Trmgali and Wilson mtDNA analysis of two stocks of Sardinella aurita 



363 



Analysis of restriction-site polymorphisms of mito- 

 chondrial DNA (mtDNA) is generally regarded as of- 

 fering a higher degree of resolution than protein elec- 

 trophoresis in population genetic studies (e.g., Ferris 

 & Berg 1987). and has recently become important in 

 studies of potential fishery stocks. The most direct ana- 

 lytical approach is to compare the frequency distribu- 

 tions of composite mtDNA haplotypes among the 

 sample populations, testing (x 2 ) whether the distribu- 

 tions are homogeneous and, therefore, indicative of 

 stock mixing (Bentzen et al. 1989). However, the usu- 

 ally skewed distribution toward a common haplotype(s) 

 coupled with the common occurrence of several rare 

 haplotypes (Billington & Hebert 1991) in a population 

 sample usually results in low sample numbers per cell 

 in contingency tables, and tabulated values of .r 2 can- 

 not be used properly. This problem can be avoided, 

 however, by using a Monte Carlo method of computing 

 the exact x 2 expected from the observed haplotype fre- 

 quencies among sample populations (Roff & Bentzen 

 1989, Bernatchez & Dodson 1990). 



In this study we employ analysis of restriction-site 

 polymorphisms of mtDNA and Monte Carlo statistics 

 to test the hypothesis that the two main fisheries in 

 the eastern Gulf of Mexico are comprised of at least 

 two genetic stocks of S. aurita. We have included speci- 

 mens of S. aurita collected off southern Brazil under 

 an a priori expectation that stronger differences in 

 mtDNA haplotype frequencies should exist between 

 specimens from the Gulf of Mexico and southern Bra- 

 zil. Considering the geographic proximity of our 

 sample populations in the eastern Gulf of Mexico, 

 and recognizing the fact that inadequate sampling of 

 the mitochondrial genome (as evidenced by the num- 

 ber of restriction enzymes used) can bias the outcome 

 in the direction of "no significant differences," inclu- 

 sion of the Brazilian specimens allows us to objec- 

 tively gauge the power of our methods to address the 

 stated hypothesis. 



Materials and methods 



Specimens of S. aurita from the eastern Gulf of Mexico 

 were obtained from the two main commercial fisheries 

 along Florida's west coast, one near Tampa Bay and 

 the other off the Panhandle. Individuals were collected 

 fresh from both sites during the summer of 1989 (Fig. 

 1). Ripe ovaries were excised soon after capture and 

 held on ice or in refrigerated buffer (MSB-Ca" of 

 Lansman et al. 1981) for up to lOd prior to purifica- 

 tion of mtDNA; some ovaries from the Tampa Bay 

 sample were stored frozen at -86 C until processing. 

 The timing of specimen collections was structured to 

 avoid resampling a single migrating school. Two 



samples were taken concurrently from Port St. Joe, 

 Florida (re=10) and Destin, Florida (n=13) in May 1989. 

 and a third sample was taken from Destin (n=6) in 

 September 1989, for a total of 29 specimens (Fig. 1). In 

 the Tampa Bay region a total of 28 individuals were 

 taken during three collection times: February (n-12). 

 May (n = 10), and August (ra=6) of 1989. 



Whole fish specimens obtained from off Santos, Bra- 

 zil were collected from the fishery and hand-carried 

 frozen on dry ice to St. Petersburg, Florida within 2d 

 and were stored at -86 C for several months without 

 removing the ovaries. Because these specimens were 



A. 



40°N 



20 C N 



20°S 



40°S 



40°W 



100°W 



60°W 



20°W 



B. 



GULF OF 

 MEXICO 



_l_ 



30°N 



25°N 



90°W 



85°W 



80°W 



Figure 1 



Collection sites of Spanish sardines Sardinella aurita in the 

 eastern Gulf of Mexico and southern Brazil: (A) The two 

 major sites, and (B) two area sites in the eastern Gulf of 

 Mexico. 



