@article{ghersa_agroptim_2024,

title = {AgrOptim: A novel multi-objective simulation optimization framework for extensive cropping systems},

author = {Felipe Ghersa and Lucas Figarola and Rodrigo Castro and Diego Ferraro},

url = {https://www.sciencedirect.com/science/article/pii/S0168169924005106},

doi = {10.1016/j.compag.2024.109119},

issn = {0168-1699},

year  = {2024},

date = {2024-01-01},

urldate = {2025-07-01},

journal = {Computers and Electronics in Agriculture},

volume = {224},

pages = {109119},

abstract = {Cropping systems should be designed to be more productive and have a smaller environmental footprint to sustainably meet the growing demand for food, fiber, and fuel. However, this requires the evaluation and ranking of many cropping system designs based on their economic and biophysical performance, which are often in conflict. Although field experiments and simple crop simulation models have been used for this purpose, studies have generally considered a limited number of agronomic decision combinations or indicators that partially capture ecosystem functions. Coupling evolutionary algorithms with process-based crop simulation models provides a less resource-intensive alternative and can incorporate many indicators to (1) quantify the trade-offs between biophysical and economic performance, and (2) identify the set of agronomic decision combinations that minimize these trade-offs. The objective of this paper was to present AgrOptim, a novel cropping system simulation optimization framework that uses genetic algorithms to optimize a holistic set of biophysical and economic performance indicators through different combinations of agronomic decision variables (i.e., crop sequence, crop structure, pesticide dose, and fertilizer dose). Indicators were derived from a process-based crop simulation model, an ecotoxicological risk simulation model, and emergy synthesis. The framework was implemented in Argentina to (1) characterize the relationship between economic and biophysical indicators and (2) evaluate the current state and potential improvements of three frequently used cropping system designs. A multi-objective optimization experiment was designed to simultaneously optimize 30-year cropping sequences based on one economic objective (return on investment) and four biophysical objectives (crop residue carbon inputs, precipitation use efficiency, nonrenewable to renewable energy ratio, and pesticide ecotoxicity). Results showed that trade-offs exist between economic and all biophysical objectives, albeit with varying intensities. Additionally, the decision variables that provided improved performance in terms of carbon residues, precipitation use efficiency, and ecotoxicological risk also presented higher levels of nonrenewable energy use. For the three frequently used cropping system designs, the decision variables that improved the performance of each indicator were identified. These findings highlight the challenges faced by agricultural producers considering the trade-offs between their economic and biophysical objectives. Additionally, they reveal potential model-aided improvements that can be obtained using crop simulation models and optimization algorithms to redesign cropping systems.},

keywords = {Crop sequence, Crop simulation, Genetic algorithm, Optimization, Sustainability gap},

pubstate = {published},

tppubtype = {article}

}