Sexual dimorphism in chelicerae size in three species of nuptial-gift spiders: a discussion of possible functions and driving selective forces


  • Editor: Philip Rainbow

Luiz Ernesto Costa-Schmidt, Núcleo de Aracnologia, Departamento de Genética, Universidade Federal do Rio Grande do Sul, Av. Bento Gonçalves, 9500; Prédio 43323; Sala 205, Porto Alegre/RS, CEP 91501-970 Brazil.


Positive allometric patterns observed for intersexual signalling characters are related to directional sexual selection, and supported by theoretical and empirical data. Recent models have shown that positive allometry may not hold as a rule if the influence of natural selection is added to the model. Here we tested these models applying traditional morphometrical techniques for the analysis of chelicerae sexual dimorphism and allometric patterns within the genus Paratrechalea: Paratrechalea azul, Paratrechalea galianoae and Paratrechalea ornata. Spider chelicerae are basically used for prey capture, but males of Paratrechalea also use the chelicerae to offer a nuptial gift during courtship, also presenting a clear size and colour sexual dimorphism supporting a possible role as a signal. Chelicerae size was male biased for all the variables studied and showed an isometric pattern, while females showed a higher variation. Our findings are in accordance with models of viability-related function for prey capture, questioning some statements proposed by the positive allometry model.


The premise that characters associated with intersexual signalling show a positive allometric pattern under directional sexual selection is supported by empirical data and theoretical models (Eberhard et al., 1998; Maynard-Smith & Harper, 2003; Funke & Huber, 2005). However, some authors have demonstrated that such arguments may not hold as a rule, and the patterns stated previously could be a consequence of sampling biases of the cases studied (Bonduriansky, 2007). Furthermore, signalling structures imply resource allocation trade-offs between body size and trait size that maximize fitness. While sexual selection could favour these signalling traits, natural selection may counteract this trend (Arnqvist, 1997; Bonduriansky & Day, 2003).

Recently, Costa-Schmidt, Carico & Araújo (2008) described the sexual behaviour of Paratrechalea azul Carico 2005 and Paratrechalea ornata (Mello-Leitão, 1943), where the male offers a nuptial gift during courtship. The same behaviour was also observed in the field for a third species of the same genus, Paratrechalea galianoae Carico 2005 (Costa-Schmidt et al., 2008). These are the first records of such nuptial gift offering for Neotropical spider species, seen previously among a few Pisauridae species (Huber, 2005).

The nuptial gift consists of a prey item wrapped in silk that can be adult stages of aquatic insects (mainly Ephemeroptera). Males perform stereotyped behaviour during courtship, starting with the adoption of a hyperflexed posture in front of the female while holding the nuptial gift with the chelicerae (Fig. 1). In case of acceptance, the female adopts the same posture in front of the male and bites the nuptial gift. While the female is biting the nuptial gift, male courtship continues with a series of stimulatory movements over the female body, with alternate mountings over the female prosoma in order to proceed with palpal insertion and sperm transfer. A detailed description of the courtship process is presented in Costa-Schmidt et al. (2008).

Figure 1.

 Male of Paratrechalea azul holding a nuptial gift with its chelicerae while in the hyperflexed position. This sequence of pictures was taken under laboratory conditions by presenting a mature female to the male, illustrating how the male chelicerae are used during courtship.

According to preliminary observations, male and female chelicerae within these species are sexually dimorphic in both size and colour aspects, with males bearing more robust and reddish chelicerae (Costa-Schmidt et al., 2008Fig. 2). There, therefore, seems to be behavioural and morphological evidence suggesting that male chelicerae could function as intersexual signals during courtship.

Figure 2.

 Overall Paratrechalea sp. chelicerae morphology. All specimens and dissected structures shown are Paratrechalea azul. (a) Male frontal view; (b) female frontal view; (c) male (left) and female (right) lateral view. BL, basal segment frontal length; BW, basal segment frontal width; BLW, basal segment lateral width; FL, fang length. Scale bars=1 mm.

This paper aims to describe the degree of chelicerae sexual dimorphism in P. azul, P. ornata and P. galianoae, and discuss the possible use of chelicerae as a signal character by analysing the data with traditional linear morphometric tools.

Material and methods

One hundred and seven female and 89 male adult Paratrechalea spiders were collected from natural populations. One sample of P. azul and another sample of P. ornata were collected at Pedra de Amolar River (Maquiné Municipality, Rio Grande do Sul State, Brazil – 29°32′20.52″S; 50°14′46.83″W) from May to August 2005. Paratrechalea galianoae was collected from a single population located at Pedra Branca Fall (Pedras Brancas River, Itati Municipality, Rio Grande do Sul State, Brazil – 29°23′45.59″S; 50°02′42.44″W) in January 2006.

For the measurement of chelicerae dimensions, we used a Nikon SMZ600 (Melville, NY, USA) stereomicroscope with a scaling eyepiece for recording the size of the following chelicerae dimensions: basal segment frontal length (BL), basal segment frontal width (BW) basal segment lateral width (BLW) and fang length (FL) (Fig. 2). All measurements are presented in millimetres.

We also considered the cephalothorax centroid size as an indicator of overall body size (CSize), because this structure has a well-known correlation with body size in spiders (Eberhard et al., 1998; Prenter et al., 1999; Walker & Rypstra, 2001). The centroid size corresponds to the square root of summed squared distances of landmarks from their centroid (Swiderski, 2003). Cephalothorax centroid size is also presented in millimetres.

In order to estimate cephalothorax centroid size, digital images from the dorsal view of cephalothorax were taken using a digital camera (Nikon Coolpix 5400, Tokyo, Japan) attached to the eyepiece of a stereomicroscope (Nikon SMZ600). All photographs were taken at the same magnification, and an eyepiece scale was used to make a posterior pixels/millimetres conversion. Only Type II landmarks (Slice et al., 1996) were plotted in each image, except for two Type I landmarks in the cephalothorax (Fig. 3, landmarks #11–12). The identified landmarks for the cephalothorax were: #1 & #3, maximum width of the anterior end, which corresponds to the clypeus width; #2 & #7, cephalothorax length; #4 & #10, inflexion points between the cephalic region and the thoracic region; #5 & #9, cephalothorax width; #6 & #8, maximum curvature of the posterior end; and #11 & #12: thoracic furrow length.

Figure 3.

 Locations of landmarks on the cephalothorax in dorsal view. The specimen corresponds to a Paratrechalea ornata male. Scale bar=1 mm.

Variables were tested for normal distributions using the Kolmogorov–Smirnov model test, and most of the variables showed a normal distribution (P-values ranging from 0.05361 to 0.9945). Only the chelicerae BW of P. ornata males and FL of both P. ornata males and P. azul females showed deviations from normal distributions (P-values equal to 0.03594, 0.00372 and 0.0104, respectively). Mean differences between sexes for each species were analysed using the Student t-test in order to detect chelicerae sexual size dimorphism. When any of the two assumptions of normality or equal variances were not met, Mann–Whitney U tests were performed.

The coefficient of variation (CV), which is the standard deviation divided by the mean, was estimated for each variable within each sex. We applied a test proposed in Zar (1999, p. 141) to compare the CV of each variable between sexes. However, Eberhard et al. (1998) direct special attention to the fact that CV is affected by two factors that should be analysed separately, particularly for comparative studies involving correlated quantitative traits.

One factor is the slope of the estimated allometric line between the trait of interest and a particular body size trait, which will be presented below. The other factor is the dispersion of points around this allometric line, which we estimated using the alternative coefficient of variation [CV′=CV(y) × (1−r2)1/2] proposed by Eberhard et al. (1998). CV′ represents the coefficient of variation that y would have if x were held constant (Eberhard et al., 1998), which is the portion of the total CV that is more prone to variation in the context of sexual selection.

Allometric indices (the regression coefficient of the study variable over an overall body size variable) were obtained from linear regressions of the log-transformed original data. Two sets of comparisons were made for each structure and for each sex. Firstly, we used the CSize as the overall body size variable. Secondly, we used the size of the structure to infer the relations among dimensions within the structure. We arbitrarily chose the BL as a measure of overall chelicerae size.

Two models of linear regression were used: the ordinary least-squares (OLS) model, which implies the assumption of a dependent variable error only; and the reduced major axis (RMA) model, a model that assumes some amount of error for both variables (dependent and independent). The RMA model seems to be more realistic, because all the measurements were taken using the same technique, one that does not control for measurement errors within the independent variable, although it has some statistical requirements, like significant levels of correlation coefficients (Eberhard, Huber & Rodríguez, 1999; Ohno et al., 2003; Bonduriansky, 2007). The OLS model still holds as the most-used model for the comparative data available, although there is a possible underestimation of the regression coefficient. The allometric indices were tested for the null hypothesis of isometry, that is, assuming a regression coefficient equal to one (β=1). For each species, two sets of intersexual comparison of the estimated regression slopes were performed using a one-way ANCOVA test for homogeneity of regressions, with CSize (body size allometry) and BL (chelicerae size allometry) as covariates.

Regression models were run with PAST v. 1.75b (Hammer, Harper & Ryan, 2001), using the default bootstrap procedures for the estimation of confidence limits (2000 repetitions). Cephalothorax CSize was estimated using the RMorph library (Baylac, 2007) developed for R environment (R Development Core Team, 2007).


Chelicerae sexual size dimorphism

The overall chelicerae dimensions confirmed the already expected male-biased sexual size dimorphism (Table 1). The only exceptions were the inversion of this pattern for the FL in P. azul and the absence of a statistical difference for this same variable in P. galianoae.

Table 1.   Mean and standard deviation (mm) for each chelicerae variable and overall body size in males and females and the corresponding result of the statistical comparison
 CharacterMalesFemalesStatistic comparisonP
  1. BL, basal segment length; BW, basal segment width; BLW, basal segment lateral width; FL, fang length; CSize, cephalothorax centroid size.

Paratrechalea azul n=35n=38  
BL2.57 (± 0.1393)2.41 (± 0.1359)t=4.86; d.f.=71<0.00001
BW1.26 (± 0.0670)1.04 (± 0.0694)t=13.54; d.f.=71<0.00001
BLW1.68 (± 0.0890)1.37 (± 0.0881)t=15.18; d.f.=71<0.00001
FL1.16 (± 0.0634)1.23 (± 0.0715)U=299.5<0.0001
CSize8.06 (± 0.4089)8.45 (± 0.4943)t=−3.68; d.f.=71<0.001
Paratrechalea ornata n=37n=35  
BL2.21 (± 0.1055)1.87 (± 0.1345)t=12.00; d.f.=70<0.00001
BW1.07 (± 0.0644)0.82 (± 0.0565)U=1290<0.00001
BLW1.39 (± 0.0740)1.08 (± 0.0763)t=17.34; d.f.=70<0.00001
FL1.01 (± 0.0478)0.97 (± 0.0714)U=8840.0056
CSize6.96 (± 0.3004)6.68 (± 0.3978)t=3.33; d.f.=700.0014
Paratrechalea galianoae n=17n=34  
BL2.10 (± 0.1350)1.93 (± 0.1705)t=3.83; d.f.=49<0.001
BW1.07 (± 0.0620)0.85 (± 0.0703)t=10.81; d.f.=49<0.00001
BLW1.36 (± 0.1196)1.13 (± 0.1068)t=6.99; d.f.=49<0.00001
FL0.96 (± 0.0538)0.98 (± 0.0876)U=2410.3382
CSize6.47 (± 0.4137)6.57 (± 0.4939)t=−0.71; d.f.=490.4823

Females showed higher CV values for all measurements in the three species. CV comparisons between sexes within each species showed different significant levels (Table 2). For P. azul, the CV values are statistically equal between males and females. An opposite pattern was observed for P. ornata, which showed significant differences in CV for all measurements, except for BW and BLW. The other species, P. galianoae, showed a pattern closer to P. azul, with different CV values for FL.

Table 2.   Coefficients of variation (× 100) for males and females for each chelicerae variable and overall body size
 Paratrechalea azulParatrechalea ornataParatrechalea galianoae
  1. BL, basal segment length; BW, basal segment width; BLW, basal segment lateral width; FL, fang length; CSize, cephalothorax centroid size.


The absolute values of the alternative CV′ did not have a consistent pattern (Table 3). From 12 comparisons, the only observed statistical significance was for BLW in P. galianoae, with males showing more variation than females (Table 3). If we consider marginally significant results, P. ornata and P. galianoae showed a tendency towards an increase of male variation in BW in relation to overall body size, while P. azul showed a female variation bias for FL in relation to chelicerae size (Table 3).

Table 3.   Alternative coefficient of variation (CV′× 100), for males and females, for each chelicerae variable in relation to overall body size (CSize) and chelicerae size (BL)
 Paratrechalea azulParatrechalea ornataParatrechalea galianoae
  1. CSize, cephalothorax centroid size; BL, basal segment length; BW, basal segment width; BLW, basal segment lateral width; FL, fang length.

Body size
Chelicerae size

Chelicerae allometric analyses

Males have a constant isometric pattern for both regression models (OLS and RMA) in relation to CSize (Table 4). We found only two significant coefficients in opposite directions from 24 estimated coefficients, that is, one hyperallometric coefficient estimated using RMA (BW in P. ornata) and a hypoallometric coefficient using OLS (FL in P. galianoae). As expected, RMA tended to show higher allometric coefficients than OLS. When the measured variables were regressed against the BL (Table 5), they maintained the isometric pattern, except for the BW of P. ornata, which showed the same hyperallometric pattern for the RMA estimate.

Table 4.   Ordinary least squares (bOLS) allometric coefficients, reduced major axis (bRMA) allometric coefficients and correlation coefficients (r) between the analysed chelicerae dimensions and cephalothorax centroid size (Csize)
  • All correlation coefficients were statistically significant for P<0.0001.

  • °

    Marginally significant.

  • *


  • BL, basal segment length; BW, basal segment width; BLW, basal segment lateral width; FL, fang length.

 Paratrechalea azul (n=35)0.901.060.850.811.040.770.861.040.830.921.080.85
 Paratrechalea ornata (n=37)0.911.100.831.031.40*0.730.931.250.740.831.090.76
 Paratrechalea galianoae (n=17)0.951.010.940.770.920.831.171.46°0.800.68*0.870.77
 Paratrechalea azul (n=38)0.82*0.950.860.941.120.840.951.090.870.84°0.990.85
 Paratrechalea ornata (n=35)1.111.21*0.911.031.150.901.091.18*0.921.121.24*0.90
 Paratrechalea galianoae (n=34)1.111.16*0.961.041.090.961.041.100.951.091.18*0.93
Table 5.   Ordinary least squares (bOLS) allometric coefficients, reduced major axis (bRMA) allometric coefficients, and correlation coefficients (r) between the analysed chelicerae dimensions and chelicerae basal segment length (BL)
  • All correlation coefficients were statistically significant for P<0.0001.

  • °

    Marginally significant.

  • *


  • BW, basal segment width; BLW, basal segment lateral width; FL, fang length.

 Paratrechalea azul (n=35)0.850.980.860.880.980.900.941.020.92
 Paratrechalea ornata (n=37)1.081.27*0.850.881.130.770.81°0.990.82
 Paratrechalea galianoae (n=17)0.810.910.890.941.050.900.710.860.83
 Paratrechalea azul (n=38)1.051.18°0.891.051.14°0.920.891.040.86
 Paratrechalea ornata (n=35)0.81*0.950.850.880.980.900.961.030.93
 Paratrechalea galianoae (n=34)0.88*0.940.930.90°0.950.950.961.020.94

Females showed different patterns for each species. From the 12 estimated RMA slopes, five were significant for hyperallometry (Table 4). Two measures showed the same relation for both P. ornata and P. galianoae (BL and FL), and a third was only observed for P. ornata (BLW). A single case of hypoallometry using OLS was observed for BL in P. azul. However, within chelicerae size regressions showed a distinct pattern (Table 5), where P. azul presented a slight tendency for hyperallometry (marginally significant RMA slopes for BW and BLW) and P. ornata and P. galianoae showed a hypoallometric slope for BW.

Male and female comparisons showed statistically the same allometric slopes for most of the traits in all the species considered (Table 6). The only exceptions were related to BW and FL allometric slopes for P. galianoae when considering the overall body size. However, three other comparisons showed marginally significant results: P. ornata FL in relation to overall body size, P. ornata BW and P. galianoae FL in relation to chelicerae size (Table 6).

Table 6.   Comparison between allometric coefficients (b values) of males and females in relation to cephalothorax centroid size (CSize:body size) and basal segment length (BL:chelicerae size)
 Paratrechalea azulParatrechalea ornataParatrechalea galianoae
  1. BL, basal segment length; BW, basal segment width; BLW, basal segment lateral width; FL, fang length.

Body size
Chelicerae size


Chelicerae sexual dimorphism

The present results confirm that chelicerae size dimorphism is male biased in the three species studied, which strongly coincides with the hypothesis of a selective force acting over this structure. Chelicerae sexual size dimorphism could be explained in terms of sexual selection acting on that trait. A possible and intuitive interpretation, supported by behavioural (Costa-Schmidt et al., 2008) and now by morphometric data, could be that it is advantageous for males to develop larger chelicerae to be used for signalling during courtship. If so, theory predicts that male chelicerae would not only be larger but should also show a higher coefficient of variation. However, these predictions were only partially met by our data. Even though we found a male-biased sexual size dimorphism for most chelicerae dimensions, CV values (numerical and/or significant) were female biased (Table 2) and CV′ values did not show a regular pattern for a higher variation level for either sex (Table 3).

Such discordance from theoretical predictions leads us to imagine another possible evolutionary scenario where sexual selection is not necessarily the main driving force. We may interpret this pattern as a consequence of a natural selection process on the chelicerae evolution of these Paratrechalea species. Differences in chelicerae size could be a product of specialization for prey capture (Walker & Rypstra, 2001), where males can be more prone to capturing larger and/or more difficult prey than females. This could lead to an ecological outcome of niche divergence, a mechanism proposed by Hedrick & Temeles (1989) for the origin of sexual size dimorphism within a species. In fact, this hypothesis does not have empirical support. Field observations (L. E. C. S.) suggest that both sexes have the same prey capture ability and a complete diet overlap. The investigation of differences in prey capture abilities between sexes deserves an experimental approach.

Alternatively, we also have another interpretation where both selective forces, sexual and natural selection, may be working concomitantly. Sexual selection could indeed act for intersexual signalling, while natural selection may act as a constraint to male chelicerae development, in order to maintain the functionality of this structure for prey capture (viability-related functional selection). Reduction in chelicerae functionality for prey capture has been observed in cases where males have morphological modifications of their chelicerae for male–male combat (Walker & Rypstra, 2001), even though male–male combat was never observed for P. azul and P. ornata (L. E. C. S. field observations). The lower CV levels observed for males fit with this model and can be interpreted as the occurrence of such a constraint, that is, most males have reached an optimal level of chelicerae size and are now constrained to this optimal (maximum) limit. However, it is important to emphasize the fact that three of the four CV′ values that were significant or marginally significant were male biased for higher variation, which may be interpreted as the presence of a sexual selection force over these chelicerae traits.

Chelicerae allometric coefficients

The overall findings of an isometric pattern along male allometric coefficients direct our attention towards the possibility that Paratrechalea male chelicerae are not a signalling trait driven by sexual selection. In this case, the hypothesis of a signalling trait was proposed under the assumption of visual communication during the courtship process – but according to the present results the observed dimorphism would be guided by natural selection.

However, we insist that because there is a consistent chelicerae sexual dimorphism (shape and size) and a full behavioural description of the use of this trait use during courtship (Costa-Schmidt et al., 2008), the data provided do not exclude the male chelicerae indeed being a sexual signalling trait. As Bonduriansky (2007) states, allometric patterns cannot necessarily support the action of sexual selection over a specific trait, that is, the fact that we found an isometric relation along the measured chelicerae dimensions – which are not different from those compared with females – does not mean that sexual selection is not acting over them. Furthermore, we should pay special attention to the life history of the species in order to draw our conclusions, because an isometric pattern had already been modelled for sexual selected traits that also have viability-related functions (Bonduriansky & Day, 2003).


The fact that females showed higher coefficients of variation agrees with our hypothesis that male chelicerae did not present higher allometric slopes because they are under viability-related functional selection. This was supported by the CV′ data, which showed little evidence for differences in variation in the context of sexual selection. The sexual dimorphism of Paratrechalea chelicerae can be an example of model number 3 presented by Bonduriansky & Day (2003). Paratrechalea male chelicerae probably have reached an optimal variation level, and so have a restricted CV for chelicerae traits. Females would be under a natural selection regime for this structure, maintaining the functionality of the structure with a higher level of variation than males.


We thank Alfredo Peretti, Lucas C.T. Silveira and Cristine S. Trinca for reading the paper and suggesting important improvements to previous versions, Matthias Foellmer,Sídia Callegari-Jacques and Janaína Jaegger for statistical support Luiza Schmidt for English review and all those who have helped us during field collections. We also thank Anita Aisenberg and an anonymous referee for their analytical and theoretical suggestions. The present study was supported by a grant from the National Council for Scientific and Technological Development (CNPq – Brazil), awarded to L.E.C.S. All animal collections comply with the current laws of the Brazil government, represented by IBAMA, which issued all the collection permits.