|
<br />
<br />656
<br />
<br />COPEIA, 1989, NO.3
<br />
<br />Dl
<br />
<br />Description and numerical ranking
<br />
<br />
<br />TABLE 1. DIAGNOSTIC QUA LIT A TIVE CHARACTER STATES PROPOSED FOR THREE SPECIES OF THE CVPRINID
<br />GENUS Gila FROM THE COLORADO RIVER, WESTERN UNITED STATES. The first five character states are dia-
<br />grammed in Figure 2.
<br />
<br />Characters
<br />
<br />Mouth (snout and
<br />upper lip)
<br />
<br />Jaw length rela-
<br />tive to orbit
<br />
<br />Predorsal (nuchal)
<br />hump
<br />
<br />Skull depression
<br />
<br />Caudal peduncle
<br />
<br />Nuchal scalation
<br />
<br />Breast scalation
<br />
<br />Sub-terminal, oblique, and associated with acute snout (G. elegans; state I), through
<br />clearly sub-terminal, oblique, with obtuse snout (G. Tobusta; 3), to inferior and ap-
<br />proaching horizontal, beneath overhanging or bulbous snout (G. cypha; 5).
<br />Posterior extension of upper jaw (as measured by vertical position of posterior tip of
<br />mandible beneath orbit) either anterior to eye (G. Tobusta; state I), beneath ante-
<br />rior third of eye, or perhaps as far posterior as front of pupil (G. cypha; 3), to
<br />clearly beneath pupil (G. elegans; 5).
<br />Posterior to skull, nuchal development varies from absent (nape and nuchal region
<br />rounded smoothly from skull) to slightly developed (G. Tobusta; states 1 and 2),
<br />through pronounced as a convexity rising smoothly from skull (G. elegans; 3), to an
<br />abruptly rising (4), or even forward-directed (5), fleshy hump (G. CYPha).
<br />Dorsal surface of head varies (in lateral view) from flattened to slightly concave (G.
<br />Tobusta; states I and 2), through smoothly and deeply concave, with deepest point
<br />of depression near middle of skull above or slightly behind orbits (G. elegans; 3), to
<br />markedly concave, with deepest point of concavity clearly posterior to orbits (G.
<br />cypha; 5).
<br />With body horizontal, a line projected along venter parallel with base of anal fin
<br />either intersects near middle of upper lobe of caudal fin (G. Tobusta; states 1 and
<br />2), falls on or near dorsal margin of that fin (G. CYPha; 3), or passes across caudal
<br />peduncle anterior to caudal fin (G. elegans; 5).-
<br />Scale development on nuchal region varies from complete, normal in imbrication,
<br />and relatively exposed (G. Tobusta; state I), through scattered, with substantial dis-
<br />tances between scales and/or deep embedding in skin (G. elegans; 3), to so deeply
<br />embedded in skin as to appear absent (4), or scaleless (5) (G. CYPha).
<br />Scale development on breast varies as in character state 6 (above), in same sequence
<br />of species.
<br />
<br />~
<br />~
<br />r2
<br />~
<br />~
<br />
<br />Fig. 2. Semi-diagram
<br />cyprinid fishes of the gel
<br />
<br />excluded, resulting in a total of 243 fish in the
<br />formal analysis. Assessment of the discriminat-
<br />ing ability of field-collected data was performed
<br />on the Arizona State University IBM-3090
<br />mainframe computer. Compiled data were en-
<br />tered and preliminary univariate analyses ac-
<br />complished using the Statistical Analysis System
<br />(SAS, 1982). Data sets were then analyzed using
<br />principal components analysis (PCA) (NT-SYS,
<br />Rohlf et aI., 1972, unpublished manual, SUNY-
<br />Stony Brook, N.Y.).
<br />Three separate PCA of qualitative rankings
<br />were performed: to assess data as recorded in
<br />the field; to estimate effects of coding variation;
<br />and to evaluate robustness of characters on the
<br />ability to segregate species. In the first, all in-
<br />formation was retained and analyzed as record-
<br />ed, with scores varying from 1-5. For the sec-
<br />ond, data were recorded from 1-3, as follows:
<br />scores recorded as 1-1.5 became I; scores of
<br />2-4 became 2, and scores of 4.5-5 became 3.
<br />The third analysis employed only the first five
<br />
<br />character states (exc
<br />ters) , with scores var
<br />PCA (Book stein et al
<br />as per Rohlf and Boo
<br />to quantitative data to
<br />size and produce size-
<br />ting. Group member!
<br />cedure was based on
<br />in analyses of qual ita I
<br />
<br />. This character reflects a complex relationship between caudal peduncle and body shape. while also describing elongation and attenuation (or
<br />lack of same) of the caudal peduncle i,self.
<br />
<br />treme. Intermediate rankings (i.e., 1.5,2.5, and
<br />so forth) were allowed. Data were collected by
<br />six personnel, all of whom had university de-
<br />grees in biological sciences and work experience
<br />varying from a few weeks to 3 yr. None was
<br />specifically trained in systematic ichthyology,
<br />and no more than 15-30 min of instruction
<br />constituted preparation for an individual. Per-
<br />sonnel worked in pairs, with one individual tak-
<br />ing measurements and the second recording
<br />data. In practice, decisions on qualitative rank-
<br />ings constituted effort on the part of both. A
<br />span of 3-5 min was required to collect data
<br />from each specimen, and fish were released near
<br />the site of capture. Diagrams of the various
<br />combinations of features (Fig. 2) were prepared
<br />to aid in field estimation of qualitative character
<br />states.
<br />A total of 358 Gila was collected. Of these,
<br />only specimens greater than 199 mm and less
<br />than 30 I mm SL were statistically evaluated.
<br />Specimens with missing information also were
<br />
<br />RJ
<br />
<br />PCA of the seven (
<br />erated three compor
<br />of cumulative variati
<br />vectors I-III [with I
<br />cated] are 4.02 [57~
<br />[12%], respectively). .
<br />component (PC.) vs I
<br />3A) demonstrates th
<br />groups of fishes, sep
<br />a contrast among m
<br />sion, and nuchal hu.
<br />loadings) and the tW(
<br />high negative loadir
<br />characters are diagn
<br />with the remainder (
<br />as G. robusta (n = 216
<br />
|