836 



Fishery Bulletin 92(4), 1994 



indirectly the potential of the technique to resolve 

 spawning stock structure. Specifically, we sought 

 evidence of regionally different patterns in otolith 

 composition that might reflect stock structure. 



In that regard, concentrations measured near the 

 primordium differed significantly for four of six ele- 

 ments among adults from the six sites. In all four 

 cases, the range of mean values among sites exceeded 

 an empirically derived 'minimum significant differ- 

 ence' by at least 50% (Table 2). Furthermore, the 

 pattern of differences among sites appeared to be 

 regionally based: the two Tasmanian samples pooled 

 together in the discriminant analysis, as did the geo- 

 graphically contiguous NSW and Victorian samples. 

 Such a grouping of sites could imply any of four dif- 

 ferent mechanisms: 1) all sites differ, and the group- 

 ing is a statistical artefact of the small number of 

 sites and individuals sampled; 2) regional differences 

 result from retrospective changes in otolith chemis- 

 try in response to the latest conditions encountered 

 by each adult, and adjacent sites pool because their 

 environmental characteristics are more similar than 

 those of widely separated sites; 3) the sites pool be- 

 cause each regional set derives uniquely from a com- 

 mon spawning ground or spawning population; and 

 4) each set is derived from a number of spawning 

 grounds or populations that have similar chemical 

 fingerprints, within which individuals mix widely 

 and the boundaries of which are set by constraints 

 on adult or larval mixing. 



The possibility that the regional groupings are an 

 artefact is difficult to evaluate without knowing the 

 range of chemical fingerprints possible and their like- 

 lihood of occurrence. Assuming three chemical phe- 

 notypes randomly distributed among six individu- 

 als (=sites), then the probability that at least two 

 adjacent sites will have identical characteristics is 

 extremely high. However, given the number of pos- 

 sible permutations, the probability that all pooling 

 of sites will be only among nearest neighbors is less 

 than 0.01. Therefore, we reject the hypothesis that the 

 apparent regional groupings are a statistical artefact. 



We also think it unlikely that the groupings (and 

 similarity offish within sites) are the result of retro- 

 spective modification of otolith chemistry. It is a 

 consistent assumption of otolith-based aging stud- 

 ies that otolith structure is not modified after depo- 

 sition. A similar assumption underlies chemical stud- 

 ies, although there are no experimental data to verify 

 the point (as opposed to studies on scales, the chemi- 

 cal compositions of which are modifiable retrospec- 

 tively, e.g. Sauer and Watabe, 1989). In fact, it is 

 likely that at least some water- and alcohol-soluble 

 compounds are transported into or out of otoliths 

 during preservation. However, our data are not con- 



sistent with such retrospective modification of the 

 micro-constituents. Nemadactylus macropterus col- 

 lected at the same site and time show little evidence of 

 convergence on a common marginal composition. This 

 implies that recent environmental history has little or 

 no effect on the composition of the otolith margin and 

 presumably even less on the interior. Where a common 

 marginal composition was evident, as in Sr levels 

 among juveniles collected in the same area, it appears 

 to be related to an environmental effect during deposi- 

 tion rather than to retrospective modification. 



Distinguishing between the other two hypoth- 

 eses — a single spawning ground for each regional 

 phenotype or multiple spawning grounds with region- 

 ally restricted mixing — is not possible without addi- 

 tional information. As noted, information on the re- 

 productive biology of Australian N. macropterus is 

 sparse. Smith (1989) found running-ripe individu- 

 als in autumn (February-March) off NSW; we found 

 large numbers of relatively young larvae present 

 along the east, but not the west, coast of Tasmania 

 (for sampling sites and protocol, see Thresher et al., 

 1989), and several unpublished reports indicate simi- 

 lar larvae off Victoria and South Australia (in the 

 Bight). These scattered observations suggest that N. 

 macropterus spawn at a number of sites along the 

 southeastern Australian coast and certainly spawn 

 in each of the three regional groupings of sites iden- 

 tified by otolith chemical analysis. But sampling is 

 not yet detailed enough to determine whether there 

 are discrete spawning areas, or whether spawning 

 occurs in a continuous band of activity all along the 

 coast. Genetic data provide little additional informa- 

 tion. Richardson's (1982) samples were drawn from 

 Tasmanian and NSW/Victorian sites and hence ap- 

 pear to bracket two otolith-based regional groupings 

 but indicate no significant genetic differences across 

 this range. This result has recently been confirmed 

 by Elliott and Ward ( 1994) for allozymes and Grewe 

 et al. (1994) for mitochondrial DNA. 



The lack of genetic differentiation in southeastern 

 Australian N. macropterus populations is consistent 

 with our observations of apparent examples of lar- 

 val mixing. Probe microanalysis of otoliths of juve- 

 niles from Victorian and Tasmanian coastal habitats 

 indicated that most are similar in composition to 

 those of adults collected at the same sites, which 

 suggests regional, self-recruiting populations. How- 

 ever, the distribution of the probabilities that each 

 juvenile originated in the region where it was col- 

 lected was conspicuously bimodal. Four out of ten 

 juveniles caught off Victoria had chemical phenotypes 

 more typical of Tasmanian (3) or Bight (1) origin, 

 whereas 8 of 106 Tasmanian-caught juveniles clas- 

 sified mainly with the NSW/Victorian adult sample. 



