The first three PC’s accounted for 69% (33, 20 and 16% respectively) and 62% (32, 16 and 14 respectively) of the total variation in shell shape of L. saxatilis and M. neritoides respectively. Ordinations of these axes reveal a clear spatial segregation of sites for both species, although this is more evident for L. saxatilis than M. neritoides with the degree of separation more pronounced along the first and third axes than the second axis (Fig. 3). There is, however, substantial overlap, particularly among locations in the central part of the Iberian coast.
In the case of L. saxatilis (Fig. 4), PC1 summarised variation in the height of the apical whorls (relative positions of landmarks 1, 7 and 8 along the shell axis) and in the width of the aperture (separation between landmarks 2 and 4), with a narrowing of the apical whorls associated with a widening of the aperture. PC2 was associated with the shape of the right side of the last whorl (a displacement of landmark 4 towards the bottom of the shell was accompanied by an approach of landmark 2 along a perpendicular direction) and with the curvature of the labrum (displacement of landmark 8 perpendicularly to the shell axis). PC3 was also mostly related to the shape of the last whorl, but this time on the left side (migration of landmark 3 along the shell axis). Regarding M. neritoides (Fig. 5), PC1 was also related to the height of the apical whorls (relative positions of landmarks 1, 7 and 8 along the shell axis), as was the case with L. saxatilis. PC2 was associated with overall elongation of the shell (simultaneous displacement of landmarks 3 on the left and 4 and 8 on the right in a direction normal to the shell axis). Finally, PC3 was associated with a change in aperture shape (reflected by displacements of landmark 4 along the shell axis) and with a reduction in height of the apical whorls (displacement of landmarks 5 and 6 along the profile of the shell relative to the apex).
Shape differed significantly among sites for L. saxatilis (Approx-F5,90 = 7.77, P < 0.001) and M. neritoides (Approx-F5,90 = 5.42, P < 0.001). There was also a significant effect of allometry on shape for both L. saxatilis (Approx-F1,9.4 = 3.73, P < 0.001) and M. neritoides (Approx-F1,9.4 = 5.73, P < 0.001). Mean centroid size, furthermore, differed significantly among populations of both species (LS: F5,90 = 8.28, P < 0.001, R2 = 0.277; MN: F5,90 = 12.68, P < 0.001, R2 = 0.413). The importance of allometry in structuring spatial variation in M. neritoides can be seen in Figs 6-8. There was, however, in contrast to the multivariate result no significant linear relationship between any of the first three PC axes and allometry in L. saxatilis (PC1: F1, 94 = 2.22, P = 0.139; PC 2: F1, 94 = 0.15, P = 0.704; PC 3: F1, 94 = 0.083, P = 0.773) but highly significant associations between PC 1 (F1, 94 = 28.87, P < 0.001, R2 = 0.235) and PC 3 (F1, 94 = 7.20, P < 0.01, R2 = 0.071) and centroid size for M. neritoides. There was no significant relationship between PC 2 and centroid size (F1, 94 = 0.238, P = 0.627) for M. neritoides. The main allometric effect in M. neritoides was a marked increase in the height of the apical whorls with centroid size.
Fig. 6. Regression of the first PC axis on centroid size for (a) of Littorina saxatilis and (b) Melaraphe neritoides. In (b) there was a significant linear relationship between centroid size and shape.
Fig. 7. Regression of the second PC axis on centroid size for (a) of Littorina saxatilis and (b) Melaraphe neritoides.