Experimental evidence for beneficial effects of projected climate change on hibernating amphibians

Amphibians are the most threatened vertebrates today, experiencing worldwide declines. In recent years considerable effort was invested in exposing the causes of these declines. Climate change has been identified as such a cause; however, the expectable effects of predicted milder, shorter winters on hibernation success of temperate-zone Amphibians have remained controversial, mainly due to a lack of controlled experimental studies. Here we present a laboratory experiment, testing the effects of simulated climate change on hibernating juvenile common toads (Bufo bufo). We simulated hibernation conditions by exposing toadlets to current (1.5 °C) or elevated (4.5 °C) hibernation temperatures in combination with current (91 days) or shortened (61 days) hibernation length. We found that a shorter winter and milder hibernation temperature increased survival of toads during hibernation. Furthermore, the increase in temperature and shortening of the cold period had a synergistic positive effect on body mass change during hibernation. Consequently, while climate change may pose severe challenges for amphibians of the temperate zone during their activity period, the negative effects may be dampened by shorter and milder winters experienced during hibernation.

surprisingly scarce (for an exception see 30 ). Further, studies on potential effects of climate change on amphibians generally neglect effects on juveniles (for exceptions see e.g. 27,30 ), even though a sufficiently high survival rate and good body condition of juveniles are of fundamental importance for population persistence 27,31,32 .
Here we present a full factorial experimental test of the effects of rising winter temperature and shortening winter length on survival and body mass change of hibernating juveniles of an anuran amphibian of the temperate zone, the European common toad (Bufo bufo). We performed the experiment on a set of toadlets we had reared under controlled semi-natural conditions using mesocosms during the larval stage and outdoor enclosures after metamorphosis. In this way we controlled for individual variation in genetic relatedness and environmental conditions during early ontogeny that can have carry-over effects after metamorphosis 33,34 . We hibernated toadlets at current (1.5 °C) or elevated (4.5 °C) hibernation temperature in combination with current (91 days) or shortened (61 days) hibernation length. We weighed toadlets and documented mortality one week after termination of hibernation ( Fig. 1).
Overwintering survival was higher in scenarios with the higher hibernation temperature and in scenarios with shorter hibernation period (Table 1, Fig. 2). Also, survival increased with larger body mass before hibernation (mean ± s.e.m. for survivors: 1389 ± 28.64 mg, N = 325; died: 1070 ± 52.57 mg, N = 46) and differed between larval environments (Table 1, Supplementary Table S1, Fig. S1). None of the tested interactions were significant (Supplementary Table S2). Number of toes clipped had no significant effect on survival (Supplementary Table S3).
Body mass change during hibernation. Body mass one week after hibernation was positively related to mass before hibernation (linear mixed effects model, t = 36, r = 0.90, N = 325), and negatively to hibernation length (Table 1, Fig. 3). However, the interaction between hibernation temperature and hibernation length was also significant (Table 1): body mass of toadlets did not differ between cold and mild temperature regimes when exposed to shortened hibernation, but when toadlets were exposed to longer hibernation, animals overwintering at 4.5 °C weighed more after hibernation than toadlets that had overwintered at 1.5 °C (Fig. 3).

Discussion
Survival of toadlets during overwintering was highest when exposed to a shortened hibernation period and elevated average winter temperature (61 days at 4.5 °C), and lowest under current average overwintering conditions (91 days at 1.5 °C). This result contradicts the hypothesis that milder winters could lead to increased mortality in poikilothermic amphibians due to elevated metabolic rates during a long period without feeding 14,23 . Positive effects of milder winter temperatures and/or shorter winters on overwintering survival of anurans have also been found by a few correlative studies [26][27][28] . The only other experimental study testing how climate change may affect hibernation success of an amphibian 30 registered negligible mortality (3 out of 160 animals). Garner, et al. 30 manipulated overwintering conditions by exposing juvenile common toads to different hibernation lengths of 51 and 11 days at 4 °C, which might have already exposed toads to more beneficial conditions than currently experienced in nature [35][36][37] , thereby possibly underestimating effects of climate change on hibernation success. Although we applied a fairly constant temperature treatment in our study, our results likely are valid for natural populations. Anholt, et al. 26 found negative effects of highly variable winter temperatures on survival in a field study, but the recent report of the Intergovernmental Panel on Climate Change shows that low-percentile winter temperatures are predicted to increase faster than the mean, thereby decreasing temperature variability 17 . This way, if we had included variability in our treatments, our simulated climate change scenario would have been not only warmer but also less variable than the current climate scenario; so it is likely that we would have found even more pronounced differences in favor of the climate change group. Our results, thus, bolster up conclusions of previous correlative studies by providing experimental evidence for the hypothesis that temperate-zone anurans, including common toads, may benefit from mildly increasing temperatures due to climate change via increased winter survival. Nevertheless, further experiments are needed to scrutinize the effects of temperature variability coupled with warming.
Body mass after hibernation was very similar among toadlets hibernating for 61 days, regardless of temperature, and was higher than in animals hibernating for 91 days. This may have simply resulted from a shorter time period without food intake pertaining smaller loss of body mass. On the other hand, low temperatures seemed to enhance the negative effects of a long hibernation period. Glycogen and lipids are the main energy source for amphibians during hibernation 38 , but glycogen can also be utilized in the production of cryoprotective sugars 39,40 . Although there is no direct evidence that common toads can harness glycogen this way, they can survive subzero temperatures of − 3 °C for up to 22 hours 40 . Thus, one possible contributor to the observed pattern may be that toadlets hibernating at 1.5 °C depleted their energy reserves twofold: by maintaining normal physiological  functions and also deploying glycogen as a response to low temperature. Body mass before hibernation was also positively related to survival and post-hibernation mass, which is in accordance with results of a previous field-based study on juvenile anurans 33 . However, Garner, et al. 30 did not find a significant effect of hibernation regime and mass before hibernation on proportional mass change of toadlets during hibernation, perhaps because overwintering conditions were benign in all of their scenarios (see above). Also, while in our study T min was − 0.77 °C, already low enough to trigger physiological mechanisms preventing freezing, Garner, et al. 30 never exposed toads to temperatures approaching or below 0 °C (T min was 4 °C), rendering the production of cryoprotective agents unnecessary.
In conclusion, under simulated current hibernation conditions in our study 67% of toadlets survived, which is within the range of overwintering survival estimated for natural conditions 26,28,41 , whereas under simulated climate change conditions survival was close to 100% one week after hibernation. Moreover, when exposed to a shorter and/or milder winter, survivors also had larger body mass after hibernation, which is likely to have long-lasting effects by enhancing further survival and fecundity 14,23,42 . While other consequences of climate change, such as an increasing probability of mass mortalities in aquatic larvae due to premature pond-desiccation, the spread of invasive predators and infectious diseases are likely to have negative effects on anuran populations located in the temperate zone and at high altitudes and latitudes 1,3,13,25,30,43,44 , the benefits arising from milder and shorter winters may somewhat dampen these effects on population persistence 27,31,32 . Further studies testing the general applicability of our results across several amphibian taxa, under different climate change scenarios and during several life stages are urgently needed. These investigations will deliver important insights regarding the expected effects of future climate change on amphibians, allowing the design and execution of informed and effective conservation actions protecting this highly threatened group of animals.  (Notonecta sp.), larvae of the southern hawker (Aeshna cyanea), juvenile sticklebacks (Gasterosteus aculeatus) and adult, male smooth newts (Lissotriton vulgaris).
We transferred metamorphosing toadlets into slightly tilted 45 L boxes containing egg boxes as shelter, and fed them ad libitum with springtails (Folsomia sp.), woodlice (Trichorhina tomentosa), fruit flies (Drosophila hydei) and hatchling crickets (Gryllodes sigillatus). After all individuals metamorphosed, we transported them to the Konrad Lorenz Institute of Ethology, Vienna, Austria. We marked toadlets of 10 families individually by toe-clipping, a method widely used for marking of anuran amphibians 45,46 , and assigned them randomly to 16 outdoor enclosures (set up similar to 33 ). Each enclosure received one toadlet from each family by larval environment combination, resulting in 50 animals per enclosure.
We recaptured toadlets from enclosures on 14-15 October and let them defecate in groups of 15-20 individuals in 45 L plastic outdoor containers filled with moist leaves to prevent desiccation. On 17 October after individual identification and measuring their body mass to the nearest mg, we transferred each toadlet separately into a 50 ml centrifuge tube (11.5 cm × 3 cm, length × diameter) filled with 20 ml of a 1:1 mixture of sterilized and dried (2 hours at 170 °C) soil and sand moistened with 5 ml aged tap water to allow burying and prevent desiccation. Tubes were covered with a mosquito net to allow air transfer and stored in a close to horizontal position (~5°). Toadlets had only a limited space available to move around during the experiment, but not much more may be available to animals in natural hibernacula 47,48 . We placed tubes into a lab refrigerator with forced air convection (Miele, Gütersloh, Germany) set to 10 °C. Three days later, we lowered the temperature to 7.5 °C and maintained it for three weeks to simulate transient autumn temperatures (Fig. 1).
We randomly selected 372 individuals out of 403 survivors and assigned 93 toadlets randomly to each of the four hibernation scenarios, while balancing for body mass, larval environment, family and enclosure (for details see Supplementary Table S4). The simulated current average hibernation temperature of 1.5 °C is well within the natural range experienced by common toads during winter 35,41 and a 3 °C increase in hibernation temperature (from the current 1.5 °C to 4.5 °C) could be expected to be of significance for the toadlets' performance during hibernation 36,37 , while not exceeding projected temperature rise due to climate change 17 (for further justification of the chosen temperatures see the Supplementary Information). We set the length of the current hibernation scenario based on average monthly temperatures registered in Hungary between November and March during the period of 1950-2000 49 , and on reported durations of hibernation in common toads 35,41 . Shortening of the hibernation period by 30 days is supported by previous publications on shorter snow-cover duration 50 and elongated growing season in the Northern Hemisphere 51 .
We initiated hibernation on 10 November 2015 by relocating toadlets into two laboratory refrigerators with forced air convection (Pol-Eko-Aparatura, Wodzisław Śląski, Poland). We recorded temperature hourly using data loggers (Onset, Cape Cod, MA, USA; for measured temperatures see Supplementary Table S5). Once a week we sprinkled aged tap water into tubes to prevent desiccation and relocated tubes within refrigerators haphazardly to account for spatial variation in temperature. Sixty-one and 91 days after initiation of hibernation, we transferred toadlets from both 1.5 and 4.5 °C to a refrigerator set to 7.5 °C (Fig. 1). Apart from phases of sprinkling and relocating, toadlets overwintered in complete darkness and without food throughout the study. For further details on the materials and methodology see the online Supplementary Information and Supplementary Table S4 Statistical analyses. We analyzed survival using generalized linear mixed-effects modeling procedures with binomial error distribution and logit link function. Effects on post-hibernation body mass were evaluated using general linear mixed-effects models. Mass was log 10 -transformed prior to analysis. We applied a forward model-selection approach based on P-values (α = 0.05), testing the main effects of larval environment, hibernation length and hibernation temperature as fixed factors, body mass before hibernation as a covariate, tadpole family and enclosure as crossed random factors, and all testable two-and three-way interactions among fixed effects, including the covariate. We used a forward model-selection approach to ensure a standardized method of analysis for all studied response variables, since some interactions with larval environment could not be estimated for survival because in some subgroups there was no variance (all individuals survived or died, Supplementary Tables S2, S6, S7), which rendered backwards model selection unfeasible. We included larval environment as a fixed effect in the analysis to control for potential carry-over effects after metamorphosis 33,34 . We ran all analyses in R 3.1.3 52 . For the analysis of survival, we used the 'glmer' function in the 'lme4' package 53 , with the 'bobyqa' optimizer function to handle computational problems arising during the model-selection process. When analyzing variation in body mass we used the 'lmer' function in 'lme4' . P-values were calculated using ' Anova' in the 'car' package 54 . We conducted post-hoc pairwise comparisons by calculating linear contrasts corrected for false discovery rate 55 using 'glht' in the 'multcomp' package 56 . We discarded one toadlet from the analyses, due to missing data (Supplementary Table S4).