Variation in morphological traits between North Uist and South Uist
All measured armour traits (except number of lateral plates) were size-standardized by calculating the residuals of a regression of each trait against standard length to obtain their allometric relationship with body size. The resident (including freshwater) (hereafter ‘resident’) and anadromous fish were analysed separately to quantify variation in morphological data [lateral plate count (hereafter ‘plate count’), standard length, residuals of 1st dorsal spine length (first dorsal spine), 2nd dorsal spine length (second dorsal spine), pelvic spine length (pelvic spine), length of pelvis (pelvis length), height of pelvis (pelvis height)]. Principal component analysis (PCA) was performed using the singular value decomposition method to explore the axis of greatest variation in the measured armour data. All variables were scaled and centred for PCA and grouped for visualisation purposes according to three variables: location (North Uist and South Uist), pH (high pH, neutral pH and low pH) and salinity (freshwater and saltwater).
A one-way analysis of variance (ANOVA) was performed to test for mean differences in each morphological trait between the two different locations (North Uist and South Uist). Modified signed-likelihood ratio tests (MSLRT) (Krishnamoorthy and Lee 2014) were used to test for differences in coefficients of variation (CV) between North Uist and South Uist for each morphological trait. Pearson’s correlations were used to test for relationships between measured armour traits. To compare the mean fish length of the anadromous and resident stickleback, a Wilcoxon rank sum t-test was performed for all samples of North Uist and South Uist.