Spatial Variation of Shell Crushing Resistance and Pigmentation
Mean and standard errors of crushing resistance and shell size are presented in Table 1. The five linear morphological variables had high positive loadings (> 0.96) on the first principal component, and over 94% of the total variance was explained by this first PC, which we interpret as a measure of overall shell size. Overall shell size (PC1 scores) showed considerable variation among populations (Table 1). Covariance PCA analysis revealed one major shape component representing a contrast between aperture size and spire length. This shape component explained about 5% of the total variance, and individuals with positive scores for this component have relatively large apertures and shortened spires. Multiple regression analysis of crushing resistance revealed that shell size and shape were both significantly positively related to crushing resistance. Shell size explained most of the variation in crushing resistance (standardized β = 0.662, df = 1, 503; R2 = 43.9%, p β
= 0.077, df = 2, 503; R2
= 0.006%, p = 0.02). We next used analysis of covariance to examine the effect of population on crushing resistance using size as a covariate. The interaction term (shell size by population) was not significant (p = 0.24), and there was a highly significant effect of population on size-adjusted crushing resistance (F = 6.3, df = 18, 467, p 3a
). The correlograms of size-adjusted crushing resistance and all distance measures were not significant (see Fig. 4
for correlogram using cytochrome b
distance), indicating the absence of autocorrelation among populations at any scale.
There was considerable variation in the frequency of pigmentation among populations (Fig. 3b) and there was a significant association between frequency of pigmentation and population of origin (likelihood ratio χ2 = 310.7, n = 19, p 4 for correlogram using cytochrome b distance classes), again indicating the absence of autocorrelation among populations at any scale.
Abiotic and Biotic Influences on Crushing Resistance and Pigmentation
There was no significant correlation between size-adjusted crushing and temperature (r = 0.32, n = 19, p = 0.19) or conductivity (r = -0.40, n = 19, p = 0.09). Similarly, there was no significant correlation between pigmentation frequency and temperature (r = -0.01, n = 19, p = 0.17) or conductivity (r = -0.23, n = 19, p = 0.13). In contrast, there were significant negative correlations between molariform frequency and both size-adjusted crushing resistance (r = -0.63, n = 10, p = 0.05; Fig. 5) and average frequency of pigmentation (r = -0.78, n = 10, p 5). Using mean population estimates, we examined the correlation between pigmentation patterns and crushing resistance. There was a highly significant positive correlation between number of bands and crushing resistance (r = 0.58, n = 19, p
In the two of the three paired adjacent populations that differ in Nymphaea abundance (RM1/RM3 and Tio Candido S/N), crushing resistance was higher where Nymphaea was abundant (Figure 3a). For example, Tio Candido N had significantly higher size-adjusted crushing resistance than Tio Candido S (90.0 and 51.8, respectively; F = 20.82, df = 1, 58, p p p > 0.5). In the paired adjacent populations, the frequency of pigmented shells was significantly higher in darker versus lighter substrates in all 3 comparisons: Rio Mesquites 1 and 3 (90% versus 25%, likelihood ratio = 19.1, d.f. = 1, p p p