There is increasing evidence that conditions experienced by a female can influence her offspring’s phenotype via non-genetic maternal effects [1]. Environmentally-induced maternal effects typically include maternal oviposition decisions [2], maternal care [3] and egg size [4]. Active or passive transmission of maternal trace elements or hormones to eggs or embryos may provide another mechanism to adjust female strategy to environmental conditions [5, 6].

Changes in reproductive strategies and maternal investment of females in response to the environment must be addressed from a life-history perspective because adaptive maternal effects for offspring can come at a cost to females [6]. Increased investment of resources and energy in eggs and altered timing of reproduction entail direct costs for the female and will reduce her future reproductive success [7]. Other maternal effects, such as transfer of hormones to eggs or embryos, may incur indirect costs due to negative pleiotropic effects of increased hormone levels on the physiology or behaviour of the mother [8]. Thus, females are expected to adjust their maternal strategy in relation to their current and future reproductive returns as well as offspring fitness. Indeed, it has been shown in soil mites (Sancassania berlesei) that egg and clutch size shift dynamically with female age, creating asymmetry in the competitive abilities of older and younger offspring from the same mother [9]. This suggests that the significance of female strategies may be misinterpreted when studies consider only a snapshot of maternal effects, based on a single reproductive attempt.

Increased temperatures due to current climate change often have negative impacts on fitness of ectotherms whose body temperatures conform to environmental temperatures, particularly in populations that live close to the species’ thermal maximum [10–12]. In ectotherms, a higher temperature leads to a higher resting metabolic rate [13], and so adjustments to oxygen delivery or oxidative damage may be critical in coping with thermal stress [14–17]. Increased temperatures also have a negative effect on the body size of aquatic ectotherms from the individual to community structure levels [18]. Since metabolic rate increases as a function of body size as well as environmental temperature [19], it is expected that increased temperatures should lead to a decrease in the mean body size if the supply of resources does not increase in parallel [20]. Transgenerational plasticity is a potentially important mechanism by which animals can compensate for negative impacts of thermal stress. Indeed, some previous studies on fish species have shown that parental exposure to thermal stress can induce changes in offspring phenotype [21–23]. Adaptive maternal strategies in response to increased temperatures are expected to alter physiological or life-history phenotypes of offspring, for example by enhancing stress resistance or reducing body size. In fishes, maternal antioxidants deposited in eggs, such as carotenoids and vitamin E, provide protection for offspring against deleterious effects of free radicals produced during normal metabolism and growth [24, 25]. However, it is unknown whether females can adjust maternal transfer of antioxidants to eggs in response to expected thermal conditions for their offspring.

In this study, we experimentally tested whether female three-spined sticklebacks (Gasterosteus aculeatus), originating from a population close to the low-latitude margin of the species’ distribution (and hence at risk of exposure to increasingly stressful environmental temperatures), change their seasonal reproductive strategies in response to increased water temperatures experienced during juvenile life. In this annual population, females spawn more frequently throughout a single relatively long breeding season compared to more northern populations [26, 27]. F1 full-sib families were reared under normal and elevated winter temperatures prior to the reproductive season. Our previous studies have shown that this manipulation of winter temperature influenced the reproductive strategy of males, altering the seasonal pattern of carotenoid investment in nuptial colouration [27] and the expression of genes related to metabolic processes and oxidation reduction [28]. In the three-spined stickleback, maternal transfer of carotenoids to eggs can trade off with other critical functions such as somatic maintenance or sexual selection [29]. Here, we examined whether exposure to warmer winter temperatures prior to the breeding season altered the seasonal patterns of investment by females in clutch mass, clutch size, egg mass and egg carotenoids, and so tested whether females adjust maternal strategies according to the expected environmental conditions. Since individual variation in the expectation for future reproduction is an important determinant of reproductive trajectories, we also tested how overall productivity interacted with seasonal variation in maternal investment under the contrasting thermal regimes. We predicted that, in general, females would reduce egg size but increase egg carotenoids when they experience warmer temperatures because small body size and maternal antioxidants can allow offspring to reduce their metabolic costs and oxidative damage under warm developmental conditions. The temperature treatment was also expected to influence the body condition of the females, thereby modulating the seasonal patterns of their maternal strategies.

Our study demonstrates that female sticklebacks adjust their seasonal patterns of reproductive investment in response to environmental conditions that they experienced earlier in life. Warmer winter temperatures prior to the breeding season did not influence either their body size at maturation or their timing of first reproduction. However, the warm-acclimated females had a lower fecundity and reduced lifespan compared to those reared at normal winter temperatures. The warm-acclimated and control females spawned at similar intervals within the season, but their reproductive investment trajectories in clutch mass, clutch size and concentration of egg carotenoids differed.

In contrast to the response by females, male sticklebacks in the same warm winter temperature treatment (i.e. the brothers of the females analysed in the present paper) showed a clear plasticity in body size and timing of reproduction (i.e. expression of red nuptial colouration): the warm-acclimated males were smaller at the beginning of the breeding season [28] but started to express red colouration earlier than the control males [27]. Female sticklebacks begin mobilizing energy into ovarian and support tissue in winter then continue to grow throughout the breeding season, whereas males stop growing at the onset of maturation [34, 35]. Thus, contrasting energy allocation strategies could explain the different life-history plasticity between the two sexes. In ectotherms, the approximately exponential effects of temperature on energy expenditure are well established. These effects are partly due to direct influences on the kinetic energy of cellular components [19, 36]. But while energetic costs increase exponentially with increasing temperature, food intake only increases up to a peak, above which food intake declines steeply with further increases in temperature [37]. Feeding and digestion rates of fish in a warming environment are thus metabolically constrained even if food is freely available [38]. Food was provided only once a day to both control and warm-acclimated individuals, so the possibility that fish increased feeding to compensate for the metabolic costs of living in a warm environment was probably none or minor. Thus, energy for somatic growth and maintenance was probably limited during winter in both control and warm environments. It is possible that the warm-acclimated females compensated for the increased energetic costs of living in warmer water by reducing their allocation of energy to reproductive functions during the winter (while maintaining their investment in somatic growth) [19, 36]. In accordance with this hypothesis, the warm-acclimated females produced fewer eggs (primarily because they laid fewer clutches), terminated reproduction earlier and suffered higher mortality during the season than the control females, probably due to increased energetic costs for somatic maintenance and growth [39]. Warmer conditions during the winter may have promoted physiological changes in metabolic rate and cardiac function in a manner that incurred higher energetic costs to the fish [23, 40]. It is also possible that higher temperatures either caused more oxidative damage or reduced the amount of resources available to the fish to repair damage [41].

The warm-acclimation not only reduced the overall fecundity and lifespan of females but also altered their strategic allocation of resources among reproductive attempts. In iteroparous organisms the cost of reproduction leads to a trade-off between reproductive effort at any given time and expectation of future reproduction [42, 43]. Thus, life-history theories predict that breeders will optimize the temporal pattern of reproductive investment according to both their life expectancy and the time-dependent fitness gain from a given reproductive attempt [44–46]. Females reared in the control environment invested relatively more in their first spawning, producing heavier and larger clutches, compared to their later spawning events, whereas the warm-treated females produced increasingly heavier clutches with a similar number of eggs across clutches. This suggests that thermal conditions during early life can shape a female’s strategy for allocating energy and resources between early and late reproductive attempts. The warm-acclimated females may have had fewer resources available for initial reproductive investment than the control females due to their higher energetic costs prior to the breeding season. They then invested more in reproduction as the season progressed probably because their prospects for survival and reproduction were lower than for the control females.

An alternative explanation is that the differential allocation strategies of females reflect their expectation of the conditions that their offspring will encounter during growth. This life-history plasticity can be adaptive if fitness benefits of reproducing early in the season [47, 48] are greater in colder or more seasonal environments. Fish born late in the season may incur a greater risk of overwinter mortality (due to small size) and/or a reduced likelihood of attaining a size that allows reproduction the following year [49, 50], but these handicaps may be reduced for juvenile fish that live in environments with warm winters. Thus, the warm-treated females may benefit by reducing their investment in early reproduction and so delaying reproductive senescence. In accordance with this observation, our previous study has shown that male sticklebacks grown in a warmer environment also subsequently reduced their investment in costly sexual signals early in the breeding season, and consequently senesced at a slower rate [27].

Our results show that any transgenerational effects of the maternal thermal environment on offspring phenotype may be mediated by the allocation of antioxidants into eggs but not by egg size. A previous study of marine sticklebacks also found that female acclimation temperature did not influence egg size [23], and in general egg size in sticklebacks shows less phenotypic plasticity than do other reproductive traits such as the number of spawnings [51]. In our study, egg mass increased with clutch number at a similar rate in the control and warm-treated females. Since the breeding season in this annual fish population is extremely long, producing larger eggs in the later clutches may be an adaptive strategy for a female to compensate for the competitive disadvantages and shorter growing season of late-hatching offspring. In fish, larvae hatching from larger eggs are more likely to survive than those hatching from smaller ones, but there is no general evolutionary trend towards greater egg size [52]. Because juvenile fish that attain a larger size at the onset of winter suffer less overwinter mortality [53], selection may favour the female strategy of producing larger eggs in the later clutches. On the other hand, we found evidence that female sticklebacks modulate maternal transfer of antioxidants to eggs according to the predicted thermal environment of the offspring. In the control females, the concentration of egg carotenoids was maintained at a similar level across subsequent clutches despite the senescent decline in clutch size and increase in individual egg mass. However, the females reared in warmer temperatures increased the investment of these maternal antioxidants in the first and last clutches of the season.

Carotenoids are important for reducing oxidative stress and provide the pigment for colour signals used in sexual selection in many bird and fish species [33, 54, 55]. Some experimental studies of birds have shown that maternal carotenoids deposited into eggs can have profound effects on offspring fitness by enhancing their survival during embryonic and nestling periods and the eventual ornamental coloration as adults [56, 57]. In fishes, egg carotenoids positively influence survival and disease resistance of larvae during development and juvenile body size during growth ([25, 58], but see [59, 60]). Carotenoids transferred from the mother may enhance the physiological condition of offspring during their development and growth, for example, by affecting the expression of phenotypic variation in antioxidant capacity or endocrinal signal [61]. Such phenotypic programming by mothers may be important for offspring to cope with increased oxidative stress due to greater rates of metabolism in the predicted warm environment [13]. Transgenerational effects of the maternal environment on metabolic or antioxidant capacities should shape thermal reaction norms of offspring traits in a warming environment [23]. However, it is unclear why in our study the warm-acclimated female sticklebacks increased the investment of maternal antioxidants only in their earliest and last clutches. Females may be able to adjust their investment of carotenoids between egg composition, influencing offspring phenotypic variation among clutches, and their personal use for self-maintenance during the reproductive season [29, 62]. It is possible that the warm-acclimated females tended to invest more heavily in their last clutch (i.e. terminal investment) because they had a reduced life expectancy than those reared at normal winter temperatures [46, 63].

In general, elevated temperature due to climate change has negative consequences in reproduction and survival of ectotherms. Our results clearly indicate the importance of studying reproductive strategies from a life-history perspective to understand both developmental and transgenerational effects of temperature. Female strategy for allocating energy and resources among consecutive reproductive events may be an adaptive mechanism mediating transgenerational phenotypic plasticity for preparing offspring to grow and survive better in the unfavourable environment. However, our results cannot explicitly demonstrate whether the temperature-induced changes in maternal strategy are due to constraints imposed by higher energetic costs for self-maintenance of the mother or an adaptation. Future studies are needed to test whether this female strategy is indeed beneficial for offspring in warming environments.