Work Description: Glyphosate-based herbicides (GBH) are the most widely used pesticides, considered ubiquitous in ecosystems. Among contaminated environments, soil plays a pivotal role in glyphosate dissipation to other natural resources. The presence of GBH in environments is associated with ecological imbalances and risks to human and other animal health. However, the methods to detect glyphosate are expensive and time-consuming, resulting in little or no monitoring in many countries. In this context, mathematical and computational models, which allow inferring outcomes and classifying variables, can be used for detecting environmental contaminants. Therefore, in this project, I propose the application of mathematical models and machine learning algorithms to identify the presence of GBH in soil, based on biological (earthworm) and physical-chemical soil variables. Earthworms were chosen as representatives of soil fauna because they are bioindicators sensitive to pollutants. Specifically, the general objectives of this project are: 1) Verify whether exposure to dosages of agronomic importance of GBH alters biological and physical-chemical properties of the soil environment, and 2) Verify whether and which mathematical and computational models can be used to indicate the presence of glyphosate-based herbicide in the soil using biological variables of earthworms and physical-chemical soil data as input. Among the specific objectives, it is planned to identify the most important variables to achieve general objective 2, using a random forest algorithm.