Geographic variation in maternal investment and trade-offs between egg size and clutch size in an endemic toad of the Qinghai-Tibet Plateau.

Life history theory predicts that animals often produce fewer offspring of larger size and indicate a stronger trade-off between the number and size of offspring to cope with increasing environmental stress. In order to evaluate this prediction, we tested the life history characteristics of Bufo minshanicus at eight different altitudes on the eastern Tibetan Plateau, China. Our results revealed a positive correlation between female SVL and clutch size or egg size, revealing that larger females produce more and larger eggs. However, high-altitude toads seem to favor more offspring and smaller egg sizes when removing the effect of female SVL, which is counter to theoretical predictions. In addition, there was an overall significantly negative relationship between egg size and clutch size, indicative of a trade-off between egg size and fecundity. Therefore, we suggest that higher fecundity, rather than larger egg size, is a more effective reproductive strategy for this species of anuran living at high-altitude environments.

selective pressure. Thus, we investigated altitudinal variations in Bufo minshanicus life history traits, focusing on a trade-off between egg size and clutch size. So far, the taxonomic status of B. minshanicus has been controversial for over half a century. Fu et al. (2005) 24 suggested B. minshanicus and Bufo andrewsi were treated as a single species 25 , Bufo gargarizans, without subspecies division. However, Fei and Ye (2000) 26 treated B. minshanicus as a subspecies of B. gargarizans. In this study, we defined B. minshanicus as a valid species. Bufo minshanicus is a species endemic to high altitude regions of the eastern Tibetan Plateau in China. They are widely distributed in forests, fields, and open alpine marshes from 2500 to 3700 meters (m) altitude and one of the few amphibian species to live at such high altitudes globally (Fei and Ye 2001). Geographically widespread species that occupy many thermal environments provide testable models for understanding the evolution of life-history responses to altitude. As a typical explosive breeder 27 , Bufo minshanicus exhibits a short breeding season (5-18 days). Females are larger than males 28 and clutch size increases with increasing female SVL 29,30 . Low latitude populations in the eastern Qinghai province select deep swamps as hibernation site, while high latitude populations in the southern Gansu province prefer rabbit or otter holes as hibernating sites. Studies across altitudes provide a good approach to understand life-history evolution in stressful environments. Here, we predict that altitude would be (1) negatively correlated with clutch size, (2) positively associated with egg size, and (3) positively correlated with female SVL. We also tested the prediction that (4) there is a trade-off between egg size and clutch size.

Materials and methods
Data collection. We collected Bufo minshanicus individuals from eight populations (all collected individuals from the same pond during the breeding period were defined as a population, at elevations ranging from 2506 to 3478 m) along the eastern Tibetan Plateau, China. At these sites, annual mean air temperatures decrease with increasing altitude or decreasing latitude, whereas annual total precipitation increases with increasing altitude or decreasing latitude ( Table 1). The Gahai-Zecha National Nature Reserve Management Bureau approved this project (approval number GHZCRMB/03-212014), and gave permission for fieldwork. All animals were collected under the guidelines for animal care in China. Handling and processing of toads followed approved protocols from the Animal Scientific Procedures Act 1988 by the State Department of China. A total of 683 amplectant pairs were collected during the peak period of breeding activity from 2009 to 2015 (Table 1; Dataset 1). We transported them to our field lab located close to breeding sites. Each amplexed pair was put into a plastic container (2-liter) until the eggs were deposited. After counting eggs for each clutch and measuring SVL of both sexes to the nearest 1 mm using a ruler, the eggs and adults were returned to the breeding pond.
About 100 eggs were randomly selected from each clutch and placed on a glass plate with calipers to take a digital image. We obtained measurements of individual egg sizes (n = 20-30; ±0.01 mm) excluding the jelly using a tpsdig2 computer software. Clutch volume was calculated by multiplying the number of oviposited eggs with the volume of each egg (Vs = 4/3πr 3 , where r is the radius). We obtained average ambient temperatures and precipitation from 2009 to 2015 from the Chinese Meteorological Administration (http://www.cma.gov.cn) and the Gansu Gahai-Zecha National Nature Reserve Management Bureau.
Statistical analyses. Clutch size, egg size and clutch volume were analyzed using general linear mixed models (GLMMs) where population was treated as a random factor, and altitude and latitude as covariates. Egg size and clutch size were used as covariates of one another when testing for a trade-off between them. To investigate variation in body size of females, we used GLMMs where female SVL was the dependent variable, latitude and altitude as covariates, and population as a random factor.
In the clutch-by-female analyses, we treated clutch size, egg size and clutch volume as the dependent variable and female SVL, latitude and altitude as covariates. Finally, to investigate the trade-off between egg size and clutch size, we used egg size or clutch size as the dependent variable, the other as the independent variable, and then the other way around. Prior to analyses, we used log-transformed raw data on body size, clutch size, egg size and clutch volume. Meanwhile, we removed one population (3211 m) because only 3 female toads were collected. The analyses were started with a full model, with all possible interactions, and then removing non-significant interactions (P > 0.1, for details, see Räsänen et al. 2008 21 ) because interaction effects in many cases may be statistically weaker, but biologically important. Thus, in this case, it would be important to report the effect sizes rather than actual P values. For the R 2 values, as we used GLMM with both fixed effects and random factors, it thus was www.nature.com/scientificreports www.nature.com/scientificreports/ needed to provide the R 2 marginal (R 2 m ) as the amount of variation explained by fixed factors, and R 2 conditional (R 2 c ) as amount of variation explained by both fixed and random factors. All the analyses were done with IBM SPSS Statistics 20.0 (IBM Corp, Armonk, NY, USA).

Discussion
In order to maximize fitness, life history traits vary under different environmental pressures 31,32 . In this study, our results showed that maternal investment strategies differ among altitudes. When the effect of body size is removed, the clutch data across seven B. minshanicus populations indicated that females are more fecund but produce smaller egg sizes at higher altitudes than at lower altitudes. The clutch-by-female size further showed a population-level trade-off between fecundity and egg size.
However, if the outlier population (at 3211 m) was not removed, the clutch data across eight B. minshanicus populations indicated that egg size and clutch volume tend to decrease or increase with increasing altitude (  Relationship between altitude or latitude and egg size (a), clutch size (b) and total clutch volume (c) in 7 Bufo minshanicus populations. Solid lines: P < 0.05; dashed line: P > 0.05. Data points are population means. Relative log 10 was generated from regression of log 10 (egg size) or log 10 (clutch size) or log 10 (total clutch volume) on log 10 (female body size). Statistical details are as in Table 3.
In this study, female SVL was not significantly positive correlated with latitude or altitude. This result was not consistent with previous evidence considering six populations 28 . The shorter seasonal period of activity, weaker digestion and lower energy of prey may limit larger body size for appearing at high altitudes or latitudes. Moreover, females living in higher-altitudes must store enough energy during the short activity period in order to survive the harsh and extended winter, which may become more difficult for larger individuals. Therefore, Chen et al. (2013) 4 suggested that larger females at high altitudes may be costly.
Life history theory suggests that organisms should allocate the limited energy between growth and development, thus creating a trade-off between growth and development 1 . The clutch data across seven B. minshanicus populations revealed that clutch attributes (clutch size, egg size, and clutch volume) were positively correlated with female size, revealing that larger females produce more eggs, larger eggs and increase reproductive effort. In fact, most amphibians and reptiles, in general, exhibit this relationship between body size and reproductive output. Thus, it is attributed to positive fecundity selection, where female body size is larger for larger clutches 33 . Like most ectotherms with indeterminate growth, female B. minshanicus fit the von Bertalanffy's model 28 , which describes rapid somatic growth for earlier stages followed by slower growth thereafter. This implies that a larger fraction of energy would be devoted to reproduction as individuals become older 34,35 , thus resulting in age-specific reproductive output.
In some anurans, larger clutches in high-altitude or latitude populations are due to the larger size of high-altitude or latitude females 16,22,36,37 . When removing the effect of female body size, females at high altitudes produce relatively fewer clutches, but larger eggs than conspecifics from low altitudes 5,16,23,38 . However, in this study, after controlling for the effect of female size, higher altitude females had higher fecundity and smaller egg sizes with similar reproductive effort than lower altitudes females, thus the endemic plateau toad has evolved a different breeding strategy. This result was consistent with a previous study 39 , which showed that the phrynosomatid lizard Phrynosoma cornutum had smaller females at higher latitudes that produced larger clutches but similarly sized eggs when compared to larger females at lower latitudes and removing the effect of female size. It is well known that rainfall can directly affect the abundance of food, and may indirectly affect fecundity by reducing fat stores, which is necessary for development of a holistic egg number 40 . Females may have more resources for reproduction and clutch production if food availability is higher and physiological stress is lower. Interestingly, in this study, rainfall increases with increasing altitude, indicating higher food availability at higher altitudes. In this case, we speculated that resource may not limit clutch production at higher altitudes. Meanwhile, this species selects rabbit or otter holes as hiding and hibernating sites in the field. In this case, they reduce chances of predation and protect against extreme temperature fluctuations, especially in summer and winter. Additionally, disease and predation tend to decline when increasing altitude 41 . Thus, we speculate that this species does not allocate as much energy to survive the harsh and extended winter, and suggest females living at high altitudes likely allocate more energy for producing clutches. Similarly, Dobzhansky (1950) 42 suggested that r-selection strategies were more likely to appear in changing environments (e.g. temperate and high altitude regions), whereas K-selection more likely to be found in relatively stable situations (e.g. tropics). As a consequence, larger clutch size is regarded  Table 3. Analyses of covariance of egg size, clutch size, and total reproductive output (clutch volume) in females from 7 high-altitude Bufo minshanicus populations in a model including female size. **P < 0.001,*P < 0.05.

Source of variation
www.nature.com/scientificreports www.nature.com/scientificreports/ as an attribute that populations have evolved through time as an adaptation to their general environment and their ecological niche 43,44 .
Life-history theory emphasizes that high-altitude females are expected to select for the production larger eggs 2-6 because larger eggs or offspring will lead to larger initial sizes, faster growth speeds, higher developmental speeds, or both 7-12 . However, our results showed egg size across seven B. minshanicus populations decreased with increasing altitude, which provide support for the notion that life-history strategies respond to environmental conditions. Similarly, in some instances smaller eggs are better 45 or some studies report no significant difference between egg size and altitude 46,47 . Other studies found that some species with small eggs will increase the number of offspring 48 , especially in the uncertain larval environment. Additionally, small eggs have a relatively large surface-to-volume ratio and require relatively little oxygen during embryonic development 49 . In this case, this animal may better be able to adapt to the anoxic environment of the Tibetan Plateau.
Although some studies suggest that trade-offs between clutch size and egg size in amphibians seldom occurs 4,50-53 , some previous findings 5,9,54,55 support the existence of a trade-off. In this study, although intra-population-level trade-off between fecundity and egg size was uncommon, there was a significant inter-population-level trade-off between fecundity and egg size, suggesting that females producing greater numbers will produce smaller eggs to compensate for the energy put into the greater numbers, which are consistent with the general principle of MacArthur and Wilson's theory of r-versus K-selection in populations 56,57 . Moreover, higher altitudes influenced investment in egg size, as suggested by the fact that there were stronger negative effects on egg size than on clutch size and reproductive output. Our data also showed that the strength of the trade-off between clutch size and egg size was not different among altitudes and latitudes. However, theoretical 53,58 and empirical 9,21,54 studies showed that harsh environments promote larger maternal investment in per-individual offspring to improve survival of individual offspring. For B. minshanicus, at higher altitudes, the advantage associated with producing smaller eggs because of increasing fecundity, which potentially maximizes the number of offspring surviving to reproduction 10 . Therefore, different environmental pressures will lead to geographical variation in life history traits aimed at maximizing fitness 31,32 . Bufo minshanicus have evolved a different life history strategy to adapt their environments, which may be inconsistent with our expectations. On the Tibetan Plateau, it appears as though more fecundity and smaller egg size are traits that enhance fitness for this animal.