Genital and body allometry in two species of noctuid …Genital and body allometry in two species of noctuid moths (Lepidoptera: Noctuidae) MOHAMMAD MAHDI RABIEH1*, MEHDI ESFANDIARI1,
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Genital and body allometry in two species of noctuidmoths (Lepidoptera: Noctuidae)
MOHAMMAD MAHDI RABIEH1*, MEHDI ESFANDIARI1, ALI ASGHAR SERAJ1 andRUSSELL BONDURIANSKY2
1Department of Plant Protection, College of Agriculture, Shahid Chamran University of Ahvaz,Ahvaz, Iran2School of Biological, Earth and Environmental Sciences, University of New South Wales,Sydney, NSW, 2052, Australia
Received 2 February 2015; revised 31 March 2015; accepted for publication 31 March 2015
ADDITIONAL KEYWORDS: body size – genitalia – Helicoverpa armigera – sexual selection – Spodopteraexigua.
INTRODUCTION
Huxley (1932) was first to employ the term ‘allome-try’ to define the differential analysis of size relation-ships between different body parts. This is still apowerful quantitative approach, used to examinehow selection impacts the relationship between mor-phology and performance. The observed allometricslope (a slope in log-log regression on an indicator ofbody size) is presumably a consequence of selection
that favours one slope over others (Eberhard et al.,1998). Static allometry is the relationship between abody part and body size among conspecific individu-als at a particular developmental stage. If the slopeexceeds 1.0, the trait is said to be positively allo-metric, such that larger individuals have relativelylarger traits. Allometric slopes around 1.0 showisometry, where relative trait size remains constantacross the range of body sizes. Slopes significantlylower than 1.0 indicate negative allometry, wherelarger individuals express relatively smaller traits.Some sexually-selected morphological characters*Corresponding author. E-mail: [email protected]
Biological Journal of the Linnean Society, 2015, 116, 183–196. With 4 figures.
used as weapons, or as visual advertisements, exhi-bit positive static allometries when scaled againstother body parts (Vencl, 2004). Selection acts inthree general ways: directional, disruptive or stabi-lizing. Stabilizing selection favours phenotypes withtraits of intermediate size (Hosken & Stockley,2004). Hypotheses based on selection pressures sug-gest that an allometric relationship will have a slopeof one when selection on the morphological characteris the same as on overall body size and a slope otherthan one when selection acts differently on the twotraits (Gould, 1966; Bonduriansky & Day, 2003).Genitalic allometries might be important for femalechoice because large male body size may be associ-ated with superior abilities to accumulate resourcesand to survive (Andersson, 1994). The idea that sex-ual selection influences genital evolution has beenwidely developed in the context of postcopulatorysexual selection. Eberhard et al. (1998; Eberhard,2009) studied the allometry of the sizes of male andfemale genitalia and other body parts of many spe-cies of arthropods (including different orders ofinsects and several species of other arthropods( andfound that allometric slopes for male genitalia wereconsistently lower than 1.0 and lower than the slopesfor the other body parts. Based on the findings, theone-size-fits-all hypothesis was proposed, which pos-its that, within a species, sexual selection favoursmales with genitalia of average size, and suggeststhat approximate size-invariance of genitalia isachieved by shallow static allometry. Eberhard et al.(1998) suggested that females typically ‘perceive amale’s genitalia at close range by more or less pre-cisely aligned touch’, so that visual assessment isunlikely to play a major role. Eberhard (2009) sug-gested a more general explanation, whereby low allo-metric slopes in arthropods reflect selection for bothmechanical fit and stimulation. Subsequent to Eberhardet al. (1998), several studies on beetles, dipterans,moths, and spiders have similarly detected negativeallometry for male genital size (Palestrini, Rolando& Laiolo, 2000; Schmitz, Reinhold & Wagner, 2000;Uhl & Vollrath, 2000; Iwahashi, 2001; Iwahashi &Routhier, 2001; Tatsuta, Mizota & Akimoto, 2001;Bernstein & Bernstein, 2002; Eberhard, 2002; Ohnoet al., 2003; Mutanen & Kaitala, 2006; Mutanen,Kaitala & M€Onkk€Onen, 2006). Low allometric slopesfor male genital structures therefore appear to becommon among insects and spiders. However, someexceptions to this general pattern have also beenobserved (Johnson, 1995; Cayetano et al., 2011;Nava-Bola~nos et al., 2014). This suggests that genitaltraits may evolve via different kinds of selectivepressures in different species, and perhaps withinspecies. Further research is thus required to gain abetter understanding of the diverse functional roles
of genitalic traits, and the selective pressures drivingtheir evolution.
Green (1999) challenged Eberhard’s idea, arguingthat the appropriate technique for estimating suchlines of ‘organic correlation’ is Model II regression(RMA) rather than ordinary least square regression(OLS), and emphasizing that Model II regressionyields substantially steeper allometries for genitalictraits. In response, Eberhard, Huber & Rodriguez(1999) suggested that the weak correlations betweengenitalia and somalic structures may have biologicalsignificance in that they may indicate the unreliabil-ity of genitalia size in providing information to thefemale about overall body size of the male. Nonethe-less, genitalic allometries might be important forfemale evaluation of male abilities to accumulateresources and to survive (Andersson, 1994). Accord-ing to the ‘good viability genes’ model, females usemale genitalia to assess male’ heritable quality.Strong negative allometry between body size andgenital size is proposed by some other hypotheses,such as the lock-and-key hypothesis and the crypticfemale choice hypothesis (Eberhard et al., 1998;Eberhard, 2009), and may be produced by othermechanisms of sexual selection as well (Bondurian-sky & Day, 2003). According to the cryptic femalechoice hypothesis, sperm of males with superior stim-ulating ability during copulation is preferred byfemales. Although Eberhard et al. (1998) and Eber-hard (2009) found that female genitalia may show thesame patterns of variation as male genitalia, previousstudies that investigate the allometry relationships ininsects have focused almost exclusively on males andset females aside (Palestrini et al., 2000; Schmitzet al., 2000; Tatsuta et al., 2001; Bernstein &Bernstein, 2002; Eberhard, 2002; Ohno et al., 2003;Mutanen & Kaitala, 2006; Mutanen et al., 2006;Al-Wathiqui & Rodriguez, 2011; Nava-Bola~nos et al.,2014). Comparison of static allometries of male andfemale traits could be used to determine whethermale genitalia are really distinctive from other typesof traits in their shallow static allometries: if selectionfavours shallow allometry in male genitalia becauseof sexual selection, then female genitalic traits wouldnot be expected to exhibit similarly shallow allome-tries. Rather, female genitalia should be similar tosomatic traits in their allometries. Conversely, if bothmale and female genitalic traits have similarly shal-low allometries, then this would suggest that genita-lia have shallow static allometry slopes for reasonsunrelated to sexual selection.
According to Eberhard (1985, 1996), the male geni-talic traits of many animals function as copulatorycourtship devices that mechanically stimulatefemales in a specific way during copulation, in con-trast to the male ornaments that stimulate females
visually. Eberhard et al. (1998) proposed that sexualselection can cause low allometry for male genitalsizes, as follows. Because females assess male genita-lia at close range by touch, rather than visually,selection may favour males with average-sized geni-talia because such genitalia may be most effective atstimulating average-sized females (i.e. the most com-mon class of females in the population). If male bodysize varies considerably but selection favours a rela-tively invariant, average genital size across therange of male body size, a negative static allometryin male genitalic traits is expected to evolve.
To determine whether genitalic traits exhibitstrong negative allometry, consistent with stabilizingselection, in noctuid moths, we examined the allo-metry of genitalic and somatic morphological traitsin males and females of two pest species, thebeet armyworm, Spodoptera exigua (H€ubner, 1808)(Noctuidae: Xyleninae) and the cotton bollworm,Helicoverpa armigera (H€ubner, 1808) (Noctuidae:Heliothinae). According to Fibiger & Lafontaine(2005), subfamilies Heliothinae and Xyleninae areconsidered as two sister-groups and are included inthe ‘pestclade’ of Mitchell, Mitter & Regier (2006).The studied species belong to different subfamiliesand investigation of both species allowed us to deter-mine whether similar patterns occur in these distinctnoctuid lineages. They differ in the male genitalicstructure: one species has a valvae elaborated bymorphological structures (S. exigua) and the otherhas a rather simple valval structure (H. armigera).
For similar reasons, we also predicted that genita-lic traits would exhibit lower variability [quantifiedas the coefficient of variation (CV) and standarderror of estimate (SEE)] than somatic traits.
MATERIAL AND METHODS
INSECTS
The study species, S. exigua and H. armigera, arecommon agricultural pests in Iran (Fibiger & Hacker,2007; Matov, Zahiri & Holloway, 2008). Sample sizes(34 males and 34 females for S. exigua and 31 malesand 31 females for H. armigera) of adults were col-lected during 2010–2011, from agricultural fieldsaround Mashhad city, north-eastern Iran. Night sam-plings were carried out by using light traps. Materialswere deposited in the Insect and Mite Collection ofAhvaz, Plant Protection Department, Shahid Cham-ran University of Ahvaz, Iran.
PREPARATION PROCEDURE
The body parts were first boiled in water and thenwashed with ethanol. After cleaning their scales and
setae with a camel’s hair brush, they were then pre-pared between microscope slides in Canada balsam fix-ative. The abdomen was first removed and preserved in10% caustic potash for 24 h and then washed withwater. Genitalia of both sexes were then removed fromthe softened surrounding tissues, dehydrated with eth-anol, and mounted on Euparal between the microscopeslides and cover slips. After preparation, the genitaliawere photographed through a microscope using a C-5050ZOOM digital camera (Olympus).
MORPHOLOGICAL TRAITS AND MEASUREMENTS
We examined both male and female adults; 19 (10somatic and nine genitalic) traits in males and 12 (10somatic and two genitalic) traits in females ofS. exigua, as well as 10 (five somatic and five genita-lic) traits in both sexes of H. armigera were measuredin the present study. In all parts of the body, the leftone was measured. In S. exigua, the ten measuredsomatic traits in both sexes were the lengths of thebody parts: forewing, hindwing, fore femur, fore tibia,mid femur, mid tibia, hind femur, hind tibia, and thefirst and second segments of the labial palp (Fig. 1).In S. exigua, the measured nine genitalic traits ofmales were the lengths of the genitalic parts: valva,sacculus, clasper, aedeagus, cornutus of vesica, uncus,ampulla, tegumen, and diameter of aedeagus (Fig. 1).In the female genitalia of this species, the lengths ofthe papilla analis and apophysis posterior were mea-sured (Fig. 1). In H. armigera, the five measuredsomatic traits in both sexes were the lengths of thebody parts: forewing, fore femur, fore tibia, hindfemur, and hind tibia (Fig. 2). The five measured geni-talic traits in H. armigera males, were the lengths ofthe body parts: valva, aedeagus, uncus, tegumen, andthe diameter of aedeagus (Fig. 2). In the female geni-talia, the measured lengths of the genitalic partswere: papilla analis, apophysis posterioris, apophysisanterioris, bursa copulatrix, and ostium bursa diame-ter (Fig. 2). All the measurements were performedusing TPSDIG, version 2.16 (Rohlf, 2004).
MEASUREMENT ERROR
To evaluate measurement error, three measurementsof each measured trait were repeated nonconsecu-tively. The percentage measurement error was calcu-lated as (Yezerinac, Lougheed & Handford, 1992):
%ME ¼ S2within
S2within þ S2
among
�100
The percentage measurement error values weremostly less than 1% of the total variance (mean 0.15,
range 0.00017–1.83) and did not differ significantlybetween the two species (Mann–Whitney U-test,P = 0.946) and between genitalic and somatic traitgroups in both sexes (Mann–Whitney U-test,P = 0.383). For these reasons, variables for furtheranalyses were calculated by averaging the repeatedmeasurements.
ESTIMATION OF ALLOMETRIC SLOPE AND OTHER
FEATURES OF MORPHOLOGICAL VARIATION
We used principal component analysis (PCA), con-ducted separately for each species/sex, to examine
the covariation structure of the trait matrix (Figs 3and 4). Several studies on genital allometry used asingle somatic trait as the indicator of body size,although the use of a different body-size indicatormay give rise to different results (Green, 1999). Weselected forewing, which loads most strongly on PC1,for use as index of body size and to calculate all ofthe allometric slopes. Using PC1scores as body sizeindicator is problematic because all traits areincluded in the PCA, and the x and y variables inthe regressions are thus not independent.
Correlations between body size and the morpholog-ical traits were then calculated one by one using
A
B
C
E
G
H
I J
K
M
N
L Q
R
S
T
U
O
P
F
D
Figure 1. Measured body parts of Spodoptera exigua male and female (A–J: somatic, K–U: genitalic). A, forewing;
B, hindwing; C, hind femur; D, hind tibia; E, mid femur; F, mid tibia; G, fore femur; H, fore tibia; I; J, first and second
segments of labial palp; K, valve length; L, sacculus length; M, ampulla length; N, clasper length; O, tegumen length;
P, uncus length; Q, length of aedeagus; R, diameter of aedeagus; S, length of spine of vesica; T, papilla analis; U, apoph-
Pearson’s product–moment correlation coefficient.For each trait, the allometric slope was estimated bylinear regression analysis of the log10-transformedvalue of the trait on the log10-transformed forewinglength values.
We used type I (OLS) regression analysis (Eberhardet al., 1998, 1999; Cuervo & Møller, 2001; Al-Wathiqui& Rodriguez, 2011). Because OLS regression assumesthat the values in x are determined without error, itis generally expected to underestimate slopes unlessthe error in x is much smaller than in y (Sokal &Rohlf, 1981). The absolute values of OLS slope esti-mates should therefore be interpreted with caution,although Al-Wathiqui & Rodriguez (2011) concludedthat OLS regression is relatively robust to measure-ment error in x. However, in the present study, we
are mainly interested in comparing slopes betweengenitalic and somatic traits and between male andfemale traits, rather than in estimating the absolutevalues of these slopes. Because all slopes are calcu-lated in a similar way (using OLS regression withforewing as the body size index), the tendency of OLSregression to yield shallower slope estimates thanRMA regression is therefore unlikely to lead to biasedresults in our analysis. In addition, the use of RMAregression suffers from problems of interpretability(Cuervo & Møller, 2001; Ohno et al., 2003; Al-Wathiqui& Rodriguez, 2011). In particular, RMA slopesshould not be used when the correlation betweentrait size and body size is nonsignificant (Wartonet al., 2006), as is often the case for genitalic traits.Therefore, we use slopes from OLS regression
A
B
D
F
G
H
I
J
K
L M
NO
E
C
Figure 2. Measured body parts of Helicoverpa armigera male and female (A–E: somatic, F–O: genitalic). A, forewing;
B, hind femur; C, hind tibia; D, fore femur; E, fore tibia; F, length of aedeagus; G, diameter of aedeagus; H, valve length;
I, tegumen length; J, uncus length; K, papilla analis; L, apophysis posterioris; M, ostium bursa diameter; N, apophysis
throughout (Table 1) (Cuervo & Møller, 2001). How-ever, for comparison, we provide a table includingthe allometric slopes of morphometric traits in thetwo examined species using three regression methodsOLS, RMA and MA (see Supporting information,Table S1). The null hypothesis of b = 1 was tested byt-tests, and by examining 95% confidence intervals(Tables 1, 2).
Another, potentially independent source of infor-mation about trait variation is the CV, which is cal-culated as the SD/mean. The CV denotes the relativeamount of variation in the size of a body part, inde-pendent of the magnitude of the mean. A high vari-ance, as shown by a high CV value, is considered tobe an important indicator of sexual selection com-pared to measures of overall body size (Vencl, 2004).
The difference in CV values among traits isaffected by the degree of dispersion of data pointsaround the allometric line, in addition to the allo-metric slope (Eberhard et al., 1998). To estimate thedegree of dispersion of points around the line, we cal-culated the CV’. Because CV’ is invalid when r is not
significant (Eberhard et al., 1998; Cuervo & Møller,2001), we also calculated SEE, which is another mea-sure of the degree of dispersion of points (Cuervo &Møller, 2001). SEE is unaffected by r, although ithas the disadvantage of not being dimensionless(Eberhard et al., 1998). The allometric slope, CV,CV’, and SEE were respectively compared betweentrait categories (genitalic vs. somatic) in two sexes byMann–Whitney U-tests within each species.
RESULTS
PCA ANALYSIS
In H. armigera males, the first component of thePCA explained 92.3% (eigenvalue 0.63) and the sec-ond explained 2.6% (eigenvalue 0.018) of the totalvariance. In H. armigera females, the first compo-nent of the PCA explained 93.4% (eigenvalue 1.01)and the second explained 3.2% (eigenvalue 0.034) ofthe total variance. For H. armigera, plots of traitloadings on PC1 vs. PC2 that reveal the covariation
A C
DB
Figure 3. Principal component (PC) analysis plots of trait loadings on PC1 vs. PC2 showing the covariation structure
in the trait matrix of the males and females. A, Helicoverpa armigera male. B, H. armigera female. C, Spodoptera ex-
structure in the trait matrix are shown in Figure 3A,B. In S. exigua males, the first component of thePCA explained 86.0% (eigenvalue 0.78) and thesecond explained 0.041% (eigenvalue 0.037) of thetotal variance. In S. exigua females, the first compo-nent of the PCA explained 93.3% (eigenvalue 0.72)and the second explained 2.09% (eigenvalue 0.022) ofthe total variance. For S. exigua, plots of trait load-ings on PC1 vs. PC2 that reveal the covariationstructure in the trait matrix are shown in Figure 3C,D. For examined species/sexes, matrix scatterplotsthat illustrate the allometric relationships betweenall measurements are given (Fig. 4).
CORRELATION BETWEEN BODY AND EXAMINED TRAITS
SIZES
All somatic traits in each species/sex were signifi-cantly correlated with body size (r = 0.249–0.970, allP < 0.01) (Tables 1, 2, 3, 4). In H. armigera, all geni-talic traits in males and females were also signifi-cantly correlated with body size (r = 0.521–0.761, allP < 0.01). However, in S. exigua, only two out of 11genitalic traits of males and females had significantcorrelations (r = 0.54–0.74, P < 0.01). When a correlation
was observed, it was statistically highly significant(P < 0.01 in all cases).
TEST OF ISOMETRY
The relationship between male genitalic traits andthe body size indicator was in general strongly nega-tively allometric (i.e. all of the slopes in both specieswere significantly less than 1.0) (Tables 1, 3). Infemales, three genitalic traits showed strongly nega-tively allometric relationship with body size and fourtraits showed isometry (Tables 2, 4). The relation-ship between somatic traits and the body size indica-tor did not differ significantly from one (isometry) in15 cases (N = 26). In both species, most of the allo-metric slopes of somatic traits in both sexes showedallometric values equal to or slightly less than 1.0.However, in S. exigua, the allometric slope of labialpalps showed a strong negatively allometric relation-ship with the body size indicator in three out offour cases (Tables 1, 2). However, in four traits inS. exigua males and four traits in S. exigua females,the slopes showed significant negative allometry,whereas two traits in H. armigera females showedsignificant negative allometry and the hind tibia in
A C
DB
Figure 4. Matrix scatterplots of two examined species that illustrate the allometric relationships between all measure-
ments. A, Helicoverpa armigera male. B, H. armigera female. C, Spodoptera exigua male. D, S. exigua female. PC, prin-
The mean slope for the nine genitalic traits was sig-nificantly lower than the mean slope for the ninesomatic traits in S. exigua males (Mann–Whitney U-test, P < 0.01). Similarly, the mean slope for the fivegenitalic traits was significantly lower than the meanslope for the four somatic traits in H. armigera males(Mann–Whitney U-test, P < 0.01). In the females,although the mean slope of genital traits was lowerthan the mean slope for somatic traits, the differ-ence was not significant (Mann–Whitney U-test forS. exigua, P = 0.085; Mann–Whitney U-test forH. armigera, P = 0.172). The slopes for somatic traitsdid not differ significantly between males and femalesin either species (Mann–Whitney U-test for S. exigua,
P = 0.545; Mann–Whitney U-test for H. armigera,P = 0.173). The slopes for sexual traits of the maleswere significantly lower than the slopes for sexualtraits of the females in each species (Mann–WhitneyU-test for S. exigua, P < 0.05; Mann–Whitney U-testfor H. armigera, P < 0.05). Note that different regres-sion models give different slope estimates in severalcases (see Supporting information, Table S1).
CV MEASUREMENTS
Dispersion around the mean, as measured by the CV,is one possible indicator of sexual selection that wascalculated for all of the traits measured in the two spe-cies (Tables 1, 2, 3 4). The CVs for all structures werelow, not exceeding 10%. In S. exigua males, genitalictraits showed significantly lower CV than somatictraits (Mann–Whitney U-test, P < 0.01). Similarly, inH. armigera males, genitalic traits showed signifi-cantly lower CV than somatic traits (Mann–Whitney
Table 1. Allometric slopes and other features for 19 morphometric traits of Spodoptera exigua males
U-test, P < 0.01) (Tables 1, 3). However, in thefemales of both species, somatic traits showed lowerCV than genitalic traits. This difference was not sig-nificant in H. armigera (Mann–Whitney U-test forS. exigua: P < 0.05; Mann–Whitney U-test forH. armigera, P = 0.602) (Tables 2, 4). CV of genitalictraits in the males were significantly lower than CV ofgenitalic traits in the females of both species(Mann–Whitney U-test for S. exigua, P < 0.05;Mann–Whitney U-test for H. armigera, P < 0.05). Thedifferences in CV of somatic traits between twosexes was not significant in the both species(Mann–Whitney U-test for S. exigua, P = 0.054;Mann–Whitney U-test forH. armigera, P = 0.249).
CV’ AND SEE MEASUREMENTS
The measure of dispersion of points around theallometric line, CV’, did not differ significantlybetween the trait categories (somatic vs. genitalic) inS. exigua males (Mann–Whitney U-test, P = 0.22) orin H. armigera males (Mann–Whitney U-test,P = 0.60) (Tables 1, 3). In the females of both spe-cies, CV’ of somatic traits were significantly lowerthan CV’ of genitalic traits (Mann–Whitney U-testfor S. exigua, P < 0.05; Mann–Whitney U-test forH. armigera, P < 0.01) (Tables 2, 4). CV’ of genitalictraits in the males were significantly lower than CV’of genitalic traits in the females (Mann–Whitney
U-test for S. exigua: P < 0.05; Mann–Whitney U-testfor H. armigera: P < 0.01) and the difference in CV’of somatic traits between two sexes was not signifi-cant in each species (Mann–Whitney U-test forS. exigua, P = 0.175; Mann–Whitney U-test forH. armigera, P = 0.602).
SEE, another measure of dispersion of points aroundthe allometric line, did not differ significantly betweenthe trait categories (somatic vs. genitalic) in eitherS. exigua males (Mann–Whitney U-test, P = 0.252) orH. armigera males (Mann–Whitney U-test, P = 0.602)(Tables 1, 3). In the females of both species, SEE ofsomatic traits were significantly lower than SEE ofgenitalic traits (Mann–Whitney U-test for S. exigua,P < 0.05; Mann–Whitney U-test for H. armigera,P < 0.01) (Tables 2, 4). SEE of genitalic traits in themales were significantly lower than SEE of genitalictraits in the females (Mann–WhitneyU-test for S. exigua,P < 0.05, Mann–Whitney U-test for H. armigera,P < 0.01) and the difference in SEE of somatic traitsbetween two sexes was not significant in each species(Mann–Whitney U-test for S. exigua, P = 0.364;Mann–Whitney U-test forH. armigera, P = 0.754).
DISCUSSION
Although most studies on genitalic allometry haveexamined only one or two traits, and have typically
Table 4. Allometric slopes and other features for 10 morphometric traits of Helicoverpa armigera females
limited their analysis to males, the present studycompared static allometries and several measures ofvariation for a large number of genitalic andsomatic traits within both sexes of two insect spe-cies. Our study therefore furnishes an unusuallypowerful and comprehensive comparison of genitalicand somatic traits in these species. According to ourresults, male genitalic traits of S. exigua andH. armigera showed negative allometry, whereasthe allometric slopes of most somatic traits in themales were near-isometric (equal to or slightly lessthan 1.0 but greater than 1.0 in one case). We con-firm that, in S. exigua and H. armigera, male geni-talia exhibited a similar pattern to the resultsreported by Eberhard et al. (1998) and Eberhard(2009). This suggests that, in these species, maleswith intermediate sized genitalia, regardless of over-all body size, apparently left more surviving off-spring than males with relatively small or largegenitalia. Some of our somatic traits showed nega-tive allometry relationships with body size. Negativeallometry is typical for most traits, although it isnot clear, in most cases, why nonsexual traits varyin static allometry slope. Further research isrequired to determine the reasons for this finding.
Our conclusion that the shallow static allometryslopes of male genitalic traits reflect functional rolesassociated with sexual competition is supported bythe finding that the static allometry slopes of femalegenitalic traits were more similar to the slopes forsomatic traits than was the case in males. Femalehomologues of male genital traits can serve as a use-ful control and, in this case, the study of female geni-talia allowed us to test the idea that male genitalcomponents are under selection for low static allome-try slopes. Because female genitalia were more simi-lar than male genitalia to somatic traits, we mayconclude that male genitalia are under selection forshallow static allometry as a result of their role inmating and sexual competition, as assumed by thefunctional hypotheses discussed above.
As Eberhard et al. (1998) stated, with a given dis-tribution of body sizes, both a higher allometric valueand a greater dispersion of points around the allo-metric line will result in a larger CV. Eberhard et al.(1998) referred to the allometric slope as a ‘designfeature’ of the organism, a manifestation of thedevelopmental-genetic programme that evolvesunder selection favouring a particular scaling rela-tionship. The degree of dispersion, on the otherhand, may be related to various causes, includinggenetic differences among individuals, differences infactors such as hormone titres that affect the size ofthe structure, variation in environmental factorsduring particular stages of growth, and imprecisionin developmental programmes.
In our results, the degree of dispersion of datapoints around the allometric lines did not signifi-cantly differ between genitalic and somatic traits inthe males. However, the CV that combines theeffects of the slope and the SEE exhibited a weakerbut nevertheless significant trend towards smallervalues in genitalic than in somatic characters inmales of each of the two species. Because the genita-lic traits showed less phenotypic variation thansomatic traits, these findings show that, in S. exiguaand H. armigera, the size of male genitalia is morestable than that of somatic parts against changes inthe body size, and the difference in the degree ofphenotypic variation between genitalic and somatictraits is attributed to the difference in allometricslopes. The same conclusion can be drawn from theordination plots for males of the two species (Fig. 3):the genitalic traits exhibit much smaller loadings onPC1 (reflecting weaker correlations with body size)than the somatic traits. The present findings are con-gruent with those of previous studies that analyzedboth the allometric slope and the dispersion of pointsaround the allometric line for male genitalia andother body parts (Eberhard et al., 1998; Palestriniet al., 2000; Ohno et al., 2003). Because the shape ofmale genitalia is conspicuously diversified amongtaxa in insects and spiders (Eberhard, 1985), it isparticularly interesting that male genital size is sta-ble within single populations of diverse taxa. Thismay suggest that a common evolutionary force hasstabilized the male genital size within various spe-cies of insects and spiders. Interestingly, Pomian-kowski & Møller (1995), in their survey, found thatCVs above 20% were common for sexually-selectedcharacters, whereas those of nonsexually-selectedcharacters averaged almost 8%. It was argued thatsexual selection favours alleles that reduce thedegree of developmental control of quantitative sec-ondary-sexual traits, with the result that they mightbe expected to have higher CVs than somatic traits.This suggests that genitalic traits differ markedly intheir development (and, presumably, in the selectiveregime that shapes the developmental-genetic archi-tecture) from secondary sexual structures, despitethe role of genitalic traits in sexual interactions andempirical evidence of sexual selection on genitalia insome species (Bertin & Fairbairn, 2007).
Our findings suggest that such stabilizing sexualselection may operate on genital size in noctuidmoths. However, apart from cryptic female choice,low allometric values may also result from spermcompetition strategies that exert stabilizing selectionon male genitalic size. Male genitalic size might beoptimized for sperm removal, quick sperm transferor sperm displacement. In all of these cases, malesmay have an advantage if their genitalia fit best to
the most typical female. Stabilizing selection on malegenitalic size through selection on sperm competitioncapability is thus as plausible as stabilizing selectionthrough cryptic female choice (Schmitz et al., 2000).
Allometric slopes for genitalia of the females in sev-eral species examined by Eberhard et al. (1998) andEberhard (2009) were lower than the median allomet-ric slope for somatic traits for the same species. Inaccordance with their results, they stated that femalegenitalia may show the same patterns of variation asmale genitalia (Eberhard et al., 1998; Eberhard,2009). Eberhard et al. (1998) reported that, just as inmales, the sizes of female genitalia were slightly lessvariable than those of somatic characters. In ourresults, somatic traits of females showed lower CVthan genitalic traits and this difference was signifi-cant in S. exigua. Moreover, CV of genitalic traits inthe males was significantly lower than CV of genitalictraits in the females in each species. Also, Eberhardet al. (1998) showed that the SEE was larger for theallometric line of genitalia than the median for non-genitalia in 11 of 12 genitalic traits. In our results,CV’ and SEE of somatic traits were significantly lowerthan CV’ and SEE of genitalic traits in the females ofboth species. According to our results, the differencesin the CV of genitalic and somatic traits of the femalesare related to the differences of CV’ and SEE. As Eb-erhard et al. (1998) stated, the degree of dispersionmay be related to various causes including genetic dif-ferences among individuals.
According to the results of the present study,female genitalia showed a similar trend to the males(especially in terms of relatively low static allometryslopes, relative to somatic traits), although the differ-ence between genital and somatic traits was not sig-nificant in females. This finding suggests thatselection is acting differently on male and femalegenitalia.
Eberhard et al. (1998) concluded that selection onfemales may favour intermediate, standard sizes ofgenitalic structures that are contacted by males. Inthe present study, the papilla analis showed strongnegative allometry in both species (Tables 2, 4). Thisstructure is the terminal part of the female’s genita-lia in the moths and is assumed to be contacted bymales during copulation. Other genitalic structuresin the females showed isometric relationships withbody size, with the exception of bursa copulatrixlength in the H. armigera, which showed negativeallometry. This part of the female genitalia receivesthe vesica of the male aedeagus during copulation.In the Noctuidae, the aedeagus and vesica (penis) inthe males, and the bursa copulatrix in the females,together form a lock-and-key mechanism (Mikkola,2008). Our results showed the same allometric rela-tionship with body size for the aedeagus in the males
and the bursa copulatrix in the females of H. armi-gera, supporting the presence of the lock-and-keymechanism in this species as well. Such a mecha-nism is consistent with the presence of selection (viaboth mechanical fit and stimulation) for low allomet-ric slopes in arthropods (Eberhard, 2009).
The hind tibia in H. armigera males showed posi-tive allometry with body size. This result suggestsselection for larger hind tibia in large armigeramales and/or relatively smaller hind tibia in smallarmigera males. If secondary sexual trait sizes areunder directional sexual selection, a large relativetrait size can yield high mating success, althoughviability costs may limit secondary sexual traitexpression in small individuals, resulting in positiveallometry (Bonduriansky, 2007). It has also beensuggested that positive allometry evolves to amplifydifferences in body size, which may be important inmale–male interactions, or female assessment ofmale mate quality. The hind tibia in H. armigerafemales showed an isometric relationship with bodysize (Table 4). Therefore, positive allometry of hindtibia in H. armigera may be a result of secondarysexual function. However, positive allometry has alsobeen observed in nonsexual somatic traits, and maybe especially common in locomotory structures(Bonduriansky, 2007). The function of the hind tibiain this species remains to be determined.
ACKNOWLEDGEMENTS
We express our appreciation to the Deputy of Research,Shahid Chamran University of Ahvaz, for providingfinancial support for this research. We sincerely thankDr Seyit Ali Kayis (Selcuk University, Turkey),Dr Marko Mutanen (University of Oulu, Finland),Professor Norman MacLeod (The Natural HistoryMuseum, London) and Dr Abdolrahman Rasekh (ShahidChamran University of Ahvaz) for their suggestionsregarding the statistical analysis. We also thankAlireza and Hossein Rabieh and Hassan Rahimi fortheir kind help in the field sampling programmes.
REFERENCES
Al-Wathiqui N, Rodriguez RL. 2011. Allometric slopes not
underestimated by ordinary least squares regression: a case
study with Enchenopa treehoppers (Hemiptera: Membraci-
dae). Annals of the Entomological Society of America 104:
562–566.
Andersson M. 1994. Sexual selection. Princeton, NJ: Prince-
ton University Press.
Bernstein S, Bernstein R. 2002. Allometry of male genita-
lia in a species of soldier beetle: support for the one-size-