8235 - Laserfiche WebLink

Home Browse Search

~ DISCRIMINATION OF COLORADO PLATEAU GILA SPP. 167 TABLE I.-Results from discriminant analysis based on meristic (10 variables), morphometric (23 loglO-transformed variables), and field-usable (23 meristic and morphometric variables) characters that were used to discriminate among Gila robusta, G. elegans, and G. cypha. Percent discrimination is presented for each model within each contrast. Characters Species compared Meristic Morphometric Field-usable G. robustalG. elegans Dorsal rays Caudal peduncle depth Caudal peduncle depth Gill rakers (2nd arch) Discrimination (100%) (100%) (100%) G. cyphalG. elegans Gill rakers (2nd arch) Caudal peduncle depth Caudal peduncle depth Dorsal rays Length of upper jaw Dorsal rays Thoracic vertebrae Length of left pharyngeal Length of upper jaw Dorsal fin base Head depth at occiput Discrimination (98.8%) (96.5% ) (95.3%) G. robustalG. cypha Anal rays Depth of frontal depression Head depth at eye Caudal vertebrae Anal origin to caudal base Dorsal fin base Gill rakers (2nd arch) Anal fin base Pectoral [0 pelvic insertion Discrimination (95.7%) (99.3% ) (99.3%) G. robustalG. cyphal Gill rakers (2nd arch) Caudal peduncle depth Caudal peduncle depth G. elegans Anal rays Depth of frontal depression Anal origin to caudal base Dorsal rays Anal origin to caudal base Snout length Thoracic vertebrae Dorsal fin base Snout to occiput Caudal vertebrae Length of upper jaw Dorsal fin base Discrimination (97.6%) (96.4%) (97.0%) (81 of 81), and G. cypha and G. elegans were spec- ified in 93.1 % (54 of 58) and in 96.4% (27 of 28), respectively. Field characters correctly allocated all specimens in 96.3% of cases (132 of 137). Again, G. robusta was correctly designated in 100% (81 of 81), and G. cypha and G. elegans were classified in 96.6% (56 of 58) and in 89.3% (25 of 28), respectively. Only 2 (of 24) individuals incorrectly allocated among the three species were shared between the three data subsets. The five field characters (Table 1) were used to derive a parametric discriminant function and to perform a canonical variate analysis. With regard to the former, the jackknife procedure failed to improve the original classification of individuals into species (five were misidenified: two G. cypha as G. elegans and three G. elegans as G. cypha). This classification was calculated to be 94% better than chance alone (kappa = 0.9417; Z = 19.463; P < 0.00001). A three-dimensional plot of indi- viduals on the first three canonical axes is pre- sented in Figure 2. Only axes 1 and 2 were sig- nificant (eigenvalue for canonical variate 1 = 8.6068 [87.4% of total variation]; eigenvalue for canonical variate 2 = 1.24 [12.6% of total varia- tion]). When the discriminant analysis was per- formed with equal sample sizes (N = 28) and with the use of within-group covariance matrices, only three individuals were misrepresented (one G. cy- pha was allocated to G. elegans and two G. elegans to G. cypha). Classification of Juveniles and Putative Hybrids Sixteen putative hybrids and 142 juveniles were designated to species by inserting relevant loglO- transformed measurements corrected for overall body size into the three formulae below. An un- known specimen is allocated to the species that achieves the highest of the three scores. Gila robusta = [(546.15. snout length) + (382.88. snout to occiput) + (966.72.dorsal base) + (752.3. caudal peduncle depth) + (1,573.58. anal origin to caudal base)] - 1,663.41. Gila cypha = [(470.37. snout length) + (481.35. snout to occiput) + (840.49. dorsal base) + (863.59. caudal peduncle depth) + (1,441.2. anal origin to caudal base)] - 1,630.01. Gila elegans = [(529.18. snout length) + (459.55. snout to occiput) + (820.2. dorsal base) + (941.2. caudal peduncle depth) + (1,489.33. anal origin to caudal base)] - 1,783.62. With the use of these functions, 88% (126 of 142) of the juveniles were correctly classified. For individual species, the results were as follows: G. robusta, 97.2% (103 of 106); G. cypha, 60.0% (18 of 30), and G. elegans, 66.7% (4 of 6). The three misclassified G. robusta juveniles were allocated