A potential microevolutionary life-history trade-off in White-Footed Mice (Peromyscus leucopus)


†Author to whom correspondence should be addressed. E-mail: pdheid@wm.edu


  • 1Reproductive costs can affect survival and future reproduction. In winter and in short photoperiods, some individuals within populations of temperate-zone rodents inhibit reproduction and decrease food intake, while others do not.
  • 2Two lines derived from a natural population of White-Footed Mice and selected to maintain or inhibit reproduction in short photoperiod were tested for two potentially correlated responses to selection: changes in food intake and body mass. Mice were raised to age 70 days in short or long photoperiod, following which testis size, body mass and food intake were measured over a 2-week period.
  • 3In both lines and both photoperiods, there was an apparent response to selection in testis size and a correlated response to selection in food intake but not body mass. In both photoperiods, mice in the line selected for winter reproduction ate 50% more than mice in the line selected for winter reproductive inhibition. Mice in an unselected control line were intermediate for testis size and food intake.
  • 4The increased food intake in a line of mice with a genetic tendency for winter reproduction suggests a potential cost of winter reproduction and a potential microevolutionary trade-off related to this cost of reproduction.


High costs of winter reproduction cause reproductive inhibition in many mammals. For short-lived small mammals, winter survival can be so low as to favour winter reproduction in spite of low success (Bronson 1985) (reviewed by Bronson & Heideman 1994). Winter reproduction increases costs of reproduction for males and females, because both sexes must obtain sufficient resources for the direct costs of gamete and embryo production as well as for mate searching, which may increase both energetic costs (Prendergast, Kriegsfeld & Nelson 2001) and predation risk (Horton & Rowsemitt 1992). Winter breeding small mammals are typically ‘income’ breeders, relying on food intake to meet reproductive needs, and thus cannot simply allocate stored fat and nutrients to meet needs of winter reproduction.

The high cost but potential fitness benefit from winter reproduction (Bronson 1989) creates the potential for life-history trade-offs. Physiological life-history trade-offs have been defined as changes in allocation of resources when resource intake is unchanged (Stearns 1992), but physiological trade-offs may also occur even when resource intake has changed if there is reallocation among competing needs. Such allocation differences are considered microevolutionary trade-offs when there is (1) genetic variation within a population for two or more traits, and (2) when a change in one trait increases fitness, but results in a change in the other trait that decreases fitness (Stearns 1992). If such genetic variation exists within a population, then selection on one trait should result in correlated response to selection in the linked trait(s). Because of this relationship, evidence for microevolutionary trade-offs can be gained by selection experiments (Stearns 1992). Thus, selection on winter reproduction (which may increase fitness) that results in a change in another trait that may decrease fitness (such as a need for more food) would be evidence for a microevolutionary life-history trade-off.

Levels of phenotypic variation in winter reproductive condition in wild mice can be duplicated in the laboratory using short photoperiods (Bronson 1989). Genetic variation for winter reproduction (Desjardins, Bronson & Blank 1986; Spears & Clarke 1988; Lynch, Lynch & Kliman 1989; Heideman & Bronson 1991; Heideman et al. 1999) might be accompanied by microevolutionary trade-offs between winter reproduction and costs of reproduction. A number of previous papers have demonstrated positive correlations between the winter breeding phenotype and food intake, as well as correlations between winter reproductive phenotype and body mass (reviewed by Prendergast et al. 2001) or immune function of rodents (Demas & Nelson 1996; Prendergast & Nelson 2001), suggesting the existence of physiological trade-offs related to winter reproduction.

In this paper, we test for evidence for a microevolutionary trade-off between voluntary food intake and winter reproduction in a wild population of White-Footed Mice (Peromyscus leucopus Rafinesque). In the source population, winter reproduction occurs in some but not all individuals (Terman 1993; Heideman et al. 1999). In such a population, an increased innate tendency to increase food intake might improve winter reproductive success. If so, we would predict a genetic correlation between voluntary food intake and reproductive status in short photoperiod, and therefore a response in voluntary food intake to selection on winter reproductive phenotype. An alternative hypothesis is that winter reproduction favours ‘capital breeders’, which use stored fat and nutrients to support reproduction. If so, then a response to selection on winter reproductive phenotype might be a change in body mass. To test for these trade-offs, we compared food intake and body mass of males in artificially selected lines that had been under directional selection for or against gonadal development, respectively, in short winter photoperiods, as well as in an unselected control line. Tests were conducted in both permissive long summer photoperiods and short winter photoperiods.


animals and housing

Artificial selection on a population of P. leucopus from the Williamsburg area produced the two selected lines and one control line of mice used in the study. These lines were formed from a single, wild population trapped in 1995 in Williamsburg, VA (latitude 37·3°N, longitude 76·7°W), followed by multiple generations of mass selection for either reproductive inhibition in short-day photoperiods or a lack of reproductive inhibition in short photoperiods (Heideman et al. 1999). The criterion for selection in one selected line was gonadal development to within the range typical of breeding mice in summer photoperiods, and the criterion for selection in the other selected line was gonadal development that would be considered infertile for mice in summer photoperiods. Matings were not made between siblings. Each line in each generation included 20–50 breeding pairs. Forty-eight wild-caught mice bred successfully in the laboratory to establish the parental generation in the laboratory of 104 pairs of mice; each line in each generation included 20–50 successful breeding pairs. Most mice from the photoperiod responsive (R) line had suppressed reproductive systems in short photoperiod, while most mice from the photoperiod non-responsive (NR) line had well-developed reproductive systems in both short and long photoperiod (Heideman et al. 1999; Majoy & Heideman 2000). A control line (C) was not subject to selection and produced the full range of reproductive phenotypes (Heideman et al. 1999). The NR and R phenotypes in the selected lines in short photoperiod are a subset of those in the unselected control line. The availability of the selected lines allowed comparison of the different genotypes for winter reproduction even when there were no detectable phenotypic differences, in, for example, long photoperiod.

experiment 1: comparison of two selected lines in short photoperiod

In Experiment 1a, male mice aged 70–112 days from generations 3 and 4 of the R and NR lines were used to compare food intake (two runs; R: N = 21, NR: N = 28) and faecal mass (one run of the experiment above; R: N = 11, NR: N = 18). In Experiment 1b, male mice aged 70–112 days from generations 3 and 4 of the R and NR lines were used to compare spontaneous activity (one run; R: N = 8; NR: N = 8) in short photoperiod (SD; 8:16 light/dark; lights on at 08·00 EST). Experimental mice were born in a long photoperiod (LD; 16:8 h light :dark; lights on at 04·00 EST) and transferred within 3 days of birth to SD. Thus, in these preliminary experiments, mice were raised and tested only in SD. Only a single male was used from each litter. Mice were weaned at 21–23 days and singly housed in a 27 × 13 × 16 cm3 cage with wire top. Mice were provided with steam-treated pine shavings approximately 3 cm deep and ad libitum tap water and food (Harlan Teklad mouse and rat chow). Rooms were illuminated with fluorescent overhead lighting of approximately 100–1000 lux at cage heights. Room temperatures were 22 °C ± 4 °C during the experiments.

To measure food intake, mice were given a weighed quantity of food in a cup in a clean cage, and food intake was measured for 2 days. Faecal pellets produced during these 2 days by a subsample of these mice were collected, dried to constant weight, and weighed. For activity measures, mice were placed in a cage 36 × 24 × 19 cm3 with a plexiglass top and monitored by an overhead video camera connected to a Videomex V system (Columbus Instruments, Columbus, OH) to measure distance travelled. The Videomex V software tracked the dark mouse against the light background of the cage, recording distance moved. Activity was recorded for mice over a 22-h period beginning in the light phase, through the dark phase (the active period for these nocturnal mice; Majoy & Heideman 2000), and ending in the following light phase.

experiment 2: comparisons of two selected lines and one control line in two photoperiods

In a second experiment, male mice from generations 7, 8 and 9 of the R, NR and C lines were used to test for differences in testis size, body weight and food intake. Experimental mice were born in LD, and either kept in LD or transferred within 3 days of birth to SD. More than one male was used from some litters; for those litters, mean values for the entire litter were used in the analysis. Thus, sample size and degrees of freedom were based on the number of litters, not on the number of individuals. The number of litters in treatment groups was: LD R: N = 8; LD C: N = 11; LD NR: N = 9; SD R: N = 17; SD C: M = 17; SD NR: N = 11.

At 70 ± 3 days of age, mice were lightly anaesthetized using isoflurane, the scrotum was dampened with 70% ethanol, the length and width of the right testis were measured using dial callipers, and body mass was measured. Testis length was multiplied by width to obtain a testis index, which correlates highly with testis weight (Heideman et al. 1999) and level of spermatogenesis (P. D. Heideman and P. Zelensky, unpublished data; Desjardins & Lopez 1983). Food in the hopper was weighed at the beginning, middle, and end of a 2-week period for calculation of average daily food intake; body mass was measured again at the end of food intake measurements.

experiment 3: comparisons of breeding adult mice of two selected lines in ld

This experiment tested for differences in testis size of mature fertile adults between the two selected strains (R; N = 13, NR = 13). Body weight and testis index was measured as above from males of generations 7, 8 and 9. Males were at least 6 months in age and had sired at least one litter with a breeder female.

data analysis

Data were analysed using Statview 4·5 or Super anova (Abacus Concepts, Inc.) on a Macintosh. Data points were values from one individual or the mean of all individuals taken from a single litter. Reproductive organ mass of rodents often changes independently of body mass, and direct adjustments for body mass can be misleading (discussed in Heideman et al. 1999); therefore, we used analysis of covariance (ancova) with body mass as the covariate to test for a potential effect of body mass on food intake, reproductive measures and faecal mass. Because body mass might be affected by food intake, ancova with food intake as the covariate was used to compare body mass among groups in Experiment 2. In cases where relationships were not significant, results are reported only from anova.

In the text and in figures, all means are presented with their standard errors.


In Experiment 1a, on NR and R mice in SD, food intake was higher in NR than in R mice (NR: 0·23 ± 0·004 g g−1 body mass; R: 0·18 ± 0·006 g g−1 body mass; P < 0·05). NR also produced a greater mass of faeces (NR: 0·91 ± 0·02 g; R: 0·77 ± 0·03 g; P < 0·05). However, in Experiment 1b, the two strains did not differ in total distance moved in SD (NR: 0·75 ± 0·05 km; R: 0·74 ± 0·06 km; P > 0·10).

In Experiment 2, testis index was significantly related to line (F = 14·10; P < 0·0001) and photoperiod (F = 128·64; P < 0·0001), but there was no interaction between line and photoperiod (F = 0·02; P = 0·98). Testis index was larger in LD than in SD, and was ordered NR > C > R in both photoperiods (Fig. 1). The three lines did not differ significantly in body mass (F = 0·87; P = 0·42; Fig. 2), but mice of all three lines weighed significantly less in SD than in LD (F = 7·3; P < 0·01; Fig. 2), with no interaction between line and photoperiod (F = 0·22; P = 0·80).

Figure 1.

Testis index of mice from the non-responsive (NR), control and reproductively inhibited (R) lines from generations 7, 8 and 9 raised in LD or SD. Data are presented as mean ± SEM.

Figure 2.

Body mass of mice from the non-responsive (NR), control and reproductively inhibited (R) lines from generations 7, 8 and 9 raised in LD or SD. Data are presented as mean ± SEM.

In Experiment 2, the three lines also differed significantly in food intake (F = 23·40; P < 0·0001), with food intake ordered NR > C > R (Fig. 3a). All three lines ate slightly less in SD than in LD, but this small effect of photoperiod on food intake was not statistically significant (F = 2·45; P = 0·12). There was no interaction between line and photoperiod (F = 0·07; P = 0·93) on food intake. Surprisingly, food intake was not significantly correlated with body mass in LD (F = 0·34; P = 0·57) or SD (F = 3·69; P = 0·06; R2 = 0·04).

Figure 3.

(a) Food intake and (b) food intake/gram body mass of mice from the non-responsive (NR), control and reproductively inhibited (R) lines from generations 7, 8 and 9 raised in LD or SD. Data are presented as mean ± SEM.

In Experiment 2, when food intake was assessed relative to body mass by ancova, with body mass as covariate, there was a significant effect of line on food intake (F = 13·53; P < 0·0001; Fig. 3b). Relative to R mice, C mice ate approximately 30% more and NR mice ate approximately 60% more food per gram body mass in both SD and LD (Fig. 3b). Within each line, food intake relative to body weight was nearly identical in LD and SD (Fig. 3b), with no significant effect of photoperiod (F = 0·25; P = 0·62). There was no interaction between line and photoperiod (F = 0·04; P = 0·97).

Food intake was not correlated with testis size in LD (P > 0·10 in all three lines; data not shown). However, food intake was correlated with testis index in SD (Fig. 4) in all three lines (C: P < 0·01, R2 = 0·29, Fig. 4a; R: P < 0·005, R2 = 0·43, Fig. 4b; NR: P < 0·0001, R2 = 0·60, Fig. 4c).

Figure 4.

Testis index in relation to food intake in SD for: (a) mice in the control line; (b) mice in the R line; and (C) mice in the NR line from generations 7, 8 and 9. The attained level of significance and R2 are shown for each panel. Note that the scales for the x and y axes are different for each panel.

In Experiment 3, breeder males of the NR line had significantly larger testes than breeder males of the R line (testis index of R: 62·1 ± 2·0; testis index of NR: 81·4 ± 3·7; t = 4·53, P < 0·0001).


Either genetic drift or selection on gonadal development in short winter photoperiod resulted in changes in testis size (Fig. 1), with no response in body mass (Fig. 2) or distance moved (Experiment 1), but an apparently correlated response in food intake (Experiment 1 and Fig. 3) and faecal mass (Experiment 1). We cannot rule out the possible effects of drift on these unreplicated selection lines and control line. However, the results suggest a correlated response to selection that implies a physiological link between winter reproduction and food intake in this population. These results are consistent with the hypothesis that high food intake and winter reproduction are elements of a microevolutionary trade-off, in which winter reproduction is enhanced by increased food intake. Interestingly, selection on gonadal development in SD altered testis size even in LD (Fig. 1; Heideman et al. 1999). This suggests that high food intake might involve a life-history trade-off with reproductive potential in all seasons, not just in winter.

In winter, populations of this omnivorous species sometimes encounter abundant food that is easy and safe to obtain (e.g. acorn mast, Gashwiler 1979), but sometimes individuals must forage much further and at more dangerous sites to obtain food. Thus, higher food intake might allow winter reproduction, but could entail a cost due to increased foraging and higher predation risk. In mast years with high acorn production, it might be possible for an individual under a mast tree to find sufficient food within minutes each day, and with low risk of predation, while at different sites or in other years individuals might require long periods of foraging in more open areas for small and scarce seeds. Such spatial and temporal heterogeneity might result in variable selection on traits such as winter breeding and food intake (Bell 1997; Mitton 1997).

Previous laboratory studies on other species of reproductively photoresponsive temperate-zone rodents have reported decreases in food intake in SD (reviewed by Prendergast, Nelson & Zucker 2002). Studies comparing food intake in photoresponsive and non-responsive phenotypes have found higher food intake in non-responsive individuals (Blank et al. 1994; Ruf et al. 1997) or no differences between phenotypes (Moffatt, DeVries & Nelson 1993). Non-responsive phenotypes of P. maniculatus have higher food intake than responsive phenotypes in SD (Blank et al. 1994; Ruf et al. 1997) consistent with our results. Genetic variation exists in that population, but those studies only assessed phenotypic variation. In our study, we found a correlation in all three lines between food intake and testis size in SD (Fig. 4), consistent with these other studies in suggesting that decreased testis size in winter permits lower food intake.

Nelson (1987) argued that genetic variation in photoresponsiveness might occur as a balanced polymorphism involving a life-history trade-off in which winter reproduction increased mortality. Our results are consistent with Nelson's hypothesis, and suggest a mechanism for such a trade-off: that a genetic tendency for winter reproduction may be associated with high food intake, causing increased foraging risk, and thus increased mortality.

A microevolutionary trade-off of this type could be caused by a single pleiotropic gene exhibiting antagonistic pleiotropy (Mitton 1997), such that one allele simultaneously increased food intake and stimulated reproductive development, while another allele at the same locus reduced food intake and inhibited reproduction. Alternatively, linkage or epistatic interactions among genes may also cause correlated responses to selection (Bell 1997).

A caveat with respect to our conclusions is that mice in both LD and SD were tested in a protected laboratory environment: ad libitum food and protection from cold temperature. Photoperiod alone is likely to be only one aspect of the environmental inputs that regulate winter reproduction in nature. Phenotypic responses we observed in the laboratory may be relevant only to the outcome of selection in that environment (Bell 1997), while not accurately representing the full range of responses in nature. A very important caveat is that founder effects or genetic drift in an unreplicated selection line could produce results like ours.

The phenomenon of variable winter reproduction is widespread in temperate-zone rodents. If genetic variation in food intake is correlated with genetic variation in winter reproduction, this may represent a microevolutionary life-history trade-off that may be relevant to many populations of rodents.


We thank K. King and L. L. Moore for assistance with mouse care and data collection, and J. P. Swaddle and C. D. Jenkins for suggestions and comments. Support was provided by the National Science Foundation (IBN-CAREER-9875866).