Modeling nitrous oxide mitigation potential of enhanced efficiency nitrogen fertilizers from agricultural systems

Highlights

• We developed an EENF model within DayCent to investigate N₂O emission reduction potentials compared to conventional fertilizers.

• Bayesian method was used to calibrate model parameters and quantify modeling uncertainty.

• DayCent produced similar results for the effect of EENFs on N₂O emission reductions as the experimental dataset.

• Freeze-thaw dynamics appear to influence N₂O reductions associated with EENFs, suggesting meta-analyses may have overestimated its benefit.

Abstract

Agriculture soils are responsible for a large proportion of global nitrous oxide (N₂O) emissions—a potent greenhouse gas and ozone depleting substance. Enhanced-efficiency nitrogen (N) fertilizers (EENFs) can reduce N₂O emission from N-fertilized soils, but their effect varies considerably due to a combination of factors, including climatic conditions, edaphic characteristics and management practices. In this study, we further developed the DayCent ecosystem model to simulate two EENFs: controlled-release N fertilizers (CRNFs) and nitrification inhibitors (NIs) and evaluated their N₂O mitigation potentials. We implemented a Bayesian calibration method using the sampling importance resampling (SIR) algorithm to derive a joint posterior distribution of model parameters that was informed by N₂O flux measurements from corn production systems a network of experimental sites within the GRACEnet program. The joint posterior distribution can be applied to estimate predictions of N₂O reduction factors when EENFs are adopted in place of conventional urea-based N fertilizer. The resulting median reduction factors were − 11.9% for CRNFs (ranging from −51.7% and 0.58%) and − 26.7% for NIs (ranging from −61.8% to 3.1%), which is comparable to the measured reduction factors in the dataset. By incorporating EENFs, the DayCent ecosystem model is able to simulate a broader suite of options to identify best management practices for reducing N₂O emissions.

Introduction

Nitrogen (N) fertilizer applied to agriculture soils accounts for a majority of anthropogenic nitrous oxide (N₂O) emissions (Bouwman et al., 2002; Mosier and Kroeze, 2000; Reay et al., 2012) and agriculture is responsible for about 70% of global anthropogenic sources (Tian et al., 2020). Production of N₂O in soil is primarily from microbial processes of nitrification and denitrification (Firestone and Davidson, 1989), although other processes may also contribute to emissions (Butterbach-Bahl et al., 2013). Agricultural soil N₂O emissions are driven by N management practices, particularly the addition of inorganic fertilizers and manure (Mosier et al., 1998). Nitrous oxide is an important greenhouse gas (GHG) with approximately 298 times (100 year time horizon) the global warming potential of carbon dioxide (CO₂) on a mass basis while also contributing to the depletion of stratospheric ozone (Crutzen and Ehhalt, 1977; Ravishankara et al., 2009).

Enhanced-efficiency N fertilizers (EENFs) have shown a potential for mitigating N₂O emissions from N-fertilized agricultural soils (Akiyama et al., 2010; Eagle et al., 2017; Thapa et al., 2016), and have emerged as an important management option for mitigating N losses from agroecosystems (Halvorson et al., 2014; Sha et al., 2020; Trenkel, 2010). We studied two categories of EENFs that can reduce N₂O emissions: (1) controlled-released N fertilizers (CRNFs) and (2) nitrification inhibitors (NIs). Best management practices can incorporate EENFs based on the 4R paradigm for nutrient stewardship (right source at right rate, right placement and right timing of fertilizer application). The rate, pattern, and duration of N release from CRNFs vary depending on the coating materials and thickness, as well as the field conditions (Halvorson et al., 2014; Shaviv, 2001; Timilsena et al., 2015; Trenkel, 2010). CRNFs gradually release N fertilizers into the soil, maintain low mineral N concentration, and extend the availability for plant uptake, ideally with a release pattern in synchrony with the crop's N requirements that minimizes environmental losses (Naz and Sulaiman, 2016; Shaviv and Mikkelsen, 1993; Trenkel, 2010). NIs stabilize N fertilizer in the form of NH₄⁺ in the soil (Trenkel, 2010). Specifically, NIs inhibit the biological process of nitrification and delay the transformation of NH₄⁺ to NO₃⁻ for a certain period of time (four to ten weeks). Maintaining N in the form of NH₄⁺ prevents gaseous loss from both nitrification and denitrification as well as leaching of NO₃⁻ below the rooting zone to the groundwater.

Meta-analyses have been performed to quantify the effect of EENFs on N₂O emissions compared to conventional fertilizers (Akiyama et al., 2010; Eagle et al., 2017; Han et al., 2017; Thapa et al., 2016; Wolt, 2004; Zhang et al., 2019). Significant reduction in N₂O emissions have been reported in these studies for CRNFs and NIs (Eagle et al., 2017; Thapa et al., 2016; Wolt, 2004; Zhang et al., 2019). However, the N₂O reduction from EENFs varies due to a combination of factors including type and rate of N applied, soil properties, climatic factors, and management practices. Despite meta-analyses synthesizing results from many field studies reporting N₂O reductions, the measurements are not spatially continuous resulting in data gaps for some region, making it difficult to derive empirical reduction factors from EENFs for all farms that may adopt these types of fertilizers.

In contrast to an empirical method to estimate reduction in N₂O from EENFs, we developed a process-based mechanistic approach within the DayCent ecosystem model for EENFs. The dynamic N release from CRNFs and the effectiveness of NIs on inhibiting nitrification were modeled as influenced by environmental conditions. There are several advantages of a process-based biogeochemical model (DayCent) relative to empirical methods using meta-analysis. DayCent accounts for N₂O emissions during both nitrification and denitrification in a daily time step and accounts for factors influencing emissions patterns (N inputs, climate, soil, plant growth) at a finer time scale than empirical methods. Furthermore, DayCent predictions of N₂O emissions have been validated and agreed reasonably well with field measurements (Del Grosso et al., 2005, Del Grosso et al., 2010). Previously, Del Grosso et al. (2009) implemented an algorithm to represent the impact of NIs in DayCent with a simple approach defined by two parameters: reduction in nitrification and duration of the effect. Similar approaches have been adopted for the DNDC model in a modified version developed specifically for conditions in New Zealand (Giltrap et al., 2010). The model used a simplified exponential function of time to represent the degradation of NIs that did not account for impacts of soil properties. Further development of DNDC has been proposed by Li et al. (2020) for modeling the effect of NIs by incorporating soil properties, ratio of fertilizer to NIs, and soil parameters (i.e. temperature, moisture, and pH). To our knowledge, more advanced approaches have not been published for modeling CRNFs in process-based models.

The objective of this study is to develop a dynamic modeling approach for CRNFs and NIs as influenced by soil properties and other important drivers, including weather patterns and irrigation management. We focused on the N₂O emissions from agricultural soils and reduction potentials of EENFs compared to conventional fertilizers. Furthermore, we implemented a Bayesian framework developed by Gurung et al. (2020) to calibrate the model parameters and evaluate results using field data from the Greenhouse gas Reduction through Agricultural Carbon Enhancement network of experimental sites in the United States (GRACEnet).