Search

Near-infrared reflectance (NIR) spectroscopy was evaluated for use in predicting the plant-available N (PAN = NH4-N + NO3-N) and mineralizable N (MN) from poultry litter (PL) samples on an "as received" water content basis. Near-infrared reflectance scans were collected on 118 PL samples of variable water content without drying. The samples were not fresh but had been stacked or otherwise stored for various lengths of time and were closely representative of the PL applied to crop and pasture soils in Georgia (United States). Average PAN for the samples studied was 6.3 g kg-1 PL (dry weight). One-hundred-day soil incubations with a loamy fine sand and a sandy clay loam were used to determine MN. An average of 19% of the organic N was mineralized after 100 d. Total PAN after 100 d (PAN100 = PANinitial + MN) averaged 33% of the total N for these samples. Near-infrared reflectance calibration models were developed with a modified partial least squares procedure. The results show that NIR calibration models can be developed on PL with variable water contents for total N (r² = 0.896, ratio of performance to deviation [RPD] = 3.5), NH4-N (r² = 0.795, RPD = 3.2), NO3-N (r² = 0.926, RPD = 2.7), uric acid N (r² = 0.909, RPD = 4.0), organic N (r² = 0.821, RPD = 3.6), water-soluble organic N (r² = 0.897, RPD = 4.0), MN (r² = 0.842, RPD = 2.6), initial PAN + MN (r² = 0.888, RPD = 2.6), total C (r² = 0.926, RPD = 3.7), and water content (r² = 0.996, RPD = 15.6). We conclude that drying is not necessary to develop highly successful NIR calibrations for N forms from PL. [ABSTRACT FROM AUTHOR]