Marine depth use of sea trout Salmo trutta in fjord areas of central Norway
Abstract
The vertical behaviour of 44 veteran sea trout Salmo trutta (275–580 mm) in different marine fjord habitats (estuary, pelagic, near shore with and without steep cliffs) was documented during May–February by acoustic telemetry. The swimming depth of S. trutta was influenced by habitat, time of day (day v. night), season, seawater temperature and the body length at the time of tagging. Mean swimming depth during May–September was 1·7 m (individual means ranged from 0·4 to 6·4 m). Hence, S. trutta were generally surface oriented, but performed dives down to 24 m. Mean swimming depth in May–September was deeper in the near-shore habitats with or without steep cliffs (2·0 m and 2·5 m, respectively) than in the pelagic areas (1·2 m). May–September mean swimming depth in all habitats was slightly deeper during day (1·9 m) than at night (1·2 m), confirming that S. trutta conducted small-scale diel vertical movements. During summer, S. trutta residing in near-shore habitat progressively moved deeper over the period May (mean 1·1 m) to August (mean 4·0 m) and then reoccupied shallower areas (mean 2·3 m) during September. In winter (November and February), individuals residing in the innermost part of the fjords were found at similar average depths as they occupied during the summer (mean 1·3 m). The swimming depths of S. trutta coincide with the previously known surface orientation of salmon lice Lepeophtheirus salmonis. Combined with previous studies on horizontal use of S. trutta, this study illustrates how S. trutta utilize marine water bodies commonly influenced by anthropogenic factors such as aquaculture, harbours and marine constructions, marine renewable energy production or other human activity. This suggests that the marine behaviour of S. trutta and its susceptibility to coastal anthropogenic factors should be considered in marine planning processes.
Introduction
In rivers with a two-way connectivity for fish to the sea, brown trout Salmo trutta L. 1758 populations often consist of both river-resident and anadromous individuals (Klemetsen et al., 2003). Anadromous S. trutta perform marine migrations in order to maximize their feeding opportunities and ultimately enhance their fitness (Jonsson & Jonsson, 1993). Although the better at-sea feeding opportunities can improve individual fitness, there are also stressors and risks associated with migration, including higher energy expenditures for locomotion and osmoregulation, increased predation risk and exposure to novel pathogens (Thorstad et al., 2016).
Marine survival and behaviour of salmonids are highly variable spatially and temporally, with some of the most influential factors being temperature, physiological state and fish size (Drenner et al., 2012). Salmo trutta is among the lesser studied salmonids in the marine environment (Drenner et al., 2012) and present knowledge of the marine distribution and behaviour of the species is incomplete, limiting ability to assess potential threats and implement appropriate conservation measures for this valued species.
In Norway, potential marine stressors and threats for S. trutta, such as salmon lice Lepeophtheirus salmonis infections associated with Atlantic salmon Salmo salar L. 1758 farming activity (Thorstad et al., 2015), may not only vary in time and among sites, but also with depth (Heuch et al., 1995). Knowledge of the use of the water column by S. trutta in the marine environment, however, is limited (Rikardsen et al., 2007; Gjelland et al., 2014; Aarestrup et al., 2015). Further knowledge of marine migration, habitats occupied and depth use of S. trutta will help in implementing appropriate conservation measures.
The aim of this study was to use acoustic telemetry to examine spatial and temporal variation in marine depths occupied by free-ranging wild S. trutta in the Hemnfjord–Snillfjord system in Norway. At this site the horizontal movements and selected marine behaviour of S. trutta individuals included in this study were previously analysed (Eldøy et al., 2015). Marine residence time and area use varied among S. trutta individuals and marine behaviour was influenced by individual morphology and life-history characteristics (Eldøy et al., 2015). The previous analysis, however, did not consider use of the water column by S. trutta. Lack of previous knowledge about S. trutta habitat occupancy at sea made accurate predictions difficult and simple null-hypotheses (no effect) were used as background for the different habitat occupancy comparisons (estuary, pelagic, near shore with and without steep cliffs), time of day comparisons (day v. night) and inter-season comparison (month during summer, summer v. winter) in the present study. Furthermore, the null-hypothesis that the variance in individual daily swimming depth did not vary among habitats during day or night was tested.
Materials and methods
Study area
The study was performed in a fjord system in Sør-Trøndelag County, central Norway. The fjord's inner part is divided in two interconnected fjord arms, the Hemnfjord and Snillfjord. Together, the two inner fjords cover >60 km2 of sea surface and have 65 km of shoreline (Fig. 1). The fjord system is connected to the open sea through a 36 km long strait. Water column depths in the study area ranged from c. 0–100 m in the near shore areas, to a maximum of c. 400 m in the deepest parts. The study was conducted from 22 April to 4 December 2013 and from 4 February to 1 March 2014. Summer was defined as the period from May to the end of September and winter as November to the end of February (but see below for small site-specific variations in these definitions). The other months of the study were considered transitional periods between seasons.
, estuary; 
, near shore area without steep cliffs; 
, near shore area with steep cliffs; 
, pelagic area) and aqua culture locations (
) in the study area. Data loggers are indicated by (T) temperature and (S) salinity. There were two active S. salar farms within the fjord system during the study (Fig. 1). Weekly salmon lice counting conducted by the farmers showed average lice values ranging from 0 to 1·25 salmon lice in motile stages and from 0 to 0·3 adult female salmon lice per S. salar in the salmon pens within the study period.
Environmental variables
Three temperature and salinity recorders (DST milli-CT; www.star-oddi.com), recordings stored every 10th minute, were deployed at 1 m depth in the fjord system. The first was 1 km from the mouth of the River Søa in the Hemnfjord (array H1), the second 600 m from the mouth of the River Snilldalselva in the Snillfjord and the third at the middle receiver of the outermost receiver array (array H3; Fig. 1). Profiles of conductivity and temperature were recorded from 0 to 15 m in the central part of Hemnfjord on 1 May and 15 August 2013 using an SD204 conductivity–temperature–depth recorder (SAIV A/S; www.saivas.no).
Fish capture and tagging
Forty-four S. trutta that had previously undertaken marine migrations (e.g. were veteran migrants, of which some were previous spawners and some had not spawned before) were tagged with individually coded ADT-9-long acoustic transmitters equipped with depth sensors (9 mm× 39 mm, mass in air 6·8 g, minimum tag duration 267 days, depth resolution 10 cm, max depth 100 m, power output 146 dB re 1 μPa at 1 m, nominal delay 30–90 s; Thelma Biotel AS, www.thelmabiotel.com). The S. trutta came from two watercourses draining into the fjord. At the mouth of River Snilldalselva, 15 S. trutta (mean total length, LT = 381, s.d. = 53, range 275–460 mm) were captured using constantly monitored gillnets (35–42 mm mesh width) and tagged during 22–23 April 2013. An additional 29 S. trutta (LT mean = 417, s.d. = 55, range 330–580 mm) were captured using gillnets and tagged at the mouth of River Søa during 3–12 May 2013. The tagged S. trutta had a mean body condition factor of 0·95 (s.d. = 0·13, range 0·75–1·33). Based on scale growth-ring deposition patterns, the S. trutta had a mean LT at smoltification (back-calculated LT assuming linear relationship between body length and scale growth, Závorka et al., 2014) of 143 mm (s.d. = 39, range 96–246 mm), a mean smolt age of 2·4 years (s.d. = 0·6, range 2·0–4·0 years), had undertaken a mean of 3·1 previous marine migrations (s.d. = 0·9, range 2·0–6·0) and had a mean age of 5·6 years (s.d. = 1·0, range 4·0–8·0) at the time of tagging. The tags were surgically inserted through a 1·5–2·0 cm incision into the body cavity of S. trutta anaesthetized with 2-phenoxyethanol (EC No. 204-589-7; Sigma; www.sigmaaldrich.com; 0·5 ml l−1 water). The recovered S. trutta were released close to their capture site (Eldøy et al., 2015).
Tracking of tagged S. trutta
The tagged S. trutta were tracked using 39 VR2W acoustic receivers (Vemco, Inc.; www.vemco.com) moored 5 m below the sea surface (Fig. 1). The receivers operated from 22 April to 1 December 2013 and again from 6 February 2014 to 1 March 2014. The distance between receivers in the receiver arrays ranged between 332 and 660 m (Array H1, H2, H3 and S1; Fig. 1). The receivers were not operational during December—January due to logistical constraints. Tag detection range was tested at the central receiver of array H1 (Fig. 1) on 22 August 2013 (calm, clear weather, high tide) by deploying acoustic transmitters similar to those used in the study at 3 and 5 m depth and at increasing distance to the receiver in steps of 50 m. Maximum detection range was 350 m at both depths. The detectability at acoustic monitoring arrays H1 and H2 (Fig. 1) were further investigated by controlling if any S. trutta had passed the array without being registered. Here, all the individuals were registered at the investigated arrays prior to first registration outside the arrays, indicating that the arrays successfully detected passing S. trutta.
Data filtering
Salmo trutta depths could only be recorded when a tagged individual was within the range of an acoustic receiver. Before use, the depth sensing acoustic tags were tested at water surface and at 5 m depth. The results indicated that there was no need for subsequent calibration of the recorded depth data. The 42 S. trutta generated 1 161 998 valid detections during the study. In addition, there were 1137 registrations containing false transmitter identifications (ID), which were discarded. Two S. trutta were never recorded. As 136 821 (12%) of the registrations were suspected to be from either dead S. trutta or a shed tag (Thorstad et al., 2013), these registrations were excluded from analyses. A data filter was added to the data from the two receivers in the mouth of the River Søa and the three innermost receivers in Snillfjord, because these receivers contained higher frequencies of false detections, probably due to code collisions when a large number of tagged S. trutta were within the receiver range simultaneously. The data filter required at least two registrations from a tagged individual within a time span of 10 min to accept the registrations, which excluded an additional 16 881 (1·5%) of the registrations from analyses.
Defining habitat at receiver locations
The habitat at each receiver (i.e. the area within its range) was categorized as either estuarine, near shore without steep cliffs, near shore with steep cliffs or pelagic habitat (Fig. 1). Receivers deployed near river mouths in the fjord were highly influenced by the freshwater outflow and categorized as estuarine habitat. Receivers deployed near shore (<200 m), or in areas with shallow water (<15 m depth), were defined as near-shore habitat without steep cliffs. Receivers deployed over deep water with steep cliffs, but no shallow areas, along the shoreline were described as near-shore habitat with steep cliffs. Finally, receivers deployed over deep water, >600 m from the shore and without shallow areas (<15 m depth) within the receiver range were defined as pelagic habitat.
Data analyses
Analyses of swimming depth during summer for S. trutta tagged at the River Snilldalselva covered the period 1 May–1 October 2013 and for S. trutta tagged at the River Søa from tagging (3–12 May) until 1 October 2013. Analyses of swimming depth during winter covered the periods 1 November–1 December 2013 and 6 February–1 March 2014 (acoustic receivers were not deployed during 4 December 2013–6 February 2014). To distinguish between summer and winter residence at sea, data from October were not included because some individuals were showing transitional behaviour and only spent a few days at sea, probably due to spawning. Swimming depths were compared among habitats and between day and night. Day time was defined as the time between sunrise and sunset using the calculator of the Astronomical Applications Department of the US Naval Observatory (aa.usno.navy.mil) for the coordinates 63° 22′ 00·0″ N; 9° 13′ 00·0″ E. Night was defined as the time between sunset and sunrise.
All statistical analyses were conducted using R Studio 1.0.44 (RStudio Team, 2015). For analysis of factors influencing the swimming depth of S. trutta during summer, a mixed-effect model was constructed using the lme function in the R package lnme (Pinheiro et al., 2017), where S. trutta ID was assigned as a random factor. The initial global model included log-transformed swimming depth as the response variable, habitat, body length, month during summer, water temperature and time of day (day v. night) as fixed effects, as well as fish ID as a random effect. Akaike's information criterion was used to identify the best fitting model using the dredge function in the R package MuMin (Barton, 2013). The function cor.test in R was used to evaluate whether S. trutta LT correlated with the individual difference in swimming depth between day and night. For analysis of difference in means between two groups, paired t-tests were conducted assuming unequal variance between groups. For non-normally distributed data, paired Mann–Whitney U-tests were applied for comparisons of two groups. To test for differences in swimming depth among habitats, Tukey's HSD (R-package Multcomp; Hothorn et al., 2008) tests were conducted on a mixed-effects model (lme function in the R package lnme; Pinheiro et al., 2017) including log-transformed swimming depth as response variable, habitat as fixed effect and fish ID as random effect. Individual mean values were used to calculate the population mean values in order to keep the data points independent.
The proportions of detections at different depths were compared between the littoral and pelagic habitat during summer. Individual daily movement activity within habitats during day and night, hereafter referred to as vertical movement activity (VMA), was investigated by calculating individual standard deviation values from daily mean swimming depth. VMA values generated from <10 registrations were excluded from further analyses. To test for significant differences in VMA among habitats and between day and night, Tukey's HSD (R-package Multcomp; Hothorn et al., 2008) were conducted on a mixed-effect model (lme function in the R package lnme; Pinheiro et al., 2017) including daily s.d. of individual swimming depth as the response variable, Julian day, habitat and time of day (day v. night) as fixed effects and fish ID as a random effect. In order to account for data heterogeneity, a combination of variance structures was added to the model (varComb function in the R package lnme; Pinheiro et al., 2017), combining the varConstPower for Julian day and the varIdent function for habitat. The data and residuals of the models and groups in the statistical tests were visually inspected to check for assumption violations.
Results
Mean daily water temperature during summer at 1 m depth in the outer part of the study area was similar to or slightly lower than that in the inner parts of both Hemnfjord and Snillfjord (Table I). Water temperatures increased towards the end of July and thereafter declined (Fig. 2). Salinities at 1 m depth during summer were brackish in both the outer study area and in the inner parts of Hemnfjord and Snillfjord (Table I). The water column temperature and salinity in the central Hemnfjord were relatively stable from 0 to 15 m depth in both April (5° C; salinity 33) and August (13° C; salinity 34–37, Fig. 3).
| Temperature (°C) | Salinity | ||
|---|---|---|---|
| Mean ± s.d. | Range | Mean ± s.d. | |
| May–September 2013 | |||
| Inner parts of Hemnfjord | 13·4 ± 2·9 | 4·8–18·3 | 26 ± 7 |
| Inner parts of Snillfjord | 12·9 ± 3·0 | 5·3–17·9 | 24 ± 5 |
| Outer areas | 12·6 ± 2·5 | 5·1–16·2 | 21 ± 2 |
| November 2013 | |||
| Inner parts of Hemnfjord | 9·3 ± 0·6 | 6·7–10·8 | 31 ± 2 |
| Inner parts of Snillfjord | 8·0 ± 2·3 | 2·0–10·8 | 16 ± 7 |
| Outer areas | 8·8 ± 0·8 | 6·1–10·2 | 24 ± 1 |
| February 2014a | |||
| Inner parts of Hemnfjord | 6·9 ± 0·3 | 5·8–7·7 | 35 ± 0 |
- a 4 February to 1 March.
), inner part of Snillfjord (
) and outer part of the study area (
) in the period 1 May 2013–1 1 October 2013.
) and 22 August 2013 (
).The temperature at 1 m depth during November was similar in the different fjord areas. The water was less saline in the inner parts of Snillfjord, than the other areas during November (Table I). The mean water temperature at 1 m depth in the inner parts of Hemnfjord was 6·9° C and was close to full saline sea water during 4 February–1 March (Table I). Temperature and salinity were not recorded at the outer site and in Snillfjord in February due to logistical constraints.
The four best mixed effect models all indicated that the swimming depth of S. trutta during summer was influenced by habitat, time of day (day v. night) and month, while the best model (marginal R_GLMM2 = 0·37, P < 0·001) also included water temperature and LT as explanatory variables (Tables II and III). Mean swimming depth for all S. trutta in all habitat types during summer was 1·7 m (s.d. = 1·3, range of individual means = 0·4–6·4). The deepest depths recorded for the different individuals varied between 4·6 m and 24·0 m (mean of deepest recordings from all individuals 14·1 m, s.d. = 4·9). In the near-shore habitat without steep cliffs, S. trutta moved progressively deeper from May until August, but thereafter moved closer to the surface again. The number of S. trutta recorded in the pelagic and the near-shore habitat with steep cliffs, however, was low from June to September (Fig. 4). The S. trutta resided slightly deeper during day than at night within all types of marine habitats (P < 0·05; Table IV and Fig. 5). LT did not correlate with the swimming depth difference between day and night (Pearson's correlation = 0·08, n = 40, P > 0·05).
| Model | AIC | ΔAIC | AIC weights | d.f. |
|---|---|---|---|---|
| [D, H, M, L, T] | 1643565 | 0 | 0·998 | 11 |
| [D, H, M, T] | 1643577 | 12·31 | 0·002 | 10 |
| [D, H, M, L] | 1651227 | 7662·29 | 0 | 10 |
| [D, H, M] | 1651241 | 7676·01 | 0 | 9 |
| [] | 1981745 | 338 180·32 | 0 | 3 |
- a The models estimate the relative contributions to the swimming depth from the variables time of day (day v. night, D), habitat (H), month during summer (M), body length (L) and water temperature (T). AIC is the score based on Akaike's information criterion.
| Model | Effect | Estimate (95% C.I.) | t | d.f. | P |
|---|---|---|---|---|---|
| Swimming depth during summer | (Intercept) | 0·53 (0·41, 0·65) | 8·91 | 825266 | <0·001 |
| Habitat: Estuary | −0·73 (−0·75, −0·71) | −68·58 | 825266 | <0·001 | |
| Habitat: Near shore without cliffs | −0,03 (−0·05, −0·01) | −3·22 | 825266 | <0·05 | |
| Habitat: Pelagic | −0·44 (−0·47, −0·41) | −29·49 | 825266 | <0·001 | |
| Body length | 0·18 (0·10, 0·27) | 4·37 | 38 | <0·001 | |
| Month: May | −0·70 (−0·71, −0·70) | −165·66 | 825266 | <0·001 | |
| Month: June | −0·50 (−0·50, −0·50) | −182·7 | 825266 | <0·001 | |
| Month: July | −0·11 (−0·11, −0·10) | −41·63 | 825266 | <0·001 | |
| Month: September | −0·24 (−0·24, −0,23) | −90·97 | 825266 | <0·001 | |
| Water temperature | 0·14 (0·13, 0·14) | 87·3 | 825266 | <0·001 | |
| Time of day: Night | −0·36 (−0·37, −0·36) | −220·53 | 825266 | <0·001 | |
| VMA | (Intercept) | 0·72 (0·52, 0·92) | 7·08 | 5137 | <0·001 |
| Julian Day | 0·00 (0·00, 0·00) | 13·15 | 5137 | <0·001 | |
| Habitat: Estuary | −0·39 (−0·58, −0·20) | −4·09 | 5137 | <0·001 | |
| Habitat: Near shore without cliffs | −0·04 (−0·23, 0·14) | −4·46 | 5137 | >0·05 | |
| Habitat: Pelagic | −0·13 (−0·41, 0·16) | −0·88 | 5137 | >0·05 | |
| Time of day: Night | −0·16 (−0·18, −0·14) | −16·69 | 5137 | <0·001 |
), the interquartile ranges (box) and the 5th and 95th percentiles (whiskers) and outliers (O).| Habitat | Mean swimming depth (m) | W | P | ||
|---|---|---|---|---|---|
| Day | Night | Difference | |||
| Estuary | 1·0 | 0·7 | 0·3 | 399 | <0·001 |
| Near shore without steep cliffs | 2·1 | 1·3 | 0·8 | 805 | <0·001 |
| Near shore with steep cliffs | 2·7 | 1·7 | 1·0 | 45 | <0·05 |
| Pelagic | 1·3 | 0·8 | 0·5 | 257 | <0·05 |
| Total marine | 1·9 | 1·2 | 0·7 | 794 | <0·001 |
- a Paired Mann–Whitney U-test P-values describe level of statistical significance between the swimming depths during day and night.
) and night (
) during summer (1 May–1 October 2013). The box-and-whisker plots show median values (
), the interquartile ranges (boxes) and the 5th and 95th percentiles (whiskers) and outliers (
).The Tukey HSD post hoc test on the mixed-effect model of swimming depth revealed that the S. trutta had greater swimming depth in both near-shore habitat without (mean = 2·0, s.d. = 1·5, range = 0·4–6·7 m), and with steep cliffs (mean = 2·5, s.d. = 1·6, range = 0·3–7·1 m), compared with the estuarine habitat (mean = 0·9, s.d. = 0·5, range = 0·4–2·6 m, P < 0·001). The Tukey HSD post hoc test also showed that S. trutta were recorded deeper in both the near-shore habitat without steep cliffs and the near-shore habitat with steep cliffs compared with the pelagic habitat (mean = 1·2, s.d. = 1·1, range = 0·1–4·9 m, P < 0·001). Finally, the Tukey HSD post hoc test revealed that swimming depths of S. trutta were greater in the pelagic habitat than in the estuarine habitat (P < 0·001) and deeper in the near-shore habitat with steep cliffs compared with near-shore habitat without steep cliffs (P < 0·001).
The depth distribution of individual registrations varied between habitats (Table V). Vertical movements were generally larger during day than night (P < 0·001; Table VI). The Tukey HSD post hoc tests on the mixed effect VMA model (Table III) revealed that the VMA were greater in the near-shore habitats compared with the estuarine and pelagic habitats during day (P < 0·01). Similary, the Tukey HSD post hoc tests showed that the VMA were greater in the near-shore habitat without steeps cliffs compared with the estuarine and pelagic habitat (P < 0·001) and greater in the near-shore habitat with cliffs compared with the pelagic habitat (P < 0·05) during night.
| Depth (m) | Estuarine habitat (n = 29) | Near-shore habitat without steep cliffs (n = 40) | Near shore habitat with steep cliffs (n = 15) | Pelagic habitat (n = 28) | |
|---|---|---|---|---|---|
| 0–0·5 | Mean | 8 | 10 | 7 | 19 |
| Range | 0–47 | 0–46 | 0–71 | 0–78 | |
| 0·5–1·5 | Mean | 76 | 42 | 22 | 57 |
| Range | 41–96 | 1–70 | 0–62 | 13–100 | |
| 1·5–3·0 | Mean | 14 | 25 | 42 | 14 |
| Range | 2–46 | 0–48 | 6–88 | 0–62 | |
| 3·0–5·0 | Mean | 1 | 12 | 14 | 4 |
| Range | 0–13 | 0–52 | 0–50 | 0–28 | |
| 5·0–10·0 | Mean | 1 | 9 | 13 | 4 |
| Range | 0–20 | 0–87 | 0–78 | 0–33 | |
| >10 | Mean | 0 | 1 | 1 | 1 |
| Range | 0–3 | 0–19 | 0–6 | 0–13 |
Six individuals were recorded at near-shore or river mouth receivers in the innermost parts of Hemnfjord (two individuals) and Snillfjord (four individuals) during winter. Despite winter temperatures, the fjord was not ice covered. Mean individual swimming depth was 1·3 m (s.d. = 0·5, range 0·5–1·9. The deepest depths recorded during winter for these individuals varied between 9·7 m and 34·2 m (mean of deepest recordings 20·9 m, s.d. = 9·5). The S. trutta had greater individual mean swimming depth during day (mean = 2·0 m, s.d. = 0·6 m, range 1·3–2·9 m) than at night (mean = 1·0 m, s.d. = 0·5 m, range 0·3–1·7 m) also during the winter period (Paired t-test, t > 0·05, d.f. = 5, P < 0·001).
Discussion
Swimming depths of S. trutta individuals in this study were generally surface oriented, but influenced by habitat (estuary, pelagic, near shore with or without steep cliffs), time of day (day v. night), month of summer, water temperature and S. trutta size. Salmo trutta had deeper swimming depths in near-shore areas compared with pelagic areas. The swimming depth in estuaries was shallower than in other habitats, possibly influenced by the shallow nature of these areas. The S. trutta followed in winter showed similar depth-use patterns in the winter as during the summer, but it should be noted that the number of tagged individuals detected at sea during winter was low. The surface-orientation of S. trutta in this study is similar to that observed for electronically tagged, free-ranging wild S. trutta in previous studies. Rikardsen et al. (2007) found that S. trutta stayed close to the surface in the Alta fjord in northern Norway, recording a mean swimming depth of 1·8 m (eight S. trutta tagged with data storage tags). Gjelland et al. (2014) observed that S. trutta infested with salmon lice (L. salmonis) were mainly recorded between 1 and 3 m depth in the Etnefjord, south-western Norway. The present study confirms that this surface orientation is exhibited by the species across a range of different marine habitat types and remains consistent across seasons.
| Habitat | Time of day | Paired Mann–Whitney U–test | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Day | Night | Hypothesis (H0) | W | n1 | n2 | P | |||
| Mean (m) | Range (m) | Mean (m) | Range (m) | ||||||
| Estuary | 0·4 | 0·2–1·3 | 0·2 | 0·0–0·7 | VMA day ≤ VMA night | 369 | 29 | 28 | <0·001 |
| Near shore without steep cliffs | 0·8 | 0·3–1·8 | 0·4 | 0·1–1·0 | VMA day ≤ VMA night | 814 | 40 | 40 | <0·001 |
| Near shore with steep cliffs | 0·9 | 0·4–1·3 | 0·4 | 02.–0·6 | VMA day ≤ VMA night | 45 | 15 | 9 | <0·001 |
| Pelagic | 0·5 | 0·0–1·6 | 0·2 | 0·0–0·6 | VMA day ≤ VMA night | 186 | 27 | 21 | <0·001 |
- a Paired Mann–Whitney U-test statistics describe differences in VMA of S. trutta during day and night.
Salmo trutta are vulnerable to infestation with salmon lice because of their surface-oriented behaviour, which overlaps with the distribution of salmon lice larvae (Heuch et al., 1995; á Norði et al., 2016). On the other hand, in coastal areas with a surface layer of fresh water, owing to high influx of fresh water from rivers, S. trutta may be protected against salmon lice due to their surface orientation, because salmon lice do not cope well with freshwater and brackish water (Wootten et al., 1982; Johnson & Albright, 1991; Bricknell et al., 2006). The local S. salar farmers reported varying salmon lice concentrations in the pens during the study period, indicating that the S. trutta individuals probably experienced variable lice infection pressure depending on their spatial and temporal use of the marine habitat. Salmon lice infestation has earlier been found to alter the marine behaviour of S. trutta regarding timing of return to estuaries and rivers (Birkeland, 1996; Bjørn et al., 2001; Gjelland et al., 2014). Furthermore, Bui et al. (2016) found that farmed S. salar in salmon pens altered their swimming depth in response to high infestation rates of salmon lice. It remains unknown, however, whether depth distributions of S. trutta individuals observed in the current study were in whole or in part a response to salmon lice infestation, as their actual infection rates could not be documented.
Despite the statistical significance of the diel movements of S. trutta observed in this study, the biological significance of these movements is unclear since the difference in swimming depth between day and night was not large (1 m or less). The small difference in depth use between day and night could be explained by the minor differences in light between day and night during mid-summer due to the bright summer nights at the latitude of the study area. The small-scale difference in depth use between day and night, however, was consistent among seasons, suggesting it is a general pattern rather than resulting from summer light conditions in the study area.
Vertical day–night movements are common in many aquatic taxa (Hays, 2003). The phenomenon is thought to be governed by light (Appenzeller & Leggett, 1995) and earlier studies have suggested that such migratory behaviour might be triggered by body temperature regulative behaviour, feeding activity and anti-predator behaviour (Clark & Levy, 1988). Diel vertical movement patterns have previously been recorded by Davidsen et al. (2008) for S. salar post-smolt and they suggested that the observed vertical movement was a result of a trade-off between avoiding predation by avian and marine predators, feeding or osmoregulatory advantages. This might also be the case for the S. trutta in the present study. As the size of the veteran S. trutta tagged in the present study was larger than post-smolt of S. salar, the tagged individuals in the present study might have shown less anti-predator behaviour than the S. salar post-smolt in the study by Davidsen et al. (2008). No effect of body size on diurnal changes in swimming depth was found. This may suggest that although large individuals are less prone to predation than smaller ones, predation still has highly undesirable effects on individual fitness and that antipredator behaviour therefore may be a basic instinct that remains active through all S. trutta life stages.
Pemberton (1976) investigated the diel feeding of S. trutta and found increased intake of mid-water and surface prey and a decrease in benthic prey during the night. They suggested that S. trutta were more likely to detect prey that were above the substratatum or at the surface at night and that there could be varying conspicuousness of littoral prey between day and night. In the present study, an increased variability in depths used during day compared with at night was observed, possibly reflecting episodic benthic feeding activity, at least in the near-shore and estuary habitats. Hence, the difference between day and night in the present study might partly be explained by a shift in prey type and be linked to changes in the detectability of the different prey.
The increased swimming depth from May towards August coincided with an increase in water temperature. Rikardsen et al. (2007) observed similar patterns in the Altafjord in northern Norway, where S. trutta progressively resided in deeper water as temperatures increased during the month of June. It might be suggested that S. trutta actively regulated their ambient temperature by seeking water layers of preferred temperatures, as earlier suggested by Rikardsen et al. (2007) and Jensen et al. (2014). Jensen et al. (2014) suggested that S. trutta actively sought out the warmest areas in the fjord. In contrast, S. trutta in the present study tended to reside deeper when the temperature increased, suggesting that they moved into colder water during the warmest periods in summer. The study of Jensen et al. (2014) was performed in the northern portion of the species range in an area with lower sea temperatures, which may explain the species differences between the two sites. The warmest temperatures (16 and 18° C) recorded at 1 m depth in this study were in the outer part of the study area and in the inner parts of both Hemnfjord and Snillfjord, respectively.
Rikardsen et al. (2007) reported that individual mean ambient temperatures for S. trutta in the Alta Fjord in June and July ranged between 12 and 13° C. Another possible reason for the deeper swimming towards late summer might be a shift in prey type, or that the prey also moved deeper due to changing water temperatures. Stomach analyses of S. trutta caught in previous studies showed that polychaetes and marine crustaceans were important early in the season, while fishes were more important during late summer (Knutsen et al., 2001; Rikardsen et al., 2006). This may suggest that prey type varies with seasonal changes in prey availability and potentially explain the difference in swimming depth over the summer.
In conclusion, S. trutta were surface-oriented during their marine migration, both during summer and also in winter. Body size influenced the depth use of S. trutta during summer. Body size also influenced horizontal movements of the same individuals, showing that individual morphology and life history influenced the marine behaviour of S. trutta (Eldøy et al., 2015). Slight differences in depth use among individuals and habitats, between day and night, as well as during the summer season, were observed. These are clearly important to the animals, but the biological significance of these small changes are at present unclear. Water temperature was positively correlated with swimming depth, which might suggest that S. trutta actively adjusted their body temperature by seeking preferred ambient temperatures. Collectively, the present study and that of Eldøy et al. (2015) illustrate that S. trutta utilize coastal water bodies commonly influenced by anthropogenic factors such as aquaculture, harbours and marine constructions, marine renewable energy production or other human activity. This suggests that the effects from anthropogenic factors may vary both among S. trutta populations and among individuals within a population according to their temporal and spatial movements in marine coastal areas. Behavioural differences among populations are thus important to account for when assessing the effects of human activities in the coastal zone and when introducing fishing regulations for S. trutta and for fisheries where S. trutta might be captured as bycatch.
The study was financed by the Hemne Municipality, the County Governor of Sør-Trøndelag, Sør-Trøndelag County Authority, the Norwegian Environment Agency, the Lake Rovatnet Landowners Association, Trønder Energi AS, Aqua Gen AS, Dalhousie University (Ocean Tracking Network), Norwegian Institute for Nature Research, DTU Aqua, Uit the Arctic University of Norway and NTNU University Museum. The crew of R.V. Gunnerus, M. G. Hansen, O. M. Taftø, H. Erlandsen, V. P. Sollien, P. Skarsvåg, S. H. Hemmingsen and K. Lian, are thanked for extensive help during the field work. M. Daverdin (NTNU University Museum), A. G. Finstad (NTNU University Museum) and R. J. Lennox (Carleton University) are thanked for assistance with data analyses. The acoustic receivers were part of the Ocean Tracking Network (www.oceantrackingnetwork.org). The experimental procedures concur with the national ethical requirements and were approved by the Norwegian National Animal Research Authority. Five anonymous reviewers are thanked for commenting on earlier drafts of the manuscript.
