834 



Fishery Bulletin 92(4). 1994 



Point number 



Figure 18 



Individual life history trajectories (interior-most 40 points) for each 

 of the six elements (Ca, Na, Sr, K, S, and CI) detected in the sagittae 

 of juvenile Nemadactylus macropterus collected at Cygnet (SE 

 Tasmania). To minimize potentially confounding effects of varia- 

 tion in date of settlement, the comparison was restricted to juve- 

 niles collected on the same day (19 December 1987) and in the 

 same size range (7-11 cm standard length). The data for each in- 

 dividual have been filtered through a five-point moving average. 



nator. Two discriminant functions (2 and 3) load onto 

 Sr at r>0.5 in the original analysis, whereas Sr does 

 not achieve this load level for any function in the 

 point 2-6 analysis. Reflecting this, mean Sr concen- 

 trations differ among nursery areas at P<0.001 

 (F 5no =6.21) in the original analysis, but at only 

 P<6.02 CF 5iU0 =2.82) for the point 2-6 analysis. By 

 comparison, differences among nursery areas for the 

 other five elements are significant at similar levels 

 for the two analyses. 



Discussion 



Effects of data quality on stock delineation 



Electron-probe microanalysis with WD-spectrom- 

 eters revealed extensive variability in the concen- 



trations of six elements in N. macropterus otoliths. 

 Some of this variability is induced by the inherent, 

 small-scale compositional heterogeneity of otoliths 

 and some is noise that reflects the limits of detect- 

 ability and precision of the electron probe. However, 

 comparisons of life history scans along similar growth 

 axes of left and right otolith pairs indicate signifi- 

 cant ontogenetic variability for all elements. For most 

 elements, there is also evidence of geographic vari- 

 ability in composition. 



The extent to which this ontogenetic and geo- 

 graphic variability can be used to detect differences 

 in either life histories or population structure criti- 

 cally depends on the scale of the life history or popu- 

 lation 'signal' relative to analytical 'noise.' In that 

 regard, data quality varies widely among elements. 

 Two identifiable sources of this 'error' are the effects 



