Nonlinear effects of water stress on peanut photosynthesis at crop and leaf scales

Crop models are being increasingly used in agricultural and environmental studies of marginal environments where water availability limits crop growth. The PNUTGRO model (a precursor of CROPGRO and CSM) has systematically underestimated rainfed pod yield and aboveground biomass, while accurately predicting the same variables under irrigation, in the frequently drought-stricken Argentine peanut-growing region. This happened although the model was previously optimized to properly simulate atmospheric water demand and soil/plant water supply, suggesting that the mechanisms of peanut drought tolerance are not adequately expressed in the model. Crop models such as PNUTGRO typically describe water stress using a linear function to penalize carbon assimilation when water supply falls below a certain limit, assuming that stomatal control affects transpiration and photosynthesis proportionally. We analyzed the feasibility of a linear transpiration—protosynthesis relationship at the leaf and crop scales. At the leaf scale we used two leaf gas exchange models (a conductance model and the anatomy-based 2DLEAF). At the crop scale we replaced the linear equation linking transpiration supply/demand and photosynthesis in PNUTGRO with an equation of the form PG/PGMAX = 1 − (1 − SWFAC)∧WSFEXP, where SWFAC is PNUTGRO’s transpiration demand/supply ratio and WSFEXP is an empirical nonlinearity constant that was determined by simultaneously fitting simulated and observed biomass and plant-extractable soil water (PESW) content of several field experiments. An independent data set was used for validation. Both leaf models showed that linearity is infeasible, primarily due to the greater contribution of stomatal aperture to the total pathway resistance of water vapor versus CO2. At the crop-level, simulations of biomass, PESW, and pod yield in rainfed experiments improved the most when we used the nonlinear function with WSFEXP = 2.5. Mean final biomass error improved from −20 to −6.5%; mean final pod yield error went from 20 to 0.07%; mean PESW error went from 5 to −0.2%. Our results support the idea that water use efficiency (WUE) is a nonlinear function that increases under conditions of water stress. This agrees with experimental evidence from the literature and with theory integrating quasi-steady-state stomatal ∗ Corresponding author. Tel.: +1-352-392-1864. E-mail address: [email protected] (R.A. Ferreyra). 0304-3800/$ – see front matter © 2003 Elsevier B.V. All rights reserved. doi:10.1016/S0304-3800(03)00195-9 58 R.A. Ferreyra et al. / Ecological Modelling 168 (2003) 57–76 closure due to low soil water availability, short-timescale midday adaptive behavior, and peanut-specific drought-avoidance mechanisms. © 2003 Elsevier B.V. All rights reserved.