. Bonner zoologische Beiträge : Herausgeber: Zoologisches Forschungsinstitut und Museum Alexander Koenig, Bonn. Biology; Zoology. Bonner zoologische Beiträge 56 (2007) 27 2. MATERIALAND METHODS In 313 specimens, representing 43 species (Table 1), 42 equivalent skeletal variables (measurements of bone lengths, widths and depths) were taken, as described in detail in Nemeschkal (1999). Twenty measurements stem from the feeding apparatus, 13 represent the flying appa- ratus and 9 hindlimb locomotion (Table 2). Firstly, centroids of 43 species over 42 variables were built from the log scaled origi
. Bonner zoologische Beiträge : Herausgeber: Zoologisches Forschungsinstitut und Museum Alexander Koenig, Bonn. Biology; Zoology. Bonner zoologische Beiträge 56 (2007) 27 2. MATERIALAND METHODS In 313 specimens, representing 43 species (Table 1), 42 equivalent skeletal variables (measurements of bone lengths, widths and depths) were taken, as described in detail in Nemeschkal (1999). Twenty measurements stem from the feeding apparatus, 13 represent the flying appa- ratus and 9 hindlimb locomotion (Table 2). Firstly, centroids of 43 species over 42 variables were built from the log scaled original data matrix consisting of 313 specimens. This procedure was essential, because body mass data were available as species means only. Morpho- logical variation between species was then quantified by the total variance of species centroids. Body mass varia- tion in the actual ftingilline-cardueline sample ranges from a maximum of 54 g in the Hawfinch Coccothraustes coc- cothraustes to a minimum of 9 g in neotropical siskins (Table 1). To test hypotheses about size impact, linear re- gression analysis was applied. Log transformed species means of body mass were taken as predictor variable and each of the 42 skeletal measurements as criterion variable. The 42 resulting residuals are used as the variables under study - the variables corrected for body mass (= variables of which size was partialized out; BCM). The coefficients of determination between body mass and original variables (Table 2) were tested for significance using random boot- strap (1000 replicates each; for computer programme package see Nemeschkal (1999)). They are figured as profiles of variance (Figs 1, 2). Single linkage R-cluster analyses were chosen to figure correlations of original da- ta (Figure 3- BO; based upon the variable intercorrelation matrix between species means for original data) and cor- relations data corrected for body mass (Fig. 3- BCM; vari- able intercorrelation matrix between species mea
Size: 2165px × 1154px
Photo credit: © Library Book Collection / Alamy / Afripics
License: Licensed
Model Released: No
Keywords: ., bookcentury1900, bookcoll, booksubjectbiology, booksubjectzoology
