Gail G. Wilkerson

Modeling Soybean Growth for Crop Management

ABSTRACT Asoybean (Glycine max (L.) Merr.) crop growth simulation model (SOYGRO) was developed to aid farm managers in making irrigation and pest management decisions. Non-linear first order differential equations describe dry matter rates of change, accumulation and depletion of protein pools, and changes in shell and seed numbers. Two data sets from defoliation and irrigation experiments were used for calibration and validation of the model. The model responds well to drought and defoliation stresses for two test cases. Sensitivity analyses of SOYGRO revealed that simulated yield was most sensitive to changes in gross photosynthesis and growth respiration. The sensitivity of simulated yield to changes in model parameters was increased by the occurrence of either water or defoliation stress.

Weed management decision models: pitfalls, perceptions, and possibilities of the economic threshold approach

Abstract The use of scouting and economic thresholds has not been accepted as readily for managing weeds as it has been for insects, but the economic threshold concept is the basis of most weed management decision models available to growers. A World Wide Web survey was conducted to investigate perceptions of weed science professionals regarding the value of these models. Over half of the 56 respondents were involved in model development or support, and 82% thought that decision models could be beneficial for managing weeds, although more as educational rather than as decision-making tools. Some respondents indicated that models are too simple because they do not include all factors that influence weed competition or all issues a grower considers when deciding how to manage weeds. Others stated that models are too complex because many users do not have time to obtain and enter the required information or are not necessary because growers use a zero threshold or because skilled decision makers can make better and quicker recommendations. Our view is that economic threshold–based models are, and will continue to be, valuable as a means of providing growers with the knowledge and experience of many experts for field-specific decisions. Weed management decision models must be evaluated from three perspectives: biological accuracy, quality of recommendations, and ease of use. Scientists developing and supporting decision models may have hindered wide-scale acceptance by overemphasizing the capacity to determine economic thresholds, and they need to explain more clearly to potential users the tasks for which models are and are not suitable. Future use depends on finding cost-effective methods to assess weed populations, demonstrating that models use results in better decision making, and finding stable, long-term funding for maintenance and support. New technologies, including herbicide-resistant crops, will likely increase rather than decrease the need for decision support.

A proposal to standardize soil/solution herbicide distribution coefficients

Abstract Herbicide soil/solution distribution coefficients (Kd) are used in mathematical models to predict the movement of herbicides in soil and groundwater. Herbicides bind to various soil constituents to differing degrees. The universal soil colloid that binds most herbicides is organic matter (OM), however clay minerals (CM) and metallic hydrous oxides are more retentive for cationic, phosphoric, and arsenic acid compounds. Weakly basic herbicides bind to both organic and inorganic soil colloids. The soil organic carbon (OC) affinity coefficient (Koc) has become a common parameter for comparing herbicide binding in soil; however, because OM and OC determinations vary greatly between methods and laboratories, Koc values may vary greatly. This proposal discusses this issue and offers suggestions for obtaining the most accurate Kd, Freundlich constant (Kf), and Koc values for herbicides listed in the WSSA Herbicide Handbook and Supplement. Nomenclature: Readers are referred to the WSSA Herbicide Handbook and Supplement for the chemical names of the herbicides.

Evaluation of the CROPGRO-Soybean model over a wide range of experiments

Crop simulation models are increasingly being used to predict yield responses to soil, weather, and management conditions. This requires that the models be evaluated for their abilities to accurately respond to those factors. Our objective in this paper was to evaluate the recently released CROPGRO-Soybean model for its ability to simulate soybean growth, seed yield, flowering dates, and season lengths over a wide range of conditions. Inputs (weather, soil characteristics, management practices) and data on growth and yield were assembled for several cultivars from various locations in the USA. The Bragg cultivar was evaluated in multiple years at Gainesville under varying water supply and showed the model ability to predict water limitations on growth. The Williams cultivar was evaluated in multiple years at sites in Iowa, Ohio, and Florida and illustrated CROPGRO ability to predict in different locations and climatic environments. Planting date studies were simulated for three cultivars in North Carolina to evaluate model ability to predict yield response to planting date. Model ability to predict growth and yield in two ‘on-farm’ soybean trials was evaluated. Three of the experiments (Williams cultivar in Florida, planting date trials, and on-farm trials) represent independent data never used in model calibration and illustrate the ability of CROPGRO to predict growth and yield in new locations and environments. We conclude that CROPGRO-Soybean gives reasonable predictions under a wide range of environmental conditions.

Modeling competition for light between soybean and broadleaf weeds

Abstract Crop yield reductions due to weed competition can vary with both environmental and cultural conditions. Models of resource use by crops and weeds will likely be necessary to accurately simulate crop losses from weed competition. Many broadleaf weeds cause losses in soybean ( Glycine max (L.) Merr.) through competition for light. A simple model of canopy structure and light interception by soybean and broadleaf weeds within a weed area of influence has been developed and incorporated into SOYWEED, a dynamic soybean-weed competition model. Daily photosynthetic rates per unit ground area are calculated according to interception of direct light by the weed and crop. Interception is approximated from the simple exponential extinction of light by a plant canopy. For competing plants, interception depends on both the amount and arrangement of leaf area. Arrangement of that leaf area is described by extinction coefficients, the plant height, and the vertical distribution of leaf area. Preliminary testing with cocklebur ( Xanthium strumarium L.) competition data indicates an improved simulation of weed and crop growth and crop yield with the incorporation of this model into SOYWEED, but highlights the lack of data on weed and soybean canopy structure. Simulated crop response to changes in weed emergence date and time before weed removal is consistent with general observations reported in the literature. Simulation results indicate that several weed canopy characteristics may contribute to the differential ability of weed species to compete with soybean.

Value of information about weed distribution for improving postemergence control decisions

Abstract Weeds apparently occur in patches within fields. This spatial distribution has implications for choosing the most profitable postemergence control measure, because weed distribution influences the yield loss from competition, the design of the optimal scouting plan and the feasibility of patch spraying. Simulation models that use data on the distribution and composition of actual populations may be used to examine these implications for choosing between many potential treatments for postemergence control of a mixed species population. Simulation experiments were carried out to investigate the value of information about weed patchiness for improving the recommendations of a decision model (HERB) for postemergence weed control in soybean. Information about weed patchiness was more valuable when used to account for the possible error in the density estimates obtained by scouting than when used to increase the accuracy of the yield loss prediction. Accurate scouting was shown to be important for choosing treatments when control is required, as well as determining if control is necessary. Simulation results may be used to identify the optimal scouting plan once information about the cost of scouting becomes available.

Yield and Physiological Response of Peanut to Glyphosate Drift

Five experiments were conducted during 2001 and 2002 in North Carolina to evaluate peanut injury and pod yield when glyphosate was applied to 10 to 15 cm diameter peanut plants at rates ranging from 9 to 1,120 g ai/ha. Shikimic acid accumulation was determined in three of the five experiments. Visual foliar injury (necrosis and chlorosis) was noted 7 d after treatment (DAT) when glyphosate was applied at 18 g/ha or higher. Glyphosate at 280 g/ha or higher significantly injured the peanut plant and reduced pod yield. Shikimic acid accumulation was negatively correlated with visual injury and pod yield. The presence of shikimic acid can be detected using a leaf tissue assay, which is an effective diagnostic tool for determining exposure of peanut to glyphosate 7 DAT. Nomenclature: Glyphosate; peanut, Arachis hypogaea L. ARHHY.

HADSS™, Pocket HERB™, and WebHADSS™: Decision Aids for Field Crops1

Row crop weed management decisions can be complex due to the number of available herbicide treatment options, the multispecies nature of weed infestations within fields, and the effect of soil characteristics and soil-moisture conditions on herbicide efficacy. To assist weed managers in evaluating alternative strategies and tactics, three computer programs have been developed for corn, cotton, peanut, and soybean. The programs, called HADSS™ (Herbicide Application Decision Support System), Pocket HERB™, and WebHADSS™, utilize field-specific information to estimate yield loss that may occur if no control methods are used, to eliminate herbicide treatments that are inappropriate for the specified conditions, and to calculate expected yield loss after treatment and expected net return for each available herbicide treatment. Each program has a unique interactive interface that provides recommendations to three distinct kinds of usage: desktop usage (HADSS), internet usage (WebHADSS), and on-site usage (Pocket HERB). Using WeedEd™, an editing program, cooperators in several southern U.S. states have created different versions of HADSS, WebHADSS, and Pocket HERB that are tailored to conditions and weed management systems in their locations. Nomenclature: Corn, Zea mays L.; cotton, Gossypium hirsutum L.; peanut, Arachis hypogea L; soybean, Glycine max L. Additional index words: Bioeconomic models, computer decision aids, decision support systems, weed management. Abbreviations: HADSS, Herbicide Application Decision Support System; PDS, postemergence-directed; POST, postemergence; PPI, preplant-incorporated; PRE, preemergence.

Decision analysis as a tool for integrating simulation with expert systems when risk and uncertainty are important

Abstract Risk and uncertainty are important components of agricultural decision making. The methodology of applied decision analysis is especially useful in addressing such problems, but has not been widely integrated with expert systems. One problem has been the difficulty of handling uncertainty within the expert system framework in a way which is logically consistent with rational decision criteria. An additional problem in agriculture is the need to combine uncertain or incomplete information from simulations and statistical studies with the subjective knowledge of one or several experts. In this paper, it is argued that Bayesian probability theory provides a natural approach, and a methodology is developed for combining diverse sources of information within the framework of an expert system. The methodology is developed within the context of an expert system for protection of soybeans against corn earworm, using information from HELSIM, a heliothis population model (R. Stinner) and from the SOYGRO soybean crop model (G. Wilkerson).

Evaluating the fidelity of downscaled climate data on simulated wheat and maize production in the southeastern US

Crop models are one of the most commonly used tools to assess the impact of climate variability and change on crop production. However, before the impact of projected climate changes on crop production can be addressed, a necessary first step is the assessment of the inherent uncertainty and limitations of the forcing data used in these crop models. In this paper, we evaluate the simulated crop production using separate crop models for maize (summer crop) and wheat (winter crop) over six different locations in the Southeastern United States forced with multiple sources of actual and simulated weather data. The paper compares the crop production simulated by a crop model for maize and wheat during a historical period, using daily weather data from three sources: station observations, dynamically downscaled global reanalysis, and dynamically downscaled historical climate model simulations from two global circulation models (GCMs). The same regional climate model is used to downscale the global reanalysis and both global circulation models’ historical simulation. The average simulated yield derived from bias-corrected downscaled reanalysis or bias-corrected downscaled GCMs were, in most cases, not statistically different from observations. Statistical differences of the average yields, generated from observed or downscaled GCM weather, were found in some locations under rainfed and irrigated scenarios, and more frequently in winter (wheat) than in summer (maize). The inter-annual variance of simulated crop yield using GCM downscaled data was frequently overestimated, especially in summer. An analysis of the bias-corrected climate data showed that despite the agreement between the modeled and the observed means of temperatures, solar radiation, and precipitation, their intra-seasonal variances were often significantly different from observations. Therefore, due to this high intra-seasonal variability, a cautious approach is required when using climate model data for historical yield analysis and future climate change impact assessments.