Environmental factors affecting the variation of morphological traits in North Uist and South Uist
To quantify associations between morphological traits of resident stickleback and different environmental factors of the loch water in North Uist and South Uist, GLMs were fitted with a Gaussian distribution and identity link function. PC1 and PC2 from the PCA of armour traits were fitted as response variables, with location (North or South Uist), pH, salinity and conductivity as predictor variables. GLMs using the mean of each morphological trait (standard length, 1stdorsal spine length, 2nd dorsal spine length, pelvic spine length, length of pelvis, height of pelvis and plate count) of each population as the response variable were fitted in relation to pH value, conductivity and salinity of loch water. A Scheffe post-hoc test (Scheffe 1951) for multiple comparisons of mean with unequal sample sizes was performed to observe the differences in the size of measured armour traits among four low pH lochs in North Uist.
3. Results
3.1 Variation in morphological traits between North Uist and South Uist
The first two PCs from the PCA of the armour traits for the anadromous stickleback from both North and South Uist accounted for 58% of the total variation. All measured armour traits had positive loadings for PC1 and PC2 except pelvic spine length and length of pelvis (Fig. 2). Overall, anadromous fish from North and South Uist were morphologically similar, with some exceptions. The mean height of the pelvis of South Uist anadromous fish was significantly greater than North Uist (one-way ANOVA: F = 5.19, df = 1, 44, p = 0.02; Table 1). The standard length of South Uist anadromous stickleback showed significantly less variation than North Uist fish (MSLRT = 4.15, p = 0.04) (Fig. 3).
Resident populations of stickleback from North and South Uist were significantly smaller (smaller mean standard length) than the anadromous populations (W = 9742, p < 0.001; Fig. 3). Mean standard length of the North Uist resident stickleback (mean ± S.E. 36.44 ± 0.64 mm) was significantly greater than South Uist fish (34.55 ± 0.53 mm) (one-way ANOVA: F= 5.21, df = 1, 216, p= 0.02; Fig. 3), although the difference in length is small.
The first two principal components of a PCA of armour traits for resident fish collected from North and South Uist explained approximately 85% of the variance among individuals (Fig. 4A). PC1 was strongly correlated with all measures of armour traits so that high values of PC1 are associated with more armour. Plate count was highly correlated with PC2 and had a stronger influence on the variation in North Uist fish than South Uist. All measured armour traits (except plate count) were strongly correlated with each other Pearson’s correlation (r)>0.5 (Fig. 4B). For resident fish, despite no significant differences in trait means, there was significantly more variation (CV) in all traits (except standard length) in North Uist than South Uist (Table 1).
3.2 Factors affecting the variation of morphological traits in North Uist and South Uist
There was wide variation in many of the environmental factors including salinity and pH across the 8 lochs of South Uist and 10 lochs of North Uist: loch pH ranged from low (6.5, acidic) to high (~ 9, alkaline) pH of water (one-way ANOVA: F = 10.28, df = 1, 216, p = 0.001) in the freshwater lochs of North Uist compared to mostly neutral (~ 7) lochs of South Uist (Table A1 and A2).
There were significant associations between the first two PCs of armour traits and the pH of the 18 populations of resident fish collected from North and South Uist (Table 2). There was striking morphological variation in fish found in acidic freshwater lochs relative to other lochs (Fig. 4).
All morphological traits of the resident fish showed significant associations with pH levels in both North and South Uist (Table 2). Total plate count showed different associations with pH of water across the two locations with significantly lower number of plates in higher pH on South Uist, although overall across both North and South Uist the mean plate count was highest in the high pH populations (6.58 ± 1.41) and lowest in freshwater low pH lochs of North Uist (1.86 ± 0.27) (F1, 17 = 11.3, p<0.001; Fig. 5A). The mean standard length of resident fish was highest in freshwater high pH populations (37.02 ± 0.97 mm) and lowest in freshwater low pH (33.49 ± 0.92 mm) (F1, 17 = 4.85, p = 0.044; Fig. 5B). Among the 18 populations of resident fish of both islands, the mean length of all measured armour variables increased with increasing pH (Fig. 5C-F). One exception to the overall trend was in Tros (pH 6.63), were armour measurements were relatively large. Post-hoc tests (Scheffe Test) revealed that all measured armour traits Tros were significantly larger in length than in the three other freshwater low pH populations of North Uist (p < 0.001).
4. Discussion
Studies of adaptive radiation have tended to concentrate on what we can learn from adaptive radiations themselves, rather than their absence. In this study we have compared lacustrine populations of three-spined stickleback from isolated habitats on two neighbouring Scottish islands. North Uist has been known for decades to support a striking example of adaptive radiation in this species (Campbell 1985; Magalhaes et al. 2021), apparently linked to variation in the abiotic and biotic environments that is indexed by variation in pH (Giles 1983, MacColl et al. 2013, Magalhaes et al. 2016, Haenel et al. 2019). In contrast, variation in stickleback on South Uist has hardly been considered, despite apparently similar variation in aquatic environments. We have shown that in fact South Uist lacks the striking morphological variation of North Uist, and this lack of evolutionary diversification is well explained by a lack of variation in the pH of water bodies.
Anadromous stickleback, that are likely to provide a good approximation for the ancestors of Uist freshwater stickleback showed similar variation in armour traits and size on North and South Uist, consistent with close common ancestry in the marine environment, and supporting the idea that differences between the two islands arise because of differences in freshwater. In contrast, resident fish collected from North and South Uist showed variation in all measured armour traits within and between populations which indicate there is adaptive radiation of phenotypic characters among populations as a result of colonization and adaptation to the freshwater environment (Bell and Foster 1994; Magalhaes et al. 2021). There was significantly more variation in armour traits in resident fish of North Uist than South Uist which demonstrates differences in evolutionary diversification of closely-related lineages.
The pH of the lochs showed strong associations with the armour traits of resident fish collected from North and South Uist consistent with several previous studies of stickleback adaptive evolution in freshwater environments (Giles 1983; Bell and Foster 1994; Spence et al. 2012; MacColl et al. 2013; Magalhaes et al. 2016). In the present study, the armour traits of resident stickleback showed substantial variation in dorsal spines, pelvic spines and pelvis length in the freshwater populations of both islands. However, North Uist stickleback exhibited much more morphological variation than the stickleback of South Uist and this variation was directly associated with greater diversity in pH of the loch water on North Uist. This suggests that the differing extent of morphological diversification between North and South Uist is a direct consequence of differences in some aspect or aspects of environment that are related to pH.
It is not immediately obvious why there should be less environmental variation on South, than North, Uist. At first sight, the surface geology of both islands, the root cause of the pH variation (Waterston et al.1979), is rather similar, with acidic, peaty lochs in the east and alkaline, machair lochs in the west. On North Uist, the acidic lochs where the most extreme phenotypes have evolved are relatively large. Adjacent, smaller lochs with similar pH have less extreme phenotypes (ADCM, personal observations). It may be that larger stickleback populations in larger lochs facilitate evolution. On South Uist the eastern lochs are generally smaller, but even the larger ones (e.g. Druidibeg) do not contain unusual stickleback. In any case, the eastern lochs in South Uist are clearly less acidic, but again it is unclear why. Druidibeg is rather shallow, with a largely rocky bottom, and it may be that a shorter residence time of water in the loch, coupled with less contact with peat, may prevent the development of more extreme acidity. The topography of South Uist is also rather different to that of the North. North Uist is generally low-lying, and catchments drain in a radial pattern, meaning that there is little variation in surface geology within catchments. In contrast, South Uist is hilly in the east and the main catchments drain to the west across the machair. The latter pattern results in a kind of ‘environmental flow’ (movement of water from east to west) that may reduce variation in water chemistry between lochs, at least the development of more alkaline conditions in machair lochs. These linear catchments may also facilitate gene flow between lochs on South Uist that is absent on North Uist, and this could inhibit diversification, but the restricted environmental variance alone on South Uist appears sufficient to explain the reduced morphological diversification, without invoking differences in gene flow.
It is not obvious exactly why phenotypic variation should be associated with variation in pH, although on North Uist resource availability, competition and predation regimes covary with pH (MacColl et al. 2013; MacColl and Aucott, 2014; Magalhaes et al. 2016). It seems most likely that many differences in ecology between alkaline and acidic lochs are fundamentally linked to a reduction in nutrient availability in the latter. This is consistent with the reduced size and armour that is apparent in acidic populations. Body size of some low pH populations was extremely small which mirrored the previous finding of dwarfism in stickleback in association with low pH indicative of poor resource conditions (Giles 1983; MacColl et al. 2013). The dorsal spines and pelvic complexes were also either rudimentary or missing in most of the populations from low pH lochs (except Tros, see below) in North Uist. Previously, several studies have suggested that this striking variation in armour was due to calcium ion limitations in acidic conditions (Giles 1983; Bourgeois et al. 1994). There is no obvious correlation on North Uist between calcium availability and armour evolution, but it is likely that there is a more general shortage of nutrients in the acidic lochs, which are usually oligotrophic verging on dystrophic. This suggests that the evolution of extreme morphological variation on North Uist is driven more by ecological constraint than by ecological opportunity.
The striking exception provided by stickleback in Loch Trosavat (Tros) to the general pattern of relationship between armour traits and pH is illuminating, if anecdotal. Tros is linked to the sea by a short stream, and this means that large, migratory salmonids (Atlantic salmon,Salmo salar and sea trout, S. trutta ) and potentially other marine fish that prey upon stickleback, occur in this loch. This by itself could explain the more developed armour in Tros stickleback, but it is also likely that proximity to the sea results in Tros having a better nutrient status than most acid lochs, which could compound the effect of predation.
To conclude, the increased evolutionary variation of North Uist stickleback populations compared to South Uist populations can be explained by greater environmental variation on North Uist, especially linked to variation in the pH of loch water. Research is on-going to understand the role of pH in the ecology and evolution of Uist stickleback populations.
Table 1: One-way analysis of variance (ANOVA) for mean value and modified signed-likelihood ratio test (MSLRT) for coefficients of variation (CV) of each morphological trait [standard length, 1st dorsal spine length (DSL), 2nddorsal spine length (DSL), pelvic spine length (PSL), length of pelvis (LPS), height of pelvis (HPS) and plate count] of stickleback sample fish (217 residents and 45 anadromous) collected from two locations (North Uist and South Uist, Scotland). Significant P values are in bold.