Grounding pesticide use intensity and hazards with real-world data: Toward correcting global models in tropical agriculture.

Global pesticide use is increasing, especially in low- and lower-middle-income countries, posing threats to non-target organisms, ecosystems, and people. However, data on pesticide use globally and within nations are often low quality, lacking fine-grained data, including specific active ingredients used, which constrains the accurate calculation of hazards and risk assessments associated with pesticide use practices. The geographic differentiation of these shortcomings suggests that the agricultural patterns in the global South, especially in the tropics, are generally misrepresented. Here, we use Costa Rica, which relies heavily on pesticides and has a formulation industry, as a case study to develop a nuanced country-level analysis of pesticide use and associated hazards. We bring together various sources of data on trade, formulation, and use practices to estimate pesticide use and calculate its associated hazards by combining our estimations with a dataset on pesticide hazards. This way, for the first time, we assessed trends in pesticide use and hazards over time and their associations with different crops using publicly available national data. The yearly average use intensity in cropland was 21.91 ± 0.42 kg/ha and we found temporal and spatial trends in pesticide use, with higher levels in the country's Caribbean region, dominated by export crops (large-scale banana and pineapple monocrops). Our results contrast with national estimates and international datasets, such as the FAOSTAT dataset, and with global model outputs, which show lower estimates. Moreover, our hazard analyses contradict the narrative of hazard improvement through modern a.i.'s. We argue for the importance of data availability, as it enables transparency into trends over time and accountability for the state's efforts to reduce pesticide risks.