For avian embryos, the demand for oxygen generally increases as temperatures rise, leading to an increased production of CO2 as well as higher water loss through diffusion across the eggshell (17). While the ratio of CO2 lost to O2 gained through is between 0.70 – 0.75 in the avian egg, the O2 molecule also has 27% lower mass, meaning that the mass exchanged due to the diffusion of these molecules is nearly equivalent (17). Therefore, during the course of incubation, the change in mass of the embryo can be attributed almost entirely to water loss. In our experiment we used the change in mass as an indirect measure of water loss. When we compared the water loss among eggs produced by control mothers, we saw an expected pattern. Embryos from juvenile control mothers lost 5.63 x 10-2 g (±0.474, ±95% CI) more water at a higher incubation temperature compared to the control incubation temperature (Fig. 1a) (Linear mixed model (LMM): t = -2.47, d.f. = 21, P = 0.022). However, mothers who were exposed to the mild heat conditioning as juveniles produced embryos that were more resistant to heat-associated water loss. Embryos from these heat-conditioned mothers did not lose more mass at higher incubation temperatures compared to those at a control temperature. (Fig. 1a) (LMM: t = -0.223, d.f. = 21, P = 0.826). In contrast, the thermal treatment mothers received as adults approximately three weeks before breeding had no effect on the change in egg mass during incubation. This effect on water loss cannot be attributed to differential egg size alone, as there were no differences in initial egg mass among the maternal treatment groups (LMM: juvenile: t = -0.193, d.f. = 37, P = 0.848; adult: t = -0.406, d.f. = 37, P = 0.687; J x A: t = 0.280, d.f. = 37, P = 0.781).
At standardized humidity, differences in embryonic water loss can be attributed in part to metabolic rate. Therefore, we investigated how maternal heat exposure shapes the metabolic rate of offspring as embryos. To do this, we measured embryonic heart rate – which is frequently used as a proxy for metabolic rate in birds and reptiles due to its strong correlation with oxygen consumption – at ~30% and ~77% of embryonic development (30, 33). As expected, there was a negative correlation between the embryonic heart rate and change in egg mass during development across all samples. For each 1 beat per minute (bpm) increase in heart rate, we saw a 4.92 x10-4 g (±3.98 x10-4, ±95% CI) increase in mass loss (LMM: t = -2.52, d.f. = 31, P = 0.017). However, this pattern did not hold true for all treatment groups. When incubated at the high temperature, embryos from mothers exposed to both heat treatments (Heat-Heat) had a significantly lower heart rates than embryos from all other maternal treatment groups at ~77% of embryonic development (Fig. 1b) (LMM: d.f. = 141, all P < 0.01).
Non-optimal incubation temperature can also result in changes of embryonic development time and/or differential hatchling morphology (20, 29, 36, 37). In such situations, embryos may prioritize development of essential organs such as the brain and heart at the expense of others (17). Here we saw an effect of the maternal juvenile treatment on embryonic development time, however this effect was dependent on the incubation temperature the embryos were exposed to. For embryos produced by juvenile control mothers, those that were incubated at the high temperature had a development time 0.435 days (±0.451, ±95% CI) shorter than those incubated at the control temperature as expected (Fig. 2a) (LMM: t = -1.94, d.f. = 47, P = 0.058). However, embryos produced by mothers exposed to the mild heat conditioning as juveniles had longer development times than the aforementioned embryos from juvenile control mothers, even when incubated at a high temperature (Fig. 2a) (LMM: HC-CH: β = 1.01, t = 2.89, d.f. = 47, P = 0.006; HH-CH: β = 0.947, t = 2.39, d.f. = 47, P = 0.021). Rather than being shorter, embryos produced by mothers that received the mild heat conditioning as juveniles had similar development times to those incubated at the control temperature (Fig. 2a) (LMM: CC-HC: β = 0.575, t = 1.88, d.f. = 47, P = 0.067; CC-HH: β = 0.512, t = 1.42, d.f. = 47, P = 0.162). These results again indicate that mothers exposed to heat as juveniles produced embryos that have altered physiological responses to high incubation temperatures that may be beneficial.
Across all samples there was a significant negative correlation between development time and embryonic heart rate at ~77% of embryonic development, a pattern similar to that demonstrated in other species (38, 39). For each 1 bpm increase in heart rate, we saw a 5.89 minute (±3.92 x10-3, ±95% CI) decrease in embryonic development time (LMM: t = -2.10, d.f. = 46, P = 0.042). Interestingly, differential egg mass loss and heart rate patterns between treatment groups were also associated with morphology differences at hatching, but not in the manner we anticipated. Hatchlings produced by mothers who were exposed to the mild heat conditioning as juveniles had heavier pectoralis muscles when incubated at a high temperature. These hatchlings had pectoralis muscles 5.13 x10-3 g (±4.57 x10-3, ±95% CI) heavier than hatchlings produced by juvenile control mothers incubated at a control temperature (Fig. 2b) (LMM: pectoralis: t = 2.24, d.f. = 64, P = 0.029). Interestingly, this same group also exhibited a seemingly increased resistance to heat-induced water loss and development times unaffected by a higher incubation temperature as previously described. These results suggest a potential adaptive maternal effect, as the increased pectoralis mass and elongated development times at high temperatures was not at the cost of reduced overall body mass or other measured organs (heart or residual yolk mass) (See Table S1 for full statistics). However, we cannot rule out that this change in embryo phenotype did not incur a cost that only becomes apparent later on in life. Phenotypic changes that promote increased fitness early on in life can come at the expense of longevity or reproductive output later in adulthood. Whether such trade-offs occur may also depend on the thermal environment the offspring would experience post-hatch. Phenotypic changes through maternal effects may only promote offspring fitness if the environment they experience is similar to that experienced by the mother, and detrimental if there is a mismatch.
Differential embryonic water loss may not only be attributed to metabolic rate, but also to eggshell traits. Lower pore density and thicker eggshells decrease gas conductance and therefore water loss. However, such morphology would also limit the rate of diffusion of oxygen and CO2 across the eggshell, potentially decreasing the chance of survival at high temperatures. Therefore, we examined the role eggshell traits played in survival at different incubation temperatures. We found that successfully hatched individuals had 3.45 (±3.19, ±95% CI) more pores per cm2 of eggshell compared to those that died (LMM: t = -2.16, d.f. = 75, P = 0.034) as well as eggshells that were 3.23 x10-3 mm (±3.07 x10-3, ±95% CI) thinner (LMM: t = 2.10, d.f. = 80, P = 0.039). We also saw that even among embryos that died, those with thinner eggshells and more pores tended to survive until further along in embryonic development (Fig. 3) (LMM: t = -2.42, P = 0.018). However, the positive correlation between the number of pores and survival time was only found in embryos exposed to the high incubation temperature (Fig. 3b) (LMM: t = 2.04, d.f. = 77, P = 0.0446), indicating the importance of eggshell characteristics in embryonic survival in this species at high ambient temperatures.
While it is clear that eggshell characteristics can influence the physiology, development, and survival of avian embryos, little is known about their plastic potential and ability to promote anticipatory maternal effects. Therefore, we examined how eggshell pore density and thickness are influenced by thermal conditions experienced by the mother. While we found no effect on eggshell thickness, we did find that mothers exposed to both heat treatments (Heat-Heat) laid eggs with 5.27 (±4.80, ±95% CI) more pores/cm2 than Control-Control mothers (LMM: t = 2.24, d.f. = 29, P = 0.033). Interestingly, this increase in pore density did not correlate with increased water loss, even at high incubation temperatures. This may be due in part to the lower heart rates of embryos from Heat-Heat mothers at high temperatures, which paired with the increased pores potentially signals a reduced demand for oxygen rather than limitations in gas conductance.
In zebra finches, an incubation temperature of just one degree Celsius above optimum has been shown to decrease both hatching success and lean body mass in male embryos (20, 29). In this study, we saw an effect of maternal treatment on embryo survival, although not in the manner we anticipated. We presumed that the reduced heart rates and higher pore density shown by embryos produced by Heat-Heat mothers would promote increased survival at high temperatures, however this was not the case. Among embryos exposed to the high incubation temperature, maternal treatment had no effect on survival (Fig. S2). However, at the control incubation temperature, embryos from mothers that experienced the mild heat as juveniles (Heat-Control) were 4.35 and 3.53 times as likely to survive and hatch successfully than those from mothers with “matching” treatments (Control-Control and Heat-Heat, respectively) (Fig. S2) (COXME: HC-CC: HR = 4.35, 95% CI = [0.469, 2.47], z = -2.88, P = 0.004; HC-HH: HR = 3.53, 95% CI = [0.383, 2.14], z = 2.81, P = 0.005). While maternal heat exposure resulted in physiological changes in embryos and survival at control incubation temperatures, it did not confer any benefits in survival in a “matching” environmental context, as expected to see in an anticipatory maternal effect. It is possible that the potential benefits or constraints of those physiological changes are only seen over a longer time-scale, and do not present themselves until later on in life.
While interpreting the adaptive significance of maternal effects is complex due to their highly context dependent nature, our study demonstrates the importance of designing experiments that incorporate several different time points and magnitudes of stressor exposure. Many of the effects on offspring phenotype we observed were the result of an interaction between maternal exposure to a mild and prolonged heat conditioning as juveniles, and high incubation temperatures experienced by their offspring. Such changes in offspring phenotype as the result of an interaction between “matching” maternal and offspring environment may signal the presence of anticipatory maternal effects, however, further research is required. Our findings support previous studies that have established early development as an important ontogenetic window for inducing transgenerational plasticity. The juvenile mild heat conditioning protocol used was not only effective at inducing maternal effects, but has also been previously shown to result in within-generation plasticity in this species (22). The duration of the mild heat conditioning likely played a significant role in its ability to generate both types of plasticity, as the 28 day treatment period represents a significant portion of the zebra finches developmental period. This finding supports the idea that environmental cues of a longer duration are more likely to induce anticipatory maternal effects, as they present a strong indicator that there will be high correlation between the maternal and offspring environment.
However, perhaps just as interesting was our finding that exposure to the high intensity heat stressor as adults did not induce a selfish maternal effect, or any effect independent of its interaction with the juvenile mild heat treatment. This is surprising, as the time just prior to and during a reproductive bout is considered another critical window for inducing maternal effects (10). Stressful environments can induce maternal effects at a cost to the mother, creating a position where mothers face a trade-off between self-maintenance and the current reproductive bout, often resulting in a decrease in offspring fitness (3). However, we observed no apparent detrimental effects of the maternal high heat stressor alone on the offspring or maternal phenotype (see 22). It is possible that trade-offs were avoided via plentiful resources, or the stressor was not strong enough or close enough to the reproductive bout to elicit a cost. Mothers had ad lib access to seed and water except when undergoing experimental treatments, potentially allowing them to increase food consumption to negate detrimental effects of the stressor. Often a longer environmental cue near reproduction will result in a stronger transgenerational effect (9, 10, 40, 41). If the high heat stressor mothers experienced as adults was a longer duration or closer to the reproductive bout, we may have observed a more pronounced effect. Another possibility is that we did not capture the phenotypic costs because we did not measure the affected traits, or because they were not present at the time-scale at which we measured. Potentially, the cost of this stressor exposure may only be seen at a cellular level (see 22), that may only present at a macroscopic level later in life as reduced longevity (42). Similarly, we measured phenotypic traits in offspring only prenatally and immediately after hatch, and an induced selfish maternal effect may not be apparent until later on in development. Traits have different capacities for plasticity and involvement in trade-offs, which underpins the importance of measuring multiple traits and factoring in life-history in future studies on maternal effects moving forward. (43).
Our results suggest that thermal conditions experienced by the mother early on in life can act as a maternal effect, with the potential to increase embryonic survival, but not in an anticipatory manner as predicted. Such plasticity could play a key role in species that live in regions experiencing temperatures close to their upper thermal tolerance, which are thought to be at greater risk of diminished populations as a result of climate change. Our results also indicate that thermal conditions experienced by the mother, even very early on in life, elicits a phenotypic change in offspring prenatally. Such information is critical, as the embryos of many avian species develop near their upper thermal critical limits and have very little thermoregulatory capabilities own their own (15, 16). Even though avian parents can regulate egg temperature, their methods to prevent hyperthermia when ambient temperatures rise above the optimum incubation temperature often comes at a significant cost to themselves (44). As climate change causes increased temperatures worldwide as well as an increase in extreme temperature events, the importance of understanding the rescue capabilities of phenotypic plasticity for species grows (2). Understanding physiological and morphological adjustments of embryos due to maternal effects could improve our ability to predict population responses to climate change and pinpoint species that will be most at risk. Future studies in maternal effects would do well to utilize multiple ecologically relevant stressor magnitudes and exposure time points, as well as measuring a suite of traits in offspring over the course of their life-history in varying environments.