Effects of altered photoperiod regimes during winter on growth and gonadosomatic index in Arctic charr (Salvelinus alpinus) reared in freshwater

As the freshwater species with the northernmost distribution in the world, the Arctic charr (Salvelinus alpinus) experiences large variations in day length throughout the year. Photoperiod regime is known to affect production characteristics in the species, but the relationship and interaction between photoperiod, temperature and endogenous rhythms and their effects on growth, feeding and the timing of maturation are complex (Frantzen, Arnesen, Damsgård, Tveiten, & Johnsen, 2004; Imsland & Gunnarsson, 2011; Liu & Duston, 2018, 2019). There is a general agreement that an increase in day length has a positive effect on appetite and growth in the species (Johnston, 2002; Tveiten, Johnsen, & Jobling, 1996). Mortensen and Damsgård (1993) demonstrated that the main reason for increased growth in charr during spring was the change from short to Received: 22 August 2019 | Revised: 10 December 2019 | Accepted: 13 December 2019 DOI: 10.1111/are.14481

long day, rather than just the long day length. Siikavuopio, Saether, Skybakmoen, Uhlig, and Haugland (2009) confirmed those findings when Arctic charr, reared under continuous light conditions, were exposed to a period of short day length during winter and responded with a growth increase resulting in a 25% higher weight gain compared with that of fish reared at continuous light for the whole period. The potential effect these photoperiod regimes had on subsequent maturation of the fish was, however, not investigated.
Photoperiod manipulation has been validated as a tool that can be used to arrest or delay maturation in rearing of salmonids (Bromage, Porter, & Randall, 2001;Peterson & Harmon, 2005;Taranger et al., 1999). For Atlantic salmon (Salmo salar), continuous light is commonly used from December to March as part of the current production regime of Atlantic salmon in Norway to promote growth without triggering maturation (Imsland et al., 2017;Oppedal, Berg, Olsen, Taranger, & Hansen, 2006). Frantzen et al. (2004) investigated the effect of photoperiod manipulation on the timing of sexual maturation and spawning of Arctic charr and found that Arctic charr responded in a similar manner to other salmonids. Duston, Astakie, and MacIsaac (2003) found a significant increase in the proportion of higher value immature Arctic charr by applying a long photoperiod (18L:6D) for 42 days in winter followed by a short or natural photoperiod compared with fish reared on constant long day or natural photoperiod. In contrast, long-term exposure to continuous or extended light regimes has, in some cases, resulted in reduced growth and food conversion efficiency (Stefánsson, FitzGerald, & Cross, 2002), implying that the period of extended or continuous light must be harmonized with the internal rhythms of the fish in order to achieve optimal results. The present study was set up to investigate the effect of periodic exposure to continuous light at different phases during the winter on growth and subsequent maturation in Arctic charr. Fish were reared at either constant long day (16 hr of light: 8 hr of darkness, LD16:8, continuous light; LD24:0) and transferred for a short time interval from LD16:8 to LD24:0 and back to LD16:8 again. The rationale for choosing these regimes was based on previous findings on Arctic charr (Duston et al., 2003;Imsland & Gunnarsson, 2011) and the photoperiod regime in use at the production site. The main aim was to determine whether there is a specific time during winter where a shift from LD16:8 to continuous light and back may reduce maturation without compromising growth.

| Experimental fish and conditions
The fish used in the experiment were from a multiple generation farmed Arctic charr strain, commonly denoted the Hammerfest Strain, originating from Lake Storvatn (70°N, 24°E), Northern Norway. Juvenile Arctic charr with mean weight (±SEM) 3.5 ± 0.5 g were achieved from commercial farmer (Sjøblink Blokken AS) in The fish (n = 178) were acclimatized for approximately 2 months and subsequently reared throughout the experiment in 4 fibreglass tanks with a volume of 1,000 L, at ambient temperature (2.3-12.5°C; Figure 1) for 18 months. Water was supplied from a freshwater lake close to the rearing site at a depth of 30 m. Water flow was set to approx. 20 L/min and was gradually increased with increasing biomass in the tanks, and oxygen levels were always kept above 80% saturation. The fish were fed a commercial dry feed (Skretting) in excess, using automatic feeders and additional handfeeding to control appetite. A 36 W fluorescent daylight tube integrated in each tank-cover provided light, and the respective photoperiods were maintained using electronic timers. The experiment lasted from 21 April 2009 to 20 October 2010 (18 months).

| Ethics statement
The experiment described has been approved by the local responsible laboratory animal science specialist under the surveillance of the Norwegian Animal Research Authority (NARA) and registered by the Authority and thereby conforming to Directive 2010/63/EU. The fish with the initial mean weight of (SEM) 261 (7) were stocked in four tanks with two tanks reared at continuous light and two tanks reared at a photoperiod of LD16:8. Both groups were, therefore, replicated. Initial density in the rearing tanks was between 8.3 and 10.4 kg/ F I G U R E 1 Mean monthly temperature at the rearing facility (Kirkenes Charr) during the trial period (April 2009-October 2010) m 3 . As the experiment progressed, the groups being exposed to continuous light for a limited period during winter were transferred from the LD16:8 tanks to the LD24:0 tanks for the given period and back again to their original tank when the exposure period had ended. The experimental groups included one group that was held at a constant long day (LD16:8, N = 40) photoperiod for the whole period; three tagged subgroups from this group were transferred to continuous light for a period of approx. 44 days for three consecutive periods during winter (Early, Mid and Late WL (winter light)) (N = 35, 38 and 33 for Early WL, Mid WL and Late WL, respectively) followed by a long day photoperiod until slaughtering (Table 1). In each case, the three transfer groups were moved from the two LD16:8 tanks to the two LD24:0 tanks so that these three groups were also replicated. The last group was held at a continuous light regime, as commonly used in the pro- Specific growth rate (SGR) was calculated according to the formula of Houde and Schekter (1981):

| Statistical analysis
All statistical analyses were performed using STATISTICA™ 13.0. To assess normality of distributions, a Kolmogorov-Smirnov test (Zar, 1984) was used and homogeneity of variances was tested using Levene's F test (Brown & Forsythe, 1974).
Mean individual growth trajectories were analysed using a growth curve analysis (GCM) multivariate analysis of variance (MANOVA) model (Chambers & Miller, 1995;Timm, 1980). The model equation of the GCM had the form: where Y (n × p) are the growth-at-age vectors.
for each p (age) measurements on n individual fish; X (n × q) is the design matrix or the set of extraneous variables measured for each individual; that is, q = age p +photoperiod regime i (i = LD8:16, LD24:0, Early WL, Mid WL, Late WL + replicate j ); B (q × p) is the matrix of parameters estimated by the model; E (n × p) is the matrix of deviations for each individual from the expected value of Y = XB.
Separate analyses were done for each sex. As the GCM analyses included both fixed (photoperiod group) and random (replicate) factors, statistical testing was performed using the VEPAC (Variance Estimation and Precision) program in STATISTICA™. In this module of the programme, the variance components for both the random and fixed effects are estimated with a restricted maximum-likelihood estimate (REML) procedure (Demidenko, 2004).
Individual size ranking (initial size rank (21.04.09) versus. final size rank (20.10.10)) and individual growth rate ranking (initial growth rate (21.04.09-03.06.09) versus. final growth rate (26.08.10-20.10.10)) were tested using Spearman's rank correlation (r Sp ) (Zar, 1984). Possible differences in mean weight and GSI were analysed using a two-way model III nested ANOVA, where the replicates were nested within treatments.
Significant ANOVA's were followed by a Student-Newman-Keuls (SNK) multiple comparison test to identify differences among treatments. A significance level (α) of 0.05 was used if not stated otherwise.

| Growth
For the whole experimental period, the fish reared under the different photoperiod regimes did not differ in their growth patterns (MANOVA ( TREATMENT ), Wilk's Λ 7, 28 = 0.50, p > .1). But periodic differences were seen in the growth of the males as the individual growth trajectories for the males differed from February 2010 onwards (MANOVA ( TREATMENT ), Wilk's Λ 4, 16 = 0.75, p < .05), whereas no differences were seen in female growth at any time. Significant differences were also found in growth-at-age trajectories of male SGR = (e g − 1) × 100. GSI = (Gonad weight∕Body weight) × 100.
(1) Y(n × p) = X(n × q) B(q × p) + B(n × p) (2) y = (y 1 , y 2 , … y p ) The initial mean weight (SEM) was 261 (7) g and did not differ significantly between tanks or groups (two-way nested ANOVA, p > .45). The overall final mean weight did not differ significantly when data from both sexes were combined and were as follows:

TA B L E 1 Schematic overview over the experimental groups and treatments
1.50 kg (LD24:0 group), 1.49 kg (Early WL group), 1.40 kg (Mid WL group), 1.32 kg (LD16:8 group) and 1.20 (Late WL group). When analysed for each sex, no differences in mean weights between experimental groups were found for the females (two-way nested ANOVA, p > .20; Figure 2b). For the males, the group reared on constant

| Size and growth ranking
Apart from males in the Mid WL group (r Sp = 0.61, p < .01), no significant size rank correlation (initial weight vs. final weight) was found in any experimental group of either sex (r Sp < 0.13, p > .10). Positive correlations between initial and final growth rates were not found in any experimental group of either sex (r Sp < 0.28, p > .09).

| Final GSI
Exposure to continuous light during early winter significantly influenced the final GSI in female Arctic charr (two-way nested ANOVA, p < .05), whereas no differences were found between males. Female GSI (%) was lowest (SNK post hoc test, p < .05; Figure 3b) in the Mid WL group (1.7) and highest in the LD24:0 group (7.0).

| Growth and size dispersal
Altered photoperiods affected the growth pattern of male Arctic charr, whereas no changes in growth were found for females.
Growth of Arctic charr is often stimulated following changes in day length, and the growth of fish reared under a changing photoperiod may be better than in fish kept on constant photoperiod. Mortensen and Damsgård (1993) reported that Arctic charr (4-50 g) reared at constant short or long days grew equally well but a group of fish reared for a period on a short photoperiod followed by long photoperiod showed a significant increase in growth. In a long-term study, Siikavuopio et al. (2009) reported a 25%-30% higher growth rate of wild Arctic charr held under culture condition and exposed to intervals of short day length in between periods of continuous light compared with a group reared at continuous light. Gunnarsson et al. (2012) reported that Arctic charr reared using a 6-week short photo-  (Imsland et al., 2017).
Possible formation of size hierarchies was performed by comparing individual size ranking at the start and termination of the trial.
It should be noted that no behaviour observation was performed during the experimental period, but according to staff at rearing facility, aggressive behaviour was rarely seen during the trial. The present size ranking data show that, with the exception of one group (male Mid WL group), no size rank correlations were seen in any experimental group of either sex indicating low, or no, formation of size hierarchies within the experimental tanks. Such size hierarchies are common under culture conditions (Brännäs, 2009;Imsland, Nilsen, & Folkvord, 1998;Jobling & Baardvik, 1994;Petursdottir, 2002) and can lead to higher growth size heterogeneity with negative effect on growth as growth can be suppressed by competition under such conditions (Brännäs, 2009;Imsland et al., 1998). For Arctic charr, it is possible to reduce aggression leading to formation of size hierarchies by applying forced exercise (high current in tanks) (Brännäs, 2009) or by avoiding to rear too size similar fish together (Baardvik & Jobling, 1990;Wallace & Kolbeinshavn, 1988). Under conditions promoting hierarchy formation, the largest fish at the beginning is expected to get the largest share of the feed, grow the fastest and have the highest weight at the end (Brännäs, 2009;Petursdottir, 2002) and the fact that almost no stable size ranking was found in the present study indicates that dominance hierarchies did not develop within the experimental groups. A possible explanation can be related to the experimental layout of this study. The experimental fish was divided into five experimental groups in four experimental tanks where fish were transferred from LD16:8 to LD24:0 for 44 days and then back to LD16:8. This frequent transfer of fish between tanks may have counteracted formation of stable size and growth rankings thereby suppressing the formation of size hierarchies within the experimental tanks.
The ambient rearing temperature in the trial varied between 2.3 and 12.5°C. For most of the study period, the growth was uniform, whereas the growth of the fish increased in the second summer of the experimental period when temperature was between 10 and 12.5°C. Arctic charr will grow at temperatures as low as 0.3°C (size range 200-300 g, Brännäs & Linnér, 2000; size range 2-25 g, Borgstrøm, Isdahl, & Svenning, 2015), and the upper limits for growth are near 20°C (size range 1-5 g, Lyytikäinen, Koskela, & Rissanen, 1997;size range 15-26 g, Thyrel, Berglund, Larsson, & Naslund, 1999). Accordingly, the rearing temperatures in this trial are well within the temperature tolerance range of the species. Rearing temperature was between 2.7 and 3.1°C in the period when fish were moved from LD16:8 to LD24:0 so it is unlikely that a possible interaction effect between temperature and photoperiod may have occurred as all transfer groups experienced very similar temperatures in the photoperiod transfer period. For Atlantic salmon, the findings of Døskeland et al. (2016) suggest that the magnitude of the effect of continuous light on growth is inversely related with temperature which results in significant interaction between temperature and photoperiod. Clarke, Shelbourn, and Brett (1978) suggested that the rate-controlling effect of temperature might be the reason for the short duration of the growth-enhancing effect of long photoperiod at higher temperature in sockeye, Oncorhynchus nerka, and coho, O. kisutch, salmon.

| Final GSI
For salmonids, it is well established that photoperiod can be used to either advance or delay the time of maturation (Bromage et al., 2001) and photoperiod is also used in commercial fish farming to postpone or suppress maturation. Duston et al. (2003) reported lower maturation rate and increased proportion of high-value fish (>1 kg) in Arctic  .
Restricted feeding reduces growth and adiposity (Shearer, Silverstein, & Plisetskaya, 1997), but if applied for a short time during the window of 'critical decision point' for maturation, it can lead to delayed maturation with little effects on the final weight due to compensatory growth during the following full feeding period (Taranger et al., 2010). Imsland and Gunnarsson (2011) subjected groups of Arctic charr to two different ration levels, 100% (full ration) and 50% (half ration) in two 6-week periods during autumn (September-November) and winter (December-February). In between the restricted ration periods and from February onwards, all fish were fed full ration. In the following summer and autumn, signs of lower maturation were seen for females in the feed-restricted group. The latter period of that study is in line with the Mid WL group of current study where changes in photoperiod regime resulted in lower maturation. Thus, it may be hypothesized that this period is the critical period for maturation in the subsequent year. Periods of food restrictions can affect the maturation rate of Arctic charr although results are contrasting (e.g. Liu & Duston, 2019). Similar results have been presented for Atlantic salmon (Thorpe, Talbot, Miles, & Keay, 1990), Chinook salmon, Oncorhynchus tshawytscha (Shearer, Parkins, Gadberry, Beckman, & Swanson, 2006) and Atlantic halibut (Foss et al., 2009).

| CON CLUS ION
At present, Arctic charr is commonly reared under continuous light throughout both during the juvenile stage and during the on-growing phase, whereas present data indicate that a 44-day rearing period on continuous light in January and February (Mid WL group) resulted in lower female GSI but no changes in female growth. Male growth was highest in the group reared at continuous light throughout the study period. Accordingly, we recommend the application of brief continuous light treatments during the juvenile phase (January to February) as a tool to lower female maturation in Arctic charr farming.

ACK N OWLED G M ENTS
The present study was financed by the Norwegian Research Council (Project no. 187354). The authors would like to thank the technical staff of Kirkenes Charr for valuable assistance prior to and during the experimental period. There is no conflict of interest related to the findings presented in this manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data sharing is not applicable to this article due to commercial restrictions.