The metrica R package (Correndo et al., 2022) is an open-source software designed to facilitate the quantitative and visual assessment of prediction performance of point-forecast simulation models for continuous (regression) and categorical variables (classification). The package ensembles a series of 80+ functions that account for multiple aspects of the agreement between predicted and observed values. Without the need of advanced skills on programming, metrica enables users to automate the estimation of multiple prediction performance metrics including goodness of fit, error metrics, error decomposition, model efficiency, indices of agreement, and to produce stylish data visualization outputs. This article introduces metrica, an R package developed with the main objective of contributing to transparent and reproducible evaluation of point-forecast models performance.

This article aims to discuss the arcsine–log calibration curve (ALCC) method designed for the Better Fertiliser Decisions for Cropping Systems (BFDC) to calibrate relationships between relative yield (RY) and soil test value (STV). Its main advantage lies in estimating confidence limits of the critical value (CSTV). Nevertheless, intervals for 95% confidence level are often too wide, and authors suggest a reduction in the confidence level to 70% in order to achieve narrower estimates. Still, this method can be further improved by modifying specific procedures. For this purpose, several datasets belonging to the BFDC were used. For any confidence level, estimates with the modified ALCC procedures were always more accurate than the original ALCC. The overestimation of confidence limits with the original ALCC was inversely related to the correlation coefficient of the dataset, which might allow a relatively simple and reliable correction of previous estimates. In addition, because the method is based on the correlation between STV and RY, the importance to test it for significance is emphasised in order to support the hypothesis of a relationship. Then, the modified ALCC approach could also allow a more reliable comparison of datasets by slopes of the bivariate linear relationship between transformed variables.

Continuous potassium (K) removal without replenishment is progressively mining Argentinean soils. Our goals were to evaluate the sensitivity of soil-K to K budgets, quantify soil-K changes over time along the soil profile, and identify soil variables that regulate soil-K depletion. Four on-farm trials under two crop rotations including maize, wheat and soybean were evaluated. Three treatments were compared: (1) control (no fertilizer applied); (2) application of nitrogen, phosphorus, and sulfur fertilizers -NPS-; and (3) pristine condition. After nine years, crops removed from 258 to 556 kg K ha⁻¹. Only two sites showed a decline in the exchangeable-K levels at 0–20 cm but unrelated to K budget. Topsoil exchangeable-K levels under agriculture resulted 48% lower than their pristine conditions, although still above response levels. Both soil exchangeable-K and slowly-exchangeable K vertical distribution patterns (0–100 cm) displayed substantial depletion relative to pristine conditions, mainly concentrated at subsoil (20–100 cm), with 55–83% for exchangeable-K, and 74–95% for slowly-exchangeable-K. Higher pristine levels of exchangeable-K and slowly-exchangeable-K and lower clay and silt contents resulted in higher soil-K depletion. Soil K management guidelines should consider both topsoil and subsoil nutrient status and variables related to soil K buffer capacity.