Previous PageTable Of ContentsNext Page

Model-based analysis of maize plastic response to water stress for screening traits in improving breeding efficiency

Youhong Song1, Colin Birch1, 4, Shanshan Qu2, Jim Hanan3


1The University of Queensland, School of Land, Crop and Food Sciences, Gatton Campus, Gatton, 4343, Australia
Agricultural Science and Technology Extension Station of Qingdao, 266071, China
The University of Queensland, Centre for Biological Information Technology, Brisbane, 4072, Australia
Present Address: The University of Tasmania, Cradle Coast Campus, Burnie, 7320, Tasmania


To assist in finding traits for improved breeding efficiency, modelling approaches characterising crop phenotypic plasticity by unravelling the interaction of crop growth and environments can be used, therefore the response of crop model parameters to the environments of interest such as water stress needs to be adequately assessed. For this purpose, maize was grown under five water regimes from adequately watered to severely stressed in a glasshouse in 2007. Data on canopy morphology (individual leaf dimensions and internode length) and biomass accumulation were collected destructively at 1-2 day intervals. A carbon-driven functional-structural model Greenlab was used to characterize maize growth and development. Model parameters estimated by optimizing the simulation and observation of plant growth were compared across water regimes. Parameters governing biomass allocation were relatively insensitive to water stress. The hydraulic parameter r2 progressively increased with increasing water stress due to lesser irrigation, and its value was inversely exponentially related to leaf stomatal conductance. Overall, this study revealed information on the sensitivity of plant traits regarding model parameters to water stress by model-assisted phenotyping crop performance under differing watering conditions. These findings will be able to be used for heuristic screening of traits to inform plant breeding for tolerance of water stress.

Key Words

Zea mays, canopy, drought, functional-structural, model, phenotype


There is increasing interest in linking genotype to phenotype through crop and plant modelling approaches to improve breeding efficiency for complex traits such as adaptation and yield (Tardieu 2003; Hammer et al. 2005; Dingkuhn et al. 2005). A major challenge in using the modelling approaches is to select plant traits that relate to underlying processes within specific environments (Tardieu 2003; Hammer et al. 2005; Dingkuhn et al. 2005). GreenLab has been developed to simulate plant morphophysiological plasticity in response to environmental variation in a mechanistic manner (de Reffye et al. 1999; Guo et al. 2006). This model improved the limitations in prediction of plant structure encountered by conventional crop models when linking genotypes and phenotypes (Dingkuhn et al. 2005). Within this context, we aim to quantify responses of model parameters in Greenlab to water stress and propose their use for heuristic screening of traits for breeding.


Glasshouse experiment

A glasshouse experiment with maize grown under five water regimes was carried out during vegetative stages from September 13, 2007 to November 04, 2007. A completely randomized block design with five water treatments and three replicates was used. Three reference plants in each treatment were chosen to observe plant development. Eight pots plus guard pots were used in each replicate for each treatment, and arranged to produce a plant population density of 6 plants m-1 with 75cm interval between rows. Pots were placed in a rectangular tray to minimise loss of irrigation water. Pots were 25 cm high with a 24.5 cm top diameter and 19 cm bottom diameter. The soil was a black vertosol obtained from the field at Gatton Campus, The University of Queensland (Latitude 2734’S, longitude 15220’E). Each pot was 90% filled to 23 cm deep with the 8 kg of soil. Maize hybrid Pioneer 31H50 was sown at 3 seeds per pot, 5 cm deep. One well grown seedling per pot was retained after establishment at 3 leaf stage, the other two removed by cutting at the soil surface. A Tiny-tag(Gemini Loggers, UK) was placed above the canopy to monitor daily minimum and maximum temperatures and humidity. Field capacity of the soil was 38.8%, measured with a pressure (Ceramic Plate, USA), and used in determining water stress treatments that were imposed.

Water treatments

Five water treatments (T1-T5) were implemented by imposing differing amounts of water during different growth stages. T1 received least irrigation, and T2, T3, T4 and T5 received progressively more water, with T5 being the fully watered control treatment. Figure 1 shows irrigation accumulation plotted against thermal time after sowing (with 8C base temperature) calculated from temperatures recorded in the glasshouse for the five treatments.

Figure 1. Irrigation accumulation after sowing for five water regimes

Plant samplings

The leaves at positions five and ten on the plants, used to monitor leaf development, were tagged to ensure accuracy in leaf identification after senescence of lower leaves commenced. Plants were destructively sampled nine times from the 5 leaf stage at the interval of 1-2 days to obtain data on individual organ extension under differing water regimes. At each sampling time, two plants were selected from two of the three replicates of each treatment, and dissected. Data on the length of leaves and internodes and width of lamina at the widest point as well as biomass of individual organs were collected. Leaf stomatal conductance was measured using a porometer (Li-1600; LICOR, inc) between 11am-1.30pm from Oct 19, 2007 to Oct 31, 2007 by five times, the interval of measurement depending on atmospheric conditions in the glasshouse and leaf development.

Modelling studies

Details of Greenlab have been described extensively (de Reffye et al. 1999; Guo et al. 2006), thus this paper only briefly introduces the embedded parameters for characterizing biomass production and allocation that drive organ morphological development. The parameters embedded in Greenlab are primarily involved in sink strength of organ compartments (Pb, Pp and Pe, respectively refer to lamina (b), sheath (s) and internode(e)), sink strength variation parameterized by a beta law function with given constants N (e.g. 5), B (Bb, Bs and Be) and T (Tb, Ts and Te, equal to organ extension duration, directly measurable), and plant resistance r1 (denoting the resistance due to leaf size effect) and r2 (leaf hydraulic resistance, representing the resistance of water movement through plants (de Reffye et al. 1999; Guo et al. 2006). The values for embedded parameters were estimated by optimizing the comparisons between simulated and observed plant phenotypic characteristics through a generalized least square method (Zhan et al. 2003). Target files ie observed data at different stages such as dimensions and biomass of individual organs were required for the optimization.

Results and discussion

Plant characteristics

Rates of lamina, sheath and internode extension and final organ length were significantly affected by water stress. The reduced extension rate of organs was mainly responsible for the reduction in organ length. The duration of organ extension was significantly increased in severe water stress treatments (T1 and T2, data not shown), but not by mild water stress (T3 and T4) (Song et al. 2007).

Leaf stomatal conductance

Leaf stomatal conductance was measured in expanding leaves on 5 sampling dates for each watering regime, leaf stomotal conductance was shown to decline with diminishing water supply, from 0.420.11 mmol m-2s-1 (T5) to 0.030.01 mmol m-2s-1 (T1) (Figure 2). The variation of leaf stomotal conductance (standard errors of 20% of mean or less) indicated that stable water regimes were achieved.

Figure 2. Leaf stomatal conductance under five water regimes (vertical bars indicate s.e.).

Optimized parameters

Results from 9 samplings during rapid canopy development for each treatment were used as target files (observed data) for estimating the embedded parameters for Greenlab. The organogenesis was characterized by leaf initiation rate, estimated as 0.029C-1d-1 from the observation of tassel initiation. Duration of organ extension was not affected by water stress for T3 and T4, but it was extended for T1 and T2, the more severely water stressed treatments. The increase in duration of organ extension by severe water stress was included in parameter estimation. The relationship between lamina fresh weight and lamina area was also calculated automatically by the optimization program according to the measurements.

Table 1. Comparisons of parameter value of optimized parameters across five water regimes.









CV (%)


































































Figure 3. (a) Relationship between r2 and leaf stomatal conductance, and (b) biomass production by fixing leaf area and resistance.

Table 1 presents optimized parameters across five water regimes. The stability of parameter values was assessed by the coefficient of variation (CV, %) as in Guo et al (2006). Bb, Bp and Be had very low CVs (less than 10%), indicating that these parameters governing sink variation were not affected by water stress. Pp and Pe had CVs of 10% and 12% respectively, showing that sink strength of organs were only slightly affected by water stress, thus the relative amounts of photosynthate directed to each would not vary greatly. However, there are no data on root sink strength, which would need to be considered if the modelling was used for whole plants, rather than above ground plant parts. The parameter r1, for resistance due to leaf size was significantly reduced by water stress, consistent with reduced leaf size in stressed treatments, and r2, hydraulic resistance, was increased substantially in T1 and T2 with only small increases over the control (T5) in T4 and T3. The large increase between T3 and T2 indicates that there is a threshold value of leaf conductance near (just below) leaf stomatal conductance of 0.1 mmol m-2s-1 (Figure 3a) below which the value of r2, increases sharply, indicating plants are conserving their limited water supplies. Reduced leaf area and increased r2 both contributed to reduce biomass production. Separation of contribution of r1 and r2 by fixing the value of one factor at latest sampled stage indicated that increased hydraulic resistance (r2) resulted in much less biomass production than reduced leaf area in stressed treatments (Figure 3b).


The parameters governing biomass allocation among individual organs were relatively insensitive to water stress. However, the parameters used to calculate biomass production (r1 and r2) were significantly decreased and increased respectively by water stress, indicating that maize adaptation to water stress was mainly through adjusting resistance for biomass production rather than biomass allocation among organs. This study indicated that traits selection could be made more efficient using modelling-assisted approach.


de Reffye P, Blaise F, Chemouny S, Jaffuel S, Fourcaud T, Houllier F (1999) Calibration of hydraulic growth model on the architecture of cotton plants. Agronomie, 19: 265-280.

Dingkuhn M. Luquet D, Quilot B, de Reffye P (2005) Environmental and genetic control of morphogenesis in crops: Towards models simulating phenotypic plasticity. Australian Journal of Agricultural Research 56: 1-14.

Guo Y, Ma YT, Zhan ZG, Li BG, Dingkuhn M, Luquet D, de Reffye P (2006) Parameter optimization and field validation of the functional–structural model GREENLAB for Maize. Annals of Botany 97: 217-230.

Hammer GL, Chapman SC, Van Oosterom EJ, Podlich (2005) Trait physiology and crop modelling as a framework to link phenotypic complexity to underlying genetic systems. Australian Journal of Agricultural Research 56(9): 947-960.

Tardieu F (2003) Virtual plants: modelling as a tool for the genomics of tolerance to water deficit. Trends in Plant Science 8, 9-14.

Song Y, Birch C, Hanan J (2007) Architectural analysis and modeling of maize growth and development under water stress. 5th International Workshop on Functional Structural Plant Models (FSPM07), Napier New Zealand, 04-10, Nov, 2007.

Zhan ZG, de Reffye P, Houllier F, Hu BG (2003) Fitting a structural-function model with plant architectural data. In: Hu BG and M. Jaeger (Eds) Plant growth modelling and application. PMA03 Proceedings. Beijing, China: Tsinghua University Press, Springer, pp. 236-249.

Previous PageTop Of PageNext Page