A tripartite study on Bayesian estimation of photosynthetically active radiation, impacts of future climate, and adaptation strategies on crop production: a spatial model framework for the Eastern Kansas River Basin

Abstract

Climate change impacts and adaptation strategies on rainfed and irrigated cropping systems in the Eastern Kansas River Basin (EKSRB; a historically wetter portion of the Great Plains) have not been well-reported. While there are regional-to-global scale climate impact analyses, they rarely provide insights into management and adaptation strategies needed to tackle climate risk impacts on crop production in the EKSRB. As climate warms with its negative impacts, producers in the region continually look out for the latest management and adaptation options to optimize crop production, while saving resources. Traditional methods for quantifying climate change impacts and studies done in the western part of the Great Plains have relied on point-scale process-based crop modeling analysis to generalize regional insights. However, a comprehensive outlook to provide specific solutions and insights into climate change impacts and adaptation options will require a spatially explicit approach, treating each field in the region as an independent landscape capable of providing location-specific insight on crop productivity and water use management. More so, improvement in process-based crop model inputs such as photosynthetically active radiation (PAR) is essential to enhance model accuracy, while also documenting new estimation methods. The ultimate goal of this dissertation is to develop climate change adaptation strategies for rainfed and irrigated maize production in EKSRB using a combination of county-level and fine-scale spatial protocols. We quantified location-specific impacts under heterogeneous responses to environmental variability and different water management techniques using process-based crop models. In addition, this dissertation also developed a Bayesian regression framework to improve the estimation of PAR, which is required to enhance process-based crop model accuracy. The specific research objectives are: (1) assess the impacts of future climate conditions and root proliferation adaptation strategy on rainfed maize production; (2) develop a fine-scale spatial protocol for quantifying climate change impacts on maize productivity under different irrigation management strategies; (3) improve irrigated maize productivity based on genotype-specific and agronomic adaptation strategies under climate change; and (4) develop a Bayesian regression framework to improve estimation of PAR for crop yield prediction in EKSRB. A county-level CERES-Maize crop model was calibrated and validated in the EKSRB to study the impacts of late century (2071-2100) future climate scenarios under the SSP245 emission pathway using the CMIP6 climate models. A novel root proliferation adaptation strategy was tested to assess the impact of these climate change signals on regional rainfed maize yield and rainfall productivity relative to baseline climate conditions (1985–2014). Results showed that baseline yield values ranged from 6,522 to 12,849 kgha-1, with a regional average value of 9,270 kgha-1. Projections for the late century scenario indicate a substantial decline in maize yield (36% to 50%) and rainfall productivity (25% to 42%). Introducing a hypothetical maize cultivar by employing root proliferation as an adaptation strategy resulted in a 27% increase in regional maize yield, and a 28% increase in rainfall productivity compared to the reference cultivar without adaptation. These findings offer valuable insights for the US Great Plains maize growers and breeders, guiding strategic decisions to adapt rainfed maize production to the region's impending challenges posed by climate change. To assess the location-specific response of irrigated maize production to future climate conditions, a fine-scale spatial protocol was developed. This protocol was tested using Shawnee County of the EKSRB as a case study for a comprehensive assessment of climate change impacts under different future representative concentration pathways (RCPs) of CO2 emission and irrigation management techniques. Simulation results showed that future irrigated maize yield decline may have occurred due to shortened growing season length, with values ranging from 21% to 38% (early and late century, respectively) under RCP 4.5, and from 22% to 70% (early and late century, respectively) under RCP 8.5. Despite the yield declines, irrigation water use under full allocation increased by 9% to 23%, with a significant reduction in average farmers’ net returns ranging from 53% to 80% (RCPs 4.5 and 8.5, respectively), and a significant decline in irrigation water productivity. Assessing the impacts of deficit irrigation resulted in water savings ranging from 3 to 15% without further diminishing overall maize productivity. Our results indicate that climate-smart irrigation management strategies implemented at a fine-scale spatial resolution could help save water despite negative climate impacts on future maize production in the region. The future of maize production in the region necessitates integrating yield-advancing cultivars and proper agronomic management with improved water management in order to meet the expected grain demand over the next decades. Impacts of in silico genotype-specific adaptation for selected maize genotype-specific parameters, treated as quantitative traits (improve canopy photosynthesis for radiation use efficiency [RUE], light extinction [KCAN], and heat tolerance [HEAT]), and agronomic management (shifting planting window [PD1 & PD2], and non-limiting nutrient allocation [NNL]) on yield and irrigation water use were assessed. Results showed that incorporating individual levels of either genotype-specific or agronomic adaptation strategies resulted in significant yield and water savings improvements relative to no adaptation scenario under future climate conditions. However, an integrated adaptation strategy linked to improved RUE, HEAT, KCAN, NNL, and PD2, resulted in the early 21st century (2025-2049) yield gain of 0.6 to 3% (RCPs 4.5 and 8.5, respectively), with water savings of 10 to 13% (RCPs 4.5 and 8.5, respectively), relative to historical condition (1991-2015). Going forward into the 21st century, marginal yield deviations were observed (especially under RCP 4.5) with further increase in water savings. The overall findings emphasize the transformative potential of all-inclusive integrated adaptation strategies in mitigating the impacts of climate change on irrigated maize production in the Great Plains. Finally, a practical Bayesian regression framework for PAR estimation was developed for the EKSRB. Results revealed that daily PAR values generated from the posterior predictive distribution using the Gibbs sampler algorithm outperformed the PAR estimates from the traditional linear modeling approach. In addition, soybean yield estimates using PAR from the Bayesian framework as inputs, captured the year-to-year variability in observed yield better than using PAR estimates from the traditional linear modeling approach. The results of this study support the use of the Bayesian regression framework for robust estimation of PAR inputs to crop models. This dissertation underscores the utility of climate-smart strategies for the EKSRB maize production by assessing climate impacts, adaptive irrigation, and climate adaptation strategies, and in addition, enhancing crop model accuracy using Bayesian-estimated PAR. In general, it offers location-specific solutions to improve maize productivity and resilience under climate change.