Anderson: Systematlcs of the North American menhadens in the genus Brevoortia 



369 



sniithi were western and eastern Gulf cognates (Christ- 

 mas and Gunter, 1960). Dahlberg (1970) conducted an 

 extensive morphological investigation of the Atlantic 

 and Gulf menhadens, including compilation of older 

 data as well as 14 novel physical character assessments 

 that provided morphological support for the notion of 

 small-scaled and large-scaled groups. Subsequent mor- 

 phological assessments of menhaden at various life 

 stages support this generally accepted taxonomic ar- 

 rangement (Hettler, 1984; Ahrenholz, 1991; Tolan and 

 Newstead, 2004). Avise et al. (1989) and Bowen and 

 Avise (1990) examined mitochondrial DNA (mtDNA) 

 fingerprints of the large-scaled menhaden complex 

 {B. pat/'onus+tyrajinus), and two divergent lineages of 

 large-scaled menhadens were identified, one of which 

 was shared between the two species, and the other oc- 

 curring only in B. tyrannus. Avise (1992) suggested that 

 this type of pattern is what would be expected from 

 historical divergence, followed by secondary contact be- 

 tween the species. Anderson and McDonald (2007) used 

 microsatellites to characterize populations of western 

 Gulf menhadens (S. patronus and B. gunteri) and found 

 significant genetic differences between these species. 

 However, the latter work was limited in its phylogenetic 

 scope because 1) it included only two of the four species 

 of Brevoortia found in North America, and 2) it did not 

 include genomic or mtDNA sequence data, which are 

 typically more reliable than microsatellites in discern- 

 ing relationships among species. 



In the present study, the taxonomic relationships 

 among the four North American menhaden species are 

 examined by using both nuclear satellite DNA (micro- 

 satellites) and mitochondrial DNA sequences (mtDNA). 

 Microsatellites are short runs of repetitive sequence 

 (usually a two to four base motif repeated multiple 

 times) that tend to be highly variable in vertebrates, 

 are codominant (both parental copies of the marker 

 can be scored for progeny), and are selectively neutral. 

 Because of these properties, they are used extensively 

 for examinations of population structure, paternity, 

 and kinship in vertebrates (Wright and Bentzen, 1994; 

 Jarne and Lagoda, 1996). Five DNA microsatellites 

 were scored in populations of menhaden comprising the 

 range of the four North American species. In addition, 

 mtDNA sequencing was conducted on a short segment 

 of the mtDNA control region for a subset of menhaden 

 representing the four species. Mitochondrial DNA as- 

 says are beneficial because the mitochondrial genome of 

 fishes is presumably not affected by recombination, and 

 direct maternal lineages can be examined in contrast 

 to the potentially admixed genotypes of nuclear DNA. 

 In addition, genome-wide genetic studies (i.e., those that 

 include both nuclear and mitochondrial data) are fa- 

 vored over single-locus or single-genome treatments be- 

 cause different genetic loci may yield conflicting results 

 (Hare and Avise, 1998). These data revealed distinctive 

 patterns which, when examined in the context of the 

 biogeographic setting of each species, broaden the exist- 

 ing information with which the evolutionary history of 

 North American menhadens may be elucidated. 



Materials and methods 



Sample collection and DNA isolation 



Fin clips from individual menhaden were collected from 

 bay systems covering the extent of the North American 

 range of Brevoortia (Fig. 1) from 2002 through 2004 

 (see Brown et al., 1996 for a discussion of the implica- 

 tions and assumptions of pooling multiple year genetic 

 data in clupeids). Adult fish samples from Texas were 

 collected with a 184 m gill net partitioned into four sec- 

 tions ranging in mesh sizes from 76 mm to 152 mm, and 

 all juveniles and young adult menhaden were collected 

 in bag-seine hauls. Menhaden from locations outside 

 of Texas were obtained through appropriate state and 

 federal agencies (see "Acknowledgments" section). Fin 

 tissue was excised from larger fish and placed in 70% 

 ethanol; for smaller fish, whole individuals were pre- 

 served in ethanol. DNA was extracted from 200 mg of 

 each fin clip or whole fish by using a Puregene® miniprep 

 kit (Gentra Systems, Minneapolis, MN) and following 

 the manufacturers instructions. 



Microsatellite data set 



Genomic DNA from 20 sampling locales (Fig. 1) was 

 used as a template to amplify microsatellite loci by 

 touchdown polymerase chain reaction (PCR) with fluo- 

 rescent end-labeled forward primers. Primer pairs were 

 designed around microsatellite regions initially isolated 

 from another clupeid, American shad (Alosa sapidissirna 

 Wilson, 1811); these primers have subsequently been 

 evaluated for examinations of the genus Brevoortia 

 (Anderson and McDonald, 2007). PCR products were 

 electrophoresed with a Beckman-Coulter CEQ'" 8000 

 automated capillary system, with a 400 base pair (bp) 

 standard (Beckman Coulter, Inc., Fullerton, CA), accord- 

 ing to the manufacturer's suggested protocol. Microsatel- 

 lite allele sizes were estimated with Beckman-Coulter 

 fragment analysis software (Beckman Coulter, Inc., 

 Fullerton, CA), by using allelic bins that were based on 

 the analysis of 1276 DNA samples from the four North 

 American menhaden species collected over four years 

 (data not shown). 



The freeware program Fstat vers. 2.9.3.2 (Goudet, 

 1995) was used to calculate observed iHJ and expected 

 (H^) heterozygosity and allelic diversity ik^), and to 

 identify microsatellite loci which deviated significantly 

 from Hardy- Weinberg equilibrium expectations (HWE). 

 Deviation from HWE was tested by calculating the 

 inbreeding coefficient {F^J in the overall data set, and 

 within each individual sampling locale. Statistical sig- 

 nificance was assessed in a randomization procedure 

 with 100 iterations. For this analysis, 13 populations 

 of B. patronus (/i=20) ranging from southern Texas 

 to southern Florida were used. Individuals were then 

 pooled into species and sampling locales (n=20 in each 

 sampling locale, in locales where two species were cap- 

 tured, 20 of each species were used). The genetic diver- 

 gence (F^f, Weir and Cockerham, 1984) among the four 



