Temperature during embryonic development in brown trout inﬂuences juvenile behaviour in encounters with predators

Variation in thermal conditions during embryogenesis can have far-reaching impact throughout ontogeny and may give rise to behavioural variation. Many animals, such as salmonids, exhibit behavioural trade-offs related to foraging and predator avoidance. How embryonic temperature affects these behaviours has remained unexplored. Not only abiotic conditions during embryogenesis but also biotic factors such as predator conditioning may affect ﬁ sh behaviour, especially anti-predator responses. We examined how elevated temperatures and predator odours throughout embryogenesis affect the behaviour of 28 – 37 mm young-of-the-year brown trout ( Salmo trutta ) in encounters with predators, namely Atlantic salmon ( Salmo salar ; 20 cm) and burbot ( Lota lota ; 40 cm). Juvenile brown trout were more active and aggressive if they were incubated in warmer water as eggs than if they were incubated in colder water, and trout remained inactive longer when encountering predators if they were cold incubated. Brown trout were less active and aggressive when an Atlantic salmon was present than when a burbot or no predator was present. Behavioural responses did not differ between trout that had been subjected to water with versus without predator odours during embryogenesis, possibly because brown trout were not subjected to conspeci ﬁ c alarm cues during egg incubation. This study shows that thermal conditions during embryogenesis can in ﬂ uence ﬁ sh behaviour early in life and thus contribute to behavioural variation, with potential effects on life history. Considering the rapid warming of northern regions, elevated embryonic temperatures may contribute substantially to variation in salmonid behaviour in the near future.


Introduction
Temperature during embryogenesis of salmonids affects the development of embryos (Cingi et al., 2010;Johnston, 2006;Jonsson & Jonsson, 2019;Ojanguren & Braña, 2003), which may influence various traits of juvenile and adult life stages (Jonsson & Jonsson, 2014, 2019).Both climate models and empirical data show that climate warming is predicted to be pronounced during winter in northern regions (Choi et al., 2010;Hansen et al., 2010;IPCC, 2021;Magnuson et al., 2000), with effects on river water temperatures (Dokulil, 2014;Leach & Moore, 2014).Many salmonid species have eggs that are incubated during winter, and effects of warming during salmonid embryogenesis are likely to become more evident in the near future.Embryonic temperature may affect behaviour, physiology and consequently life-history traits of juvenile and adult fish, and for example in Atlantic salmon (Salmo salar), elevated temperatures during embryogenesis result in enhanced growth and earlier smoltification and sea migration (Finstad & Jonsson, 2012;Jonsson et al., 2005), possibly through epigenetic effects (Burgerhout et al., 2017).Differences in growth caused by embryonic temperature are still evident in adult salmon and also are reflected in reproductive traits (Jonsson et al., 2014).Recently, Durtsche et al. (2021) reported that brown trout (Salmo trutta) juveniles incubated at elevated temperatures as eggs have lower metabolism than trout from eggs incubated at natural winter temperatures, which results in reduced activity and exploration (Takatsu et al., 2022).
Behavioural alterations attributed to embryonic temperature have not received much attention, and earlier work has mainly focused on birds and reptiles (Bertin et al., 2018;Booth, 2006;Burger, 1990Burger, , 1991;;Mueller et al., 2019;Siviter et al., 2019).For fish, Jonsson and Jonsson (2018) found that Atlantic salmon subjected to 3°C warmer water during egg incubation returned to the Norwegian coast about 2 weeks later to spawn compared to salmon from eggs incubated at normal winter temperatures.Incubation temperature has also been shown to affect downstream migration, with a greater proportion of warm-incubated brown trout moving downstream after stocking than cold-incubated trout (Jonsson & Greenberg, 2022).Considering the major influence that embryonic temperature seems to have on life-history traits overall, it seems plausible that thermal effects during embryogenesis could contribute to intraspecific behavioural variation during other stages of a species' life history.
Predation threat itself can affect fish embryos.Elevated heart rate of embryos when predators are nearby have been reported in several species of teleosts (Atherton & McCormick, 2015;Oulton et al., 2013).In addition to physiological responses, conditioning to predator cues during embryogenesis can alter the behaviour of juvenile fish.For example, when conditioned to both northern pike (Esox lucius) odours and alarm cues from conspecifics as embryos or alevins, juvenile rainbow trout (Oncorhynchus mykiss) exhibited anti-predator behaviours in the presence of pike, whereas juveniles that had not been exposed to predator cues as embryos or alevins did not (Horn et al., 2019).Further, the retention of the anti-predator response was longer if the juvenile trout were subjected to predator cues as embryos compared to if trout were first subjected to predator cues as alevins.Acquired predator recognition during embryogenesis has not only been observed in salmonids but also in cichlids (Nelson et al., 2013), amphibians (Ferrari & Chivers, 2010;Garcia et al., 2017) and gastropods (Dalesman et al., 2015).Hence, biotic conditions such as predator conditioning during embryogenesis, in addition to abiotic factors such as temperature, can affect the behaviour of animals in subsequent life stages.
The objective of this study was to examine how elevated temperatures and predator odours during embryogenesis affect the behaviour of juvenile salmonids.Specifically, we tested how brown trout fry differed in their behaviour in encounters with predators.Based on the results from Durtsche et al. (2021), Takatsu et al. (2022) and Greenberg et al. (2023) we predicted that incubation temperature would influence risktaking behaviour, such as swimming activity and aggression, and we expected that activity and aggression levels would be reduced by elevated temperatures.We also hypothesized that trout would exhibit anti-predator behaviours in the presence of piscivorous fish and that these responses would be more pronounced if trout had been subjected to predator cues as embryos.

Egg incubation
We incubated brown trout eggs between October 2018 and March 2019.Eggs and milt came from wild-caught brown trout (eight females and 10 males; mean body mass and length AE SD = 2.6 AE 0.7 kg and 64 AE 5.3 cm; origin River Klar€ alven, 59°31.561 0N 13°29.971 0E).We placed 100 trout eggs into each of 12 incubation aquaria (75 9 40 9 40 cm; length 9 width 9 depth).Water was constantly filtered and chilled by separate recirculating systems for each aquarium.Incubation environments in the aquaria resembled natural spawning redds, where trout eggs were distributed in interstitial spaces in between smooth river stones.
The eggs were subjected to four treatments using a 2 9 2 design: cold/warm-incubation temperature 9 predator odours present/absent, and each treatment was replicated in three incubation aquaria (n = 12 aquaria in total).The cold and warm incubation temperature regimes were maintained at 2.5 and 5.5°C, respectively, throughout most of winter, and thereafter temperatures were increased during spring (Fig. 1).From mid-May and onwards, when all trout had hatched and emerged from the incubation gravel, temperatures were kept stable at 12°C in all aquaria, which is a typical temperature in small streams in central Sweden during this season (Watz et al., 2016).As salmonids predate eggs and fry from other salmonids (Br€ ann€ as, 1995;Johnson & Ringler, 1979), we subjected the eggs to Atlantic salmon (Salmo salar) odours throughout embryogenesis.The predator odour treatment was maintained by the addition of water from a 200 L tank containing five first-generation hatchery-reared Atlantic salmon (mean AE SD body mass and total length = 5.9 AE 1.0 g and 86 AE 6 mm).We replenished 20% of the water in each incubation aquarium once per week, and thereby, half of the aquaria received predator odours from the tank with Atlantic salmon and the other received water from a tank without salmon (Yl€ onnen et al., 2007).We removed dead fry daily.We kept the eggs in total darkness throughout winter (apart from short periods of dim light when husbandry and aquarium maintenance work was carried out), but increased light duration during spring in accordance with natural daylight cycles.A more detailed description of the egg incubation can be found in Filipsson et al. (2024).From emergence to the onset of the behavioural experiment, we fed trout daily with conventional fish feed for salmonid fry (Aller performa, Aller aqua A/S, Christiansfeld, Denmark).Brown trout were fed ad libitum twice a day, in the morning and afternoon.

Experimental execution
We conducted the behavioural experiment between 22 June and 8 July 2019.The experimental setup consisted of three aquaria (63 9 48 9 40 cm; length 9 width 9 depth; Fig. 2), positioned next to each other in a temperature-regulated room (14°C).Water in the aquaria was constantly filtered (EHEIM 2217 classic canister filter; Eheim GmbH & Co KG, Deizisau, Germany) and chilled (Teco TK 2000, Teco, Ravenna, Italy).We kept the water temperature at 12°C and the water depth at 20 cm.All behavioural trials were conducted in daylight with a light intensity of 150 lx at the water surface level.We divided each experimental aquarium into two equally sized sections (31.5 9 48 9 40 cm; length 9 width 9 depth), by using net screens attached to wood frames.The bottom of one section in each aquarium was covered with gravel (5-20 mm), and these sections functioned as enclosures for the predatory fish.
We used one Atlantic salmon and one burbot (Lota lota) as predators in the behavioural trials.We selected these predators as they both are known to prey on trout fry, but use different hunting tactics to some extent: piscivorous salmonids in slowflowing water often forage using an active search tactic (Juanes et al., 2002), whereas burbot mainly is a nocturnal ambush predator (P€ a€ akk€ onen, 2000).The salmon (20 cm and 52.2 g) was a first-generation hatchery fish, was kept in a 200 L aquarium before the experiment commenced and was fed 0.5% of its body mass with commercial fish feed (Aller performa, Aller aqua A/S, Christiansfeld, Denmark) three times per week.We kept the burbot (40 cm and 550 g) inside the uppermost section (95 9 320 cm) of a 7-m-long stream flume before the experiment commenced, and we fed the burbot 15 g thawed, previously frozen, fish filets twice per week.Both the Atlantic salmon and the burbot were kept in the same light and temperature regimes as the trout.
We placed the salmon and the burbot in two of the experimental aquaria 72 h before the behavioural experiment commenced.One aquarium was kept free from Atlantic salmon and burbot and served as a control for the predator treatments.Each predator treatment was tested in the same aquarium until one-third of the trials were conducted.After this, we emptied all aquaria of water, including filters and chillers, cleaned the equipment with ethanol and left it to dry.Thereafter, we reused the aquaria but re-distributed the three treatments so that each aquarium was used for a different treatment.The procedure was then repeated after two-thirds of the trials were completed.This design allowed us to test the three predator treatments in all aquaria, while taking the welfare of the animals into consideration (i.e.not moving the predators between aquaria more than necessary).
Semicircle-shaped net cages (34.5 9 25.5 9 26 cm; width 9 radius 9 height; mesh size = 2 mm) were used for holding the trout placed in the section of the aquaria designated for the trout (Fig. 2).One net cage was placed in each aquarium, with the flat side placed directly against the net screen that divided each aquarium.The bottoms of the net cages were covered with white polyester paper, which provided a clear background for observations.A mirror (20 cm high and 8.5 cm wide) was placed in a corner of each trout cage, next to the predator enclosure (Fig. 2).This position of the mirror ensured that only the trout, and not the predatory fish, was reflected in the mirror.
We tested brown trout from the different incubation treatments according to a randomization protocol.Before a behavioural trial, we netted three brown trout from different incubation treatments and based on the randomization protocol placed them individually inside the net cages in the three experimental aquaria.Hence, one brown trout at a time was tested in each aquarium.After 40 min, we moved the brown trout to an aquarium with another predator treatment so that each brown trout was subjected to all three predator treatments.In total, we tested 73 brown trout: 19 brown trout from the warm 9 predator odour absent incubation and 18 brown trout from the cold 9 predator odour present, cold 9 predator odour absent and warm 9 predator odour present treatments, respectively.Brown trout sizes were (mean AE SD, min-max) 0.19 AE 0.03, 0.13-0.27g and 31 AE 1, 28-34 mm for coldincubated brown trout and 0.24 AE 0.06, 0.15-0.37g and 32 AE 2, 29-37 mm for warm-incubated brown trout.

Data collection and statistical analysis
During a trial, brown trout were filmed (GoPro Hero 7 Black, GoPro, San Mateo, CA, USA) for 40 min, with cameras mounted 30 cm directly above the trout cages.From the video recordings, we quantified four behavioural parameters from the brown trout: (1) time from start of trial until first movement; (2) proportion of time spent swimming; (3) proportion of time that trout exhibited aggressive behaviours towards their mirror reflections and (4) position expressed as distance from the predator enclosure.Time to first movement was noted directly after trout were placed inside the experimental cages.We defined movement as when brown trout moved more than one body length in less than a second.For behaviours 2-4, we first let brown trout acclimatize for 20 min and thereafter used the subsequent 15 min of each recording for data collection, with observations of brown trout behaviour every 15 s.We defined time spent interacting with mirror reflections as the time when brown trout were exhibiting display behaviours or were actively attacking the mirror.Yearling salmonids exhibit aggressive behaviours towards their mirror reflection, and mirror image simulation tests are a frequently used method to measure aggression (Adriaenssens & Johnsson, 2013;Johnsson & N€ aslund, 2018).We also measured the distance between the brown trout and the predator demarcation net every 15 s, and we calculated trout's mean position for the entire 15 min period.
We analysed all data with full-factorial generalized linear mixed models, which included incubation temperature, predator odour incubation treatment (between-subject factors) and predator treatment during the behavioural trials (within-subject factor) as fixed factors.Incubation aquarium was included in all models as a random factor, nested within the incubation temperature 9 predator odour incubation interaction term.We used a negative binomial model to analyse time to first movement, as this variable had skewed positive distributions and did not meet assumptions of normality.We arcsine square root transformed proportion of time spent swimming.Data for swimming activity and trout distance from the predator enclosure conformed to assumptions of normality, and data were therefore fitted to Gaussian models.Many brown trout did not show any aggressive behaviour, and we therefore used the analysed mirror aggression as a 0/1 variable, where individual brown trout were categorized as either displaying aggression (1) or not (0).We used compound symmetry covariance structures in all models for the within-subject factor (i.e.repeated measures; Littell et al., 2000).Any effect of incubation conditions (in particular temperature) on the four response variables could potentially consist of an effect mediated by size.We therefore analysed all models both with and without brown trout mass as a covariate.All statistical analyses were performed in IBM SPSS Statistics 26 (Armonk, NY, USA).

Ethical note
We adhered to the ASAB/ABS guidelines for the treatment of animals in behavioural research throughout this project.The experiment was carried out at the aquarium facility of Karlstad University (permit 5.2.18-2527/2018 from the Swedish Board of Agriculture) with approval from the Ethical Committee on Animal Experiments, Gothenburg (5.8.18-03818/2018 and 5.8.18-03819/2018).
Figure 2 Birds-eye view of an experimental aquarium (63 9 48 cm).Net screens divided each aquarium into two equally sized sections (31.5 9 48 cm), one for trout and one for the predatory fish.We placed semicircle-shaped net cages (34.5 9 25.5 cm) in the trout sections, with the flat side against the net screen.A mirror (20 cm high and 8.5 cm wide) was placed in the corner of each trout cage, next to the demarcation to the predator enclosure.Water depth was 20 cm.

Results
Without a predator present, the mean (AE SE) time to first movement was 69 s (AE 22), whereas with a burbot and Atlantic salmon present the corresponding times were 96 (AE 24) and 131 (AE 31) s, respectively.This difference was mainly found for cold-incubated brown trout, and there was a significant interaction effect on the first movement between incubation temperature and predator presence (Table 1, Fig. 3).Coldincubated brown trout started to move earlier when no predator was present (mean AE SE, 29 AE 17 s) than when in the presence of burbot (114 AE 48 s) or Atlantic salmon (141 AE 61 s), but for warm-incubated brown trout, time to first movement did not differ considerably between predator treatments (Fig. 3).None of the other fixed terms, interaction terms or body mass had a significant effect on time to first movement, and the results were similar regardless of whether or not mass was included as a covariate (Table 1).
There was a significant difference in swimming activity between cold-and warm-incubated brown trout (Table 1), where warm-incubated brown trout spent a greater proportion of time swimming (51 AE 6%) than cold-incubated brown trout (33 AE 5%) (Fig. 3).Swimming activity was also affected by predator presence (Table 1).Brown trout were less active in the presence of Atlantic salmon (28 AE 5%) than in the presence of burbot (51 AE 6%) or in the control treatment (46 AE 6%) (Fig. 3).Pairwise contrasts between predator treatments indicated that the swimming activity of brown trout differed statistically between Atlantic salmon and burbot treatments (P < 0.001), Atlantic salmon and control treatments (P < 0.001), but not between burbot and control treatments (P = 0.22).Body mass also had an effect on the proportion of time that brown trout kept swimming (Table 1), where swimming activity tended to increase with increasing brown trout mass.None of the other main or interaction terms had a significant effect on brown trout activity, and the results were similar regardless of whether or not mass was included as a covariate (Table 1).
Predator presence affected the probability that brown trout displayed aggressive behaviour towards their mirror images (Table 1).Fewer brown trout were aggressive when Atlantic salmon was present (46%) than when burbot was present (69%) or in the absence of a predator (71%).Body mass had an effect on brown trout aggression, where aggression increased with increasing size.The size effect resulted in that incubation temperature had an effect on aggression only when mass was not included as a covariate (Table 1), indicating that incubation temperature affected aggression only through its effect on fry size.No other main effects or interaction terms had a significant effect on mirror aggression.
Predator presence and brown trout mass influenced how far brown trout positioned themselves from the predator enclosure (Table 1).Brown trout kept a greater distance from the predator enclosure in Atlantic salmon treatments (mean AE SE = 9.7 AE 0.7 cm) than in burbot (8.6 AE 0.5 cm) or control treatments (8.9 AE 0.6 cm) (Fig. 3).Pairwise contrasts between predator treatments showed that the position of trout differed statistically between Atlantic salmon and burbot treatments (P = 0.003), Atlantic salmon and control treatments (P = 0.026), but not between burbot and control treatments (P = 0.43).Large brown trout fry generally stayed closer to the predator enclosure than small brown trout, and this size effect resulted in an effect of incubation temperature when brown trout mass was excluded from the model as a covariate (Table 1), because incubation temperature had an effect on size.

Discussion
In this study, we showed that a temperature increase of 3°C during embryogenesis resulted in more active and aggressive brown trout fry.This result was the opposite of what we predicted, and it contrasts with those of Takatsu et al. (2022) and Greenberg et al. (2023).They found that increased temperature during egg incubation resulted in less distance moved by newly hatched brown trout fry in an open field test.Effects on egg incubation temperature on distance moved in an open field test seem, however, to be ephemeral (Greenberg et al., 2023).We measured the proportion of time spent swimming as a measure of activity, and we did not use an empty container as our test arena (as is the case in an open field test).Instead, we used a cage with a mirror in an aquarium with a predator, and it is possible that the different physical environments led to contrasting results between the studies.This explanation is supported by N€ aslund et al. ( 2015), as they found that the design of experimental arenas may affect the results of behavioural assays.The differing results may also relate to how activity was defined (i.e. in Takatsu et al., 2022 andGreenberg et al., 2023 vs. in our study).Distance in an open field test mainly measures exploratory behaviour (Stewart et al., 2012), whereas in our study, time spent swimming likely related more to a trade-off between avoiding conspicuous behaviour and aggression towards the mirror image than exploration.
We found that time to first movement was delayed when a predator was present, but only for cold-incubated brown trout, indicating that variation in embryonic temperature may contribute to behavioural variation in anti-predator responses.Thus, the degree of anti-predator responses (in our case movement latency) may be augmented by low incubation temperatures, an effect previously demonstrated in reptiles (Flatt et al., 2001) and birds (Bertin et al., 2018), and the underlying mechanism may be related to hormonal regulation of fear-related behaviours (Bertin et al., 2018).
In animals, certain traits are often positively correlated, such as activity, aggression and growth (R eale et al., 2010;Stamps, 2007).Salmonids grow faster post-hatching when eggs are incubated at elevated temperatures (Burgerhout et al., 2017;Finstad & Jonsson, 2012;Jonsson et al., 2005).Trout subjected to variation in temperature as embryos can therefore be expected to exhibit variation in swimming activity and aggression, where high levels of these may increase vulnerability to predation.Previous studies examining the effects of incubation temperature in salmonids (e.g.Durtsche et al., 2021;Greenberg et al., 2023;Takatsu et al., 2022) have not considered how behaviour is influenced under high predation-risk conditions.
Our study shows that, under predator risk, variation in some behaviours (e.g.aggression and predator avoidance) seem to be linked to fish size as influenced by incubation temperature, whereas others (swimming activity) may be affected by mechanisms other than size.Aggressive fish usually have high foraging success under predictable conditions (Reid et al., 2012) and may therefore exhibit accelerated growth, but are at the same time more susceptible to predation.These behaviours may thus be beneficial for fitness under certain conditions, but may also have negative effects under others, for instance in areas with high predation pressure.
Contradictory to our hypothesis, anti-predator behaviour of juvenile brown trout in this study did not differ between trout that had been subjected to salmon odours during embryogenesis or not.This result was surprising, because previous studies show that predator conditioning during egg incubation can result in anti-predator responses of juvenile fish (e.g.Horn et al., 2019;Nelson et al., 2013).Earlier work suggests that predator conditioning may only be acquired when developing embryos are subjected to predator odours in combination with conspecific alarm cues, or in certain cases heterospecific alarm cues from closely related species (Atherton & McCormick, 2015;Horn et al., 2019;Nelson et al., 2013).A similar pattern often is observed in juvenile salmonids, where fish usually respond strongly to imposed predation threat when predators have consumed conspecifics of the prey species (Hirvonen et al., 2000;Jones et al., 2003;Vilhunen & Hirvonen, 2003).In this study, Atlantic salmon were fed conventional fish feed, both when providing predator odours during brown trout embryogenesis and during the behavioural experiment.Possibly, the absence of conspecific alarm cues during embryogenesis (in addition to the predator cues) resulted in trout not becoming conditioned to salmon as predators post-hatching.Brown trout in this study did however exhibit anti-predator behaviours in the presence of Atlantic salmon, regardless of whether trout had experienced Atlantic salmon cues during embryogenesis or not.Hence, the innate anti-predator response may be so strong that predator conditioning during egg incubation is difficult to detect in the behaviour of the brown trout in our study.How anti-predator responses during embryogenesis are associated with predator conditioning and avoidance behaviours post-hatching is currently not clear.This topic thus deserves future attention, to elicit how predator presence during embryogenesis may affect fish during later life stages.Visual cues were likely important for detecting the presence of the predator.Brown trout were less active and aggressive in the presence of Atlantic salmon than in the presence of burbot or in control treatments.Brown trout also positioned themselves further away from the predator enclosure when salmon was present.Larger salmonids prey on smaller individuals, both con-and heterospecifics (Br€ ann€ as, 1995;Jones et al., 2003;Vilhunen & Hirvonen, 2003).Juvenile salmonids do however also exhibit anti-predator responses in the presence of burbot (Enefalk et al., 2017;Filipsson et al., 2019Filipsson et al., , 2020;;Hirvonen et al., 2000;Jones et al., 2003).In this study, trout did not do so (Prokkola et al., 2021), except for the time to first movement, where cold-incubated trout became active later when a predator (either Atlantic salmon or burbot) was present.There are several possible explanations as to why brown trout in this study did not respond to the presence of burbot in the same way as they did to Atlantic salmon.First, this study was conducted in daylight and burbot is mainly nocturnal (Hackney, 1973), but this explanation seems unlikely as juvenile brown trout exhibited anti-predator behaviours in the presence of burbot in daylight in a previous study (Filipsson et al., 2019).Second, as brown trout were newly emerged, they were considerably smaller than trout used in other studies that have examined anti-predator responses of brown trout in the presence of burbot (Enefalk et al., 2017;Filipsson et al., 2019Filipsson et al., , 2020)).The small-sized brown trout used here may have resulted in burbot and brown trout not considering each other as predator and prey.Although we did not quantify the behaviour of the predatory fish, we noted that Atlantic salmon was active in the experimental aquaria, whereas burbot was generally inactive.Juvenile Atlantic salmon generally feed on drifting invertebrates, but may also forage by actively searching for prey, including salmonid fry (Beall et al., 1989).Possibly, the active behaviour of the salmon stimulated the antipredator behaviour of the trout.Moreover, individual differences in predator behaviour may affect the anti-predator behaviour of prey.To shed more light on this, multiple individuals of each predator species should be used to secure robust conclusions of how juvenile salmonids distinguish between fish predators (Brown et al., 2011;Hawkins et al., 2007;Hirvonen et al., 2000;Kuehne & Olden, 2012).
This study showed that salmonid fry subjected to elevated temperatures as embryos exhibit an altered behavioural pattern, namely higher swimming activity and aggression levels, compared to trout incubated at natural temperatures.Variation in temperature at egg incubation may be an important driver of behavioural variation within salmonid populations, and climate warming may result in behavioural alteration, with possible implications for life-history traits such as dispersal, migration and reproduction (Jonsson & Jonsson, 2019;R eale et al., 2010).These behavioural alterations most likely constitute only a fraction of the changes induced by embryonic temperature (Jonsson & Jonsson, 2019;R eale et al., 2010).Future research efforts should continue to explore to what extent embryonic temperature influences behaviour and life history, including its interaction with other environmental conditions, throughout ontogeny (Groothuis & Taborsky, 2015).This knowledge may thus provide insight into how thermal effects early in ontogeny caused by climate warming can influence animals throughout the course of their lives.

Figure 1
Figure 1 Temperature curves for cold-incubated (light grey) and warm-incubated (dark grey) brown trout eggs and juveniles throughout winter and spring.

Figure 3
Figure 3 Mean (a) time to first movement, (b) proportion of time spent actively swimming and (c) distance from predator enclosures of brown trout fry during behavioural experiments.Treatments on the xaxis represent different conditions during egg incubation: water temperature (cold/warm) and the absence or presence of predator (Atlantic salmon) odours (control/predator).The filled bars indicate the predatory fish species that was present during the behavioural trials: control (no predator), burbot and salmon.Error bars denote AE1 SE.

Table 1
Results from generalized linear mixed models, presenting the effects of embryonic temperature, predator odours during embryogenesis (between-subject factors), predator presence trials (within-subject factor) during behavioural and body mass (covariate) on brown trout fry (1) time to first movement, (2) swimming activity, (3) aggression towards their mirror reflections and (4) distance from predator enclosures Note: P values in bold font indicate statistical significance (a = 0.05).