RESUMO
Forage production is primarily limited by weather conditions under dryland production systems in Brazilian semi-arid regions, therefore sowing at the appropriate time is critical. The objectives of this study were to evaluate the CSM-CERES-Pearl Millet model from the DSSAT software suite for its ability to simulate growth, development, and forage accumulation of pearl millet [Pennisetum glaucum (L.) R.] at three Brazilian semi-arid locations, and to use the model to study the impact of different sowing dates on pearl millet performance for forage. Four pearl millet cultivars were grown during the 2011 rainy season in field experiments conducted at three Brazilian semi-arid locations, under rainfed conditions. The genetic coefficients of the four pearl millet cultivars were calibrated for the model, and the model performance was evaluated with experimental data. The model was run for 14 sowing dates using long-term historical weather data from three locations, to determine the optimum sowing window. Results showed that performance of the model was satisfactory as indicated by accurate simulation of crop phenology and forage accumulation against measured data. The optimum sowing window varied among locations depending on rainfall patterns, although showing the same trend for cultivars within the site. The best sowing windows were from 15 April to 15 May for the Bom Conselho location; 12 April to 02 May for Nossa Senhora da Gloria; and 17 April to 25 May for Sao Bento do Una. The model can be used as a tool to evaluate the effect of sowing date on forage pearl millet performance in Brazilian semi-arid conditions.
RESUMO
A potato crop multimodel assessment was conducted to quantify variation among models and evaluate responses to climate change. Nine modeling groups simulated agronomic and climatic responses at low-input (Chinoli, Bolivia and Gisozi, Burundi)- and high-input (Jyndevad, Denmark and Washington, United States) management sites. Two calibration stages were explored, partial (P1), where experimental dry matter data were not provided, and full (P2). The median model ensemble response outperformed any single model in terms of replicating observed yield across all locations. Uncertainty in simulated yield decreased from 38% to 20% between P1 and P2. Model uncertainty increased with interannual variability, and predictions for all agronomic variables were significantly different from one model to another (P < 0.001). Uncertainty averaged 15% higher for low- vs. high-input sites, with larger differences observed for evapotranspiration (ET), nitrogen uptake, and water use efficiency as compared to dry matter. A minimum of five partial, or three full, calibrated models was required for an ensemble approach to keep variability below that of common field variation. Model variation was not influenced by change in carbon dioxide (C), but increased as much as 41% and 23% for yield and ET, respectively, as temperature (T) or rainfall (W) moved away from historical levels. Increases in T accounted for the highest amount of uncertainty, suggesting that methods and parameters for T sensitivity represent a considerable unknown among models. Using median model ensemble values, yield increased on average 6% per 100-ppm C, declined 4.6% per °C, and declined 2% for every 10% decrease in rainfall (for nonirrigated sites). Differences in predictions due to model representation of light utilization were significant (P < 0.01). These are the first reported results quantifying uncertainty for tuber/root crops and suggest modeling assessments of climate change impact on potato may be improved using an ensemble approach.