<br />2750
<br />
<br />Evolution: DeMarais et al.
<br />
<br />would produce coincidental sorting of unique features in
<br />morphology, allozymes, and mtDNA to form G. r. robusta
<br />and G. elegans, as well as the morphological and allozymic
<br />intermediacy of MRN and G. seminuda.
<br />Morphology, allozymes, and mtDNA are all consistent
<br />with past hybridization between G, elegans and G. r. ro-
<br />busta. Various features of body shape place G. seminuda
<br />almost exactly intermediate between G. elegans and G. r.
<br />robusta (Fig, 1) and similar to artificially produced hybrids of
<br />that same parentage, while MRN, although clearly G. r.
<br />robusta-like, tends toward G. elegans. Allozymes provide
<br />the same pattern of intermediacy. MRN and G. seminuda
<br />exhibit a mixture of alleles diagnostic for G. elegans and G.
<br />r. robusta in the Colorado River basin below Grand Canyon
<br />(B.D.D., unpublished data). Results for G. elegans are,
<br />however, based on only a single population of this rare
<br />species.
<br />Unlike morphology or allozymes, mtDNA is nonrecombi-
<br />nant, and strict maternal inheritance prevents its expression
<br />as a mosaic of characters within individuals. The mtDNA of
<br />only one of the hybridizing species may be present within
<br />hybrid zones, depending on the direction and extent of
<br />introgression (e.g., see refs. 39-41), or both mtDNAs may
<br />persist long after the disappearance of one parental taxon
<br />(e.g., see refs. 42 and 43). By itself, the presence of only G.
<br />elegans-Iike mtDNA in G. seminuda and MRN has little
<br />meaning; however, the strong contrast between morpholog-
<br />ical and mtDNA characteristics in these forms (such that G.
<br />r. robusta-like or intermediate morphologies are coupled with
<br />G. elegans-like mtDNAs) is compelling evidence for their
<br />hybrid origin.
<br />Hybridization and Evolution. Acquisition of ecological di-
<br />vergence and attainment of reproductive isolation are the two
<br />major problems that must be overcome before hybrid speci-
<br />ation can successfully occur (3). If either fails, recombinant
<br />phenotypes are swamped out of existence via gene flow with
<br />parental forms. Hybrid recombinants generally show ecolog-
<br />ical divergence from both parental forms and seem most
<br />suited for hybridized habitats (1). In fact, the presence of
<br />such a suitable habitat has been argued as the primary limiting
<br />factor for hybrid speciation (ref. 44 and references therein).
<br />Modern examples of widespread hybridization between
<br />fish species can usually be traced to stocking of alien species
<br />and/or anthropogenic habitat alterations. Interspecific (and
<br />even intergeneric) hybridization is common (45, 46), with
<br />most such incidents resulting only in first-generation hybrids
<br />or local and transitory hybrid swarms.
<br />The early discovery and description of G. seminuda (19)
<br />preclude the possibility of introductions and/or man-induced
<br />habitat change as promoting their hybrid origin. Further-
<br />more, neither G. r. robusta nor G. elegans has ever been
<br />collected from the Virgin-River system (47), LaRivers (16)
<br />noted G. elegans in the lower Virgin River, but this may have
<br />been speculation since it was common in the adjacent main-
<br />stem Colorado River before impoundment of Lake Mead and
<br />persisted in the reservoir until the 1960s (ref. 48; W.L.M.,
<br />unpublished data).
<br />Natural phenomena surely had consequences similar to
<br />those of human perturbations (45) and perhaps with far
<br />broader implications, such as bursts of speciation (44). Drain-
<br />age transfers through stream capture or tectonism forced
<br />allopatric faunas into contact, and progressive desertification
<br />shrank aquatic habitats more than enough to hinder ecolog-
<br />ical segregation (49). Dramatic environmental change has
<br />been the rule in western North America throughout the
<br />Tertiary period (50), and repeated cycles of isolation and
<br />sympatry must have been common during the evolutionary
<br />history of modem western fishes. Differentiating taxa, re-
<br />peatedly forced into contact, almost certainly exchanged
<br />genetic material, and some stocks of hybrid origin must have
<br />
<br />~
<br />
<br />Proc. Natl. Acad. Sci. USA 89 (1992)
<br />
<br />persisted. These may have had a selective advantage over
<br />parental forms since interspecific gene flow increases genetic
<br />variability faster than mutation alone, thereby allowing rapid
<br />responses to changing environments (1, 51, 52),
<br />Before human perturbations, both G. elegans and G, r,
<br />robusta were widely distributed in the Colorado basin. The
<br />former lived in mainstem rivers and their largest tributaries,
<br />and the latter occurred syntopically, while occupying smaller
<br />streams as well (53). Opportunities for hybridization were
<br />thus abundant, and the origin of G. seminuda/MRN through
<br />such an event would be predicted in intermediate-sized
<br />tributaries during regional drought or other natural catastro-
<br />phes,
<br />No physical barrier is known that would have impeded
<br />continual gene flow or actual migration of either G, elegans
<br />or G. r. robusla into the Virgin River. Yet, G. seminuda has
<br />established and continues to maintain independence from
<br />these taxa, It has stabilized morphologically, maintained
<br />Hardy-Weinberg equilibrium, and sustained a distinct and
<br />unique distribution parapatric to both parents for the > 100
<br />years since it came to the attention of ichthyologists,
<br />The taxonomic status of stabilized hybrid derivatives such
<br />as G. seminuda will likely remain a point of contention.
<br />However, current relegation of G. seminuda to a subspecies
<br />of G. r. robusta is inappropriate, iff or no other reason than
<br />arbitrariness; it could just as easily be considered a subspe-
<br />cies of G. elegans. More importantly, subspecific designation
<br />obscures both the evolutionary origin and independence of G.
<br />seminuda as a distinct lineage. Because G. seminuda exhibits
<br />phenotypic, genetic, and geographic integrity, its reelevation
<br />to specific rank is warranted.
<br />We thus conclude that G. seminuda and MRN comprise a
<br />distinct species, originating, through Pleistocene or post-
<br />Pleistocene hybridization between G. elegans and a local
<br />form of G. robusta and promoted by habitat changes brought
<br />on by regional aridification.
<br />Conservation Implications. Demonstration of hybrid origin
<br />for G. seminuda and MRN has specific ramifications for their
<br />conservation status. When G. seminuda was listed as endan-
<br />gered (54), MRN was considered an "undescribed" subspe-
<br />cies of G. robusta and specifically excluded from protection.
<br />However, information presented here indicates a close rela-
<br />tionship between these forms. Therefore, the listing of G,
<br />seminuda should be amended to include the Moapa River
<br />population.
<br />Evidence for hybrid origin also has general implications for
<br />conservation of the entire grouj). Most Gila native to the
<br />Colorado River basin are threatened with extinction, with
<br />several members listed or candidates for listing as endan-
<br />gered (G. cypha, G. elegans. G. r. robusta, Gila robusta
<br />jordani. G. seminuda). The tremendous morphological di-
<br />versity of this group is paralleled by its genetic complexity,
<br />and we anticipate that hybridization played an important role
<br />in generating and maintaining this complexity, While the
<br />general validity of this hypothesis remains to be tested, it still
<br />requires serious consideration, particularly when recovery
<br />programs are implemented and further studies are designed,
<br />If hybridization is indeed an important mode of evolution in
<br />western fishes, then protection of distinct forms coupled with
<br />an active bias against suspected hybrids could prove detri-
<br />mental to the entire complex. Not only could valid species of
<br />hybrid origin (such as G, seminuda) be eliminated, but a
<br />valuable mode of evolution could be truncated. Given the
<br />potential impact of such a program, research emphasis should
<br />instead be placed on determining the extent of hybridization
<br />and its effect on the evolution of these fishes. Meanwhile, it
<br />would be prudent to protect the entire complex so that
<br />valuable genetic variation distributed throughout the basin
<br />will not be lost forever.
<br />
|