Cotton Water Productivity and Growth Parameters in the Humid Southeast: Experimentation and Modeling

A global challenge for the coming decades will be increasing food and fiber production with less water. This can be partially achieved by increasing water productivity (WP) or crop water use efficiency (WUE), i.e., yield or biomass produced per unit water used. Cotton (Gossypium hirsutum L.) is a major rainfed and irrigated crop in the U.S. and worldwide. While there is abundant information on cotton water use and yield relationships in drier regions, relevant field data and modeling in humid regions is not well developed. Robust crop models help complement field experimentation and predict the impact of irrigation management on production. The Food and Agriculture Organization (FAO) of the United Nations has developed a crop model (AquaCrop) that has been parameterized for a number of crops including cotton, but not in a humid region. The objective of this study was twofold: (1) to quantify cotton water use and in-season biomass and growth parameters, and (2) to parameterize and validate AquaCrop to simulate cotton water use and productivity in the humid Southeast U.S., where water conservation and sustainability are emerging issues. Irrigation experiments were conducted from 2009 to 2011 under field conditions as well as under a newly installed automated rainout shelter at a Clemson University research site near Blackville, South Carolina. Extensive season-long soil, crop, and climate data were collected under irrigation regimes ranging from rainfed (i.e., no irrigation) to fully irrigated (i.e., meeting 100% of crop water needs). Season-long data collected in 2009 and 2011 under different irrigation regimes were used to parameterize the model. Cotton seasonal water use ranged from 320 mm at 33% irrigation to 718 mm at full irrigation, corresponding to average seed cotton yields of 2134 to 3682 kg ha(-1). Water use efficiency in the three years ranged from 0.48 to 0.71 kg of seed cotton per m(3) of water applied (irrigation and rainfall). Normalized water productivity for the local climate was nearly the same during the three years, averaging 12.5 g m(-2). Model parameterization was less demanding than expected, requiring adjustments of a few model parameters. Using the parameterized model with independent field data from 2010, the model simulated canopy cover, soil water content, aboveground dry biomass, and yield values that were highly correlated with measured values, with coefficients of determination (R-2) greater than 0.66. Seasonal crop water use was predicted within 5% for most treatments but underpredicted by as much as 18% for the 100% treatment at the shelter and for all irrigation treatments in 2010. The underprediction of crop water use corresponded to values of deep percolation simulated by AquaCrop. Field measurements were inadequate to verify deep percolation, but such measurements are suggested for future model investigations. Results of this study suggest that a properly parameterized AquaCrop model provides a useful tool to study yield response to water stress and evaluate the effects of drought and alternate management practices on productivity.