Evolutionary associations between sand seatrout (Cynoscion arenarius) and silver seatrout (C. nothus) inferred from morphological characters, mitochondrial DNA, and microsatellite markers.(Report)

Abstract--The evolutionary associations between closely related fish species, both contemporary and historical, are frequently assessed by using molecular markers, such as microsatellites. Here, the presence and variability of microsatellite loci in two closely related species of marine fishes, sand seatrout (Cynoscion arenarius) and silver seatrout (C. nothus), are explored by using heterologous primers from red drum (Sciaenops ocellatus). Data from these loci are used in conjunction with morphological characters and mitochondrial DNA haplotypes to explore the extent of genetic exchange between species offshore of Galveston Bay, TX. Despite seasonal overlap in distribution, low genetic divergence at microsatellite loci, and similar life history parameters of C. arenarius and C. nothus, all three data sets indicated that hybridization between these species does not occur or occurs only rarely and that historical admixture in Galveston Bay after divergence between these species was unlikely. These results shed light upon the evolutionary history of these fishes and highlight the genetic properties of each species that are influenced by their life history and ecology.

**********

The molecular genetic associations between populations of sand seatrout (Cynoscion arenarius) and silver seatrout (C. nothus) have not been specifically examined on a large scale with DNA methods despite the close ties between the respective fisheries for the two species. In particular, the possibility of contemporary hybridization or historical admixture between these species remains to be explored by using a large panel of unlinked DNA markers. Sand and silver seatrout are so morphologically similar that they are collectively known as white trout by fishermen (Ginsburg, 1931). Both species are abundant throughout the Gulf of Mexico (hereafter, GOM); the distribution range for sand seatrout extends into the Atlantic Ocean, north to Georgia, and the distribution range for silver trout extends to Massachusetts (Hildebrand and Schroeder, 1928; Cordes and Graves, 2003). These seatrout make up a modest proportion of bycatch in shrimp and other commercial trawl operations (Warren, 1981), although commercial landings have decreased dramatically in the last 30 years (Fig. 1). Weinstein and Yerger (1976) completed perhaps the most comprehensive study of molecular evolution in the genus Cynoscion; they assessed protein electrophoresis variants in all four western North Atlantic species (C. arenarius, C. nothus, spotted seatrout [C. nebulosus], and gray weakfish [C. regalis]). Although these methods provided some insight into the evolutionary relationships among the species, the data of Weinstein and Yerger (1976) were insufficient to answer direct questions about rates of contemporary and recent historical gene flow within and among Cynoscion species. Enzyme electrophoresis has since been superceded by DNA-based methods on a broad scale. Microsatellite markers are likely more sensitive for studies involving high rates of gene flow and low levels of population identity (Wright and Bentzen, 1994). This is particularly true for marine fishes, whose populations are often characterized by enormous census sizes and higher rates of migration between subpopulations than freshwater and terrestrial organisms (DeWoody and Avise, 2000).

Previous morphological comparisons of sand and silver seatrout have resulted in a suite of distinct characteristics that vary between species in larval (Ditty, 1989) and adult stages (Ginsburg, 1929, 1931; Gunter, 1945; Moshin, 1973; Chao, 2002). However, both display a similar streamlined and fusiform body shape, and the ranges of numerous commonly used morphometric and meristic measures overlap between the species. Additionally, hybridization, regional differentiation, or a combination of both, may often confound the trademark diagnostics used to distinguish between the two species. In any event, the superficial similarity of these species indicates that morphological divergence has been minimal and the concurrent bimodal timing of spawning indicates overlapping life history parameters (Sheridan et al., 1984). Distributional data seem to indicate that these species exhibit some habitat partitioning (primarily by water depth and distance from the shore) with the result that silver seatrout are found more frequently in deeper water farther from shore (Ginsburg, 1931; Byers, 1981).

[FIGURE 1 OMITTED]

These life history and distributional data yield a framework for devising hypotheses to test the influence of niche overlap on historical associations between sand and silver seatrout populations. In particular, if hybridization between these species occurs, it is likely to occur in areas of contact such as nearshore marine waters used commonly by both species. Hybridization in the genus Cynoscion has been previously documented on the Atlantic coast of Florida (Cordes and Graves, 2003). In their initial examinations, Cordes and Graves (2003) characterized populations of gray weakfish using genetic techniques and in doing so also identified putative hybrids between gray weakfish and either sand or silver seatrout. This identification was accomplished by using four microsatellite markers and two nuclear intron gene regions (restriction fragment length polymorphisms, or RFLP's). Although these markers were appropriate for identification of hybrids with gray weakfish, they were ineffective for determining conclusively whether the second gametic contribution was made by sand or silver seatrout.

Here, both morphological and molecular (nuclear microsatellites and mitochondrial restriction fragments) data are used to characterize populations of sand and silver seatrout from the nearshore Gulf waters outside of Galveston Bay, Texas. Three competing hypotheses regarding genetic associations between these species are evaluated. First, in the case of contemporary hybridization, hybrids would appear as proportionate admixtures of both parental forms in microsatellite assignment tests. Moreover, the directionality of hybridization could be assessed with the use of mtDNA haplotypes (Wirtz, 1999), and hybrids would likely be found to be intermediate for diagnostic morphological characters (Hubbs, 1955; Campton, 1987). Second, in the case of historical association, such as lineage overlap during speciation or lineage admixture after speciation, mtDNA haplotypes might be shared between the species despite a mutually exclusive assignment of microsatellite genotypes (with the assumption of no contemporary hybridization). In such a case, assignment based on microsatellite genotypes should be more reliable than assignment by means of mtDNA haplotypes if mtDNA lineages have not been sorted categorically into contemporary populations (species). Third, in the case of no gene flow between the species, microsatellite assignment should be conclusive, mtDNA haplotypes should sort conclusively by species, and specimens should not reveal morphological intermediates for characters previously described as diagnostic among species. Each of these competing hypotheses was examined in light of evidence from the three data sets. The morphological and genetic similarities and differences between the species were examined as evidence for hybridization, and as an aid for future species identification. Finally, aspects of the ecology and life history of each species are invoked to explain the patterns of genetic variability within and between these congeneric species.