Determination of carbon, carbonate, nitrogen, and phosphorus in freshwatersediments by near-infrared reflectance spectroscopy: Rapid analysis and a check on conventional analytical methods

Sediments are typically analyzed for C, N, and P for characterization, sedi ment quality assessment, and in nutrient and contaminant studies. Cost and time required for analysis of these constituents by conventional chemical t echniques can be limiting factors in these studies. Determination of these constituents by near-infrared reflectance spectroscopy (NIRS) may be a rapi d, cost-effective method provided the technology can be applied generally a cross aquatic ecosystems. In this study, we explored the feasibility of usi ng NIRS to predict total C, CO3-2 organic C, N, and P in deep-water sedimen t cores from four Canadian lakes varying over 19 degrees of latitude. Conce ntration ranges of constituents in the samples (dry weight basis) were tota l C, 12-55; CO3-2, 6-26; organic C, 7-31; N, 0.6-3.1; and P, 0.22-2.1 mg g( -1). Coefficients of determination, r(2), between results from conventional chemical analysis and NIR-predicted concentrations, based on calibrations across all the four lakes, were 0.97-0.99 for total C, organic C, and N. Pr ediction for CO3-2 was good for the hard water lake from a calibration acro ss all four lakes, but this constituent in the three soft water lakes was b etter predicted by a calibration across the soft water lakes. The NIR calib ration for P fell below acceptable levels for the technique, but proved use ful in the identification of outliers from the chemical method that were la ter removed with the re-analysis of several samples. This study demonstrate d that NIRS is useful for rapid, simultaneous, cost-effective analysis of t otal C, CO3-2, organic C, N, and P in dried sediments from lakes at widely varying latitudes. Also, this study showed that NIRS is an independent anal ytical tool useful for the identification of outliers that may be due to er ror during the analysis or to distinctive composition of the samples.