Reference # 6466253


Geometric morphometric analysis of shape variation in the seed scales of Pinus pumila populations in the Kamchatka peninsula

Valentina VETROVA
Kamchatka Branch of Pacific Geographical Institute FEB RAS, Petropavlovsk-Kamchatsky, Russia

Methods of multivariate analysis of data obtained employing geometrical morphometry were used to compare samples from the 12 Kamchatka populations of Pinus pumila (Pall.) Regel. The similarity of the sample plots by the shape of seed scales was estimated based on the set of 7 landmarks. Scales belong to symmetric structures, therefore landmarks have been placed on one side of the scales: 3 fixed landmarks at the base, at the apex and at the greatest width and 4 outline points (semilandmarks) placed at regular intervals using angular algorithm. The shapes of the scales were compared using coordinates of the landmarks obtained by generalized least squares Procrustes superimposition and sliding baseline registration. For an estimation of the statistical significance of shape differences between locations, we used Goodall's F-test, Hotelling's T2-test, and Bootstrap resampling. Principal components analysis and canonical variates analysis were performed. Two principal components of variability accounted for about 80 % of variability in the scale shape. The first component is related mainly to the variability in length and reflection of scale tips and width of scales (43,5 % of total variability), and the second - to a variation in scale stretching (35,2 % of total variability). Partial Procrustes distances between pairs of mean shapes from different populations varied from 0.009 to 0.125. Results of a geometrical morphometry of seed scales suggest that P. pumila was dispersed in the region from the coniferous forest refugium - the central Kamchatka depression (CKD). Cluster analysis yielded two groups of populations originating from north and south of CKD. The method of geometrical morphometrics provides a more accurate estimation of seed scale shape variability than classical morphometrics and, thus, is a promising analytical approach to the analysis of phenotypic variability of conifers.

© 2012 Organizing Committee