Thresher et al.: Otolith analysis of Nemadactylus macropterus 



833 



each specimen the mean composition of 

 points 35—39, inclusive, in a standard life 

 history scan (680 to 780 urn from the pri- 

 mordium). All juveniles used in these analy- 

 ses had otoliths at least this large. 



Discriminant analysis of the juveniles 

 indicated highly significant differences 

 among all six areas sampled. The weak- 

 est discriminator (the fifth root of the dis- 

 criminant analysis) was significant at 

 P<0.01. The preliminary conclusion then 

 is that there are signatures specific to each 

 nursery-area that could be sought in the 

 adult population. 



Further analyses of the data, however, 

 indicated this conclusion was premature. 

 Specifically, if there are nursery-area-spe- 

 cific environmental signals in the otolith, 

 then we would expect them to be manifest 

 ontogenetically in either or both of two 

 ways. First, we would expect that at the 

 end of the larval period (approximately 

 points 25-30), the mean concentrations of 

 various elements would diverge among 

 sites, reflecting the specific environment 

 at each (i.e. the nursery area 'fingerprint'). 

 Second, we would also expect that among 

 individuals, concentrations of these same 

 elements would converge within sites, re- 

 flecting recruitment into a common envi- 

 ronment. Again, this convergence should 

 occur at approximately points 25—30. For the 

 second prediction, we analyzed in detail one 

 site (Cygnet) for which the sample size of 

 juveniles was large enough that we could 

 reduce possible variability due to differences 

 in date of recruitment. This was done by 

 examining juveniles caught on the same day 

 and falling within a narrow size range (7-11 cm SL). 



For most elements (all but Sr), neither prediction 

 is supported by the data (Figs. 17 and 18). Although 

 variance is high at all points, there is little or no 

 indication of either divergence among sites (in the 

 case of mean concentrations) or convergence among 

 individuals (in the case of variation within the single 

 site) at or near points 25-30 for any element other 

 than Sr. For Sr, however, both predictions appear to 

 be borne out. Mean concentrations overlap broadly 

 among juveniles from all sites during the larval stage 

 but diverge significantly among sites at about point 

 25. Among individuals, juveniles at Cygnet appear 

 to converge on two different postrecruitment Sr tra- 

 jectories, also beginning at about point 25. The avail- 

 able evidence suggests that concentrations of ele- 

 ments other than Sr are largely unaffected by the 



Derwent « Nutgrove ° Maria l.«W. Tas. 'Phillip l.°Cygnet 



Point number 



Figure 1 7 



Group mean concentrations of each of the six detected elements 

 (Ca, Na, Sr, K, S, and CI) for the interior-most 40 points analyzed 

 along the long growth axis of sagittae of juvenile Nemadactylus 

 macropterus caught at each of the six nursery areas sampled (see 

 Fig. 1 and Table 1). Point 1 is on the primordium. Data for each 

 individual were filtered through a five-point moving average be- 

 fore the group mean was calculated. 



transition from the larval to the juvenile stages and 

 that the apparent discrimination among nursery- 

 areas is the manifestation of differences among in- 

 dividuals already evident in their larval stages. 



We tested this conclusion by reanalyzing for "nurs- 

 ery-area-specific signals" using data for points 2-6 

 (early larval life) rather than for points 35-39 (early 

 juvenile stage). In general, the results were similar 

 to those obtained with points 35-39, with good dis- 

 crimination among most nursery grounds and a com- 

 parable level of overall site separation (Wilk's lam- 

 bda=0.20 for points 2-6 vs. 0.22 for points 35-39, 

 P<0.001 in both cases). However, the accuracy of cor- 

 rectly assigning juveniles to nursery areas was less 

 in the point 2-6 analysis (51% vs. 82%), which re- 

 flects the divergence of Sr concentrations in the nurs- 

 ery areas and its increased importance as a discrimi- 



