Nitrogen management

Exercise objective:

·    Manipulating nitrogen (N) management in the experimental data file.

Suggested reading:

Chapters 7.2 (the part on N management) of the book, ORYZA2000: modelling lowland rice.

Exercise: 

Ex-III.1. Start up the Shell. Open project C:\COURSE\NITROGEN\ORYZAWin. Read CONTROL.DAT and verify its contents:

FILEIT = 'C:\COURSE\NITROGEN\IR72DS2.T92'   ! Experimental data

FILEI1 = 'C:\COURSE\NITROGEN\IR72.D92'      ! Crop data

FILEIR = 'C:\COURSE\NITROGEN\RERUNS.DAT'    ! Rerun file

Read the crop data file IR72.D92 and scroll to Section 10 ‘Nitrogen parameters’ at the bottom. These parameters are used when the crop is run with an N balance (see below). Most of these parameters have been derived from a large number of experiments at different locations, using various N levels and varieties. It is assumed that these parameters are generic for most modern high-yielding varieties and do not need to be calibrated for new situations. A detailed treatment of these parameters is beyond the scope of this exercise; they are fully explained in Chapters 5 and 7.3 of the book “ORYZA2000: modelling lowland rice”.

Ex-III-2. Read the experimental data file IR72DS2.T92 and verify that the nitrogen simulation environment is set to simulate potential production (NITROENV = 'POTENTIAL'). Run ORYZA2000, and check the default simulation values in column Sim1 in Table III.1.

When ORYZA2000 is run with an N balance, the management of N needs to be specified in the experimental data file. There are three data that need to be supplied: the soil N supply capacity, SOILSP (kg N ha-1 d-1); the fertilizer N application rate, FERTIL (kg N ha-1 d-1); and the recovery fraction of the applied fertilizer N, RECNIT (fraction). The recovery fraction is the fraction of applied fertilizer N that the crop takes up, the complementary fraction is lost through processes such as leaching and volatilization, or is accumulated in the soil at the end of the season. In the experimental data file, SOILSP is a single value, whereas FERTIL and RECNIT are tables.

In ORYZA2000, it is assumed that the soil N supply capacity is a constant value throughout the growing season. Fertilizer application, on the other hand, is applied on a daily basis. The table FERTIL is linearly interpolated at each day of simulation and, therefore, each day of fertilizer application should be preceded and succeeded by a day with zero fertilizer application. The days of fertilizer application in FERTIL are defined as the number of days after emergence in the seedbed (or in the main field if the crop is direct-seeded). There should always be a 0 entry for day 1 and for the last day of simulation (can be safely set to day 366), to avoid errors in the linear interpolation. Fertilizer-N recovery is dependent on the development stage (DVS) of the crop. As FERTIL, RECNIT is linearly interpolated at each day of simulation.

Table III.1.

View ANSWERS from the Tutorial_answer_sheet.pdf file.

Ex-III.3. In IR72DS2.T92, change the nitrogen simulation environment to NITROENV = 'NITROGEN BALANCE'. Scroll down to Section 6 ‘Nitrogen parameters’ and examine SOILSP, FERTIL and RECNIT. Recognize in FERTIL the fertilizer application rate for treatment 2 in Box III.1. The values for SOILSP were derived from the zero-N treatment and the values for RECNIT from a study of the N balance of a number of field experiments at IRRI (see Chapter 8.3 of the book “ORYZA2000: modelling lowland rice” for details). Run ORYZA2000, fill out column Sim2 in Table III.1. Notice in OP.DAT a new variable: ANCR, which is the total amount of N in the above-ground canopy (kg ha-1).

In Table III.1, there are differences in simulated yield when ORYZA2000 is run in potential N production mode (Sim1) and when ORYZA2000 is run with an N balance (Sim2), even if the fertilizer N application is high enough to guarantee unlimited N supply. The difference is caused by the differences in simulation of leaf N content (which strongly affects leaf photosynthesis rate). In ORYZA2000, the leaf photosynthesis rate is influenced by the concentration of N in the leaves on a leaf area basis (NFLV; g N m-2 leaf). In the potential N production mode, ORYZA2000 simply reads NFLV as input data from the crop data file. These data should be derived from measurements on a crop grown under conditions of potential production with ample supply of N. In the simulation with an N balance, ORYZA2000 calculates the leaf N content NFLV from daily supply and demand dynamics. The demand for N by the crop is partly derived from the concentration of N in the leaves on a leaf weight basis, FNLV (kg N kg-1 leaf dry matter). The maximum leaf N concentration on a leaf weight basis, NMAXLT, is read from the crop data file, and should (as NFLV) be derived from measurements on a crop grown under conditions of potential production with ample supply of N. The values of NFLV and NMAXLT in our crop data file IR72.DAT have been derived from a large number of data sets. Measurement inaccuracies and errors in green leaf weight and green leaf surface area have resulted in small discrepancies between maximum N concentration in leaves on area basis and on weight basis. This difference is the cause of the difference in simulated yield in the POTENTIAL PRODUCTION mode and in the NITROGEN BALANCE mode with ample supply of N fertilizer. Users of ORYZA2000 that enter their own measurements on leaf N concentration should expect similar discrepancies.

Ex-III.4. Using the results of the simulation from Ex-III.3, make a graph of simulated and observed weights of green leaves, panicles and total above-ground biomass versus development stage. Also notice the presence of some new variables that are being generated by the nitrogen balance modules of ORYZA2000. Make a graph of total N in the leaves (ANLV), in the panicles (ANSO) and in the total above-ground crop (ANCR) versus development stage. Note that all units are in kg N ha-1.

In Figure III.1a, we see good agreement between simulated and observed biomass using ORYZA2000 with an N balance. In Figure III.1b, we see the dynamics of N uptake by the crop. The time course of N in the whole crop shows an a typical pattern: periods with rapid uptake follow each fertilizer N application, and periods with slow accumulation follow when the applied N has been fully taken up and the crop can only satisfy its N demand from the natural soil N supply. In ORYZA2000, there is no dynamic simulation of the N balance in the soil, and in reality the pattern of N uptake by the crop probably looks more smoothly than the simulated curves in Figure III.1b. Nitrogen in the green leaves rapidly declines after flowering because of leaf death (shedding of leaves by the crop) and because of remobilization of N from the leaves and transfer to the panicles.

Figure III.1a. Simulated and observed dry weights (kg ha-1) of total above-ground biomass (WAGT), green leaves (WLVG) and storage organs (WSO); cv. IR72 with 225 kg N ha-1, IRRI, Los Baños, 1992.

Figure III.1b. Simulated total N (kg N ha-1) in the leaves (ANLV), in the panicles (ANSO) and in the whole above-ground crop (ANCR); cv. IR72 with 225 kg N ha-1, IRRI, Los Baños, 1992.

Ex-III.5. Make a graph of simulated and observed maximum and minimum nitrogen concentrations in the leaves (NMAXL, NMINL), with simulated (FNLV) and observed (FNLV_OBS) nitrogen concentrations in the leaves versus time. Remember that these N concentrations are on weight basis (kg N kg-1 leaf dry matter).

Figure III.1c. Simulated actual (FNLV), observed (FNLV_OBS) and theoretical maximum (NMAXL) and minimum (NMINL) N concentrations in the leaves (kg N kg-1 leaf dry matter); cv. IR72 with 225 kg N ha-1, IRRI, Los Baños, 1992

In Figure III.1c, the maximum and minimum leaf N concentrations are the values supplied in the crop data file (check this in IR72.D92!). The simulated N concentrations are close to the maximum and, as in Figure III.1b, periods of stabilizing N concentrations follow a fertilizer application, whereas periods of declining N follow the exhaustion of the pool of supplied fertilizer N. An exact similarity between simulated and observed leaf N concentration is very difficult to realize because (i) the dynamics of N in a crop are complex and difficult to model, and (ii) the measurement of N concentration in green leaves is inaccurate, since it is impossible to unequivocally distinguish green, yellow and dead leaves. In ORYZA2000, schematically only ‘green’ and ‘dead’ leaves are distinguished, whereas in reality, all shades of yellow occur in the phase of ripening.

After studying the simulation of crop growth and development at high fertilizer N application rates, we now turn to the simulation of crop growth under conditions of N deficiency. The files IR72DS1.T92 and IR72DS0.T92 contain the experimental information of the treatments 1 (180 kg N ha-1) and 0 (0 kg N ha-1), respectively, of the N experiment (see Box III.1).

Ex-III.6. Study the files IR72DS1.T92 and IR72DS0.T92 and verify that the fertilizer management information (FERTIL) corresponds to treatments 1 and 0, respectively, of the experiment (see Box II.1). Open the file RERUNS.DAT and create the following two reruns:

FILEIT = 'C:\COURSE\NITROGEN\IR72DS1.T92'   ! Experimental data

FILEIT = 'C:\COURSE\NITROGEN\IR72DS0.T92'   ! Experimental data

In these reruns, we identify complete data files as rerun option, instead of individual parameters from the data files! Run ORYZA2000 and fill out columns Sim3 (run 1, 180 kg N) and Sim4 (run 2, 0 N) in Table III.1. Explain the differences in Sim1, Sim2 and Sim3.

Under declining N availability, crop growth is increasingly restricted by N limitation and yield decreases. In the absence of fertilizer N supply, the crop contains 79 kg of N that has been supplied purely by the soil. This quantity of N has mainly been taken up after transplanting in the main field in 98 days (in OP.DAT, we read that the time between emergence and maturity, DAE, is 110 days and that 12 of these days were spent in the seedbed). A good estimate of the daily soil N supply capacity is obtained as total N taken up divided by the number of days = 79/98 = 0.8. This is the value given for SOILSP in the experimental data file (verify this!).

Ex-III.7. Use the graphics mode to make Figures III.2a, 2b and 2c for run 1 (180 kg N) and run 2 (0 N). Comment on the goodness-of-fit between the simulated and observed values. Try to relate the patterns of the curves to the timing of the fertilizer applications.

Especially noteworthy is the (simulated and observed) low leaf N concentration without N fertilizer application (Figure III.2c). This low leaf N concentration is the result of severe N limitation and is the reason for the low yield.

For comparison among the treatments, it is convenient to plot variables from different treatments in one graph.

Ex-III.8. Make graphs of different variables across treatments. For example, simulated and observed biomass and/or simulated and observed leaf N concentrations.

Figure III.2a. Simulated and observed dry weights (kg ha-1) of total above-ground biomass (WAGT), green leaves (WLVG) and storage organs (WSO); cv. IR72 with 0 kg N ha-1, IRRI, Los Baños, 1992.

Figure III.2b. Simulated total N (kg ha-1) in the leaves (ANLV), in the panicles (ANSO) and in the whole above-ground crop (ANCR); cv. IR72 with 0 kg N ha-1, IRRI, Los Baños, 1992.

Figure III.2c. Simulated (FNLV), observed (FNLV_OBS) and theoretical maximum (NMAXL) and minimum (NMINL) N concentration in the leaves (kg N kg-1 leaf dry matter); cv. IR72 with 0 kg N ha-1, IRRI, Los Baños, 1992.

Figure III.3a. Simulated and observed dry weights (kg ha-1) of total above-ground biomass (WAGT) of cv. IR72 with 225, 180 and 0 kg N ha-1, resp. Run 0, Run 1 and Run 2, IRRI, Los Baños, 1992.

Figure III.3b. Simulated N (kg ha-1) in the above-ground crop (ANCR) of cv. IR72 with 225, 180 and 0 kg N ha-1, IRRI, Los Baños, 1992.

Figure III.3c. Simulated and observed N concentration in the leaves (FNLV; kg N kg-1 leaf dry matter) of cv. IR72 with 225, 180 and 0 kg N ha-1, IRRI, Los Baños, 1992.

In Figure III.3a, we see that simulated differences in above-ground dry weight between treatment 1 (180 kg N) and treatment 2 (225 kg N) are very small. The experimental data also show little difference between these treatments. For both treatments, the simulation underestimates crop growth somewhat after flowering. In Figure III.3c, we again see that simulated differences in leaf N concentrations between treatments 1 and 2 are small, just as in the observations. In both, simulations and observations, leaf N concentration without fertilizer N (treatment 0) is considerably lower than in treatments 1 and 2.