A comparative analysis of four models of photosynthesis for 11 plant species in the Loess Plateau

https://doi.org/10.1016/j.agrformet.2004.07.004Get rights and content

Abstract

We compared performance and behavior of two suites of models of photosynthesis based on field data of 11 plant species in the semiarid Loess Plateau in northern China. Diurnal net photosynthesis rates and stomatal conductance of two C3 trees of Robinia pseudoacacia L. and Malus pumila Mill., two C3 shrubs of Caragana korshinskii Kom. and Hippophae rhamnoides L., a C3 subshrub of Lespedeza davurica (Laxm.) Schindl., two C3 forbs of Artemisia gmelinii Web. Ex Stechm. and A. giraldii Pamp., two C4 grasses of Panicum virgatum and Bothriochloa ischaemum (L.) Keng, and two C4 crops of Zea mays L. and Setaria italica (L.), were observed in field in three months of May, July and September in 2002. Net photosynthesis rates of the 11 species were then fitted with Farquhar's biochemical model of photosynthesis for C3 species, Berry and Farquhar's intercellular transport model for C4 species, and modified Thornley and Johson's leaf photosynthesis models for C3 and C4 species, and the observed stomatal conductance was fitted with a stomatal conductance model, by means of nonlinear least square regression. The four photosynthesis models were then coupled with the stomatal conductance model to predict net assimilation rates of these plants under extrapolated micro-environmental conditions. The results of nonlinear least-square regression showed that the biochemical models explained on average 66% and 82% of variations in observed net photosynthesis rates for the C3 and C4 plants, respectively, in comparison with the explanations of 72% and 76% by the leaf photosynthesis models. The more mechanistic, detailed treatment of biochemical processes in the biochemical photosynthesis models did not offer significant advantage over the simpler leaf photosynthesis models in the explanation of the field data, but tended to under predict net assimilation rates at high ranges for most C3 species. The subsequent extrapolation analysis indicated that net assimilation rates calculated with the leaf models are more strongly affected by stomatal conductance and moisture conditions, than the biochemical models.

Introduction

Models of photosynthesis have been playing key roles in predicting primary production of vegetation and crop yields under variable climatic environmental conditions, and thus have been extensively used in ecosystem simulations and crop modeling at various scales from spatially homogeneous patches to heterogeneous landscapes, regions, and the whole globe (Running and Coughlan, 1988, Raich et al., 1991, Running and Gower, 1991). Not only are ecosystem researchers concerned with the performance of carbon assimilation models under contemporary climate, but also it is important to know the behavior of these models under extrapolated future environmental conditions.

Mechanistic biochemical models of photosynthesis, represented by (Farquhar et al., 1980) and (Berry and Farquhar, 1978, Collatz et al., 1992), are in general derived from known quantitative relationships between various kinds of molecules involved in the biochemical processes of photosynthesis. These models emphasize detailed biochemical reaction processes such as carboxylase–oxygenase dynamics and electron transport. A common approach used in these models is to consider carbon assimilation as a process limited by a number of factors, each of which controls a sub-process, such as rubisco-limited carboxylation, light-limited electron transport and carboxylase–oxygenase production, etc. Each limiting factor gives a maximum allowable end assimilation rate, and the minimum of these allowable rates is considered as realized assimilation. This approach to take the minimum of a number of maximums has been used in a few other similar models (Collatz et al., 1991, Collatz et al., 1992). Among these biochemical models, Farquhar's model for C3 species (Farquhar et al., 1980), and Berry and Faruhar's intercellular transport model for C4 species (Berry and Farquhar, 1978), have been extensively applied to various ecosystem analyses and simulations (Medlyn and Dewar, 1996, Medlyn et al., 1999). On the other hand, leaf-level photosynthesis models by (Thornley and Johnson, 1990) featuring simplified gas exchange but emphasizing stomatal control of assimilation, received less attention in the literature. No comparison on characteristics and behavior between these two kinds of models has been made available to researchers. Is it a real gain to be more mechanistic to consider detailed biochemical processes such as the approach used in the biochemical models? What is the expense to use the simplified approaches of (Thornley and Johnson, 1990)? Questions of this kind are still open to investigations.

On the other hand, modeling ecophysiological processes for plants in arid and semiarid regions is especially challenging because plants in arid regions are often under extreme conditions so that ecosystem nonlinearity is most likely to come into play (Reynolds et al., 1996, Reynolds et al., 1999). A model may perform well when the variation ranges of driving variables (temperature, light intensity, and water stresses) are small or moderate, but may fail when these variation ranges are exceedingly large. Plants in arid regions are subject to large diurnal and seasonal variations of light intensity, temperature, and moisture stresses in both soil and air. These large variations in environmental variables may lead to complicated interactions of physical processes such as stomatal dynamics and biochemical processes of carboxylation–oxylation that control leaf-level gas exchanges and water use efficiency (Ball et al., 1987, Tezara et al., 1998, Yu et al., 1998).

The objectives of this paper are to compare performance and behavior of two biochemical photosynthesis models (one for C3, and the other for C4 plants) with those of two leaf photosynthesis models, by applying the models to 11 plant species of trees, shrubs, forbs, grasses and crops in the semiarid Loess Plateau, northern China. The models were fitted to the field-observed data of net photosynthesis to obtain physiologically significant parameters for carbon assimilation processes. A model of stomatal conductance (Gao et al., 2002) was fitted to the observed stomatal conductance of these plants, and was then coupled with the photosynthesis models to analyze behavior of the photosynthesis models under altered micro-environmental conditions.

Section snippets

Field measurements and data preparation

Our field experiments were conducted in a small rural watershed of Zhifanggou (109.25°E, 36.75°N) in the central Loess Plateau, northern China. The area has undergone serious deforestation and then re-vegetation over the past few decades. Turning formerly cultivated hilly agricultural crop fields into grasslands or forests is the most important measure adopted in ecosystem restoration programs for the area (Lu et al., 1997). With water as the major limitation for plant growth and severe

Parameters of biochemical models

Table 1 illustrates the estimated parameters of biochemical models, and Fig. 1 plots predicted net assimilation rates by the models against the observed values for the 11 species. The regressions using the biochemical models are shown all statistically significant with a mean R2 of 72.0%. The model for C4 performed better with average R2 = 82% than that for C3 plants with average R2 = 66%. The rubisco capacity varies from 13.1 for Z. mays to 87.0 μmol m−2 s−1 for A. giraldii, and average Vcmax25

Summary and conclusions

Performance of two suites of models of photosynthesis in explanation of variations in net assimilation rates measured in fields, and behavior of these models under extrapolated environmental regimes, were analyzed and compared for plant species from C3 trees, shrubs and forbs, to C4 grasses and crops, in the Loess Plateau, northern China. The biochemical models (Farquhar et al., 1980, Berry and Farquhar, 1978) and the leaf photosynthesis models (Thornley and Johnson, 1990) explained on average

Acknowledgments

This research was jointly supported by the Chinese Ministry of Science and Technology grant #G2000018605, the National Science Foundation of China grants #90202008 and #90211002, and BNU Creation Team Fund for the Project of Synthetic Landscape Dynamics.

References (27)

  • J.R. Cooke et al.

    A finite element shell analysis of guard cell deformations

    Trans. ASAE

    (1976)
  • D.G.G. dePury et al.

    Simple scaling of photosynthesis from leaves to canopies without the errors of big-leaf models

    Plant Cell Environ.

    (1997)
  • G.D. Farquhar et al.

    A biochemical model of photosynthetic CO2 Assimilation in leaves of C3 species

    Planta

    (1980)
  • Cited by (21)

    • Characteristics of E. japonicus stomatal conductance under water-deficit stress using a nonlinear Jarvis modified model

      2013, Mathematical and Computer Modelling
      Citation Excerpt :

      The relationship between stomatal movement and environmental factors has also been extensively investigated [1–7]. Some domestic researchers also made some meaningful research on the dynamics of the changing characteristics of stomatal conductance of different plant leaves, and on the effect of environmental factors on stomatal conductance [8–15]. However, the physiological mechanism of stomatal behavior is not fully understood.

    • Parsimoniously modelling perennial vegetation suitability and identifying priority areas to support China's re-vegetation program in the Loess Plateau: Matching model complexity to data availability

      2010, Forest Ecology and Management
      Citation Excerpt :

      In some regions, such as the CSHC, where floristic databases of species distribution, presence–absence and abundance field data are not available, the stochastic approaches described by methods 2 through 7 (outlined in Section 1) cannot be used. Additionally, while isolated cases of mechanistic modelling of plant growth have been reported for three tree species (H.C. Zhang et al., 2001; Zhang et al., 2003), and two tree and two shrub species in small catchments (the 8.27 km2 Zhifanggou catchment Gao et al., 2004) in the Loess Plateau, these methods are not ‘scaleable’ to the 113,000 km2 CSHC. Consequently we cannot consider models that: (1) require presence or presence–absence data or (2) are mechanistic.

    View all citing articles on Scopus
    View full text