A Generic Pest Submodel for Use in Integrated Assessment Models

Abstract

<abstract> <b><sc>Abstract.</sc></b> The sustainability of agricultural land use has been the subject of integrated assessments for some time. In these assessments, knowledge of various disciplines is combined to assess, e.g., the drawbacks and benefits of new agro-technologies, changes in management systems, and the introduction or abolition of agricultural policies. Crop models are often used in these assessments, and the more details the models can provide, the more factors can be considered. For example, the crop model EPIC can determine fertilizer and irrigation water use based on crop demand. However, pesticides can only be applied manually and have no impact on the crop. We therefore developed a pest submodel providing estimates of pesticide consumption based on pest pressure and the corresponding reactions of crops, thus completing a process loop analogous to the application of fertilizers or irrigation water. We distinguish three pest categories: insects, fungi, and weeds. For insects, a biological temperature response function is used to model the daily feeding rate, from which population growth is derived. Insect mortality is dependent on temperature and nutrient availability. Fungal disease spread depends on coefficients of primary and secondary infection, abundance of spores, and host density. The number of spores increases depending on a humidity and temperature function and decreases according to a decay rate. Crop stress is dependent on the ratio of healthy to infected plant tissue. Weeds are grown in EPIC the same way as crops and compete for the same resources. Herbicide applications are triggered when a specified ratio of weed to crop LAI is reached. Pesticide-induced mortality is modeled as a pesticide-specific dose-response curve. We tested the submodel by simulating five crops with the highest use of pesticides in the U.S. (cotton, maize, potatoes, soybeans, and wheat). Among other findings, the mean simulated application rates approached the reported mean application rates, and the spatial distributions of simulated and reported application rates matched well. We conclude that the new pest submodel is appropriate to estimate pesticide use on larger scales and can be employed to enhance integrated assessments.