Juvenile trout behavior in relation to piscivores, surface ice and light

These are data for the behavior of juvenile brown trout (Salmo trutta) when subjected to the threat of predation from piscivorous fishes at 4 C ("winter") in artificial flumes. There were three predator treatments (burbot Lota lota present, northern pike Esox niger present, no predator present), two light levels (dark, light) and two conditions for surface ice cover (surface ice present, surface ice absent).

We captured 76 juvenile brown trout (12 yearling and 64 one-year-old fish) in late September 2020, by electrofishing in the forest stream, Barlingshultsälven, Sweden (59°31.356'N 12°18.728'E). The fish were subsequently transported to the aquarium facility at Karlstad University and kept in four 200 L aquaria (~20 trout in each aquarium). Trout were acclimated to aquarium conditions for approximately 4 months before the experiment commenced. Water in the aquaria was constantly filtered (EHEIM 2217 Classic canister filter; Eheim GmbH & Co KG) and cooled (Teco TK 2000, Teco), and 25 % of the water in each aquarium was changed once a week. The water temperature in all aquaria was initially maintained at 11 °C, the temperature in the Barlingshultsälven when trout were captured. After one week of acclimatization, we reduced the water temperature by 1 °C/week, until temperatures reached 4 °C in the beginning of December. Photoperiod followed natural daylight cycles. During the acclimatization period, we fed trout approximately 1 % of their body mass three times a week with thawed, previously frozen, red chironomid larvae. At the beginning of the experiment, mean wet mass ± SD of the trout was 10.02 ± 4.66 g (min-max = 2.67-23.17 g). Mean total length ± SD was 109 ± 19 mm (min-max = 71-151 mm).

One burbot and one pike were used as predators. The burbot (53 cm and 1064 g) was captured using net cages during winter 2019, and was kept in a 4000 L flume until the start of this experiment. The pike (53 cm and 1179 g) was captured by angling in early October 2020 and kept in a second 4000 L flume. Burbot and pike were captured at the same location, at the mouth of the River Klarälven, close to the Lake Vänern, Sweden (59°21.905'N 13°33.075'E). Light conditions and the water temperature regime for these fish followed the same protocol as for the brown trout. One week prior to the experiment started, both burbot and pike were fed two thawed, previously frozen, 5 g brown trout.

This experiment was conducted in the mid-sections of three 7-meter-long stream flumes (Fig. 1). Glass windows on one side of the flumes enabled observations of fish during the behavioral trials. We demarcated the sections of the flumes used for the experiment with stainless steel mesh screens (mesh size 5.35 mm, thread 1 mm, 71 % open area). Two experimental arenas were constructed in each flume, separated by green plastic net screens (mesh size 5.35 mm, thread 1 mm, 71 % open area) attached to wood frames (95 cm wide and 60 cm high). Upstream and downstream arenas had areas of 95 × 50 cm and 95 × 130 cm, respectively. Burbot and pike were kept in the upstream sections, placed there 48 h before the experiment started. The third flume was kept completely free from burbot and pike and functioned as a predator-free control. Downstream arenas served as experimental arenas for the trout. Water depth in all the flumes was 25 cm, and the average water velocity was 5 cm/s. Water temperature was kept constant at 4 ± 0.1 °C throughout the experiment, and the substrate consisted of 5-20 mm gravel. In the middle of each trout arena, we put one large stone (~10 × 5 × 3 cm) to provide a focal point for the trout when foraging. We also attached a measuring scale to the flume’s glass panel, with tick marks to enable measurements of the upstream-downstream position of the trout.

We constructed one large shelter for the trout per stream by attaching a thin black opaque plastic sheet (975 × 190 × 5 mm) to three concrete blocks (10 × 16 × 3 cm). Four iron legs (7 cm) supported each concrete block. We positioned the concrete blocks at each end and in the middle of the shelters. The shelter structure was placed in upstream-downstream direction along the side opposite to the glass panels (Fig. 1). Surface ice cover was simulated by using pieces of transparent multiwall polycarbonate (6 layers, 30 mm thick in total, hereafter referred to as plastic boards), cut to fit in the stream flumes. These plastic boards reduced the illuminance by ~200 lx.

We conducted the experiment during January 2021. Prior to each trial, we removed six trout from the holding aquaria and sorted them into three size-matched pairs. Trout were anesthetized (benzocaine, 0.1 g/L-1), weighed and measured before the trials. Size differences within each pair did not exceed 15 mm or 3 g and trout sizes did not differ between the three predator treatments (One‐way ANOVA, p > 0.8 for both wet mass and total length). We placed one pair of trout into each of the three flumes. Thereafter, trout were left in the flumes overnight for c. 12 h before observations of fish started. Trout were not fed during the 48 h prior to the experimental trials. During the experiment, we kept the light regime at 17 h darkness and 7 h daylight, which reflects the natural daylight cycle for January in the area from which the trout originated. Trout spent c. 20 h in the stream flumes during each experimental trial. During the experimental trials, each pair of trout was video recorded (Canon XA10; Canon Inc.) during four 10-min-long recording sessions throughout the day. Two of the recording sessions were done in darkness (< 0.05 lx), and occurred early in the morning and late in the afternoon. The remaining two recording sessions were conducted in daylight (300 lx) in the morning and afternoon, between the two recordings in darkness. We used infrared illuminators (IR illuminator No. 40748, Kjell & Co Elektronik AB) and the infrared function on the camera to enable observations of fish in darkness. Four infrared illuminators were used for each flume, attached to wood posts 50 cm directly above the water surface. During two of the four recording sessions, one in darkness and one in daylight and either in the morning or in the afternoon, fish were subjected to simulated ice cover by placing the plastic boards directly on the water surface. We placed or removed artificial ice at least 5 h before any observations of fish behavior. We randomized whether trout were subjected to surface ice in the morning or in the afternoon. We fed trout one red chironomid larvae (> 10 mm long) every 15 s during the first three minutes of each recording session. Prey items were flushed with water through a funnel and delivered through a plastic tube, which entered the flume through an opening in the net screen in the middle of the upstream cross section of the trout arenas at a water depth of 5 cm. During the last 7 minutes of each recording session, trout were not fed but behavioral observations still continued. We decided to feed trout for three minutes to prevent fish from becoming satiated and thus be less willing to forage during the remaining recording sessions throughout the day. When four 10-min recording sessions (darkness/daylight ᵡ ice/no ice) had been carried out for all three predator treatments (burbot, pike, control), trout were removed from the flumes and replaced with new pairs of trout. In total, 12 pairs of trout were tested for each of the three predator treatments, resulting in 72 trout being used f0r this experiment. When all experimental trials had been carried out, all fish, including the pike and burbot, were returned to the sites where they were caught.

We quantified the following response variables: (1) if trout foraged or not; (2) the number of consumed prey; (3) if trout exhibited aggressive behaviors or not; (4) proportion of time that trout were active; (5) proportion of time that trout sheltered; (6) distance between the trout's anterior end and the upstream predator arenas (hereafter referred to as distance from predator arena, regardless of predator presence/absence) and (7) the average distance between the two trout in each pair.

Foraging behavior was quantified both as a binomial response, if trout within a pair foraged or not, and as the total number of prey that trout consumed (0 to 13 prey per trial). Aggression was measured as a binomial response, whether trout pairs exhibited aggressive behaviors or not. Activity was measured as the proportion of time that trout were actively moving, and shelter use was quantified as the proportion of time that trout spent under the overhead shelter. We analyzed all variables expressed as proportions as arcsine transformed proportions, based on the arithmetic mean for each pair of trout. Position of each trout in the upstream-downstream direction was measured every 15 s, and from these values we could estimate average trout distance from the predator demarcation and the average distance between trout in the upstream-downstream direction during every recording session. Trout distances were measured in body lengths, based on the average length of the two trout in each pair.

The Swedish Ethical Committee on Animal Welfare approved this study, both fish collection in the field (5.8.18-03819/2018; ID no. 001530) and the laboratory experiments (5.8.18-03818/2018; ID no. 001500).