Murray cod and modern fish screens: influence of water velocity and screen design on the entrainment and impingement of larval and young-of-year fish at water offtakes

Experimental design

Three experiments were conducted in a laboratory flume to examine the entrainment (propensity to be sucked through a screen or into the pump inlet) and impingement (propensity to be held against the screen) of larval and YOY Murray cod at a simulated offtake structure.

Each replicate in all experiments were run at 340 lx, measured using an ExTech Easy View 30 light meter. The water temperature for Experiment 1 was 17.9°C and 0 NTU turbidity. For Experiments 2 and 3 the water temperature was 21.3°C and turbidity was 0.3 NTU.

Fish collection and holding

Larvae and YOY Murray cod were obtained from the Narrandera Fisheries Centre Hatchery (NSW, Australia). Egg trays were removed from spawning boxes situated within earthen ponds and transferred to holding tanks within the hatchery. Larvae emerged after 7–10 days and were kept in aerated trays (50 cm long, 50 cm wide and 15 cm deep) before experimentation. Up to 17 days post-hatching, Murray cod larvae were sustained on their own yolk reserves, after which time this was supplemented with daily feeds of newly hatched nauplii. For YOY juvenile Murray cod used in the experiments, larvae were stocked into a large earthen pond (3600 m² and ~3 ML) at 3 days post-hatching, where they fed naturally on the available plankton. At ~28 days post-hatching, 24 h before Experiment 2 commenced, Murray cod YOY were collected from the pond with a hand net.

Larvae and YOY were kept alive in 900-L troughs with flow-through river water at ambient temperatures for the duration of the experiments. Within the troughs, replicate groups of 10 fish were acclimated in 500-mL perforated plastic holding containers (AquaOne Midi Float Guppy tank) for 24 h prior to the experiment. Feeding was ceased 24 h prior to experimentation. Water quality parameters including temperature, dissolved oxygen, conductivity, and turbidity were monitored throughout the experiment using a water quality meter (Horiba Multi Water Quality Checker U-50 Series) to ensure conditions remained within an acceptable range (Barker et al. 2009).

The experiment was conducted under animal research authority number 14/02 granted by the Primary Industries (Fisheries) Animal Care and Ethics Committee and scientific collection permit P01/0059(A).

Experimental flume apparatus

The swimming flume apparatus used in this experiment was similar to the apparatus described in Stocks et al. (2019) and Boys et al. (2013a). The apparatus was comprised of an open top, partially submerged rectangular swimming flume constructed from a perforated aluminium plate (3 mm holes with 30% porosity; Fig. 1b). A removable, cylindrical fish-holding cradle constructed from the 3-mm perforated aluminium plate wrapped in woven fibreglass lattice mesh (0.28-mm diameter wire and 1.22-mm aperture) was used to allow easy removal of fish from the flume after each replicate treatment (Fig. 1). The fish-holding cradle was butted against the removable fish screen. The flume was positioned within the centre of a 2.5-m diameter circular flow-through tank filled to a depth of 0.70 m (containing ~1800 L). Water was drawn from the surrounding tank through the perforated mesh flume by a pump capable of delivering 1950 L min⁻¹. Water drawn from the tank through the flume was recirculated into the holding tank through a network of 100-mm high-pressure plastic polymer piping. Water flow was controlled by two ball valves fitted within the pipe network allowing for manipulation of AV values by 0.01 m s⁻¹. The velocity profile perpendicular to the screen was measured at eight increasing distances from the screen, up to a maximum distance of 0.50 m using a hand-held propeller-driven flow meter.

Fig. 1.

Schematics of experimental flume apparatus: (a) closeup of experimental flume; (b) experiment flume apparatus.

The flume was designed to allow for interchangeable fish screens. The two screen materials used in this study were a 2-mm-wide vertical wedge-wire stainless steel screen (hereafter referred to as ‘2-mm screen’) and a perforated aluminium plate with 3-mm diameter holes and 30% porosity (hereafter referred to as ‘3-mm screen’; Fig. 2). Both screen types were recommended for the protection of native fish in the current Australian screen design guidelines (Boys 2021). The velocity 8 cm in front of the screen was denoted as the AV for standardised comparison to earlier studies by Boys et al. (2013a) and Stocks et al. (2019) (Fig. 1 and 3) and was varied as outlined in the following experiments.

Fig. 2.

The two screen types assessed in this study: (a) 2-mm-wide vertical wedge-wire stainless steel screen; (b) perforated aluminium plate with 3-mm diameter holes and 30% porosity.

Fig. 3.

Velocity profile measured within the flume at eight increasing distances along the centreline from the fish screen (0.5, 5, 8, 10, 20, 30, 40 and 50 cm). Each datapoint shows the average of three velocity measurements with standard error. The treatment approach velocity of 8 cm in front of the screen is shown with the dashed line.

Experiment 1 – larval entrainment

The larval entrainment experiment was conducted from 24 to 27 October 2016. Entrainment was compared at seven AV treatments (0, 0.05, 0.075, 0.1, 0.125, 0.15 and 0.2 m s⁻¹) in the presence of the 2-mm screen, 3-mm screen and absence of a screen, hereafter referred to as ‘no screen’. Five replicates were tested for each treatment, with each replicate consisting of a group of 10 fish in the flume. The order of replicates was assigned using a randomised block design. Additionally, five replicate 0.0 m s⁻¹ AV treatment controls were included.

To begin a treatment, a replicate group of fish was placed in the flume cradle and held at the end farthest from the screen using a 2-mm perforated plate divider covered in woven fibreglass lattice mesh with 0.28-mm diameter wire and 1.22-mm aperture, that allowed fish to acclimate after the transfer (velocity 0.0–0.01 m s⁻¹ at holding area). After 1 min of acclimatisation, the divider was removed and the fish were slowly guided with the divider, preventing erratic fleeing behaviour, to a distance of 8 cm from the exclusion screen, being the distance where fish were judged to interact with the a priori defined AV. Fish were guided to the a priori defined AV at 8 cm in front of the screen to ensure all fish were exposed to the treatment AV, providing a balanced experimental design, and ensuring no fish could occupy potential eddies or vortices in the corners of the fish cradle. Once all the fish were within 8 cm of the screen, the divider was removed, and the pump operation continued for another 1 min. After that time, the pump was turned off and all fish that were not entrained were counted to provide a proportional measure of the dependent variable ‘entrainment’ through the screen.

Experiment 2 – YOY impingement or entrainment

The YOY impingement or entrainment experiment was conducted from 21 to 24 November 2016. Impingement was compared at six AV treatments (0, 0.1, 0.2, 0.3, 0.35, and 0.4 m s⁻¹) in the presence of the 2- and 3-mm screen, and entrainment was examined in the absence of a screen. Five replicates were tested for each treatment, with each replicate consisting of a group of 10 fish in the flume. The order of replicates was assigned using a randomised block design. Five replicate 0.0 m s⁻¹ AV treatment controls were included. Each replicate treatment was performed as described in Experiment 1 above. After 1 min at the treatment AV, the number of fish impinged against the screen was counted and the pump was turned off. Screen impingement was a binary measure (impinged or free swimming) and was assessed at 1 min of exposure at the treatment velocity (i.e. instantaneous measure of the count of fish stuck to the screen at 1 min of pumping), providing a proportional measure of the dependent variable ‘impingement’. For the ‘no screen’ treatments, the pump was turned off after 1 min of pumping and the fish that were not entrained were counted to provide a proportional measure of the dependent variable ‘entrainment’.

Experiment 3 – YOY impingement duration

The YOY impingement duration experiment was conducted from 21 to 24 November 2016. A 0.3 m s⁻¹ AV was identified as the minimum velocity required for 100% impingement to occur within 1 min. Five replicates, each containing 10 Murray cod, were run for four durations (0, 1, 10 and 20 min) for the 2- and 3-mm screen at 0.3 m s⁻¹ AV. Additionally, five replicate holding controls were kept in the holding troughs but not transferred to the experimental flume to ensure any mortality was due to the impingement duration treatments rather than animal husbandry. The 0 m s⁻¹ treatment, when the pump was not operating, acted as a handling control. For each treatment replicate, 10 fish were transferred into the end of the flume cradle farthest from the screen and held there to acclimate for 1 min as detailed above. After this time, the pump velocity was increased to create an AV of 0.3 m s⁻¹ and the divider was removed and then used to slowly guide the fish to 8 cm from the exclusion screen. In all but the 0 m s⁻¹ treatment, 100% impingement occurred within the first minute. At the end of each replicate, the flume was turned off and the number of impinged fish suffering acute mortality was counted. Fish were considered dead if no opercular movement was observed. All fish were transferred back into their holding containers within the troughs and kept for 24 h before they were inspected for further deaths, providing a proportional measure of the dependent variable ‘total mortality’ (acute mortality + 24 h mortality).

Statistical analysis

For Experiment 1 (entrainment), a two-factor ANOVA was used to compare the proportion of larvae entrained at each AV (0, 0.05, 0.075, 0.1, 0.125, 0.15 and 0.2 m s⁻¹) for the two screen types (2- and 3-mm screen) and no screen.

For Experiment 2 (impingement) a two-factor ANOVA was used to compare the proportion of YOY impinged after 1 min of pump operation at each AV (0, 0.1, 0.2, 0.3, 0.35 and 0.4 m s⁻¹) for the two screen types (2- and 3-mm screen).

For Experiment 3 (impingement duration) a two-factor ANOVA was used to compare ‘total mortality’ at each impingement duration (0, 1, 10 and 20 min) for the two screen types (2 mm screen and 3 mm screen) when the AV was 0.3 m s⁻¹.

All data were assessed for normality and homogeneity of variances prior to analyses. Data that did not meet these assumptions were log₁₀ transformed and reassessed. Post hoc pair-wise comparisons were performed using Tukey’s HSD with a P-value of <0.05 considered significant. All analyses were performed in the ‘R’ package (ver. 4.3.1, R Foundation for Statistical Computing, Vienna, Austria, see https://www.r-project.org/) and figures were produced using ggplot2 (ver. 3.4.3, see https://CRAN.R-project.org/package=ggplot2; Wickham 2009).

Experiment 1 – larval entrainment

A significant interaction between the effects of screen type and AV was observed for larval entrainment (F_12,86 = 3.02; P < 0.01). Both the 2- and 3-mm screen reduced entrainment of larval Murray cod at 0, 0.05, and 0.075 m s⁻¹ AV compared to no screen (P < 0.05). No significant differences in entrainment were observed between the two screens at the 0, 0.05, and 0.075 m s⁻¹ AV (Tukey’s HSD P > 0.05). At 0.1 m s⁻¹ AV, only the 2-mm screen reduced entrainment in comparison to no screen. The 2- and 3-mm screen had no significant effect on reducing entrainment for AV ≥0.125 m s⁻¹ (0.125, 0.15, and 0.2 m s⁻¹ AV; Fig. 4). At an AV of 0.075 m s⁻¹, the 2-mm screen reduced larval entrainment by 84%, compared to no screen (screened = 10% entrainment, unscreened = 62% entrainment; Fig. 4).

Fig. 4.

Mean percentage (±s.e.) of larval Murray cod (mean SL = 10.6 mm, minimum SL = 10 mm, maximum SL = 11.4 mm) entrainment when exposed to the six treatment approach velocities for a duration of 1 min using a 2-mm vertical wedge-wire fish screen (red line) and a perforated aluminium plate fish screen with 3-mm diameter holes and 30% porosity (green line). Mean percentages of fish entrainment when no fish screen was implemented are displayed for each treatment velocity (blue line). n = 5 replicates of 10 fish for each treatment. Approach velocity treatment controls are displayed as 0 m s⁻¹ approach velocity.

Experiment 2 – YOY impingement or entrainment

Approach velocity (F_5,68 = 83.81; P < 0.001) had a significant effect on YOY impingement rates. Increasing AV led to a higher proportion of impinged fish after 1 min of pumping for the 2- and 3-mm screens (Fig. 5). At each AV, impingement was not significantly different between the 2- and 3-mm screens (F_1,68 = 4.22; P > 0.05). The mean proportion of YOY entrained during the ‘no screen’ treatment ranged from 96 to 100% for AV of 0.10–0.40 m s⁻¹ (Fig. 5). Both screens prevented the entrainment of all YOY.

Fig. 5.

Mean percentage (±s.e.) of young-of-year Murray cod (mean SL = 26.6 mm, minimum SL = 23.6 mm, maximum SL = 28.9 mm) impingement when exposed to the five treatment approach velocities for a duration of 1 min using a 2-mm vertical wedge-wire fish screen (red line) and a perforated aluminium plate fish screen with 3-mm diameter holes and 30% porosity (green line). Mean percentages of fish entrainment when no fish screen was implemented are displayed for each treatment velocity (blue line). n = 5 replicates of 10 fish for each treatment. Approach velocity treatment controls are displayed as 0 m s⁻¹ approach velocity.

Experiment 3 – YOY impingement duration

Total mortality was positively related to impingement duration (F_3,32 = 19.24; P < 0.0001). For both the 2- and 3-mm screen, total mortality for impingement durations greater than 1 min were significantly greater than the holding controls (Fig. 4). Total mortality was not significantly different between the 2- and 3-mm screen for each impingement duration (P > 0.05, Tukey’s HSD pairwise compassions), (Fig. 6).

Fig. 6.

Mean percentage (±s.e.) of young-of-year Murray cod (mean SL = 26.6 mm, minimum SL = 23.6 mm, maximum SL = 28.9 mm) that suffered mortality at the four impingement durations when exposed to a 0.3 m s⁻¹ approach velocity at a 2-mm wedge-wire fish screen (red histogram) and a perforated aluminium plate fish screen with 3-mm diameter holes and 30% porosity (green histogram). n = 5 replicates of 10 fish for each treatment. Handling controls are displayed as 0-min impingement duration.

Larval entrainment

The present study identified reduced entrainment of larval Murray cod at <0.125 m s⁻¹ AV compared to treatments with no screen. This was likely a behavioural response to the presence of the screen, rather than the screen physically preventing entrainment (given screen aperture was greater than larval head width). This demonstrates that fish larvae are not purely passively drifting organisms and even very young Murray cod can avoid screens if the hydraulic conditions are suitable. Previous research by Boys et al. (2013a) demonstrated that juvenile silver perch and golden perch utilised visual cues to navigate the screen face, thereby enhancing their ability to avoid contact with the screen. However, as the AV increases towards the swimming capacity of larval fish, the physiological limitations of their swimming capacity likely overpower their avoidance behaviour, resulting in entrainment. Similarly, Carter et al. (2023) concluded that while physical exclusion is an important consideration when screening intakes, other biological or hydraulic processes may also influence entrainment.

Results indicated that the 2-mm screen outperformed the 3-mm screen. However, at ≥0.125 m s⁻¹ AV, the fish screens trialled in this study did not reduce entrainment of larval Murray cod. This demonstrates that if the AV is not low enough, the mesh type used at the screen has little relevance to larval fish entrainment. It is important to note that susceptibility to entrainment varies between species (Kelso and Leslie 1979; McLaren and Tuttle 2000). This variability is influenced by factors such as swimming performance and the behavioural response of a species to the presence of a screen (Carter et al. 2023). Swimming performance of fish can be influenced by a range of biological and physical factors (Webb 1975), including morphology (Hammer 1995; Plaut 2001), muscle function (Kieffer 2000), swimming mode (Muller et al. 2001), body condition (Muller et al. 2001), and water temperature (Ojanguren and Brañta 2000). Moreover, ontogenetic changes in swimming performance during larval development, and interspecies variations attributed to precocial life-history strategies, may also influence entrainment susceptibility (Kopf et al. 2014).

To further explore the swimming capabilities of larval fish species, Kopf et al. (2014) tested the swimming performance of the larvae of six Australian freshwater fish species. The results revealed considerable interspecific and ontogenetic variation in swimming performance. For instance, metalarvae of trout cod (Maccullochella macquariensis) exhibited the highest swimming speed, with a maximum critical swimming speed of 0.464 m s⁻¹ (defined as prolonged swimming speed for ≤60 min by Beamish 1978). By contrast, preflexion larvae of silver perch (Bidyanus bidyanus) displayed the slowest swimming speed, with a minimum critical swimming speed of 0.001 m s⁻¹.

Kopf et al. (2014) reported the critical swimming speed of postflexion larval Murray cod ranged ~0.1–0.25 m s⁻¹. Consistent with that research, the present study observed larval entrainment percentages exceeding 90% at AV ≥0.15 m s⁻¹, which aligns with the lower end of the reported critical swimming speed for this species. Hutchison et al. (2020) suggest that burst and sprint speed, defined as maximum swimming speed obtained by anaerobic metabolism in <20 s according to Beamish (1978), may serve as a more explicit indicator of an individual fish’s ability to avoid entrainment. However, limited data on burst and sprint speed for larval species in the MDB are currently available (Watson et al. 2019a, 2019b; Hutchison et al. 2020). In the present study, the flume apparatus design aligns more so with a critical swimming speed assessment. In the wild, a fish may be able to employ a quick swimming burst until it is swept past the screen. To better test this, we recommend that future experiments use a flume design that allows for this behaviour. This may include letting fish volitionally approach the screen and allowing them to avoid impingement by using short bursts of swimming until they have passed by the screen.

In the Pacific Northwest of America, when developing juvenile fish screen criteria, fisheries agencies established criteria that would protect the weakest swimming species during their most vulnerable life-stage under adverse environmental conditions (Nordlund 1996). Kopf et al. (2014) demonstrated that preflexion silver perch larvae displayed the slowest critical swimming speed of Australian native species. Therefore, it is recommended that further fish screening trials be conducted using preflexion silver perch larvae given they are likely a highly vulnerable species and life-stage.

In addition to biophysical studies of swimming performance and screen interactions, studies on egg and larval drift dynamics within natural river systems may further refine design specifications for fish screens to protect native fish larvae. Drifting larval fish have an active component to their movement and are capable of altering their position in the water column (Braaten et al. 2008; Lechner et al. 2014); similarly, drifting fish eggs have shown spatial variability in drift abundances throughout the water column (Faulkner and Copp 2001; Tonkin et al. 2007). Understanding the depths at which native larval fish drift could guide the appropriate placement of offtakes within the water column to minimise exposure to entrainment.

Young-of-year impingement

Modern fish-protection screens have shown effectiveness in reducing entrainment. However, they still pose risks for injury and mortality due to impingement on the screen face (Peake 2004; Stocks et al. 2019). It is critical to design screens that do not exceed the AV at which fish can actively avoid contact with the screen (White et al. 2007). The present study observed impingement of YOY Murray cod at ≥0.1 m s⁻¹ AV, with impingement rates increasing at higher AV. Previous literature supports the positive correlation between impingement and AV (Danley et al. 2002; Swanson et al. 2005; White et al. 2007; Young et al. 2010; Boys et al. 2013a; Stocks et al. 2019; Carter et al. 2023). Consistent with findings by Boys et al. (2013a), there was no statistically significant difference between the two screen types in the proportion of impinged fish, indicating that AV has a greater effect on impingement than screen type.

Fish injury and mortality are positively corelated with impingement duration (Peake 2004; Stocks et al. 2019). In our study, mortality of YOY Murray cod increased significantly with impingement duration for the 2- and 3-mm screens at 0.3 m s⁻¹ AV. Similar findings were reported by Stocks et al. (2019) regarding YOY golden perch, suggesting that reducing impingement time through pump cycling (turning pump on or off to allow impinged fish to escape) could mitigate mortality while maintaining AV. However, implementing pump cycling from an engineering and maintenance perspective may be impractical. Therefore, the most effective approach to minimise impingement-induced mortality is to limit AV and avoid impingement.

Fish screening criteria

Our results indicate that the prescribed Australian fish screening guidelines provide protection for postflexion larvae and YOY Murray cod. Boys (2021) recommended limiting AV to ≤0.1 m s⁻¹ and that aperture size of wedge-wire and perforated plate screens should not exceed 2 mm and 3 mm, respectively. However, it was recognised that further investigation was required to determine the specifications necessary for protecting larval and earlier life-stage fish (Stocks et al. 2019; Boys 2021). Results from the present study show that a 0.1 m s⁻¹ AV did reduce the impingement of YOY Murray cod within an experimental flume. However, at 0.1 m s⁻¹ AV, only the 2-mm screen reduced entrainment of postflexion larval Murray cod. The 3-mm screen at 0.1 m s⁻¹ AV did not reduce entrainment of larval Murray cod when compared to the unscreened treatment. At ≥0.125 m s⁻¹ AV, the fish screens trialled in this study did not reduce entrainment of larval Murray cod. This demonstrates that if the AV is not low enough, the mesh type used at the screen has little relevance to larval fish entrainment.

The present study also demonstrated that when using a 2-mm screen there was limited benefit in further reductions to the maximum prescribed AV of 0.1 m s⁻¹. For the 2-mm screen at 0–0.1 m s⁻¹ AV, there was no significant difference in larval entrainment. Only when AV was greater than 0.1 m s⁻¹ were larval entrainment values significantly different from 0 m s⁻¹. Consequently, we recommend that pump offtakes preferentially be fitted with 2 mm vertical wedge-wire screen over the 3-mm perforate plate screen given their superior performance in larval exclusion.

The positioning of offtake pumps within rivers also has the capability of further protecting fish (Hutchison et al. 2020). Positioning offtakes in rivers with flows acting parallel to the screen, creating sweep velocities across to the screen face, have been shown to reduce fish impingement and entrainment (Danley et al. 2002; Swanson et al. 2004; White et al. 2007; Young et al. 2010). As previously alluded to, the present study did not incorporate sweep velocities. Consequently, impingement and entrainment at the treatment AV are likely overestimated compared to scenarios in flowing water bodies.