MLH, JH, and RB conceived and designed the experiments. MLH and JH performed the experiments. MLH, MDJ, and RB analyzed the data. JH and RB contributed reagents/materials/analysis tools. MLH, JH, MDJ, and RB wrote the paper.
The authors have declared that no competing interests exist.
The fitness consequences of mate choice are a source of ongoing debate in evolutionary biology. Recent theory predicts that indirect benefits of female choice due to offspring inheriting superior genes are likely to be negated when there are direct costs associated with choice, including any costs of mating with attractive males. To estimate the fitness consequences of mating with males of varying attractiveness, we housed female house crickets,
Experiments reveal that female crickets that choose an attractive mate have lower survival, but that this can be cancelled out or even outweighed by the increased fitness of her offspring.
Whether mate choice can be maintained by indirect selection when females incur direct costs by being choosy is the subject of ongoing theoretical controversy [
In order to understand how mate choice evolves, it is necessary to estimate the overall effect of mate choice on female fitness [
Evidence from studies that have measured one or a few fitness components has been invoked to support direct benefits [
How fitness should be estimated is controversial [
Here, we measured both direct and indirect fitness components of female house crickets,
Our treatment did not affect the number of grandchildren produced via daughters, via sons, or in total (
In each reduced model individual females' scores for the component(s) listed were replaced with experiment-wide mean scores.
The overall difference between the treatments on
Females housed alone (black line) survived longer than females housed with either type of male (Cox regression Wald1 = 29.636,
When we combined a female's egg number, egg width, and egg length (from the first week of egg laying) into a single index of reproductive effort, we found that females mated to attractive males exerted greater reproductive effort in the first week of the experiment than those mated to unattractive males (principal component 1: attractive = 0.239 ± 0.116, unattractive = −0.233 ± 0.199, randomisation test
To provide an inclusive estimate of the total fitness consequences of mating with an attractive or unattractive male, we quantified both the direct costs to females and the indirect benefits to their offspring. We made two main findings. First, the mating-associated costs borne by females are greater when mating to attractive males throughout their life than when they are mated to unattractive males. Second, these costs are cancelled out (when we use the rate-insensitive measure of the number of grandchildren) and may be outweighed by (when we use the rate-sensitive estimate of the intrinsic rate of increase) the benefits of having offspring with elevated fitness (i.e., indirect benefits).
Contrary to some theoretical predictions [
The costs of choice, including the costs of mating with attractive males, are of central importance to theoretic models of mate-choice evolution [
While we do not know the exact mechanisms driving the survival cost seen in our experiment, our finding that females mated to more attractive males experience lower survival is consistent with sexual conflict between males and females over mating decisions [
Several studies have reported fitness benefits of mating with attractive males. Females mated to such males have been reported to have offspring that have greater longevity [
A number of studies have proposed the use of an aggregate measure of male attractiveness rather than a single morphological indicator [
Due to the nature of our experimental design we were unable to measure all the costs and benefits associated with choosing and mating with attractive males. First, we did not measure sons' ability to compete with other males for access to females. However, in this population of
The fitness estimate of choice in empirical studies may depend on the importance of reproductive timing in the system in question [
There are several reasons why reproducing early and having short maturation times is likely to be advantageous in crickets. First, extrinsic mortality of crickets in the wild is likely to be high. Second, females become less choosy [
Our research constitutes one of the first attempts to directly and simultaneously test the combined direct and indirect effects of mating with males that differ in attractiveness. Only by following the effects of mating with attractive or unattractive males through at least two generations, and through both sons and daughters, is it possible to observe the combined direct effects on female lifetime fecundity and the genetic effects on offspring fitness [
We obtained approximately 1,000 final-instar
In the cultures from which the insects have been derived, crickets are raised in densities ranging from 23,000–34,000 m−3 and fed grain ad libitum. In these conditions males and females mate multiply. Males fight with other males and court females, and there is a positive relationship between male dominance and attractiveness [
The attractiveness trials throughout our experiment were based on latency to mounting for pairs of crickets. While this protocol does not allow all elements of female choice to be measured, in
To obtain males that were either attractive or unattractive to females we ran a two-round tournament. In round one, each male was placed in a clear plastic container (7 × 7 × 5 cm) with a single randomly assigned female, at night, under red lighting. When a female mounted a male, but before spermatophore transfer, they were separated. Once half of the females had mounted, all remaining pairs were separated. Round two commenced with a new female assigned at random to each male. The first half of first-round mounted males to be remounted became our “attractive” treatment males. The half of first-round unmounted males that remained unmounted longest in round two became the “unattractive” treatment males. Only males that courted females during the tournament were used. This biological assay of male attractiveness incorporates all traits that make a male attractive during short-range courtship, rather than a single trait correlated with attractiveness (see [
Forty females were randomly assigned to each of three treatments: attractive, unattractive, and an unmated control. Females were weighed and placed individually in a small plastic container (as above) with food, water, and a petri dish of moist sand for egg laying. Males from the appropriate treatment were randomly assigned to a female. Every 7 d, or whenever a male died, a new male from the same treatment (but from a new tournament) was placed with the female. This allowed us to measure the fitness consequences of the strategy of mating with attractive or unattractive males, rather than the consequences of mating with a given individual male. Food, water, and sand were replaced every 7 d.
Female survival was monitored daily, and the number of eggs laid was counted weekly. Hatching success was estimated as the proportion of eggs that hatched within 14 d of the first egg hatching in each collection. Hatchlings were collected every 3 d, and their mean weight was recorded. Each week, 50 hatchlings per female were separated into two boxes (20 × 13 × 13 cm), each containing 25 nymphs. We monitored offspring survival every 7 d and recorded the time to mature and sex and body weight at eclosion.
If a female had fewer than 50 hatchlings in a given week these were discarded. For these females, the actual number of hatchlings was multiplied by the overall experimental mean for each subsequent offspring fitness measure, to predict the number of grandchildren produced. This is a conservative approach to missing values because it reduces the difference between the treatments.
Offspring generation times were calculated from the time females were first placed with a male until the offspring matured. This takes into account not only the time it takes for the offspring to mature, but also the timing of egg laying. Mature offspring were housed individually, and their survival monitored daily. Ten days after eclosion each son's attractiveness was estimated by placing him with a stock virgin female in a small plastic container for 90 min. Mounted males were separated from females before spermatophore transfer occurred. We used the proportion of a female's sons that were successful in this assay as our measure of sons' average attractiveness (e.g., if 8 of 16 sons were mounted, we assumed that, on average, each son had a 50% chance of mating per encounter with a female).
Ten days after eclosion daughters were placed with a stock male for 12 h to allow mating. Afterwards, survival of sons and daughters was again monitored daily, and sand was collected from daughters weekly. Eggs from daughters were counted to estimate lifetime fecundity.
We calculated two estimates of female relative net fitness when mating with either an attractive or unattractive male. A rate-insensitive estimate, the relative number of grandchildren produced by a female (
To estimate the absolute number of grandchildren each female had (
to the number she had through sons,
The attractiveness of a female's sons was estimated as the proportion of her sons that were mounted in attractiveness trials; longevity is the mean adult life span of a female's sons and
We estimated the absolute intrinsic rate of increase for each female as
where
Due to the non-normal distributions of many fitness components, we tested the significance of treatment differences for each fitness component using two-tailed randomisation tests. In each randomisation test the observed data was randomly assigned to the two treatments 10,000 times.
To explore the sensitivity of our estimates of
We used principal components analysis to investigate the effects of mating with attractive or unattractive males on week 1 reproductive effort via egg number, egg width, and egg length. All three measures showed a strong positive loading on the first principal component, which explained 66% of the variation in the constituent measures. We then tested for differences in female reproductive effort between the treatments using a randomisation test.
We thank L. Bussière, J. Kelley, K. Savage, J. Evans, S. Griffith, H. Kokko, A. Lindholm, K. Monro, S. Zajitschek, F. Zajitschek, and M. Blows for valuable discussion and comments on the manuscript. This research was funded by an ARC grant to JH, RB, and MDJ, and an Australian Postgraduate Award to MLH.
relative intrinsic rate of increase