7861 - Laserfiche WebLink

Home Browse Search

DOUGLAS ET AL.-COLORADO RIVER CHUBS 657 Orbit-Jaw Relationship ~ V Y ~ ~t. ~J t-, 3 f-<-5 ~ predor~ Hump ;: J ~ r 1'1)_____ :J)--- ~ ~3 --5 Skull Depress i on & 1 ~- r"~ ~ . --~..- - -::: 3 5 Caudal Peduncle Fig. 2. Semi-diagramatic illustrations used to guide ranking of qualitative morphological characters for cyprinid fishes of the genus Gila from the Yampa River, Colorado. character states (excluding scalation charac- ters), with scores varying from 1-5. Sheared PCA (Bookstein et aI., 1985, with corrections as per Rohlf and Bookstein, 1987) was applied to quantitative data to remove influences of body size and produce size-free components for plot- ting. Group membership for the shearing pro- cedure was based on discrete clusters revealed in analyses of qualitative data. RESULTS peA of the seven qualitative characters gen- erated three components that explained 84% of cumulative variation in the data set (eigen- vectors I-III [with percentage variation indi- cated] are 4.02 [57%], 1.06 [15%], and 0.81 [12%], respectively). A plot of the first principal component (PC.) vs PC2 from this analysis (Fig. 3A) demonstrates the presence of two distinct groups of fishes, separated on the PCl axis by a contrast among mouth/snout, skull depres- sion, and nuchal hump (all with high positive loadings) and the two scalation characters (with high negative loadings). Extremes of all these characters are diagnostic of G. cypha (n = 27), with the remainder of the specimens identified as G. Tobusta (n = 216). There are no indications that G. elegans, G. elegans hybrids, or G. Tobusta x G. cypha hybrids are represented in the data set. Analysis of the seven qualitative characters with scores recoded from 1-3 resulted in the plot shown in Figure 3B. Eigenvectors I-III from this analysis were 3.73 (53%), 1.05 (15%), and 0.85 (13%), respectively, and explained 81 % of cumulative variation. Components 1 and 2 were defined as above. Gila cypha is again clearly dif- ferentiated from G. Tobusta. The third analysis, using only five qualitative characters, produced the PCI vs PC2 plot de- picted in Figure 3C. The first three eigenvec- tors accounted for 87% of the cumulative vari- ation, and were recorded as 2.51 (50%), 1.04 (21%), and 0.79 (16%), respectively. The clear separation between component scores for the two species is reduced, and the clouds of points touch tangentially. The first two components of PCA of the eight quantitative characters explained 84% of cu- mulative variation (eigenvectors I and II are 5.85 [74%] and 0.86 [10%], respectively). Over- all body size of the specimens (as indicated by the first eigenvector) greatly influenced results of this analysis. Figure 4 is a plot of two size- free components, sheared PC2 vs sheared PC,.