Predicting growth and development of pigeonpea: A simulation model

Introduction

The development and application of simulation models of crops is well established in studying crop response to changes in genotype, cultivar, soil, weather, climatic patterns and management practices (Penning de Vries, 1977; Monteith and Virmani, 1991). Crop simulation model for pigeonpea has not been developed so far. A model of pigeonpea with the capacity to simulate the range of cultivar types from extra-short to medium duration phenology in response to weather, soil conditions and agronomic management was therefore developed under a collaborative endeavor of Agricultural Production Systems Research Unit (APSRU), Australia and ICRISAT-Patancheru. The model has been coded as a module in the cropping systems simulator APSIM (Agricultural Production System sIMulator) and has been prepared for publication (Robertson et al. 2001).

Brief description of the model

The pigeonpea model is a module of APSIM ( Agricultural Production Systems Simulator), (McCown et al., 1995). APSIM allows models of crop and pasture production, residue decomposition, soil water and nutrient flow to be readily configured to simulate various production systems, including crop sequences and intercropping, and soil and crop management to be dynamically simulated using conditional rules. The model uses a daily time-step, and is designed to simulate a uniform field and predict on an area basis grain yield, crop biomass, crop nitrogen uptake (including fixation) and partitioning within the plant. Different cultivars are defined in the model in terms of phenological development and partitioning of biomass to grain.

The parameters and relationships that were necessary to build the functions in the model were derived from 34 data sets derived from previously published studies in the literature (Carberry et al., 2000; Ranganathan et al., 2000, Robertson et al., 2000a) and unpublished reports of the experiments conducted at ICRISAT-Patancheru. These were essentially for phenology, leaf area development, biomass accumulation and partitioning, plant nitrogen accumulations, root growth and development and nitrogen accumulation, and effects of water deficits.

Progression through the crops' phenological phases triggers changes in the growth and partitioning of biomass between roots and above ground parts, including grain. It is simulated from sowing through ten stages:

1. Sowing,
2. Germination,
3. Emergence,
4. End of juvenile stage,
5. Floral initiation,
6. Flowering,
7. Start of grain filling,
8. End of grain filling,
9. Maturity,
10. Harvest ripe.

For emergence, 18.0 0 Cd is required plus 1.4 0 Cd per mm of sowing depth. Pigeonpea is a short-day plant and when the plant experiences photoperiods beyond a second critical value (Pc, h), progression to flowering ceases (i.e. a qualitative response) (Carberry et al., 2000). It is only when the photoperiod returns below this critical value that progression to flowering continues.

Leaf area development is described using functions for the appearance, expansion and senescence of leaves. Node and leaf appearance occur can potentially occur from emergence until maturity, depending upon the supply of assimilate available for leaf growth. Leaf appearance is determined from the rate of node appearance on the main stem, taken from Ranganathan et al. (2000) as 33 0 Cd per node, and the potential number of leaves per main stem node, taken as 10 leaves per node after the appearance of node 15 (Ranganathan et al., 2000).

Potential above ground biomass production is predicted from leaf area index, extinct_coef and the crop's radiation use efficiency (RUE). The value of RUE, 0.9 gMJ -1, is derived from Robertson et al. (2000b). RUE can be limited by sub-optimal or supra-optimal average daily temperature. The cardinal temperatures for this relationship are usually derived from the literature where leaf photosynthetic rate is measured as a function of temperature, however this information is not available for pigeonpea. In the absence of other information, RUE is assumed to reduced linearly by a factor varying between 0 and 1 when the mean daily temperature is between 0° C and 20° C and from 1 to 0 between 35° C and 45° C, respectively. These cardinal temperatures will need to be verified, especially in situations were the crop is grown during the cool season (rabi season in India). Daily-synthesized biomass is partitioned to roots, leaf, stems, pods and grain.

Soil nitrogen transformations and movement is simulated by other modules in the APSIM framework (Probert et al., 1997) for two soil types, an alfisol and a vertisol. Roots are grown daily in a fixed proportion to the tops production. Soil water infiltration and redistribution, evaporation and drainage is simulated by other modules in the APSIM framework (Probert et al., 1997). Water stress in the model reduces the rate of leaf area expansion, RUE and nitrogen fixation via three soil water deficit factors. When the legume module is coupled to SOILWAT (Probert et al., 1997), potential soil water uptake is calculated as described by Robertson et al. (2000b).

Model evaluation

Examples of the time-course of simulated and observed growth attributes of crops of extra-short, and short are given in Fig. 1 and medium duration cultivars Fig. 2. Comparisons were made between the simulated (Y) and observed (X) data with regression analyses of the form Y = a + b X. Measures of accuracy were made with the adjusted coefficient of determination (R 2 ) and the root mean squared deviation of between simulated and observed (Appendix-1). All of these examples were of crops sown at ICRISAT in the wet season, (kharif) on alfisol soils with irrigation applied to minimize water deficit. The model captures the pattern of LAI and the expected differences in peak LAI across the three main cultivar groups. The extent of leaf senescence and detachment in pigeonpea near maturity can be variable, therefore both the simulated total biomass (inclusive of senesced leaves) and green biomass (exclusive of senesced leaves) is plotted for comparison with observed values. The time-course and final values of total biomass (ca. 700, 900 and 1200 g m -2 ) are well simulated for the three groups, and within the simulated total and green biomass values. Fig. 2 shows extra growth and development detail for simulation of a medium-duration cultivar, cv. C11. It can be seen that not only is biomass simulated well, but the partitioning between leaf and stem is also simulated satisfactorily. Node number increases continuously until the end of grainfilling in medium-duration cultivars, and this phenomenon is reproduced well by the model (Fig. 2d).

Figure 1: Observed and simulated (a) leaf area index, and (b) total biomass and grain biomass in an irrigated extra-short duration pigeonpea crop (cv. ICPL 84023) (top two figures) and in an irrigated short-duration pigeonpea crop (cv. ICPL 87) (bottom two figures) sown 20 June 1994 on an alfisol at Patancheru at a plant density of 33 plants m -2 .

Figure 2: Observed and simulated (a) leaf area index, (b) total biomass and grain biomass, (c) green leaf and stem biomass, and (d) main stem node number in an irrigated medium-duration pigeonpea crop (cv. C11) sown 21 June 1991 on an alfisol at Patancheru at a plant density of 4.4 plants m -2 .

Data used for model evaluation were independent from those used to derive model parameters. These datasets (34) were obtained from published and unpublished field experiments from Patancheru (lat. 17 0 31'N, 78 0 16'E; elev. 530m), India. Briefly, the data comprise of experimental treatments with varying cultivars (extra-short, short and medium-duration), soil water conditions (irrigated and dryland), soil type (alfisol and vertisol) and plant population. The trials generally had minimum datasets required for model, but in case they were not available, data from nearby experiment was used to initialize the model. Where soil water contents could not be estimated from nearby fields plant extractable water was set to 10% for all soil layers at sowing. Since the experiments did not contain significant surface residues, errors in these estimates will have little consequence for the soil water and nitrogen balance in these validations. Of the 34 independent datasets used for model testing, 24 had recorded dates of 50% flowering and physiological maturity. Fig. 3 shows the excellent agreement between observed and simulated days to flowering and days to physiological maturity, with RMSD values being 4.3 and 9.8 days, and the coefficient of determination of 96 and 92% for flowering and maturity, respectively. The time to maturity was simulated with less accuracy than flowering, possibly reflecting the additive effect of errors in simulating the intermediate flowering and grain filling stages. Also, I a crop such as pigeonpea, there can be difficulty in accurately determining when physiological maturity occurs in the field, due to the occurrence of multiple flushes of pods.

Grain yield was simulated well by the model with a RMSD of 332 g m -2 , which was 18.2% of the observed mean grain yield (Fig. 4). The R2 value for observed vs. simulated regression was 76%. Green biomass at maturity was simulated less well - overall the model over-predicted biomass as shown by the large positive intercept for the regression of observed vs. simulated. The difficulty in gaining good agreement between simulated and observed biomass at maturity is due, in part, to the issue of variable leaf loss in pigeonpea, referred to above. It was possible to identify some of the reasons for poor agreement between simulated and observed biomass and yield in Fig. 4. In particular, waterlogging was thought to limit crop growth in some datasets - those that experienced heavy rain in the first half of the season and the observed data shows a cessation in increase or a decrease in LAI in apparently well-watered conditions. Usually the suspected waterlogging influence was more obvious on vertisol soil types, under irrigation, where waterlogging is often observed at ICRISAT (Chauhan, 1987).

Figure 3. Observed and simulated (a) days to first flower, and (b) days to physiological maturity. ESD = extra-short duration cultivars, SD = short-duration cultivars, M = medium-duration cultivars. The 1:1 line is shown.

Figure 4: Observed and simulated (a) grain yield, (b) non-senesced (green) biomass at maturity. ESD = extra-short duration cultivars, SD = short-duration cultivars, M = medium-duration cultivars. The 1:1 line is shown. Datasets with suspects waterlogging limitation to growth are identified by the diamond symbol.

Conclusions

The model described here gives good predictive capability for pigeonpea phenology, leaf area, biomass and grain yield, despite the relatively simple approach taken. A key feature of this model is that relatively few parameters (phenology, partitioning during grainfilling) are used to define each maturity type, yet satisfactory predictions of phenology and yield are possible across a wide range of cultivars.

The set of experiments used to test the model, while all conducted at ICRISAT Center, Patancheru in India, encompassed a wide range in yield levels due to manipulation of the key agronomic variables of sowing date, density, cultivar, irrigation, and soil type. Hence, the model can be used with some confidence outside central India. Carberry et al. (2000) showed that the phenology sub-model predicted flowering dates at locations in northern India and eastern Africa, and with some verification against local data there is no reason why it should not be able to also predict growth and yield in these environments.

Currently much of the world's pigeonpea cultivation is in intercropping situations. APSIM, with its intercropping capability (Carberry et al., 1996), is a useful tool to explore crop management issues in mixed-species systems.

We have identified further areas of model improvement include waterlogging effects on crop growth. In addition, the ability to simulate crop N accumulation (fixation and mineral N uptake) needs to be tested.

Acknowledgments

This work was funded by the Australian Center for International Agricultural Research and ICRISAT core funds in a collaborative project (CARMASAT). The scientist involved were MJ Robertson PS Carberry from APSRU; YS Chauhan, R Ranganathan, GJ O'Leary from ICRISAT. Dr JP Dimes helped collate the APSIM soil parameters for the ICRISAT soils used in this study and John Hargreaves, Anil Kumar Varma and Gayatri Devi provided helpful programming assistance.

References

Carberry PS, Adiku SGK, McCown RLand Keating BA. 1996. Application of the APSIM cropping systems model to intercropping systems. In: O Ito, C Johansen, JJ Adu-Gyamfi, K Katayama, JVDK Kumar Rao, and TJ Rego (Eds.) Dynamics of Roots and Nitrogen in Cropping Systems of the Semi-Arid Tropics, pp. 637-648. Japan International Research Center for Agricultural Sciences.

Carberry PS, Ranganathan R, Reddy LJ, Chauhan YS and Robertson MJ. 2000. Predicting growth and development of pigeonpea: flowering response to photoperiod. Field Crops Research (accepted)

Chauhan YS. 1987. Screening for tolerance to salinity and waterlogging: case studies with pigeonpea and chickpea. In: "Adaptation of Chickpea and Pigeonpea to Abiotic Stresses. Proceedings of the Consultant's Workshop, 19-21 December 1984, ICRISAT Center, India". Patancheru, A. P., India: ICRISAT, pp. 99-103.

McCown RL, Hammer GL, Hargreaves JNG, Holzworth DPand Freebairn DM. 1996. APSIM: a novel software system for model development, model testing and simulation in agricultural systems research. Agricultural Systems 50: 255-271.

Monteith JL and Virmani SM. 1991. Quantifying climatic risk in the semi-arid tropics: ICRISAT experience. In: R.C. Muchow and J.A. Bellamy (Editors), Climatic Risk in Crop Production: Models and Management for the Semiarid Tropics and Subtropics. C.A.B. International, Wallingford, UK., pp. 183-204.

Penning de Vries FWT. 1977. Evaluation of simulation models in agriculture and biology: conclusions of a workshop. Agricultural Systems, 2, 99-107.

Probert ME, Dimes JP, Keating BA, Dalal RCand Strong WM. 1997. APSIM's water and nitrogen modules and simulation of the dynamics of water and nitrogen in fallow systems. Agricultural Systems 55: 1-28.

Ranganathan R, Chauhan YS, Flower DJ, Robertson MJ, Sanetra C, and Silim SN. 2000. Predicting growth and development of pigeonpea: leaf area development. Field Crops Research (in review)

Robertson MJ, Carberry PS, Probert ME, Poulton PL, Bell M, Wright GC, Turpin JE, Yeates SJ and Brinsmead B. 2000a. Simulation of growth and development of diverse legume species in APSIM II. Model validation. (in preparation).

Robertson MJ, Silim SN, Chauhan YS and Ranganathan R (2000b). Predicting growth and development of pigeonpea: biomass accumulation and partitioning. Field Crops Research (in review)

Robertson MJ, Carberry PS, Chauhan YS, Rangnathan R and O'Leary GL. 2001. Predicting growth and development of pigeonpea: A simulation model. Prepared for Field Crops Research.

For more information please contact:

Dr Chauhan Y S
Principal Scientist (Crop Physiology),
ICRISAT-Patancheru 502 324.