Li et al a key to selected Sebostes spp, based on mitochondrial DNA restriction fragment analysis 



189 



product with eiidonuclease BstN I. Continue digesting 

 with the specified restriction endonucleases, identifying 

 the resultant haplotype, and continue to the next step 

 until the species has been identified. Between one and 

 five restriction enzymes will be needed to achieve the 

 separation. Examining the fragment patterns of addi- 

 tional restriction enzymes can increase confidence in the 

 identifications (see key). 



Although variation observed in the 12S/16S region 

 was not used in the key, it does provide alternatives to 

 species identification that may be useful for resolving 

 the identification of some species or for corroborating 

 identifications, especially when there is intraspecific 

 variation that has not been previously observed for the 

 enzymes applied in resolving species. In particular, Rsa 

 I, Dde I, and Mbo I exhibited substantial interspecific 

 variation in the 12S/16S region. In addition, if one has 

 reduced the possible species by other means, a single 

 12S/16S digest can be used in some instances. For ex- 

 ample, S. aleutianus and S. borealis differ in fragments 

 produced by Bs^N I and by Rsa I, and S. caurinus and 

 S. maliger also differ in fragments produced by Rsa I. 



Discussion 



Restriction site analysis of mtDNA is a simple and effec- 

 tive tool for identifying juveniles of Sebastes species. 

 Because inexpensive equipment is used to obtain data 

 from restriction fragment patterns, the analyses can 

 be conducted in most laboratories, including many high 

 school laboratories. 



Although the restriction fragment key identified most 

 (58 or 81.7%) of the 71 different rockfish species we 

 evaluated, it failed to identify 13 species. The identities 

 of those 13, however, were narrowed to five small groups 

 of species. One group, S. carnatus and S. chrysomelas, 

 are very closely related; they differ obviously only in 

 body coloration as adults: S. carnatus has flesh-col- 

 ored blotches on an olive-brown background, and S. 

 chrysomelas has yellow blotches on a black background 

 (Love et al., 2002). They are included in the subgenus 

 Pteropodus (Kendall, 2000). Another group includes S. 

 chlorostictus, S. eos, and S. rosenblatti, which are also 

 morphologically similar, occur sympatrically, and are 

 closely related (Chen, 1971; Love, 1996; Rocha-Olivares 

 et al., 1999; Love et al., 2002); all are members of the 

 subgenus Sebastomus. Estimates of divergence times 

 of the subgenus Sebastomus suggest that the three 

 species are the result of the most recent speciation 

 events within the subgenus, which may have begun less 

 than 140,000 years ago (Rocha-Olivares et al., 1999). 

 Members of a third group, S. emphaeus-S. variegatus- 

 S. wilsoiii, are assigned to the subgenus Allosebastes 

 (Kendall, 2000). The relationships between species in 

 the other two unresolved groups are not as clear be- 

 cause the subgenera of the members differ. In one of 

 those groups, S. entomelas is in subgenus Acutomen- 

 tum and S. mystinus is in subgenus Sebastosomus. In 

 the other subgroup, S. polyspinis remains unassigned 



to a subgenus, S. crameri is in Eosebastes, and iS. cil- 

 iatus (subgenus Sebastosomus) has only recently been 

 separated from S. variabilis (Orr and Blackburn, 2004). 

 The similarities observed may actually reflect genetic 

 relationships because the subgenera assignments are 

 probably inaccurate reflections of phylogeny (Kendall, 

 2000). Our inability to resolve within those five groups 

 of rockfish to species indicates the need for additional 

 markers. Approaches for obtaining such markers include 

 screening additional regions of the mtDNA and appli- 

 cation of additional restriction enzymes. If additional 

 mtDNA regions and restriction enzymes do not provide 

 species-specific information, other molecular techniques 

 such as microsatellites should be considered. 



We applied the baseline data from which our key was 

 developed to identification of recently extruded rock- 

 fish larvae in Southeast Alaskan waters (Gray et al., 

 2006). That application, which was made while the key 

 presented in this article was being developed, evolved 

 over subsequent years of application. The key was also 

 used to delineate juvenile rockfishes from the southern 

 California Bight (Li et al., in press). We were able to 

 identify all the specimens to species or to narrow iden- 

 tification to one of the five small groups of genetically 

 similar species. The identifications of juvenile rockfish 

 were concordant with genetic and morphological crite- 

 ria, except that the genetic key did resolve species in 

 the closely related subgenus Sebastotyius, which could 

 not be resolved with morphological criteria. The larval 

 specimens, in contrast, could not be identified to spe- 

 cies with morphological criteria. We detected previously 

 unobserved intraspecific variation in both studies as 

 well. Neither the variation observed in the larval and 

 juvenile studies, the intraspecific variation observed in 

 developing the key, nor intraspecific variation observed 

 among large numbers of individuals of several species 

 examined as part of a population genetic analysis (Li, 

 2004) obscured species detection. 



Molecular genetic keys can remove much of the guess- 

 work for species determination in ichthyoplankton and 

 juvenile surveys. Because DNA-based characters re- 

 main constant throughout the life of an individual, 

 genetic divergence can provide unequivocal markers 

 for species delineation. For specific questions involving 

 the identification of small numbers of species, such as 

 distinguishing between the two sibling species included 

 in S. aleutianus (Gharrett et al., 2005), it is possible 

 to develop assay methods (e.g., single nucleotide poly- 

 morphism — referred to as SNP (Collins et al., 1996), 

 that can rapidly identify large numbers of specimens. 

 Finally, such a key can be used to identify larval and 

 juvenile species of Sebastes so that the variation in the 

 morphological characters can be determined. 



Acknowledgments 



This work represents, in part, the master's thesis work 

 of Z. Li at the University of Alaska Fairbanks. The pro- 

 ject was supported by funding from the U.S. Geological 



