Pollard et al.: Chemical marking of Pagrus auratus 



121 



Source 



an hour prior to weighing. The tubes 

 were sealed during this time to reduce 

 moisture uptake which could have 

 caused an increase in weight. Samples 

 were weighed to the nearest 0.001 mg. 

 The remaining steps took place in a 

 laminar flow cabinet to avoid contami- 

 nation. With a pipette we placed 50 |iL 

 of 1% nitric acid (H2NO3) into the ep- 

 pendorf tube containing the spines for 

 10 seconds and then removed the acid. 

 This procedure cleaned possible exter- 

 nal contaminants from the surface of 

 the sample. The time and concentration 

 needed for cleaning were determined by 

 microscopic examination of the spines 

 to check for excessive etching and loss 

 of bone. The contents of the eppendorf 

 tube were then dissolved in a proportional amount 

 of 75% nitric acid to produce a standard 1 mg of spine 

 tissue per 10 |iL of solution. After allovraig an hour or 

 more for the sample to fully dissolve in the sealed tube, 

 10 |iL of solution was removed and added to a sample 

 tube containing 4950 |iL of milliQ water and 40 |iL of 

 concentrated nitric acid. This resulted in a standard- 

 ized 0.2 mg/|iL sample solution in 1% nitric acid 

 which became the sample concentration used in our 

 experiment after preliminary analyses with ICP-MS. 



Measurement of isotopes 



Inductively coupled plasma-mass spectrometry (ICP- 

 MS) was used to measure concentrations of stron- 

 tium in the dorsal spine tissue. The model used was 

 a Perkin-Elmer SCIEX ELAN 5000 with a Gilson 

 212B autosampler for sample introduction, and pro- 

 cedures were similar to those described in Dove et. 

 al. (1996). The ICP-MS was optimized with a 0.010 

 mg/L standard solution covering the extremes of the 

 mass range. By using the graphics application, we 

 were able to adjust the nebulizer flow to maximize 

 the i°^Rh signal and balance the signals for ^'^Mg and 

 208pij fj.pj^ ^^Q standard solution. The quantitative 

 analysis function of the ICP-MS was externally cah- 

 brated by using calibration standards with concen- 

 trations of 1, 10, 100, and 500 parts per billion, i.e. 

 those covering the range of expected concentrations. 

 The correlation coefficient of the calibration curve was 

 generally 1.000 for *^Sr, the isotope used for analysis. 

 ICP-MS interferences such as isobaric overlaps or 

 plasma-induced polyatomic ions are generally negligible 

 for strontium. A blank solution containing only the 1% 

 nitric acid matrix was also analyzed prior to the sample 

 solutions, so that the isotopic concentration of the acid 

 matrix could be subtracted. 



Table 1 



Analysis of variance for experiment 1: persistence over 36 days. Three- 

 way partially hierarchical analysis of variance to test for differences in 

 dorsal spine *^Sr among sample times 0, 12, 24, and 36 days (fixed 

 factor=time). treatments (treat) (Sr^* zero, 5x and lOx ambient; fixed fac- 

 tor), and tanks (random factor). Nonsignificance at P=0.05 is indicated by 

 n.s. Cochran's test; df = 1; no. variances (*) = 48; and C = 0.1565; variances 

 homogeneous; no transformation. 



df 



Mean square F-ratio Probability 



Time 3 13699.2 6.690 <0,005 



Treatment 2 568635.5 105.485 <0.001 



Time X treat 6 4536.0 2.215 n.s. 



Tanks (treat) 9 5390.7 2.632 <0.050 



Time X tanks (treat) 27 2047.8 0.695 n.s. 



Residual 48 2946.5 



Quality control solutions were analyzed for *^Sr at 

 the start, middle, and end of each analytical run for 

 each date of anaiysis, allowing the stability of the 

 anal3^e signal to be monitored both within and be- 

 tween runs. The quality control readings obtained 

 were not constant over time, owing perhaps to clogged 

 sampling cones. The results for the spine samples, 

 therefore, required adjustment in relation to the 

 quality control results. We assumed that there was 

 no change in the *^Sr concentration of the quality 

 control solution over time. External drift correction 

 can significantly improve both accuracy and preci- 

 sion (Jarvis et al., 1992). 



Treatment of data 



Data were analyzed by analysis of variance accord- 

 ing to the general recommendations of Underwood 

 (1981). Normality of data was tested with Cochran's 

 test. Student-Newman-Keuls (SNK) multiple range 

 tests were used for a posteriori comparisons among 

 treatments where appropriate. Where data were 

 missing (<2 per design), the missing values were re- 

 placed by the means for their group, and the degrees 

 of freedom of the residual were reduced by the num- 

 ber of missing values. 



Results 



Persistence of strontium over 36 days 



Differences in the ^^Sr concentrations of dorsal spines 

 were found among treatments, despite differences 

 among tanks within treatments (Table 1; Fig. 1). Fish 

 exposed to strontium at lOx ambient incorporated 

 more strontium into their spines than fish at 5x 



