9336

• UCREFRP Catalog Number: 9336
• Author: Douglas, M. E., M. R. Douglas, J. M. Lynch and D. M. McElroy
• Title: Use of Geometric Morphometrics to Differentiate
• USFW Year: 2001
• Copyright Material: YES

394

COPEIA, 2001, NO.2

TABLE 3. CLASSIFICATION RESULTS FOR Gila robus/a FROM A CANONICAL ANALYSIS OF SHAPE COORDINATES. Num-
bers in parentheses are classification results from truss analysis of McElroy and Douglas (1995). Key to sample
codes is given in Table 1.

BR CC DC WW YR DB Rl
BR 19 (18) o (0) o (0) 0(0) o (1) o (0) 0(0)
CC o (0) 6 (6) o (0) 0(0) o (0) o (0) o (0)
DC o (0) o (0) 24 (24) 0(0) o (0) o (0) o (0)
WW 2 (0) 0(0) o (0) 52 (52) 2 (1) o (2) 0(0)
YR 3 (1) o (0) o (0) o (0) 62 (63) o (1) o (0)
DB o (0) 0(0) o (0) o (0) o (1) 20 (19) 0(0)
Rl 0(0) o (0) o (0) o (0) ] (1) o (0) 24 (24)

with higher values corresponded to a robustar
like form (i.e., larger head with no nuchal
hump and broader peduncle; Fig. 5B). The sec-
ond relative warp (RW2) is an axis of within-
species variation. Here, decreasing values cor-
respond to increasing prominence of the nu-
chal hump and concavity of the head. Little Col-
orado River G. cypha exhibit extreme CYPharlike
morphology when compared to specimens from
Y<J,mpa River, Black Rocks, and Westwater Can-
yons. Variation in G. robusta is also exhibited
along this axis, with atypical morphology (with
respect to. relative nuchal development) found
in Cataract Canyon. Patterns of shape variation
for each species (depicted in Fig. 4) are statis-
tically cO!lgruent (Mantel Z = 10.834; r = 0.721;
p = 0.02).

Congruence among truss and geometric analyses.-
McElroy and Douglas (1995) did not present
Mahalanobis distance (D2) and classification

0.015 b.
YR
.
YR
. b. b.
BR BR ww
.
ww b.
0 DC b.
.
RW2 DC DB b.
RI
b.
. CC
CC
.G.01S .
LC
U02 0.D02

RW1

Fig. 4. Shape variation among 13 populations of
Gila cypha and Gila robus/a based upon the first two
relative warps (RWs) of a relative warp analysis of
shape coordinates using the grand mean as the illn-
gent configuration. Population abbreviations follow
Table 1.

matrices for their truss analyses. These are pre-
sented in Tables 2 and 3. For G. cypha, good
agreement exists between D2 values obillined
from geometric and truss analyses (Mantel Z =
0.88; / = 3.15; P = 0.007). The same significant
relationship was observed for analyses of G. ro-
bus/a. Therefore, both geometric and truss anal-
yses support a single relative pattern of varia-
tion. In addition, classification matrices for geo-
metric and truss analyses closely resemble one
another (Tables 2-3) . For G. CYPha, the sole dif-
ferenceoccurred at Westwater Canyon, where
the truss analysis correctly reallocated 55 (of 57)
specimens, whereas the shape coordinate anal-
ysis reallocated 54 specimens. Although num-
bers of specimens correctly reallocated are sim-
ilar in both analyses of G. robus/a (diagonal of
Table 3), the placement of incorrectly allocated
specimens differs, particularly for Westwater
Canyon.

~.:::I::.:L..L:::.C~:.;,[':'~...[.E.f:.',"::".in':";::::;...j.nn..!...::..L....j~...:,T::::;

l : :
. : : :. ....--, -;. .+-m.f...-i....n~.....i......f.......~.. ...~.......L._..l.......l

r]11l1ffit~~~t~fE;:~=t~-~~-J

A

....,.;.....;......":'......-:....... -........-..r......~_.

;=!;I~nH~[u:iI~{tt~~iS:j

...i.......:.................... .. .:; .' : :
.... j .r....t..-..j
B ....,.....'-.....!

Fig. 5. Splines illustrating shape variation among
the first relative warp axis of Figure 4. (A) corre-
sponds 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.