424 



Fishery Bulletin 90|2). 1992 



CALCIUM 



Sr/Ca RATIO 



2 4 6 8 10 12 



TRANSECT POSITION 

 Figure 2 



Example of changes in Sr and Ca concentration and the Sr/Ca 

 ratio along one of 12 transects made on a Dover sole otolith. 

 A Tfjm beam and 20-sec counting time were used. Position 

 1 is the point closest to the central primordium, and position 

 12 is closest to the otolith edge. 



was >100nm inside the OTC mark. Starting and end- 

 ing points for all transects were referenced to specific 

 growth areas identified by dark continuous bands, and 

 the remaining points were evenly spaced between these 

 two points. Locations at the start of the transects were 

 in a translucent area of the otolith assumed to have 

 little organic material (Dannevig 1955), while the end 

 points were in a more opaque area, which probably con- 

 tained more organic material. 



Exposure time and accuracy 



Counting time and exposure time were distinguished 

 in this experiment. Counting time is the minimum time 

 the specimen is exposed to the electron beam, while 

 exposure time also includes the time necessary to col- 

 lect background counts and counts of other elements. 

 Six sequential analyses were made at each of six loca- 

 tions (combined N 36) to determine changes in elemen- 

 tal concentration. The locations were the start and end- 

 points of each 20-sec transect used for the precision 

 analysis. Sequential analyses at each location were 



45 

 4 



3.5 

 3 



25 

 2 



1,5 



02 04 0,6 08 1 1.2 1.4 



Sr MOLE FRACTION (K 01) 



Figure 3 



Relation of coefficient of variation (measurement error) to 

 elemental concentration for transects from one Dover sole 

 otolith. Each regression represents three transects with 12 

 points each (A^ 36). Regression equations are presented in 

 Table 2. 



made in increments of 20-sec counting times, which cor- 

 responded to exposure times of 65, 130, 195, 260, 325, 

 and 390 sec. These exposure times were approximate- 

 ly twice as long as those which would result from an 

 analysis of Sr and Ca alone, because S was also ana- 

 lyzed (results not reported). 



Statistical analyses 



The effect of elemental concentration on Sr precision 

 (CV's) was examined with linear and nonlinear regres- 

 sions. To determine if beam-power density affected Sr 

 precision, multiple regressions containing normalized 

 concentration and "dummy variables" corresponding 

 to beam size were analyzed with partial-F tests (Neter 

 et al. 1989:364-370). Each of the four counting times 

 was analyzed separately. 



