Use of Geometric Morphometrics to Differentiate Gila (Cyprinidae) within the Upper Colorado River Basin

Video images of 215 adult Gila robusta and 148 endangered Gila cypha were collected from May 1991–October 1992 at eight Colorado River basin localities (seven upper basins and one lower basin). The two species were sympatric at five of these locations; G. robusta was absent at one site, whereas G. cypha was missing at two others. Saggital views of each individual were videotaped and 25 morphological points (15 anatomical landmarks and 10 helping points) identified. Bookstein shape coordinates were calculated from Cartesian coordinates of these landmarks and points, whereas centroid size was used as a measure of body size. Shape differences were evaluated among populations of each species using MANOVA and canonical variates analysis. In G. cypha, variation encompassed three aspects: nuchal hump (most pronounced in Grand Canyon forms), relative head size (larger in Cataract Canyon forms), and caudal peduncle dimensions (shorter with a tapering depth in Cataract Canyon forms but longer and uniformly deeper in those from Desolation Canyon). Nuchal development in G. robusta is slight, hence only head and peduncle dimensions distinguished populations. Those individuals from Cataract Canyon had relatively shorter peduncles that (again) tapered in depth from anterior to posterior, whereas G. robusta from Desolation Canyon possessed peduncles that were much longer and of uniform depth. Specimens from Debeque and Rifle Canyons had proportionally smaller heads. Variation among all 13 populations (i.e., both species together) was evaluated using relative warp analysis, with G. cypha and G. robusta clearly separated at all sympatric locations except those from Desolation and Caratact Canyons. Here, body shapes of the two species converged. Overall, shape variation in both species is clinal. Although results from our geometric morphometric analysis were statistically similar to those based on distances derived from a truss analysis, the geometric approach visually demonstrated phenotypic differences among populations and species and this, in turn, has management implications.

Fig. 1.

Landmarks used in the current study. Definitions are given in text or in McElroy and Douglas (1995)

Fig. 2.

Shape variation among six Gila cypha populations based on the first two variates derived from a canonical analysis of shape coordinates. Population abbreviations follow Table 1. Circles represent 95% confidence intervals for the group mean

Fig. 3.

Shape variation among seven Gila robusta populations based on the first two variates derived from a canonical analysis of shape coordinates. Population abbreviations follow Table 1. Circles represent 95% confidence intervals for the group mean

Fig. 4.

Shape variation among 13 populations of Gila cypha and Gila robusta based upon the first two relative warps (RWs) of a relative warp analysis of shape coordinates using the grand mean as the tangent configuration. Population abbreviations follow Table 1

Fig. 5.

Splines illustrating shape variation among the first relative warp axis of Figure 4. (A) corresponds to negative values of the warp and depicts a phenotype indicative of Gila cypha; (B) corresponds to positive values of the warp and demonstrates a Gila robusta phenotype

Accepted: 20 Nov 2000