136 



Fishery Bulletin 98(1) 



lation. Contrary to the study of Gold et al. (1994), 

 we found no evidence to support the hjrpothesis that 

 red drum in Mosquito Lagoon are reproductively iso- 

 lated from other Atlantic red drum. In our survey 

 of the mtDNA control region, there was no nucleo- 

 tide divergence between Mosquito Lagoon and other 

 Atlantic samples, whereas divergence values between 

 the Mosquito Lagoon and the Gulf samples were 

 among the highest. Furthermore, other investigators 

 also found no significant differences at the allozyme 

 loci that putatively distinguish Mosquito Lagoon red 

 drum from other red drum (Campton*; Crawford and 

 Bert'M. This lack of difference suggests that the allo- 

 zjrme frequency differences observed by Gold and Rich- 

 ardson ( 1994) may not be temporally stable. A lack of 

 samples geographically proximal to Mosquito Lagoon 

 may have also influenced the outcome of that study. 

 Considering all the evidence, it seems more likely 

 that red drum from Mosquito Lagoon belong to the 

 larger genetic population occupying nearshore Atlan- 

 tic waters of the southern United States. Although 

 it is generally better to obtain hatchery broodstock 

 from locations within or near the intended release 

 site (Utter, 1998), the use of Mosquito Lagoon red 

 drum as a source of broodstock in Atlantic coast stock 

 enhancement programs should not produce a nega- 

 tive genetic impact on wild red drum. 



Implications for fishery and hatchery management 



Because red drum support valuable fisheries through- 

 out the Gulf of Mexico and Atlantic seaboard, the 

 species is of special concern to state and regional fish- 

 ery management agencies. Our results support the 

 hypothesis that a genetic transition in red drum pop- 

 ulation structure occurs in south Florida. In theory, 

 it requires the regular exchange of only a few indi- 

 viduals between breeding populations to homoge- 

 nize their genetic composition (Slatkin, 1987). Thus, 

 genetic exchange between Atlantic and Gulf stocks by 

 any recruitment process must be sufficiently low to 

 allow genetic differences to accumulate or, if the dif- 

 ferences reflect a historical disassociation, for them 

 to be maintained. The two red drum stocks can best 

 be described genetically as demes, separate semi-iso- 

 lated groups between which the structuring of het- 

 erogeneity differs from the assumption of panmixia 

 (Hartl and Clark, 1989). Therefore, the Atlantic and 

 Gulf stocks are likely to respond independently to 

 harvest regulations and these fisheries should con- 

 tinue to be managed separately. 



" Crawford, C, and Bert, T. 1997. Florida Marine Research 

 Institute, Department of Environmental Protection, St. Peters- 

 burg, FL. Unpublished data. 



Gold et al. (1993) observed a pattern of isolation 

 by distance in the distribution of mtDNA haplotypes 

 among Gulf samples that ranged from the south- 

 east coast of Texas to the southwest coast of Florida. 

 Although we did ^ot observe a similar pattern for 

 Gulf samples ranging from Apalachicola Bay to Flor- 

 ida Bay, the probability value for the Mantel coeffi- 

 cient was nearly significant at the 0.05 level and the 

 AMOVA value for the Apalachicola sample versus all 

 other samples was highly significant. This indicates 

 that the minimum geographic scale at which the 

 isolation-by-distance mechanism operates is greater 

 than the distance between Apalachicola Bay and 

 Florida Bay (approximately 670 km) or that genetic 

 discontinuity also exists between Florida Gulf red 

 drum and red drum inhabiting the northern and 

 western Gulf of Mexico. Accordingly, cooperative man- 

 agement of the Gulf fishery on a regional basis is 

 appropriate. No pattern of isolation-by-distance was 

 evident for red drum along the southern Atlantic sea- 

 board; the fishery between South Carolina and south- 

 east Florida should be managed as a single unit. 



Our principal objective for undertaking this study 

 was to improve upon available genetic information 

 relating to red drum population structure for fish- 

 ery management purposes. Our most informative 

 statistical tools were those that assessed relation- 

 ships among the samples. Genotype frequency differ- 

 ences accumulate quickly in subdivided populations 

 compared with the rate at which distinct phyletic 

 lineages emerge and sort geographically. Moreover, 

 mitochondrial DNA restriction fragment length poly- 

 morphism and sequence data for marine populations 

 are typically characterized by haplotype distributions 

 which consist of a few numerically and geographi- 

 cally prevalent haplotypes and many rare, geograph- 

 ically restricted haplotypes that may be important 

 with respect to population structure. Therefore, as 

 our results and the results of Tringali and Bert ( 1996) 

 demonstrate, genetically-based management units 

 (sensu Moritz, 1994), especially in marine fishery 

 stocks, may be more easily identified by applying pop- 

 ulation-level analyses that take full advantage of both 

 differences in genotype frequencies among samples 

 and phylogenetic relatedness of individual genotypes. 

 Statistical tests of association when applied to skewed 

 haplotype distributions often lack the power to detect 

 the low levels of population divergence that may char- 

 acterize marine populations. Moreover, these tests 

 ignore the interrelatedness of haplotypes in terms of 

 sequence similarity or difference. The two principal 

 analytical methods we employed, clustering of inters- 

 ample nucleotide divergence values and AMOVA, are 

 based on both the occurrence of haplotypes at particu- 

 lar locations and their sequence similarity to other 



