SEARCH

SEARCH BY CITATION

Keywords:

  • climate change;
  • habitat;
  • nocturnal activity;
  • passive integrated transponder;
  • predation;
  • sheltering behaviour

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. References

This study examines seasonal (winter v. summer) differences in space-time budgets, food intake and growth of Atlantic salmon Salmo salar parr in a controlled, large-scale stream environment, to examine the direction and magnitude of shifts in behaviour patterns as influenced by the availability of overhead cover and food supply. Salmo salar parr tested in the presence of overhead cover were significantly more nocturnal and occupied more peripheral positions than those tested in the absence of overhead cover. This increase in nocturnal activity was driven primarily by increased activity at night, accompanied by a reduction in daytime activity during winter. The presence of overhead cover had no effect on rates of food intake or growth for a given food supply in a given season. Growth rates were significantly higher for fish subjected to a high food supply than those subjected to a low food supply. Food supply did not affect the extent to which S. salar parr were nocturnal. These results were consistent between winter and summer. The use of riparian shading as a management technique to mitigate the effects of warming allows the adoption of more risk-averse foraging behaviour and may be particularly beneficial in circumstances where it serves also to increase the availability of food.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. References

The effects of climate change on freshwater ecosystems are predicted to be diverse and widespread and encompass alterations to temperature, flow and hydrological regimes, modifications to water quality and changes to the relative success of non-native species (Schindler, 2001; Boyer et al., 2010; Britton et al., 2010; Kirillin, 2010; Suen, 2010). Furthermore, climate change may exacerbate, confound or complicate existing problems affecting freshwater ecosystems (Durance & Ormerod, 2009; Ormerod et al., 2010) and has significant negative effects on freshwater fisheries (Ficke et al., 2007). Since fishes are poikilothermic animals affected strongly by ambient water temperature, the effects of climate change on their ecology and behaviour are likely to be particularly pronounced and widespread, with changes to temperature regimes being of fundamental importance.

The use of riparian tree shading to mitigate the effects of warming is of great interest to fisheries managers since shading provided by riparian vegetation is an important component in moderating stream temperature (Blann et al., 2002; Ebersole et al., 2003; Webb & Crisp, 2006). Greater levels of stream shading have been shown to reduce both maximum water temperature and the magnitude of daily temperature fluctuations relative to less shaded streams with positive influences on densities and biomass of redband trout Oncorhynchus mykiss gairdneri (Richardson) (Zoellick, 2004). Management strategies aimed at ameliorating warming, however, demand an understanding of probable effects extending beyond that concerned solely with moderating water temperature. For example, shading affects a range of factors besides temperature, one of which is food supply. Riley et al. (2009) reported that low levels of in-stream macrophyte growth as a result of shading by a closed tree canopy diminished macroinvertebrate production and therefore the availability of potential prey for juvenile salmonids. Similarly, low prey biomass for young Atlantic salmon Salmo salar L. was found to be associated with heavy shading from the riparian forest canopy by Ward et al. (2009). Conversely, terrestrial contributions to invertebrate drift from riparian trees may be crucial in supplementing food availability in situations in which aquatic prey are scarce (Ormerod et al., 2004). Therefore, despite the potential benefits of shading with respect to stream temperature, the effects of interactions between shading and food supply are unclear but are likely to affect other aspects of fish ecology that are linked to fitness, including space–time budgets.

In common with many animal taxa, fishes need to balance energy requirements and food intake (growth) against the costs of foraging and exposure to predation risk (mortality) in order to maximize fitness. The balance between growth and mortality is typically expressed via space–time budgets of habitat use. Growth may be maximized by foraging in a location offering the greatest net energy gain (Fausch, 1984) or at a time of day when food availability is greatest and foraging efficiency is highest (Fraser & Metcalfe, 1997). Mortality may be minimized by utilizing physical structures as shelters or refuges during periods of inactivity as an effective means of avoiding predators (Culp & Scrimgeour, 1993; Cowlishaw, 1997; Martín et al., 2003). Since refuges typically offer fewer foraging opportunities than more exposed locations resulting in a reduction in total food intake (Sih et al., 1988), habitats of intermediate quality typically offer a balance in growth and mortality. This is frequently achieved by animals modifying the extent and timing of foraging and refuge use (Gilliam & Fraser, 1987; Lima & Dill, 1990; Metcalfe et al., 1999; Allouche, 2002; Martín et al., 2003).

Salmo salar parr have become an important model for exploring the balance between growth and mortality in relation to habitat use and daily activity cycles (Fraser et al., 1993, 1995; Burns et al., 1997; Gries et al., 1997; Valdimarsson et al., 1997; Gries & Juanes, 1998; Metcalfe et al., 1999; Imre & Boisclair, 2004; Johnston et al., 2004; Orpwood et al., 2006) with fish typically utilizing the interstitial habitat for shelter when not foraging (Rimmer et al., 1983; Cunjak, 1988; Bremset, 2000). Salmo salar parr are highly reliant on visual acuity for capturing prey (Keenleyside, 1962; Wankowski, 1981; Stradmeyer & Thorpe, 1987) and suffer reduced foraging efficiency under low light conditions (Fraser & Metcalfe, 1997). Despite this effect, S. salar parr are frequently nocturnal, with the extent of day activity being inversely related to food availability (Metcalfe et al., 1999; Orpwood et al., 2006) and directly related to temperature (Fraser et al., 1993, 1995). The adoption of nocturnal activity appears to be an important anti-predator strategy, whereby fishes minimize their activity during the day when they may be particularly vulnerable to diurnal avian predators (Cramp et al., 1977; Cramp, 1985; Greenwood & Metcalfe, 1998; Valdimarsson & Metcalfe, 1998; Metcalfe et al., 1999).

The aim of this study was to determine how overhead cover affects growth and habitat use by S. salar parr. The combined effects of shading and food supply on space–time budgets and growth were investigated in a controlled multi-factorial design using replicated mesocosm test arenas across two seasons (winter v. summer). A fuller understanding of how overhead cover and food supply affects growth and habitat use by S. salar parr is necessary if the use of riparian vegetation is to become an important management strategy to mitigate the effects of warming without inadvertently affecting other aspects of their ecology.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. References

Experimental animals

The experiment used wild-reared S. salar parr that were the offspring of wild adults. In the autumn of 2003, wild sea-run adult fish were captured by electrofishing from the River Almond, Perthshire, U.K., and transported in a container of oxygenated river water to the Marine Scotland Almondbank Field Station, Perthshire. Full-sib families (n = 6) were created by fertilizing the eggs from six females with the milt from six males. Eggs were incubated over the winter at ambient water temperature. On 28 April 2004, the resulting fry from all six families were mixed together and released at the time of first feeding into the Tombane Burn (56° 32′ N; 3° 41′ W), a tributary of the River Braan, Perthshire, which did not contain S. salar due to an impassable waterfall. Fish were subsequently captured by electrofishing from the Tombane Burn on either 7 October 2004 (winter experiment; n = 32) or 5 May 2005 (summer experiment; n = 32) and transported to the Marine Scotland Almondbank Field Station in a container of oxygenated river water. After capture, fish were kept in a 1 m diameter flow-through holding tank and fed with thawed chironomid larvae until the start of the experiment. The day after capture, each fish was assigned a unique identity code by subcutaneous insertion of a passive integrated transponder (PIT) tag into the abdominal cavity.

Test arenas

Two replicate experiments (one each during winter and summer) were carried out in a glass-sided indoor stream channel at the Marine Scotland Almondbank Field Station. Each experiment was carried out over two runs, each of 2 week duration [winter run 1:12 to 26 October 2004; winter run 2:27 October to 10 November 2004; summer run 1:16 to 30 May 2005; summer run 2:30 May to 13 June 2005). Part of the stream channel was divided into 16 test arenas using stainless steel wire mesh screens. Each test arena was 0·9 m long and 0·6 m wide. Water from the River Almond flowed through the test arenas at ambient temperature [range: 5·0–9·0° C (winter); 6·5–17·5° C (summer)]. Daily water temperature [mean ±s.d.: 7·2 ± 0·7° C, n = 15 (winter run 1); 7·3 ± 1·0° C, n = 15 (winter run 2); 10·2 ± 1·5° C, n = 15 (summer run 1); 12·5 ± 1·8° C, n = 15 (summer run 2)] was calculated as the mean value of eight measurements per day taken 3 h apart from 0000 hours. Mean daily water temperature did not differ significantly between runs during the winter experiment (one-way ANOVA: F1,28 = 0·03, P > 0·05) but was significantly higher during run 2 of the summer experiment than during run 1 of the summer experiment (one-way ANOVA: F1,28 = 14·42, P < 0·01). Water depth (mean ±s.d.: 0·18 ± 0·02 m, n = 32) and flow rate (mean ±s.d.: 0·13 ± 0·04 m s−1, n = 32) were measured twice in each test arena at the end of run 2 in each experiment. Flow rate was measured 0·1 m above the substratum. The substratum of each test arena was formed from fine gravel and pebbles (c. 25 mm diameter) to minimize the availability of interstitial habitat (Rimmer et al., 1983, 1984; Bain et al., 1985; Cunjak, 1988; Gries & Juanes, 1998) and therefore discourage fish from using refugia other than that provided by experimental (PIT antenna-equipped) shelters (Wyre Micro Design Ltd; www.wyremicrodesign.co.uk).

An ambient photoperiod (56° N) was maintained during the experiments by supplementing natural light from windows between 0800 and 1700 hours during the winter experiment and between 0530 and 2100 hours during the summer experiment. Supplementary lighting was provided by 18 Son-T Agro sodium lamps (400 W each) positioned c. 2 m above the substratum and spread out evenly along the length of the stream channel. A single shelter comprising a tubular PIT antenna surrounded by a dark grey plastic shell was provided in each test arena (Orpwood et al., 2006). Shelters were buried in the substratum immediately adjacent to the glass side of the stream channel c. 0·4 m downstream of the upstream mesh screen in each test arena. Shelters were opaque, tubular in shape (115 mm total length ×55 mm internal diameter), open at both ends and positioned parallel to the direction of flow as S. salar parr prefer dark shelters with high levels of water flow (Valdimarsson & Metcalfe, 1998). Each shelter was constructed around a PIT antenna that was connected to a multi point decoder (MPD) unit (model number: DEC-MPD-16; Wyre Micro Design Ltd). The MPD was linked to a computer and was set to interrogate the PIT antennae once per min. The presence of a tagged fish in the shelter was automatically recorded by the computer in a data file that gave the day, date, time, test arena number and PIT tag code. Each record in the data file represented 1 min of shelter occupancy by a given fish.

The arrangement of test arenas was contiguous, with test arenas numbered sequentially from 1 (most upstream) to 16 (most downstream). Four treatments were tested: (1) overhead cover present, low food (test arenas 1–4); (2) overhead cover absent, low food (test arenas 5–8); (3) overhead cover present, high food (test arenas 9–12); (4) overhead cover absent, high food (test arenas 13–16). It was not possible to assign treatments randomly to test arenas because those subjected to a high food supply could not be situated upstream of those subjected to a low food supply. Although the arrangement of test arenas was contiguous, the indoor stream channel was designed specifically for the detailed study of fish behaviour on a semi-natural spatial scale, and so all test arenas were matched for possible confounding factors such as lighting. Observations were made from within a darkened hide immediately adjacent to the test arenas to avoid disturbance. Food supply (low v. high) was controlled by a custom built automatic fish feeder (ASU 2000) operated via a computer (MacLean et al., 2003). A single bucket containing a suspension of thawed chironomid larvae in river water formed the food reservoir for each test arena. The suspension of thawed chironomid larvae was maintained by bubbling air through the water in the food reservoir. Food reservoirs were restocked with fresh larvae daily between 0830 and 0930 hours by adding two cubes of frozen larvae to each reservoir [mean ±s.d. dry mass of a cube: 0·22 ± 0·02 g, n = 50; 99·31 ± 18·07 larvae g−1, n = 50 (MacLean et al., 2003)]. Larvae were delivered to each test arena via a black feeding tube (3·2 mm diameter bore) that emerged from the substratum 0·2 m from the upstream mesh screen and equidistant from each side of the test arena. The rate of delivery of larvae to test arenas was controlled by solenoid valves (Burkert, 24 V direct current, 3 mm bore; www.burkert.co.uk) which were set to open for 2·5 s every 3 min (high food supply) or every 30 min (low food supply) so that fish subjected to a high food supply received ×10 as many opportunities to feed as those subjected to a low food supply. The rate at which the food reservoirs drained was set based upon pilot trials and earlier work, indicating that the chosen high food supply regime (i.e. valves open for 2·5 s every 3 min) allowed the entire contents of a food reservoir to drain over a 24 h period. Thus, the low food supply regime (i.e. valves open for 2·5 s every 30 min) meant that only about a tenth of the contents of a food reservoir drained over a 24 h period. Food was therefore available throughout the diel cycle. Empirical data collected during behavioural observations of fish showed that the number of larvae per delivery (mean ±s.d.: 1·7 ± 1·6 larvae, n = 1825) ranged from 0 to 10, although the cumulative frequency distribution was positively skewed with >97% of deliveries containing ≤5 larvae.

Overhead cover (absent v. present) was determined by the amount of cover provided above the water surface. Test arenas designated as ‘overhead cover absent' had no cover provided above the water surface, while those designated as ‘overhead cover present' were partially covered by a sheet of black polyfoam positioned to shade approximately two-thirds of their surface area. The sheet of polyfoam had a hole cut in it to create an unshaded portion that extended along the length of a test arena in a strip parallel to the glass side, beginning 0·2 m from the glass side and ending 0·4 m from the glass side. Two very thin strips of the polyfoam cover, each c. 20 mm wide and positioned immediately adjacent to the mesh screens separating adjoining test arenas, linked the two shaded portions of a test arena. Thus, the unshaded portion of a test arena was situated above the feeding tube so that fish had to leave the vicinity of shade to intercept food items. The unshaded region was designated the central zone (C). The sheet of polyfoam rested on two pieces of taut 1 mm diameter tie wire (Netlon Sentinel plastic coated garden wire, Apollo Gardening Ltd; www.netlon.co.uk) c. 30 mm above the surface of the water so as not to impede water flow.

Experimental protocol

On the first day of each run (day 0), 16 fish were removed from the holding tank and initial fork length [LF1: mean ±s.d.: 75·6 ± 7·0 mm, n = 32 (winter); 83·6 ± 8·6 mm, n = 31 (summer)] and initial wet mass [MW1: mean ±s.d.: 4·6 ± 1·3 g, n = 32 (winter); 6·3 ± 2·2 g, n = 31 (summer)] were measured. Fish used in the winter experiment were significantly smaller than those used in the summer experiment (two-sample t-test assuming equal variances on log 10LF1 and log 10MW1: P < 0·001). Within an experiment, fish size did not differ significantly among treatments or between runs (GLM two-way ANOVA on log 10LF1 and log 10MW1: P > 0·05). A single fish was then introduced into each test arena and allowed to acclimate to experimental conditions overnight. Shelter use was recorded continually by the MPD for 13 consecutive 24 h periods. The timing of these 24 periods differed between the winter and summer experiments due to natural seasonal variation in absolute day length. Thus, shelter use was recorded from 0700 hours on day 1 to 0700 hours on day 14 during the winter experiment and from 0400 hours on day 1 to 0400 hours on day 14 during the summer experiment. In addition, five sets of behavioural observations were carried out on each day, beginning at 0930, 1030, 1130, 1330 and 1430 hours. Each set of observations involved observing each test arena for 3 min during which time it was noted whether or not the fish was visible. When the fish was not visible, the data file being generated by the computer connected to the MPD was checked to see whether the fish was in the sub-gravel shelter. When the fish was visible, feeding was recorded by counting the number of larvae in a delivery and noting the number eaten by the fish. When fish were not occupying the sub-gravel shelter, their position in the test arena (to the nearest 50 mm) was recorded by estimating the distance of the fish's head from the glass side of the stream channel. Fish were deemed to be occupying a central position in the test arena if they were observed between 0·2 and 0·4 m from the glass. All other positions in the test arenas were considered to be peripheral. In this way, a central position indicated that a fish was in the highly risky open part of the test arena where, regardless of treatment, there was no overhead cover. Behavioural observations were carried out using a rolling protocol, beginning observations at test arena 1, moving to test arena 2 and so on to test arena 16. The next set of observations started at test arena 2, moving to test arena 3 and so on to test arena 16, finishing with test arena 1. In practice, while five observation sets per day were planned, heavy sediment load in the water following rainfall constrained the number of observations. Out of a possible 65 observation sets per run, 56 and 54 were made in runs 1 and 2 of the winter experiment, respectively, while 54 and 55 were made in runs 1 and 2 of the summer experiment, respectively. On the last day of each run (day 14), fish were removed from the test arenas and killed prior to final fork length (LF2, mm) and final wet mass (MW2, g) being measured. Four replicates of each treatment were tested during each run, giving a total of eight replicates per treatment during both winter and summer. One fish, however, escaped during run 1 of the summer experiment, meaning that only seven replicates were obtained for the overhead cover absent, low food treatment. A total of 63 S. salar parr was used during the two experiments. Each fish was tested in only one run and subjected to just one combination of overhead cover and food supply during either winter or summer.

Statistical analysis

Fish activity (time spent out of shelter) was derived from the data generated by the PIT apparatus. Analyses considered data collected during 13 consecutive 24 h periods from 0700 hours on day 1 to 0700 hours on day 14 of each run for the winter experiment, and from 0400 hours on day 1 to 0400 hours on day 14 of each run for the summer experiment. For each fish during each 24 h period, day activity (AD; min), night activity (AN; min) and total activity (AT; min, where AT = AD + AN) were calculated by subtracting the amount of time spent sheltering from the amount of time available during the time period in question. Day was defined as the period of time between 0900 and 1600 hours during the winter experiment and between 0600 and 2100 hours during the summer experiment. Night was defined as the period of time between 1800 and 0700 hours during the winter experiment and between 2300 and 0400 hours during the summer experiment. Day and night were defined in this way to avoid considering PIT tag data collected during the outwith periods between true daytime (i.e. when supplementary lighting was provided in addition to natural light) and true night time (i.e. when no supplementary lighting was provided and there was no natural light). Thus, 20 h of data were used from each 24 h period to ensure that the results were truly representative of day and night behavioural patterns. To account for the difference in length of day and night both within and between experiments, AD, AN and AT were subsequently expressed as a percentage of the time available during the time period in question (day activity: D, %; night activity: N, %; total activity: T, %). Values of D and N were used to calculate a nocturnal activity index (INA), such that INA = 100 N (N + D)−1, thus expressing the extent to which fish were nocturnal as a ratio of nocturnal activity to total activity (Fraser et al., 1993). Analyses of D, N, T and INA used mean values calculated from 13 consecutive 24 h periods.

When not utilizing the sub-gravel shelter, the tendency of the fish to occupy the central zone (C) was calculated as the number of times the fish was observed in a central position as a proportion of the total number of positions the fish was observed occupying during behavioural observations. Similarly, daytime feeding was analysed from the observed number of larvae delivered to (FD) and eaten by (FE) each fish during behavioural observations. The percentage of observations during which fish were visible was high during both winter (mean ±s.d.: 89·0 ± 15·1%, n = 32 fish) and summer (mean ±s.d.: 93·3 ± 6·7%, n = 31 fish). The variability in the percentage of observations during which fish were visible, however, meant that an unequal number of observations were obtained for each fish during both winter [mean (range): 49 (18–55) observations per fish, n = 32 fish] and summer [mean (range): 51 (39–55) observations per fish, n = 31 fish]. Therefore, analyses of C,FD and FE used data pooled from all observations obtained for each fish. Specific growth rate (GM, % day−1) was calculated as GM = 7·1429[ln(MW2MW1−1)].

GLM two-way ANOVAs were used to test the effects of overhead cover (absent v. present) and food supply (low v. high) on D, N, T, INA, C, FE and GM. Preliminary analyses incorporating run number as an additional factor revealed no significant effects of run number during the winter experiment. A run effect was noted in summer due to the significantly higher mean daily water temperature experienced during run 2. Since the effects of overhead cover and food supply were consistent between runs, however, all analyses were conducted on pooled data from runs 1 and 2 during both the winter and summer experiments to aid clarity of presentation and interpretation. Data were examined for normality of residuals (Anderson–Darling test) and homogeneity of variances (Bartlett's or Levene's test). When necessary, an appropriate transformation was applied [C: arcsin square root; FE: log 10(x + 1)]. Figures are constructed from untransformed data. Statistical values are for two-tailed tests at P < 0·05 and adjusted for ties. Quoted times are GMT. Analyses were conducted using MINITAB 13 (Minitab Inc.; www.minitab.com) for Windows.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. References

Total activity

During winter, S. salar parr tested in the presence of overhead cover were significantly more active than those tested in the absence of overhead cover (F1,28 = 5·98, P < 0·05). Mean ±s.d. T for fish tested in the presence of overhead cover was 63·9 ± 17·9% (n = 16) compared with 49·9 ± 13·8% (n = 16) for fish tested in the absence of overhead cover during winter. No such effect was observed during summer (F1,27 = 0·90, P > 0·05) when mean ±s.d. T for fish tested in the presence of overhead cover was 59·6 ± 15·6% (n = 16) compared with 65·5 ± 19·4% (n = 15) for fish tested in the absence of overhead cover. T did not differ significantly according to food supply during either winter (F1,28 = 1·20, P > 0·05) or summer (F1,27 = 1·24, P > 0·05), nor were the interaction terms between food supply and overhead cover significant in either season (P > 0·05).

Nocturnal activity

The presence of overhead cover had a significant effect on the extent to which S. salar parr were nocturnal. During winter, fish tested in the presence of overhead cover were significantly more nocturnal than those tested in the absence of overhead cover [F1,28 = 9·76, P < 0·01; Fig. 1(a)]. This effect was driven by a significant reduction in day activity [F1,28 = 5·01, P < 0·05; Fig. 1(b)] and a significant increase in night activity [F1,28 = 22·40, P < 0·001; Fig. 1(c)]. Similarly, during summer, fish tested in the presence of overhead cover were also significantly more nocturnal than those tested in the absence of overhead cover [F1,27 = 11·85, P < 0·01; Fig. 2(a)]. This effect was driven by a tendency towards reduced day activity [F1,27 = 3·76, P > 0·05; Fig. 2(b)] and a significant increase in night activity [F1,27 = 4·88, P < 0·05; Fig. 2(c)].

image

Figure 1. Effects of overhead cover and food supply on (a) the nocturnal activity index, (b) day activity and (c) night activity of Salmo salar parr during winter. Mean ±s.e., averaged over 13 days, is presented for fish tested in the absence (inline image) or presence (inline image) of overhead cover and subjected to a low or high food supply (n = 8 in all cases). (a) A value of 50% indicates that the fish were equally likely to be active at night as during the day, a value of 0% indicates that the fish were active exclusively during the day and a value of 100% indicates that the fish were active exclusively at night.

Download figure to PowerPoint

image

Figure 2. Effects of overhead cover and food supply on (a) the nocturnal activity index, (b) day activity and (c) night activity of Salmo salar parr during summer. Mean ±s.e., averaged over 13 days, is presented for fish tested in the absence (inline image) or presence (inline image) of overhead cover and subjected to a low or high food supply (n = 8 except for those fish tested in the absence of overhead cover and subjected to a low food supply where n = 7). (a) A value of 50% indicates that the fish were equally likely to be active at night as during the day, a value of 0% indicates that the fish were active exclusively during the day and a value of 100% indicates that the fish were active exclusively at night.

Download figure to PowerPoint

Food supply did not affect the extent to which S. salar parr were nocturnal during either winter [F1,28 = 0·32, P > 0·05; Fig. 1(a)] or summer [F1,27 = 0·95, P > 0·05; Fig. 2(a)] with neither D nor N varying significantly according to food supply [P > 0·05; Figs 1(b), (c) and 2(b), (c)]. All interaction terms between overhead cover and food supply were non-significant (P > 0·05; Figs 1 and 2).

Position when not occupying a sub-gravel shelter

When not occupying a sub-gravel shelter, fish tested in the presence of overhead cover were observed in central positions on a significantly smaller proportion of occasions than those tested in the absence of overhead cover [winter: F1,28 = 21·51, P < 0·001; Fig. 3(a); summer: F1,27 = 5·53, P < 0·05; Fig. 3(b)]. The effect of food supply on the positions adopted by fish when not occupying a sub-gravel shelter differed between winter and summer. During winter, fish subjected to a high food supply were observed in central positions on a significantly smaller proportion of occasions than fish subjected to a low food supply [F1,28 = 8·83, P < 0·01; Fig. 3(a)]. The interaction between overhead cover and food supply was significant during winter [F1,28 = 5·04, P < 0·05; Fig. 3(a)], indicating that the effect of overhead cover differed according to food availability. In contrast, during summer, food supply did not have a significant effect on the proportion of occasions that fish were observed in central positions [F1,27 = 0·55, P > 0·05; Fig. 3(b)]. Furthermore, the interaction between overhead cover and food supply was not significant during summer [F1,27 = 0·05, P > 0·05; Fig. 3(b)], indicating that the effect of overhead cover did not differ according to food availability.

image

Figure 3. Effects of overhead cover and food supply on the spatial positions adopted by Salmo salar parr when not occupying a sub-gravel refuge during the day in (a) winter and (b) summer. Mean ±s.e. proportion of observations where fish were observed in the centre of test arenas is presented for fish tested in the absence (inline image) or presence (inline image) of overhead cover and subjected to a low or high food supply (n = 8 except for those fish tested in the absence of overhead cover and subjected to a low food supply in summer where n = 7).

Download figure to PowerPoint

Daytime feeding

Empirical data collected during the behavioural observations confirmed that the applied feeding regime had the desired effect on FD (high food supply: mean ±s.d.: 90·9 ± 35·4 larvae delivered to each fish, n = 32; low food supply: mean ±s.d.: 5·8 ± 6·3 larvae delivered to each fish, n = 31). Despite the large difference in the amount of food available between feeding regimes, food supply did not have a significant effect on FE (winter: F1,28 = 0·44, P > 0·05; summer: F1,27 = 2·46, P > 0·05). FE did not differ significantly according to the level of overhead cover (winter: F1,28 = 0·07, P > 0·05; summer: F1,27 = 0·02, P > 0·05). The interaction between overhead cover and food supply was not significant (winter: F1,28 = 0·72, P > 0·05; summer: F1,27 = 0·76, P > 0·05).

Specific growth rate

GM was significantly higher for fish subjected to a high food supply than fish subjected to a low food supply [winter: F1,28 = 37·24, P < 0·001; Fig. 4(a); summer: F1,27 = 27·15, P < 0·001; Fig. 4(b)]. GM did not differ significantly according to the level of overhead cover [winter: F1,28 = 0·02, P > 0·05; Fig. 4(a); summer: F1,27 = 0·42, P > 0·05; Fig. 4(b)]. The interaction between overhead cover and food supply was not significant [winter: F1,28 = 3·05, P > 0·05; Fig. 4(a); summer: F1,27 = 0·42, P > 0·05; Fig. 4(b)].

image

Figure 4. Effects of overhead cover and food supply on specific growth rate in mass (GM) of Salmo salar parr during (a) winter and (b) summer. Mean ±s.e. percentage wet body mass per day is presented for fish tested in the absence (inline image) or presence (inline image) of overhead cover and subjected to a low or high food supply (n = 8 except for those fish tested in the absence of overhead cover and subjected to a low food supply in summer where n = 7).

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. References

Salmo salar parr were more nocturnal in the presence than in the absence of overhead cover during both winter and summer. The adoption of a nocturnal activity pattern reduces the exposure of fishes to perceived risk from diurnal avian predators (Cramp et al., 1977; Cramp, 1985; Greenwood & Metcalfe, 1998; Valdimarsson & Metcalfe, 1998; Metcalfe et al., 1998, 1999). It might be assumed that the presence of overhead cover would itself provide protection from predators. Nevertheless, the presence of overhead cover caused increased nocturnal foraging, which, although likely to be less efficient than diurnal foraging (Fraser & Metcalfe, 1997), allowed fish to attain growth rates similar to those observed when shelter was absent. During winter, increased nocturnal foraging was driven by a significant increase in night activity and a significant reduction in activity during the day. Therefore, it appears that the presence of overhead cover allowed fish to reduce foraging effort during the most risky part of the diel cycle (Cramp et al., 1977; Cramp, 1985; Greenwood & Metcalfe, 1998; Valdimarsson & Metcalfe, 1998; Metcalfe et al., 1998, 1999) by foraging for longer at night to compensate. This may have been further facilitated by the presence of shelter causing a reduction in resting metabolic rate of the fish (Millidine et al., 2006) and so reduced overall food demands.

When not occupying a sub-gravel shelter, fish used central positions less in the presence than absence of overhead cover during both winter and summer. The presence of overhead cover therefore affected S. salar parr both temporally (by making them more nocturnal) and spatially (by reducing their use of more central locations). Both these effects might be expected to reduce growth via either a reduction in activity (and therefore presumably foraging effort) during the day when foraging efficiency is highest (Fraser & Metcalfe, 1997) and a reduction in occupancy of the most profitable central part of the test arena where the food outlet was located (Fausch, 1984). This was not the case, however, as the presence of overhead cover had no effect on either the observed number of larvae eaten by each fish or specific growth rate for a given food supply in a given season. Thus, the presence of overhead cover allowed S. salar parr to maintain a growth trajectory similar to that achieved by fish tested in the absence of overhead cover, despite adopting a more risk-averse pattern of activity and space use, by increasing activity (and therefore presumably foraging effort) at night.

Specific growth rates of S. salar parr were affected by food availability during both winter and summer. It is interesting to note in this regard, however, that food supply did not affect the number of larvae observed being eaten by fish throughout daytime observations. It is therefore reasonable to assume that fish subjected to a high food supply were able to eat more food items at night than fish subjected to a low food supply in order to attain a faster growth rate overall.

The present study revealed that food supply did not affect the extent to which S. salar parr were nocturnal during either winter or summer. Although this result contradicts earlier findings by Metcalfe et al. (1999) and Orpwood et al. (2006), it is important to view the results of the present study within the wider context of a generic balance between growth and mortality. As outlined by Orpwood et al. (2006), S. salar parr may respond to a reduction in food availability in one of four ways. First, there may be no response if neither growth nor foraging effort is affected by a reduction in food availability. Such a circumstance might be observed if both the pre and post-reduction level of food availability were sufficient for the fish under a particular set of circumstances. Second, growth rate could decrease if the level of exposure to food and feeding opportunities (via foraging effort) remains unchanged. Third, growth could remain unchanged if the level of exposure (foraging effort) increases to compensate. Finally, growth rate could decrease despite an increased level of exposure (foraging effort). As exemplified by the findings of Metcalfe et al. (1999) and Orpwood et al. (2006), relatively high food availability allows S. salar parr a greater degree of flexibility in temporal patterns of activity and growth which may simply not exist in situations of low food availability.

Greater food availability resulted in a reduction in the proportion of occasions fish were observed in central positions when not occupying a sub-gravel shelter, but only during winter. Furthermore, there was a significant interaction between overhead cover and food supply during winter, indicating that the effect of overhead cover differed according to food availability. In contrast, during summer, food supply had no significant effect on the proportion of occasions that fish were observed in central positions. The reason for this seasonal difference is not immediately obvious. One possible explanation, however, is that fish with little food available to them in winter were particularly vulnerable to a food intake deficit that may have resulted from sheltering and occupying less profitable peripheral foraging positions in their test arenas. Such fish may therefore have been forced to exploit the most profitable central portions of their test arenas almost exclusively when out of shelter. During summer, however, faster swimming speeds facilitated by warmer water temperatures (Webb, 1978) may have allowed even those fish subjected to a low food supply to occupy peripheral regions of their test arenas while out of shelter, for example by allowing a ‘sneaky feeding’ strategy (Höjesjöet al., 2005) to intercept drifting food items.

The effects of climate change are predicted to have significant negative effects on freshwater fisheries (Ficke et al., 2007) with alterations to temperature regimes likely to be of fundamental importance. An increase in ambient water temperature is likely to increase the extent to which fishes need to be active in order to attain sufficient food. Such increase in activity, manifest for example via increased diurnal activity and occupation of more profitable but more risky foraging patches, is likely to place individuals at higher risk of predation. Management strategies such as the use of riparian cover to moderate stream temperature (Blann et al., 2002; Ebersole et al., 2003; Zoellick, 2004; Webb & Crisp, 2006) may also affect food supply (Ormerod et al., 2004; Riley et al., 2009; Ward et al., 2009), thus affecting space–time budgets and therefore fitness. This study has shown that the presence of overhead cover allows increased adoption of nocturnal activity and peripheral space use patterns, both likely to reduce the exposure of S. salar parr to potential predators, suggesting great benefits of overhead cover. Thus, in addition to mitigating the effects of warming in streams (Blann et al., 2002; Ebersole et al., 2003; Zoellick, 2004; Webb & Crisp, 2006), the addition of riparian cover as a management technique to mitigate the effects of warming may also allow fishes to adopt more risk-averse foraging behaviour and may be particularly beneficial in circumstances where riparian cover increases the availability of food.

We thank D. Stewart and S. Mackay for stocking and catching fish, M. Miles, J. Muir and S. Keay for fish husbandry and assistance with the experimental work, P. Rycroft for building the shelters and offering advice about PIT equipment, and A. Ojanguren for helping to landscape the indoor stream channel and for helpful discussions. We are grateful to I. Winfield, N. Metcalfe and an anonymous referee for their constructive comments on the manuscript. J.E.O. was funded by a Cardiff School of Biosciences PhD studentship.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. References