Influence of Sample Size on Measurement of Soil Denitrification

The influence of sample size on the magnitude and variability of soil denitrification was studied by collecting soil cores, ranging in size from 1.7 to 5.4 cm in diameter, from no-till and conventionaltill corn plots. Estimates of natural denitrification rates were obtained by incubating intact soil cores with C2H2 and monitoring gaseous N2O production. In addition, maximum denitrification potential was determined by monitoring N,O production in anaerobic slurries amended with glucose, NOj and C,H2. Natural rate estimates were highly skewed and approximated lognormal distributions. The spatial variability of denitrification was characterized by large variation at small distances of < 10 cm and only weak spatial dependence at distances of 10 to 100 cm. Studies of the effect of sample size on denitrification suggest that soil cores >4.2 cm in diameter yielded the most reliable estimates of natural denitrification rates. Using a 1 United States Department of Agriculture-Agricultural Research Service, Soil Nitrogen and Envir. Chem. Lab., Beltsville Agric. Res. Cent., Beltsville, MD 20705. Received 27 Aug. 1986. 2 Research Microbiologist, Research Soil Scientist, and Research Soil Scientist, respectively. Mention of a product by company or name is not an expressed or implied endorsement by the USDA. computerized random resampling technique, we estimated that approximately 10 to 15 kg of soil was necessary to obtain a representative soil mass for estimating natural denitrification rates. The results of this study are consistent with the hypothesis that the source of variability associated with the natural denitrification rates is the patchy distribution of denitrifying "hot spots" in soil. Some implications associated with the application of classical statistical methods to lognormal data are also discussed. Additional Index Words: lognormal, spatial variability, geostatistics, nitrous oxide. Parkin, T.B., J.L. Starr, and J.J. Meisinger. 1987. Influence of sample volume on measurement of soil denitrification. Soil Sci. Soc. Am. J. 51:1492-1501. /"^ ASEOUS N LOSS from denitrification is one of the vJ least well quantified of the soil N cycle processes. A better understanding of soil and environmental controls as well as better quantitative estimates of denitrification are required before PARKIN ET AL.: INFLUENCE OF SAMPLE VOLUME ON SOIL DENITRIFICATION 1493 management practices can be developed to moderate the effect of this process on agricultural systems. Difficulties associated with quantification of denitrification can be attributed to the lack of methodology to both measure the process and to deal with the spatial and temporal variability. Recently, a variety of techniques have been developed in an attempt to estimate natural rates of soil denitrification. These include measurements of: (i) Ngas flux rates from N-fertilized soils (Rolston et al., 1982; Siegel et al. 1982) (ii) N2O flux rates from the soil surface in C2H2-amended soil chambers (Ryden and Dawson, 1982; Burton and Beauchamp, 1984; McConnaughey and Duxbury, 1986), and (iii) N2O production rates from C2H2-amended intact soil cores (Rice and Smith, 1982; Aulakh et al., 1984; Parkin et al., 1985). Regardless of the relative merits and limitations of these approaches, a problem inherent to all attempts to quantify natural rates of denitrification has been the high spatial variability associated with this process. This high variability has been acknowledged, yet few studies have focused on this problem (Folorunso and Rolston, 1984; Folorunso and Rolston, 1985). Recent research on the variability of other soil properties and the application of novel statistical methods has aided in the characterization of this variability. The purpose of this study was to investigate the influence of sample size on estimates of natural denitrification rates and on the variability associated with these rate estimates. The statistical nature of these analyses also requires a discussion and critical evaluation of the commonly used statistical methods for hypothesis testing. MATERIALS AND METHODS Field Site and Sampling Samples were collected from a field site located on the University of Maryland Plant Research Farm, Beltsville, MD. The site had long-term (12 yr) plots of no-till and conventional-till continuous corn. The soil at this site was a Beltsville silt loam (Typic Fragiudult) having a pH of 6.5 and total organic N (Kjeldahl) and C (persulfate digestion) contents of 0.8 g/kg and 5.1 g/kg, respectively. All plots received 168 Kg N/ha of NH4NO3 fertilizer in April, 1984. The conventional-till plots were plowed at the end of May 1984 and all plots were planted to corn in early June 1984. An experiment consisted of collecting soil samples from 36 blocks (20 by 30 cm experimental units utilizing two block spacing patterns; Table 1). In the two spring experiments (prior to planting) the blocks were spaced adjacent to one another in a 6 block by 6 block grid which covered a 1.2 m by 1.8 m area. In the two summer sampling experiments the blocks were spaced between corn rows at 76-cm intervals. Five different sized soil cores were collected at random Table 1. Description of sampling dates, treatments, and experimental unit spacing (block spacing) for the sample size experiment. from within each block for each experiment (Table 2). Cores were obtained by pounding steel tubes, fit with hardened cutting bits, 16 cm into the ground. The intact soil cores contained within the steel tubes were then transferred into plastic tubes which were then stoppered. The intact soil cores fit loosely in the plastic tubes to facilitate gas diffusion into and out of the soil. After all five soil cores were removed from the block, the remaining soil was removed using rectangular steel templates (20 by 30 cm), which were also driven 16 cm into the soil. The soil from the large blocks was not maintained in an intact state but sieved (0.5 cm mesh) and stored at 4°C overnight. This sampling protocol produced a total of 216 soil samples (36 of each of the 5 different size soil cores plus the 36 blocks of soil) for each experiment. Natural Denitrification Rate Measurements Natural denitrification rate estimates of the intact soil cores were begun immediately upon returning to the laboratory using an C2H, block technique. The gas pressure in each core was first brought to atmospheric levels by venting with a needle. Then an appropriate volume of C2H2 was added to each sized core to achieve a final C2H2 partial pressure of approximately 10 kPa. All C2H2 was generated by reacting CaC2 with distilled water immediately prior to use. The pressure increase from injecting the known volumes of C2H2 was recorded for each core using a pressure transducer. The pressure values were used to calculate the total gas-filled volume in each of the samples as described by Parkin et al. (1984). Gas in the soil cores was then mixed to distribute C2H2 throughout the core by alternately drawing and releasing a vacuum on the samples using a 60-mL syringe. This mixing procedure, in addition to the loose fit of the intact soil cores in the tubes, facilitated C2H2 distribution into, and N2O distribution out of, the soil pores. Following the gas mixing procedure, the gas overpressure in the soil cores resulting from the initial C2H2 injection was vented. Cores were incubated at 24 to 26 °C and gas samples withdrawn 3, 6, and 18 h following the C2H2 additions. Gas samples were obtained by adding 5 mL of air to each core, mixing the gas in the cores using the above procedure, then removing a 5-mL gas sample. The 5-mL gas samples were stored in 3-mL evacuated vials for later N2O analyses.