. Bonner zoologische Beiträge : Herausgeber: Zoologisches Forschungsinstitut und Museum Alexander Koenig, Bonn. Biology; Zoology. 214 A. Hille & H. Meinig i 2 MGPC-1 Fig. 4: Bivariate plot of the 95 °7o confidence ellipses for the sample means of the first two principal component factor scores of 13 populations of A. agrarius derived from a multiple- group PCA. The proportion of total variation explained by each component is indicated. The inset illustrates character vectors, based on their respective correlations with these axes. When the samples were grouped only after their geograp
. Bonner zoologische Beiträge : Herausgeber: Zoologisches Forschungsinstitut und Museum Alexander Koenig, Bonn. Biology; Zoology. 214 A. Hille & H. Meinig i 2 MGPC-1 Fig. 4: Bivariate plot of the 95 °7o confidence ellipses for the sample means of the first two principal component factor scores of 13 populations of A. agrarius derived from a multiple- group PCA. The proportion of total variation explained by each component is indicated. The inset illustrates character vectors, based on their respective correlations with these axes. When the samples were grouped only after their geographic origin, the separation of the three subspecies turned out less clearly (Fig. 7); 12 (19 °/o) animals were ill- classified, but 48 % could be classified correctly out of 13 populations, showing a high degree of variability. The same result is shown by the Neighbour-Joining tree based on the Mahalanobis distances of individual canonical variable scores from group centroids. Only a separation of the istrianus/kahmanni group on the one hand and agrarius on the other hand can be ensured. The branching pattern within the agrarius group displays no significant evidence. In general, discriminant analysis is a very useful tool to find or contrast differences between groups. However, a pre-allocation to a certain group should always be based on hard evidence. It is not advisable to introduce information into the calculations that should be confirmed by the following analysis. One should avoid an assignment of individuals to subspecies if the purpose of the analysis is to look for subspecific differentiation. Results of the discriminant analysis therefore should not be taken as affirmative because they are biased by a priori information. In this case multiple group principal component analysis revealed a convergent size shaping of W Slove-. Please note that these images are extracted from scanned page images that may have been digitally enhanced for readability - coloration and app
Size: 1612px × 1551px
Photo credit: © Library Book Collection / Alamy / Afripics
License: Licensed
Model Released: No
Keywords: ., bookcentury1900, bookcoll, booksubjectbiology, booksubjectzoology
