Hostname: page-component-74656f9884-skzmx Total loading time: 0 Render date: 2026-11-05T23:25:05.657Z Has data issue: false hasContentIssue false
  • English
  • Français

Wing shape variations in an invasive moth are related to sexual dimorphism and altitude

Published online by Cambridge University Press:  27 January 2010

N. Hernández-L. ,
Á.R. Barragán ,
S. Dupas ,
J.-F. Silvain  and
O. Dangles
N. Hernández-L.
Affiliation:
PUCE, Facultad de Ciencias Exactas y Naturales, Quito, Ecuador
Á.R. Barragán
Affiliation:
PUCE, Facultad de Ciencias Exactas y Naturales, Quito, Ecuador
S. Dupas
Affiliation:
PUCE, Facultad de Ciencias Exactas y Naturales, Quito, Ecuador IRD-UR 072, Biodiversité et évolution des complexes plantes – insectes ravageurs – antagonistes, LEGS, UPR 9034, CNRS 91198Gif-sur Yvette Cedex, France and Université Paris-Sud 11, 91405Orsay Cedex, France
J.-F. Silvain
Affiliation:
IRD-UR 072, Biodiversité et évolution des complexes plantes – insectes ravageurs – antagonistes, LEGS, UPR 9034, CNRS 91198Gif-sur Yvette Cedex, France and Université Paris-Sud 11, 91405Orsay Cedex, France
O. Dangles*
Affiliation:
PUCE, Facultad de Ciencias Exactas y Naturales, Quito, Ecuador IRD-UR 072, Biodiversité et évolution des complexes plantes – insectes ravageurs – antagonistes, LEGS, UPR 9034, CNRS 91198Gif-sur Yvette Cedex, France and Université Paris-Sud 11, 91405Orsay Cedex, France
*
*Author for correspondence Fax: (593) 2991687 E-mail: dangles@legs.cnrs-gif.fr
Rights & Permissions [Opens in a new window]

Abstract

Wing morphology has great importance in a wide variety of aspects of an insect's life. Here, we use a geometric morphometric approach to test the hypothesis that variation, in insect wing morphology patterns, occurs between sexes and along altitudinal gradients for invasive species, despite their recent association to this environment. We explored the variation in wing morphology between 12 invasive populations of the invasive potato pest, Tecia solanivora, at low and high altitude in the central highlands of Ecuador. After characterizing sexual dimorphism in wing shape, we investigated if moths at higher elevations differ in wing morphology from populations at lower altitudes. Results indicate wing shape and size differences between sexes and between altitudinal ranges. Females showed larger, wider wings than males, while high altitude moths showed larger, narrow-shaped wings by comparison to low-altitude moths. GLM analyses confirmed altitude was the only significant determinant of this gradient. Our study confirms a sexual dimorphism in size and wing shape for the potato moth. It also confirms and extends predictions of morphological changes with altitude to an invasive species, suggesting that wing morphology variation is an adapted response contributing to invasion success of the potato moth in mountainous landscapes. Ours is one of the first studies on the morphology of invasive insects and represents a valuable contribution to the study of insect invasions because it both offers empirical support to previous genetic studies on T. solanivora as well as proving broader insight into the mechanisms behind morphological evolution of a recently introduced pest.

Keywords

altitudegeometric morphometricspotato tuber mothsexual variationTecia solanivorawing shape

Information

Type
Research Paper
Information
Bulletin of Entomological Research , Volume 100 , Issue 5 , October 2010 , pp. 529 - 541
Copyright
Copyright © Cambridge University Press 2010

Introduction

Wing morphology has been used largely in taxonomic, ecological and evolutionary studies in insects. (Moraes et al., Reference Moraes, Manfrin, Laus, Rosada, Bomfin and Sene2004; Carreira et al., Reference Carreira, Soto, Hasson and Fanara2006; Soto et al., Reference Soto, Hasson and Manfrin2008). It is an especially attractive study trait because of its importance in a wide variety of aspects of an insect's life, such as sexual and territorial display, foraging, defense mechanisms, thermal regulation, and the aerodynamics and the energetic costs of flight (Betts & Wootton, Reference Betts and Wootton1988; Wootton, Reference Wootton1992; Berwaerts et al., Reference Berwaerts, Van Dyck and Aerts2002, Reference Berwaerts, Aerts and Van Dyck2006). Wing configurations relate to the varying ecological roles and physiological constrains of flight that are particular to each sex. In general, female wing design would confer superior load bearing, by being larger and/or broader than that of males (e.g. Willmott & Ellington, Reference Willmott and Ellington1997). Wing morphology, shape in particular, can further be used as an indicator of changing – and often stressful – environmental conditions (Hoffmann et al., Reference Hoffmann, Collins and Woods2002, Reference Hoffmann, Woods, Collins, Wallin, White and McKenzie2005). Decreasing temperature, atmospheric pressure and oxygen availability, and increased solar radiation related to high-altitude environments can lead to morphological changes in insects (Hodkinson, Reference Hodkinson2005; Dillon & Frazier, Reference Dillon and Frazier2006). Previous studies have related the effect of temperature in wing morphology, observing larger wings and reduced wing loadings (body mass/wing area) at lower temperatures/higher altitudes (e.g. Miller, Reference Miller1991a; Norry et al., Reference Norry, Bubliy and Loeschchke2001; Altshuler & Dudley, Reference Altshuler and Dudley2002; Gilchrist & Huey, Reference Gilchrist and Huey2004).

Surprisingly, wing morphology studies of invasive insects have been rare (e.g. Loh et al., Reference Loh, David, Debat and Bitner-Mathé2008); however, these species are interesting models to study environmental adaptations. The environmental factors to which they are exposed act on the invasive population for relatively few generations and provide an opportunity to test for the speed of the response to new environmental gradients. In addition, new environments are often considered stressful, which may affect wing developmental canalization and, thereby, patterns of variations (Hoffmann et al., Reference Hoffmann, Woods, Collins, Wallin, White and McKenzie2005). Finally, invasive insects often have undergone rapid genetic change due to bottlenecks during introduction to and/or selection within the new environments. Wing developmental regulation may be affected by these genetic changes. For all these reasons, invasive insects are of great interest because not only does their study offer the prospect to understand the effects of environmental and genetics stress on wing shape patterns, but it also allows testing predictions of such patterns within relative short periods of time.

An interesting case of well-documented invasion is that of the potato tuber moth, Tecia solanivoraPovolny (Reference Povolny1973) (Lepidoptera: Gelechiidae), which has spread from Guatemala into Central America, northern South America and the Canary Islands during the past 30 years, attacking Solanum tuberosum L. tubers in the field and in storage and becoming one of the most damaging crop pests in these regions (Niño, Reference Niño2004; Puillandre et al., Reference Puillandre, Dupas, Dangles, Zeddam, Capdevielle-Dulac, Barbin, Torres-Leguizamon and Silvain2007). As with other invasive species, T. solanivora has suffered a significant genetic bottleneck during the invasion process (Puillandre et al., Reference Puillandre, Dupas, Dangles, Zeddam, Capdevielle-Dulac, Barbin, Torres-Leguizamon and Silvain2007) but has established successful populations in a wide variety of mountainous landscapes and a wide altitudinal range (from 200 m to >3500 m a.s.l.). In Ecuador, the potato moth is found from 2300 m to almost 3800 m a.s.l., its lowest elevation coinciding with some of its highest ranges in Central America. Another interesting characteristic of this invasive pest is that the rapid upward expansion of the agricultural frontier in the Ecuadorian highlands represents a constant opportunity for new populations of the potato moth to become established. It also involves climatic conditions that become more extreme the higher this frontier goes, thus posing new ecological, physiological and behavioral challenges to the moths.

Here, we use a geometric morphometric approach to test the hypothesis that insect wing morphology patterns among sexes and along altitudinal gradients are observed for invasive species despite their recent association to a new environment. To achieve this goal, we explored the variation in wing morphology between 12 populations of T. solanivora, at low and high altitude in the central highlands of Ecuador. With traditional morphometrics, biological measures are limited to linear distances, ratios or angles, thus failing to capture the geometrical relations between the anatomical points analyzed (Rohlf, Reference Rohlf1990). Geometric morphometrics, on the other hand, offer a more comprehensive approach to the study of shape through the multivariate statistical analysis of anatomical landmarks of biological homology (Bookstein, Reference Bookstein1991; Rohlf & Marcus, Reference Rohlf and Marcus1993; Adams et al., Reference Adams, Rohlf and Slice2004). It preserves the information about the relative spatial arrangement of the data throughout the analysis (Zelditch et al., Reference Zelditch, Swiderski, Sheets and Fink2004), making it possible to find and analyze shape variations in the organisms within and between populations (Walker, Reference Walker2000). Moreover, geometric morphometric tools present the advantage of laying results that not only have high statistical power but also have easily visualized results, helping with their interpretation and communication (Rohlf & Marcus, Reference Rohlf and Marcus1993; Adams et al., Reference Adams, Rohlf and Slice2004; Zelditch et al., Reference Zelditch, Swiderski, Sheets and Fink2004). In this study, we first characterized sexual dimorphism in wing shape and then investigated whether the populations of the invasive moth at higher altitudes show changes in the wing morphology that can differentiate them from the populations at lower altitudes. We expect to find that previously observed sexual dimorphism in coloration and shape extends to variations of the wing shape between the males and females of T. solanivora. We also expect that if there is variation of the wing shape between altitudinal ranges, there would be a relation with environmental factors characteristic of each altitude.

Materials and methods

Study area and background

Tecia solanivora attacks potato tubers both in the field and in store, burrowing deep feeding tunnels into the tuber (Pumisacho & Sherwood, Reference Pumisacho and Sherwood2002). Adults show sexual dimorphism in both size and coloration, with generally bigger, light-brown colored females versus smaller, deep-brown colored males (EPPO/OEPP, 2005; see Pumisacho & Sherwood, Reference Pumisacho and Sherwood2002; Barragán, Reference Barragán2005 for further details on T. solanivora's biology). In Ecuador, potatoes are cultivated in the highlands, where climatic conditions are rather stable across seasons and allow year round growth of this crop (Dangles et al., Reference Dangles, Carpio, Barragan, Zeddam and Silvain2008). Mean temperature has little seasonal variation: in the upper elevations, the warmest month averages 18°C and the coolest month averages 12°C (Cáceres et al., Reference Cáceres, Mejía and Ontaneda1998). Instead, daily temperature variation is much higher: air temperature can vary daily between 0° and 30°C. Under such favorable climatic conditions and extended food resources available, populations of the potato tuber moth are active throughout the year (Pollet et al., Reference Pollet, Barragán, Zeddam and Lery2003; Pruna Reference Pruna2004; Dangles et al., Reference Dangles, Carpio, Barragan, Zeddam and Silvain2008).

Sampling

Populations of T. solanivora were sampled in 12 sites located on the western slope in the central Ecuadorian highlands (table 1). Although moth population occurs continuously along the elevation gradient, moth sampling was designed in order to explore two altitudinal extremes: sites below 2750 m and sites above 3000 m (see fig. 1 and fitted bimodal distribution). This allowed us sampling populations living under contrasting environmental (especially thermal) conditions. Intermediate altitude sites were discarded as they could potentially be influenced by both high and low altitude climates indistinctively. This design implies that populations were treated as nested samples within altitudinal classes as low or high altitude, allowing the use of discriminant analyses with a priori established groups. Sampling took place between October of 2006 and January of 2007 using dome traps baited with species-specific pheromones (Pherobank, Wageningen, The Netherlands) placed at about 1 m height in potato fields. Pheromone traps for T. solanivora make adult male sampling comparatively easier to female sampling. For this reason, only male individuals were used to test for wing shape variation related to altitude. However, during January of 2006, we also collected adult moths from potato sacks in the localities of San Pablo and San Simón (table 1) that we used to study sexual variation in wing morphology. Preliminary exploration showed that there was no bias induced by pheromone trapping in the samples. Sampled specimens were kept in individual vials with 90% ethanol.

Fig. 1.

Methodology used to group the study sites into two altitude classes. Frequency distributions of altitudes were decomposed into Gaussian distributions using a combination of a Newton-type method and expectation maximization algorithms. The mean (μ) and variance (σ) of the altitude distribution for both groups of sites were calculated using the ‘mixture distribution’ package of R software (R Development Core Team, 2008). For low altitude sites, μ=2548 and σ=108, and for high altitude sites, μ=3075 and σ=77.

Table 1.

Data on localities where Tecia solanivora specimens were collected for this study (localities considered for sex-related variation analysis are marked with*).

Detail of number of individuals used for each locality and natural abundance (calculated based on the number of male catch in pheromone traps per week during the period November 2006–March 2007, see Dangles et al., Reference Dangles, Carpio, Barragan, Zeddam and Silvain2008 for more details).

Specimen's preparation and data collection

In all cases, forewings of adult moths were separated from the body at the base and the scales removed by gently sweeping them away with a fine brush in a Petri dish containing 90% ethanol. In some cases, the base of the wings was damaged during extraction, due to fragility caused by being retained in the pheromone traps for long durations. This also affected the quality of the whole body, making it impossible to obtain other morphometrics related to body measures. Photographs of each wing were taken using a digital camera (Powershot S40, Canon, Tokyo, Japan) attached to a stereo microscope (Leica M275, Bannockburn, IL, USA) with a black background and a millimeter grid for size reference.

Following Zelditch's et al (Reference Zelditch, Swiderski, Sheets and Fink2004) criteria of homology and comprehensive coverage, a total of 18 landmarks were identified on each wing and their coordinates obtained using tpsDig 2.10 (Rohlf, Reference Rohlf2006). Of these, 17 correspond to type I landmarks (whose claimed homology from case to case is supported by morphological evidence, e.g. vein-vein and vein-edge intersections) and one to type II landmarks (whose claimed homology from case to case is supported only by geometric evidence, e.g. structure tip), according to Bookstein's (Reference Bookstein1991) classification (fig. 2). Landmarks at the base of the wing were not considered because wing veins in this area are quite thick, and positioning of landmarks was not easily repeatable; moreover, due to sampling and preparation constrains, wing base of some specimens was not in suitable condition. All landmark coordinates were superimposed with a Generalized Procrustes Analysis (GPA) algorithm using tpsSmall 1.20 (Rohlf, Reference Rohlf2003) to remove the effect of scale, position and orientation from the coordinates (Rohlf, Reference Rohlf1999; Zelditch et al., Reference Zelditch, Swiderski, Sheets and Fink2004). The centroid size (CS), calculated as the square root of the sum of squared distances from the landmarks to their centroid (Bookstein, Reference Bookstein1991), was obtained as an estimator of size. The coordinates were also divided by CS to obtain size-free, though not allometry-free, data (Baylac et al., Reference Baylac, Villemant and Simbolotti2003). Adjusted coordinates and CS were averaged between the right and left wings of each individual when both wings were available, in order to reduce measurement error (Chris Klingenberg, personal communication). These repeated measurements also provided an estimate of the digitizing error of >1 mm (see Arnqvist & Martensson, Reference Arnqvist and Martensson1998).

Fig. 2.

Right wing of a male specimen of Tecia solanivora depicting the position of the 18 landmarks used for the analyses. Vein nomenclature from Borror et al. (Reference Borror, De Long and Triplehorn1981).

Morphometric and statistical analyses

Projection of data from Kendall's shape space into a linear tangent space is necessary for using standard multivariate analyses (Rohlf, Reference Rohlf1999). We confirmed that data variation due to this projection was small enough for data set to correctly represent the original distribution in the shape space. A regression of the Procrustes distance (Pd) from the consensus configuration in the shape space on the PD calculated on the linear tangent space was performed using tpsSmall.

Sex related variation

To evaluate wing shape dimorphism, shape variables (also called partial warps: Rohlf, Reference Rohlf, Marcus, Bello and García-Valdecasas1993) were used for a principal components analysis called relative warps analysis (RWA), performed with tpsRelw 1.45 (Rohlf, Reference Rohlf2007). Using the total shape variables (both uniform and non-uniform components) the variation between male and female wings was assessed with a multivariate analysis of variance (MANOVA) and a canonical variate analysis (CVA) (following Cardini & O'Higgins, Reference Cardini and O'Higgins2004). Statistical significance was evaluated with a two-group multivariate permutation analysis (N=10,000). All statistical analyses were performed using PAST 1.6 (Hammer et al., Reference Hammer, Harper and Ryan2001) and groups were established a priori to maximize the differences (Rohlf et al., Reference Rohlf, Loy and Corti1996; Cavalcanti et al., Reference Cavalcanti, Monteiro and Duarte-L.1999). To visualize the existing variation, a thin-plate spline analysis was performed using tpsSpline (Rohlf, Reference Rohlf2004) with the consensus configurations of each sex on a deformation grid based on the reference configuration calculated from the total sample (n=112).

Altitude related variation

A RWA was also used for evaluating wing shape variation among high and low altitude groups with the overall sample from the 12 localities. A second RWA was performed using the consensus configuration of each locality to give an equal weight to each one of the localities instead of to each specimen (Rohlf et al., Reference Rohlf, Loy and Corti1996). A combined MANOVA/CVA along with a multivariate permutation analysis (N=10,000) were run to test the significance of the separation between the two groups. Visual analysis of the variation was performed between the consensus configurations of each locality on a deformation grid based on the reference configuration calculated from the total sample (n=281) using tpsSpline. Although we were aware that cross-validation would have improve our discriminant analyses results and made them more robust (e.g. Klimov et al., Reference Klimov, Bochkov and O'Connor2006), limitation of the number of available specimens prevented us from doing it. However, random removal of 1–5 individuals did not change the significance of our results.

Allometry and allometric effect

Forewing morphometry can be taken as a reliable body size index in Lepidoptera (Miller, Reference Miller1991b). Consequently, to test for overall body size differences between sexes and between specimens from different sites, we performed a one-way ANOVA using the isometric estimator centroid size (CS). This was obtained from the coordinates' data with tpsSmall and averaged across left and right wings of individuals (see Gomez & Monteiro, Reference Gomez and Monteiro2008). We did not use thorax mass for size related exploration, as recent findings show that it decreases with age in some butterflies (Stjernholm et al., Reference Stjernholm, Karlsson and Boggs2005), and our sampling method did not allow for determining age in the captured individuals. Additionally, we ran two analyses using tpsRegr 1.31 (Rohlf, Reference Rohlf2005) to examine the influence of allometry on any variation found between the studied groups. First, a multivariate regression of the total shape variables, using the CS logarithm (CS loge) as the independent variable. Second, we ran a multivariate analysis of covariance (MANCOVA), using the total shape variables as dependent variables and the CS loge as the independent variable.

Environmental impact on wing morphology

Generalized Linear Model analysis (GLM) were used to test the effect of environmental variables on mean values of two wing morphometric variables (size: CS, and shape: CV1 scores) for T. solanivora populations at each sampling point. The Poisson log-linear model included the following (Log10+1) transformed variables: region, altitude, mean temperature, min temperature, max temperature, moth density and relative humidity. These explanatory variables were chosen based on their potential use as surrogate measures of processes and factors that might have a direct effect on wing morphology along the gradient. The analysis was run using mean, minimum and maximum temperatures, but the most significant results were always obtained with mean temperature. We also included a ‘site’ variable to allow within-site comparisons while controlling for variation resulting from unmeasured site-specific parameters. Change in wing morphometrics due to each factor was modeled considering each factor independently and in combination with other factors, including biologically reasonable two-way interactions and squared variables. The more parsimonious model was identified using the Akaike's Information Criterion (AIC: see Venables & Ripley, Reference Venables and Ripley2002) in likelihood ratio tests to find the difference between the initial model (including all terms) and the reduced model (in which one effect term was removed). All analyses were performed using the mass library for R (R Development Core Team, 2008).

Results

Wing shape variation and sexual dimorphism

The first three relative warps, which are the principal components of the distribution of wing shapes, explain 50.92% of the total wing shape variation between the 63 female and 49 male adult moths studied (rw1=24.62%, rw2=14.94%, rw3=11.35%). A plot of rw1 and rw2 showed a tendency of females to be grouped at the negative side of the first relative warp and males at the positive side; deformation grids for both extremes showed slender wings on the positive end and wider wings at the negative end (fig. 3). Sexual dimorphism in wing shape was highly significant (MANOVA/CVA of shape variables, sex, Wilks' lambda=0.129, F=16.62, df=32, P<0.001). Both sexes were effectively separated along the discriminant axis (fig. 4), with 99.11% of correctly classified specimens, as did too the two-group multivariate permutation analysis (Mahalanobis distance=0.4944, P<0.0001). The grid deformations of the consensus configuration of each sex from the reference configuration revealed that male and female wings were mostly differentiated by their width, and not so much by their longitude (fig. 5). This result is consistent with what was observed with the RWA.

Fig. 3.

Upper part: Scatter plot for the RWA on male and female wings. Distribution of specimens for (a) x: rw1, y: rw2 and (b) x: rw2, y: rw3 (males: n=49, in grey; females: n=63, in black). Lower part: Deformation grids for both negative and positive extremes of the three relative warps: rw1, rw2 and rw3.

Fig. 4.

Histogram of the specimens grouped by sex (males: n=49, in grey; females: n=63, in black) along the Discriminant axis. Percent of individuals correctly assigned to their original groups: 99.11%.

Fig. 5.

Deformation grids for the consensus configurations and displacement vectors of a. males and b. females, from the reference configuration calculated on the overall sample (n=112, top). Grids and vectors obtained using tpsSpline, exaggerated by a factor of 5 for better visualization of changes.

Wing shape differentiation among low and high altitude populations

A total of 281 adult males were collected from our pheromone sampling: 192 corresponding to localities underneath 2800 m a.s.l. (low altitude group) and 89 to localities above 3000 m a.s.l. (high altitude group) (table 1). The RWA for all the individuals of the 12 localities showed that 45.33% of the overall variation in wing shape was explained by the first three relative warps: rw1=21.03%, rw2=13.10%, rw3=11.20% (fig. 6). As expected, a RWA for the consensus configurations of each site showed a better differentiation of the groups, with a 75.51% of the overall variation explained by the first three relative warps (data not shown). Deformation grids for rw1 showed that wing shapes narrow at the positive extreme, and widen at the negative extreme. High altitude specimens tended to group at the positive end of the first relative warp, while low altitude specimens grouped around the center and negative end. The effect of altitude on the wing shape variation was highly significant (MANOVA/CVA of shape variables, altitude, Wilks' Lambda=0.565, F=5.958, df=32, P<0.001). Low and high altitude groups were effectively separated along the discriminant axis (fig. 7), with 83.99% correctly classified specimens. Similar results were obtained with the multivariate permutation (Mahalanobis distance=0.1124 and P<0.0001). TpsSpline deformation grids and displacement vectors from the consensus configurations show that, for the high altitude group, wing shape is narrower than wing shape of the low altitude group, especially at the middle part of the wing (fig. 8).

Fig. 6.

Scatter plot for the RWA on males from the 12 localities studied. Breakdown of the distribution of specimens for each locality on x: rw1, y: rw2. (a) San Pablo, (b) Pumamaqui, (c) San Simón, (d) Salache, (e) San Miguelito, (f) Salache, (g) La Hoya, (h) Anchilibí, (i) San Francisco, (j) Palama Bajo, (k) Carbón Chimipamba, (l) Palama Medio. The figure shows a rw1 and rw2 plot breakdown of each site for better visualization of the distribution.

Fig. 7.

Histogram of the specimens grouped by altitude (high altitude: grey, low altitude: black) along the discriminant axis. Percent of individuals correctly assigned to their original groups: 83.99%.

Fig. 8.

Deformation grids for the consensus configurations and displacement vectors of specimens from low and high altitude sites, from the reference configuration calculated on the overall sample (n=281, top). Grids and vectors obtained using tpsSpline, exaggerated by a factor of ten for better visualization of changes.

Allometry and allometric effect

One-way ANOVA was highly significant for centroid size variation (i.e. variation of wing size as calculated through the isometric estimator of size, centroid size) between sexes (F=19.24, df=1, P<0.001) and between altitudinal sites (F=9.979, df=1, P=0.0017). Female wings were generally larger than male wings (fig. 9), and high altitude wings tended to present higher values for centroid size (i.e. larger wings; fig. 10). Regression of shape variables on size for male and female wings showed significant allometric relationship (Wilks' Lambda=0.318, F=5.288, df=32, P<0.001), but MANCOVA indicated allometric slopes were not significantly different (Wilks' Lambda=0.732, F=0.401, df=64, P=1.00) and, when size was held constant, shape variables proved to differ significantly (Wilks' lambda=0.151, F=13.697, df=32, P<0.001). When looking for allometric effect on the altitude group's variation, similar outcomes were obtained. The regression of shape variables on size showed highly significant values (Wilks' Lambda=0.618, F=4.797, df=32, P<0.001) and again no difference was observed between allometric slopes with MANCOVA (Wilks' Lambda=0.857, F=0.614, df=63, P=0.992). The difference observed on wing shape between low and high altitude groups remained significant once size was held constant (Wilks' lambda=0.151, F=13.697, df=32, P<0.001).

Fig. 9.

Regression of the natural logarithm of centroid size (CS loge) on the first canonical axis (CV1) of the shape variables of male (n=49, in grey) and female (n=63, in black) wings.

Fig. 10.

Regression of the natural logarithm of centroid size (CS loge) on the first canonical axis (CV1) of the shape variables of high altitude (n=89, in grey) and low altitude (n=192, in black) wings.

Environmental impact on wing morphology

Of the eight factors included in the GLM analysis, only altitude significantly affected the two wing morphometric variables (CS: ΔAIC=3.1, P=0.035, CV1 scores: ΔAIC=2.5, P=0.045). The following most important predictors of wing morphometrics were ‘mean temperatures’ and ‘moth abundance’ but these were not significant (P>0.12). We found neither significant effect of ‘site’ nor ‘region’ on wing morphometrics (P>0.66).

Discussion

Sexual dimorphism in wing shape

Our results not only confirm a female-biased sexual dimorphism in size for T. solanivora but also show dimorphism in the shape of the wings. Several studies have shown that sexual dimorphism in life-history traits of insects, such as size and shape, may be adaptive (reviewed in Nylin & Gotthard, Reference Nylin and Gotthard1998). The larger size of females may be favored by natural selection for increased fecundity (Gilchrist, Reference Gilchrist1990; Calvo & Molina, Reference Calvo and Molina2005) and suggests a higher fitness importance of size advantages for female's fecundity compared to those offered to males (Nylin & Gotthard, Reference Nylin and Gotthard1998). Small size of males can be the result of sexual selection for rapid development that leads to early eclosion (Gilchrist, Reference Gilchrist1990). Although populations with high overlapping of generations (as is the case of populations studied here) were not usually expected to present protandry (Singer, Reference Singer1982; Nylin et al., Reference Nylin, Wiklund, Wickman and Garcia-Barros1993), more recent findings show that aseasonal moths can present it (Muralimohan & Srinivasa, Reference Muralimohan and Srinivasa2008). Elongated wings observed in males can be related to the longer flying periods they require for finding sexually active females (Douwes, Reference Douwes1976; Kingsolver, Reference Kingsolver1983; Shreeve, Reference Shreeve1984; Gilchrist, Reference Gilchrist1990). The observed tendency of larger wing size of female wings than that of males is probably a consequence of the wider wing configuration in females; similar wing configurations were observed in other Lepidoptera, the speckled wood butterfly, Pararge aegeria (Berwaerts et al., Reference Berwaerts, Van Dyck and Aerts2002).

Wing shape variation with altitude

Our study confirms and extends predictions of morphological changes with altitude to an invasive species. We observed significant differences in size and wing morphology between populations from localities under 2750 m and above 3050 m a.s.l., with larger moths at higher altitudes and most of the variation in shape occurring on the width of the middle region of the wings; higher altitude populations tend to have slender wings, while the lower altitude populations show broader wings by comparison. One of the hypotheses behind our research was that the differences in temperature we observed in the different sampled sites might somehow affect the wing morphology of the moths. This hypothesis was sustained by previous studies on D. melanogaster (Dillon & Frazier, Reference Dillon and Frazier2006; Frazier et al., Reference Frazier, Harrison, Kirkton and Roberts2008) and D. suboscura (Gilchrist & Huey, Reference Gilchrist and Huey2004), which suggest aerodynamic reasons for larger (and proportionally longer) wings at lower temperatures and lower atmospheric pressure. Slender wings might compensate for the environmental constraints on flight at higher altitudes, since they can help in reducing the energetic requirement for flight as well as dragging forces while yielding greater aerodynamic forces (Wootton, Reference Wootton1992; Berwaerts et al., Reference Berwaerts, Van Dyck and Aerts2002; Frazier et al., Reference Frazier, Harrison, Kirkton and Roberts2008 and references therein). We also expected moths with lager body sizes at lower temperatures – and, therefore, higher altitudes – as is predicted by Bergmann's rule (see Miller, Reference Miller1991a).

Despite the fact that our observations fit the predictions for body size and wing morphology variations at different altitudes, we found no evidence of a strong relationship between such changes and the decreasing temperature at higher elevations. The GLM analysis showed that the only significant factor behind the observed variation was altitude; and, although the second-most important predictor was mean temperature, it was not significant. However, altitude is a combination of temperature, atmospheric pressure, humidity and solar radiation variables, each of them potentially exerting separate selective pressures on wing shape and size.

Wing shape of an invasive species

The invasive potato moth developed altitudinal differences in wing morphology consistent with patterns observed for non-invasive species. Our results show that such patterns are able to develop in a short period of time. To our knowledge, the T. solanivora's wing size altitudinal cline in Ecuador is the only described example of an invasive insect rapidly developing altitudinal Bergmann clines in their invasive range (D. subobscura and other invasive Drosophila clines are latitudinal). Below, we discuss three hypotheses that may explain such a process.

The first hypothesis is a rapid genetic adaptation. In case of genetic adaptation, larger and broader thoraxes, smaller abdomens and higher wing aspect ratios may reflect increased dispersal ability of high-altitude potato moth populations – which are also those that are the most recently established (see Hughes et al., Reference Hughes, Dytham and Hill2007; Breuker et al., Reference Breuker, Brakefield and Gibbs2007). However, the rapidity of the establishment of the cline (ten years for T. solanivora versus 20 years for D. subobscura) and the arguments detailed below question the genetic nature of this cline. The second, and most likely, hypothesis is that these patterns are a purely plastic response, and the differences between altitudes are not genetically based as observed for another invasive insect species (Loh et al., Reference Loh, David, Debat and Bitner-Mathé2008). Potato moths show a significant decrease in development time at higher temperatures (Notz, Reference Notz1996; Dangles et al., Reference Dangles, Carpio, Barragan, Zeddam and Silvain2008) and a reduced number of generations at high altitudes (Dangles et al., Reference Dangles, Carpio, Barragan, Zeddam and Silvain2008), which can directly relate to size variation between low and high altitudes. Wing morphology can also respond to environmental cues, in particular during the developmental period, when genes that need to be activated during wing development can be altered by environmental conditions (Hoffmann et al., Reference Hoffmann, Woods, Collins, Wallin, White and McKenzie2005 and references therein). Another argument against genetically-based variation is the absence of non-altitudinal differentiation; we indeed ruled out geographic distance as a variability predictor because sampled localities did not differ significantly between provinces of origin (i.e. the largest geographical distance; data not shown). This lack of significance was also reinforced by the results of GLM showing that province and site were not influencing factors in morphological variation. The third hypothesis is that, in addition to being plastic, the response may be non-adaptive (i.e. it does not provide any advantage to the insect to have long wings at high altitude) but due to purely developmental and physiological mechanisms (see Blanckenhorn & Demont, Reference Blanckenhorn and Demont2004 for a review). For example, a differential ratio between cell division and cell growth increment with temperature lead to smaller adults at higher temperature (with no adaptive value of this reaction norm). Another physiological explanation for these clines involves slower oxygen diffusion at increasing temperatures in large cells compared with their oxygen consumption, possibly inducing malfunction of such cells at higher temperatures.

Acknowledgements

Funding for this work was provided by Project Investigation Grant C13-019 from the Pontificia Universidad Católica del Ecuador. We thank Carlos Carpio for his help in the field phase and Jonathan Parker Bedrava for his help in making the text grammatically correct and clear. We are also grateful to the sub-editor and two anonymous reviewers for their helpful comments on a previous version of the manuscript.

References

Adams, D.C., Rohlf, F.J. & Slice, D.E. (2004) Geometric morphometrics: ten years of progress following the ‘revolution’. Italian Journal of Zoology 46, 180–194.Google Scholar
Altshuler, D.L. & Dudley, R. (2002) The ecological evolutionary interface of hummingbird flight physiology. Journal of Experimental Biology 205, 2325–2336.CrossRefGoogle ScholarPubMed
Arnqvist, G. & Martensson, T. (1998) Measurement error in geometric morphometrics: empirical strategies to assess and reduce its impact on measure of shape. Acta Zoologica Academiae Scientarum Hungaricae 44, 73–96.Google Scholar
Barragán, A. (2005) Identificación, Biología y Comportamiento de las Polillas de la papa en el Ecuador. 12 pp. Quito, Ecuador, Boletín PROMSA, MAG-PUCE.Google Scholar
Baylac, M., Villemant, C. & Simbolotti, G. (2003) Combining geometric morphometrics with pattern recognition for the investigation of species complexes. Biological Journal of the Linnean Society 80, 89–98.CrossRefGoogle Scholar
Berwaerts, K., Van Dyck, H. & Aerts, P. (2002) Does flight morphology relate to flight performance? An experimental test with the butterfly Pararge aegeria. Functional Ecology 16, 484–491.CrossRefGoogle Scholar
Berwaerts, K., Aerts, P. & Van Dyck, H. (2006) On the sex-specific mechanisms of butterfly flight: flight performance relative to flight morphology, wing kinematics, and sex in Pararge aegeria. Biological Journal of the Linnean Society 89, 675–687.CrossRefGoogle Scholar
Betts, C.R. & Wootton, R.J. (1988) Wing shape and flight behaviour in butterflies (Lepidoptera: Papilionoidea and Hesperioidea): a preliminary analysis. Journal of Experimental Biology 138, 271–288.CrossRefGoogle Scholar
Blanckenhorn, W.U. & Demont, M. (2004) Bergmann and converse Bergmann latitudinal clines in arthropods: two ends of a continuum? Integrative and Comparative Biology 44, 413–424.CrossRefGoogle ScholarPubMed
Bookstein, F.L. (1991) Morphometric Tools for Landmark Ddata: Geometry and Biology. 435 pp. New York: Cambridge University Press.Google Scholar
Borror, D.J., De Long, D.M. & Triplehorn, C.A. (1981) Introduction to the Study of Insects. 5th edn.827 pp. Philadelphia, PA, USA, CBS College Publishing.Google Scholar
Breuker, C.J., Brakefield, P.M. & Gibbs, M. (2007) The association between wing morphology and dispersal is sex-specific in the glanville fritillary butterfly Melitaea cinxia (Lepidoptera: Nymphalidae). European Journal of Entomology 104, 445–452.CrossRefGoogle Scholar
Cáceres, L., Mejía, R. & Ontaneda, G. (1998) Evidencias del cambio climático en Ecuador. Bulletin Français des Etudes Andines 27, 547–556.CrossRefGoogle Scholar
Calvo, D. & Molina, J. (2005) Fecundity-body size relationship and other reproductive aspects of Streblote panda (Lepidoptera: Lasiocampidae). Annals of the Entomological Society of America 98, 191–196.CrossRefGoogle Scholar
Cardini, A. & O'Higgins, P. (2004) Patterns of morphological evolution in Marmota (Rodentia, Sciuridae): geometric morphometrics of the cranium in the context of marmot phylogeny, ecology and conservation. Biological Journal of the Linnean Society 82, 385–407.CrossRefGoogle Scholar
Carreira, V.P., Soto, I.M., Hasson, E. & Fanara, J.J. (2006) Patterns of variation in wing morphology in the cactophiilic Drosophila buzzatii and its sibling D. kowpferae. Journal of Evolutionary Biology 19, 1275–1282.CrossRefGoogle Scholar
Cavalcanti, M.J., Monteiro, L.R. & Duarte-L., P.R. (1999) Landmark-based morphometric analysis in selected species of Serranid fishes (Perciformes: Teleostei). Zoological Studies 38, 287–294.Google Scholar
Dangles, O., Carpio, C., Barragan, A.R., Zeddam, J.L. & Silvain, J.F. (2008) Temperature as a key driver of ecological sorting among invasive pest species in the tropical Andes. Ecological Applications 18, 1795–1809.CrossRefGoogle ScholarPubMed
Dillon, M.E. & Frazier, M.R. (2006) Drosophila melanogaster locomotion in cold thin air. Journal of Experimental Biology 209, 364–371.CrossRefGoogle ScholarPubMed
Douwes, P. (1976) Activity in Heodes virgaureae (Lepidoptera: Lycaenidae) in relation to air temperature, solar radiation, and time of day. Oecologia 22, 287–298.CrossRefGoogle Scholar
EPPO/OEPP (2005) Data sheet on quarantine pests: Tecia solanivora. EPPO/OEPP Bulletin 35, 399–401.CrossRefGoogle Scholar
Frazier, M.R., Harrison, J.F., Kirkton, S.D. & Roberts, S.P. (2008) Cold rearing improves cold-flight performance in Drosophila via changes in wing morphology. Journal of Experimental Biology 221, 2116–2122.CrossRefGoogle Scholar
Gilchrist, G.W. (1990) The consequences of sexual dimorphism in body size for butterfly flight and thermoregulation. Functional Ecology 4, 475–487.CrossRefGoogle Scholar
Gilchrist, G.W. & Huey, R.B. (2004) Plastic and genetic variation in wing loading as a function of temperature within and among parallel clines in Drosophila suboscura. Integrative and Comparative Biology 44, 461–470.CrossRefGoogle Scholar
Gomez, J.L.JR. & Monteiro, L.R. (2008) Morphological divergence patterns among populations of Poecilia vivipara: test of an ecomorphological paradigm. Biological Journal of the Linnean Society 93, 799–812.CrossRefGoogle Scholar
Hammer, Ø., Harper, D.A.T. & Ryan, P.D. (2001) PAST: Paleontological Statistics software package for education and data analysis. Palaeontologia Electronica 4, 9 pp.Google Scholar
Hodkinson, I.D. (2005) Terrestrial insects along elevation gradients: species and community responses to altitude. Biology Reviews 80, 489–513.CrossRefGoogle ScholarPubMed
Hoffmann, A.A., Collins, E. & Woods, R.E. (2002) Wing shape and wing size changes as indicators of environmental stress in Helicoverpa punctigera (Lepidoptera: Noctuidae) moths: comparing shifts in means, variances and asymmetries. Environmental Entomology 31, 965–971.CrossRefGoogle Scholar
Hoffmann, A.A., Woods, R.E., Collins, E., Wallin, K., White, A. & McKenzie, J.A. (2005) Wing shape versus asymmetry as an indicator of changing environmental conditions in insects. Australian Journal of Entomology 44, 233–243.CrossRefGoogle Scholar
Hughes, C.L., Dytham, C. & Hill, J.K. (2007) Modelling and analyzing evolution of dispersal in populations at expanding range boundaries. Ecological Entomology 32, 437–445.CrossRefGoogle Scholar
Kingsolver, J.G. (1983) Ecological significance of flight activity in Colias butterflies: implications of reproductive strategy and population structure. Ecology 64, 546–551.CrossRefGoogle Scholar
Klimov, P.B., Bochkov, A.V. & O'Connor, B.M. (2006) Host specificity and multivariate diagnostic of cryptic species in predacious cheyletid mites of the genus Cheletophyes (Acari: Cheyletidae) associated with large carpenter bees. Biological Journal of the Linnean Society 87, 45–58.CrossRefGoogle Scholar
Loh, R., David, J.R., Debat, V. & Bitner-Mathé, B.C. (2008) Adaptation to different climates results in divergent phenotypic plasticity of wing size and shape in an invasive drosophilid. Journal of Genetics 87, 209–217.CrossRefGoogle Scholar
Miller, E.E. (1991a) Positive relation between body size and altitude of capture site in tortricid moths (Tortricidae). Journal of the Lepidopterists' Society 45, 66–67.Google Scholar
Miller, E.E. (1991b) Body size in North American Lepidoptera as related to geography. Journal of the Lepidopterists' Society 45, 158–168.Google Scholar
Moraes, E.M., Manfrin, M.H., Laus, A.C., Rosada, R.S., Bomfin, S.C. & Sene, F.M. (2004) Wing shape heritability and morphological divergence of the sibling species Drosophila mercatorum and Drosophila paranaensis. Heredity 92, 466–473.CrossRefGoogle ScholarPubMed
Muralimohan, K. & Srinivasa, Y.B. (2008) Occurrence of protandry in an aseasonal multivoltine moth: Implications for body-size evolution. Current Science 94, 513–518.Google Scholar
Niño, L. (2004) Revisión sobre la polilla de la papa Tecia solanivora en Centro y Sudamérica. Suplemento Revista Latinoamericana de la papa, 4–22.Google Scholar
Norry, F.M., Bubliy, O.A. & Loeschchke, V. (2001) Developmental time, body size and wing loading in Drosophila buzzatii from lowland and highland populations in Argentina. Hereditas 135, 35–40.CrossRefGoogle ScholarPubMed
Notz, A. (1996) Influencia de la temperatura sobre la biología de Tecia solanivora Povolny (Lepidoptera: Gelechiidae) criadas en tubérculos de papa Solanum tuberosum L. Boletín Entomología Venezolana 11, 49–54.Google Scholar
Nylin, S. & Gotthard, K. (1998) Plasticity in life-history traits. Annual Review of Entomology 43, 63–83.CrossRefGoogle ScholarPubMed
Nylin, S., Wiklund, C., Wickman, P.O. & Garcia-Barros, E. (1993) Absence of tradeoffs between sexual size dimorphism and early male emergence in a butterfly. Ecology 74, 1414–1427.CrossRefGoogle Scholar
Pollet, A., Barragán, A., Zeddam, J.-L. & Lery, X. (2003) Tecia solanivora, a serious biological invasion of potato cultures in South America. International Pest Control 45, 139–144.Google Scholar
Povolny, D. (1973) Scrobipalpopsis solanivora sp. n. –A new pest of potato (Solanum tuberosum) from Central America. Acta Universitatis Agriculturae, Facultas Agronomica 21, 143–146.Google Scholar
Pruna, A.M. (2004) Comportamiento y control de polillas de la papa (Tecia solanivora, Symmetrischema tangolias y Phthorimaea operculella) en el valle de Salcedo. PhD thesis, Technical University of Cotopaxi, Latacunga, Ecuador.Google Scholar
Puillandre, N., Dupas, S., Dangles, O., Zeddam, J.-L., Capdevielle-Dulac, C., Barbin, K., Torres-Leguizamon, M. & Silvain, J.F. (2007) Genetic bottleneck in invasive species: the potato tuber moth adds to the list. Biological Invasions 10, 319–333.CrossRefGoogle Scholar
Pumisacho, M. & Sherwood, S. (2002) El Cultivo de la papa en Ecuador. 229 pp. Quito, Ecuador, INIAP and CIP.Google Scholar
R Development Core Team (2008) R: A language and environment for statistical computing. Vienna, Austria, Foundation for Statistical Computing.Google Scholar
Rohlf, F.J. (1990) Morphometrics. Annual Review of Ecology and Systematics 21, 299–316.CrossRefGoogle Scholar
Rohlf, F.J. (1993) Relative warp analysis and an example to its application to mosquito wings. pp. 131–159in Marcus, L.F., Bello, E. & García-Valdecasas, A. (Eds) Contributions to Morphometrics. Madrid, Spain, Museo Nacional de Ciencias Naturales, CSIC.Google Scholar
Rohlf, F.J. (1999) Shape statistics: Procrustes superimposition and tangent spaces. Journal of Classification 16, 197–223.CrossRefGoogle Scholar
Rohlf, F.J. (2003) TpsSmall. Version 1.20. New York, Department of Ecology and Evolution, State University of New York. Available at http://morph.bio.sunysb.edu/morph/index.html (accessed 2 July 2008).Google Scholar
Rohlf, F.J. (2004) TpsSpline. Thin-plate spline, Version 1.20. New York, Department of Ecology and Evolution, State University of New York. Available at http://morph.bio.sunysb.edu/morph/index.html (accessed 2 July 2008).Google Scholar
Rohlf, F.J. (2005) TpsRegr. Version 1.31. New York, Department of Ecology and Evolution, State University of New York. Available at http://morph.bio.sunysb.edu/morph/index.html (accessed 2 July 2008).Google Scholar
Rohlf, F.J. (2006) TpsDig. Version 2.10. New York, Department of Ecology and Evolution, State University of New York. Available at http://morph.bio.sunysb.edu/morph/index.html (accessed 2 July 2008).Google Scholar
Rohlf, F.J. (2007) TpsRelw. Version 1.45. New York, Department of Ecology and Evolution, State University of New York. Available at http://morph.bio.sunysb.edu/morph/index.html (accessed 2 July 2008).Google Scholar
Rohlf, F.J. & Marcus, L.F. (1993) A revolution in morphometrics. Trends in Ecology and Evolution 8, 129–132.CrossRefGoogle Scholar
Rohlf, F.J., Loy, A. & Corti, M. (1996) Morphometric analysis of Old World Talpidae (Mammalia, Insectivora) using partial warp scores. Systematic Biology 45, 344–362.CrossRefGoogle Scholar
Shreeve, T.G. (1984) Habitat selection, mate location, and microclimatic constraints on the activity of the speckled wood butterfly, Pararge aegeria. Oikos 42, 371–377.CrossRefGoogle Scholar
Singer, M.C. (1982) Sexual selection for small size in male butterflies. American Naturalist 119, 440–443.CrossRefGoogle Scholar
Soto, I.M., Hasson, E.R. & Manfrin, M.H. (2008) Wing morphology is related to host plants in cactophilic Drosophila gouveai and Drosophila antonietae (Diptera, Drosophilidae). Biological Journal of the Linnean Society 95, 655–665.CrossRefGoogle Scholar
Stjernholm, F., Karlsson, B. & Boggs, C.L. (2005) Age-related changes in thoracic mass: possible reallocation of resources to reproduction in butterflies. Biological Journal of the Linnean Society 86, 363–380.CrossRefGoogle Scholar
Taylor, P.D. & Merriam, G. (1995) Wing morphology of a forest damselfly is related to landscape structure. Oikos 73, 43–48.CrossRefGoogle Scholar
Venables, W.N. & Ripley, B.D. (2002) Modern Applied Statistics with S. 4th edn.495 pp. New York, Springer.CrossRefGoogle Scholar
Walker, J.A. (2000) Ability of geometric morphometric methods to estimate a known covariance matrix. Systematic Biology 49, 686–696.CrossRefGoogle ScholarPubMed
Willmott, A.P. & Ellington, C.P. (1997) The mechanics of flight in the hawkmoth Manduca sexta. II. Aerodynamic consequences of kinematic and morphological variation. Journal of Experimental Biology 200, 2723–2745.CrossRefGoogle ScholarPubMed
Wootton, R.J. (1992) Functional morphology of insect wings. Annual Review of Entomology 37, 113–140.CrossRefGoogle Scholar
Zelditch, M.L., Swiderski, D.L., Sheets, H.D. & Fink, W.L. (2004) Geometric Morphometrics for Biologists: A Primer. 443 pp. New York, Elsevier Academic Press.Google Scholar
Figure 0

Fig. 1. Methodology used to group the study sites into two altitude classes. Frequency distributions of altitudes were decomposed into Gaussian distributions using a combination of a Newton-type method and expectation maximization algorithms. The mean (μ) and variance (σ) of the altitude distribution for both groups of sites were calculated using the ‘mixture distribution’ package of R software (R Development Core Team, 2008). For low altitude sites, μ=2548 and σ=108, and for high altitude sites, μ=3075 and σ=77.

Figure 1

Table 1. Data on localities where Tecia solanivora specimens were collected for this study (localities considered for sex-related variation analysis are marked with*).

Figure 2

Fig. 2. Right wing of a male specimen of Tecia solanivora depicting the position of the 18 landmarks used for the analyses. Vein nomenclature from Borror et al. (1981).

Figure 3

Fig. 3. Upper part: Scatter plot for the RWA on male and female wings. Distribution of specimens for (a) x: rw1, y: rw2 and (b) x: rw2, y: rw3 (males: n=49, in grey; females: n=63, in black). Lower part: Deformation grids for both negative and positive extremes of the three relative warps: rw1, rw2 and rw3.

Figure 4

Fig. 4. Histogram of the specimens grouped by sex (males: n=49, in grey; females: n=63, in black) along the Discriminant axis. Percent of individuals correctly assigned to their original groups: 99.11%.

Figure 5

Fig. 5. Deformation grids for the consensus configurations and displacement vectors of a. males and b. females, from the reference configuration calculated on the overall sample (n=112, top). Grids and vectors obtained using tpsSpline, exaggerated by a factor of 5 for better visualization of changes.

Figure 6

Fig. 6. Scatter plot for the RWA on males from the 12 localities studied. Breakdown of the distribution of specimens for each locality on x: rw1, y: rw2. (a) San Pablo, (b) Pumamaqui, (c) San Simón, (d) Salache, (e) San Miguelito, (f) Salache, (g) La Hoya, (h) Anchilibí, (i) San Francisco, (j) Palama Bajo, (k) Carbón Chimipamba, (l) Palama Medio. The figure shows a rw1 and rw2 plot breakdown of each site for better visualization of the distribution.

Figure 7

Fig. 7. Histogram of the specimens grouped by altitude (high altitude: grey, low altitude: black) along the discriminant axis. Percent of individuals correctly assigned to their original groups: 83.99%.

Figure 8

Fig. 8. Deformation grids for the consensus configurations and displacement vectors of specimens from low and high altitude sites, from the reference configuration calculated on the overall sample (n=281, top). Grids and vectors obtained using tpsSpline, exaggerated by a factor of ten for better visualization of changes.

Figure 9

Fig. 9. Regression of the natural logarithm of centroid size (CS loge) on the first canonical axis (CV1) of the shape variables of male (n=49, in grey) and female (n=63, in black) wings.

Figure 10

Fig. 10. Regression of the natural logarithm of centroid size (CS loge) on the first canonical axis (CV1) of the shape variables of high altitude (n=89, in grey) and low altitude (n=192, in black) wings.