Flow hydrodynamics drive effective fish attraction behaviour into slotted fishway entrances

Effective fishways rely on attracting fish, utilising the natural rheotactic behaviour of fish to orient into an attraction flow near the entrance. Despite the critical importance of attraction, understanding of the hydrodynamics of vertical slot entrances in relation to fish behaviour remains poor. Herein, hydrodynamic measurements of flows at slotted fishway entrances were experimented with two different designs, two velocities, three water depths, and two fish species, silver perch (Bidyanus bidyanus) and Australian bass (Percalates novemaculeata). Fish behaviours were tracked in relation to hydrodynamic measures of three-dimensional velocity and turbulent kinetic energy (TKE). There were distinct differences in the attraction flow between entrance designs, irrespective of velocity and water depth. A plain slotted entrance produced a more symmetric flow in the centre of the flume, causing fish to approach the entrance by skirting the core of the attraction jet flow and areas of high turbulence. In contrast, streamlined slotted entrance design resulted in an asymmetric attraction flow which guided fish along the wingwall towards the slotted entrance, improving attraction for both species. There were clear patterns in swimming trajectories for silver perch, swimming along the sidewalls of the observation zone towards the entrance, but Australian bass were less predictable, using random routes on their way to the slotted entrance. Both species preferred areas of low turbulence (TKE < 0.02 m2 / s2). This work has important implications for design of vertical slotted entrance systems.


Introduction 
Dams, weirs and levees are important for flood protection, hydropower generation, and for water supply [1] but contribute to significantly reduced fish migration and fish population diversity in rivers and estuaries [2] .Fishways mitigate this problem, enabling fish movement past barriers.Most constructed fishways, such as vertical slot and pool fishways, use sloped open channels divided by cross-walls into a series of pools to allow fish to swim past a barrier [3] .Other fishway types transport fish through closed conduit system of pipes such as the Whooshh system [4] and the tube fishway [5][6][7] .
Most fishways rely on the natural rheotactic behaviour of fish to attract them into a fishway entrance, before they move past a barrier.Rheotaxis orientates fish into a current using the sensory cells of the fish's lateral line system [8] .Therefore, fish movement depends on fishway design (e.g., pool dimension and slot design), channel slope [9] , and associated flow conditions (e.g., water depths, velocities and turbulence).Turbulence can affect fish swimming capacity, reducing fish swimming speed and stability [10] .Inadequate attraction flow can reduce overall fishway effectiveness, and may delay fish attraction which increases predation risk [11] .It is clear that improved understanding of relationships between fish behaviour and the hydrodynamics of attraction flows could significantly improve fishway performance [12] .Complicating this relationship, behaviour of different fish species also varies, reflecting differences in swimming speeds, swimming path selection, and response to turbulent flows [13] .
Despite this importance, there is currently little documented understanding on suitable flow hydrodynamics for effective fish attraction [14] , particularly in relation to velocity and turbulence.Velocity of attraction flows needs to be sufficient to induce fish rheotaxis and swimming through the fishway entrance, but not too high to affect swimming ability [15] .Salmonids are attracted by up to 10% of the main river discharge into fishways [16] .Attraction flow velocities of more than 2 m/s are recommended for economically significant salmonid fish species, such as Pacific lamprey (Lampetra tridentata), American shad (Alosa sapidissima), and Pacific salmon (Oncorhynchus) [17] .However, such velocities exceed the recommended swimming capabilities of some non-salmonoid species: 0.4 m/s for Prenant's schizothoracin (Schizothorax prenanti) [18] , 0.25 m/s for Iberian barbel (Luciobarbus bocagei) [19] , and 0.2 m/s to 0.3 m/s for perch barbel (Percocypris pingi) and grass carp (Ctenopharyngodon idella) respectively [14,20] .Juvenile silver perch (Bidyanus bidyanus) and Australian bass (Percalates novemaculeata) had preferred attraction flow velocity of 0.15 m/s [21] .
Turbulent kinetic energy (TKE) is most commonly used hydrodynamic descriptor for fish attraction into fishways [9] , measuring mean kinetic energy associated with velocity fluctuations.Fishway entrances, designed for salmonoids, operate with TKE values of 0.1 m 2 /s 2 -1.2 m 2 /s 2 [22] , possibly too high for non-salmonid species with lower swimming capabilities [23] .Chen et al. [24] investigated attraction flows for six endemic fish species in China showing that 0.02 m / s [25] , while 0.003 m / s allowed optimal attraction of three socio-economical important fish species in the Jing River in China (Cyprinidae Phoxinus lagowskii, Opsariichthys bidens and cobitidae Triplophysa stoliczkae) in an entrance channel with a sluice gate [26] .In this study, the relationships between hydrodynamics of the attraction flow and fish behaviour were investigated in two different vertical slotted entrance designs (plain and streamlined) for different velocities and water depths for two fish species (silver perch and Australian bass).Firstly, the visual observations of the flow patterns of the attraction flows as well as detailed measurements of the velocity field were reported upstream and downstream of the slotted entrance.Secondly, the fish attraction behaviour in relation to the attraction flows were examined for the two slotted entrance designs as well as the range of velocities a V and water depth d to improve attraction of fish to vertical slotted fishway entrances.

Experiments
Experiments were conducted at the UNSW Water Research Laboratory in an open-channel flume with glass sidewalls of 6.0 m length, 0.6 m width, and 0.6 m height.The flume had a 1.1 m long observation channel (OC), with a slotted entrance (SE) of 0.03 m width at the upstream end.Fish were attracted into a pipe T-section, called transfer chamber (TC) as used in tube fishways [5] (Fig. 1(a)).The pipe T-section had an inlet pipe of 0.05 m diameter on the upstream side which expanded gradually towards the T-section (Fig. 1(a)).The T-section of the transfer chamber was open at the top to provide similar light conditions as in the observation channel.Flow direction was through the transfer chamber and into the observation channel (Fig. 1), reflecting placement of a fishway in a river.Experiments were conducted for three water depths ( ) d at the slotted entrance (Fig. 1), 0.08 m, 0.18 m and 0.32 m, respectively corresponding to transfer chamber diameters =  1).Using a recirculation system, flows were pumped through an inlet pipe into the transfer chamber, before passing through the slotted entrance into the observation channel (Fig. 1(a)).The resulting attraction flow emerged as jet (hereafter referred to as a jet flow), designed to attract fish.Flow was measured with a Yamatake Honeywell flowmeter (accuracy of 0.5% of the flow rate).Water flowed from the observation channel via gravity into a ground reservoir before being pumped to a 6 m 3 large recirculation reservoir.Experiments used either a plain entrance design (as the most common entrance for vertical slot fishways) (Fig. 1(b), Table 1), or a streamlined entrance design with two 45 wingwalls next to the slotted entrance (which previously proved to improve attraction for silver perch and Australian bass [21] (Fig. 1(c), Table 1).

Hydrodynamic measurements
Measurements of the hydrodynamic flow properties were conducted with a SonTek 16 MHz Micro acoustic Doppler velocimeter (ADV).Velocities were measured across a grid of 80 points in the observation channel, with focus on the attraction jet flow region, just downstream of the slotted entrance and at 19  and z v , Fig. 1), with a sampling frequency of 200 Hz and for four minutes at each measurement point.Velocity sampling duration was the convergence of mean and standard deviation, identified using sensitivity analysis at three representative locations: Just downstream of the slotted entrance ( = 0.05 m) X , in the middle of the observation chamber ( = X 0.55 m) , and close to the weir ( = 1.05 m) X (Fig. 1).Measurements were taken in three horizontal planes at depths of = Z 0.1d , 0.3d and 0.5d for = d 0.18 m and at one horizontal plane of = 0.3 Z d for = d 0.08 m, 0.32 m, respectively.These depths represented fish preferences in entering the lower half of the slotted entrance [21] .In total, 1 500 measurement points were taken for all the combinations of d , a V , and slotted entrance design in this study.Vertical profiles of velocities were also measured at three locations downstream of the slotted entrance ( = X 0.05 m, 0.10 m and 0.20 m) for = d 0.18 m, 0.32 m to check variability of velocity with depth.The data quality was improved through seeding the flume water with neutrally buoyant clay powder [11] , resulting in signal-to-noise ratios (SNR) mostly in the range of 40 dB-60 dB.Win-ADV was used to filter the raw data removing low quality data, with correlation coefficients below 70% and SNR below 5 dB [28] .A cut-off of 15 dB would have resulted in 0.3% 0.1  difference in velocities and TKE for the present data.All data were filtered with the despiking filter [29][30] .The resulting velocity time series were post-processed, yielding mean velocities x v , y v and z v in x , y and z directions at each measurement point, their corresponding standard deviations ( x v , y v and z v ), and turbulent kinetic energy TKE.
To visualize the flow patterns of the attraction flows downstream of the entrance, experiments with blue dye complemented the quantitative ADV measurements.High-concentration dye was injected through a stainless-steel nozzle into the flow at the slotted entrance at a water depth of 5 ~0.z d for all experiments.The dye cloud motion was recorded with two GoPro (HERO8 black edition) video cameras (1080p, 60 fps), from top view and through the sidewall visualizing the attraction flow patterns.

Fish behaviour
Silver perch (total length, TL = 67 mm -87 mm ) and Australian bass (TL = 117 mm -152 mm) were maintained in 0.14 m 3 or 0.20 m 3 tanks at the UNSW Water Research Laboratory [21] .In that study, seven replicates with groups of five silver perch and five replicates with groups of five Australian bass were conducted for one hour for all the combinations of = d 0.08 m, 0.18 m, and 0.32 m, = a V 0 m/s, 0.15 m/s and 0.5 m/s, and slotted entrance design (plain and streamlined).A total of 136 trials showed that fish could be attracted irrespective of the water depth and attraction flow velocity with = 0.18 m d and the streamlined entrance with = 0.15 m / s a V representing the most successful attraction flow conditions [21] .
In the present study, top and side view recordings were manually re-analysed to record the pre-entry fish locations in the vicinity of the slotted entrance (see yellow and green shaded areas in Fig. 1(a)) every second during the 1 h filming of each observation trial on a fine-scale grid: (jet-centre) corresponding to the number of fish entries from left, right, and centre of the attraction flow.The pre-entry fish location was defined based on the average fish body length of both fish species of approximately 0.1 m.For the streamlined entrance, the attraction jet was deflected towards one of the sidewalls downstream of the slotted entrance, and the regions for the pre-entry fish locations were divided into the centre, the jet region on one side and a non-jet region on the other side.For easier readability, the jet is shown on the left sidewall of the streamlined entrance, while the jets were also observed on the opposing side.
Locations of fish before they entered the entrance (pre-entry) were modelled using linear models (LMs) and ANOVA to identify the relative importance of predictors (attraction direction, velocity, and water depth) for both species.To meet assumptions of normality, a log transformation was required.For a significant interactive effect, the estimated marginal means package was used for contrast analysis to identify specific differences [31] .All statistical analyses were performed using the R software (1.1.456,R Core Team, 2021).Statistical significance was at P < 0.05.
In addition, videos of the swimming trajectories of 60 fish (30 silver perch and 30 Australian bass) of the first fish entries (frame by frame, 60 fps) across the observation channel were analysed, using opensource video analysis software Tracker [32]

Results: Hydrodynamics of attraction flows
This section presents representative results of the flow pattern observations and quantitative hydrodynamic measurements for plain and streamlined entrances, supported by comprehensive results (Figs.S2-S5 in supplementary material).

Plain entrance
Generally, there were clear flow patterns for the plain slotted entrance, with a jet flow emerging out of the slotted entrance, gradually expanding in downstream direction and eventually dispersing (Fig. 2).The overall jet patterns were consistent with free jets in open channel flows [33] (Fig. 2).Downstream of the entrance, the main jet trajectory diverged towards one side of the observation channel (indicated by solid lines, grey and black in Fig. 2 Figure 3 shows typical results of the streamwise velocities.The velocities were normalised with a V and the same colour scheme was used for all flow conditions to enable comparison between flow conditions (Fig. 3).The maximum normalised streamwise velocity (1.15 < / < 1.3) x a V V occurred just downstream of the slot, in centreline, and the jet decayed rapidly along the jet (0.2 < / < 0.8) x a V V , confirming the overall flow pattern observations (Fig. 2).The velocity decay was consistent with those observed for single vertical slot pools [22] .Further downstream, the jet propagated along one of the sidewalls initiating recirculation motions ( 0.6 < / < 0.2) , which was consistent with the observations of Ref. [34].A smaller recirculation zone with very low velocities occurred in the corner, adjacent to the main jet (Fig. 3).While the overall velocity distributions were consistent, irrespective of d (Fig. 3  connected to the lower layers of the transfer chamber, leading to comparatively higher velocities near the bottom.There was a similar variation in the vertical profiles of velocities at three points along the observation channel ( = X 0.05 m, 0.10 m and 0.20 m downstream of the slotted entrance) across water depths.Increasing the attraction flow from = a V 0.15 m/s to 0.50 m/s had little impact on the streamwise, transverse, and vertical velocity distributions in the observation channel for the respective water depth (Fig. S2 in supplementary material).
Inside the transfer chamber, the flows were more complex, compared with the observation channel (Figs.3(a)-3(d), Fig. S2 in supplementary material).There were two velocity regions: One with higher velocity (0.2 < / < 1.3) x a V V and one with lower/ negative velocity (reverse flow) ( 0.6 < / < 0.2) V , TKE values were relatively higher inside the transfer chamber compared , consistent with velocity observations (Fig. S3 in supplementary material).

Streamlined entrance
For the streamlined entrance, there was a distinct pattern of the attraction jet clinging to one of the two 45 wingwalls, immediately downstream of the slotted entrance (Fig. 5).Irrespective of the water depth, the jet followed the sidewall, initiating a steady recirculation motion across the observation channel.This jet deflection was probably caused by the wingwalls initiating a "Coanda effect".The deviation of a jet flow from its original path when it encounters a nearby wall is the "Coanda effect" [35] .The presence of sidewalls caused pressure gradients perpendicular to the jet, deflecting the attraction flow.
The distinctive jet deflection to one side and recirculation motion were reflected in the velocity distributions in the observation channel, irrespective of flow conditions (Fig. 6, Fig. S4  The velocity decay along the deflected jet occurred faster compared with the streamlined entrance, which is likely linked to the Coanda effect which slows the velocity much more than a free jet [35] (Fig. A1, Appendix).Inside the transfer chamber, velocities varied strongly as well (Figs.

Results: Fish response to attraction flow
In this section, the pre-entry fish locations at the vicinity of the slotted entrance are compared with the attraction jet flow patterns (Section 3.1) and the fish swimming trajectories are compared with the distributions of TKE (Section 3.2) providing guidance on the interaction of fish behaviour with attraction flow hydrodynamics.

Pre-entry location
The distribution of locations of silver perch and Australian bass before they entered (pre-entry locations) the two slotted entrance designs varied with flow conditions (Fig. 8).Statistical analysis showed complex 3-way interactions among the attraction direction taken by fish and different velocity and water depth, which was significant ( = 0.001 P , 6, 66 = F 4.06 ) for silver perch (Table A1, Appendix), albeit not significant ( = 0.96 P , 3, 60 = 0.09 F ) for Australian bass (Table A2, Appendix).There were also significant 2-way interactions between the attraction direction and velocity ( 0.001) P < for both species (Tables A1, A2, Appendix), revealing that pre-entry locations varied with velocities and entrance designs.Overall, most fish entered from the jet region of the streamlined entrance, irrespective of the water depth ( ratio = 9.40 T , 0.001 P < for silver perch, Table A3, Appendix, ratio = 5.24 T , 0.001 P < for Australian bass, Table A4, Appendix).Contrastingly, there was no significant tendency for silver perch or Australian bass to enter via jet-region for plain entrance ( ratio = 0.40 T  , = 0.63 P for silver perch, ratio = T  , = 0.24 P for Australian bass).When considering the attraction direction in the vicinity of the streamlined entrance, more than 80% of fish entries occurred via the jet with only a small fraction attracted through the centre and the non-jet side, irrespective of the fish species and the water depth (dark green colour in Fig. 8).For the plain entrance with a centre jet just downstream of the slotted entrance, the number of pre-entry locations for the silver perch and Australian bass were similar among the three entry categories, with a slight dominance for the entry at the jet-centre for = a V 0.15 m / s for all water depths (orange colour in Fig. 8).For the condition with no attraction velocity ( = 0 m / s) a V , the results were similar with a balanced re-entry location.However, for the largest attraction flow velocity ( = 0.50 m / s) a V , there was a tendency for fish entry via the jet-centre for both fish species and most water depths (Fig. 8).

Fish swimming trajectories
There were clear patterns in swimming trajectories for silver perch, but Australian bass used varied swimming paths for different slotted entrance design (Fig. 9).Body morphology, sensory systems, and individual behaviour of silver perch and Australian bass could all contribute to these differences [36] .Silver perch tended to swim along the sidewalls and corners with no preference for any specific side, for the plain entrance and the control flow condition ( = 0 m / s) a V (Fig. 9(a)), while Australian bass selected more direct paths within the centre part of the observation channel (Fig. 9(b)).When an attraction velocity was present ( = 0.15m /s) a V , most silver perch swam along the sidewall where there were slightly higher values of TKE, i.e., TKE 0.0002 m / s  on the opposite sidewall (Fig. 9(c)).It appeared that silver perch responded to the slightly higher TKE values indicating that small changes in hydrodynamic flow conditions can have a large effect on the fish swimming trajectory.Along their trajectory, silver perch preferably passed through the recirculation zone in the corner and avoided to swim along the main jet flow with 0.35 < TKE / < 0.60  cues [37] .Also, their trajectories were similar to the no-flow control condition (Figs.9(b), 9(d)).One Australian bass followed the edge of the main jet (0.002 m / s < TKE < 0.008 m / s ) , while another bass crossed through the jet region towards the slotted entrance (Fig. 9(d)).
The observation for the streamlined slotted entrance were substantially different for both fish species.Most silver perch and Australian bass followed the main trajectory of the jet with regions of maximum TKE (Figs.Further, when silver perch approached the streamlined entrance from the far end of the observation channel for = 0.15 m / s a V , most swam along the jet into the slotted entrance, whatever the water depth (Fig. 10(a)).This finding was consistent with the more detailed fish swimming trajectory observations for = 0.18 m d (Fig. 9(e)) highlighting that silver perch are strongly responsive to attraction jet flows.In contrast, the observations of Australian bass suggested less Australian bass followed the jet flow trajectory for = d 0.08 m, 0.32 m (Fig. 10(b)).

Discussion
There is relatively poor understanding of fish attraction to slotted fish entrances in relation to the flow hydrodynamics.The present results showed how fish behaviour varied between two juvenile Australian fish species and two slotted entrance designs, reflecting on the response of fish to flow hydrodynamics, There were strong responses from one of the fish species, juvenile silver perch, while the other species, Australian bass, was less responsive.Importantly, hydrodynamics explained the movements of the fish, providing some novel insights into fish behaviour when attracted to flow.As expected, flow hydrodynamics was particularly sensitive to the entrance design.Plain and streamlined entrances generated different jet flows (Figs.2-7).The plain slotted entrance created a central jet with typical decay in velocity magnitudes and turbulent kinetic energy in downstream direction, while the streamlined slotted entrance deflected the jets towards a sidewall.Most juvenile fish preferred to swim into the main attraction flow irrespective of the water depth, following the distinct jet flow along one of the sidewalls (Figs. 9, 10).The interaction of the jet with the sidewall boundary layer likely created optimum attraction conditions of flow for the streamlined entrance as it was more effective for fish than the plain entrance [21,38] .In streams, fish use the area near structures because they not only provide regions of low flow but also an opportunity to swim more efficiently given the effects of solid walls [39] .Wall-like structures are also wellknown refuges suitable for fish predator avoidance [39] .
For the plain entrance, most silver perch also tended to swim along the sidewalls of the observation zone towards the entrance, but Australian bass were less predictable, moving along different paths, even when there was no flow (Figs.9(b), 9(d)).Both fish species generally moved along low-velocity zones, avoiding areas of high turbulence.This reflects fish movements in other rivers, concentrated along reduced velocity areas, minimising energy expenditure [40] .Fish respond to hydrodynamic processes with rheotactic responses mediated through lateral line organs, vision, and vestibular functions [8] , guiding fish with hydraulic signals.
TKE is a key factor governing attraction flow hydrodynamics for fish [9] .For both slotted entrances, the attraction jet created areas with varying flow velocities and TKE distributions in the observation channel providing diverse attraction flow conditions which allowed our tested fish to freely select attraction paths, probably reflecting body morphology and biology [41] .The successful use of such asymmetric attraction flows occurs in other vertical slot fishways [22] .For improved attraction, several parameters need to be considered, including the residence time of fish in the fishway without escaping downstream into the observation channel which is negatively correlated with TKE.Manipulating depth and velocity can be critical, with silver perch and Australian bass exhibiting the highest residence time (up to 95%) for = 0.18 m d , = 0.15 m / s a V compared with other depths and velocities [21] .Based on our analysis, juvenile silver perch and Australian bass moved best when 2 2 TKE < 0.02 m / s .This is similar to TKE values favouring attraction of tropical fish species at vertical slot entrance channels [24][25][26] .In our study, TKE was significantly higher than 0.20 m 2 /s 2 inside the observation channel and the transfer chamber when = 0.50 m / s a V , irrespective of water depth ( ) d .This probably resulted in fewer fish entrances with significantly lower residence time in the transfer chamber under these flow conditions [21] , beyond the fish swimming capability [16] .
This study provided important guidance to optimise fish attraction, particularly when jet flows were directed along a sidewall.Obviously, there are a range of scale dependent issues [42] , given the limitations to two juvenile fish species and extrapolation of experimental results to field applicability.Future research should test whether other fish species and fish sizes at different life stages and in field settings behave similarly.Expanding beyond slotted entrance designs, to other opening geometries [43] and their effects on hydrodynamics and fish attraction behaviour are also important.

Conclusion
Investigations of the relationships between fish behaviour at slotted fishway entrances and hydrodynamics are rare.Interactions between flow hydrodynamics and fish behaviour of two Australian fish species, one coastal and the other inland, were investigated.There were complex interactions but clearly fish were attracted to flow.Importantly, there were interesting ramifications for improving the effectiveness of slotted fishway entrances.Different attraction flow hydrodynamic scenarios created improved fish attraction behaviour into fishways.Streamlined entrances performed better than plain entrances with fish capitalising on predictable zones of low turbulent kinetic energy for the former, as they were attracted to the entrances.Once fish were at the vicinity of the streamlined entrance (after approach), they frequently sought out the main attraction jet flow.Fish seem to be exploiting patterns of jet flows, swimming where swimming costs are probably minimised.There remains much to be learnt about the role of flow hydrodynamics in slotted and other fish entrances, which could further improve effectiveness of fishways and their ability to attract the range of species and different times of their life history.

a D 0
.100 m, 0.225 m and 0.400 m.To ensure open channel flows inside the transfer chamber, the ratio of water depth to the transfer chamber diameter was 80% ( / = 0.8) a d D .The water depths were controlled by adjusting a weir at the downstream end of the observation channel (Fig. 1(a)).The transfer chamber lengths ( ) a L (Fig. 1(a)) were 0.25 m, 0.60 m and 1.00 m corresponding to = a D 0.100 m, 0.225 m and 0.400 m, respectively.Experiments were conducted for a range of attraction flows: m/s and 0.5 m/s (Table

Fig. 1 (
Fig. 1 (Color online) Sketch of the experimental setup for testing relationships between fish behaviour and hydrodynamics of slotted fishway entrances, showing: (a) Top view of the setup with the transfer chamber (TC), slotted entrance (SE) and observation channel (OC) for the plain slotted entrance design, with the arrow showing the direction of flow and the area adjacent to the slotted entrance (yellow and green shadings, defining the pre-entry location) segmented into jet (centre) and non-jet regions (left and right) and the same experimental setup but for the two designs of (b) Plain entrance and (c) Streamlined entrance, with the coordinate system defined at the centre bottom of the slotted entrance Table 1 Summary of experimental flow conditions, water depth , transfer chamber diameter , attraction flow , attraction flow velocity at the slotted entrance , the corresponding cross-sectional average velocity in the transfer chamber , the Froude number at the slotted entrance , the Reynolds number at the slotted entrance ( is the water density, is the dynamic viscosity of water), and the momentum flux at the slot . The time series of the resulting coordinates of the fish swimming trajectories in the x-y plane were plotted across the observation channel.The trajectory analysis focussed on = 0.18 m d for the plain entrance with = a V 0 m/s, 0.15 m/s and the streamlined entrance for = 0.15 m / s a V for both fish species.To confirm the strong preference of fish to swim along the jet trajectory for the streamlined entrance, additional analysis of the approach swimming trajectories of 45 silver perch and 45 Australian bass for first fish entries with the streamlined entrance ( = a V 0.15 m / s) were conducted for all water depths ( = d 0.08 m, 0.18 m and 0.32 m).The fish swimming trajectories were divided into two distinct areas of jet and non-jet regions downstream of the slotted entrance.
(a) and the blue dye in Figs.2(b), 2(c), creating a recirculation pattern in the observation channel (dashed lines in Fig. 2(a)).Depending on the water depth and the attraction velocity, a second recirculation zone was maintained in the corner, next to where the main jet first emerged (Figs.2(a), 2(b)).
, Fig. S2 in supplementary material), the jet for d = 0.18 m tended to travel downstream in a straighter direction compared with other two water depths (Fig. S2 in supplementary material), creating two more equal recirculation zones on either side of the jet.This is also reflected in the transverse velocity distributions, with recirculation motions of almost similar size, on either side of the jet for = 0.18 m d , but differently sized for = d 0.08 m, 0.32 m (Fig. S2 in supplementary material).

Fig. 2 (Fig. 3 (
Fig. 2 (Color online) Flow patterns of jet flow downstream of a plain entrance: (a) Conceptional jet flow patterns, with main jet trajectories (solid arrows, black showing main direction and grey showing dispersed/reduced flow) and recirculation motions (dashed arrows), (b) Dye (dark blue) trajectories for = 0.08 m d , = a V 0.15 m / s and (c) = 0.32 m d , = 0.15 m / s a V A comparison of velocities across different horizontal layers ( ) Z for = 0.18 m d , showed higher d .Increasing the water depth from = 0.88 m d to = 0.32 m d reduced the magnitude and the extent of / z a V V inside the transfer chamber.There was also a strong change in magnitude of streamwise and vertical velocity components across the water depth for = 0.18 m d (Fig. S3 in supplementary material), showing a highly threedimensional flow inside the transfer chamber.Typically, the strongest dimensionless TKE ( TKE / ) a V was just downstream of the slot, decaying in value and widening along the jet (Fig.4, Fig.S2in supplementary material).The values of TKE / a V in this region were similar for all flow conditions and water depths, indicating similar jet flow properties irrespective of water depth (Fig.S2in supplementary material).TKE decayed along the jet, consistent with the observations of the velocities.In the remaining part of the observation channel (the blue zone in Fig.4), TKE / a V m / s ) than the main jet trajectory, consistent with observations of reductions in flow velocity away from the attraction jet.Magnitudes of TKE / a V varied across different horizontal layers (Fig.S3in supplementary material).For example, for = 0material).This was consistent with variation in velocities across the attraction flow depth, indicating strong three-dimensionality in the attraction jet.For a given d , a

Fig. 4 (
Fig. 4 (Color online) Normalised turbulent kinetic energy observations for a plain slotted entrance at for in supplementary material).Comparison of the velocity distributions for different d , showed consistent maximum streamwise velocities just downstream of the slotted entrance (Figs.6(a), 6(b), Fig. S4 in supplementary material).The jet extended across a wider area of the observation channel when d was smaller (with magnitudes of 0.2 < / < 1.8 .6(a), 6(b), Fig. S4 in supplementary material).Across the three horizontal planes ( = Z 0.1d , 0.3d and 0.5d ) for = 0.18 m d , velocity varied with depth.It was highest near the bottom ( = 0.1 ) Z d (Figs.6(c), 6(d), Fig. S5 in supplementary material).
6(c), 6(d), Fig. S5 in supplementary material).The flows inside the transfer chamber were characterised by a strong velocity gradient and a recirculation motion (Fig. 6, Fig. S4 in supplementary material).

Fig. 5 (V
Fig. 5 (Color online) Flow patterns of attraction flow for a streamlined entrance: (a) Conceptional jet flow patterns with the main jet trajectory (solid arrows) and recirculation motions (dashed arrows), (b) Dye (dark blue) trajectories for = 0.08 m d , = 0.15 m / s a V and (c) = 0.18 m d , = 0.15 m / s a V

Fig. 6 (
Fig. 6 (Color online) Normalised streamwise velocities upstream and downstream of a streamlined slotted entrance

Fig. 7 (
Fig. 7 (Color online) Normalised TKE for a streamlined entrance at m / s ) (Fig.9(c)).They then entered the transfer chamber by swimming towards the jet and making a 90° turn into the slotted entrance (Fig.9(c)).Australian bass used more varied routes than silver perch, with a slight preference for areas with higher TKE values in the observation channel (Fig.9(d)) due to their natural behavior and response to environmental

Fig. 8 (
Fig. 8 (Color online) Pre-entry locations for repetitive entries of silver perch (a)-(c) and Australian bass (d)-(f) for variation in attraction velocities and water depths in relation to directions of the attraction jet for plain (grey, orange, and blue colours) and streamlined (green colours) entrances 9(e), 9(f)).Both species swam along the jet path with the highest TKE values in the observation channel ( 0.2 < TKE / < 0.6 m / s < TKE < 0.008 m / s ).A few fish sought out a path away from the jet, with much lower values of TKE ( 0 < TKE / < 0.2 .001m / s ), skirting the jet just before heading to the slotted entrance (Figs.9(e), 9(f)).

Fig. 9 (
Fig. 9 (Color online) Fish movement trajectories (right to left), with flow (left to right), projected onto for selected trials of at

Fig. 10 (
Fig. 10 (Color online) First fish entry trajectories (percentages) for the streamlined entrance in relation to jet and nonjet areas for various depths ( ) d and = 0.15 m / s a V

Fig. A1 (
Fig. A1 (Color online) Normalised decay in maximum jet velocity along jet trajectory as function of relative distance for streamlined (blue symbols) and plain entrances (red symbols) for and various water depths, showing comparison with velocity decay in a single vertical slot pool[22]

Fig. S2 (
Fig. S2 (Color online) Detailed three-dimensional velocities and TKE result plots at the horizontal plane of for 0.08 m, 0.18 m and 0.32 m and 0.15 m/s, 0.50 m/s, plain entrance ( refers to normalised streamwise velocity, refers to normalised transverse velocity, refers to normalised vertical velocity, and refers to normalised TKE)

Table A4 Statistical results for interaction between attraction directions and velocities on Pre-entry location for Australian bass
a Streamlined entrance.