Sexual Selection on Females: Comparing Two Estimates of Mating Success in a Sex-Role Reversed Insect

SEXUAL SELECTION ON FEMALES: COMPARING TWO

ESTIMATES OF MATING SUCCESS IN A SEX-ROLE

REVERSED INSECT

Laura Jane Robson

A thesis submitted in conformity with the requirements for the degree of Master of Science Graduate Department of Ecology and Evolutionary Biology University of Toronto

Abstract please enjoy this thesis and con ranting e

Sexual selection on females: comparing two estimates of mating success in a sex-role reversed insect

Laura Jane Robson

Master of Science

Department of Ecology and Evolutionary Biology

University of Toronto

While there has long been interest in the form of sexual selection in males, studies characterizing this selection in females remain sparse. Sexual selection on females is predicted for sex-role reversed Mormon crickets, where males are choosy of mates and nutrient-deprived females compete for matings to gain nutritious nuptial gifts. I used selection analyses to describe the strength and form of sexual selection on female morphology. There was no positive sexual selection on the female body size traits predicted to be associated with male preferences and female competition. Instead, I detected selection for decreasing head width and mandible length. Additionally, I tested the validity of a commonly-used instantaneous measure of mating success (mated vs. unmated) by comparing selection results with those determined using a more detailed fitness measure (cumulative mating rate). The two fitness measures yielded similar patterns of selection, supporting the common sampling method comparing mated and unmated fractions.

ii Acknowledgements I hope at you ca n find l

I must give my sincerest thanks to my lab-mates, family and friends whose help and encouragement made this degree so rewarding. Above all, I am grateful to Darryl Gwynne for giving me the opportunity to work in his lab. While I can’t thank Darryl enough for letting me chase Mormon crickets and for vastly improving my thesis, I am equally grateful for the exceptional lab that he has created. Kevin Judge is endlessly generous with his time, and I benefitted from his guidance constantly. Kyla Ercit was the best field companion imaginable, and without her help I would have collected half the crickets while being twice as panicked.

Edyta Piascik helped measure eggs and was always available for advice and editing. Jill

Wheeler and Andrew MacDonald were incredibly patient in coaching me through R, Murray

McConnell livened up lab meetings and David Punzalan provided eleventh-hour statistics advice.

As always, I must thank my parents, Debbie and Gord Robson, for their support and encouragement throughout this degree and all of my endeavours. I am grateful for too many reasons to list here, but please know that I’m aware of how lucky I am to have you. Thank you. I am also thankful to Emily Robson and Brock Hart for feeding and entertaining me, and to Ian Young for always letting me “talk it out”. Many thanks to Emma Stewart, Stephanie

Scott, Jacqueline Saint-Onge, Jamie Munro, Emma Rogers, Emily Macleod and Brechann

McGooey for enjoyably distracting me from my thesis. I owe you.

Finally, I am grateful to the Mormon crickets. While some of their habits may not be appealing, they are fascinating animals and I hope to see them in the field again someday.

This research was supported by a Natural Sciences and Engineering Research Council

(NSERC) Canadian Graduate Scholarship to Laura Jane Robson and a NSERC Discovery

Grant to Darryl Gwynne.

iii Table of Contents I hope that you can find eve r

ABSTRACT ii ACKNOWLEDGEMENTS iii TABLE OF CONTENTS iv LIST OF TABLES vi LIST OF FIGURES vii LIST OF APPENDICES viii

CHAPTER 1: Sexual selection on females: comparing two estimates of mating success in a sex-role reversed insect

1.1 Introduction 1

1.2 Materials and methods 5

Life history of Mormon crickets 5

Sample collection 7

Calculation of cumulative mating rate 8

Morphological traits measured 8

Measurement of selection 9

Comparison of fitness measures 11

1.3 Results 11

Comparison of collection sites 11

Selection on original trait axes 12

Canonical analyses of selection 12

Correlations of morphological and fecundity traits 13

Comparison of fitness measures 13

iv 1.4 Discussion 14

REFERENCES 21

TABLES 29

FIGURES 33

APPENDIX A 41

v List of Tables T

1.1 Selection coefficients estimating sexual selection on nine female morphological traits

using cumulative mating rate as the measure of fitness 29

1.2 Eigenvectors from the canonical rotation of the γ matrix presented in Table 1.1 30

1.3 Spearman rank correlations among the female morphological traits and fecundity

measures 31

1.4 Selection coefficients estimating sexual selection on nine female morphological traits

using instantaneous mating success (mated vs. unmated) as the measure of fitness 32

vi List of Figures F

1.1 Relationship between female spermatodose number and adult age 33

1.2 Thin-plate spline of sexual selection on female head width and mandible length 34

1.3 Contour map showing points distribution of sexual selection on female head width and

mandible length 35

1.4 Thin-plate spline of sexual selection along the first and ninth major axes of nonlinear

selection 36

1.5 Contour map showing points distribution of sexual selection along the first and ninth

major axes of nonlinear selection 37

1.6 Thin-plate spline of sexual selection along the eighth and ninth major axes of nonlinear

selection 38

1.7 Contour map showing points distribution of sexual selection along the eighth and ninth

major axes of nonlinear selection 39

1.8 Frequency of mated an unmated females by spermatodose number 40

vii List of Appendices F

APPENDIX A: Supplementary Figures 41

Figure A.1: Length measurement of the female’s foreleg spine 41

Figure A.2: Width measurement of the female’s last abdominal plate 42

viii Chapter 1 1

Sexual selection on females: comparing two estimates of mating success in a sex-role reversed insect

Laura J. Robson

1.1 Introduction

Given the ornaments, armaments and incredible displays often associated with mating

(Darwin 1871), there has long been interest in understanding sexual selection in the field

(Endler 1986) and this area of research has shown no decline in popularity in recent years

(Owens 2006). The use of selection analyses (Lande and Arnold 1983) to measure selection on individual traits and combinations of traits (Philips and Arnold 1989, Blows and Brooks

2003) has led to a surge in the number of studies quantifying the strength and form of sexual selection in the wild (reviewed in Kingsolver et al. 2001, Hunt et al. 2009). Many of these studies rely on cross-sectional sampling of their mating population, comparing the phenotypes of mated and unmated individuals (e.g. Arnqvist 1992, Sadowski et al. 1999, LeBas et al.

2003, Bertin and Fairbairn 2005, Bussière et al. 2008). In this study, I test the validity of this commonly used fitness classification by comparing it to a more detailed measure of mating success (cumulative mating rate).

The second goal of this study is to describe sexual selection on females in a system in which the sex roles are reversed: females compete for matings rather than males. Studies of sexual selection have focused mainly on males as they are nearly always the more sexually competitive sex with greater variation in mating success (Shuster and Wade 2003). However,

1 there is accumulating evidence of widespread sexual selection on females (Cunningham and

Birkhead 1998; Clutton-Brock 2007, 2009). In most animal mating systems, females are more selective of mates than males, and males are the more competitive sex (Darwin 1871). The operational sex ratio is generally male-biased due to the typical differences in parental effort among the sexes: female parental effort almost always exceeds that of the male due to the large energetic investment in the nutrient-rich egg compared to that of sperm and often also due to the increased brood care provided by females (Trivers 1972). However, male investment can offset the high female parental investment via the contribution of nutrients or protection to the mating female or her offspring. In many insect taxa, these contributions take the form of male nuptial gifts, which can be prey items (e.g. dance flies, Downes 1970), seminal chemicals (e.g. pyrochoid beetle, Eisner et al. 1996), male haemolymph (e.g. nemobiine crickets, Fedorka and Mousseau 2002) and even male tissue (e.g. sagebrush cricket,

Dodson et al. 1983). Some of the most nutritious mating gifts are the large glandular spermatophylaces of katydid species in which females compete for matings (Orthoptera:

Tettigoniidae) (Gwynne 2001).

One way to measure sexual selection on females and males is to examine the relationship between mating success (number of mates) and reproductive success

(fecundity/fertility) (i.e. Bateman gradient: Bateman 1948, Arnold and Duvall 1994). For individuals of the sex with the steeper Bateman gradient, reproductive success is limited mainly by mating opportunity and they must often engage in intrasexual competition for mates

(Bateman 1948). In species where males provide for their mates or offspring and show sex- role reversal, the female Bateman gradient can exceed that of males (e.g. Jones et al. 2000).

Paternity data are required to calculate the male gradient (but see Lorch 2005), and regardless of the gradient slope, sexual selection will only occur when mating success depends on the

2 expression of a given trait such as an ornament or armament (Bateman 1948, Arnold 1994).

Sexual selection on these traits can be measured directly by determining how the expression of the trait influences mating success using selection analysis (Lande and Arnold 1983). While selection analyses have been used to measure sexual selection on males in a diverse array of vertebrate and invertebrate taxa (Hunt et al. 2009), studies using such analyses to examine sexual selection in females are limited to nuptial-feeding dance flies (LeBas et al. 2003,

Wheeler 2008). In this study I quantify the strength and form of selection on females of the sex-role reversed Mormon cricket, Anabrus simplex , a species in which there appears to be intense sexual selection on females in nature (Gwynne 1981).

Males of the Mormon cricket ( Anabrus simplex ), a shield-back katydid from the western United States, provide their mates with a proteinaceous spermatophylax that constitutes 20-30% of their body weight (Gwynne 1981). High-density band-forming populations are found in the sage-brush desert of the western slopes of the Colorado Rocky

Mountains. Here, they are sex-role reversed (Gwynne 1981). Given the protein-deficient diet of Mormon crickets in the sagebrush desert (Simpson et al. 2006), the large spermatophylax appears to be particularly costly for the male to produce, and hungry females are thought to forage for males bearing nuptial gifts (Gwynne 1993). Experiments with other katydids exhibiting large nuptial gifts (e.g. Requena verticalis, Gwynne 1990) have revealed that starvation (and the resulting rarity of males available for mating) may cause a female-bias in the operational sex ratio. Given this reversal in the sex roles, I predict that the opportunity for sexual selection on females will be high. Early studies supported this prediction: Gwynne

(1984) found a size advantage (measured as both weight and pronotum length) in mating success for female crickets at gregarious sites, but not at sites where Mormon crickets are not role-reversed. Similarly, Gwynne (1981) found that gregarious males favoured heavier

3 females as mates, with female weight correlating positively with ovarian egg number. Despite behavioural evidence of sexual selection on females, no formal analyses of characters under sexual selection have been done.

Given that female Mormon crickets have been observed racing toward calling males and grappling with other female competitors for males that often reject their female suitors

(Gwynne 1981), I predict that there will be a mating advantage to females possessing traits that enable them to effectively compete in pre-mating scrambles and skirmishes and traits that might signal fecundity to males. Despite evidence for strong sexual selection on females, there are no obvious sex-specific secondary sexual traits in Mormon crickets. This is not surprising given that female-specific secondary sexual traits are rare. One hypothesis to explain this scarcity is that female investment in secondary sexual traits may be constrained by their large investment in egg production, and that males may avoid ornamentation that comes at the cost of fecundity (Fitzpatrick et al. 1995). However, even under these constraints, some level of ornamentation is expected if males cannot assess fecundity directly and require ornaments as signals (Chenoweth et al. 2006). For example, in long-tailed dance flies ( Rhamphomyia longicauda ) swarming females display inflated abdomens and scaled legs to males bearing nuptial gifts of prey, and sexual selection for this ornamentation has been detected (Wheeler

2008). Preferences for female colour ornaments also occur in pipefish showing exclusive male care and role-reversal (Berglund et al. 1986) and in two-spotted gobies with operational sex ratios that are occasionally female-biased (Amundsen and Forsgren 2001).

In addition to examining traits under sexual selection in female Mormon crickets, I also test the validity of a commonly-used instantaneous (cross-sectional) measure of mating success by comparing estimates of sexual selection determined using cross-sectional data to those generated using an estimate of cumulative mating rate. While sexual selection in the

4 wild is best estimated using lifetime mating success (Endler 1986), there are relatively few longitudinal studies of mating success from the field (e.g. Clutton-Brock et al. 1982, Madsen and Shine 1994, McElligott and Hayden 2000, Purse and Thompson 2005), in part because of the difficulty of observing every mating in an individual’s lifetime. Instead, instantaneous measures of mating success determined by cross-sectional sampling are often used (Arnqvist

1992, LeBas et al. 2003, Mason 1964), where individuals are classified as either “mated” or

“unmated”. This provides a conservative measure of differences between the two mating categories (Mason 1964) because, for example, “unmated” individuals may have already mated or be about to mate. In shield-back katydids (subfamily: Tettigoniinae), a self contained sperm-sac (spermatodose) forms within the female’s spermatheca following each mating (Boldyrev 1915, Vahed 2003), allowing us to determine the number of matings that the female gained prior to collection. By counting spermatodoses, I calculated cumulative mating rate (number of matings per day since sexual maturity) as a second and more detailed measure of mating success for the samples of females collected as “mated” or “unmated” in the field.

With these data, I assessed whether the commonly used “mated vs. unmated” field classification provides a valid proxy for more accurate (but more difficult to obtain) measures of mating success in the field.

1.2 Materials and methods

Life history of Mormon crickets

Mormon crickets are flightless katydids that form two population types: gregarious

(band-forming, high density) populations are typically found west of the continental divide from Montana to Nevada, and solitary (non band-forming, low density) populations are common on the eastern slopes of the Colorado Rocky Mountains (MacVean 1987; Gwynne

5 1984). Solitary Mormon crickets live at low densities that do not appear to have the level of food limitation observed in the bands and they show typical sex-roles of choosy females and competitive males (Gwynne 1984). Meanwhile, the gregarious Mormon crickets form enormous migratory bands in the sagebrush desert and show sex-role reversal (Gwynne 1981,

1984). Population differences in food availability appear to affect the sex roles in a similar manner in two other tettigoniids, Metaballus litus (Gwynne 1985) and Kawanaphila nartee

(Simmons and Bailey 1990, Gwynne and Simmons 1990).

From the observed mating behaviours of the role reversed male and female Mormon crickets in the field, predictions can be made about which morphological characters may influence female mating success. I predict positive selection on hind femur length, as longer legs may be advantageous in the apparent race to find a calling male (Gwynne 1981, see also

Gwynne and Bailey 1999). Furthermore, as large head and mouthpart size allow males to defeat rivals in intrasexual fights among some orthopterans with typical sex roles (e.g. Kelly

2006, Judge and Bonanno 2008), I predict positive selection for increasing female mandible and maxilla lengths in Mormon crickets, as females are known to grapple, bite and head butt for access to sexually receptive males (Gwynne 1981, 1984). Additionally, femur length, head width and maxilla span are sexually dimorphic in band-forming Mormon crickets, with females possessing larger trait sizes than their male counterparts (Judge et al. unpublished).

While dimorphism between the sexes may be a result of the differing selective regimes experienced by males and females occupying slightly dissimilar ecological niches (Shine

1989), these dimorphisms are often thought to arise from the differing sexual selection on males and females (Darwin 1871) , as is predicted for these traits. Both male and female

Mormon crickets possess a large forward-facing spine on the coxae of their forelegs (Figure

A.1). I predict that this trait could be used both as a defence against predation and as weapon

6 in grapples between females, the latter predicting positive sexual selection for spine size in this sex. I predict a mating advantage to females possessing traits signalling body size (such as pronotum and ovipositor length), as choosy males are known to prefer larger females

(measured as mass, Gwynne 1981; measured as pronotum length, Gwynne 1984).

Unfortunately, mass could not be reliably measured in this study given the crickets’ tendency to vomit and defecate upon capture. Given the nutritional value of the male’s nuptial gift, I anticipate selection for traits that ensure the complete transfer of the large spermatophylax gift which is exuded from the male’s accessory glands at the end of copulation (Gwynne 2001) . In the field, I noted that the female’s last abdominal sternite (Figure A.2) appeared to protrude against the male’s back during copulation, and I predict a mating advantage to increased size of this structure as it may aid in maintaining or enforcing intromission. Finally, given the high nutrient value of the male’s nuptial gift, I predict that increased mating rate will be followed by increased fecundity, resulting in a positive correlation between sexually selected female characters and egg number or egg size.

Sample collection

I sampled crickets from two collection sites of sage-brush desert in Idaho and Nevada: near Marsing, Idaho (N 43 26.403, W 116 51.572) from June 21, 2008 to June 29, 2008; and near Tuscarora, Nevada (N 41 12.703, W 116 00.282) from July 3, 2008 to July 9, 2008. Both roadside sites were on U.S. Bureau of Land Management rangeland. Collections at both sites began at the onset of the mating season within the bands and continued until mating pairs were infrequently found. Sample size differed greatly between the two collection sites, although roughly half of the crickets collected from each site were classified as mated in the field

(Idaho: n total =478, n mated =255, n unmated =223; Nevada: n total =281, n mated =139, n unmated =142).

7 Females scored as mated were captured in copula or carrying a spermatophore, indicating recent mating. For each mated female that was collected, the nearest female without a mate or spermatophore was collected and scored as unmated. Crickets were killed in

95% isopropyl alcohol and were then preserved in 75% ethanol.

Calculation of cumulative mating rate

The cumulative mating success for each female was determined by counting the number of spermatodoses within the female’s spermatheca. Cricket age was determined by counting the number of daily growth rings within the cuticle of the tibia (Neville 1963) using phase-contrast microscopy (Leitz Laborlux K). Ring number represents the number of days since the cricket’s eclosion to adulthood. In agreement with behavioural observations from the field (Gwynne, personal observation), I determined that the onset of sexual maturity occurred between five to eight days post eclosion (Figure 1.1). Therefore, to eliminate sexually immature females from the analysis, I excluded females under six days post-eclosion from the study. Cumulative mating rate was then calculated as the number of matings a female gained per day since she reached sexual maturity.

Morphological traits measured

Nine morphological traits were measured for the analysis. Hind femur length (mean length of left and right legs) and ovipositor length were measured using digital hand callipers

(Mastercraft). Head width, maxilla span (distance between the cardo-stipes articulation of the left and right maxillae), mandible length (mean of left and right mandibles), maxilla length

(mean of left and right maxillae), pronotum length (mean length of left and right sides at longest point), abdominal plate width (measured for the most distal abdominal ventrite, Figure

A.2) and foreleg spine length (mean of left and right spines, Figure A.1) were measured from digital photographs using the digital imaging program, Image J (version 1.38x). Images were

8 taken using a digital camera (Leica DFC290) fitted to a dissecting microscope (Wild

Heerbrugg M5A). All morphological measurements were made by the same observer.

Measurement repeatability (Lessells and Boag 1987) with a new observer was greater than

85% for all traits except foreleg spine length (repeatability=66%) (Robson, unpublished data).

I counted the total number of eggs within each female’s ovaries and measured the egg length and width for ten randomly selected eggs using digital imaging as above. If the female had fewer than ten eggs, the lengths and widths of all eggs were measured. A composite measure of egg size was calculated as the square root of the product of mean egg width and mean egg length. Egg size measures were made by two separate observers, and repeatability between observers for both egg length and width was >98% (Robson, unpublished data).

Measurement of selection

To investigate how the morphological characters predicted female mating success, I conducted a selection analysis (Lande and Arnold 1983) using multiple regressions of relative fitness (measured as cumulative mating rate) on the standardized morphological traits. All morphological traits were z-score standardized and relative fitness was calculated by dividing each female’s mating rate by the sample mean. The partial regression coefficients of the linear regression give the linear selection gradients ( β), which are measures of the direct selection on the trait after correction for correlation between traits. A separate nonlinear regression including linear, quadratic and correlational terms was used to generate the nonlinear selection gradients ( γ) (Lande and Arnold 1983). The quadratic regression coefficients were doubled to give the quadratic selection gradients (Stinchcombe et al. 2008). In order to further examine any nonlinear selection, I conducted a canonical analysis, which rotates the coordinate space to generate the major axes of nonlinear selection (Blows and Brooks 2003). The canonical rotation yields the same number of major axes as original traits (nine for this analysis); these

9 major axes are composed of linear combinations of the original traits multiplied by their eigenvector loadings. Linear ( θ) and nonlinear ( λ) selection on these canonical axes was determined using linear and quadratic multiple regressions of relative fitness (mating rate) on the individual scores (m-scores) on the major axes. Pair-wise non-parametric thin-plate splines

(“fields” package in R) were used to visualize the shape of the selection on the major axes.

Contour mapping was used to determine the distribution of points in relation to the observed fitness optima. In order to compare the selection surfaces between figures, the same smoothing parameter (sp) was used for all thin-plate splines. A smoothing parameter of sp=4.2 was used as it was intermediate to the smoothing parameters that minimized the GCV (generalized cross-validation) criterion for Figures 1.3 (sp=4.321) and 1.4 (sp=4.09). The same smoothing parameter was applied to Figure 1.2, although the GCV-estimated smoothing parameter for this thin-plate spline was extremely high (sp=126200), suggesting that the GCV model predicted a flat plane.

Univariate selection differentials (s) were calculated for each trait using linear regressions of relative fitness (cumulative mating rate) on the individual trait. These univariate selection differentials represent the overall selection (both direct and indirect) acting on the trait without correction for any correlation between characters.

The significance of all linear, quadratic and correlational selection coefficients and gradients was determined using permutation testing, where I calculated the F-statistic distribution from 1000 iterations of the regression model in which the fitness variable of the dataset was shuffled (Legendre and Legendre 1998).

To test for differences in selection between the two collection sites (Idaho and

Nevada), the sequential model building (partial F-test) procedure described by Chenoweth and

10 Blows (2005) was used to examine the effect of the inclusion of location on the linear, quadratic and correlational selection.

Correlations between the nine unstandardized morphological traits and the fecundity traits of egg number and egg size were calculated using Spearman’s rank correlations.

Significance testing was corrected for multiple comparisons using sequential Bonferroni adjustments (Rice 1989) with α= 0.05.

Comparison of fitness measures

A separate selection analysis was conducted as above, using the instantaneous mating success of “Mated vs. Unmated” as the fitness measure: mated females were assigned a fitness value of one and unmated a fitness value of zero. The partial F-test procedure described above was used to test for differences in the selection when the two different mating success metrics

(mated/unmated vs. cumulative mating rate) were used as the measure of fitness. A side by side frequency plot showing number of matings (measured as spermatodose number) for females classified as mated and unmated was produced to illustrate the accuracy of the instantaneous mating success classification in the field.

1.3 Results

Comparison of collection sites

The linear, quadratic and correlational selection gradients were not found to differ significantly between the two collection locations (linear: F= 0.545, df=9, 739, p=0.842, quadratic: F= 1.318, df=9, 721, p=0.224, correlational: F=0.825, df=36, 649, p=0.758). Data from the two sites were therefore pooled together for the subsequent analyses.

11 Selection on original trait axes

Contrary to my prediction of sexual selection for increasing mouthpart size, I found significant net (univariate) and direct linear selection for decreasing head width and mandible length (Table 1.1, Figures 1.2 & 1.3). I also detected significant negative quadratic selection on mandible length (Table 1.1). Thin-plate spline visualization shows that the negative nonlinear selection on mandible length is not stabilizing but instead directional selection for shorter mandibles with a slight plateau at the smallest values of mandible length (Figure 1.2).

No significant linear or quadratic selection was detected for the seven other traits predicted to be under sexual selection (femur length, ovipositor length, maxilla length and span, abdominal plate width, pronotum length and foreleg spine length). No significant correlational selection between any of the characters was found.

Canonical analysis of selection

Canonical analysis revealed three major axes (m1, m8 and m9) under significant nonlinear selection (Table 1.2). Thin-plate spline visualization shows accelerating directional selection for increasing values of m1 (Figures 1.4 & 1.5), while the negative nonlinear selection on m9 appears to stabilizing, as seen by the saddle shaped ridge through the data when m9 is plotted with both m1 and m8 (Figures 1.4 & 1.6 respectively). While the selection surface of m8 appears to have two fitness peaks (which can indicate disruptive selection)

(Figure 1.6), the abrupt peak at high values of m8 is driven by a single point (Figure 1.7), and the overall selection on m8 is strongly stabilizing. While canonical analyses are primarily used to measure the selection on the multivariate phenotype instead of the individual traits alone

(Blows & Brooks 2003), the loadings of the original traits on the major axes may be used to determine the influence of individual traits on the nonlinear selection observed in canonical space (Brooks et al. 2005, Bentsen et al. 2006, Hall et al. 2008). In this analysis, the

12 biological significance of the trait combinations on most of the canonical axes is unclear and, with the exception of m1, breakdown of the trait loadings on the major axes does not lend itself to easy biological interpretation. However, the directional selection for increasing values of m1 appears to favour crickets with decreasing head width, maxilla length and maxilla span and increasing abdominal plate width and foreleg spine length (Table 1.2, Figures 1.4 & 1.5).

This selection is consistent with the selection for smaller head and mouthpart size observed in the original trait space and opposes my predictions of selection for increased head-weaponry size.

Correlations of morphological and fecundity traits

None of the nine morphological traits were significantly correlated with either egg number or size (Table 1.3). Therefore, contrary to my predictions, neither the traits under significant sexual selection (head width and mandible length) nor those indicating body size

(pronotum length) appear to be predictors of female fecundity. However, as mated female

Mormon crickets appear to lay eggs continuously throughout their sexually mature lives

(Cowan 1929), ovarian egg number may be a poor measure of fecundity as it fails to account for eggs laid prior to collection. The two fecundity traits of egg number and egg size were positively correlated (r=0.550, p<0.001). Each morphological trait showed significant positive correlations with the other morphological characters, with the exception of abdominal plate width, which was not significantly correlated with ovipositor length, maxilla length or foreleg spine length (Table 1.3).

Comparison of fitness measures

The linear, quadratic and correlational selection gradients were not found to differ significantly between the two fitness measures (linear: F= 0.942, df=9, 1498, p=0.487, quadratic: F= 0.490, df=9, 1480, p=0.882, correlational: F=0.484, df=36, 1408, p=0.996).

13 Generally, the magnitude and direction of selection on the traits were very similar between the two fitness measures (Table 1.4 gives the β and γ values for the selection analysis using the

“Mated/Unmated” fitness measure). Of the 365 females classified as unmated upon collection,

75% were virgins (lacking spermatodoses), supporting the validity of the unmated classification (Figure 1.8).

1.4 Discussion

While longitudinal studies of complete lifetime mating success are the gold standard for measuring sexual selection using selection analysis (Lande and Arnold 1983), the cross- sectional comparison of mated and unmated individuals is commonly used in the field for systems where it is difficult to observe lifetime mating success (e.g. Mason 1964, McLain

1981, Arnqvist 1992, Sadowski et al. 1999, LeBas et al. 2004, Bertin and Fairbairn 2005,

Bussière et al. 2008, Wheeler 2008). Despite the frequent use of cross-sectional data, only the present study and that of Punzalan et al. (2009) have compared such measures to cumulative estimates of mating success. In their field study of sexual selection on male ambush bugs

(Phymata americana ), Punzalan et al. (2009) found the selection measured using lifetime mating success to be qualitatively similar to the selection estimated using cross-sectional data.

In this study, I found that both cross-sectional and cumulative mating rate data gave very similar estimates of sexual selection. Taken together, these two studies suggest that cross- sectional mating success can be a valid proxy for longitudinal or other cumulative, more detailed estimates of fitness and these findings increase my confidence in the selection results of similar studies comparing mated and unmated individuals. The discussion below addresses only the cumulative mating rate results, as this is the more detailed fitness measure.

14 Species with sex-role reversed mating systems are ideal for the study of sexual selection on females (LeBas et al. 2003, Wheeler 2008) and this research provides a rare description of the strength and form of this selection using gregarious female Mormon crickets. As the more competitive sex in the sex-role reversed populations (Gwynne 1984,

1993), I expected females to exhibit high variation in mating success, resulting in selection for traits that provide an advantage in winning direct sexual competition or attracting males.

Female mating success among sexually mature crickets varied from zero to nine matings

(indicating the potential for intense sexual selection, Shuster and Wade 2003) and I detected sexual selection on two of the nine candidate traits. Contrary to prediction, significant sexual selection for females with narrower heads and shorter mandibles was found. I anticipated positive selection for maxilla length and span, foreleg spine length, hind-leg femur length, body size (measured as pronotum length and ovipositor length) and abdominal plate width, but detected no significant sexual selection for any of these traits. The absence of sexual selection on femur length and maxilla span is noteworthy, as these traits are sexually dimorphic, being larger in females than males for a given body size (measured as pronotum length) (Judge et al. unpublished). These findings could indicate that these dimorphisms are not caused by differences in sexual selection acting on the two sexes but instead may result from other sex differences in natural selection (Shine 1989). For example, females may be at greater risk of predation than males due to their increased movement during mate-searching or oviposition, resulting in greater selection in females for increased leg length to enable quick escape.

While not statistically significant, there was a trend towards direct linear selection for increasing ovipositor length. During mating, the female bends her genitalia downward to engage those of the male and her abdomen appears to hinge around his genitalia at the base of the ovipositor, with the ovipositor often contacting the underside of the male’s abdomen

15 (personal observation). A longer ovipositor may aid this hinge in grasping the male. This in turn may ensure copulation until the spermatophylax is transferred. There is evidence of a grasping device at the base of the ovipositor in another sex role reversed tettigoniid,

Kawanaphila nartee (Gwynne 2001). Alternatively, male Mormon crickets may prefer females with long ovipositors in order to optimize oviposition depth in the soil for maximum egg survival (the latter is known for gryllid ground crickets: Masaki 1979, Bradford et al.

1993).

When competing for mates, female Mormon crickets grapple face-to-face (Gwynne

1981, 2001, personal observation). These fights are reminiscent of those seen among males of non sex-role reversed male field crickets, where there is evidence that larger head weaponry

(head width, maxilla span, maxilla length and mandible length) provides an advantage in intrasexual competition (Gryllus pennsylvanicus , Judge and Bonanno 2008). Thus, I anticipated that larger head and mouthpart size would provide a mating advantage to fighting females in the field. Contrary to my predictions, I instead found significant sexual selection for small head width and mandible length. While it is possible that there is some unforeseen mating advantage to smaller mouthparts, the observed selection may instead be a result of selection on an unmeasured trait that is correlated with mouthpart size. For example, mouthpart size and shape influence an individual’s ability to process its food, and it is possible that females with smaller mouthparts are better able to manipulate their food into particle sizes that optimize nutrient assimilation (Clissold 2007). If so, females with smaller mouthparts may be heavier due to their improved food uptake. Female mass was not measured in this experiment, but a previous study in the field has shown that male Mormon crickets favour heavier females as mates (Gwynne 1981). Perhaps smaller heads and mouthparts are simply correlated with female weight, which could be the actual trait under sexual selection. All

16 selection analyses are subject to the influence of unmeasured traits. While this common problem can be addressed by the addition of more candidate traits (Lande and Arnold 1983), most studies have inadequate sample sizes to both include all possibly selected traits and maintain the statistical power required to detect selection.

Alternatively, the observed sexual selection for small heads and mouthparts may result from condition dependence in female reproductive strategies. From experiments examining the effect of nutritional environment on Mormon cricket sexual behaviour, Gwynne (1993) found that the increased food-stress typical of gregarious bands led to a reversal in sex roles.

Given these findings, I propose that variation in nutrient acquisition among individuals could lead to differences in mating behaviour within the same population. Within the gregarious bands, competition for food is intense (Simpson et al. 2006), presumably resulting in variation in body condition among individuals. Cannibalism is common (Simpson et al. 2006) and larger females with greater weaponry are predicted to attain the best condition through cannibalism and fights over food. This may lead to females of different conditions also differing in their sexual strategies: large, high-condition females mate only to gain fertilization, whereas small, low-condition females mate many times to meet the energetic demands of egg production. Indeed, in a preliminary study, Lorch (unpublished) found a trend for a negative relationship between mating rate and nutritional status (as measured by stable nitrogen isotope ratio, which indicates the degree of herbivory, carnivory or cannibalism in the cricket’s diet). Similarly, for the pollen-feeding butterfly Heliconius cydno , in which males supply nutritious ejaculate to their mates, Boggs (1990) found an inverse relationship between female mating rate and pollen acquisition, suggesting a trade-off between mating and feeding rates. In the case of these alternate reproductive strategies, mating rate would no longer provide an accurate estimate of a female’s reproductive success, and fitness would need to be

17 estimated using some post-copulatory measure such as number of eggs laid or offspring hatched, neither of which could be determined for this study.

Stabilizing sexual selection was the prevalent form of nonlinear selection observed in the canonical analysis. Two of the three major axes under significant nonlinear selection exhibited strong stabilizing selection on the multivariate phenotype. In this study, I examined only total sexual selection. Therefore, all of the observed selection is the product of some unknown combination of intrasexual competition and mate choice (Hunt et al. 2009). It is possible that the two mechanisms of selection are reinforcing; both male choice and female- female competition select for intermediate values of the major axes, and the overall selection is likewise stabilizing (Hunt et al. 2009). Alternatively, stabilizing sexual selection on females may result from the opposition of male choice and female-female competition: certain traits may aid females to win intrasexual competitions or indicate quality, but males may select against any females who over-invest in weapons or ornaments at the expense of fecundity

(Chenoweth et al. 2006, Fitzpatrick et al. 1995). For example, Chenoweth and Blows (2005) proposed that the stabilizing selection for female cuticular hydrocarbon (CHC) expression in

Drosophila serrata results from male choice for females with adequate CHC expression to signal quality and condition, but not so much as to reduce offspring quality or number. Indeed, this potential trade-off between sexual trait expression and fecundity is thought to be one of the major factors limiting the evolution of female secondary sexual characters, accounting for the rarity of cases of female sexual ornamentation and the absence of female-specific sexual weaponry in nature (Fitzpatrick et al. 1995).

Interestingly, female Mormon crickets do not exhibit any obvious ornamentation despite the presence of the contemporary sexual selection detected in this study. Unlike male dance flies, who select signalling females in twilight swarms prior to coupling, male Mormon

18 crickets have been found to assess female fecundity directly upon being mounted, favouring heavier females with more ovarian eggs (Gwynne 1981). However, in this study I found no correlations between traits indicating body size and fecundity. These results could possibly be explained by the potential inadequacy of ovarian egg number as a measure of fecundity

(because of uncertainty as to whether the female laid eggs prior to collection) or because males assess female mass instead of size. The lack of elaboration of female secondary sexual traits despite selection could also result from variation in the strength and form of sexual selection within and across seasons. The nutrient availability of the environment likely varies across years or even within a season, thereby changing the value of the male’s nuptial gift and resulting in variation in the intensity of sexual selection on males and females. For example,

Gwynne (1984) observed that upon moving into a rare patch of nutritious seed-bearing plants, some gregarious crickets from the main band exhibited evidence of typical sex-roles.

Variation in sexual selection has been noted in other studies from the field. In their study of sexual selection in ambush bugs, Punzalan et al. (2008) found that the form of the selection on males changed from approximately stabilizing to directional over the course of the mating season, and Fiske et al. (1994) found that both the strength and direction of selection changed across years in lekking great snipes, Gallinago media . Without consistent sexual selection over time, evolution of the trait may be limited, and this may account for the observed lack of elaboration of traits found to be under contemporary sexual selection.

In conclusion, while sexual selection on females is believed to be widespread

(Bonduriansky 2001, Clutton-Brock 2009), there are few studies that have measured the intensity and form of such selection (but see LeBas et al. 2003, and Wheeler 2008). The finding of sexual selection for females with small heads and mouthparts contrasts that observed in fighting males of cricket species with more typical sex roles. The sexual

19 advantage of smaller head traits is currently not understood. Mormon crickets do not exhibit any female-specific secondary sexual traits and a better understanding of the temporal variability in sexual selection may aid in explaining this lack of trait elaboration. Finally, this study has provided a confirmation of the validity of the instantaneous “mated vs. unmated” measure of mating success.

20 References l

Amundsen, T., and E. Forsgren. 2001. Male mate choice selects for female coloration in a fish.

Proceedings of the National Academy of Sciences of the United States of America

98(23):13155-13160.

Arnold, S.J. 1994. Bateman’s principles and the measurement of sexual selection in plants and

animals. The American Naturalist 144:S126-S149.

Arnold, S.J., and D. Duvall. 1994. Animal mating systems: a synthesis based on selection

theory. The American Naturalist 143(2):317-348.

Arnqvist, G. 1992. Spatial variation in selective regimes: sexual selection in the water strider,

Gerris odontogaster . Evolution 46(4):914-929.

Bateman, A.J. 1948. Intra-sexual selection in Drosophila . Heredity 2:349-368.

Bentsen, C.L., J. Hunt, M.D. Jennions, and R. Brooks. 2006. Complex multivariate sexual

selection on male acoustic signalling in a wild population of Teleogryllus commodus .

The American Naturalist 167(4):E102-116.

Berglund, A., G. Rosenqvist, and I. Svensson. 1986. Mate choice, fecundity and sexual

dimorphism in two pipefish species (Syngnathidae). Behavioral Ecology and

Sociobiology 19:301-307.

Bertin, A., and D.J. Fairbairn. 2005. One tool, many uses: precopulatory sexual selection on

genital morphology in Aquarius remigis . Journal of Evolutionary Biology 18(4):949-

961.

Blows, M.W., and R. Brooks. 2003. Measuring nonlinear selection. The American Naturalist

162(6):815-820.

21 Boggs, C.L. 1990. A general model of the role of male-donated nutrients in female insects’

reproduction. The American Naturalist 136(5): 598-617

Boldyrev, B.T. 1915. Contributions à l’étude de la structure des spermatophores et des

particularitès de la copulation chez Locustodea et Gryllodea. Horae Societatis

Entomologicae Rossicae 6: 1-245.

Bonduriansky, R. 2001. The evolution of male mate choice in insects: a synthesis of ideas and

evidence. Biological Reviews of the Cambridge Philosophical Society 76:305-339.

Bradford, M.J., P.A. Guerette, and D.A. Roff. 1993. Testing hypotheses of adaptive variation

in cricket ovipositor length. Oecologia 93(2):263-267.

Brooks, R., J. Hunt, M.W. Blows, M.J. Smith, L.F. Bussiere, and M.D. Jennions. 2005.

Experimental evidence for multivariate stabilizing sexual selection. Evolution

59(4):871-880.

Bussière, L.F., D.T. Gwynne, and R. Brooks. 2008. Contrasting sexual selection on males and

females in a role-reversed swarming dance fly, Rhamphomyia longicauda Loew

(Diptera:Empididae). Journal of Evolutionary Biology 21:1683-1691.

Chenoweth, S.F., and M.W. Blows. 2005. Contrasting mutual sexual selection on homologous

signal traits in Drosophila serrata . The American Naturalist 165(2):281-289.

Chenoweth, S.F., P. Doughty, and H. Kokko. 2006. Can non-directional male mating

preference facilitate honest-female ornamentation? Ecology Letters 9:179-184.

Clissold, F.J. 2007. The biomechanics of chewing and plant fracture: mechanisms and

implication. Advances in Insect Physiology 34:317-372.

Clutton-Brock, T.H. 2007. Sexual selection in males and females. Science 318:1882-1885.

------2009. Sexual selection in females. Animal Behaviour 77:3-11.

22 Clutton-Brock, T.H., F.E. Guinness, and S.D. Albon. 1982. Red deer: behaviour and ecology

of two sexes. University of Chicago Press, Chicago, Illinois, USA.

Cowan, F.T. 1929. Life history, habits, and control of the Mormon cricket. Technical Bulletin

of the U.S. Department of Agriculture 161:1-28.

Cunningham, E.J.A., and T.R. Birkhead. 1998. Sex roles and sexual selection. Animal

Behaviour 56:1311-1321.

Darwin, C. 1871. The descent of man and selection in relation to sex. John Murray, London,

UK.

Dodson, G.N., G.K. Morris, and D.T. Gwynne. 1983. Mating behaviour of the primitive

Orthopteran genus Cyphoderris (Haglidae). Pages 319-336 in D.T. Gwynne and G.K.

Morris, editors. Orthopteran mating systems: sexual competition in a diverse group of

insects. Westview press, Boulder, Colorado, USA.

Downes, J.A. 1970. Feeding and mating behaviour of specialized Empidinae (Diptera):

observations on four species of Rhamphomyia in the high Arctic and a general

discussion. Canadian Entomologist 102: 769-791.

Eisner, T. S.R. Smedley, D.K. Young, M. Eisner, B. Roach, and J. Meinwald. 1996. Chemical

basis of courtship in a beetle ( Neopryrochroa flabellata ): Cantharidin as “nuptial gift”.

Proceedings of the National Academy of Science USA 93:6499-6503.

Endler, J.A. 1986. Natural selection in the wild. Princeton University Press, Princeton, New

Jersey, USA.

Fedorka, K. M., and T.A. Mousseau. 2002. Tibial spur feeding in the ground crickets: larger

males contribute larger gifts (Orthoptera: Gryllidae). Florida Entomologist 85(2):317-

323.

23 Fiske, P. J.A. Kalas, and S.A. Saether. 1994. Correlates of male mating success in the lekking

great snipe ( Gallinago media ): results from a four-year study. Behavioural Ecology

5:210-218.

Fitzpatrick, S., A. Berglund, and G. Rosenqvist. 1995. Ornaments or offspring: costs to

reproductive success restrict sexual selection processes. Biological Journal of the

Linnean Society 55:251-260.

Furrer, R., D. Nychka, and S. Sain. 2009. fields:Tools for spatial data. R package version 5.02.

http://www.image.ucar.edu/Software/Fields.

Gwynne, D.T. 1981. Sexual difference theory: Mormon crickets show role reversal in mate

choice. Science 213: 779-780.

------1984. Sexual selection and sexual differences in Mormon crickets (Othoptera:

Tettigoniidae, Anabrus simplex ). Evolution 35(5): 1011-1022.

------1985. Role-reversal in katydids: habitat influences reproductive behaviour

(Orthoptera: Tettigoniidae: Metaballus sp.). Behavioral Ecology and Sociobiology

16:355-361.

------1990. Testing parental investment and the control of sexual selection in katydids: the

operational sex ratio. The American Naturalist 136:474-484.

------1993. Food quality controls sexual selection in Mormon crickets by altering male

mating investment. Ecology 74(5): 1406-1413.

------2001. Katydids and bush-crickets: reproductive behaviour and evolution of the

Tettigoniidae. Cornell University Press. Ithaca, New York, USA.

Gwynne, D.T., and W.J. Bailey. 1999. Female-female competition in katydids: sexual

selection for increased sensitivity to a male signal? Evolution 53(2):546-551.

24 Gwynne, D.T., and L.W. Simmons. 1990. Experimental reversal of courtship roles in an

insect. Nature 346:172-174.

Hall, M.D., L.F. Bussière, J. Hunt, and R. Brooks. 2008. Experimental evidence that sexual

conflict influences the opportunity, form and intensity of sexual selection. Evolution

62(9):2305-2315.

Hunt, J., C.J. Brueker, J.A. Sadowski, and A.J. Moore. 2009. Male-male competition, female

mate choice and their interaction: determining total sexual selection. Journal of

Evolutionary Biology 22:13-26.

Jones, A.G., G. Rosenqvist, A. Berglund, S.J. Arnold, and J.C. Avise. 2000. The Bateman

gradient and the cause of sexual selection in a sex-role-reversed pipefish. Proceedings

of the Royal Society of London Biology 267:677-680.

Judge, K.A., and V.L. Bonanno. 2008. Male weaponry in a fighting cricket. PLoS ONE

3(12):e3980.

Kelly, C.D. 2006. Fighting for harems: assessment strategies during male-male contests in the

sexually dimorphic Wellington tree weta. Animal Behaviour 72:727-736.

Kingsolver, J.G., H.E. Hoekstra, J.M. Hoekstra, D. Berrigan, S.N. Vignieri, C.E. Hill, A.

Hoang, P. Gibert and P. Beerdi. 2001. The strength of phenotypic selection in natural

populations. The American Naturalist 157:245-261.

Lande, R., and S.J. Arnold. 1983. The measurement of selection on correlated characters.

Evolution 37(6): 1210-1226.

LeBas, N.R., L.R. Hockham, and M.G. Ritchie. 2003. Nonlinear and correlational sexual

selection on ‘honest’ female ornamentation. Proceedings of the Royal Society of

London Biology 270:2159-2165.

25 LeBas, N.R., L.R. Hockham, and M.G. Ritchie. 2004. Sexual selection in the gift-giving dance

fly, Rhamphomyia sulcata , favors small males carrying small gifts. Evolution

58(8):1763-1772.

Legendre, P., and L. Legendre. 1998. Numerical Ecology. 2 nd English edition. Elsevier

Science BV. Amsterdam, Netherlands.

Lessells, C.M., and P.T. Boag. 1987. Unrepeatable repeatabilities: a common mistake. Auk

104:116-120.

Lorch, P.D. 2005. Using upper limits of “Bateman Gradients” to estimate the opportunity for

sexual selection. Integrative and Comparative Biology 45:924-930.

MacVean, C. M. (1987). Ecology and management of the Mormon cricket, Anabrus simplex

Haldeman. Pages 116-136 in J. L. Capinera, editor. Integrated pest management on

rangeland: a shortgrass prairie perspective. Westview Press. Boulder, Colorado, USA.

Madsen, T., and R. Shine. 1994. Components of lifetime reproductive success in adders,

Vipera berus . Journal of Animal Ecology 63:561-568.

Masaki, S. 1979. Adaptation and species status in the lawn ground cricket. III. Ovipositor

length. Oecologia 43(2):207-219.

Mason, L.G. 1964. Stabilizing selection for mating fitness in natural populations of Tetraopes .

Evolution 18:492-497.

McElligott, A.G., and T.J. Hayden. 2000. Lifetimes mating success, sexual selection and life

history of fallow bucks ( Dama dama ). Behavioral Ecology and Sociobiology 48:203-

210.

McLain, D. K. 1981. Interspecific interference competition and mate choice in the soldier

beetle, Chauliognathus pennsylvanicus . Behavioral Ecology and Sociobiology 9: 65-

66.

26 Neville, A.C. 1963. Daily growth layers for determining the age of grasshopper populations.

OIKOS 14:1-8.

Owens, I.P.F. 2006. Where is behavioural ecology going? Trends in Ecology and Evolution

21(7), 356-361.

Philips, P.C., and S.J. Arnold. 1989. Visualizing multivariate selection. Evolution 43:1209-

1222.

Punzalan, D., F.H. Rodd, and L. Rowe. 2008. Contemporary sexual selection on sexually

dimorphic traits in the ambush bug Phymata americana . Behavioral Ecology

19(4):860-870.

Punzalan, D., F.H. Rodd, and L. Rowe. 2009. Temporally variable multivariate sexual

selection on sexually dimorphic traits in a wild insect population. In review, The

American Naturalist.

Purse, B.V., and D.J. Thompson. 2005. Lifetime mating success in a marginal population of a

damselfly, Coenagrion mercuriale . Animal Behaviour 69:1303-1315.

R Development Core Team. 2008. R: A language and environment for statistical computing. R

Foundation for Statistical Computing,Vienna, Austria. http://www.R-project.org.

Rice, W.R. 1989. Analyzing tables of statistical tests. Evolution 43(1):223-225.

Sadowski, J. A., A. J. Moore, and E.D. Brodie. (1999). The evolution of empty nuptial gifts in

a dance fly, Empis snoddyi (Diptera : Empididae): bigger isn't always better.

Behavioral Ecology and Sociobiology 45: 161-166.

Shine, R. 1989. Ecological causes for the evolution of sexual dimorphism: a review of the

evidence. The Quarterly Review of Biology 64:419-461.

Shuster S.M., and M.J. Wade. 2003. Mating Systems and Strategies. Princeton University

Press. Princeton, New Jersey, USA.

27 Simmons, L.W., and W.J. Bailey. 1990. Resource influenced sex roles of a zaprochiline

katydid (Orthoptera: Tettigoniidae). Evolution 44:1853-1868.

Simpson, S.J., G.A. Sword, P.D. Lorch, and I.D. Couzin. 2006. Cannibal crickets on a forced

march for protein and salt. Proceedings of the National Academy of Science 103(11):

4152-4156.

Stinchcombe, J.R., A.F. Agrawal, P.A. Hohenlohe, S.J. Arnold, and M.W. Blows. 2008.

Estimating nonlinear selection gradients using quadratic regression coefficients: double

or nothing? Evolution 62(9):2435-2440.

Trivers, R.L. 1972. Parental investment and sexual selection. Pages 136-179 in B. Campbell,

editor. Sexual selection and the descent of man, 1871-1971. Heinemann, London, UK.

Vahed. K. 2003.Structure of spermatodoses in shield-back bushcrickets (Tettigoniidae,

Tettigoniinae). Journal of Morphology 257:42-52.

Wheeler, J. 2008. Sexual selection for female ornaments and their relationship to fecundity in

the dance fly Rhamphomyia longicauda (Diptera: Empididae). Master’s thesis.

Tables 1

Table 1.1 Univariate(s), linear ( β) and nonlinear ( γ) selection on female Mormon crickets using cumulative mating rate (number of

spermatodoses/day since sexual maturity) as the fitness measure. Only females older than 5 days post-adult moult are included. Quadratic

terms are doubled. Bolded terms represent selection that is significant at p<0.05.

Morphological Femur Ovipositor Maxilla Head Mandible Maxilla Abdominal Pronotum Spine s p β p traits length length span width length length plate width length length Femur length -0.071 0.095 -0.012 0.827 -0.068

Ovipositor 0.035 0.411 0.081 0.071 0.029 0.023 Length 29

Maxilla span -0.015 0.755 0.041 0.416 -0.026 -0.010 0.090

Head width -0.136 0.003 -0.173 0.014 -0.052 -0.036 0.061 0.023

Mandible -0.158 0.000 -0.168 0.004 0.167 -0.103 -0.066 0.147 -0.208 length Maxilla length -0.028 0.510 0.091 0.111 -0.036 0.000 0.067 0.075 0.027 -0.081

Abdominal 0.014 0.778 0.048 0.284 0.101 -0.016 0.032 -0.129 -0.047 -0.038 -0.082 plate width Pronotum -0.046 0.270 0.075 0.210 0.028 0.052 -0.046 -0.028 -0.015 0.000 0.040 -0.108 length Spine length -0.065 0.125 -0.009 0.851 -0.063 0.018 -0.006 -0.044 0.033 -0.085 -0.009 0.060 0.073

Table 1.2 M matrix of eigenvectors for the canonical analysis of the γ matrix given in Table 1.1. The linear ( θ) and quadratic ( λ)

gradients of selection for each axis are given. Fitness is measured as mating rate (number of spermatodoses/days since sexual maturity).

Only females older than 5 days post-adult moult are included. Bolded terms represent selection that is significant at p<0.05.

M Selection

Femur Ovipositor Maxilla Head Mandible Maxilla Abdominal Pronotum Spine θ p λ p length length span width length length plate width length length m1 0.145 0.261 -0.328 -0.632 -0.194 -0.343 0.320 0.228 0.308 0.146 0.011 0.264 0.001 m2 0.362 -0.291 -0.711 0.187 0.466 -0.115 -0.096 0.032 -0.055 -0.180 0.000 0.143 0.061 m3 0.441 -0.015 0.139 -0.232 -0.021 0.118 0.387 -0.077 -0.749 0.073 0.212 0.123 0.169 m4 0.188 -0.662 0.425 -0.085 0.199 -0.184 0.340 -0.081 0.378 -0.067 0.173 0.052 0.262

30 m5 0.523 0.522 0.302 0.314 0.301 0.085 0.071 0.299 0.267 -0.026 0.395 0.003 0.873 m6 -0.163 -0.256 -0.114 -0.055 -0.030 0.630 0.198 0.673 0.038 0.109 0.078 -0.109 0.134 m7 -0.183 -0.078 0.059 0.421 -0.170 -0.620 0.176 0.508 -0.279 -0.053 0.361 -0.169 0.163 m8 0.216 -0.168 0.255 -0.393 0.046 -0.147 -0.730 0.361 -0.139 0.035 0.530 -0.222 0.006 m9 0.487 -0.193 -0.118 0.268 -0.763 0.104 -0.125 -0.057 0.165 0.053 0.390 -0.422 0.001

Table 1.3 P matrix of the Spearman’s rank correlations of unstandardized morphological and fecundity traits. Sequential Bonferroni

adjustment was used with α=0.05. Significant correlation terms are bolded.

Femur Ovipositor Maxilla Head Mandible Maxilla Abdominal Pronotum Spine Egg Egg length length span width length length plate width length length number size Femur length

Ovipositor 0.388 length Maxilla span 0.364 0.246

Head width 0.646 0.347 0.496

Mandible 0.515 0.280 0.317 0.591 length Maxilla 0.493 0.269 0.344 0.578 0.538

3 length 1 Abdominal 0.150 0.070 0.118 0.196 0.146 0.109 plate width Pronotum 0.634 0.325 0.358 0.625 0.489 0.440 0.159 length Spine length 0.443 0.249 0.296 0.452 0.388 0.292 0.104 0.435

Egg number -0.025 -0.012 0.077 -0.021 0.010 0.056 0.019 0.005 -0.024

Egg size -0.058 0.036 -0.045 -0.089 -0.126 -0.034 0.028 -0.073 -0.053 0.550

Table 1.4 Univariate(s), linear ( β) and nonlinear ( γ) selection on females when both the fitness is measured as the instantaneous mating

success in the field (mated females have a fitness of 1, unmated females have a fitness of 0). Only females older than 5 days post-adult

moult are included. Quadratic terms are doubled. Bolded terms represent selection that is significant at p<0.05.

Morphological Femur Ovipositor Maxilla Head Mandible Maxilla Abdominal Pronotum Spine s p β p trait length length span width length length plate width length length Femur length 0.030 0.398 0.005 0.919 -0.159

Ovipositor 0.105 0.005 0.108 0.002 0.068 0.011 Length Maxilla span 0.029 0.413 0.026 0.509 0.014 0.082 0.020

32 Head width -0.021 0.578 -0.121 0.021 -0.053 -0.036 -0.035 -0.061

Mandible -0.011 0.756 -0.046 0.300 0.184 -0.086 0.011 0.069 -0.114 length Maxilla length 0.055 0.116 0.081 0.063 0.036 -0.013 0.060 -0.015 0.026 -0.093

Abdominal -0.013 0.702 -0.006 0.851 0.084 -0.013 -0.015 -0.002 -0.047 -0.049 -0.031 plate width Pronotum 0.045 0.212 0.082 0.087 -0.035 0.005 -0.052 0.098 -0.022 -0.008 0.028 0.007 length Spine length -0.015 0.664 -0.033 0.411 0.026 -0.007 0.045 -0.012 0.072 -0.123 -0.030 -0.020 0.000

Figures f Spermatodose Number Spermatodose 0 2 4 6 8

0 5 10 15

Age

Figure 1.1 Relationship between spermatodose number and female age (days since eclosion to adulthood). The number of spermatodoses equals the cumulative number of matings. The data points were jittered to provide an indication of the frequency within each age category.

2.0

R 1.5

1.0

-6 -4 h -2 dt wi -2 0 d ea Man 0 2 H dible 2 length

Figure 1.2 Thin-plate spline “perspective” visualization of the selection surface for female mandible length and head width when cumulative mating rate is used as the fitness measure.

0 .6

0 .8 1.4

1 1.2

1 1.0 .2

Mandiblelength 0.8

1 .4

0.6 -3 -2 -1 0 1 2

-6 -4 -2 0 2

Head width

Figure 1.3 Contour-map visualization showing the points distribution for the selection on female mandible length and head width when cumulative mating rate is used as the fitness measure. Fitness increases with colour “warmth”, with dark blue representing the lowest fitness and dark red representing the highest fitness. The predicted fitness of a given trait combination is given on the black contour lines.

2.0

a 1.5

t e 1.0

0.5 3 -1 2 1 0 0 1 M -1 2 9 1 -2 M -3 3

Figure 1.4 Thin-plate spline “perspective” visualization of the selection surface for the first and ninth major canonical (m1 and m9) when cumulative mating rate is used as the fitness measure.

2.0

5 . 1

1.5 M9

1 1.0

-1 0 1 2 3

0.5

-3 -2 -1 0 1 2 3

Figure 1.5 Contour-map visualization showing the points distribution for the selection on the first and ninth major canonical axes (m1 and m9) when cumulative mating rate is used as the fitness measure. Fitness increases with colour “warmth”, with dark blue representing the lowest fitness and dark red representing the highest fitness. The predicted fitness of a given trait combination is given on the black contour lines.

1.0

0.8 R

0.6

-1 0 1 9 2 2 M 0 3 M8 -2

Figure 1.6 Thin-plate spline “perspective” visualization of the selection surface for the eighth and ninth major canonical axes (m8 and m9) when cumulative mating rate is used as the fitness measure.

1.1

0.8 0.6 1.0

0.9

M9 0.8

0.7

1 0.6 -1 0 1 2 3

1 0. 0.5 6

-2 -1 0 1 2 3

Figure 1.7 Contour-map visualization showing the points distribution for the selection on the eighth and ninth major canonical axes (m8 and m9) when cumulative mating rate is used as the fitness measure. Fitness increases with colour “warmth”, with dark blue representing the lowest fitness and dark red representing the highest fitness. The predicted fitness of a given trait combination is given on the black contour lines.

Number of Females of Number 0 50 100 150 200 250

0 1 2 3 4 5 6 7 8 9

Spermatodose Number

Figure 1.8 Frequency of mated (light bars) and unmated (dark bars) females for mating number, as determined by the number of spermatodoses found within the female’s spermatheca. A small number of mated females had no spermatodoses within their spermathecae, and they were classified as having one mating as this failure of spermatodose formation is assumed to be uncommon.

Appendix A: Supplementary Figures f

Figure A.1 Photograph showing the spine on the foreleg coxa of a female Mormon cricket.

The length of the spine was measured as the length of the orange line.

Figure A.2 Photograph showing the last abdominal sternite of a female Mormon cricket. The width of the plate was measured as the length of the yellow line.