Thresher et al.: Otolith analysis of Nemadactylus macropterus 



821 



tium), the L a line. S and CI were measured on Spec- 

 trometer 1 (PET), K and Ca on Spectrometer 2 (PET), 

 and Na and Sr on Spectrometer 3 (TAP). Matrix cor- 

 rections were made by using the "PAP" (Pichou and 

 Pichoir, 1984) matrix conversion software. Minimum 

 detection limits and confidence intervals for the con- 

 centration estimates are based on equations provided 

 byAnceyetal. (1978). 



Ontogenetic variation in composition was assessed 

 by analyzing a series of points along the longest 

 growth axis of each otolith section — a "life history 

 scan." The finished section of the sagitta of N. 

 macropterus exposes a nearly straight, uninterrupted 

 growth axis, through which we ran a series of pro- 

 grammed scan lines that tracked the slight curva- 

 ture of the axis (Fig. 2). We duplicated this axis as 

 closely as possible in each specimen in order to maxi- 

 mize comparability of the data sets. The life history 

 scan line for each fish ran from the primordium to 

 the posterior ventral tip of the otolith. The size of 

 the scan points and their spacing were determined 

 by logistic considerations and, in part, by the results 

 of experimental trials. In practice, routine spacing 

 (center to center) between points was 25 urn. In the 

 case of the parallel scans (see Fig. 6), we used a beam 

 diameter of 14 urn spaced 16 urn apart and 6 |im 

 diameter beams spaced 8 um apart. Reproducibility 

 of life history scan data was evaluated by comparing 

 left and right otolith pairs from the same fish. 



Even if the data are free from conspicuous distor- 

 tions due to irregular surface features, the use of data 

 from a single-point probe analysis for evaluation of 

 stock structure still risks high error rates due to 

 measurement noise (Gunn et al., 1992). Conse- 

 quently, we analyzed stock structure in two ways: 

 based on comparisons of single-point data collected 

 adjacent to the primordium (point 2); and based on 

 mean concentrations of the first five probe points 



adjacent to the primordium (points 2-6 inclusive). 

 The latter method filters out high-frequency mea- 

 surement noise but risks low discriminant power by 

 including information from relatively late in larval 

 life. Point 6 is about 125 um from the primordium; 

 increment counts suggest this corresponds to a lar- 

 val age of about 45-55 days. 



Statistical procedures in general follow Sokal and 

 Rohlf (1981). Test for normality were made by using 

 Lilliefors K-S test. Differences in mean concentra- 

 tions among sites were tested by means of a Kruskal- 

 Wallis nonparametric AN OVA. Groupings of sites and 

 specimens were tested and quantified by linear dis- 

 criminant function analysis (LDFA) with the SYSTAT 

 statistical software package. General procedures for 

 and assumptions underlying LDFA are described by 

 Klecka (1980); Cameron (in press) reviews the ap- 

 plication of discriminant analysis to studies of otolith 

 and skeletal composition. 



Results 



Composition of N. macropterus otoliths; 

 data quality and reproducibility 



Six elements could be reliably detected in N. macro- 

 pterus otoliths by wavelength-dispersive electron 

 probe microanalysis (WD-EPMA): Ca, Na, Sr, K, S, 

 and CI, in order of decreasing mean concentration. 

 The elements in N. macropterus sagittae constitute 

 three distinct sets separated in concentration from 

 other less abundant elements by a difference of one 

 to three orders of magnitude (Fig. 3): Ca, carbon, and 

 oxygen (the last two are not routinely measured be- 

 cause of methodological difficulties) constitute the 

 'macro-constituents' present in concentrations >10% 

 (10 5 ppm) by weight; Na, Sr, K, S, and CI constitute 

 the 'micro-constituents,' which occur in mean con- 



Figure 2 



Diagram of section of a sagitta of an adult Nemadactylus macro- 

 pterus, as prepared for electron probe microanalysis, depicting pro- 

 grammed transects used to analyze points along the main growth 

 axis from the primordium (p) to the otolith's posterior edge (pe). 



