Claudia M. Rauter, Department of Entomology, University of Kentucky, S-225 Agricultural Science Center North, Lexington, KY 40546-0091, USA. Tel: +1859 257 7472; fax: +1859 323 1120; e-mail: email@example.com
Indirect genetic effects (IGE) of parental care performance and the direct–indirect covariance contribute substantially to total heritability in domesticated and laboratory mammals. For animals from natural populations empirical estimates of IGE are sparse. Thus, despite recent models relating IGE to evolution, evolutionary interpretations of IGE are limited. To address this deficit, we used a reciprocal cross-fostering breeding design to estimate environmental influences, direct and indirect genetic effects, and direct–indirect genetic covariances in the burying beetle Nicrophorus pustulatus to determine the evolutionary importance of IGE arising from variation in parental care performance. Carrion size positively affected adult mass and time on carrion, but had no effect on total development time. Males were slightly larger than females. For both mass and development, independent of these environmental influences, direct and indirect genetic effects were of moderate magnitude. Total genetic effects explained 36–50% of the phenotypic variance in mass and size and 27–37% of phenotypic variance in development time. Direct–indirect genetic covariances were zero or close to zero. Thus, for both mass and development time, the response to natural selection arising from environmental variation may be accelerated by the presence of IGE in N. pustulatus. The generality of this pattern and the evolutionary significance of IGE arising from parental care awaits further study of natural populations.
Variation in parental care, often described in terms of variation in the effects of parental care or as `parental performance', can be an important evolutionary influence (Kirkpatrick & Lande, 1989, 1992; Lande & Kirkpatrick, 1990; Cheverud & Moore, 1994; Mousseau & Fox, 1998; Wolf et al., 1998). A consideration of the evolutionary effects of parental performance, however, requires an explicit consideration of both environmental and genetic influences over multiple generations (Cheverud & Moore, 1994). Like any parental trait, parental performance may be influenced by the environment to which the parent is exposed (Fig. 1: path 1) and the genes expressed in the parent (Fig. 1: path 2). Parental genes and environment indirectly affect trait expression in offspring through their effects on parental performance (Fig. 1: Path 3) and are therefore called indirect genetic effects and indirect environmental effects, respectively. Typically, mothers have the greatest effect on offspring, and indirect genetic effects are symbolized Am and indirect environmental effects Em. In addition to maternal performance, the offspring phenotype is determined by the offspring's genes (i.e. direct genetic effects, Ao; Fig. 1: path 4) and the other environmental effects that the offspring experiences (i.e. direct environmental effects, Eo; Fig. 1: path 5). The impact of direct genetic and environmental effects may, however, depend on indirect genetic and environmental effects, respectively (i.e. correlations between direct and indirect genetic effects [cov(Ao,Am); Fig. 1: path 9] and between direct and indirect environmental effects [cov(Eo,Em); Fig. 1: path 8]).
Both the genes determining parental performance and the genes determining expression of the trait of interest are transmitted from the parent to the offspring (Fig. 1: paths 6 and 7). Thus, although the effects of parental performance genes are experienced as an environment, both indirect genetic effects, Am, and correlation between direct and indirect genetic effects, cov(Ao,Am), may influence the phenotype of the offspring thus affecting response to selection. In contrast to genetic correlations between traits expressed within a generation (e.g. body size and body mass), the genetic correlation between direct and indirect genetic effects has a cross-generational effect and can therefore lead to unusual evolutionary dynamics (Cheverud & Moore, 1994; Wolf et al., 1998).
where VP is the total phenotypic variance and S is the selection differential. VAo are direct genetic effects in the offspring, VAm are the indirect genetic effects of the parent influencing offspring phenotypes through parental performance, weighted by the relatedness, r, between interactants, and cov(Ao,Am) is the covariance between direct and indirect genetic effects weighted by (1 + r) (Cheverud & Moore, 1994).
Indirect genetic effects can have a strong influence on the rate of evolution depending on their magnitude. The contributions of indirect as well as direct genetic effects to responses to selection can be described by `total heritability' (Willham, 1963) which, in mother–offspring interactions (where the coefficient of relationship, r, equals 1/2), is
As this equation illustrates, not only do indirect effects have an influence on evolutionary responses, but also the covariance between direct and indirect genetic effects can be very important (Cheverud & Moore, 1994; Lynch & Walsh, 1998). If cov(Ao,Am) is positive, the response to selection is accelerated. Similarly, if cov(Ao,Am) is negative, then the response to selection is slowed. If the covariance is negative and >|2/3VAo + 1/3VAm| in mother–offspring interactions, the response to selection will be in the opposite direction of natural selection. In addition, indirect genetic effects of parents can cause a time lag in response to selection or even a response after selection has ceased (Kirkpatrick & Lande, 1989, 1992; Lande & Kirkpatrick, 1990). Thus, both the rate and direction of response to selection are influenced by genetic variances and covariances associated with indirect genetic effects of parents on offspring phenotypes.
Although many insects provide extended parental care (Wilson, 1971; Herman, 1981; Tallamy, 1984), studies quantifying the magnitude of indirect genetic effects in insects are lacking. There is, however, some evidence that indirect genetic effects of parental care performance may also be important for insect growth. A recent study on dung beetles (Onthophagus taurus; Hunt & Simmons 2000) reports a large repeatability for parental performance. Agrawal et al. (2001) inferred genetic influences on parental provisioning in a burrower bug (Sehirus cinctus Palisot).
To increase our knowledge of indirect genetic effects in nondomesticated animals with extended parental care, we investigated the genetics and evolutionary importance of parental care performance on growth and development time in the burying beetle Nicrophorus pustulatus Herschel. Burying beetles provide an excellent system for this type of study. Burying beetles have highly developed parental care behaviours (Eggert & Müller, 1997; Scott, 1998). Burying beetles use carrion as the food resource for their offspring. Parents, singly or together, prepare and attend the carrion and feed the larvae with predigested carrion. The behaviour of parents in the laboratory closely resembles field observations (e.g. Wilson & Fudge, 1984; Scott & Traniello, 1990; Trumbo, 1991), providing an opportunity to test evolutionary predictions.
We performed a reciprocal cross-fostering breeding experiment (see Lynch & Walsh, 1998). In this breeding design, we randomly paired two unrelated mothers and exchanged half of the offspring between the two mothers. The experimental design allowed us to partition the phenotypic variance into direct genetic effects, indirect genetic effects due to parental performance, genetic effects, effects due to the covariance between direct and indirect genetic effects of parental performance (Cheverud & Moore, 1994; Lynch & Walsh, 1998).
Life history of Nicrophorus pustulatus
The burying beetle N. pustulatus occurs in forested habitats in southern Canada east of the Rocky Mountains and in the eastern United States from the Canadian border south to Texas and Florida. Adult beetles are active from March to October and reproduction takes place in spring (Anderson & Peck, 1985; Peck & Kaulbars, 1987). Although the reproductive biology of N. pustulatus in the field is not well known, preliminary field data (C. M. Rauter, unpublished data) and breeding experiments in the laboratory (Robertson, 1992; Trumbo, 1992; Rauter & Moore, 2002) indicate that N. pustulatus behaves in the same manner as other burying beetles (reviews by Eggert & Müller, 1997; Scott, 1998). Provided with a dead mouse, N. pustulatus bury the body, remove the fur, work the carrion into a compact ball, and create and maintain a burial chamber. Within 36–72 h after the female has detected the carrion, she oviposits in the earth near the carrion (Robertson, 1992). About three days later, the larvae hatch and crawl to the carrion where they gather in a crater on top of the carrion ball prepared by the parents (Robertson, 1992). The first few days parent beetles feed the larvae predigested carrion. Later the larvae feed by themselves from the carrion and predigested carrion that the parents regurgitate into the crater on top of the carrion. The parents also attend the carrion to slow down decay and growth of mould. Male parents desert the carrion first, whereas female parents usually stay until larval development is complete (Robertson, 1992).
The larvae disperse from the carrion, when most of the carrion is consumed. The subsequent wandering phase lasts several days. Afterwards they prepare a small chamber in the soil where they pupate (Robertson, 1992). Three to four weeks later, adult beetles emerge, that become sexually mature within 1 month (Robertson, 1992).
Reciprocal cross-fostering breeding design
For our breeding experiment we used fourth, fifth, and sixth generation offspring of N. pustulatus from our laboratory colony that we derived from 30 beetles collected in the research forest of Berea College, Kentucky, USA. Each generation 20–50 pairs of unrelated beetles gave rise to the new generation of the laboratory colony. The beetles of the laboratory colony were kept individually in plastic containers (15 × 10 × 5 cm) filled two-thirds with moist peat and fed with previously frozen mealworms or crickets ad libitum twice a week. They were maintained at 24 °C and under a 15L : 9D photoperiod.
Detecting indirect as well as direct genetic effects requires somewhat more complicated breeding designs than most quantitative genetic experiments (Cheverud & Moore, 1994; Lynch & Walsh, 1998). To estimate the direct and indirect causal components of phenotypic variation, we performed a reciprocal cross-fostering breeding experiment. In this breeding design, we matched pairs of families of which the offspring were hatching at the same time as reciprocal cross-fostering units. Within each reciprocal cross-fostering unit, half of the offspring of each family were exchanged with half of the offspring of the other family. Thus for each family, a random half of the offspring was raised by their genetic mother together with foster siblings from the other family. The other half was raised by the other mother of the cross-fostering unit as foster offspring together with half of the genetic offspring of this mother. We raised all offspring with maternal care only. This breeding experiment resulted in 27 reciprocal cross-fostering pairs with 575 offspring.
We initiated the breeding experiment by mating virgin males and females chosen randomly from our laboratory colony. At the beginning of the beetles' activity period, we placed a male and a female into a plastic container (15 × 10 × 5 cm) containing a previously frozen mouse (27.0–34.8 g) and filled half with moist peat. We marked males with a dot of white correction fluid on each elytron and a dot on the pronotum to facilitate distinction between the sexes. Paired beetles were kept in a dark room equipped with red lights at 24 °C. Three times a day the containers were checked for eggs. The day after first spotting eggs, we transferred the female beetle and the mouse to a new container, whereas the male was removed. The old container with the eggs was placed on top of the female's container and monitored for newly hatched larvae three times a day. When the larvae hatched, we transferred the female and the mouse a second time to a new container. The larvae were placed on top of the mouse about 1 h after the transfer of female and mouse to the new container.
To be able to distinguish between genetic and foster offspring within a brood, we marked newly hatched larvae by cutting off the first segment of either the left middle tarsus or the right hind tarsus using a pair of fine surgery scissors. All larvae received family specific marks. The marking technique produced no mortality and did not affect larval growth and development time (paired t-tests, all P > 0.160; growth and development time of 12 marked and 12 unmarked larvae of 11 broods were compared; all larvae were reared without parental care). The only exception was development time from hatching to adult emergence where development of marked larvae took longer than of unmarked larvae (37.8 vs. 36.9 days; t7=2.426, P=0.046). The mark persisted over the two moults that occur during larval development. During pupation the missing tarsus segment was regenerated and the mark was not visible in adult beetles.
Marked larvae were weighed to the nearest 0.01 mg before they were placed on the mouse. A random half (six larvae) of each family was added to the mouse with their genetic mother and the other half (six larvae) was added to the mouse with the foster mother. To control for effects of brood size (Bartlett, 1987; Wilson & Knollenberg, 1987; Bartlett & Ashworth, 1988; Scott & Traniello, 1990), we standardized the number of larvae placed onto a mouse to 12 larvae which is slightly less than the average brood size of 15 larvae for the size of carrion used in the experiment (Rauter & Moore, 2002). Seventy-two hours after placing the larvae onto the mouse, we weighed all larvae on a mouse to the nearest 0.1 mg. To avoid desertion of the carrion by the female, we removed only six larvae at a time.
We determined the onset of larval dispersal by monitoring all containers for dispersing larvae three times a day. All larvae on a mouse were considered as dispersing when at least two larvae were wandering on the surface of the peat away from the mouse. At dispersal, we counted the number of genetic and foster larvae surviving from hatching to dispersal for each mouse and weighed all larvae to the nearest 0.1 mg. To determine the duration of wandering phase and time spent in the pupal chamber, we placed the larvae individually into round containers (diameter: 5.5 cm; height: 3.5 cm) filled two-thirds with moist peat. These containers were kept in a dark room under the same conditions as larvae developing on a mouse. We checked the containers twice a day to determine the end of the wandering phase, the time spent in the pupal chamber and adult emergence. We defined the end of wandering phase and thus the onset of the time spent in the pupal chamber as the time when a larva was laying motionless on the bottom of the pupal chamber for at least two consecutive observations. Two weeks after the onset of pupation, each container was monitored twice a day for newly emerged beetles, which were weighed immediately to the nearest 0.1 mg. We also measured pronotum width of all newly emerged beetles using an optical micrometer on a 20× dissecting scope. Pronotum width is a highly repeatable measurement (repeatability ± SE: 0.994 ± 0.000; 1638 beetles, three measurements each).
Maternal discrimination between genetic and foster larvae
To investigate whether female N. pustulatus discriminate between genetic and foster larvae while providing care, we tested the following prediction. If female N. pustulatus provide preferentially more or higher quality parental care to genetic offspring, survival and/or growth of genetic offspring should be higher.
We estimated the variance components of the observed phenotypic variation using the statistical model:
where μ is the grand mean, sexi is the average effect of the ith offspring sex, Pj is the average effect of the jth reciprocal cross-fostered unit, Mjk is the direct effect of the kth genetic mother within the jth unit, Njl is the direct effect of the lth foster mother within the jth unit, Ijkl is the effect of the interaction between genetic and foster mother within the jth unit, and eijklm is the residual error for the mth offspring of the sex i, born by the kth mother and raised by the lth foster mother within the jth cross-fostering unit.
Factors affecting phenotypic variation independently from parental performance can bias estimation of genetic parameters if they are not included into the statistical model (Robinson, 1996a,b). We therefore controlled for offspring sex, mass of the mouse, and brood size at dispersal by including these factors as covariates in the statistical model. Although our design minimized the impact of brood size or carcass mass by experimentally eliminating as much variation as possible, these two factors are known to affect growth and development time in burying beetles (Bartlett, 1987; Wilson & Knollenberg, 1987; Bartlett & Ashworth, 1988; Scott & Traniello, 1990; Trumbo, 1990, 1992; Eggert & Müller, 1997). While estimating the variance components, however, we included only those covariates that showed a significant effect in the full model, because including factors without effects in the model can bias the estimate of genetic parameters (Meyer, 1992). We used restricted maximum-likelihood to estimate all variance components (Lynch & Walsh, 1998) using the function varcomp in S-Plus version 4.5 (MathSoft, 1997) with the specification REML.
To estimate the genetic parameters, we partitioned the phenotypic variances of offspring growth and development time into causal components following Lynch & Walsh (1998; Table 1). Our breeding design does not allow us to estimate all the potential contributions to variation in offspring traits illustrated in Fig. 1, requiring us to make some necessary assumptions (Lynch & Walsh, 1998). As has been done in previous studies (e.g. Cheverud et al., 1983), we used full-sibs and so assumed that direct and maternal dominance effects are zero. Although the bias arising from this assumption is likely to be small for morphological traits such as body weight, for life history traits such as development time dominance effects can be substantial (Crnokrak & Roff, 1995). Estimates of direct genetic effects on these traits from a previous half-sib study (Rauter & Moore, 2002) allows us to evaluate the seriousness of this potential bias. Our design also requires that we assume that there are no indirect environmental effects and violations of this assumption would lead to biased estimates of indirect genetic variance. Studies in mammals report values for indirect environmental effects expressed in proportion to the total phenotypic variance ranging from –0.1 to 0.32 (Hanrahan & Eisen, 1974; Robinson, 1981, 1996a; Meyer et al., 1994) but we have no information for burying beetles. Finally, we assumed that there was no interaction between direct and indirect environmental effects, which would affect estimates of covariance between direct and indirect genetic effects. There is some evidence that interactions between direct and indirect environmental effects are common in other insects (Rossiter, 1998).
Table 1. Coefficients of causal variance components of phenotypic (co)variation (after Lynch & Walsh, 1998). Capital letters denote causal components: A = additive genetic; D = dominance; E = environmental variance. Small case letter o refers to direct effects on the offspring phenotype, whereas m denotes indirect effects of parent on the offspring phenotype. It is assumed that environmental effects, Em, of a parent m consists of two components: (1) a fraction b of the maternal effect that m's parent exerted on m (i.e. m is the foster mother in the cross-fostering experiment) and which is transferred to the offspring and (2) a unique contribution of m. In the presence of sib competition the unique contribution of m on one sib might depend on m's unique contribution on the other siblings. The constant c accounts for this effect.
These four assumptions simplify the theoretical model illustrated in Fig. 1 by dropping path 1 (i.e. indirect environmental effects, Emw) and therefore path 8 (covariance between direct, Eox, and indirect environmental effects, Emw). As a consequence path 2 (the indirect genetic effects, Amw) is now identical with path 3 (the parental performance, Mw). With our experimental design we were able to estimate paths 2=3, 4, 5, and 9 (Fig. 1).
In addition to parental care, indirect effects may also arise as a result of maternal investment in eggs affecting size of offspring (Clutton-Brock, 1991; Mousseau & Fox, 1998). To assess a potential effect of maternal investment in eggs, we measured body mass of newly hatched larvae. None of the response variables were significantly correlated with larval body mass at hatching (all P > 0.10). Therefore we did not include larval body mass at hatching as a covariate in our analyses.
We assessed the impact of the genetic and environmental contributions to offspring phenotype, and thus the potential for evolution of a specific trait, by comparing the proportions of the phenotypic variation accounted for by each factor. For direct genetic effects, we calculated the heritability (direct heritability, ho2) for each trait as:
where VM is the among-genetic mother variance component and VP is the total phenotypic variance. The heritability for indirect genetic effects (indirect heritability, hm2) was calculated as:
where VN is the among-foster mother variance component. The proportion of the phenotypic variance explained by the covariance between direct and indirect genetic effects was calculated as:
where VMxN is the variance component of the interaction between genetic and foster mother. Direct environmental effects were estimated as:
where VError is the error variance and VM is the among-genetic mother variance component. Heritability of the total genetic effects was determined following Willham (1963) as:
As a consequence of our definition of larval dispersal from carrion (i.e. all larvae of a mouse were considered dispersing when at least two larvae were wandering on the surface of the peat away from the mouse), there was no variation among larvae raised by the same female. Therefore, genetic influences could not be estimated for the development time from hatching to dispersal. The sex of larvae at 72 h of age is not known, because offspring could not be individually identified beyond family membership until the time of dispersal.
Because the data were slightly unbalanced, we estimated the values and standard errors of genetic and environmental effects using the jackknife method (Efron & Tibshirani, 1993; Simons & Roff, 1994; Shao & Tu, 1995). To test whether direct and maternal heritabilities and the proportions of the phenotypic variance explained by the different variance components were different from zero, we calculated 95, 99 and 99.9% confidence limits (Bortz, 1989).
Maternal discrimination between genetic and foster offspring
Survival at dispersal from the carrion did not differ between genetic and foster larvae (paired t-test: t53=0.510, P=n.s.). Mass at 72 h after hatching and mass at dispersal from carrion did not differ either between genetic and foster offspring (paired t-tests: at 72 h: t53=0.159, P=n.s.; at dispersal: t53=0.090, P=n.s).
Phenotypic characterization of growth and development time
Average mass of hatched larvae was 1.39 ± 0.01 mg (mean ± SE; 646 larvae). Within the first 72 h of their life the larvae increased their mass over 200-fold to 307.3 ± 3.1 mg (549 larvae). Thereafter, growth rate was reduced and the larvae dispersed from the carrion with an average mass of 402.4 ± 2.7 mg (542 larvae). During wandering and pupal phase they lost some mass. As a consequence the average mass (346.5 ± 2.6 mg; 524 beetles) of a newly emerged adult beetle was lower than at dispersal. Body mass of newly emerged adult beetles was highly correlated with pronotum width (6.9 ± 0.0 mm; r=0.823, d.f.=515, P < 0.001).
Sex of the offspring exerted a small, but significant effect on offspring phenotype with males being larger than females (all P < 0.001). Average mass of males at dispersal and at adult emergence was 414.1 ± 4.0 and 356.3 ± 3.9 mg, respectively. The corresponding averages for females were 391.7 ± 3.6 and 338.0 ± 3.4 mg. Pronotum width of males at adult emergence was 7.0 ± 0.0 mm. For females average pronotum width was 6.8 ± 0.0 mm.
Larvae dispersed from the carrion 7.2 ± 0.01 days after hatching. The wandering phase took 6.8 ± 0.1 days. The offspring spent 24.7 ± 0.1 days in the pupal chamber. Total development time from hatching until adult emergence was 38.0 ± 0.1 days. Males and females showed no differences in the duration of the developmental phases (all P > 0.72).
The phenotypic variance of mass increased from hatching to 72 h of age. Although the mean increased further from 72 h of age to dispersal from carrion, phenotypic variance decreased. The decrease in the mean from dispersal to adult emergence, however, was mirrored again by a decrease in phenotypic variance. For development time, phenotypic variance increased with increasing mean duration of the developmental phase except for wandering phase, which showed almost as large phenotypic variance as time spent in pupal chamber.
Effects of mouse size and brood size on offspring size and development time
The mass of mouse provided as a resource showed a small, yet significant effect on body mass and size (Table 2). Body mass and size were both positively related with mouse resource for mass at dispersal (b=6.049, F1,431=58.3, P < 0.001) and mass and size at adult emergence (b=4.747, F1,417=24.9, P < 0.001; b=0.028, F1,431=23.6, P=0.001). At 72 h there was a negative relationship between mass and mouse resource (b=–4.235, F1,439=25.2, P < 0.001). Brood size at dispersal influenced only mass at 72 h where mass increased with increasing body size (b=10.996, F1,439=44.6, P < 0.001). All other measures of mass and size were independent of brood size at dispersal (all P > 0.10).
Table 2. Genetic and environmental influences on offspring mass, size, and duration of development during different developmental phases determined from on a cross-fostering breeding design. ho2 = heritability of direct genetic effects, hm2 = heritability of indirect genetic effects, cov(Ao,Am) = covariance of direct and indirect genetic effects, hT2 = total heritability (see text), eo2 = direct environmental effects, VP = total phenotypic variance. Other components of variance estimated were mouse (i.e. mass of mouse resource), brood size (i.e. number of larvae deserting the carrion), and sex (i.e. sex of offspring). Jackknifed values with SE are presented for ho2, hm2, cov(Ao,Am)/VP, and eo2. NCF: number of cross-fostering units consisting of two families; N: total number of offspring.
Mass of the mouse resource affected time spent on carrion (F1,46,=4.2, P=0.047), duration of wandering phase (F1,256=7.2, P=0.008), and time spent in pupal chamber (F1,272=35.9, P < 0.001). Mass of the mouse resource had no effect on total development time from hatching to adult emergence (F1,404=0.4, P=0.524). Time spent on carrion and in pupal chamber were positively related with mouse resource (b=0.087; b=0.202), but duration of wandering phase showed a negative relationship with mouse resource (b=–0.102). Brood size was not related to any measure of development time (all P > 0.30).
Direct and indirect heritabilities of mass and size were of moderate magnitude (Fig. 1: paths 4 and 2/3; Table 2 for all measures of mass and size. The covariance between direct and indirect genetic effects was significantly positive for mass at 72 h (Fig. 1: path 9; Table 3), but its effect on phenotypic variance was zero. All other covariances were not significantly different from zero and their impact on total phenotypic variance was negligible (Tables 2 and 3). Total genetic heritability (i.e. sum of paths 4, 2/3, and 9 in Fig. 1; Table 2, explained 36–50% of the phenotypic variance, with indirect genetic effects accounting for 24–33% of the total genetic heritability. Direct environmental effects were small for larval mass measurements, but substantial for mass of newly emerged adult beetles (Fig. 1: path 5; Table 2.
Table 3. Jackknifed values and SE of the covariance of direct and indirect genetic effects, cov(Ao,Am).
Estimates of direct heritabilities of development time were all about 0.20 (Fig. 1: path 4; Table 2 but only time spent in pupal chamber was significantly different from zero. Indirect heritabilities (Fig. 1: path 2/3; Table 2 were small for wandering phase and time spent in pupal chamber and not significantly different from zero. Indirect heritabilities were moderate, however, for total development time from hatching to adult emergence. The covariances between direct and indirect genetic effects (Fig. 1: path 9; Table 2 showed a small and not statistically significant effect on phenotypic variation. All covariances were not significantly different from zero (Table 3). Total genetic effects (i.e. sum of paths 4, 2/3, and 9 in Fig. 1; Table 2 explained 27–37% of the phenotypic variance in development time with indirect genetic effects accounting for 15–40% of the total genetic variance. Direct environmental effects were large, ranging from 0.35 to 0.57 (Fig. 1: path 5; Table 2.
Maternal discrimination between genetic and foster larvae
Theory suggests that parental care and offspring solicitation may be integrated, resulting from genetic integration of parental and offspring genotypes (Wolf & Brodie, 1998). Theoretical arguments such as these suggest that parents may respond differently to related vs. unrelated offspring. We find no evidence of behavioural biases between parents and foster or genetic offspring, at least in terms of the effects of care on offspring. This result corroborates findings of other burying beetle studies that have examined the possibility of rejection of unrelated offspring. As long as larvae arrive at an appropriate time after females lay eggs, parent beetles do not discriminate between related and unrelated conspecific larvae, or even against larvae of other species (Müller & Eggert, 1990; Trumbo & Wilson, 1993; Trumbo, 1994; Eggert & Müller, 2000).
Phenotypic characterization of growth and development time
The pattern of growth in N. pustulatus resembles other organisms, such as mammals, with extensive parental care. The decrease in the phenotypic variance of mass between 72 h of age and dispersal from carrion, despite the continued increase in mean body mass, suggests a targeted growth pattern (Tanner, 1963; Monteiro & Falconer, 1966; Atchley, 1984; Riska et al., 1984). Variation among individual growth trajectories early in offspring development is characteristic of targeted growth. This variation may be due to differences in onset of linear growth and growth rate, or in efficiency in converting food to energy (McCarthy, 1980). Later in development, growth trajectories converge to a target value indicating compensatory growth of smaller individuals (Tanner, 1963; Monteiro & Falconer, 1966; Atchley, 1984; Riska et al., 1984; Smith & Wettermark, 1995). One potential explanation for this pattern, independent of parental care, is that there is no significant difference between the sexes in the relationship between body size and fitness in burying beetles. Body size is generally under strong directional selection in both males and females (Eggert & Müller, 1997).
Countering this argument, however, is our finding that males were slightly, but significantly, larger than females. In other burying beetles males or females are either of equal size or females are slightly larger (Otronen, 1988; Scott & Traniello, 1990; Smith et al., 2000). The lack of a clear size difference between males and females for most species has been explained by the fact that in burying beetles both males and females fight for carrion with same sex beetles (Eggert & Müller, 1997; Scott, 1998). The reasons for the sexual dimorphism seen here are unknown, and suggest that there may be subtle different ecological pressures for N. pustulatus.
The rather large phenotypic variance of wandering phase may be because of a larger direct–indirect covariance and direct environmental effects compared with time spent in pupal chamber or total development time from hatching to adult emergence. In a previous study we found a similar pattern, that the wandering phase was variable and dependent in part on the speed of development on the carcass (Rauter & Moore, 2002). Evolutionary implications of variable duration of wandering phase are difficult to assess, because little is known about the ecology of this phase in the life cycle of burying beetles.
Effects of mouse size and brood size on offspring size and development time
A potentially important environmental effect on offspring characteristics is the amount of resources available during development. This may be especially true for burying beetles as the resource quantity is fixed. A positive relationship between carrion size and offspring mass has been found in other burying beetle species (Bartlett & Ashworth, 1988; Scott & Traniello, 1990; Trumbo, 1992; Eggert & Müller, 1997) indicating food limitation for the larvae. Similarly, we see a positive relationship between mass at dispersal and at adult emergence and carrion size. In contrast, we found a negative relationship between carrion size and mass at 72 h after arrival on the carrion. This negative correlation may be the result of the fact that it takes parent beetles longer to prepare a larger carrion (Trumbo, 1992).
In contrast to studies of other burying beetles (Wilson & Fudge, 1984; Bartlett, 1987; Bartlett & Ashworth, 1988; Scott & Traniello, 1990), we did not find a negative relationship between brood size and measures of mass and size. The lack of a correlation between brood size and all measures of mass and size, except mass at 72 h, in this study is probably the result of holding the initial brood size constant. The positive relationship between brood size and larval mass at 72 h may be a consequence of that N. pustulatus larvae are able to feed immediately and more larvae might be more efficient in enlarging the crater on top of the carrion thus increasing the surface area of accessible carrion.
Development time from hatching to dispersal from carrion depended strongly on the resource environment. To some extent this is unavoidable; once the carrion is depleted, larvae are forced to disperse and initiate the subsequent developmental phases resulting in a shortened larval period for larvae on smaller carrion. Shortening of larval period and initiation of subsequent development with depletion of food resources has also been shown for dung beetles (Shafiei et al., 2001). This is in contrast to insect larvae with unlimited food supply which extend larval period at low food availability (e.g. Yasuda, 1995; Nijhout, 1999; Xia et al., 1999). Subsequent measures of development time, except the total development time from hatching to adult emergence, were also strongly associated with the amount of carrion available to larvae. The observed relationships between mouse size and duration of wandering phase and time spent in pupal chamber are not as easily explained because it is not known what triggers the end of the wandering phase and the onset of pupation.
For mass and size, direct and indirect heritabilities were of moderate magnitude and total genetic effects explained 36–50% of the phenotypic variation. Comparison of our previous study (Rauter & Moore, 2002) suggests little bias arises from dominance deviations. In contrast to the moderate genetic variances, covariances between direct and indirect genetic effects were not significantly different from zero with one exception, mass at 72 h. The direct–indirect covariances explained therefore very little of the observed phenotypic variations.
The direct and indirect heritability estimates for mass were of comparable magnitude to those reported for mammals and birds (Cheverud et al., 1983; Aggrey & Cheng, 1993; Smith & Wettermark, 1995; Skrypzeck et al., 2000; Neser et al., 2001). These results suggest that indirect genetic effects of parental care performance generally play an important role in increasing the genetic variance of body mass and size, independently of taxon, and thus accelerate the response of body mass and size to natural selection. As body size is often a target of selection, contributions from indirect genetic variances may be important in continued evolution of body size.
In contrast to the genetic variances, the covariances between direct and indirect genetic effects were zero or close to zero for body mass and size. Empirical estimates of the covariances between direct and indirect genetic effects for domesticated mammals are often substantial and mainly negative (Riska et al., 1985; Cheverud & Moore, 1994; Skrypzeck et al., 2000; Neser et al., 2001). The available studies of natural populations with estimates of covariances between direct and indirect genetic effects report mixed results, with a positive correlation in birds (Aggrey & Cheng, 1995; Kölliker et al., 2000) and a likely negative correlation in burrower bugs (Agrawal et al., 2001). Although theoretical considerations predict negative direct–indirect covariances as a result of stabilizing selection on maternal–offspring interactions and parent-offspring coadaptation (Wolf & Brodie, 1998), negative covariances can also be the result of environmentally induced negative dam-offspring covariances or sire-by-year interactions rather than genetic pleiotropy or linkage disequilibrium (Mallinckrodt et al., 1995; Robinson, 1996a,b). Positive covariances can be the result of genetic pleiotropy or linkage, or simultaneous directional selection on both parental care performance and body mass or size. To assess whether direct–indirect covariances are important and generally negative, further studies with larger sample sizes are needed that carefully control for environmentally induced negative parent–offspring covariances.
For development time, direct and indirect heritabilities were also of moderate magnitude, but only direct heritability for time spent in pupal chamber and indirect heritability for total development time were statistically significant. Total heritabilities accounted for 27–37% of phenotypic variance in development time. Comparison with our previous study (Rauter & Moore, 2002) suggests little bias arising from dominance deviations. No direct–indirect genetic covariance for development was statistically significant, although the direct/indirect covariance for time spent in pupal chamber was greater than two standard deviations from zero. This covariance, in contrast to all others, was negative.
Direct and indirect heritability estimates for duration of development time are largely lacking. Monteiro & Falconer (1966) found small direct heritability and large indirect effects (i.e. sum of indirect genetic variance and covariance between direct and indirect genetic effects) for age at sexual maturity in female mice. Williams-Blangero & Blangero (1995) and Kruuk et al. (2000) document substantial to large direct heritability for age at first reproduction, but found no evidence of indirect effects for olive baboons (Papio hamadryas anubis) and red deer (Cervus elaphus L), respectively. In altricial birds, there are only direct heritability estimates available for development time. Duration of growth period showed no direct heritability in great tits (Gebhardt-Henrich & van Noordwijk, 1994) and direct genetic variance of growth period in blue tits (Parus caeruleus L) was only detectable under poor conditions (Kunz & Ekman, 2000).
These findings together with the results of this study suggest that indirect genetic effects of parental care performance can contribute substantially to the genetic variation of development time, especially when direct heritabilities are small. This may be important for the evolution of development time as increased genetic variance from indirect genetic effects will allow a faster and continued response to natural selection. Further studies are needed to allow assessment whether the covariance between direct and indirect effects is generally zero for development time, as shown in this study.
In conclusion, our results suggest that indirect genetic effects of parental care performance in nondomesticated animals can play an important role for the potential of growth and development time to respond to selection, perhaps accelerating evolution. Direct–indirect covariances appear to be relatively small, and can be positive, zero or negative in nondomesticated animals. Because of the limited number of studies general conclusions on the sign and importance of direct–indirect covariances for speed and direction of response to evolution are not yet possible. Further studies in nondomesticated animals with extended parental care are needed to address this problem.
We thank Berea College for allowing us to collect beetles in their forest. We are grateful to C. W. Fox for providing laboratory space. For constructive comments on this manuscript we thank K. Haynes, M. Sharkey, and two anonymous reviewers. This work was financed by the National Science Foundation (IBN-9808629 and IBN-0112418) and federal and state money (KAES #01-08-129).