| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
School of Nat. Resour. Sci., Univ. of Nebraska, P.O. Box 830728, Lincoln, NE 68583-0728
* Corresponding author (aweiss1{at}unl.edu)
Received for publication May 1, 2001.
ASSOCIATED WITH THE INTRODUCTION of many new technologies, whether it be medical, mechanical, or Internet related, are promises; sometimes, too many promises are made about the potential capabilities of these new technologies. After the initial period of euphoria is over, reality sets in. Then the hard work begins of making promises meet expectations. Eventually, those technologies that narrow the difference between expectations and reality survive and flourish. Some technologies die and are forgotten, and others may become the butt of jokes.
In the mid-1960s, crop simulation modeling began with the pioneering work of C.T. de Wit (1965). Overly enthused scientists, engineers, and administrators raised high expectations of the many possible uses of this new technology. In 1973, John Passioura's article "Sense and Nonsense in Crop Simulation" was published in the Journal of the Australian Institute of Agricultural Science, providing a counter balance to voices that were overly optimistic. A challenge that we constantly face is to inject healthy skepticism into our work, to keep us from becoming too comfortable. To quote from Daniel Boorstin, noted historian and former head of the Library of Congress, "The greatest obstacle to discovery is not ignorance, but the illusion of knowledge."
My perception is that crop simulation modeling has reached a relative plateau; there have been no large changes in how these models operate in recent years but more of a refining of existing algorithms and the development of user-friendly interfaces. This is a dangerous situation as it can indicate a sense of arrogance and complacency; it's as good as it gets.
One way to gain insight into the current status of crop simulation modeling is by looking at the inputs that are common to the many types of crop simulation models. These categories of inputs quantify weather, soil, managerial options, and plant responses to the environment. With the development of automated weather station networks in the early 1980s, hourly values of temperature, solar radiation, precipitation, relative humidity, wind direction, and wind speed become available in addition to the long-established record of cooperative weather data usually consisting of maximum and minimum temperatures and precipitation. Satellite data have been used to model incoming solar irradiance. Models to predict weather parameters for nonuniform locations, those with different azimuth and slope than a flat surface, are available. In other words, the successful application of a crop simulation model is not limited by the availability of weather data. Similarly, there are sufficient soil data available, either through actual measurements or from databases on the web, so these data are not generally a factor that limits the successful application of a crop simulation model. Managerial options such as day of sowing, plant density, irrigation and fertilizer amounts and dates of application should be known. Obviously, these options are not a limiting factor in this situation.
Model inputs to quantify plant responses to the environment, sometimes called genetic coefficients, usually represent optimum measures of time, how long the crop will be in a specific developmental stage and, with regards to yield, what happens to the reproductive organs during this time. Going from these optimum inputs to the simulated actual plant responses is a major aspect of any simulation model. The time component of these coefficients can be determined by visible changes in the plant, relatively easily, compared with what happens physiologically to the reproductive organs. While initially, time and yield components had to be quantified, with the desire for model generality and the large effort necessary to accurately determine these yield components, it is my perception that it became easier to measure the time component and final yield and then use the model in an iterative procedure to determine the genetic coefficients for a cultivar. In determining genetic coefficients for a cultivar grown at two locations in 1 yr or at the same location for 2 yr, it is possible to obtain two different sets of coefficient values. Thus, the use of genetic coefficients in crop simulation models and how they are determined may not capture the range of possible plant responses to the environment. One must also consider the possibility that there could have been errors associated with field measurements of parameters directly related to genetic coefficients or that simulated plant responses in the model may not be entirely synthesized by current knowledge.
The plant's organ-level responses, eventually leading to yield, represent an integration of genetic responses to the environment. Until recently, these organ-level inputs were reasonable approaches in crop simulation models and worked in harmony with algorithms in the model to simulate actual responses. Given the above concerns about the consistency of genetic coefficients and resulting predictions and the rapid changes in our knowledge of the plant genome, how can we capture this genetic information and incorporate it into crop simulation models? This effort could enhance our understanding of plant responses to the environment by integrating responses from the gene level to the organ level as well as improve predictions of yield and other plant products that may have economic importance, such as oil and protein content. Another potential impact would be the decrease in the length of time necessary to develop new cultivars and the development of cultivars with unique properties for specific locations or environments, especially in the developing world. I believe we are at the beginning of an exciting new era in crop simulation modeling if we can bring together two divergent groups of scientists that, up till now, have not interacted often. Bringing modelers and plant geneticists together will be enabled by appropriate levels of funding and will be enhanced by developing a common set of goals. What is known, what do we need to know, and how can we translate results from one scientific discipline to another discipline? This will be an exciting journey, but as with any journey, frustrations along the way should be expected. There are few options but to proceed on this venture not knowing if in the end there will be greater understanding resulting in improved predictions. Have we come full circle to unbridled optimism concerning the potential of incorporating genomics into crop simulation modeling? I don't think so. I believe many scientists who engage in various aspects of crop simulation modeling are aware of this potential, yet realize the difficulties to overcome to achieve this potential.
The symposium "Crop Modeling and Genomics" was held at the 2000 ASA-CSSA-SSSA annual meetings in Minneapolis, MN. Divisions A-3, C-1, and C-2 sponsored this symposium, the goal of which was to synthesize current knowledge to establish the beginnings of this journey. The papers presented at this symposium represent a variety of ways to address the incorporation of genomics into crop simulation modeling. This diversity in approaches ranges from modifying existing algorithms to simulate new responses to addressing this topic at the most basic level of current understanding. Having a detailed knowledge of a crop simulation model provided Ritchie and Alagarswamy (p. 4–9) a way to modify and expand the ability of CERES-Maize to simulate bareness and prolificacy. They relate kernel number per plant, with new genetic coefficients, to cumulative intercepted photosynthetically active radiation around silking. With the goal of assisting plant breeders gain information about the response of spring wheat to early vigor, simulated by increasing specific leaf area, Asseng et al. (p. 10–19) evaluate spring wheat responses in several environments in Australia using APSIM-Nwheat. Yields could increase or decrease depending on the climate, soil type, and nutrient management. Using the simulation model Cropsim to study wheat responses in different environments, Hunt et al. (p. 20–31) conclude that unknown factors may influence yield, such that incorporating a genetic response to growth will be difficult compared with simulating crop development. Rather than trying to deal with individual gene responses, they suggest the current focus of research should be on "grouping of genes into packages that can be characterized in terms of discrete coefficients." Boote et al. (p. 32–51) investigate a wide range of simulated soybean responses by varying the genetic coefficients in CROPGRO-Soybean to produce changes in phenological development, crop assimilation, vegetative vigor, leaf area expansion of determinate and indeterminate cultivars, seed-fill rates, and traits to improve production under drought. They attempt to relate the changes in these coefficients to the physiology of the plant and, where possible, extend their interpretations of the results to lower levels of plant organization.
The paper by White and Hoogenboom (p. 52–64) begins with an introduction to basic functional genomics, then discusses six levels of genetic detail that can be incorporated into crop simulation models, and concludes by looking at potential outcomes, what species and traits to simulate, and the importance of the necessary collaboration between modelers and molecular biologists. Stewart et al. (p. 65–70) focuses on the development of a model that calculates the effects of alleles at six maturity loci on the photothermal response of soybean. They conclude that if the genetic makeup of a cultivar is known, time from planting to first flower can be predicted using average daily temperature, sowing date, and latitude. Welch et al. (p. 71–81) take a very different approach to predict flowering; they use a neural network model to predict flowering in contrasting strains of Arabidopsis thaliana. This approach was able to demonstrate an important characteristic that will be necessary to quantify genetic inputs in future crop simulation models, the ability to predict a temperature-dependent change in transition order. Hoogenboom and White (p. 82–89) point out the promise and frustration associated with attempting to incorporate gene responses into GeneGro, which was based on a simulation model of common bean (Phaseolus vulgaris L.). While there was a slight improvement in predicting days to flowering, days to maturity and growth-related variables were not improved with the incorporation of a gene that inhibits cold temperature response to photoperiod. A limitation to their approach was access to data that represents a wide range of environments where these genes are active.
An often sought but elusive goal of crop simulation modeling has been applications to plant breeding. Yin et al. (p. 90–98) suggest a complementary framework using quantitative trait loci (QTL) to represent input parameters in a crop simulation model while the models may assist in QTL mapping by "dissecting yield into physiological components that are more likely related directly to gene expression." Chapman et al. (p. 99–113) take a different approach to address issues associated with plant breeding by combining the crop simulation model APSIM-Sorg with QU-GENE, a genetically based simulation model for developing plant breeding strategies. With this approach, they studied four traits (stay green, phenology, osmotic adjustment, and transpiration efficiency) that took into account epistasis and genotype x environment interaction for yield.
Many interesting ideas were presented in this symposium. What is the next step? How do we proceed to pursue these ideas, assuming that these ideas are of sufficient merit that they should be pursued? How do two groups of scientists, crop modelers and plant geneticists, that work on different scales of plant organization using different methodologies cooperate to achieve mutually beneficial results? The general answer is a combination of altruism and money. As one always goes from the general to the specific, the devil is in the details.
 |
REFERENCES |
|---|
 TOP REFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  X. Yin, P. C. Struik, F. A. van Eeuwijk, P. Stam, and J. Tang QTL analysis and QTL-based prediction of flowering phenology in recombinant inbred lines of barley J. Exp. Bot., March 1, 2005; 56(413): 967 - 976. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. L. Hammer, T. R. Sinclair, S. C. Chapman, and E. van Oosterom On Systems Thinking, Systems Biology, and the in Silico Plant Plant Physiology, March 1, 2004; 134(3): 909 - 911. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| The SCI Journals | Crop Science | Vadose Zone Journal | |||
| Journal of Natural Resources and Life Sciences Education |
Soil Science Society of America Journal | ||||
| Journal of Plant Registrations | Journal of Environmental Quality |
The Plant Genome | |||
