Spatial Variation and Autocorrelation of Shell Crushing Resistance and Pigmentation
We sampled nineteen sites containing Mexipyrgus churinceanus from various drainages throughout the entire Cuatro Ciénegas basin (Fig. 1; Table 1). Five linear shell measurements (shell length, shell width, spire height, and aperture length and width) commonly used in gastropod studies [25,31,34] were measured using Image-Pro Express. All morphological traits were log10-transformed prior to analysis. Crushing resistance of adult Mexipyrgus churinceanus was measured in 19 populations. The snails were placed in water, transferred to the laboratory, and crushed within 4 hours of collection. Snails were crushed between two plates of a Chatillon DFM50 force gauge with an automated Chatillon LTMCM-6 stand. The mobile force plate was set at 2.54 cm/min crushing speed. The force in Newtons needed to crush the snail at the time of shell failure was recorded. To characterize shell size, we conducted a principal components analysis of log10-transformed linear shell measurements. To characterize shell shape, we performed a covariance PCA [35], which finds shape components that are separate from size. Covariance PCA analysis was conducted in ADE-4 [36], and PCA shape scores were obtained by an R program subroutine provided by A. Bellido. Using these PCA measures of size and shape, we conducted a stepwise multiple regression to assess whether size and shape explained significant variation in crushing resistance. Based on evidence that size explained considerable variation in crushing resistance (see Results), we next examined population-level variation in size-adjusted crushing resistance. We used analysis of covariance of crushing resistance under the following model: PCA size as a covariate, population as a main fixed effect, and an interaction term. We tested for homogeneity of slopes using the interaction term. Given a non-significant interaction term, we then conducted a two-way ANCOVA to assess whether there was a significant effect of population. We used the estimated marginal means as a measure of size-adjusted crushing resistance for each population in spatial autocorrelation analyses described next.
We used spatial autocorrelation analyses to examine both the independence of population estimates of crushing resistance and pigmentation, and whether defensive traits change in mosaic fashion at small spatial scales,. We used Moran's I to test for correlations among populations for two defensive traits (size-adjusted crushing resistance and frequency of pigmented shells), and whether these correlations change as a function of geographic and/or genetic distance. Under the spatial mosaic hypothesis, we predicted no positive autocorrelations of geographically-adjacent or genetically-similar populations. We estimated Moran's I using Passage [37]. Moran's I is a measure of autocorrelation, with positive autocorrelations indicating genetically-similar or geographically-proximate populations are similar in defensive traits [23,25]. We constructed plots of geographic and genetic distances against size-adjusted crushing resistance and percentages of pigmented snails (correlograms) using classes of geographic and genetic distances. We used an equal number of pairs of points for each distance class. We used GPS to determine latitude and longitude of each population and then calculated the surface distance between them in order to produce a geographic distance matrix of the Cuatro Ciénegas populations [38]. Pairwise genetic distances between populations were determined from a previously published paper on mtDNA cytochrome b sequence variation [26] using a distance matrix employing the number of pairwise differences.
Abiotic and Biotic Influences on Load Strength and Pigmentation
To assess the potential influence of abiotic factors on shell strength, we measured temperature and conductivity in the 19 populations (Oakton hand-held conductivity meter Tampa, FL). Conductivity measures were temperature compensated. Above ground flow connects a few of the populations examined here, but the abiotic factors of each are influenced by large discharge from separate isothermal springs [7]. Therefore, the abiotic factors of the 19 populations should be fairly constant, and our measurements should be representative of the temperatures and conductivity populations experience during the entire year. We determined correlations between population means of putative defensive traits (size-adjusted load strength, frequency of pigmented shells) and both temperature and conductivity.
To estimate the relationship between frequency of molariforms and both size-adjusted crushing resistance and frequency of pigmented shells, the frequency of molariforms in ten populations was obtained from two sources. The frequency in Mojarral Oeste, North Tio Candido, Tio Candido, Los Remojos Negro, and Los Remojos Blanco was estimated in 2001 by Kloeppel [22], and was estimated in 2001 in Juan Santos and determined for Churince and Mojarral Este Alta in 2003 from samples reported in Hulsey et al [15]. Fish were considered molariform if they exhibited at least one molariform tooth. The correlation between these molariform frequencies and both size-adjusted crushing resistance and frequency of pigmented shells was then determined.
To address whether the presence or absence of Nymphaea is associated with variation in snail crushing resistance and pigmentation, we compared the crushing resistance (Analysis of Variance) and frequency of banded snails (Log likelihood tests) in three pairs of adjacent populations in the Rio Mesquites, Los Remojos, and Tio Candido. These paired populations are in very close proximity to one another (0.83 km, 0.05 km, and 0.48 km, respectively). In these paired populations, substrate coloration differed dramatically due to the presence or absence of Nymphaea beds. We also examined the phenotypic correlation between mean population estimates of the number of bands and crushing resistance. We conducted all statistical analyses in SPSS [39].
Answers to all your Biology Questions
