|
<br />SPECIES RELATIONSHIPS AMONG FISHES OF THE GENUS GILA
<br />IN THE UPPER COLORADO RIVER DRAINAGEl
<br />
<br />Gerald R. Smith2, Robert Rush Miller2, and W. Daniel Sable3
<br />
<br />Taxonomic treatment of the chubs of the Gila
<br />robusta complex (family Cyprinidae) that inhabit
<br />the Colorado River basin and certain rivers in
<br />north-western Mexico (Figure 1) is problematical
<br />and requires study of the entire group for final
<br />resolution. It is intended to do this, but in this
<br />paper we present data demonstrating that there are
<br />three species of this complex in the Colorado
<br />River and its major tributaries, from the Grand
<br />Canyon region upstream. These chubs are the most
<br />extreme members of the genus in their specializa-
<br />tion for life in the unique big-river habitat of
<br />the Colorado River.
<br />
<br />Our objectives are (1) to determine the valid-
<br />ity and relationships of the roundtail, bony tail,
<br />and humpback chubs in the big-rive~ habitat;
<br />(2) to determine their relationships to other
<br />populations of Gila in the middle and upper
<br />Colorado River basin; and (3) to find characters
<br />useful for identifying young and adult Gila
<br />robusta Baird and Girard, G. elegans Baird and
<br />Girard, and G. cypha Miller, and possible hybrids.
<br />We do not at this time treat the complex repre-
<br />sentatives of Gila in the Gila River basin (see
<br />Rinne, 1976), except to note that G. intermedia
<br />(Girard) may be a separate e~olutionary line,
<br />secondarily related to G. robusta.
<br />
<br />Misinterpretation of the patterns of variation
<br />in the nuchal hump and lack of decisive, objective
<br />evidence has been the main cause of disagreement
<br />among those studying the G. robusta complex. The
<br />nuchal convexit~ has never been measured precise-
<br />ly or quantitatively related to other characters.
<br />For example, Holden and Stalnaker (1970) used
<br />subjective code values for the form of the nuchal
<br />
<br />~mp~etion of this paper was delayed by clo-
<br />sure of upper Green River to scientists other than
<br />federally funded personnel during the five-year
<br />period 1963-1967. Considerable aid was received
<br />from a number of organizations and individuals.
<br />Included are research grants to R. R. Miller from
<br />the Horace H. Rackham School of Graduate Studies
<br />(1950), the National Science Foundation (G-15914,
<br />24129, 24465, GB-3271, 4854, 6272X), and the Na-
<br />tional Park Service (1975), with cooperation from
<br />the various western states for collecting permits.
<br />Frances H. Miller recorded and calculated data,
<br />helped prepare the distribution map, and criti-
<br />cized drafts of the manuscript. Jeffrey N.
<br />Taylor prepared the ~riginal plots for Figure 1.
<br />Loans of specimens were received from the U.S.
<br />National Museum (E. A. Lachner, S. H. weitzman,
<br />W. R. Taylor, Susan Karnella), Arizona State Uni-
<br />versity (W. L. Minckley), Bell Museum of Natural
<br />History (Samuel Eddy, H. B. Tordoff, C. W, Huver),
<br />and the Museum of Northern Arizona (5. E. Caroth-
<br />ers). Mark J. Orsen (of the Museum of Zoology)
<br />and Karna M. Steelquist (of the Museum of Paleon-
<br />tology) prepared the drawings and other illustra-
<br />tions, except Figure 10, taken by Bruce J. Turner.
<br />Abbreviations are: UMMZ (University of Michigan
<br />Museum of Zoology) and USNM (U.S. National Museum
<br />of Natural History).
<br />
<br />2Museum of Zoology, University of Michigan,
<br />Ann Arbor, MI 48109.
<br />
<br />3Wisconsin State University, Whitewater,
<br />wise. 53190.
<br />
<br />hump, and coded continuous morphometric variables
<br />(23.1 15.8, etc.) into discontinuous states ("A",
<br />"B", etc.), thus losing much of the resolving
<br />power of the measurements. The clustering illus-
<br />trated by them shows only the level at which spec-
<br />imens join clusters; it does not show intermedi-
<br />acy, and it cannot display the morphological re-
<br />lationship of a specimen between other specimens.
<br />Thus there is no way to evaluate the supposed
<br />"intergrades" cited by Holden and Stalnaker: in
<br />their paper the "intergrades" may appear as a
<br />result of the method of coding data. They were
<br />unable to determine whether the so-called inter-
<br />grades were part of one variable complex or two
<br />incompletely separated species. In contrast, we
<br />find that there are very few intermediates, in
<br />the Green River at least, and that most specimens,
<br />including the "intergrades" shown by Holden and
<br />Stalnaker (1970: Figure 4), can be assigned to
<br />one of the two species cypha or elegans.
<br />
<br />Our analysis consists of two parts. First,
<br />principal components analysis of 34 meristic and
<br />morphometric characters, not including a direct
<br />measure of the nuchal hump, is used to establish
<br />clear evidence of morphological segregation of
<br />the large-river specimens into three distinct
<br />clusters. It is important to establish that the
<br />segregation is not based on the nuchal hump or
<br />other subjective characters, but on a broad range
<br />of body proportions, fin-ray counts, and verte-
<br />bral numbers, presumably with a broad genetic
<br />basis. Second, after establishing the existence
<br />of three separate populations, with little or no
<br />overlap, at least within the unmodified, large-
<br />river habitat, individual discriminating key char-
<br />acters are developed for identification of these
<br />fishes in the laboratory and field. Such charac-
<br />ters are critically needed to end the confusion
<br />that has clouded attempts to formulate a sound
<br />management policy for these endangered or threa-
<br />tened fishes. The popula..ion analy,;is and the
<br />development of key characters were conducted
<br />separately by us in order to avoid circular taxo-
<br />nomic logic and to enable the two systems to serve
<br />as useful tests of one another.
<br />
<br />Principal components analysis, a multivariate
<br />statistical method enabling taxonomic use of com-
<br />bined (correlated) information from many charac-
<br />ters, was applied to 34 characters of 140 speci-
<br />mens. These characters are: dorsal, anal, pec-
<br />toral, and pelvic fin-rays (rays from both paired
<br />fins recorded); standard length, head length, eye
<br />diameter, snout length, preanal length, head
<br />depth through eye, head depth at occiput, inter-
<br />orbital width, occiput to tip of snout, dorsal-
<br />fin basal length, anal-fin basal length, predorsal
<br />length, pectoral-fin length, pelvic-fin length,
<br />upper jaw length, mouth width, body depth over
<br />pelvic insertion, caudal-peduncle depth, anal
<br />origin to caudal base; number of vertebrae; pharyn-
<br />geal arches, total length, width, length of
<br />anterior limb, and length of posterior limb, of
<br />both left and right arches (eight characters).
<br />This analysis discriminated the populations in
<br />question (Figures 2-5) showing the major trends
<br />in variation, the characters dominant in the
<br />trends, and the characteristics of typical as well
<br />as intermediate individuals.
<br />
<br />A graph of 140 individuals plotted according to
<br />their scores on principal components I and II,
<br />
<br />613
<br />
|