Although male reproductive success is expected to increase with additional copulations and the transfer of numerous or large sperm to females, these may reduce male sperm reserves and affect future reproductive effort. The waterstrider Aquarius remigis Say (Gerridae, Heteroptera) has large sperm (5mm length), copulates frequently (averages one per day) and has a polygynandrous mating system. We test the hypothesis that sperm reserves will decrease as mating opportunities increase by comparing the behavior and sperm reserves of male A. remigis after nine days under three sex ratios (3 males:9 females, 6 males:6 females, 6 males:0 females) and for single copulation trials with two sex ratios (1 male:3 females, 2 males:2 females). We show that males with additional mating opportunities copulate more often and for shorter durations but do not become sperm depleted and frequently have copulations of insufficient duration to result in sperm transfer. The number of sperm transferred was not related to copulation duration, sex ratio, or number of eggs in females. This suggests lack of strategic sperm allocation by males. Multiple mating did not increase female fertility and the number of sperm in female reproductive tracts. Thus, non-virgin females appear to gain no direct benefit from additional sperm transfer, although past research shows females in copula receive less harassment from males. Our results indicate that variation in the success and duration of copulations is best explained by female influence over copulation rather than by male reproductive tactics.

Phenotypic differences between females and males are widely observed in organisms with separate sexes. How these sexual dimorphisms evolve despite the sexes sharing a majority of their genome remains an unresolved issue in evolutionary biology as theoretical models often disagree over the genetic mechanisms that are predicted to facilitate the evolution of sexual dimorphism. In this dissertation, I develop quantitative genetic methods for estimating sex-specific non-additive genetic and sex-linked additive genetic (co)variances. I then empirically quantify the genetic effects underlying population differences in sexually dimorphic traits using the water strider Aquarius remigis.

I show that differences between the sexes in genetic architectures bias estimates of additive genetic variance if these differences are improperly incorporated into quantitative genetic analyses. I then develop the nadiv software package for the R statistical program to facilitate estimation of non-additive genetic (co)variances using the "animal model". Next, I use simulations to demonstrate that estimates of sex-specific additive genetic variances and between-sex additive genetic correlations are biased when sex-linked additive genetic variance is ignored.

I create a genetic model of a sex-linked locus to derive general expressions for the covariance between relatives due to sex-linked genes applicable under any form of global sex chromosome dosage compensation. These expressions lead to the development of formulae and algorithms (incorporated into nadiv) to create sex-linked relatedness matrices for use in animal model analyses. I further show that the way by which sex-linked relatedness matrices under the various forms of dosage compensation differ from one another implies that unbiased estimates of sex-linked additive genetic variance can still be obtained even when the particular form of dosage compensation is unknown.

Using population crosses of water striders, I show the net genetic effects contributing to sexually dimorphism differ between female and male water striders. I demonstrate that the magnitude of this difference in epistatic and dominance maternal genetic effects positively covaries with the magnitude of sexual dimorphism across a range of morphological traits. This is the first study to show that sex-specific non-additive genetic effects correlate with the degree of sexual dimorphism across traits.

Male Aquarius remigis (Hemiptera: Gerridae), produce giant sperm with an unusual morphology: unlike most animal sperm, in which the length of the tail far exceeds that of the head, the heads and tails of A. remigis sperm are approximately equal in length. The acrosome of A. remigis sperm contains a matrix rich in flavin adenine dinucleotide (FAD), a riboflavin-containing molecule. Previous studies have shown that giant sperm are costly to produce, leading to a trade-off between sperm size and number. Furthermore, insects are unable to synthesize riboflavin and must obtain it through their diet; given their FAD-rich sperm, A. remigis males likely require large amounts of dietary riboflavin for sperm production. Thus, given their large, costly, nutrient-rich gametes, I hypothesized that male A. remigis would respond to food restriction with a decrease in sperm number.

Here, I tested the hypothesis that sperm number, and hence male fertility, is limited by food availability in A. remigis with a series of four studies examining (1) sperm numbers and egg fertility in wild males and females in natural populations, (2) the effect of food restriction on sperm number and mating behavior in wild-caught males, (3) the effect of food restriction on sperm number and mating behavior in lab-reared males, and (4) long-term sperm storage in females.

I found that males and females in natural populations often contain few or no sperm, and the fitness of both sexes may, at times, be limited by sperm supplies. Food restriction had a drastic effect on sperm numbers in wild males, with fed males containing much more sperm than unfed males. Mating behavior was also subject to the effects of food restriction, with fed males mating more frequently than unfed males. No such effect of food restriction was seen on sperm numbers or mating behavior in males reared in the laboratory with abundant food. Finally, I found that viable sperm can be stored in the female reproductive tract for at least seven weeks, which is even longer than previously known.

One of the major goals of research in evolutionary biology is explaining the biological diversity of organismal life histories. One major explanation of this diversity is that limited resources acquired by an organism must be allocated to various life processes, leading to trade-offs between these processes. In this dissertation, I use a model system-the trade-off between flight capability and reproduction in wing dimorphic insects-to investigate variation in the acquisition of resources and its subsequent allocation to competing processes.

I showed that the pattern of allocation varies in response to temperature in the water strider, Aquarius remigis and that this response has the potential to evolve. In addition, I used a quantitative genetic simulation model to show that the optimal allocation strategy depends on the predictability of resource levels across time. Specifically, this model predicts completely different responses of allocation to resource level variation in environments with low versus high predictability of resource levels across time.

Combining physiological assays with a quantitative genetic breeding experiment, I measured genetic and phenotypic variation in acquisition and allocation in and across several different resource environments in the wing dimorphic cricket, Gryllus firmus. I found a significant genetic variance for both acquisition and allocation, but I also found that independent evolution in different resource environments is constrained. I show that when allocation is measured independently of acquisition, there is evidence for the trade-off even when it may be obscured by variance in acquisition when measured only as the correlation between functions.

Using this data set, I also tested the assumptions and predictions of a theoretical model (the Y model) that predicts patterns of variation in trade-offs. An assumption of the Y model, that acquisition and allocation will be independent of one another, was not supported. However, despite this result, I found strong support for the predictions of the Y model in this system, demonstrating the Y model is robust to violations of the assumption of independence. This study is the first to use estimates of acquisition and allocation to directly test the Y model.