Understanding spatial heterogeneity of yield, and management-specific CH4 and N2O emissions in rice systems: experimental field studies and modeling

Management practices that optimize yield and resource-use efficiency, and mitigate greenhouse gas emissions (GHG) in rice (Oryza sativa L.) systems are needed to address food security and climate change challenges. The main objectives of this dissertation were (1) to evaluate the potential for various management practices to increase crop productivity and reduce environmental impacts in rice systems, and (2) to empirically evaluate a process-based model for estimating CH4 and N2O emissions in direct-seeded rice systems under varying management, soil environments, and weather conditions. In Chapter 2, the suitability of rice fields for precision management is evaluated by analyzing the stability of yield patterns and their underlying causes. Unlike other studies, no significant effect of cold water temperature on within-field yield variability was observed. However there was evidence that DOC, N, K, and salts were redistributed within fields via mass flow in flood water and accumulated in areas with restricted water movement via evapoconcentration. Soil ECe and/or P were the main drivers of yield variability, which has important nutrient and salinity management implications. Our results demonstrate that while precision management is not useful in rice fields exhibiting high temporal yield variability, in other fields exhibiting stable yield patterns with known causes, precision management may improve profitability and resource-use efficiency. In Chapter 3 the GHG mitigation potential and growth characteristics of several rice cultivars are investigated. Anaerobic microbial decomposition of soil organic matter and plant-derived C substrates leads to the emissions of CH4 and N2O, which is affected by the presence of rice plants. Thus, we hypothesized that cultivars differ in GHG emissions due to cultivar-specific variations in growth characteristics. Multi-year field studies in California and Arkansas showed that cultivars differed in seasonal CH4 emissions by a factor of 2.1 and 1.3, respectively, but were similar in N2O. Although differences in global warming potential (GWP) were observed, there were inconsistencies across sites, indicating the importance of the genotype by environment interaction. Plant growth characteristics were generally not correlated with seasonal CH4 emissions; however, the strongest correlations were observed for both shoot and total plant (root + shoot) biomass at heading (r = 0.60) at one California site, and for grain at maturity (r = -0.95) at one Arkansas site. Interestingly, the lowest emitting cultivar at one of the Arkansas sites had the highest yield, highlighting the potential for breeding high-yielding cultivars with low GWP, but environmental conditions must also be considered. In Chapter 4, the process-based model, DeNitrification-DeComposition (DNDC), is empirically evaluated for estimating cultivar-specific CH 4 and N2O fluxes in California rice systems. The model was parameterized for two cultivars, M206 and Koshihikari, and able to simulate 30% and 78% of the measured variation in yields, respectively. Modeled and observed site-level seasonal CH4 emissions were similar (R 2 = 0.85), but there was poor correspondence between modeled and observed fallow period emissions, and seasonal and fallow period N2O emissions. Management-specific CH4 emissions within sites were highly variable and uncertain, with a range of 0.2-465% relative absolute deviation across sites. Specifically, the model showed (1) over-sensitivity of CH4 emissions to N fertilizer application rates, (2) over-sensitivity of N 2O to field drainage, and (3) under-sensitivity of CH4 emissions to type of seeding system and prior fallow period straw management. In conclusion, large differences in crop productivity, soil organic C, and clay content within uniformly managed rice fields suggests that GHG emissions may also vary substantially within rice fields. These findings have implications for estimating, and potentially reducing, uncertainty of input data for process-based models used to estimate yield impacts, GHG inventories, and mitigation potentials of alternative practices. The effectiveness of GHG mitigation practices often depend on the site-specific conditions where they are implemented, such as soil type and weather. The inconsistencies in cultivar effects on GHG emissions across sites, and the model-data discrepancies pertaining to management practices within sites, indicate that further research is needed to improve model representations of plant growth and microbial decomposition processes across environmental gradients.