Introduction

Hydropower is a highly efficient electricity source, responding almost instantly to the market demand by changing the turbine discharges. Due to its flexibility and affordability, hydropower generates approximately 16% of worldwide electricity (IHA, 2022). However, the consequent artificial pulsed-flows associated with hydropower production, i.e., hydropeaking, disturb the biological, morphological, and hydrological processes of the downstream river system (Schmutz et al., 2015; Alp et al., 2020; Sukhbaatar et.al., 2020), impairing its ecological integrity. Fish represent one of the most impacted biological groups by such events, often experiencing negative effects such as downstream displacements, stranding, obstruction of migration routes, and degradation (Marmulla, 2001). These, together with habitat loss, will affect key life-cycle needs, namely migration, spawning, feeding, and sheltering (Birnie-Gauvin et al., 2020).

In recent years, the amount of research addressing hydropeaking has grown substantially (Almeida et al., 2002; Scruton et al., 2003, 2005; Alexandre et al., 2015; Costa et al., 2019a, 2019b; Boavida et al., 2020a, 2020b; Oliveira et al., 2020; Boavida et al., 2023) and have been mostly focused on salmonids. However, few studies have proposed habitat enhancement solutions to reduce its impacts based on fish behavior (Costa et al., 2019a, 2019b). For example, Iberian barbel (Luciobarbus bocagei, hereafter L. bocagei) was tested using lateral refuges as a flow-refuge against simulated hydropeaking (Costa et al., 2018, 2019a, 2019b; Moreira et al., 2020). T-shaped (Vehanen et al., 2000) and lateral refuges (Ribi et al., 2014) were also subjected to simulated hydropeaking by brown trout (Salmo trutta trutta). These studies, conducted at indoor and outdoor flumes, demonstrated that changing particular features in artificial habitat design can trigger distinct behavioral changes, dictating their efficiency.

However, there is still a gap in existing solutions to protect cyprinids, which are the most abundant fish family in European rivers, overall, as well as in Iberian. According to Goodwin et al. (2014), fish behavior may be influenced due to their capacity to feel hydraulic signals in their immediate vicinity that could be linked to their migrations and aggregation. In fact, Katopodis & Gervais (2016) mentioned that, normally, the lower velocity at the entrance of the channel will reduce the fish attraction. Katopodis et al. (2019) also referred that fish behavior and physiology, as well as hydraulic conditions, are directly linked to their capacity to be attracted. Additionally, the insertion angle of flow-refuges in relation to the flow due to the water exchange between them, is an important factor for their efficiency (use by fish), although it is still poorly studied (Ribi et al., 2014).

The main goal of the present study was to evaluate the use of two types of flow-refuge by L. bocagei at an experimental flume, during simulated pulsed-flow conditions on movement behavior and stress physiology. An integrated approach was conducted where behavioral metrics and blood physiology were quantified, in addition to the characterization of the hydraulic environment. The two flow-refuges differed in the insertion angle to the flume wall, i.e., 45° and 70° and they were considered based on previous studies (e.g., Moreira et al., 2020; Boavida et al., 2023) to improve the fish capacity on the flow-refuge utilization. The following hypotheses were tested: (i) the flow-refuge use, assessed by the metrics frequency of use and permanence time, will differ between the different insertion angles during base- and pulsed-flows; (ii) the different flow-refuge insertion angles will trigger fish physiological adjustments during base- and pulsed-flow events; (iii) the two insertion angles will create distinct hydraulic conditions in the vicinity of flow-refuges.

Methods

Ethical statement

All procedures involving animal manipulation, from capture in their natural environment to holding in the laboratory, were carried out according to European norms CEN EN 14011:2003 (CEN, 2003), and Portuguese legislation (Decreto-Lei, 2013, 2019) and guidelines (INAG, 2008). For this, a permit was issued by the Portuguese licensing entity, the Instituto da Conservação da Natureza e Florestas (ICNF). All the researchers involved with the direct manipulation of fish are authorized to carry out and design procedures and scientific projects involving animal experimentation.

The EcoPeak4Fish project is authorized by the governmental organization Direção Geral de Alimentação e Veterinária (DGAV). The Laboratory of Hydraulics of Instituto Superior Técnico is certified by DGAV as a bioterium to perform trials with live fish for conservation and scientific purposes while complying with ethical and legal rules. No fish were sacrificed for this study. The principle of replacement, reduction, and refinement (3Rs) was followed while developing the experimental protocol.

Experimental facilities

The experiments were conducted at an indoor flume located at the Hydraulics Laboratory of Instituto Superior Técnico, University of Lisbon (Fig. 1).

Fig. 1
figure 1

A Top view of the flume with the 45° flow-refuge installed and the scheme of the 45° flow-refuge. B Top view of flume with the 70° flow-refuge installed and the scheme of the 70° flow-refuge

Full size image

The flume, with a rectangular cross-section (8.0 m long, 0.7 m wide, 0.8 m high), is built on a steel frame with glass panels on both sides and it has a mobile floor that allows the placement of different structures. The usable area of the flume for fish was 6.5 m long × 0.7 m wide, delimited upstream and downstream by perforated metallic panels. The maximum discharge was 60 L/s, with a maximum water depth of 0.22 m, controlled by a downstream flap gate and an upstream sluice gate.

Two plywood flow-refuge types were constructed, differing in the angle of insertion to the flume wall, i.e., 45° or 70° (Fig. 1). For each type, two aligned flow-refuges were installed, with a 2.30 m spacing between each other. Both flow-refuge types were 0.25 m wide and 0.20 m high, differing in length, i.e., 0.45 m and 0.29 m for the 45° and the 70°- insertion angles, respectively.

Fish capture and holding

Largely spread among the northern and central basins in Portugal (Oliveira et al., 2020), the L. bocagei was the selected species to evaluate the use of two types of flow-refuge at an indoor flume. It is an endemic Iberian potamodromous cyprinid, occurring in most of the Iberian rivers, with a widespread presence in the middle and lower rivers reaches. L. bocagei is a rheophilic and other bottom-oriented species (Branco et al., 2013; Costa et al., 2019a) have their diet adapted according to their necessity, primarily consuming plant material and benthic invertebrates. Nevertheless, out of the spawning season, the adults tend to be limnophilic, showing their preferences for lower water velocities and deep available refuges (Costa et al., 2019a, b). Therefore, for the present study, young adults of L. bocagei were used because of their capacity to support fast-moving flow conditions and such phase represents the most susceptible phase of their life.

Fish sampling took place in the Sorraia River (39.011376° N, − 8.357126° W), a tributary of the lower Tagus River (central Portugal). The sampling site is not affected by artificial pulsed-flows, which makes it an appropriate site to capture fish that will be subjected to simulated pulsed-flows. Fish capture occurred on the 5th and 12th of November of 2021, through electrofishing performed with a low-voltage (400 V) unit (Hans Grassl IG-200). This period coincides with the upstream movements from this and other potamodromous species for refuge, feeding, and exploratory purposes (Benitez et al., 2015), after the reproductive season (Santos et al., 2011). In total, 125 young-adult L. bocagei were captured (Table 1), never exceeding 80 individuals per sampling occasion. After the sampling procedure, fish were placed in a permanently aerated tank (Linn Thermoport 190 L) and transported to the experimental facilities. Once in the laboratory, fish were allocated in two different 900 L tanks covered with a ventilated lid, with continuous aeration, and water filtered biologically (Fluval Canister Filters FX5 and FX6). Clay roof tiles and PVC pipes were placed at the bottom of each tank to provide sheltering areas and minimize stress.

Table 1 Summary of the experiments conducted depicting trials, number of fish tested (n), tested discharges, trial duration (min), type of flow-refuges (º), and fish total length (cm) (± SD)
Full size table

Fish acclimated under these conditions, at room temperature and natural photoperiod for a 72 h period. Using a multiparametric probe (YSI 556 MPS), the tanks water quality (mean ± SD) was monitored daily for pH (7.47 ± 0.17), conductivity (200.8 ± 17.38 μS/cm), dissolved oxygen (8.23 ± 0.59 mg/l) and temperature (19.75 ± 1.6 °C), and twice a week, for nitrites (0.35 ± 0.24 mg/l), and ammonia (0.01 ± 0 mg/l) using colorimetric methods. The water quality in the flume was monitored twice a day for pH (7.59 ± 0.11), conductivity (141 ± 9.47 μS/cm), dissolved oxygen (7.99 ± 0.82 mg/l), and temperature (21.34 ± 1.38 °C). A commercial diet for benthic fish species was administrated every night to avoid food deprivation and minimize stress. Feeding took place during the acclimation period and stopped 24 h before the experiments (Costa et al., 2019b). All fish were returned to their natural habitat after each week of experiments.

Experimental trials

Two flow events were tested, a base-flow (BF) and a pulsed-flow (PF). The BF event consisted of a continuous 7 L/s discharge, while the PF consisted of a continuous 60 L/s discharge, both events lasted 40 min. Before each event, the fish acclimated to the flume for 30 min at 7 L/s.

To create a PF event, the upstream gate was closed to fill the flume reservoir, and then, manually, the discharge was controlled until it reaches the constant 60 L/s throughout trials (maximum discharge). Afterward, the gate was rapidly opened to 12° to release the maximum flow until reaching the target flow. For the 60 L/s discharge, and with the downstream gate fixed at a 82° angle, the maximum water level was 0.22 m for both flow-refuges. The combination of flow-refuge types (i.e., 45°, and 70°) with the flow events (i.e., BF and PF) resulted in four different sets of trials: BF45, PF45, BF70, and PF70 (Table 1), each one being replicated five times, giving a total of 20 trials. For each trial, a school of five L. bocagei was tested, and each fish was tested only once. The experimental design was outlined in accordance with previous studies (Amaral et al., 2016; Costa et al., 2018; Moreira et al., 2020), which found the sample size to be representative, thereby the need for a large number of captures and, complying to the 3R’s policy (Brønstad & Berg. 2011).

To evaluate fish use of each flow-refuge insertion angle (i.e., 45º and 70º) to attract L. bocagei and provide shelter under pulsed-flows, frequency (absolute) and time of permanence were quantified. The absolute frequency corresponded to the number of times that a single fish (I), or a group of fish—two to five individuals (G) were successfully attracted to the flow-refuge. A successful use was considered when an individual or a group of individuals were observed in the downstream area outside the flow-refuge (30 cm) (I_Down; G_Down) or inside (I_Ins; G_Ins) the flow-refuge.

The Down and Ins permanence time was quantified considering the time spent on same area/distance as for the absolute frequency. The frequency of flow-refuge use was visually accessed by three trained observers. The permanence time was registered by three GoPro cameras that were installed facing the flume glass walls, covering the whole flume area (downstream, middle, and upstream reaches). The recordings were analyzed using the BORIS software (Friard & Gamba, 2016).

Physiological responses

To analyze potential physiological adjustments between flow-refuge types (i.e., 45º and 70º) and flow events (BF and PF), the concentrations of blood glucose and lactate were measured. Both are secondary-level responses of the stress axis and are expected to increase when the organism is subjected to an external disturbance (Pankhurst, 2011). Lactate is particularly interesting to address the impacts of pulsed-flows because it is associated with fatigue (Costa et al., 2019a, b), whereas glucose is a proxy for stress (Costa et al., 2018). To collect blood samples, fish were dip-netted from the flume after each trial and transferred to a bucket with continuously oxygenated water. Immediately after, the fish were transferred to a V-shaped structure in a supine position and, a blood sample (0.1–0.5 ml) was collected via caudal puncture using 23 G or 25 G pre-heparinized needles (Costa et al., 2018). This procedure was completed in less than 3 min, assuring that it did not influence primary-level responses (e.g., cortisol). The glucose and lactate levels were immediately measured using the portable meters Accu-check Aviva (Roche) and Lactate Plus (Nova Biomedical UK) respectively (Costa et al., 2019b).

Flow-field characterization

The flow-field characterization was performed using an Acoustic Doppler Velocimeter (ADV; Nortek-AS Vectrino 10 MHz). The ADV allows orthogonal measurements (x, y, z) through a four-beam down-looking probe attached to a fixed structure (Lohrmann et al., 1994; Baladron et al., 2021). Velocity measurements were conducted following a grid of 290 and 288 measuring points for the 45° and 70° flow-refuges respectively. The perpendicular direction had a spacing interval of 0.10 m, with varying spacing in the longitudinal direction (0.10 m in the refuge area and 0.15 between areas). The selected grid allowed a more detailed characterization of the flow-field over the flow-refuge area.

The velocity measurements were carried out at each point using a 100 Hz sampling rate and a sampling period of 180 s for BF and 90 s for PF, which have been considered adequate (Buffin‐Bélanger & Roy, 2005; Silva et al., 2011).

A two-way distance-based multivariate analysis with permutations (using the Euclidian distance and 999 permutations) (Oksanen, 2007) was applied to analyze if the tested trials (i.e., BF45, BF70, PF45, PF70) triggered an effect on the frequency and permanence time of the flow-refuge use. This method accepts small sample sizes (Walters & Coen, 2006) and it does not require assumptions of parametric tests (Anderson, 2001), being suitable for continuous and factor predictors (Oksanen, 2007). The R-package vegan (Oksanen, et al., 2017) was used for this analysis. When an effect was detected a Kruskal–Wallis analysis with a posthoc Nemenyi test (pairwise contrasts between trials) (Pohlert, 2014) was conducted for each metric of flow-refuge use and permanence time. To verify if there were significant physiological adjustments in L. bocagei among trials a Kruskall-Wallis followed by a posthoc Nemenyi test was conducted (Pohlert, 2014). These analyses were performed using R-Package PMCMR (Pohlert, 2014).

Water velocity data were interpolated between the points, to go from a discontinuous to a quasi-continuous. The Ordinary Kriging spatial interpolation, through Krige function from the GSTAT package (Gräler et al., 2016) was chosen, due to the best results obtained. Velocity maps were completed via the plot.grid function (R Core Team, 2021). All treatment and pre-processing of the data were performed in the RStudio software (R Core Team, 2021). The average speed ADV was plotted on a hydraulic map with each point having its reference. The calculation of the mean value was performed for each measurement and this difference was considered in the script, calculating the negative mean speed for each measurement of that line.

Results

Flow-refuge use

Overall, the absolute frequency of fish use (sum of mean ± sum of SD) at the 45º flow- refuge (96.35 ± 56.77) was higher than the 70º (69.00 ± 28.90), when pooling both “Downs” and “Ins” fish use data.

The multivariate analysis showed a significant effect of the flow on the flow-refuge use (F = 5.235; P = 0.008). There were significant differences in the frequency of individual use, inside (I_Ins; χ2(3) = 9.7968, P = 0.02) and downstream (I_Down; χ2 (3) = 11.159, P = 0.01) of the area of the flow-refuge, among events. Rank comparisons demonstrated that I_Ins was higher (25.20 ± 15.04) for fish when subjected to PF45 in comparison with BF70 (05.40 ± 2.19) (P = 0.035) (Fig. 2A). Similarly, the downstream location of the flow-refuge used by L. bocagei was higher individually during PF45 (14.20 ± 06.14) in comparison with BF45 (03.75 ± 04.99) and BF70 (03.40 ± 01.67) (P = 0.041 and P = 0.047, respectively) (Fig. 2A). The group use of the flow-refuge, did not show a significant effect in both, Inside (G_Ins; χ2(3) = 6.4573, P = 0.09) and Downstream (G_Down; χ2(3) = 6.4443, P = 0.09) locations of the flow-refuge.

Fig. 2
figure 2

A Frequency use for the 45º and 70º approaching angle structures use by L. bocagei (n = 100) for base-flow (BF) and pulsed-flow (PF) trials at the two structures areas: Downstream (Down_I and Down_G) and Inside (Ins_I and Ins_G) approaches (I: Individual; G: Group) (B) Flow-refuge time use, in total, per trial

Full size image

The permanence time (minutes) in refuge within the four trials tested is presented in Fig. 2B.

Similar to the frequency of use (Fig. 2A), the permanence time in the flow-refuge for the L. bocagei on the 45º flow-refuge (10′12″ ± 13′08″) was higher than 70º (09′08″ ± 09′52″) (Fig. 2B). On average, the individual use, on the inside area, fish were present more frequently on the 45º flow-refuge (05′27″ ± 09′00″) than 70º (04′25″ ± 07′34″) flow-refuge (Fig. 2B) (Table 2). Group use also showed the same results where 45º flow-refuge (05′21″ ± 06′41″) was higher than the 70º (04′42″ ± 04′54″). The statistical analysis did not show a significant effect of the flow on permanence time (F = 2.7747; P = 0.053).

Table 2 Average time (minutes ± SD) spent by L. bocagei per location of the flow-refuge (I_Down, G_Down, I_Ins, G_Ins), as an individual (I) and as a group (G)
Full size table

Physiological responses

The mean (± SD) levels of blood glucose in L. bocagei were 49.27 ± 12.85 mg/l (BF45), 42.10 ± 11.09 mg/l (PF45), 47.32 ± 12.56 mg/l (BF70) and 51.73 ± 18.72 mg/l (PF70) (Fig. 3). Even though glucose levels were higher in PF70, such differences were not statistically significant among trials (χ2 (3) = 3.8787, P = 0.2749). The mean (± SD) levels of blood lactate in L. bocagei were 3.88 ± 1.04 mM (BF45) 3.68 ± 1.35 mM (PF45), 3.00 ± 1.14 mM (BF70), and 3.64 ± 1.06 mM (PF70) (Fig. 3). Lactate levels did not indicate significant differences (χ2 (3) = 5.7665, P = 0.1235).

Fig. 3
figure 3

Boxplots with the variation of blood glucose (mg/dL) and lactate (mM) levels for L. bocagei subjected to trials: BF45 and BF70, which represent the base-flow event (7L/s) for the 45° and 70° flow-refuge, respectively, and PF45 and PF70, which represent the pulsed-flow (60 L/s) events for the 45° and 70° flow-refuge respectively. The asterisk corresponds to the mean value of the physiological indicator of each trial

Full size image

Flow-field

The maximum velocities for pulsed-flow (60 L/s) found were 0.914 m/s (PF45) and 0.781 m/s (PF70). Those velocities were found in the narrowest area, i.e., between the edge of the flow-refuge and the flume wall (0.35 m for the 45° and 0.42 m for the 70º) (Fig. 4). Downstream and inside areas of the flow-refuge, are characterized by lower velocities, in both configurations. As expected, the lowest velocities were observed during BF trials and, they were found in the immediate area downstream of the flow-refuge (i.e., less than 0.30 m far from the flow-refuge), for the two tested discharges (BF45—0.001 m/s and BF70—0.002 m/s) and inside of the flow-refuge 0.000 m/s

Fig. 4
figure 4

Flow velocity fields for the tested discharges: 7 L/s and 60 L/s with different approaching angles A 45º flow-refuge, 7 L/s B 45º flow-refuge, 60 L/s C 70º flow-refuge, 7 L/s and D 70º flow-refuge, 60 L/s

Full size image

Discussion

The efficiency of flow-refuges varying in the approaching angles to the flume wall to assess their attraction potential was evaluated through the quantification of behavioral and physiological responses of L. bocagei, during simulated base- and pulsed-flow conditions.

An experimental design was conducted through a multidisciplinary approach, combining behavioral and physiological responses, while correlating them with the flow-field. The approach for the flow-refuge use (frequency and permanence time) by L. bocagei differed between the different insertion angles during base- and pulsed-flows. The different flow-refuge insertion angles triggered physiological adjustments in BF and PF events.

The flow-refuge use indicated that fish behavior changed according to the flow conditions and the insertion angle. L. bocagei used the inside area of the flow-refuge more often than the downstream area for both the BF and PF conditions. Such evidence was more prominent for the 45° insertion angle than for the 70° one. Although L. bocagei used more frequently these flow-refuges (inside and downstream), this distinction was clear in the PF45 compared to the BF trials. The individual frequency of use was higher during BF45 and PF45 flow-refuge than the BF70 and PF70. Such a result was more evident during pulsed-flow events. A similar result was found by Costa et al. (2018), where, during simulated discharges conditions, L. bocagei used the flow-refuge individually more often than in groups. Although dissimilar results between trials were found in the group use frequency, the statistical analysis did not demonstrate significant differences, as mentioned before. This may be understood as a non-disruption of group behavior, which means that the simulated pulsed-flow conditions may not have affected the group behavior. In fact, schooling behavior could generate benefits, such as the frequency increase of the tail beat by the leading fish (Liao, 2007) or the reduction of the energy cost during spawning season (Standen et al., 2002; Wang & Chanson, 2018), or under turbulent flows (Enders et al., 2005). Also, according to Costa et al. (2019a, b) findings, the frequency of use in group could be explained by the fact that lower discharge and pressure magnitudes can generate stability in the group behavior. Even though the time spent within the flow-refuge areas was longer, the statistical analysis did not show any significant difference (P = 0.053).

The absence of physiological adjustments among trials, together with the characterization of the flow-field, suggests that the conditions created were not harsh enough to trigger a stress response. The connection between flow variability and stress response is difficult to identify (Costa et al., 2017) and may be justified by the type of external stimulus or the severity of the stressor (Pankhurst, 2011). Thus, the presence of both types of flow-refuge and the created hydraulic conditions favored flow-refuging (Flodmark et al., 2002; Krimmer et al., 2011; Costa et al., 2019a, b) and avoided physiological adjustments.

According to our experimental settings, the two insertion angles create distinct flow-field conditions, with higher velocities found in the narrowest distance (distance between the edge of the refuge and the channel wall) due to the refuge locations. Thus, it is possible to select the most efficient flow-refuge based on its potential to attract L. bocagei. This may be explained by the fact that the 45º flow-refuge, due to the insertion on the flume channel, has the narrowest distance (0.35 m) when compared to the 70º flow-refuge (0.42 m), creating then higher velocities, and leading to a higher attraction. This shortest distance seemed to be sufficient to trigger the rheophilic behavior of L. bocagei and increase the attractiveness of the 45° flow-refuge. According to Goodwin et al. (2014), fish detection of hydraulic signals could trigger aggregation and migration behavior. A similar result was found by Ribi et al. (2014), in their experimental study in an indoor flume simulating hydropeaking event equipped with a lateral refuge, when they compared the capability of brown trout juveniles to find different entrances installed laterally in their facility. According to their results, as long as they have flow cues toward the entrances, fish could be attracted.

The presence of the flow-refuges, under different flow conditions (BF and PF) created a differential use of flow-refuges by fish. The ADV results demonstrated that the flow-field was changed by such structures (i.e., different velocity ranges within the flume).

The collected data also allowed identifying the best flow-refuge for fish. The 45° flow-refuge demonstrated to be the most suitable flow-refuge during pulsed-flow conditions throughout the study, as it displayed the highest flow-refuge use and permanence time. Thus, considering the effects of pulsed-flows (e.g., loss of suitable habitat areas, Moreira et al., 2020), flow-refuges that create water deflection given their insertion angles are highly recommended to create suitable flow-refuging areas.

As demonstrated in this and other studies (Costa et al., 2018, 2019a, b; Moreira et al., 2020) fish can benefit from these habitat enhancement solutions. However, each approach is unique, thus, a previous analysis should consider the river morphology and hydrodynamic patterns (Schmutz et al., 2015; Costa et al., 2019a, b), as well as the fish’s ecological requirements. Thus, such an instream structure may be considered as a potential indirect mitigation measure to artificial pulsed-flows due to hydropower production.

Conclusion

This study provided novel insight regarding the attraction efficiency of two alternative flow-refuges use for the potamodromous cyprinid L. bocagei. The two approaching angles tested (i.e., 45° and 70°) created distinct flow-field conditions at the entrance of the flow-refuges and evidenced that the flow-refuge with the lowest insertion angle to the flume wall (i.e., 45°) was more used by L. bocagei in comparison with the widest angle (70°). Flume studies may not replicate natural conditions but provide knowledge regarding fish interactions under a controlled environment, which are essential for the development of more efficient solutions and their application in the field. These indirect mitigation measures can be used by cyprinids for flow-refuges during pulsed-flows associated with hydropower production. Future studies should consider extended tests with different ramping rates, different ranges of flow rates (higher ones), or even other fish guilds (e.g., water-column species).