Maternal effects are long-lasting and influence female offspring s reproductive strategy in the swordtail fish Xiphophorus multilineatus

Transcription

1 doi: /jeb Maternal effects are long-lasting and influence female offspring s reproductive strategy in the swordtail fish Xiphophorus multilineatus A.D.MURPHY,D.GOEDERT&M.R.MORRIS Department of Biological Sciences, Ohio University, Athens, OH, USA Keywords: growth rates; maternal effects; reproductive strategy; swordtail fish; Xiphophorus. Abstract The adaptive benefits of maternal investment into individual offspring (inherited environmental effects) will be shaped by selection on mothers as well as their offspring, often across variable environments. We examined how a mother s nutritional environment interacted with her offspring s nutritional and social environment in Xiphophorus multilineatus, a live-bearing fish. Fry from mothers reared on two different nutritional diets (HQ = high quality and LQ = low quality) were all reared on a LQ diet in addition to being split between two social treatments: exposed to a large adult male during development and not exposed. Mothers raised on a HQ diet produce offspring that were not only initially larger (at 14 days of age), but grew faster, and were larger at sexual maturity. Male offspring from mothers raised on both diets responded to the exposure to courter males by growing faster; however, the response of their sisters varied with mother s diet; females from HQ diet mothers reduced growth if exposed to a courter male, whereas females from LQ diet mothers increased growth. Therefore, we detected variation in maternal investment depending on female size and diet, and the effects of this variation on offspring were long-lasting and sex specific. Our results support the maternal stress hypothesis, with selection on mothers to reduce investment in low-quality environments. In addition, the interaction we detected between the mother s nutritional environment and the female offspring s social environment suggests that female offspring adopted different reproductive strategies depending on maternal investment. Introduction Evidence has been accumulating across a wide range of taxa for the adaptive value of inherited environmental effects (Rossiter, 1996), particularly in the context of environmental matching. When environmental stability creates predictability of offspring s environment based on the maternal environment, mothers can adjust the phenotype of their offspring to match local conditions and maximize offspring s fitness (Fox et al., 1997; Agrawal et al., 1999; Agrawal, 2001, 2002; Galloway, 2001; Kudo & Nakahira, 2005; Uller, 2008). For example, female crickets (Gryllus pennsylvanicus) exposed to Correspondence: Molly R. Morris, Department of Biological Sciences, Ohio University, Life Sciences Building 243, Athens, OH 45701, USA. Tel.: ; fax: ; morrism@ohio.edu predation risk produced offspring that exhibited more antipredator behaviours (Storm & Lima, 2010), and female bryozoans (Bugula neritina) exposed to the toxicant copper produced offspring with increased resistance to copper (Marshall, 2008). Adjustment of egg size, indicating the nutritional investment of the mother, is a widespread type of inherited environmental effect (reviewed by Bernardo, 1996) mediating environment phenotype matching. For example, in many species of fish, females produce larger offspring in unfavourable environments, such as high density (e.g. Gregersen et al., 2006; Leips et al., 2009; Schrader & Travis, 2012) or low productivity (e.g. Johnston & Leggett, 2002). As larger females often produce more eggs (larger broods), it has been suggested that they create an environment for their offspring of increased competition (higher density). The sibling competition hypothesis predicts that larger mothers would produce ª 2014 THE AUTHORS. J. EVOL. BIOL. 1

2 2 A. D. MURPHY ET AL. larger offspring to mitigate the negative consequences of increased sibling competition (Parker & Begon, 1986; Beckerman et al., 2006). Marshall and Uller (2007) argue, however, that the primary selection on maternal investment is maternal fitness, not offspring s fitness. Therefore, the influence females exert on their offspring s phenotype might not always be beneficial to the offspring (reviewed by Bernardo, 1996; McCormick, 2006). Reduced maternal investment in poor-quality environments could be due to constraints, or as an adaptive strategy by females to increase their lifetime fitness at the cost of the fitness of current offspring (Marshall & Uller, 2007). Reduced parental investment could result in parent offspring conflict, which can drive offspring to evolve counterstrategies to maternal strategies, such as begging and scramble competition where offspring s attempt to control maternal allocation (Parker et al., 2002). Other responses by offspring to reduced maternal investment could include compensatory growth (i.e. when offspring from smaller eggs have faster growth rates than offspring from larger eggs; Ali et al., 2003; Hector & Nakagawa, 2012) or changes in offspring s life-history strategies. We were interested in determining whether maternal investment in the swordtail fish (Xiphophorus multilineatus) fits the predictions of environment phenotype matching hypotheses. We examined how a mother s nutrition (reared on low- or high-quality diet) and her size influenced the growth rate and size at sexual maturity of her offspring, when the offspring were all reared in a low-quality environment. We also wanted to determine whether and how maternal investment would influence the response of juveniles to environmental variation other than diet: the social context of an adult male during development. Social context during development is an important environmental condition for organisms such as swordtail fishes, which live in habitats that vary from small pools of a few individuals to larger rivers where fish associate in large, mixed sex groups and individuals of all life stages interact (Endler, 1983; Reznick & Yang, 1993). Social context is known to influence age at sexual maturity in Xiphophorus helleri: when exposed to an adult male during development, females matured earlier and males later (Walling et al., 2007). The sexual difference in response to social context was hypothesized to be adaptive. For males, the presence of a large adult male represents a sexual competitor, and delaying sexual maturity would potentially produce larger males (more time to grow) that would be more competitive. For females, the adult male represents the presence of a potential mate, in which case reaching sexual maturity sooner would increase the probability that females would mate sooner (Walling et al., 2007). In this case, individuals of both sexes could have modified age at sexual maturity through growth rate and/or size at sexual maturity. We tested the hypothesis that X. multilineatus females vary their maternal investment by examining the growth rates and size at sexual maturity for the offspring of mothers raised on two different diet regimes. Even though age and thus size at sexual maturity for males in this species are influenced by copies of the MC4R gene on the Y chromosome (Lampert et al., 2010), with males unlike females ceasing to grow after sexual maturity, growth rates can influence variation in the age and size at sexual maturity to some extent for both sexes and therefore carries significant implications for offspring s fitness. As larger adults, individuals have a higher probability of evading predators (Rosenthal et al., 2001; Royle et al., 2006) and an increased reproductive success. Larger females have higher fecundity (Moyle & Cech, 1988; Nichol & Pikitch, 1994), and larger males gain an advantage through both intrasexual (Morris et al., 1992) and intersexual selection (Morris, 1998; Morris et al., 2006). In addition, growing faster increases the probability of ever reaching sexual maturity. All the offspring were raised on low-quality diets, and therefore, the environment phenotype matching hypothesis would predict that mothers raised on lowquality diets would produce offspring that were better adapted to this environment (i.e. larger fry, faster growing, larger at sexual maturity). Second, we tested the hypothesis that the growth rates and size at sexual maturity of males and females would respond differently to the manipulation of social context, based on the findings in X. helleri by Walling et al. (2007). Male fry exposed to adult males were predicted to have either higher growth rates and/or mature at larger sizes than males in isolation as X. helleri males reached sexual maturity later in the presence of an adult male competitor. Females, on the other hand, should perceive the adult male as a potential mate. Hence, to reach sexual maturity sooner, they are predicted to either grow faster and/or mature at smaller sizes. And finally, we tested the hypothesis that variation in maternal investment due to mother s diet could influence offspring s response to social context. Materials and methods Experimental individuals Offspring from 19 broods were split between a high- or low-quality diet, differing primarily in percentage of crude protein, and raised in isolation (see Lyons et al., 2013). High-quality (HQ) diet consisted of Tetra-min Tropical Flakes (47% protein) 2X/day and bloodworms 3X/week, and low-quality (LQ) diet of Nishikoi Wheat Germ Koi Food (20% protein) 2X/day. Protein content below 30% has been shown to reduce growth, reproductive performance and influence gonadosomatic and hepatosomatic indexes in X. helleri (Ling et al., 2006). Upon reaching sexual maturity, twenty-five females

3 Maternal effects and reproductive strategies 3 from this previous study, 15 from the LQ diet and 10 from the HQ diet, were each placed with two mature courter males (mating blocks, n = 7) from the largest size class (Y-L) for a period of one week. Five pairs of sisters, with one sister raised on each diet, were included in the study. The other 15 females all had different mothers. Females (n = 25) were then returned to individual tanks and allowed to drop fry. First broods were removed for use in another study (unpublished) and were thus unavailable for use in this experiment. When females dropped their second brood, the mothers were removed from the brood tank and photographed next to a millimetre scale using a Canon â Powershot SD 1200IS (Canon U.S.A. Inc., Melville, NY, USA), and the images analysed using IMAGEJ â (Rasband, W.S., NIH, Bethesda, MD, USA) to determine standard length. Size data were collected for only 22 of the 25 females. Fry from the second brood were counted and maintained as a group until they reached two weeks of age. At fourteen days of age, up to eight fry per brood (total 175, mean #individuals/brood = 7) were photographed and measured for standard length (as for mothers above), and isolated in separate ten-gallon glass aquaria ( cm) screened with opaque paperboard sheets for visual isolation. Half of each brood was assigned to a no exposure control treatment. Tanks were maintained in two environmental chambers with a constant photoperiod and temperature of 13:11 h L:D and 22 C, respectively. Fry were fed on the LQ diet treatment (see Lyons et al., 2013) of Nishikoi â (Whitehall, Wethersfield, Essex, UK) wheatgerm pellets ground finely into smaller particles ad libitum for five minutes once daily. After the allotted five minutes, any remaining food in the tank was netted and removed to prevent the growth of excess algae in the tanks that could supplement the juveniles diet. All tanks were scrubbed and half-drained accompanied by the addition of ten millilitres of StressCoat â (Aquarium Pharmaceuticals API, Chalfont, PA, USA), ten millilitres of Stress- Zyme â (Aquarium Pharmaceuticals API) and 1/8 of a cup of salt, and then refilled to inhibit the proliferation of algae as well as eliminate wastes once a month. 70 days of age. Every thirty days subsequent to this first exposure, they were exposed to a new male until every juvenile in this treatment had been exposed to a total of four distinct males (i.e. ages 70, 100, 130 and 160 days). Fry in the control treatment were exposed to empty Breed-n-Show tanks; during the same periods, treatment individuals were exposed to mature males. Juveniles were photographed on the first day of each exposure, following the same procedure used to determine the size of their mothers. All juveniles were checked weekly for signs of sexual maturity: gonopodium formation by specialization of the anal fin in males and gravid spot appearance in females (Rosen, 1960). Individuals were also photographed at age of sexual maturity. Measurements of growth Growth rates were calculated as size gain (SLage 1 SLage 0 ) over the number of days they grew (age 1 age 0 ). We calculated growth rate over two time periods so that we could compare growth before and after treatment: growth rate prior to exposure (i.e. from age 0 14 to age 1 70 days; GR1) and growth rate subsequent to exposure (i.e. from age 0 70 to age days, GR2; Fig. 1). GR2 only considered growth prior to 130 days of age as many individuals reached sexual maturity before the final exposure at age 160 days (mean SE: days, range: days in males; mean SE: days, range: days in females). Statistical analysis All analyses were performed using R statistical software (R Development Core Team, 2013). We first investigated how the diet treatments influenced the mother s standard length. Lyons (2011) had found that females raised on HQ diet were larger at sexual maturity than sisters raised on LQ diet, and therefore, we verified that Courter male exposure protocol Twenty-eight courter males measuring mm SL were used for the exposure treatment. Adult courter males were singly exposed to an individual fry via a plastic Penn Plax â (Hauppauge, NY, USA) Breed-nshow box ( cm) that was immersed inside the fry s home tank. The Breed-n-show boxes were constructed with apertures at the top and bottom of the container so that juveniles could receive both visual and chemical (pheromone) cues from the exposure males, but prevented physical contact. Fry in the exposure treatment were exposed to a mature male for three days at a time, starting at Fig. 1 Experimental design. Size (standard length, mm) was measured at three times during development: 14, 70 and 130 days. Two growth rates were calculated: GR1 which provided a rate prior to the initiation of the social context experiment, and GR2 which allowed for a comparison between individuals exposed to adult males, and control individuals that were not exposed to other fish during development.

4 4 A. D. MURPHY ET AL. the diet treatments had influenced the size of the subset of females used as mothers in the current study as well. We used a mixed-effect model with the mother s brood of origin as a random factor, correcting for pseudo-replication (Zuur et al., 2009). We then determined the factors influencing four offspring s fitness-related traits with four different mixedeffects models. If mothers reared on low-quality diets invest more in individual offspring to provide them with a fitness advantage in poor environments (all offspring were reared on LQ diets), then offspring from LQ diet mothers should have one or more of the traits associated with increased offspring s fitness (e.g. larger size at 14 days, faster growth rates before (GR1) and after (GR2) social treatment and/or larger size at sexual maturity). If males and females responded to social experience by adjusting their growth rates in different directions, we would expect an interaction between sex and social treatment for GR2. Finally, if maternal investment influenced the response of males or females to the social experience, we would expect an interaction between social treatment and maternal diet to influence GR2 and/or size at sexual maturity. For all models examining variation across offspring, we included as random factors the mother s identity, the mother s brood of origin and the mating block to which the mother was assigned. We initially fitted the complete (i.e. all fixed and random effects) model with maximum likelihood. Random effects presenting variance of zero were immediately removed, as they correspond to a model without that effect (see pp. 11 in Bates, 2010). For the remaining random effects, we performed log-likelihood ratio tests (LRT) between the most complete model (with the random effect) and an identical model with the respective random effect removed, until the final structure was obtained. The final structure was considered as the one containing all random effects that, when removed from the model, significantly decreased the likelihood estimation. We considered as the level of significance a P-value of 0.10, as the LRT between models is considered to be conservative (Pinheiro & Bates, 2000; Bates, 2010). Both mating block of the mothers and mother s brood of origin were excluded from the random structure of all four models, as LRT indicated they did not significantly improve the fitting of the model (P > 0.10). Mother s identity was always maintained in the models. For the structure of the fixed effects, we refitted the models with REML (e.g. pp in Zuur et al., 2009). An initial, complete model for the response variable size at 14 days included as fixed effects the following: brood size and mother s size (as covariates); sex, maternal diet treatment and their interaction. We also fit a second model including social treatment as an additional fixed effect to ensure no initial bias in the assortment of individuals into treatment groups. We initially included as fixed effects in the model of GR1 the following: size at 14 days; sex, maternal diet and their interaction. The GR2 complete model included as fixed effects the following: size at 70 days; sex, maternal diet, social treatment and their interactions. Finally, the complete model for size at sexual maturity included as fixed effects sex, maternal diet, social treatment and their interactions. For each model, we removed interactions among variables if P > For models including the variable mother s size at time of parturition, three mothers and all their offspring were excluded due to missing data on mother s size, resulting in a sample size of 22 mothers (from 18 broods) and 158 fry. Also, two offspring were removed from the analysis of size at sexual maturity due to missing data. Significance levels and confidence intervals for the estimates of each explanatory variable were obtained using a Markov chain Monte Carlo method with simulations using the package LANGUAGE-R (Baayen, 2011). Continuous variables were centralized for the inclusion in the analyses. We verified the assumptions of all models following Pinheiro and Bates (2000); pp We used normal probability plots of the estimated random effects to access normality of random effects and a normal probability plot of the residuals to check for normality of the within-group errors and to verify whether distribution of residuals is centred at 0. Homogeneity of variance was verified by visual inspection of a plot of the standardized withingroup residuals by the within-group fitted values and a plot of the residuals against the fitted values by each of the explanatory variables. These plots can also indicate the presence of outliers and patterns of nonlinearity in the model (e.g. Pinheiro and Bates, pp. 186). Assumptions for the mixed-effects models were met for all models presented, with two, one and four outliers removed of the models on size at 14 days, GR1 and size at sexual maturity, respectively. Finally, we calculated the marginal and conditional R 2 values, representing, respectively, the variance explained by fixed effects only and the variance explained by both fixed and random effects, following Nakagawa & Schielzeth (2013). Results There was a significant difference in the mean size of mothers from the high-quality diet treatment and mothers from the low-quality diet treatment (P MCMC = ; n = 22 mothers, 18 broods of origin). Mothers on the HQ diet were larger in size (HQ diet mean SD = mm, 95% CI = ; LQ diet mean SD = mm, 95% CI = ). Fry size at 14 days was influenced by mother s size, brood size, sex of the offspring and mother s diet (Table 1; marginal R 2 = 0.42, conditional R 2 = 0.83). Whist HQ mothers produced larger fry than LQ mothers,

5 Maternal effects and reproductive strategies 5 Table 1 Offspring s size at 14 days was influenced by brood size, maternal standard length (SL), maternal diet and offspring s sex. Estimates and standard errors obtained from mixed-effects model (random effect = mother identity), with respective P-values obtained from MCMC simulations (n = 156 individuals, 22 mothers). Estimate SE P Intercept < Brood size (scaled) < SL mother (scaled) < Diet (LQ) < Sex (M) within each diet group, larger females produced offspring that were smaller at 14 days than smaller females (Fig. 2a) and offspring born into larger broods were smaller than offspring born into smaller broods (Fig. 2b). In addition, male fry were larger than female fry at 14 days (estimates SE: male HQ = mm, female HQ = mm, male LQ = mm, female LQ = mm). Variance associated with mother s identity represented 40.86% of the variance in the model for fry size at 14 days. Growth rate prior to social context treatment (GR1) was influenced by offspring s size at 14 days, sex of the offspring and the mother s diet (Table 2; marginal R 2 = 0.32, conditional R 2 = 0.56). GR1 was negatively related to offspring s size at 14 days (slower for offspring of larger size at 14 days; Fig. 3). Male offspring grew faster than females, and offspring from mothers raised on the HQ diet grew faster than offspring from mothers on the LQ diet (male HQ = mm day 1, female HQ = mm day 1, male LQ = mm day 1, female LQ = mm day 1 ). Variance associated with mother s (a) Females Males Female size at 14 days (mm) Male size at 14 days (mm) Maternal diet High quality Low quality Mother's size (mm) Mother's size (mm) (b) Female size at 14 days (mm) Male size at 14 days (mm) Maternal diet High quality Low quality Brood size Brood size Fig. 2 Offspring s size at 14 days, as influenced of mother s size (a) and brood size (b) (males on the right panels, females on the left panels). All measurements of size are standard length (SL, mm). Lines represent estimated relationship between the variables for each combination of offspring s sex and maternal diet (solid = high-quality diet; dashed = low-quality diet), and shaded areas represent confidence intervals obtained with MCMC simulations.

6 6 A. D. MURPHY ET AL. Table 2 Growth rate prior to social context treatment (GR1) was influenced by brood size, maternal diet and offspring s sex. Estimates and standard errors obtained from mixed-effects model (random effect = mother identity), with respective P-values obtained from MCMC simulations (n = 174 individuals, 25 mothers). Estimate SE P Intercept < Size 14 (scaled) < Diet (LQ) Sex (M) < identity represented 24.28% of the variance in the model for GR1. The growth rate subsequent to exposure treatments (GR2) was influenced by an interaction of maternal diet, social experience and sex of the juvenile (Table 3; Fig. 4; marginal R 2 = 0.27, conditional R 2 = 0.37). In both the control and exposure treatments, sons of LQ mothers grew faster than females. Sons of both HQ and LQ mothers grew faster in the exposure treatment as compared to the control; however, for females, only the daughters of LQ mothers grew faster in the exposure treatment as compared to the controls, whereas daughters of HQ mothers grew slower in the exposure treatment than in the control. Variance associated with mother s identity represented 10.30% of the variance in the model for GR2. Social treatment (exposure vs. control) did not significantly influence size at sexual maturity for males or females (Table 4; marginal R 2 = 0.44, conditional R 2 = 0.66). However, both sex and diet influenced size at sexual maturity (Table 4). Males were larger at sexual maturity than females, and within males and females, fry with mothers on the HQ diet were larger at sexual maturity (control social treatment: HQ males = mm, HQ females = mm, LQ males = mm, LQ females = mm; exposure social treatment: HQ males = mm, HQ females = mm, LQ males = mm, LQ females = mm). Mother s identity was estimated as explaining 22.34% of the variance in the model for size at sexual maturity. Discussion We detected variation in maternal investment in the swordtail Xiphophorus multilineatus depending on female size and diet, effects that were long-lasting, sex specific and dependent on social context. The influence of mother s diet on their offspring was significant and independent of the influence of diet on mother s size: larger mothers produced smaller fry regardless of diet, whereas mothers on HQ diets produced larger fry. We suggest that our results support the maternal stress hypothesis (McCormick, 1998) and not the environment phenotype matching hypotheses as females raised on a low-quality diet apparently reduced maternal investment in individual fry, which by our measures was not beneficial for the offspring raised in the lowquality environment (smaller as fry, grew slower, were smaller at sexual maturity). In addition, we detected an interaction between mother s diet and social context influencing growth rates, suggesting female offspring adopted different reproductive strategies depending on the maternal investment they received. It appears that the high-quality diet allowed mothers to invest more in offspring, and they did so by investing Females Males Maternal diet High quality Low quality GR1 (mm day 1 ) GR1 (mm day 1 ) Female size at 14 days (mm) Male size at 14 days (mm) Fig. 3 Growth rate prior to the initiation of the social context experiment (GR1), as influenced by offspring s size at 14 days. Lines represent estimated relationship between the variables for each combination of offspring s sex and maternal diet (solid = high-quality diet; dashed = low-quality diet), and shaded areas represent confidence intervals obtained with MCMC simulations.

7 Maternal effects and reproductive strategies 7 Table 3 Growth rate after social context treatment (GR2) was influenced by maternal diet, offspring s sex, social treatment and the two- and three-way interactions between these variables. Estimates and standard errors obtained from mixed-effects model (random effect = mother identity), with respective P-values obtained from MCMC simulations (n = 175 individuals, 25 mothers). GR2 (mm day 1 ) Offspring sex, maternal diet Female, high quality Female, low quality Male, high quality Male, low quality Control Estimate SE P Intercept < Size 70 (scaled) Diet (LQ) Sex (M) Social (exposure) Diet (LQ): sex (M) Diet (LQ): social (exposure) Sex (M): social (exposure) Diet (LQ): sex (M): social (exposure) Social treatment Exposure Fig. 4 Growth rate after social treatment (GR2, mm days 1 ), as influenced by maternal diet by sex. Bars indicate standard errors of the estimates. Control treatment n = 87 (19 females with highquality (HQ) mothers, 26 females with low-quality (LQ) mothers, 17 males with HQ mothers and 25 males with HQ mothers). Exposure treatment n = 88 (17 females with HQ mothers, 28 females with LQ mothers, 17 males with HQ mothers and 26 males with HQ mothers). more in individual fry (larger fry) rather than more fry (Fig. 2a). The response of maternal investment to diet we detected contrasts to what has been detected in guppies, where females from both high- and low-predation populations produced larger fry in response to food limitation (Bashey, 2006; although see Sman et al., 2009), but was similar to two other lecithotrophic species of fish that responded to low food availability Table 4 Size at sexual maturity was influenced by maternal diet and offspring s sex. Estimates and standard errors obtained from mixed-effects model (random effect = mother identity), with respective P-values obtained from MCMC simulations (n = 169 individuals, 25 mothers). Estimate SE P Intercept < Diet (LQ) Sex (M) < Social (exposure) by producing smaller eggs (Reznick et al., 1996). Fishes of the genus Xiphophorus are also lecithotrophic (Constantz, 1989), which means that females provision the eggs in the form of yolk prior to fertilization. Selection on mothers can lead to maternal effects that reduce offspring s fitness when (i) maternal effects are costly to the mother, (ii) there is an opportunity for reproducing repeatedly, and/or (iii) mothers have a chance of reproducing under better conditions than the current, such that returns on investment from future reproduction would be higher (Marshall & Uller, 2007). Our results suggest that maternal investment in this species is costly, as females from the HQ diet invested more in their offspring, and there was a negative relationship between offspring number (brood size) and fry size suggesting a trade-off. In addition, swordtails are capable of reproducing repeatedly (every 30 days throughout their adult lifespans, estimated to be 2 years) and can live in environments that fluctuate in quality throughout the year (Morris & Ryan, 1992). If maternal investment decisions had predicted and produced the optimal offspring s size based on their own environment (Marshall et al., 2010), offspring from mothers reared on LQ diets would have been expected to be more successful than fry from HQ mothers, as all offspring were reared on LQ diets. Instead, our measures of success (increased growth rate and larger size at sexual maturity) suggest that the fry from HQ mothers were more successful. Therefore, we suggest that the optimal strategy for mothers in X. multilineatus may conflict with what is optimal for their offspring. Swordtail females on the low-quality diet may have been selected to invest in somatic growth over reproduction as a means to increase their lifetime reproduction, and this investment came at a cost to the current offspring s fitness, as measured by offspring s growth rate and size at sexual maturity. The optimal size for an offspring will depend not only on their nutritional environment but the density of competitors that are present in their environment. An alternative to the maternal stress hypothesis that takes into account the fact that mothers can influence the competitive environment of their offspring by the number of offspring they produce is the sibling competition hypothesis (Parker & Begon, 1986; Beckerman et al.,

8 8 A. D. MURPHY ET AL. 2006). This hypothesis suggests that high-quality mothers (often larger and producing more offspring) produce larger offspring than smaller females, as there is stronger selection on their offspring due to competition among siblings. Larger female swordtails produce larger broods, and swordtails are known to breed in very small pools in some cases, which could lead to selection due to sibling competition. However, our results do not support this hypothesis, as we detected a negative relationship between female size and offspring s size, the opposite pattern the sibling competition hypothesis would predict. Even though females reared on the HQ diet invested more in each offspring, and mothers reared on HQ diet were larger, larger females regardless of diet produced smaller fry (see Fig. 2). We did not examine, however, the response of the offspring to density, as all offspring were reared in separate tanks. It would be interesting to determine how fry size influences growth rate and size at sexual maturity in environments that differ in density of competitors during development. Both male and female offspring from mothers on LQ diets responded to the social context of being exposed to an adult male during development by increasing their growth rates. However, only the males with mothers on HQ diets responded by increasing growth rate in response to the exposure to an adult male, as their sisters decreased growth rate when exposed to courter males. We suggest that the daughters of HQ mothers were not constrained in their ability to grow in this context, as both their brothers and the daughters of LQ moms grew faster if exposed to an adult male. Instead, we suggest that the daughters of HQ moms were adopting a different reproductive strategy in this social context than the daughters of LQ moms. Females that received more maternal investment may have invested in producing more eggs rather than growing faster when exposed to courter males, whereas the daughters of LQ mothers invested in growing faster rather than investing in egg production when exposed to courter males. The difference in growth rates between daughters of LQ and HQ mothers suggests that maternal investment may have influenced the reproductive strategies of their daughters. Further study is needed to determine whether the differences in responses between these two groups of females are due to a constraint on the daughters of LQ mothers ability to respond to the presence of a potential mate and/or that this is an adaptive counter-strategy by the daughters in response to their mother s reduced investment. Whereas social context influenced growth rates, we did not find that it influenced size at sexual maturity. Changes in either growth rate or size at sexual maturity could have produced the changes in age at sexual maturity detected by Walling et al. (2007) in X. helleri. All of the offspring in this study were sired by males from the largest size class in X. multilineatus, and therefore, all sons would be expected to be similar in size, as allelic variation and copy number of the Mc4R gene on the Y chromosome influence the size differences between males (Zimmerer & Kallman, 1989; Lampert et al., 2010). One could argue that social context influenced how fast males grew to their adult size, but that the size at which males reached sexual maturity was not plastic due to the genetic influences on this trait. However, we did detect an influence of mother s diet on male size at sexual maturity for both males and females, and therefore, it is left to be explained why social context only influenced growth rate and not size at maturity. In addition, the pattern we detected in males is the opposite predicted from the study on X. helleri (Walling et al., 2007), where males reached sexual maturity later (growth slower if size at sexual maturity does not change). It will be important to determine the costs and benefits to the offspring of increased growth when developing in the presence of adult males for both sexes before this pattern can be fully understood. Maternal investment has been investigated previously in X. multilineatus (Rios-Cardenas et al., 2013), and differences in the age at which daughters reached sexual maturity depending on whether or not their mothers had mated with a male from the smallest size class (sneaker males) or one of the larger size classes (courter males) were detected. Daughters of courters reached sexual maturity sooner than daughters of sneakers, suggesting increased growth rates due to greater maternal investment when females mate with courter males. In addition, the differential allocation by females into courter broods was influenced by how long the wild-caught females had been in the laboratory, suggesting that females can adjust their reproductive investment depending on either diet and/or social context. Females housed individually in the laboratory experience a higher-quality diet and reduced future mating opportunities, due to isolation from males. As swordtail females store sperm, they can continue to produce broods in isolation. In this context (isolated in the laboratory), Rios-Cardenas et al. (2013) determined that females increased their reproductive investment by increasing investment to individual fry. Therefore, the results of this previous study support our findings that X. multilineatus females respond to higher-quality diets by increasing their investment in individual offspring and that diet and social context are important factors influencing a female s reproductive strategy. Because Xiphophorus fishes are lecithotrophic (Constantz, 1989), any facultative maternal investment allocations will be made prior to fertilization. The substances that mothers can transfer to their offspring through eggs can include hormones as well as nutrients (Love et al., 2005). Variation in yolk content in response to environmental conditions has been demonstrated extensively in birds (Love et al., 2005), but there is also some evidence that fish can adjust yolk content (e.g. Segers et al., 2011).

9 Maternal effects and reproductive strategies 9 We suggest that the possibility that female X. multilineatus vary their investment of hormones based on diet should be explored. Maternal diet influenced both growth rate and size at sexual maturity even when we controlled for offspring s size at 14 days. Moreover, because offspring with smaller size at 14 days had higher GR1, and male fry were larger than female fry, our results are consistent with a study of cichlids that demonstrated compensatory growth as a result of egg size induced differences in the gene expression of growth hormones (Segers et al., 2011). We have demonstrated that mothers reared on lowquality diets reduced maternal investment as compared to mothers reared on high-quality diets and that this was not beneficial to their offspring, which had slower growth rates and reached sexual maturity at smaller sizes. It has been suggested that the interaction between the mother s and offspring s environment can be important, such that it can be beneficial to offspring that mothers invest less if offspring will be developing in the same low-quality environment (Segers & Taborsky, 2010; Jørgensen et al., 2011) or more if the offspring will be experiencing high levels of sibling competition (Parker & Begon, 1986; Beckerman et al., 2006). However, in X. multilineatus, lower investment from mothers raised on poor-quality diets did not benefit offspring during the component of selection we measured (survival to reproduction) and in the environment we measured (low-quality diets) and may represent parent offspring conflict (Marshall & Uller, 2007). In addition, larger mothers that produced more offspring (increasing competition in the offspring s environment) produced smaller fry. As for the social environment, we detected increased growth rates by males in response to being exposed to an adult male regardless of mother s diet, but an influence of mother s diet on how daughters responded. One possible explanation for this difference is that daughters of mothers raised on low-quality diets were still below some optimal size for switching from growth to investing in reproduction, whereas the daughters of high-quality mothers that had a better start in life invested in reproduction rather than growth if exposed to an adult male, as they were already on track to reach the optimal size for sexual maturity. Regardless of the explanation, maternal diet influenced the response of daughters to a change in environment that was not nutritional, suggesting costs and benefits of maternal investment beyond predicting offspring s environment from mother s own environment. Acknowledgments We thank the following individuals for their assistance with fish care and data collection: Joseph Lyon, Brittany Tarselli, Liz Matthias, Liz Dumler, Nicole Kleinas, Will Ternes and Kara Roberts. We also thank Susan Lyons for her assistance in the set up of the experiment and use of the fish she raised in a previous study. All experiments comply with current laws and with the Animal Care Guidelines of Ohio University (Animal Care and Use approval number: L01-01). References Agrawal, A Transgenerational consequences of plant responses to herbivory: an adaptive maternal effect? Am. Nat. 157: Agrawal, A Herbivory and maternal effects: mechanisms and consequences of transgenerational induced plant resistance. Ecology 83: Agrawal, A., Laforsch, C. & Tollrian, R Transgenerational induction of defences in animals and plants. Nature 401: Ali, M., Nicieza, A. & Wootton, R.J Compensatory growth in fishes: a response to growth depression. Fish Fish. 4: Baayen, R.H languager: Data sets and functions with Analyzing Linguistic Data: A practical introduction to statistics. Available from: languager Bashey, F Cross-generational environmental effects and the evolution of offspring size in the Trinidadian guppy Poecilia reticulata. Evolution 60: Bates, D lme4: Mixed-effects modeling with R. See: Beckerman, A.P., Benton, T.G., Lapsley, C.T. & Koesters, N How effective are maternal effects at having effects? Proc. R. Soc. B Biol. Sci. 273: Bernardo, J Maternal effects in animal ecology. Am. Zool. 105: Constantz, G.D Reproductive biology of poeciliid fishes. In: Ecology and Evolution of Livebearing Fishes (Poeciliidae) (G.K. Meffe & F.F. Snelson, eds), pp Prentice Hall, Englewood Cliffs, New Jersey. Endler, J.A Natural and sexual selection on color patterns in poeciliid fishes. Environ. Biol. Fish 9: Fox, C., Thakar, M. & Mousseau, T Egg size plasticity in a seed beetle: an adaptive maternal effect. Am. Nat. 149: Galloway, L Parental environmental effects on life history in the herbaceous plant Campanula americana. Ecology 82: Gregersen, F., Haugen, T.O. & Larsen, O.N Egg size differentiation among sympatric demes of brown trout: possible effects of density-dependent interactions among fry. Ecol. Freshw. Fish 15: Hector, K.L. & Nakagawa, S Quantitative analysis of compensatory and catch-up growth in diverse taxa. J. Anim. Ecol. 81: Johnston, T.A. & Leggett, W.C Maternal and environmental gradients in the egg size of an iteroparous fish. Ecology 83: Jørgensen, C., Auer, S.K. & Reznick, D.N A model for optimal offspring size in fish, including live-bearing and parental effects. Am. Nat. 177: E119 E135. Kudo, S. & Nakahira, T Trophic-egg production in a subsocial bug: adaptive plasticity in response to resource conditions. Oikos 111: Lampert, K.P., Schmidt, C., Fischer, P., Volff, J.N., Hoffmann, C., Muck, J. et al Determination of onset of sexual

10 10 A. D. MURPHY ET AL. maturation and mating behavior by melanocortin receptor 4 polymorphisms. Curr. Biol. 20: Leips, J., Richardson, J.M.L., Rodd, F.H. & Travis, J Adaptive maternal adjustments of offspring size in response to conspecific density in two populations of the least killifish, Heterandria formosa. Evolution 63: Ling, S., Hashim, R., Kolkovski, S. & Chong Shu-Chien, A Effect of varying dietary lipid and protein levels on growth and reproductive performance of female swordtails Xiphophorus helleri (Poeciliidae). Aquac. Res. 37: Love, O.P., Chin, E.H., Wynne-Edwards, K.E. & Williams, T.D Stress hormones: a link between maternal condition and sex-biased reproductive investment. Am. Nat. 166: Lyons, S The Influence of Condition, Context, and Life History on Variation in Female Mate Preference in Xiphophorus Fishes. Master s Thesis, Ohio University, Athens OH, (August 2011). Lyons, S.M., Goedert, D. & Morris, M.R Male-trait-specific variation in female mate preferences. Anim. Behav. 87: Marshall, D.J Transgenerational plasticity in the sea: context-dependent maternal effects across the life-history. Ecology 89: Marshall, D.J. & Uller, T When is a maternal effect adaptive? Oikos 116: Marshall, D.J., Heppell, S.S., Munch, S.B. & Warner, R.R The relationship between maternal phenotype and offspring quality: do older mothers really produce the best offspring? Ecology 91: McCormick, M.I Behaviorally induced maternal stress in a fish influences progeny quality by a hormonal mechanism. Ecology 79: McCormick, M.I Mothers matter: crowding leads to stressed mothers and smaller offspring in marine fish. Ecology 87: Morris, M.R Female preference for trait symmetry in addition to trait size in swordtail fish. Proc. R. Soc. B Biol. Sci. 265: Morris, M.R. & Ryan, M.J Breeding cycles in natural populations of Xiphophorus nigrensis, X. multilineatus, and X. pygmaeus. Copeia 1992: Morris, M.R., Batra, P., Ryan, M.J., Url, S. & Carolina, N Male-male competition and access to females in the swordtail Xiphophorus nigrensis. Copeia 1992: Morris, M.R., Rios-Cardenas, O. & Scarlett Tudor, M Larger swordtail females prefer asymmetrical males. Biol. Lett. 2: Moyle, P.E. & Cech, J.J. Jr Fishes: An Introduction to Ichthyology, 2nd edn. Prentice Hall, Englewood Cliffs, New Jersey. 559 pp. Nakagawa, S. & Schielzeth, H A general and simple method for obtaining R 2 from generalized linear mixed-effects models. Methods Ecol. Evol. 4: Nichol, D.G. & Pikitch, E.K Reproduction of darkblotched rockfish off the Oregon coast. Trans. Am. Fish. Soc. 123: Parker, G.A. & Begon, M Optimal egg size and clutch size: effects of environment and maternal phenotype. Am. Nat. 128: Parker, G.A., Royle, N.J. & Hartley, I.R Begging scrambles with unequal chicks: interactions between need and competitive ability. Ecol. Lett. 5: Pinheiro, J.C. & Bates, D.M Mixed-effects models in S and S-Plus. In: Statistics and Computing. Springer-Verlag, Berlin, Germany. R Development Core Team R: A language and environment for statistical computing. See: Reznick, D.N. & Yang, A.P The influence of fluctuating resources on life history: patterns of allocation and plasticity in female guppies. Ecology 74: Reznick, D.N., Callahan, H. & Llauredo, R Maternal effects on offspring quality in poeciliid fishes. Am. Zool. 36: Rios-Cardenas, O., Brewer, J. & Morris, M.R Maternal Investment in the Swordtail Fish Xiphophorus multilineatus: support for the differential allocation hypothesis. PLoS ONE 8: e Rosen, D.E Middle-American poeciliid fishes of the genus Xiphophorus. Bull. Florida State Mus. Biol. Sci. 5: Rosenthal, G.G., Flores Martinez, T.Y., Garcıa de Leon, F.J. & Ryan, M.J Shared preferences by predators and females for male ornaments in swordtails. Am. Nat. 158: Rossiter, M.C Incidence and consequences of inherited environmental effects. Annu. Rev. Ecol. Syst. 27: Royle, N.J., Metcalfe, N.B. & Lindstrom, J Sexual selection, growth compensation, and fast-start swimming performance in green swordtails, Xiphophorus helleri. Funct. Ecol. 20: Schrader, M. & Travis, J Assessing the roles of population density and predation risk in the evolution of offspring size in populations of a placental fish. Ecol. Evol. 2: Segers, F.H.I.D. & Taborsky, B Egg size and food abundance interactively affect juvenile growth and behaviour. Funct. Ecol. 25: Segers, F.H.I.D., Berishvili, G. & Taborsky, B Egg size-dependent expression of growth hormone receptor accompanies compensatory growth in fish. Proc. R. Soc. B Biol. Sci. 279: Sman, J., Phillips, N.E. & Pfister, C.A Relative effects of maternal and juvenile food availability for a marine snail. Ecology 90: Storm, J.J. & Lima, S.L Mothers forewarn offspring about predators: a transgenerational maternal effect on behavior. Am. Nat. 175: Uller, T Developmental plasticity and the evolution of parental effects. Trends Ecol. Evol. 23: Walling, C.A., Royle, N.J., Metcalfe, N.B. & Lindstr om, J Green swordtails alter their age at maturation in response to the population level of male ornamentation. Biol. Lett. 3: Zimmerer, E.J. & Kallman, K.D Genetic basis for alternative reproductive tactics in the pygmy swordtail, Xiphophorus nigrensis. Evolution 43: Zuur, A.F., Ieno, E.N., Walker, N.J., Saveliev, A.A. & Smith, G.M Mixed Effects Models and Extensions in Ecology with R. Springer, New York. Data deposited at Dryad: doi: /dryad.b8d08 Received 3 March 2014; revised 14 April 2014; accepted 15 April 2014