Global environmental impacts driven by agricultural activities, e.g. water and air pollution, climate change, or soil acidification, are threatening human beings’ livelihood. Organic farming, which primarily depends on legumes and livestock manure for nitrogen inputs, has the potential to alleviate these environment impacts. Many studies found that organic farming can improve the quality of soil organic matter, accumulate more carbon in the soil than conventional agriculture (Gattinger et al., 2012), and reduce nutrient leaching into ground and surface water (Kirchmann and Bergstrom, 2001). However, some researchers have also criticized the lower crop yields from organic farming compared to that from conventional farming. A meta-analysis based on the literature review was published in Nature, showing that organic yields were 5-34% lower than conventional yields depending on crops, site conditions and management practices (Seufert et al., 2012). So far, most studies comparing environmental impacts and yields between organic and conventional farming systems are based on data collected from a few and rather small experimental sites. However, crop growth models have not yet been applied to quantitatively compare the crop growth processes and nutrient dynamics between organic and conventional farming systems on a large scale. Therefore, in this WFSC Ambassador project, I have tried to explore the possibility of applying a large-scale agronomic model to simulate the trade-offs between yields and environmental impacts.
First Step: Developing a Python-based EPIC (PEPIC) model
The GIS-based EPIC (GEPIC) was initially developed in 2007 at Eawag by Liu et al. and extended by Folberth et al. (2012). It was programed by using the Visual Basic Application (VBA) under the ArcGIS 9.3 platform. At the initial stage of this project, I planned to update the GEPIC model for organic farming simulation. However, I changed to develop a new model, the Python-based EPIC (PEPIC), by considering the three main limitations of the GEPIC model: 1) GEPIC requires the ArcGIS license for its application; 2) the VBA is no longer supported by ArcGIS 10.1 and its subsequent versions; and 3) the parameterization of GEPIC is quite simple. Therefore, the PEPIC model was proposed for this purpose. I had extensive discussions about how to design the PEPIC with the Agro-Environmental Systems (AES) group at IIASA, especially with Dr. Christian Folberth. After hard working, I finally completed the PEPIC model and successfully applied it to assess the global crop-water relations of maize. I am happy to tell that the results of this part have been published in the peer reviewed journal Agricultural and Forest Meteorology (Liu et al., 2016). The main features of the PEPIC model include: 1) using a world-wide and free programing language “Python” to develop the model interface and implementation framework; 2) adjusting easily some important parameters of the EPIC model; 3) calibrating and validating the model performance by using the Sequential Uncertainty Fitting (SUFI-2) method and quantifying model uncertainties by using the Latin Hypercube Sampling (LHS) method; 4) providing more crop management practice options e.g. conventional farming and organic farming.
Second Step: Adjustment of PEPIC model for nitrogen loss assessment
Many large-scale crop models were designed for assessing crop yields, crop water consumption and agricultural climate change impacts. However, application of the crop models for the assessment of environmental externalities related to pollutant loads, e.g. nitrogen and phosphorus losses, is rather limited on a large scale. Also I had discussed with the AES group at IIASA, Dr. Ligia B. Azevedo, Dr. Juraj Balkovič, Dr. Rastislav Skalský, and Dr. Christian Folberth, about the extension of the PEPIC model for nutrient loss assessment on a global scale. The most important points for such assessment include: 1) defining the proper parameterization related to nitrogen, phosphorus, and carbon dynamics; 2) finding the most updated crop-specific fertilizer data; 3) developing the suitable crop management strategies, especially nitrogen and phosphorus fertilization schedules; 4) applying uncertainty method for quantifying the scope of model uncertainties. Based on the results in this stage, I wrote a manuscript with the title “Global assessment of nitrogen losses from major crop cultivations and trade-offs with yields”. The manuscript will be submitted for review quite soon. In this paper, we found a large fraction of nitrogen fertilizer was lost into the environment. Mitigation measures are highly needed, particularly in China and India. Meanwhile, we found the nitrogen losses can be significantly reduced just through redistributing the spatial patterns of nitrogen inputs with increase of yields.
Third Step: Simulating the impacts of organic farming
On the basis of the first and second step I could set up the model for organic farming simulation by considering the manure fertilization and legume crops for rotation. I focused on Europe and considered maize and soybean as crop rotation. Manure inputs of maize were obtained from West et al. (2014). In this part, we found that maize yields will decrease 30-60% in the organic farming system compared to the conventional farming in the high nitrogen input countries (> 150 kg N ha-1), while nitrogen losses will also significantly decrease. On the other hand, in the low nitrogen input countries (< 120 kg N ha-1), maize yields will slightly decrease (< 30%), however, nitrogen losses will increase. We found no effects on soybean yields, while nitrogen losses will significantly decrease for all countries. Besides, we also found that crop residue will not significantly affect the yields, while its effects on soil organic matter are significant. However, total nitrogen losses also increase with more crop residues being left on cropland.
The concept of organic farming is of course beyond the manure fertilization and legume crop for rotation. This study is just a fist attempt to explore the possibility of crop model for organic farming simulation. I propose that more details of organic farming in crop model setting up should be considered in future studies.
In this Ambassador project, I went through three steps to explore the applicability of crop model for large-scale organic farming simulation: develop a global PEPIC model, adjust it for nitrogen loss assessment, and apply the PEPIC model for organic farming modeling. We see the high possibility to use crop models for investigating the trade-offs of yields and environmental impacts under organic farming. However, further studies are needed to consider more details of organic farming.
- Folberth C et al. (2012) Regionalization of a large-scale crop growth model for sub-Saharan Africa: Model setup, evaluation, and estimation of maize yields.
- Gattinger A et al. (2012) Enhanced top soil carbon stocks under organic farming.
- Kirchmann H and Bergstrom L (2001) Do organic farming practices reduce nitrate leaching?
- Liu J et al. (2007) GEPIC – modelling wheat yield and crop water productivity with high resolution on a global scale.
- Liu W et al. (2016) Global investigation of impacts of PET methods on simulating crop-water relations for maize.
- Seufert V et al. (2012) Comparing the yields of organic and conventional agriculture.
- West PC et al. (2014) Leverage points for improving global food security and the environment.
ABOUT THE AUTHOR
Wenfeng Liu is a Ph. candicate at both Swiss Federal Institute of Aquatic Science and Technology (Eawag) and ETH Zurich. His PhD project focuses on investigating the water-food-environment-trade nexus in the context of agricultural intensification.
“The WFSC Ambassador grant was really helpful to me. It encourages young researchers to travel around the world to have further discussion and education. With its help, I could visit EPIC experts at IIASA and successfully go through the whole story.”