Comparison of the silver nitrate and bacterial denitrification methods for the determination of nitrogen and oxygen isotope ratios of nitrate in surface water.

Nitrogen (N) and oxygen (O) isotope ratios of NO(3) (-) are often used to trace dominant NO(3) (-) pollution sources in water. Both the silver nitrate (AgNO(3)) method and the bacterial denitrification method are frequently used analytical techniques to determine delta(15)N- and delta(18)O-NO(3) (-) in aqueous samples. The AgNO(3) method is applicable for freshwater and requires a concentration of 100-200 micromol of NO(3) (-) for isotope determination. The bacterial denitrification method is applicable for seawater and freshwater and for KCl extracts of soils with a NO(3) (-) concentration as low as 1 micromol. We have carried out a thorough method comparison using 42 real surface water samples having a wide range of delta(15)N- and delta(18)O-NO(3) (-) values and NO(3) (-) concentrations. Various correction pairs using three international references and blanks were used to correct raw delta(15)N- and delta(18)O-NO(3) (-) values. No significant difference between the corrected data was observed when using various correction pairs for each analytical method. Both methods also showed excellent repeatability with high intraclass correlation coefficients (ICC). The ICC of the AgNO(3) method was 0.992 for delta(15)N and 0.970 for delta(18)O. The ICC of the bacterial denitrification method was 0.995 for delta(15)N and 0.954 for delta(18)O. Moreover, a positive linear relationship with a high correlation coefficient (r >or= 0.88) between the two methods was found for delta(15)N- and delta(18)O-NO(3) (-). The comparability of the methods was assessed by the Bland-Altman technique using 95% limits of agreement. The average difference between results obtained by the bacterial denitrification and the AgNO(3) method for delta(15)N was -1.5 per thousand with 95% limits of agreement -3.6 and +0.5 per thousand. For delta(18)O this was +2.0 per thousand, with 95% limits of agreement -3.3 and +7.3 per thousand. We found that for delta(15)N and for delta(18)O, 97% of the differences fell within these 95% limits of agreement. In conclusion, the AgNO(3) and the bacterial denitrification methods are highly correlated and statistically interchangeable.