et al., 2011;Ward, 1989).
One widely associated consequence of flow impoundments and residual flows (a minimum discharge rate that is maintained) is the excessive accumulation of fine sediment (sedimentation; typically referred to as particles <2 mm) associated with a loss of flushing flows (Baker et al., 2011;Sear, 1993). Fine sediment, when present in excessive quantities, is known to have negative implications for biodiversity, affecting all trophic levels from algae to macroinvertebrates and fish (see reviews by Jones et al., 2012Jones et al., , 2014Kemp et al., 2011;Wood & Armitage, 1997). Sedimentation can also lead to clogging of interstitial pore space and habitat homogenization (Larsen & Ormerod, 2010;Mathers, Rice et al., 2017). Loss of pore space can reduce the ability of many macroinvertebrate taxa to access subsurface habitats as large bodied taxa are limited from moving through and into subsurface sediments (Mathers, Hill, et al., 2019;Peralta-Maraver et al., 2019).
In response to the negative ecological effects of residual flows, restoration efforts are being widely advocated globally as a means to balance societal and ecosystem demands, for example by environmental flows (Webb et al., 2018;Yarnell et al., 2020). The term environmental flow is typically used to define the quality, quantity and timing of flows needed to sustain freshwater ecosystems whilst enabling human well-being and livelihoods to be maintained (see review by Acreman, 2016). One method of implementing environmental flows and which is seeing increasing recognition is the artificial release of high flows to restore ecological integrity (Konrad et al., 2011;Olden et al., 2014). By implementing high flows on a regular basis in locations of residual flow, environmental managers can seek to restore flushing flows that are otherwise absent, leading to enhanced habitat quality and subsequent improvements in ecological integrity over time Robinson et al., , 2018. Globally, there are a few long-term flood release schemes; two such examples are the Colorado River below Lake Powell in USA and the River Spöl in Switzerland -the focus of this study. Contemporary objectives of high-flow releases are system dependent (Olden et al., 2014) with artificial floods in the Spöl aimed at stimulating a more natural flow regime to restore habitat conditions for the brown trout fishery Scheurer & Molinari, 2003). Here, artificial floods have also resulted in shifting macroinvertebrate assemblages over time (Robinson, 2012;Robinson & Uehlinger, 2008;Robinson et al., , 2018. More recently, artificial floods have been used as a management tool following sediment flushing with the aim of remobilizing deposited fine sediment and accelerating biological recovery (Doretto et al., 2019).
Despite a significant body of research examining the ecological and morphological implications of artificial floods for benthic habitats (e.g., Melis et al., 2012;Robinson, 2012;Robinson & Uehlinger, 2008), our understanding of the effects of such floods for the vertical dimension of lotic ecosystems remains essentially absent (but see Hancock & Boulton, 2005 for impacts on hyporheic water quality). This represents a significant deficit in our knowledge base of how environmental flows can restore habitat quality and thus ecological integrity. Where fine sediment clogging occurs, the transfer of resources and organisms below the clogged layer may become limited with the hyporheic zone effectively becoming disconnected from surface waters and sediments (Descloux et al., 2013;Hartwig & Borchardt, 2015;Mathers et al., 2014). Importantly, a vertically connected and healthy hyporheic zone is vital to maintain wider ecosystem functioning (Stanford & Ward, 1993) and it can sustain fundamental processes such as carbon processing even without surface flows (Burrows et al., 2017).
In many streams, invertebrate production in hyporheic sediments can exceed that of the benthos (Boulton et al., 1998;Dorff & Finn, 2020;Smock et al., 1992). Furthermore, many taxa and early instars use the hyporheic zone as a refuge from intra-and interspecific predation (Mathers, Rice, et al., 2019;Vander Vorste et al., 2017) and the more stable and predictable conditions present in hyporheic substrates are important for eggs, pupae, diapausing stages of invertebrates as well as fish embryos (Geist et al., 2002;Malcolm et al., 2005). A wellconnected hyporheic zone can act as a refuge during adverse hydrological conditions and has been the focus of a large body of recent work associated with flow intermittency and droughts (Bruno et al., 2020;Maazouzi et al., 2017;Stubbington et al., 2015). The limited number of studies focussing on the use of the hyporheic zone during natural floods At present, vertical connectivity is often neglected in river restoration efforts and evaluations of management schemes despite historical calls (e.g., Boulton et al., 2010), but is essential to fully restore the health of lotic ecosystems and enhance resilience to future climatic change (Lewandowski et al., 2019;Magliozzi et al., 2019). Consequently, it is important to consider the fine sediment content of surface substrates and connectivity of surface-subsurface substrates when evaluating the success of artificial floods -knowledge that is currently absent. To better understand the integrative relationship between artificial floods and the hyporheic zone, we studied the response of a number of abiotic and biotic ecosystem properties to an artificial flow release on the River Spöl in Switzerland. We hypothesized that: 1. The artificial flow pulse would flush fine sediment (particles <2 mm) from the surface, interstitial pore spaces and the hyporheic zone of the gravel-bed river.
2. Removal of fine sediment throughout the vertical profile of the riverbed would subsequently enhance the connectivity of surface and hyporheic habitats, leading to improved physicochemical conditions following the artificial flood.
3. Improvements in vertical pathways and physiochemical conditions following the flood would lead to increased use of the hyporheic zone by macroinvertebrates.
4. Macroinvertebrates would use the hyporheic zone as a refuge during the artificial flood.

Study location
The River Spöl flows in the central Alps across the Switzerland-Italian Prior to regulation in 1970, the Spöl exhibited a natural snowmelt/glacial meltwater flow regime, with high flows in summer and low flows in winter. Periodic floods from heavy rainfall occurred during summer/early autumn with peak discharges between 20 and 60 m 3 s -1 . Average annual flow of the Spöl at the Livigno reservoir fluctuated between 12.5 and 6.6 m 3 s -1 , but post completion of the dam the residual flow was set to an average flow of 1 m 3 s -1 . In the lower flow-regulated section from Ova Spin to Zernez, the location of the present study, the flow is reduced further, being permanently set to 1 m 3 s -1 in summer and 0.3 m 3 s -1 in winter (Scheurer & Molinari, 2003).
Land use within the 295 km 2 catchment (BAFU, 2020) is predominately coniferous forest (Picea excelsa and Pinus mugo) with some grassland and sedges present in the lower floodplain. Climate in the region is continental with high seasonal variation in temperature, but with low precipitation values (Barry, 1992). River sediments originate primarily from dolomitic and calcareous scree from rocky, high gradient valley slopes and locally from remnant glacial moraines (Trümpy et al., 1997).
Bedrock is present in many areas of the riverbed.

Artificial flood and study reaches
The most notable feature of the flow regime of the regulated Spöl vious studies showed this high flow is sufficient to mobilize bed sediments and reduce algae levels with minimal fish mortality Uehlinger et al., 2003). To note, the dam is a deep release reservoir and therefore the thermal regime remains relatively constant (Jakob et al., 2003).
The effects of the 2018 artificial flood were monitored at four locations downstream of Ova Spin reservoir over a 1.5-km section of river.
Sites 1 and 2 were located upstream of the unregulated Cluozza trib- then to assess if hyporheic accessibility for macroinvertebrates was enhanced following the flood (7 days post-flood). Benthic substrate sampling was conducted immediately following the flood as changes here would be instantaneous and was conducted again 10 months later to assess the temporal longevity of flood effects. There were no changes in hydrological conditions during the 10-month period following the flood, with flow during this period being residual flow. More details for each method can be found below.

Surface and total benthic fine sediment content
To quantify the fine sediment content (particles <2 mm) of surface and shallow interstitial (ca. 10 cm) substrates, a resuspension technique using a stilling well was employed (Collins & Walling, 2007;Lambert & Walling, 1988). Sample patches were approached from down- stream and an open-ended stainless-steel cylinder (height 55 cm, diameter 34 cm) was carefully pushed manually into the riverbed until a seal with the substrate was achieved (i.e., there was no flow of surface water through the cylinder). Care was taken to introduce minimal disturbance and prevent the winnowing of fine material. The water depth within the cylinder was measured prior to sampling. To take each surface fine sediment sample, the water within the cylinder was vigorously agitated using a small metal trowel for 60 s to suspend fine sediment on the riverbed surface into the water column. Following this period, a 50-ml vial was inverted in the water column and then turned upright and brought to the surface (sensu Duerdoth et al., 2015). Subsequently, a further 60 s of agitation was undertaken with the spade, including 30 s of digging/mixing the top 10 cm of the bed substrate, with a 50-ml sample again being taken at the end of this period.
This technique provided a sample of both the surface and total benthic fine sediment content from an individual patch. To ensure that spatial variability in fine sediment deposition was accounted for at each site, samples were taken from two visually distinct erosional and three depositional areas (sensu Duerdoth et al., 2015), providing five replicates per site before (1 week) and after the flood (2 days). The fine sediment content was sampled again in June 2019 approximately 10 months following the flood, thus providing three time periods of sampling. Three background water samples were taken from each site for pre-and post-flood samples to account for natural turbidity in the water column; background samples were not taken for June 2019 due to negligible suspended sediment in the water column.
All resuspension samples were returned to the laboratory and refrigerated in the dark till processing took place. Samples were filtered using pre-washed 0.45-μm Whatman glass microfiber filters and routinely analysed for mineral content through oven drying at 105 • C overnight followed by Loss-On-Ignition (LOI) at 550 • C (2 h; Dean, 1974). Finally, water depths within the stilling well were subsequently used to convert laboratory weights to a mass of fine sediment per square metre of riverbed, with values representing the surface and the total benthic fine sediment content.

2.3.2
Hyporheic macroinvertebrates, physico-chemistry and fine sediment samples Hyporheic samples (see below for pumping details) were collected by driving a 1.2-m-long stainless steel Bou-Rouch standpipe into riverbed sediments using a sledgehammer to two sampling depths: 0.25 and 0.50 m (following Davy-Bowker et al., 2006). Samples were taken only from riffles and therefore did not follow the same criteria as the resuspension samples. A minimum of five randomly distributed replicates (six at site 1 and for two time periods at site 2 due to large spatial variability in flow and substrate conditions) were taken from each sampling depth per site (Boulton et al., 2003), and each standpipe was located at least >1 m apart to minimize any influence from adjacent wells The standpipe had an internal diameter of 5 cm and was perforated with 0.5 cm-diameter holes along the distal 5 cm of the pipe. At each sampling point, a total of 5 L (following Boulton et al., 2003) was extracted using a manual bilge pump into a bucket at the fastest consistent rate to ensure sample comparability (Hunt & Stanley, 2000).
Water temperature ( • C), pH, conductivity (μS cm −1 ) and total dissolved solids (TDS, ppm) were measured from each hyporheic water sample using a Hanna Hi 9813-6 multiprobe. Dissolved oxygen content (DO, mg L −1 ) was measured using a WTW Oxi 3310 probe. Following water chemistry measurements, water was passed through a 500-μm sieve to retain macroinvertebrates and fine sediment and preserved in the field in 70% ethanol. For each well, the strength and direction of local hydrological exchange was obtained by measuring the distance from the top of the piezometer to the water level in the piezometer using an electronic dipstick -groundwater level (A), the distance from top of piezometer to the river water level (B) and the distance from top of piezometer to sediment (C). The direction of vertical exchange (delta) was obtained via Equation (1) and the vertical hydraulic gradient (VHG) by Equation (2).
In the laboratory, hyporheic samples were fully enumerated for macroinvertebrates. Further, all fine sediments were retained, ovendried at 105 • C, gently disaggregated using a pestle and mortar and then passed through a series of sieves. Each fraction was weighed to determine the following grain size fractions: total mass <2000, 1000-2000 and 500-1000 μm (sensu Mathers, Hill et al., 2017). All hyporheic invertebrates were identified to the lowest possible taxonomic level, most to species or genus with the exception of some Diptera fami-lies (Ceratopogonidae, Chironomidae, Simuliidae, Empididae and some Limoniidae) and Oligochaeta, which were recorded as such.

Statistical analyses
2.4.1 Surface and total benthic fine sediment content

2.4.4
Hyporheic macroinvertebrate community associations with physicochemical and fine sediment content variables To assess relationships between hyporheic macroinvertebrate composition and the physicochemical parameters, redundancy analysis (RDA) was performed using the 'ordistep' function in 'vegan' (Oksanen et al., 2019). Specifically, pH, conductivity, temperature, DO, delta, mass of fine sediment 1000-2000 μm and mass of fine sediment 500-1000 μm were employed as the physicochemical parameters. Prior to analysis, a Hellinger transformation (Legendre & Gallagher, 2001) was applied to the species-abundance data. A stepwise (forward and backward) selection procedure using permutational-based significance tests (999 permutations) was used to identify factors that influenced assemblages, with only significant variables included in the final model. Final variables were checked for collinearity using the vif function in the 'car' package to ensure that all 'variance inflation factors' were <3 (Zuur et al., 2010).

Surface and total benthic fine sediment content
Total benthic fine sediment content varied significantly by time (χ 2 2,58 = 6.90, p = 0.032), whilst surface fine sediment content was marginally insignificant (χ 2 2,58 = 5.79, p = 0.060). Pairwise comparisons of benthic fine sediment content indicated that immediately following the flood, fine sediment content was significantly reduced (t = 2.63, p = 0.0113). However, there was no significant difference in benthic fine sediment content 10 months post-flood compared to preflood levels (t = −1.23, p = 0.224). Notably following the artificial flood  Table S1).

Hyporheic macroinvertebrate communities
A total of 1198 individuals comprising 17 taxa were recorded in the 113 hyporheic samples. Only one taxon was unique to pre-flood communities (Simuliidae) and three taxa to 7-day post-flood communities    (Tables 1 and S1). Taxa richness demonstrated TA B L E 1 Summary of LMMs (physicochemistry) and GLMMs (ecological) testing the influence of sample time (pre-flood, 1 day post-flood and 7 days post-flood), depth (0.25 and 0.50 m) and their interaction on a number of hyporheic physicochemical parameters, taxa richness, total abundance and abundance of selected individual taxa. Significant terms are emboldened

Hyporheic macroinvertebrate community associations with physicochemical and fine sediment content variables
When hyporheic macroinvertebrate community composition was tested against physicochemical parameters, the 0.25 m community model was significant (F = 1.50, p = 0.043) and accounted for 15.6% of variance on the first two axes. The pH was found to significantly influ- F I G U R E 6 Redundancy analysis (RDA) of 0.25 m hyporheic communities from the Spöl associated with an artificial flood in September 2018. Only significant environmental parameters (at 90% confidence level) influencing faunal data are presented. Species shown are those that plot specifically within flood groups evidence to support our first hypothesis. Surface fine sediment content also demonstrated a reduction but this trend was marginally insignificant (p = 0.06). Our results, however, provide further evidence that artificial floods can successfully be employed to flush excessive fine sediment from gravel riverbeds (Doretto et al., 2019).
The implementation of artificial floods should, however, be conducted on a regular basis, as we found evidence to suggest that fine sediment content of surface and shallow benthic substrates (ca. 10 cm) demonstrated some increases in the 10-month period following the artificial flood. Pre-flood concentrations also highlight the importance of natural flushing flows in maintaining interstitial pore space. Greater quantities of fine sediment were present pre-flood at the two sites located upstream of the unregulated tributary, Cluozza, compared to the two sites downstream that only occasionally experience periodic floods originating from this tributary. Despite being located downstream of a glacial river with high amounts of suspended sediments during natural floods Uehlinger et al., 2003), these two downstream sites (sites 3 and 4) had a lower fine sediment content in both surface and benthic sediments prior to the artificial flood.
In the River Spöl, use of the hyporheic zone by macroinvertebrates prior to the artificial floods was significantly limited and was most likely associated with the large volumes of fine sediment present in interstitial pore spaces as well as subsurface sediments. Fine sediment con- Despite alterations in the physicochemical conditions of both shallow and deep hyporheic sediments, alterations to the hyporheic community were heterogeneous in nature. We found that shallow hyporheic communities 0.25 m below the riverbed displayed altered composition after the artificial flood, whilst communities at 0.50 m displayed comparable composition to pre-flood communities (although some community metrics were altered at 0.50 m depths). Given the differences in depth, it is reasonable to assume that communities at 0.25 m would be the first to react to the changing physicochemical conditions and the restoration of interstitial pathways. Moreover, although individuals from most invertebrate groups and all insect families have been recorded in the hyporheic zone, few of these have been from depths exceeding 0.50 m, suggesting there may be an upper limit on substrate depth use (Boulton, 2000). As we only conducted ecological sampling 7 days post-flood, we are unable to ascertain if community composition shifts were evident in deeper subsurface substrates as length of time since flood increased. We are also unable to directly ascertain the duration of time in which the alterations to hyporheic physicochemical conditions and ecological communities persisted. However, we believe these beneficial effects would have diminished after a few months due to the re-accumulation of fine sediment 10 months following the flood in benthic substrates. This, in addition to the observed pre-flood conditions in both hyporheic physicochemical and ecological health despite an artificial flood being released 13 months prior (peak discharge of 35 m 3 s -1 ), suggests that regular floods are required to reinitiate/maintain surface-subsurface pathways. This finding is in line with previous research conducted in the Spöl that has documented that regular artificial floods (1-2 year) are required to maintain the benthic ecological benefits of flood pulses . Moreover, Hancock and Boulton (2005) documented no lasting effects on hyporheic water quality 49 days following an artificial flow pulse. Further research is required to understand the temporal longevity of artificial flood effects on the hyporheic zone, including other resources that may be more important in the longer term and which were not monitored here (such as organic matter).
Despite the changes in community composition at a depth of 0.25 m, the implications of the artificial flood for hyporheic communities were highly variable in space, with some sites and depths demonstrating no changes in abundance or taxa richness following the flood, whilst others demonstrated an increase (hypothesis three). Hyporheic communities are widely acknowledged to be highly variable in both space and time associated with vertical hydrological exchange, hydrological conditions, sediment composition and the provision of resources (Dole-Olivier & Marmonier, 1992;Dunscombe et al., 2018). The few studies investigating the response of hyporheic communities to natural floods and spates have similarly found that communities reacted in a patchy fashion with the hyporheic zone representing a variable refuge dependent on the direction of vertical hydrological exchange (upwelling or downwelling), substrate stability and the magnitude of the discharge event (Dole-Olivier & Marmonier, 1992;Dole-Olivier, 1997). However, we did not detect any significant associations between hyporheic communities and the direction of vertical hydrological exchange, conductivity or DO, the latter two being parameters shown to be associated to a change following the flood. We did, however, detect an association with pH and fine sediment in the fraction 500-1000 μm for shallow subsurface communities, but neither of these parameters were related to the changing community composition after the artificial flood. This result suggests another physicochemical variable not recorded in this study may have influenced the change in 0.25 m hyporheic community composition -that of riverbed instability.
Visual observations indicated that the degree of sediment deposition and erosion in the Spöl was highly variable in space and magnitude, with some locations seeing scour of up to 0.50 m in depth following the artificial flood. Mürle et al. (2003) similarly recorded erosion rates of 0.50-1.50 m from floods of varying magnitudes in the Spöl. The evidence of highly erosive substrates can be seen, in particular, at site 2 where hyporheic abundances and especially the abundance of G. fossarum were dramatically reduced. Gammarus sp. has been shown to exhibit active vertical migration in response to streambed drying events (Stubbington et al., 2011;Vadher et al., 2018) and as such it was anticipated that this taxon would seek refuge in the hyporheic zone during the flood (hypothesis four). However, it is highly likely that the low substrate stability in the Spöl prevented this activity (Matthaei & Townsend, 2000;Rempel et al., 1999), and they sought alternative refuge in channel margin habitats.
We did see some evidence of hyporheic refuge-seeking behaviour by Leuctra sp. 1 day after the flood, a taxon that is widely associated with interstitial and subsurface substrates (Dorff & Finn, 2020). This taxon also demonstrated increased abundances in deeper 0.50 m substrates 7 days following the artificial flood, a pattern also matched by taxa richness, suggesting that pore space had become available for individuals to use following the artificial flood. Importantly, improvements in the physicochemical conditions (including DO) and re-establishment of interstitial pore space show that the hyporheic zone in the Spöl would be accessible by macroinvertebrates to use as a refuge under the increasing prevalence of drying events, thereby potentially enhancing the resilience of the ecosystem (Bruno et al., 2020;Van Looy et al., 2019).
Fragmentation and flow regulation remain an ongoing threat to the integrity of river ecosystems globally (Januchowski-Hartley et al., 2020;Reid et al., 2019). In particular, by limiting longitudinal connectivity and bedload transport through the construction of dams, riverbeds are likely to become clogged with fine sediment that subsequently influences vertical connectivity and effectively disconnects the hyporheic zone. Our work, which focused on this neglected vertical dimension and which significantly advanced our scientific understanding, provides much needed evidence that artificial floods can be an effective means to restore flushing flows, thereby enhancing spatial heterogeneity in sediment composition. This flushing of fine sediment can influence and enhance physicochemical conditions within the hyporheic zone, restore interstitial pathways and ultimately influence the transfer of organisms within the hyporheic zone -even within a river that is taxa poor. The implementation of artificial high flows as a management tool should, however, be conducted on a regular basis, as we found that surface and shallow benthic fine sediment deposits had begun to accumulate again over 10 months following the artificial flood in line with similar findings for benthic taxa requiring regular artificial floods (one or two annually) to maintain the ecological benefits .