7861

• UCREFRP Catalog Number: 7861
• Author: Douglas, M. E., W. L. Minckley and H. M. Tyus
• Title: Qualitative Characters, Identification of Colorado River Chubs ‎(Cyprinidae
• USFW Year: 1989
• USFW - Doc Type: Genus
• Copyright Material: YES

DOUGLAS ET AL.-COLORADO RIVER CHUBS

titative characters, but rather to demonstrate
that under specific conditions, researchers can
be relatively confident in the use of qualitative
characters for field identification of these fishes.
Of most importance to the endangered species
manager is the rapid and reliable identification
of species, not the means whereby that identi-
fication is accomplished.
The marked success of qualitative rankings
in segregating these taxa (Fig. 3) is based on the
inherent information content of the character
type. By this we mean that qualitative rankings
are not single characters (as are the quantitative
ones), but in fact are multi-variable, depicting
overall variation within body regions critical to
separation of species of Colorado River Gila
(e.g., skull, mouth/snout, caudal peduncle, and
so forth). Although anatomical regions (espe-
cially like the nuchal hump) are composed of
numerous separate characters, the regions
themselves may not be amenable to standard
methods of quantification (Hubbs, 1946; Smith
et aI., 1979). Also, individual characters that
comprise these regions may not be particularly
informative when evaluated independently and
separately. This, in fact, may be why the quan-
titative characters were unable to successfully
separate taxa. Qualitative data provided a clear
separation because ran kings succinctly sum-
marized discriminating anatomical regions. The
. separation was maintained in spite of alteration
in scale of the data (Fig. 3B), but diminished
(although remaining) when two important sca-
lation features were deleted (Fig. 3C). We will
examine use of these and other qualitative fea-
tures in future studies of the G. robusta complex.
Does this mean that G. cypha and G. robusta
cannot be discriminated by using quantitative
morphometric data? No. Previous work using a
variety of morphometric characters collected
from preserved specimens has accomplished this
separation (Smith et aI., 1979; Bookstein et aI.,
1985). We demonstrate that under restrictions
of this study, in which a variety of personnel
collected data for a limited number of charac-
ters in a brief time on living fishes under field
conditions, two forms could not be clearly sep-
arated on the basis of the quantitative charac-
ters used (Fig. 4). In this regard, a morpho-
metric analysis of these (and other) populations
of Gila will require two things. First, clear and
unambiguous quantification must be accom-
plished of those critical anatomical regions sum-
marized by our qualitative data (this, in part,
has already been attempted by Smith et aI.,

1979). Second, a means of collecting reliable
quantitative data on live specimens must be de-
veloped, which allows uninjured release after
handling and manipulation.
From our perspective, some confusion
shrouding identification of problematic Gila
hinges on apparent misunderstanding of the
populational nature of natural selection (a sta-
tistical concept). The conservative nature of
fisheries science, with reliance on an idealized
morphological archetype to represent a species,
contributes substantially to this problem. Vari-
ation within and between populations and
species must be recognized as a natural phe-
nomenon amenable to statistical analysis, rather
than an infrequent occurrence to be either re-
defined or dismissed. The recent assimilation
of population genetics into fish management
(Vrijenhoek et aI., 1985; Ryman and Utter,
1987) may, in fact, encourage a more theoret-
ical and evolutionary expansion of this applied
and traditional science.
How do our results agree with more theo-
retical studies of the perceptual basis of phe-
netic grouping? Sokal and Rohlf (1980), in a
rather surprising result, noted that the majority
of participants in their taxonomic experiment
generated classifications that could be accu-
rately predicted a priori by analyzing only 4-5
characters of the total of 86 defined. This re-
stricted data set represented subtle features of
the Caminalcules not particularly obvious in an
initial inspection of the imaginary organisms,
but which contributed greatly to discrimina-
tion. These conclusions argue strongly that the
great majority of observers in the Sokal-Rohlf
experiment were evaluating Caminalcules not
in a univariate (or single character) sense, but
rather in a multivariate (e.g., multi-variable)
manner, using what many would term "gestalt"
in the sense of Goethe (a view of the whole
organism; Blackith and Reyment, 1971; Mayr,
1982). The assessment of different character
states by our technicians enabled species of Gila
to be separated through multivariate statistical
comparisons, the same kinds of comparisons that
either a systematist or non-systematist uses in
visually separating taxa or other categories of
objects through step-wise, conscious or uncon-
scious decisions based on combinations of char-
acters.
Our results indirectly support those of Moss
(1971), who indicated that the ability to group
organisms or to recognize characters, at least
at the phenetic level, is an eminently human

659