Modeling salinity effect on rice growth and grain yield with ORYZA v3 and APSIM-Oryza

Abstract

Development and testing of reliable tools for simulating rice production in salt-affected areas are presented in this paper. New functions were implemented in existing crop models ORYZA v3 and the cropping systems modelling framework APSIM. Field experiments covering two years, two different sites, and three varieties were used to validate both improved models. We used the salt balance module in the systems model APSIM to simulate the observed daily soil salinity with acceptable accuracy (RMSEn <35%), whereas ORYZA v3 used measured soil salinity at a given interval of days as a model input. Both models presented similarly good accuracy in simulating aboveground biomass, leaf area index, and grain yield for IR64 over a gradient of salinity conditions. The model index of agreement ranged from 0.86 to 0.99. Variability of yield under stressed and non-stressed conditions was simulated with a RMSE, of 191 kg ha^-1 and 222 kg ha^-1 _, respectively, for ORYZA v3 and APSIM-Oryza, corresponding to an RMSE_n of 14.8% and 17.3%. These values are within the bounds of experimental error, therefore indicating acceptable model performance. The model test simulating genotypic variability of rice crop responses resulted in similar levels of acceptable model performance with RMSE_n ranging from 11.3 to 39.9% for observed total above ground biomass for IR64 and panicle biomass for IR29, respectively. With the improved models, more reliable tools are now available for use in risk assessment and evaluation of suitable management options for rice production in salt-affected areas. The approach presented may also be applied in improving other non-rice crop models to integrate a response to soil salinity - particularly in process-based models which capture stage-related stress tolerance variability and resource use efficiency.

Keywords: Genotype; Photosynthesis; Soil; Transpiration; Water.

Fig. 1

Diagram representing the ORYZA v3 model and the flow of mass and information between its different modules. The tied boxes represent process rates; the boxes represent state variables, and the circles, the intermediate variables. The dashed lines represent flow of information and the continuous lines represent flow of mass. The black lines and text indicated the processes which representation was modified with the new functions accounting for the salinity effect. LAI, leaf area index; N, nitrogen. (a)–(c) indicated the processes affected by the salinity effect namely soil solute content changing soil osmotic potential component and soil water tension (a), transpiration rate (b) and leaf photosynthesis (c).

Fig. 2

Example of salinity stress factor curves for varieties with different salinity tolerances (Adapted from Radanielson et al., 2017). FSi is the salinity stress factor, where i is the considered process (photosynthesis rate or transpiration rate). Salinity stress factor curves are mainly determined by two parameters, the slope of decrease in the process rate at the inflection point (parameter a) and the critical value of salinity for 50% of loss of the process rate to occur (parameter b). In this example, the tolerant variety has parameters (a) and (b) equivalent to 0.22 and 7.83 respectively. The sensitive variety has parameters (a) and (b) equivalent to 0.22 and 13.69 respectively.

Fig. 3

Measured soil salinity using 5TE sensors (Decagon, USA) and simulated using APSIM-Oryza during the dry season 2012 in experiment 1 (a, b) and during the wet season 2012 in experiment 2 (c, d). Dashed gray lines at 4 dS m⁻¹, dashed black lines at 8 dS m⁻¹, and black solid lines at 12 dS m⁻¹.

Fig. 4

Simulated and observed values for biomass and grain yield under four different soil conditions: a) non-saline, b) 4 dS m⁻¹, c) 8 dS m⁻¹, and d) 12 dS m⁻¹. Black lines represent outputs from APSIM-Oryza simulations; dashed black lines represent outputs from ORYZA v3; square points represent the observed above-ground biomass values from the 2012 wet and dry season experiments (Expt.1& 2) at IRRI, Los Baños, Philippines, and the open circle points represent the observed grain yield. Error bars are standard deviation of the mean.

Fig. 5

Simulated and observed leaf area index under four different soil conditions. a) non-saline, b) 4 dS m⁻¹, c) 8 dS m⁻¹, and d) 12 dS m⁻¹. Black lines represent outputs from APSIM-Oryza simulation, black dashed lines represent outputs from ORYZA v3, square points represent the observed leaf area index (LAI) values from the 2012 dry and wet season experiments (Expt.1& 2) at IRRI, Los Baños, Philippines. Error bars are standard deviation of the mean.

Fig. 6

Simulated above-ground biomass and panicle biomass by the ORYZAv3 model (a, b) and by the APSIM-Oryza model (c, d) versus observed data for three contrasting varieties growing in salt stressed conditions. The observed data presented are average measurements from 3 replications in experiments 3 & 4 for each variety. Simulated data are obtained from the ORYZA v3 model (a, b) and APSIM-Oryza (c, d), considering salinity conditions of the experiments 3 & 4. Lines present the linear relationship between simulated and observed values with intercept at 0.