Behavioral Ecology Advance Access originally published online on December 3, 2007
Behavioral Ecology 2008 19(1):184-192; doi:10.1093/beheco/arm121
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Identifying a causal agent of sexual selection on weaponry in an insect
Department of Biology, University of Toronto at Mississauga, Mississauga, Ontario, Canada L5L 1C6 and School of Botany and Zoology, Australian National University, Canberra, ACT, Australia 0200
Address correspondence to C.D. Kelly, who is now at Department of Psychology, Dalhousie University, Halifax, NS Canada B3H 4J1. E-mail: clintkelly{at}dal.ca.
Received 31 July 2007; revised 13 October 2007; accepted 22 October 2007.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMethodsResultsDiscussionFundingREFERENCES |
|---|
In many animal species, males do not seek females directly but instead locate and defend sites that contain spatially or temporally limited resources essential to female survival and reproduction. Resident males that successfully repel conspecific rivals can mate with females attracted to these resources. In theory, increasing resource value increases harem size and thus increases the opportunity (Imates) for and strength of sexual selection on traits crucial to male resource-holding potential and mating success. I experimentally tested this hypothesis in the field using the Wellington tree weta, Hemideina crassidens (Orthoptera: Tettigonioidea: Anostostomatidae), a sexually dimorphic insect in which males use their enlarged mandibles as weapons in male–male contests over access to females sheltering in tree cavities (galleries). By manipulating gallery size, I showed that, compared with smaller galleries, larger galleries housed larger harems. Variation in gallery size was an important determinant of Imates, but contrary to expectation, greater opportunity existed in small galleries compared with large galleries. As predicted, male weapon size was under stronger directional selection in large galleries because the fitness benefits were greater under these conditions compared with small galleries. My results help explain the positive association between average weapon size and average gallery size observed within and among tree weta populations in New Zealand.
Key words: conservation, Hemideina, longitudinal study, resource defense, selection gradient, sexual dimorphism.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMethodsResultsDiscussionFundingREFERENCES |
|---|
The opportunity for selection identifies the upper limit to the rate of evolutionary change in the mean phenotype in a population by selection (Crow 1958
; Wade 1979; Shuster and Wade 2003). This upper bound is directly related to variance in relative fitness; greater variance in fitness in a population translates into greater opportunity for phenotypic change (Shuster and Wade 2003). When there are large differences between the sexes in the opportunity for selection, sexual dimorphism can arise (Wade 1979; Andersson 1994; Shuster and Wade 2003). For instance, in species with conventional sex roles, males typically possess more elaborate phenotypic traits than females because males have greater variance in fitness (Darwin 1871; Emlen and Oring 1977; Wade 1979; Andersson 1994; Shuster and Wade 2003; Hauber and Lacey 2005).
A high opportunity for sexual selection means that some males are more successful at acquiring mates than others (Shuster and Wade 2003), but high opportunities can arise by chance (Sutherland 1985). Successful males possess trait values of particular phenotypic characters that are either preferred by females or permit males to outcompete rivals such as in the successful defense of resources required by females for breeding (Darwin 1871; Andersson 1994; but see Sutherland 1985). Hence, although fitness variance is necessary for evolutionary change to occur, there must also be a relationship between fitness and phenotypic characters (Conner 1988). Selection gradients are used to measure the link between fitness and a particular trait, independent of other measured correlated traits (Lande and Arnold 1983; Arnold and Wade 1984b; Brodie et al. 1995). The magnitude of the link between fitness and phenotype is referred to as the strength of selection and has been quantified for a variety of traits in a number of taxa (reviewed in Kingsolver et al. 2001; see also Brooks and Endler 2001; Kwiatkowski and Sullivan 2002; LeBas et al. 2003; Mills et al. 2007).
Identifying the causal agents of selection and their interaction with the phenotype is just as important as determining how selection operates on the phenotype (Endler 1986; Wade and Kalisz 1990). However, selection gradients do not provide any information regarding why an association between fitness and trait value exists (i.e., covariation is not causation) (Mitchell-Olds and Shaw 1987; Mitchell-Olds and Shaw 1990; Wade and Kalisz 1990). To identify the agents of selection, experimental manipulations of the environment are required, which, in turn, require knowledge of a study organism's ecology (Wade and Kalisz 1990).
Examples of some of the most intense selection in nature include sexual selection that generates sexual dimorphism (Shuster and Wade 2003). One group of animals that exhibit extreme sexual dimorphism is New Zealand tree weta (Orthoptera: Tettigonioidea: Anostostomatidae)—large (ca. 65-mm body length), flightless, and nocturnal insects (Gibbs 2001). Of the 7 species of Hemideina tree weta (Gibbs 2001), the Wellington tree weta (Hemideina crassidens) appears to be the most sexually dimorphic (Field and Deans 2001) with males possessing highly exaggerated mandibles that are positively allometric (Kelly 2005). Males use their mandibles, as in other Hemideina species (Field and Sandlant 1983; Field 2001), as weapons in fights for control of females residing in tree cavities (hereafter galleries) (Kelly 2006a). Male weaponry varies considerably in length among individuals (Kelly 2005) apparently due to precocial male maturation: males can mature at either the 8th, 9th, or 10th instar, whereas females mature at the 10th instar only (Spencer 1995). Both sexes are polygamous, but there is greater variance in male mating success (Kelly CD, unpublished data).
Hemideina crassidens is a harem-defending polygynous insect in which males guard females rather than galleries per se; field evidence shows that male residency in a gallery is positively correlated with harem size (Kelly 2006c, 2006d). Galleries are used year-round as diurnal refugia and are not used as oviposition or feeding sites. A single adult male typically resides in a gallery with 1 to several females (Moller 1985; Field and Sandlant 2001; Kelly 2005), and phenotypic correlations in the wild have shown a strong positive relationship between male head length (weapon size) and how many females reside with him (harem size) (Kelly 2005). Moreover, a laboratory study showed that larger galleries (higher quality resources) house larger harems, and males with bigger weaponry control these galleries more often than those with smaller weapons (Kelly 2006d). Despite larger galleries housing larger harems, I note that galleries in nature, except for the very smallest ones (i.e., single and double occupant), typically do not reach maximum occupancy (Kelly 2005). I also note that although females prefer to settle in unoccupied galleries, evidence suggests that female settlement in galleries does not exhibit an ideal free distribution because many galleries in a patch can be empty whereas others house several individuals (Kelly 2006b, 2006c). Mating typically occurs either in the gallery or near its entrance (Field and Jarman 2001) and male–female pairs will copulate both at night and during the day, with the latter copulations always occurring within the gallery, space permitting (Field and Jarman 2001).
Gallery size limits harem size and should thus be an important ecological factor influencing the strength of sexual selection operating on male weaponry (Field and Sandlant 1983; Field and Sandlant 2001; Kelly 2006d). Field observations at several sites in New Zealand suggest that males inhabiting forests (e.g., broad-leaved forest) with larger galleries (>3 adult weta), on average, tend to have access to larger harems and exhibit larger weaponry compared with males living in forests (e.g., manuka, Leptospermum scoparium) offering smaller galleries (<3 adult tree weta), on average (Sandlant 1981; Field and Sandlant 1983; Field and Sandlant 2001; see also Discussion). The differences in gallery size between the 2 forest types is a consequence of the maximum diameter of the tree species comprising each forest (Field and Sandlant 2001; Kelly CD, personal observation): manuka is a small-diameter (<10-cm diameter at breast height; Dawson and Lucas 2000) tree species and thus accommodates small galleries, whereas broad-leaved species (e.g., kohekohe, Dysoxylum spectabile) have much larger diameters (>100-cm d.b.h; Dawson and Lucas 2000) and can thus accommodate larger galleries.
Here, I experimentally manipulated a putatively important ecological factor—gallery size—influencing the opportunity for and strength of sexual selection on male weaponry in the Wellington tree weta. More specifically, I use longitudinal field data to test the hypothesis that increasing gallery size increases the potential variation in harem size and thus increases sexual selection on traits crucial in male–male competition for female harems.
In this paper, I test the following predictions.
- Larger galleries exhibit greater opportunities for sexual selection (Imates) because these resources can host larger aggregations of females (m*).
- Males are under directional selection with individuals possessing larger weaponry having greater resource-holding potential and thus mating success.
- Because larger galleries are expected to house larger harems and provide greater reproductive opportunities, male traits (e.g., mandibular weaponry and body size) used in taking control of galleries will be under stronger selection in larger galleries. This prediction assumes that each gallery within a treatment will not be occupied to maximum with females (i.e., assumes variance in harem size among equal-sized galleries) as observed in Kelly (2005).
- If predictions (1) and (3) are met, then there should be a positive correlation between the opportunity for sexual selection (Imates) and the strength of selection on male weaponry (partial regression gradients) (McLain 1992; Bjork and Pitnick 2006).
|
Methods |
|---|
|
TOPABSTRACTINTRODUCTIONMethodsResultsDiscussionFundingREFERENCES |
|---|
Study sites
This study was conducted during April–May 2003 and 2006–2007 and March–April 2004 on Te Hoiere/Maud Island, New Zealand (41°02'S, 173°53'E), a 309-ha scientific reserve free of alien predators (e.g., rodents [Mus and Rattus spp.] and stoats [Mustela erminea]). Of the known predators of adult tree weta, only the native morepork owl, Ninox novaeseelandiae, is present on Maud Island (see Field and Glasgow 2001
for a list).
In 2002, I installed artificial galleries (Figure 1) approximately 1.5 m above the ground on mature trees at 2 sites approximately 200 m apart within a regenerating forest (ca. 40 years old) dominated by mahoe (Melicytus ramiflorus; up to 30- to 40-cm diameter) and makomako (Aristotelia serrata; up to 30-cm diameter). The natural galleries at both experimental sites were intermediate in average volume to those common in early successional (smaller galleries) or mature (larger galleries) forests. Both sites were similar in area, topography, and flora, and each was inhabited previous to this study by males representing the full range of weapon sizes known for this population (Kelly 2005, 2006c). Tree weta readily occupy artificial galleries (Trewick and Morgan-Richards 2000; Kelly 2005, 2006c; Bowie et al. 2006). Artificial galleries were installed over natural galleries that were destructively excavated and emptied of all tree weta. This field experiment comprised a treatment with 2 levels: small and large galleries. At Site A (ca. 24 x 4 m), 19 small galleries (cavity volume in cubic millimeters) were installed, whereas at Site B (ca. 22 x 3 m), 13 large galleries (cavity volume in cubic millimeters), with twice the volume of the small galleries, were installed (Figure 1). This gave approximately 1 gallery/5 m2 at each site. In 2003 and 2004, small galleries were in place at Site A and large galleries at Site B. In 2006, the galleries were switched between the sites to control for any differences between the sites.
|
Gallery census and trait measurement
I opened each gallery and censused all occupants every third morning in 2003 and 2007 and daily in 2004 and 2006. Census periods were considered independent because gallery composition can change considerably between days (Kelly 2006b
). Handling tree weta does not affect their tenure at a gallery (Kelly 2006c). I counted the number of adult males and adult females in each gallery and noted their identity. If the occupant was unmarked, I measured its head length and width and right and left hind tibia and femur lengths using digital calipers (Mitutoyo Digimatic, Mitutoyo Canada Inc., Toronto, Ontario, Canada). Head length is an excellent predictor of weapon size (i.e., mandible length; r = 0.99, Kelly 2005). To permit individual identification, I glued a uniquely numbered and colored bee tag (H. Thorne Ltd, Market Rason, UK) to each individual's pronotum. After being identified and/or measured and marked, all individuals were then immediately replaced into their gallery. No tree weta was observed at both sites in either year, and no males were observed in 2 consecutive sample years (H. crassidens has a 2-year life cycle from egg to adult, but adults do not survive more than 1 breeding season).
Opportunity for sexual selection
Of the many methods available to measure the potential for sexual selection (Fairbairn and Wilby 2001; Shuster and Wade 2003; Mills et al. 2007), I chose to use Imates approach of Shuster and Wade (2003) for 3 reasons. First, unlike the breeding sex ratio (Arnold and Duvall 1994) or Q (Ahnesjo et al. 2001), Imates takes into account that much of the variance in male reproductive success lies between mating and nonmating males. Second, although calculating Bateman gradients (relationship between fitness or fertility and number of mates) is often regarded as a superior method for determining the strength of sexual selection (Arnold 1994; Gwynne and Kelly 2004; Lorch 2005; Mills et al. 2007; but see Wade and Shuster 2005), it requires reliable paternity assignment; I was unable to hatch a sufficient number of eggs to make this approach worthwhile. Third, the Imates approach directly links the spatial distribution of females with the opportunity for sexual selection (see Emlen and Oring 1977).
I calculated female mean crowding, m*, and the opportunity for sexual selection, Imates, using the statistical framework of Shuster and Wade (2003). With this approach, only potential breeding aggregations, defined as only galleries that comprise 1 or more females with a single adult male, are considered for the analysis. The degree of female clumping is based on Lloyd's (1967) ecological measure of density-dependent competition (see Shuster and Wade 2003). Calculating mean crowding first requires determining the average density of females per breeding aggregation, m, using
 | (1) |
 | (2) |
).
The opportunity for sexual selection was calculated using
 | (3) |
). With this approach, Imates and m* are calculated for each census day (n) in each treatment, and these values are then used to calculate a mean to test for treatment differences. I note that each n sample is 1 census day and is based on that day's occupation of either 19 (small gallery treatment) or 13 (large gallery treatment) galleries. For example, in 2004, there were 25 days in which censuses provided appropriate data (i.e., there were both successful and unsuccessful males present in galleries). I determined whether response variables differed between years (random factor) and treatments (fixed factor) by using a general linear model (GLM).
Selection analysis of male weaponry
Fitness estimate for male H. crassidens
Male mating success was estimated as the number of adult females residing with an adult male in a gallery (harem size). I tested the assumption that male harem size predicts mating success by allowing wild-caught males (n = 33) access to harems varying in number of adult females. In the late afternoon on the day of a trial, I placed in a cage (plastic 5-L pail with perforated lid) a single adult male and an artificial gallery (identical to the small galleries installed on trees in experimental sites, see above) housing 1, 2, or 3 adult females, the range typically encountered by males in nature (Kelly 2005, 2006c). Harem size for each male was chosen at random. Each trial lasted for 48 h, the average duration of time a male resides in a gallery (Kelly 2006c), and each female's mating status was checked daily at approximately 18:00, 21:00, 24:00, 05:00, and 12:00 h on each day. Mating status was assessed by lifting the female's subgenital plate and looking for a spermatophore; spermatophores are fully absorbed by females after approximately 5 h (Kelly CD, personal observation), and so my 3-h observation window provided reasonable opportunity to observe a mating. If a female was mated at night, I removed her from the cage because males typically evict mated females from the gallery; however, they do not evict females if mated inside the gallery during the day, and so in this case females were put back into the gallery (Kelly CD, unpublished data). All cages were supplied daily with fresh apple ad libitum.
Calculation of gradients
No trait differed significantly in mean size between years within any treatment (Table 1), and all morphological traits (head length, head width, tibia length, and femur length) within each year in each treatment were highly correlated (Kelly 2005). Some multiple regressions using the 4 morphological traits showed multicollinearity (tolerance < 0.001), thus violating assumptions of multiple regression (Sokal and Rohlf 1995). I estimated selection gradients for a reduced set of orthogonal variables extracted from a principal components analysis (PCA) in which the component scores are uncorrelated with one another even if composed of variables that are correlated. The first 2 components were extracted from the phenotypic covariance matrix for all males in the cohort using varimax rotation. Multiple regression of mating success on the rotated factor loadings estimated the partial gradients of these composite morphological variables. Because selection acts within a generation (Endler 1986), I generated principal components for each of the treatments by each sample year.
|
I included year as an independent variable along with head and body size in the regression analysis by treatment. There was no significant effect of year on harem size in either treatment (small: F3,177 = 1.58, P = 0.196, year term, P = 0.273; large: F3,141 = 3.53, P = 0.017, year term, P = 0.054); therefore, I pooled data within each treatment and performed the regressions with the year term removed.
I estimated linear (βii) selection gradients as the partial regression coefficients from multiple regressions with mating success as the dependent variable and 2 PCA factors (head size and body size) as the predictors. To maintain statistical independence, I calculated mean harem success for males observed more than once. Because studentized residuals from regressions were nonnormal and variances were heterogeneous, parametric tests of significance were not appropriate (Mitchell-Olds and Shaw 1987). Therefore, I tested the significance of partial regression coefficients using 9999 permutation tests (Legendre 2002). I note that when independent variables are principal components, the selection differentials (S) calculated by using univariate regressions are identical to partial regression coefficients (β), and hence, I report only the latter (see also Schluter and Smith 1986). Selection gradients were visualized using nonparametric cubic splines (Schluter 1988).
All analyses of variance (ANOVAs) were conducted using the GLM and post hoc tests conducted using Ryan's Q stepwise multiple comparisons procedure. Statistical power and sample size calculations were performed after Hersch and Phillips (2004). Because the statistical power is low for most selection analyses (Kingsolver et al. 2001) and thus gradients tend to be statistically nonsignificant at the 0.05 
level, I followed Fairbairn and Preziosi (1996) and also denote (but do not discuss) selection gradients significant at the 0.10 level. Unless otherwise noted, means are presented ± 1 standard error (SE). All analyses were conducted using SPSS 11.0 for Mac OS X.
|
Results |
|---|
|
TOPABSTRACTINTRODUCTIONMethodsResultsDiscussionFundingREFERENCES |
|---|
Opportunity for sexual selection
Experimental manipulation of gallery size changed the potential mating success of resident males as the mean harem size of successful males was significantly greater in large galleries (F1,118 = 45.23, P < 0.0001; Figure 2A). Mean harem size did neither differ between years (F3,118 = 1.27, P = 0.29) nor was the interaction between treatment and year significant (F3,118 = 0.13, P = 0.95).
|
In line with my first prediction, mean crowding of females (m*) was, on average, significantly greater at large galleries compared with small galleries (F1,118 = 37.58, P = 0.001). There was no effect of sample year on mean crowding (F3,118 = 5.21, P = 0.10), and the interaction term was nonsignificant (gallery size x year, F3,118 = 0.21, P = 0.89).
Opposite to prediction, Imates was significantly greater for small galleries than large galleries (F1,118 = 43.71, P = 0.002; Figure 2B). Their effect of year (F3,118 = 9.05, P = 0.054) and the interaction term (gallery size x year, F2,118 = 0.51, P = 0.68) were nonsignificant. Recall that Imates (Equation 3) is comprised of 2 components—the opportunity for selection among successful males (1 – p0) and the opportunity for selection between successful (1 – p0) and unsuccessful males (p0). The difference in values for Imates observed here was mainly influenced by a change in the proportion of unsuccessful males rather than an increase in harem size among successful males. This component of selection (between mated and unmated males) was significantly stronger in small galleries compared with large ones (F1,118 = 32.41, P = 0.006). There was no effect of year (F3,118 = 4.58, P = 0.12), and the interaction was nonsignificant (gallery size x year, F3,118 = 0.876, P = 0.46).
The higher mean proportion of unsuccessful males in small galleries was not due to a lack of females in this treatment as the mean adult sex ratio (males:females) was significantly female biased in each year (1-sample t-test testing expectation that sex ratio differed from unity, 2003: t18 = 5.02; 2004: t24 = 7.45; 2006: t5 = 5.81; 2007: t11 = 4.52, all P < 0.002).
Harem size predicts male mating success
Male mating success increased significantly with harem size (1-way ANOVA, F2,42 = 7.79, P = 0.001). I did not detect a significant difference between males placed with either 2 or 3 females (Ryan's Q, P = 0.07), but males residing with a single female had significantly lower mating success than males with harems (Ryan's Q, P < 0.05) (Figure 3).
|
Morphological traits and fitness components
There was no effect of treatment or sample year on the size of the 4 male morphological traits (Table 1). The males observed in this study exhibited a trimodal head length frequency distribution identical to that reported in other years in this study population (Kelly 2005
).
Univariate linear selection gradients
The first component (PC1) from a PCA quantified head size, and the second component (PC2) quantified body size in 6 of 8 cases (Table 2). The overall multiple regression was significant for the large gallery treatment (r2 = 0.045, F2,142 = 3.34, P = 0.038) but not for the small gallery treatment (r2 = 0.019, F2,178 = 1.76, P = 0.175). Neither univariate nor bivariate nonlinear selection gradients were significant and are not reported here.
|
Male traits important to male mating success
My prediction that male weapon size would be favored in both treatments was only partially supported: larger male weapon size was significantly selected in the large gallery treatment only (Table 3, Figure 4A,C). There was a trend for males with smaller body size to have greater mating success in small galleries (Table 3, Figure 4B,D).
|
|
The relationship between opportunity and strength of selection
I tested the relationship between opportunity and strength by correlating the value for Imates with the harem size selection gradient calculated for each year. Contrary to prediction 4, the opportunity for sexual selection (Imates) was not related to the strength of selection (magnitude of partial regression coefficient) on body size (Pearson correlation, r = –0.59, P = 0.12, n = 8) but showed a negative trend with regard to head size (r = –0.70, P = 0.053, n = 8).
|
Discussion |
|---|
|
TOPABSTRACTINTRODUCTIONMethodsResultsDiscussionFundingREFERENCES |
|---|
My study supports the hypothesis that an ecological factor, gallery size, significantly affects sexual selection on male tree weta. Variation in gallery size was an important determinant of the opportunity of sexual selection in tree weta; however, contrary to expectation, greater opportunity existed in small galleries and not in large galleries. Also in line with prediction, significantly larger weaponry was favored when gallery size permits larger harems to form.
The experimental results reported here mirror patterns observed in the field on Maud Island and at several sites across New Zealand. For example, on Maud Island, males inhabiting a manuka (L. scoparium) forest patch (ca. 50 ha), which has only small galleries (<3 adult tree weta), possess significantly smaller weaponry (mean ± SE: 17.93 ± 0.89 mm, n = 16) than males (25.41 ± 0.84 mm, n = 16 males randomly selected from Kelly [2005]; t-test, t30 = 6.30, P < 0.0001; Kelly CD, unpublished data) in a broad-leaved forest patch (ca. 15 ha) in which large galleries (>3 adult weta) are available. In addition, male H. crassidens captured in broad-leaved forests with large galleries on Mana and Stephens Islands (Moller 1985), as well as at Palmerston North/Manawatu Gorge (Kelly CD, personal observation), exhibit head lengths greater than 20 mm. However, in 3 other New Zealand populations of H. crassidens (Lake Kaniere: 42°48'S, 171°09'E; Somes Island: 41°02'S, 173°53'E; Baton Valley: 41°17'S, 172°45'E) that tend to lack large natural galleries, typical male head lengths are less than 21 mm (Kelly CD, personal observation). Interestingly, the tree weta on Somes/Matiu Island, a scientific reserve in the Wellington Harbor that is free of nonnative predators, is descended from Mana Island individuals (Gibbs G, personal communication), a population in which males exhibit a range of head sizes similar to males on Maud Island (Moller 1985).
These observations, together with my experimental results, suggest that less intense male–male competition at sites with small galleries relaxes selection for increased weapon size. Maybe the reduced selection pressure on weaponry in small gallery sites played a role in the evolution and/or maintenance of precocial male maturation in this species. That is, perhaps selection favored males that responded to environmental cues (average gallery size in a patch) and matured early at the 8th instar, whereas males in larger gallery forests maximized their fitness by continuing their development until the 10th instar (sensu Shuster and Wade 2003).
The opportunity for sexual selection
The observed effect of resource quality on selection was opposite to that predicted despite larger galleries housing significantly larger harems and exhibiting greater clumping of females. Contrary to conventional theory (e.g., Emlen and Oring 1977), the upper limit to sexual selection was not dictated by the average harem size of successful males (i.e., operational sex ratio) or the degree to which females clump in space; instead sexual selection was most affected by the difference in the proportion of successful (males residing with one or more females) versus unsuccessful males. A larger proportion of unsuccessful males in Equation 3 produced larger values of Imates. This observation supports the claim of Shuster and Wade (2003; see also Wade 1979; Wade and Arnold 1980) that the most important determinant of the strength of sexual selection is the magnitude of the difference between successful (males residing with one or more females) and unsuccessful (males residing alone) males.
Why would smaller galleries more often house unsuccessful males than larger galleries? I posit that this phenomenon is a consequence of female-searching behavior and gallery preferences. First, larger female aggregations form in larger galleries (Kelly 2006c, 2006d), and through acoustic and/or chemical cues (Field and Jarman 2001), groups of weta likely attract other gallery-searching females (see also Wade 1995) as well as mate-searching males. Second, although females prefer to reside alone in galleries, they will take up residence with a male when other females are present likely because other conspecifics dilute the costs associated with residing with males (e.g., injury to limbs during forced copulations) (Kelly 2006c). On the other hand, because larger aggregations of females cannot occur in smaller galleries, females are less likely to enter a gallery housing an adult male as harassment costs will not be diluted. In addition, from the male perspective, a lack of large aggregations should also make it more difficult for males to find female-occupied galleries.
Linear selection on male weaponry and body size
The opportunity for sexual selection, Imates, describes the upper limit of the strength of selection. As it does not provide information on the relationship between fitness and phenotypic characters, Imates does not measure selection acting on the phenotype per se. I identified targets of phenotypic selection by using multivariate statistical techniques (Lande and Arnold 1983; Arnold and Wade 1984a, 1984b; Brodie et al. 1995). Selection gradients support my prediction that selection on weaponry should be strongest in the large gallery treatment because male–male competition should be more intense at these sites given that they likely offer greater reproductive rewards to successful males. These findings support a correlational field study conducted on the same population and using large artificial galleries of identical size (Kelly 2005). That male weaponry is a target of selection is not surprising because laboratory experiments show that male tree wetas use their weapons in fights for possession of galleries housing harems, and males with larger mandibles have a greater probability of winning a contest (Kelly 2006a). In addition, smaller weaponry, or smaller body size in general, maybe favored in patches with small galleries if smaller males are better at searching for mates (see above) and/or fitting into spaces too small to accommodate large males (Kelly 2006d).
My findings on H. crassidens are similar to those on a congener, Hemideina maori, which inhabits cavities under rocks (Gwynne and Jamieson 1998; Jamieson 2002; Leisnham and Jamieson 2004). Gwynne and Jamieson (1998) found a strong positive correlation between head size and harem size in a cross-sectional study in the wild, and, in follow-up longitudinal studies, Jamieson (2002) and Jamieson and Leisnham (2004) found weak positive selection on male weaponry (head width). The latter 2 studies were not multivariate selection analyses and so did not provide partial regression coefficients; thus, direct comparisons with selection gradients for H. crassidens are not possible.
In general, that most gradients calculated in this study are statistically nonsignificant is not surprising given that the most powerful studies (i.e., sample size > 1000) of selection gradients on a diversity of traits indicate that selection is typically quite weak (Kingsolver et al. 2001). Detecting nonlinear selection requires even greater power than needed to detect linear selection (Conner 2001). A recent literature review showed that only 26% of reported β estimates were significantly different from zero at a level of 0.05 (Hoekstra et al. 2001). Thus, one factor likely contributing to the nonsignifcant results in my study was small sample size; the need for very large sample sizes plague studies of selection in the wild (Endler 1986; Kingsolver et al. 2001). Obtaining large sample sizes is difficult with H. crassidens because weta do not occur in very high numbers in nature.
The relationship between opportunity and strength of sexual selection
Contrary to prediction, the relationship between the opportunity for sexual selection (Imates) and the strength of selection (selection gradient) on weapon size was not positive. In fact, this relationship showed a marginally nonsignifcant negative trend. Bjork and Pitnick (2006) found a negative relationship between opportunity and strength of sexual selection among Drosophila species. They found that as sperm size increased among species the strength of selection on males decreased, whereas the opportunity for sexual selection increased. Similar to my prediction, they expected measures of opportunity and strength to complement each other. Why do the patterns differ? Even though both approaches measure sexual selection, they measure different aspects of the process. Imates is a variance-based method estimating the potential of sexual selection, whereas selection gradients measure the degree to which selection operates on particular phenotypic traits. Contrary to these results, McLain (1992) found that selection opportunity indexed selection intensity, as measured using selection differentials, on traits in the seed bug Neacoryphus bicrucis. He found that when sexual selection was intense, body size accounted for a greater amount of variation in male mating success than otherwise.
Although longitudinal selection experiments on natural populations of animals provide powerful tests of sexual selection theory (Endler 1986; Wade and Kalisz 1990), this approach is rarely employed despite a call to attention more than 17 years ago (see Wade and Kalisz 1990). The present study is one of very few to experimentally manipulate, in a natural population, the environmental factors affecting the opportunity and strength of sexual selection in an animal. I have shown that refuge size directly affects harem size, which ultimately dictates the opportunity for sexual selection as well as the strength of selection on male weaponry and body size. Finally, my results have practical applications: by providing large artificial galleries in forests comprised of small galleries (e.g., early succession and/or immature regenerating forests), conservationists can maintain the full range of male tree weta phenotypes within populations.
|
Funding |
|---|
|
TOPABSTRACTINTRODUCTIONMethodsResultsDiscussionFundingREFERENCES |
|---|
Natural Sciences and Engineering Research Council and National Geographic Society (to Darryl Gwynne); Australian National University Faculty of Science (support grant to C.D.K.); National Sciences and Engineering Research Council (Postdoctoral Fellowship to C.D.K.).
|
ACKNOWLEDGEMENTS |
|---|
I thank Darryl Gwynne, Maydianne Andrade, Michael Jennions, Mark Hauber, and 2 anonymous referees for comments on the manuscript. I also thank George Gibbs for valuable discussion and insight as well as Steve Ward of the New Zealand Department of Conservation for logistical support while this work was conducted on Maud Island.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMethodsResultsDiscussionFundingREFERENCES |
|---|
Ahnesjo I, Kvarnemo C, Merilaita S. Using potential reproductive rates to predict mating competition among individuals qualified to mate. Behav Ecol (2001) 12:397–401.
Andersson M. Sexual selection. (1994) Princeton (NJ): Princeton University Press.
Arnold SJ. Bateman's principles and the measurements of sexual selection in plants and animals. Am Nat (1994) 144:S126–S149.[CrossRef][Web of Science]
Arnold SJ, Duvall D. Animal mating systems—a synthesis based on selection theory. Am Nat (1994) 143:317–348.[CrossRef][Web of Science]
Arnold SJ, Wade MJ. On the measurement of natural and sexual selection: Applications. Evolution (1984a) 38:720–734.[CrossRef][Web of Science]
Arnold SJ, Wade MJ. On the measurement of natural and sexual selection: Theory. Evolution (1984b) 38:709–719.[CrossRef][Web of Science]
Bjork A, Pitnick S. Intensity of sexual selection along the anisogamy-isogamy continuum. Nature (2006) 441:742–745.[CrossRef][Medline]
Bowie MH, Hodge S, Banks JC, Vink CJ. An appraisal of simple tree-mounted shelters for non-lethal monitoring of weta (Orthoptera: Anostostomatidae and Rhaphidophoridae) in New Zealand nature reserves. J Insect Conserv (2006) 10:261–268.[CrossRef]
Brodie ED, Moore AJ, Janzen FJ. Visualizing and quantifying natural selection. Trends Ecol Evol (1995) 10:313–318.[CrossRef]
Brooks R, Endler JA. Direct and indirect sexual selection and quantitative genetics of male traits in guppies (Poecilia reticulata). Evolution (2001) 55:1002–1015.[CrossRef][Web of Science][Medline]
Conner JK. How strong is natural selection? Trends Ecol Evol (2001) 16:215–217.[CrossRef][Medline]
Crow JF. Some possibilities for measuring selection intensities in man. Hum Biol (1958) 30:1–13.[Medline]
Darwin C. The descent of man, and selection in relation to sex. (1871) London: J. Murray.
Dawson J, Lucas R. Nature guide to the New Zealand forest. (2000) Auckland (New Zealand): Random House New Zealand.
Emlen ST, Oring LW. Ecology, sexual selection, and the evolution of mating systems. Science (1977) 197:215–223.
Endler JA. Natural selection in the wild. (1986) Princeton (NJ): Princeton University Press.
Fairbairn DJ, Preziosi RF. Sexual selection and the evolution of sexual size dimorphism in the water strider, Aquarius remigis. Evolution (1996) 50:1549–1559.[CrossRef][Web of Science]
Fairbairn DJ, Wilby AE. Inequality of opportunity: measuring the potential for sexual selection. Evol Ecol Res (2001) 3:667–686.
Field LH. Aggression behaviour in New Zealand tree wetas. In: The biology of wetas, king crickets and their allies—Field LH, ed. (2001) Wallingford (UK): CAB International. n333–n349.
Field LH, Deans NA. Sexual selection and secondary sexual characters of wetas and king crickets. In: The biology of wetas, king crickets and their allies—Field LH, ed. (2001) Wallingford (UK): CAB International. 179–204.
Field LH, Glasgow S. Defence behaviour. In: The biology of wetas, king crickets and their allies—Field LH, ed. (2001) Wallingford (UK): CAB International. 297–316.
Field LH, Jarman TH. Mating behaviour. In: The biology of wetas, king crickets and their allies—Field LH, ed. (2001) Wallingford (UK): CAB International. 317–332.
Field LH, Sandlant GR. Aggression and mating behavior in the Stenopelmatidae (Orthoptera, Ensifera), with reference to New Zealand wetas. In: Orthopteran mating systems—sexual competition in a diverse group of insects—Gwynne DT, Morris GK, eds. (1983) Boulder (CO): Westview Press. 120–146.
Field LH, Sandlant GR. The gallery-related ecology of New Zealand tree wetas, Hemideina femorata and Hemideina crassidens (Orthoptera, Anostostomatidae). In: The biology of wetas, king crickets and their allies—Field LH, ed. (2001) Wallingford (UK): CAB International. 243–257.
Gibbs GW. Habitats and biogeography of New Zealand's deinacridine and tusked weta species. In: The biology of wetas, king crickets and their allies—Field LH, ed. (2001) Wallingford (UK): CAB International. 35–55.
Gwynne DT, Jamieson I. Sexual selection and sexual dimorphism in a harem-polygynous insect, the alpine weta (Hemideina maori, Orthoptera Stenopelmatidae). Ethol Ecol Evol (1998) 10:393–402.
Gwynne DT, Kelly CD. Do the Imates have it? Evolution (2004) 58:1877–1879.
Hauber ME, Lacey EA. Bateman's principle in cooperatively breeding vertebrates: the effects of non-breeding alloparents on variability in female and male reproductive success. Integr Comp Biol (2005) 45:903–914.
Hersch EI, Phillips PC. Power and potential bias in field studies of natural selection. Evolution (2004) 58:479–485.[Web of Science][Medline]
Hoekstra HE, Hoekstra JM, Berrigan D, Vignieri SN, Hoang A, Hill CE, Beerli P, Kingsolver JG. Strength and tempo of directional selection in the wild. Proc Natl Acad Sci USA (2001) 98:9157–9160.
Jamieson IG. The relationship between male head size and harem size in the sexually dimorphic mountain stone weta Hemideina maori. Ecol Entomol (2002) 27:41–48.[CrossRef]
Kelly CD. Allometry and sexual selection of male weaponry in Wellington tree weta, Hemideina crassidens. Behav Ecol (2005) 16:145–152.
Kelly CD. Fighting for harems: assessment strategies during male-male contests in the sexually dimorphic Wellington tree weta. Anim Behav (2006a) 72:727–736.[CrossRef][Web of Science]
Kelly CD. Movement patterns and gallery use by the sexually dimorphic Wellington tree weta. New Zealand J Ecol (2006b) 30:273–278.
Kelly CD. Resource quality or harem size: what influences male tenure at refuge sites in tree weta (Orthoptera: Anostostomatidae)? Behav Ecol Sociobiol (2006c) 60:175–183.[CrossRef][Web of Science]
Kelly CD. The relationship between resource control, association with females and male weapon size in a male-dominance insect. Ethology (2006d) 112:362–369.[CrossRef][Web of Science]
Kingsolver JG, Hoekstra HE, Hoekstra JM, Berrigan D, Vignieri SN, Hill CE, Hoang A, Gibert P, Beerli P. The strength of phenotypic selection in natural populations. Am Nat (2001) 157:245–261.[CrossRef][Web of Science]
Kwiatkowski MA, Sullivan BK. Geographic variation in sexual selection among populations of an iguanid lizard, Sauromalus obesus (=ater). Evolution (2002) 56:2039–2051.[CrossRef][Web of Science][Medline]
Lande R, Arnold SJ. The measurement of selection on correlated characters. Evolution (1983) 37:1210–1226.[CrossRef][Web of Science]
LeBas NR, Hockham LR, Ritchie MG. Nonlinear and correlational sexual selection on ‘honest’ female ornamentation. Proc R Soc Lond B (2003) 270:2159–2165.[Medline]
Legendre P. Program for multiple linear regression (ordinary or through the origin) with permutation test—user's notes (2002) Montreal (QC): Departement de sciences biologiques, Universite de Montreal.
Leisnham PT, Jamieson IG. Relationship between male head size and mating opportunity in the harem-defence, polygynous tree weta Hemideina maori (Orthoptera: Anostostomatidae). New Zealand J Ecol (2004) 28:49–54.
Lloyd M. Mean crowding. J Anim Ecol (1967) 36:1030.
Lorch PD. Using upper limits of "Bateman gradients" to estimate the opportunity for sexual selection. Integr Comp Biol (2005) 45:924–930.
McLain DK. Population density and the intensity of sexual selection on body length in spatially or temporally restricted natural populations of a seed bug. Behav Ecol Sociobiol (1992) 30:347–356.[Web of Science]
Mills SC, Grapputo A, Koskela E, Mappes T. Quantitative measure of sexual selection with respect to the operational sex ratio: a comparison of selection indices. Proc R Soc B Biol Sci (2007) 274:143–150.[Medline]
Mitchell-Olds T, Shaw RG. Regression analysis of natural selection: statistical inference and biological interpretation. Evolution (1987) 41:1149–1161.[CrossRef][Web of Science]
Mitchell-Olds T, Shaw RG. Comments on the causes of natural selection. Evolution (1990) 44:2158–2158.[CrossRef][Web of Science]
Moller H. Tree wetas (Hemideina crassicruris) (Orthoptera: Stenopelmatidae) of Stephens Island, Cook Strait. New Zealand J Zool (1985) 12:55–69.
Sandlant GR. Aggressive behaviour of the canterbury weta Hemideina femorata (Blanchard) (Orthoptera: Stenopelmatidae): its adaptive significance in resource allocation (M.Sc.). (1981) Christchurch (New Zealand): University of Canterbury.
Schluter D. Estimating the form of natural selection on a quantitative trait. Evolution (1988) 42:849–861.[CrossRef][Web of Science]
Schluter D, Smith JNM. Natural selection on beak and body size in the song sparrow. Evolution (1986) 40:221–231.[CrossRef][Web of Science]
Shuster SM, Wade MJ. Mating systems and strategies. (2003) Princeton (NJ): Princeton University Press.
Sokal RR, Rohlf FJ. Biometry: the principles and practice of statistics in biological research. (1995) 3rd ed. New York: Freeman.
Spencer AM. Sexual maturity in the male tree weta Hemideina crassidens (Orthoptera: Stenopelamatidae) (M.Sc.). (1995) Wellington (New Zealand): Victoria University.
Sutherland WJ. Chance can produce a sex difference in variance in mating success and explain Bateman's data. Anim Behav (1985) 33:1349–1352.[CrossRef][Web of Science]
Trewick SA, Morgan-Richards M. Artificial weta roosts: a technique for ecological study and population monitoring of Tree weta (Hemideina) and other invertebrates. New Zealand J Ecol (2000) 24:201–208.
Wade MJ. Sexual selection and variance in reproductive success. Am Nat (1979) 114:742–764.[CrossRef][Web of Science]
Wade MJ. The ecology of sexual selection: mean crowding of females and resource-defense polygyny. Evol Ecol (1995) 9:118–124.[CrossRef]
Wade MJ, Arnold SJ. The intensity of sexual selection in relation to male sexual behaviour, female choice, and sperm precedence. Anim Behav (1980) 28:446–461.[CrossRef][Web of Science]
Wade MJ, Kalisz S. The causes of natural selection. Evolution (1990) 44:1947–1955.[CrossRef][Web of Science]
Wade MJ, Shuster SM. Don't throw Bateman out with the bathwater! Integr Comp Biol. 4 (2005) 5:945–951.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  C. D. Kelly Sperm investment in relation to weapon size in a male trimorphic insect? Behav. Ecol., September 1, 2008; 19(5): 1018 - 1024. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||