Model approaches to estimate spatial distribution of bee species richness and soybean production in the Brazilian Cerrado during 2000 to 2015.

ABSTRACT

Agricultural expansion as a main human activity has affected pollinator's habitat, causing spatial distribution changes. Meanwhile, pollinators still provide pollination service to improve crop production. However, their spatial response is unclear because of environmental changes. This study sought to estimate spatial distribution of crop production and pollinator's richness, which can provide insights as to how they interact with the environment. We acquired environmental variables from remote sensing images and used a stacked species distribution model to predict selected bee species richness and a crop simulation model to simulate and calculate soybean production at a regional scale in the Cerrado for the period 2000-2015. Then, we analyzed their potential relationship. The results showed that higher selected bee species richness and higher soybean production occurred in the southern Cerrado. From 2000/08 to 2008/15 period, the selected bee species richness significantly decreased in the western part of the state of Bahia, the state of Goiás, and the northern region of the state of Minas Gerais; while soybean production increased in the states of Mato Grosso, Goiás, Bahia, and Tocantins. Correlation results of selected bee species richness and soybean production showed that they do not follow a linear relationship during the study period. Our findings indicate that the modeling method we proposed is robust to estimate spatial distribution of bee species richness and soybean production in the Cerrado at the regional scale and that the environment has a stronger influence on selected bee species richness than on soybean production. Moreover, climate effects and agricultural expansion are the main factors that affect their spatial distribution and interaction. Finally, our methodology provides a novel spatial perspective to analyze the relationship between pollinator and agricultural expansion corresponding with the environment, but future work is needed to collect a more comprehensive data set to improve model results.