Bayesian approach for evaluating ammonia volatilization nitrogen losses in fertilizers applied to coffee plants

Abstract

Ammonia loss through volatilization is an important reason for the lower effectiveness of nitrogen fertilizers in coffee plants. The Bayesian approach uses informative prior distributions, which help improve the precision and accuracy of inferences, leading to more robust parameter estimates. In this study, we compared the performance of different nitrogen sources applied to coffee plants in terms of nitrogen loss due to ammonia volatilization, using the nonlinear von Bertalanffy model with Bayesian inference. The stabilized fertilizers used were prilled urea (45% N), urea treated with copper and boron (44% N, 0.4% B, and 0.15% Cu), and urea treated with NBPT (45% N). The controlled-release fertilizer used was urea combined with anionic polymer (41% N). The controlled-release fertilizer used was urea combined with anionic polymer. Among the sources of nitrogen, urea coated with polymer resulted in the most significant nitrogen loss, whereas urea treated with NBPT resulted in the lowest loss of nitrogen. Compared to the other fertilizers used, urea treated with NBPT resulted in the lowest nitrogen loss through volatilization, with less than 50% of the nitrogen lost relative to urea with anionic polymers. The Bayesian methodology used provided accurate estimates and enabled a direct comparison between the fertilizers based on the marginal distribution of the von Bertalanffy model parameters. © 2025, Federal University of Lavras. All rights reserved.