Análisis de riesgo asociado a la exposición crónica a plaguicidas, con el uso de Sistemas de Información Geográfica y percepción remota

Abstract

In order to assess the risk of chronic exposure to pesticide drift in populations surrounded by croplands, five compounds (chlorpyrifos ethyl, cypermethrin, atrazine, glyphosate, and 2-4-D) used in the production of corn and wheat in an irrigation area of the Bajío de Guanajuato were evaluated during the 2019-2020 agricultural cycle. A multi-criteria analysis of the variables involved in the evaluation of risk components was implemented, that is, vulnerability (intrinsic and extrinsic) and threat, establishing specific weights for each variable through the Analytical Hierarchy Process (AHP) and making use of tools such as Geographic Information Systems (GIS) and remote sensing, to facilitate the calculation and integration of the information. Data on the total population and vulnerable groups per block, as well as information on the topography, corresponding to 17 localities in the study area, was obtained from the Instituto Nacional de Estadística, Geografía e Informática (INEGI) 2020 census to determine the intrinsic vulnerability variables. In the case of extrinsic vulnerability variables, wind speed and direction data were analyzed for a period of 9 years (2013 to 2021), obtained from the network of meteorological stations of the Fundación Guanajuato Produce A.C. The propensity of the soil to erode and the geological barriers were also determined with information from INEGI and the vegetable barriers with field observations. The threat was determined by integrating the potential exposure of urban areas, and the severity of each pesticide. To establish the potential exposure, the area of crops surrounding the population points was first obtained, for which two land cover classification of satellite images were carried out for the spring-summer (SS) and autumn-winter (AW) agricultural seasons. These classifications consisted of the identification of the crops present in the study area through the training and validation of the Random Forest supervised classification algorithm, using as input the survey of 107 and 323 plots for the SS and AW seasons, respectively, 9 Sentinel-2 images from May to November 2019 (SS) and 11 from December 2019 to May 2020 (AW), and the normalized difference vegetation and water indices (NDVI and NDWI) derived from the bands of Sentinel-2 images. The results showed overall accuracies of 79.1% for the SS classification and 87.9% for the AW, which are acceptable compared to similar exercises reported by other authors. From the maps obtained, urban, corn and wheat areas were isolated, and buffers were established 500 m away from each urban pixel to calculate the area of crops within them. This information, together with the dose of the pesticide, the commercial formulation, the number of applications per season, the proportion of producers that use them, and the estimated proportion of product loss due to drift, served to determine the potential exposure of each urban pixel to each of the evaluated compounds. It should be noted that in the case of the proportion of loss due to drift, three scenarios were considered: minimum loss (14%), average (46%), and maximum (78%), according to values reported in the literature. Regarding the severity of pesticides, the environmental risk index (ERI) was used, which allowed a comparison between the physicochemical characteristics and the toxicological profile of the active ingredients evaluated, obtaining the category "very low" for glyphosate and cypermethrin, "low" for 2-4-D and atrazine, and "intermediate" for chlorpyrifos ethyl. This information was used, in conjunction with potential exposure, to determine the threat to urban areas. Once all vulnerability variables and threat components were determined, they were normalized within a 5-level scale (very high, high, intermediate, low and very low). They were weighted and integrated through aggregation levels, according to the AHP methodology, to obtain the corresponding risk assessment by drift scenario. The outcome showed that 11.8% of the urban blocks are "without risk", as they correspond to areas where there is no population, and therefore no vulnerability because they do not have croplands within 500 m, or both characteristics are present. It was also observed that the "low" risk category predominates in most localities (more than 70% of the blocks) and only a fraction (2.1% in the maximum drift scenario) is at "intermediate" risk. However, these results should not be underrated since 9% of the total population of the study area is in this last category, of which 48% corresponds to vulnerable population, that is, 722 children and 719 women of reproductive age. The localities with blocks within the “intermediate” risk category are San Nicolas de los Agustinos, San Pedro de los Naranjos, Maravatío del Encinal and Cupareo. These results show that the integration of indirect risk assessment tools, such as geographic information systems, remote sensing and multicriteria analysis, can be very useful to determine the urban areas most exposed to pesticide drift.