Kelly R. Thorp

Development and evaluation of a field-based high-throughput phenotyping platform

Physiological and developmental traits that vary over time are difficult to phenotype under relevant growing conditions. In this light, we developed a novel system for phenotyping dynamic traits in the field. System performance was evaluated on 25 Pima cotton (Gossypium barbadense L.) cultivars grown in 2011 at Maricopa, Arizona. Field-grown plants were irrigated under well watered and water-limited conditions, with measurements taken at different times on 3 days in July and August. The system carried four sets of sensors to measure canopy height, reflectance and temperature simultaneously on four adjacent rows, enabling the collection of phenotypic data at a rate of 0.84ha h-1. Measurements of canopy height, normalised difference vegetation index and temperature all showed large differences among cultivars and expected interactions of cultivars with water regime and time of day. Broad-sense heritabilities (H2)were highest for canopy height (H2=0.86-0.96), followed by the more environmentally sensitive normalised difference vegetation index (H2=0.28-0.90) and temperature (H2=0.01-0.90) traits. We also found a strong agreement (r2=0.35-0.82) between values obtained by the system, and values from aerial imagery and manual phenotyping approaches. Taken together, these results confirmed the ability of the phenotyping system to measure multiple traits rapidly and accurately.

Simulating Long-Term Effects of Nitrogen Fertilizer Application Rates on Corn Yield and Nitrogen Dynamics

Thoroughly tested agricultural systems models can be used to quantify the long-term effects of crop management practices under conditions where measurements are lacking. In a field near Story City, Iowa, ten years (1996-2005) of measured data were collected from plots receiving low, medium, and high (57-67, 114-135, and 172-202 kg N ha-1) nitrogen (N) fertilizer application rates during corn (Zea mays L.) years. Using these data, the Root Zone Water Quality Model linked with the CERES and CROPGRO plant growth models (RZWQM-DSSAT) was evaluated for simulating the various N application rates to corn. The evaluated model was then used with a sequence of historical weather data (1961-2005) to quantify the long-term effects of different N rates on corn yield and nitrogen dynamics for this agricultural system. Simulated and measured dry-weight corn yields, averaged over plots and years, were 7452 and 7343 kg ha-1 for the low N rate, 8982 and 9224 kg ha-1 for the medium N rate, and 9143 and 9484 kg ha-1 for the high N rate, respectively. Simulated and measured flow-weighted average nitrate concentrations (FWANC) in drainage water were 10.6 and 10.3 mg L-1 for the low N rate, 13.4 and 13.2 mg L-1 for the medium N rate, and 18.0 and 19.1 mg L-1 for the high N rate, respectively. The simulated N rate for optimum corn yield over the long term was between 100 and 150 kg N ha-1. Currently, the owner-operator of the farm applies 180 kg N ha-1 to corn in nearby production fields. Reducing long-term N rates from 180 to 130 kg N ha-1 corresponded to an 18% simulated long-term reduction in N mass lost to water resources. Median annual FWANC in subsurface drainage water decreased from 19.5 to 16.4 mg N L-1 with this change in management. Current goals for diminishing the hypoxic zone in the Gulf of Mexico call for N loss reductions of 30% and greater. Thus, long-term simulations suggest that at least half of this N loss reduction goal could be met by reducing N application rates to the production optimum. However, additional changes in management will be necessary to completely satisfy N loss reduction goals while maintaining acceptable crop production for the soil and meteorological conditions of this study. The results suggest that after calibration and thorough testing, RZWQM-DSSAT can be used to quantify the long-term effects of different N application rates on corn production and subsurface drainage FWANC in Iowa.

SIMULATING THE LONG-TERM PERFORMANCE OF DRAINAGE WATER MANAGEMENT ACROSS THE MIDWESTERN UNITED STATES

Drainage water management (DWM) has been proposed as a solution to reduce losses of nitrate (NO3) from subsurface drainage systems in the midwestern U.S.; however, tests of DWM efficacy have only been performed over short time periods and at a limited number of sites. To fill this gap, the RZWQM-DSSAT hybrid model, previously evaluated for a subsurface-drained agricultural system in Iowa, was used to simulate both conventional drainage (CVD) and DWM over 25 years of historical weather at 48 locations across the midwestern U.S. Model simulations were used to demonstrate how variability in both climate and management practices across the region affects the ability of DWM to reduce losses of NO3 in subsurface drainage. The regional average simulated reduction in drain flow was 151 mm yr-1 when using DWM instead of CVD, and the regional percent reduction over the long term was 53%. Reductions in drain flow were offset mainly by increases in surface runoff and evapotranspiration. Similarly for nitrogen (N), the regional average simulated reduction in NO3 losses through subsurface drains was 18.9 kg N ha-1 yr-1, and the regional percent reduction over the long term was 51%. Subsurface drain NO3 loss reductions were counterbalanced mainly by increases in stored soil N, denitrification, and plant N uptake. The simulations suggest that if DWM can be practically implemented throughout the region, particularly in the southern states, then substantial reductions in the amount of NO3 entering surface waters from agricultural systems can be expected.

Field-Based High-Throughput Plant Phenotyping Reveals the Temporal Patterns of Quantitative Trait Loci Associated with Stress-Responsive Traits in Cotton

The application of high-throughput plant phenotyping (HTPP) to continuously study plant populations under relevant growing conditions creates the possibility to more efficiently dissect the genetic basis of dynamic adaptive traits. Toward this end, we employed a field-based HTPP system that deployed sets of sensors to simultaneously measure canopy temperature, reflectance, and height on a cotton (Gossypium hirsutum L.) recombinant inbred line mapping population. The evaluation trials were conducted under well-watered and water-limited conditions in a replicated field experiment at a hot, arid location in central Arizona, with trait measurements taken at different times on multiple days across 2010–2012. Canopy temperature, normalized difference vegetation index (NDVI), height, and leaf area index (LAI) displayed moderate-to-high broad-sense heritabilities, as well as varied interactions among genotypes with water regime and time of day. Distinct temporal patterns of quantitative trait loci (QTL) expression were mostly observed for canopy temperature and NDVI, and varied across plant developmental stages. In addition, the strength of correlation between HTPP canopy traits and agronomic traits, such as lint yield, displayed a time-dependent relationship. We also found that the genomic position of some QTL controlling HTPP canopy traits were shared with those of QTL identified for agronomic and physiological traits. This work demonstrates the novel use of a field-based HTPP system to study the genetic basis of stress-adaptive traits in cotton, and these results have the potential to facilitate the development of stress-resilient cotton cultivars.

DRAINMOD-N II: evaluated for an agricultural system in Iowa and compared to RZWQM-DSSAT.

A new simulation model for N dynamics, DRAINMOD-N II, has been previously evaluated for only a few sites. We evaluated the model using ten years (1996-2005) of measured data from a subsurface-drained, corn-soybean agricultural system near Story City, Iowa. Nitrogen fertilizer was applied to plots at low, medium, and high rates (57 to 67 kg N ha-1, 114 to 135 kg N ha-1, and 172 to 202 kg N ha-1, respectively) during corn years, and nitrate (NO3) losses from subsurface drains under each plot were monitored biweekly for ten years. Average annual simulated and measured NO3 losses in drainage water were 21.9 and 20.1 kg N ha-1 for the low N rate, 26.6 and 26.5 kg N ha-1 for the medium N rate, and 36.6 and 37.0 kg N ha-1 for the high N rate, respectively. The model efficiency statistics for DRAINMOD-N II simulations of annual subsurface drain NO3 losses were 0.89, 0.95, and 0.94 for the low, medium, and high N rates, respectively. For the same experimental dataset, a comparison of DRAINMOD-N II simulations to that of another model that simulates hydrologic and N dynamics of agricultural systems, the RZWQM-DSSAT hybrid model, demonstrated that the two models were most different in their simulations of soybean N fixation, plant N uptake, and net N mineralization. Future field investigations should focus on generating better understandings of these processes. The results suggest that DRAINMOD-N II can reasonably simulate the effects of different corn-year N rates on losses of NO3 through subsurface drainage lines and that simulations of subsurface drainage NO3 losses by DRAINMOD-N II are comparable to that of RZWQM-DSSAT.

Winter Cover Crop Effects on Nitrate Leaching in Subsurface Drainage as Simulated by RZWQM-DSSAT

Planting winter cover crops such as winter rye (Secale cereale L.) after corn and soybean harvest is one of the more promising practices to reduce nitrate loss to streams from tile drainage systems without negatively affecting production. Because availability of replicated tile-drained field data is limited and because use of cover crops to reduce nitrate loss has only been tested over a few years with limited environmental and management conditions, estimating the impacts of cover crops under the range of expected conditions is difficult. If properly tested against observed data, models can objectively estimate the relative effects of different weather conditions and agronomic practices (e.g., various N fertilizer application rates in conjunction with winter cover crops). In this study, an optimized winter wheat cover crop growth component was integrated into the calibrated RZWQM-DSSAT hybrid model, and then we compared the observed and simulated effects of a winter cover crop on nitrate leaching losses in subsurface drainage water for a corn-soybean rotation with N fertilizer application rates over 225 kg N ha-1 in corn years. Annual observed and simulated flow-weighted average nitrate concentration (FWANC) in drainage from 2002 to 2005 for the cover crop treatments (CC) were 8.7 and 9.3 mg L-1 compared to 21.3 and 18.2 mg L-1 for no cover crop (CON). The resulting observed and simulated FWANC reductions due to CC were 59% and 49%. Simulations with the optimized model at various N fertilizer rates resulted in average annual drainage N loss differences between CC and CON increasing exponentially from 12 to 34 kg N ha-1 for rates of 11 to 261 kg N ha-1, but the percent difference remained relatively constant (65% to 70%). The results suggest that RZWQM-DSSAT is a promising tool to estimate the relative effects of a winter crop under different conditions on nitrate loss in tile drains, and that a winter cover crop can effectively reduce nitrate losses over a range of N fertilizer levels.

SIMULATING LONG-TERM IMPACTS OF WINTER RYE COVER CROP ON HYDROLOGIC CYCLING AND NITROGEN DYNAMICS FOR A CORN-SOYBEAN CROP SYSTEM

Planting winter cover crops into corn-soybean rotations is a potential approach for reducing subsurface drainage and nitrate-nitrogen (NO3-N) loss. However, the long-term impact of this practice needs investigation. We evaluated the RZWQM2 model against comprehensive field data (2005-2009) in Iowa and used this model to study the long-term (1970-2009) hydrologic and nitrogen cycling effects of a winter cover crop within a corn-soybean rotation. The calibrated RZWQM2 model satisfactorily simulated crop yield, biomass, and N uptake with percent error (PE) within ±15% and relative root mean square error (RRMSE) 0.50, ratio of RMSE to standard error (RSR) <0.70, and percent bias (PBIAS) within ±25% except for the overestimation of annual drainage and NO3-N in CTRL2. The simulation in soil water storage was unsatisfactory but comparable to other studies. Long-term simulations showed that adding rye as a winter cover crop reduced annual subsurface drainage and NO3-N loss by 11% (2.9 cm) and 22% (11.8 kg N ha-1), respectively, and increased annual ET by 5% (2.9 cm). Results suggest that introducing winter rye cover crops to corn-soybean rotations is a promising approach to reduce N loss from subsurface drained agricultural systems. However, simulated N immobilization under the winter cover crop was not increased, which is inconsistent with a lysimeter study previously reported in the literature. Therefore, further research is needed to refine the simulation of immobilization in cover crop systems using RZWQM2 under a wider range of weather conditions.

USING CROSS-VALIDATION TO EVALUATE CERES-MAIZE YIELD SIMULATIONS WITHIN A DECISION SUPPORT SYSTEM FOR PRECISION AGRICULTURE

Crop growth models have recently been implemented to study precision agriculture questions within the framework of a decision support system (DSS) that automates simulations across management zones. Model calibration in each zone has occurred by automatically optimizing select model parameters to minimize error between measured and simulated yield over multiple growing seasons. However, to date, there have been no efforts to evaluate model simulations within the DSS. In this work, a model evaluation procedure based on leave-one-out cross-validation was developed to explore several issues associated with the implementation of CERES-Maize within the DSS. Five growing seasons of measured yield data from a central Iowa cornfield were available for cross-validation. Two strategies were used to divide the study area into management zones, one based on soil type and the other based on topography. The decision support system was then used to carry out the model calibration and validation simulations as required to complete the cross-validation procedure. Results demonstrated that the models ability to simulate corn yield improved as more growing seasons were used in the cross-validation. For management zones based on topography, the average root mean squared error of prediction (RMSEP) from cross-validations was 1460 kg ha-1 when two growing seasons were used and 998 kg ha-1 when five years were used. Model performance was shown to vary spatially based on soil type and topography. Average RMSEP was 1651 kg ha-1 on zones of Nicollet loam, while it was 496 kg ha-1 on zones of Canisteo silty clay loam. Spatial patterns also existed between areas of higher RMSEP and areas where measured spatial yield variability was related to topography. Changes in the mean and variance of optimum parameter sets as more growing seasons were used in cross-validation demonstrated that the optimizer was able to arrive at more stable solutions in some zones as compared to others. Results suggested that cross-validation was an appropriate method for addressing several issues associated with the use of crop growth models within a DSS for precision agriculture.

Proximal hyperspectral sensing and data analysis approaches for field-based plant phenomics

Field-based proximal hyperspectral data was collected for cotton phenotyping.3.68 billion PROSAIL runs on a supercomputer were used for model inversion.Partial least squares regression models best estimated four cotton phenotypes.High-throughput capability will improve hyperspectral methods for phenomics. Field-based plant phenomics requires robust crop sensing platforms and data analysis tools to successfully identify cultivars that exhibit phenotypes with high agronomic and economic importance. Such efforts will lead to genetic improvements that maintain high crop yield with concomitant tolerance to environmental stresses. The objectives of this study were to investigate proximal hyperspectral sensing with a field spectroradiometer and to compare data analysis approaches for estimating four cotton phenotypes: leaf water content ( C w ), specific leaf mass ( C m ), leaf chlorophyll a + b content ( C ab ), and leaf area index (LAI). Field studies tested 25 Pima cotton cultivars grown under well-watered and water-limited conditions in central Arizona from 2010 to 2012. Several vegetation indices, including the normalized difference vegetation index (NDVI), the normalized difference water index (NDWI), and the physiological (or photochemical) reflectance index (PRI) were compared with partial least squares regression (PLSR) approaches to estimate the four phenotypes. Additionally, inversion of the PROSAIL plant canopy reflectance model was investigated to estimate phenotypes based on 3.68 billion PROSAIL simulations on a supercomputer. Phenotypic estimates from each approach were compared with field measurements, and hierarchical linear mixed modeling was used to identify differences in the estimates among the cultivars and water levels. The PLSR approach performed best and estimated C w , C m , C ab , and LAI with root mean squared errors (RMSEs) between measured and modeled values of 6.8%, 10.9%, 13.1%, and 18.5%, respectively. Using linear regression with the vegetation indices, no index estimated C w , C m , C ab , and LAI with RMSEs better than 9.6%, 16.9%, 14.2%, and 28.8%, respectively. PROSAIL model inversion could estimate C ab and LAI with RMSEs of about 16% and 29%, depending on the objective function. However, the RMSEs for C w and C m from PROSAIL model inversion were greater than 30%. Compared to PLSR, advantages to the physically-based PROSAIL model include its ability to simulate the canopys bidirectional reflectance distribution function (BRDF) and to estimate phenotypes from canopy spectral reflectance without a training data set. All proximal hyperspectral approaches were able to identify differences in phenotypic estimates among the cultivars and irrigation regimes tested during the field studies. Improvements to these proximal hyperspectral sensing approaches could be realized with a high-throughput phenotyping platform able to rapidly collect canopy spectral reflectance data from multiple view angles.

A model-independent open-source geospatial tool for managing point-based environmental model simulations at multiple spatial locations

A novel geospatial tool box named Geospatial Simulation (GeoSim) has been developed, which can be used to manage point-based model simulations at multiple locations using geospatial data within a geographic information system (GIS). The objectives of this paper were to describe GeoSim and demonstrate its use. GeoSim has been developed as a plug-in for Quantum GIS, and both of these software programs are open-source and freely available. An important feature of GeoSim is its model-independent nature, meaning any point-based simulation model that uses ASCII files for input and output can be managed spatially. GeoSim facilitates the transfer of geospatial data from the GIS database to the model input files and from the model output files back to the GIS database. GeoSim presently includes six software tools, each with a graphical user interface. A case study demonstrates the use of GeoSim for processing geospatial data layers at a field site, conducting spatial model simulations, and optimizing model parameters for site-specific conditions. Two cropping system models, AquaCrop and the DSSAT Cropping System Model, were implemented to simulate seed cotton yield in response to irrigation management, nitrogen management, and soil texture variability for a 14 ha study area near Lamesa, Texas. Geoprocessing tools within GeoSim were able to summarize 5592 data points within 405 polygon features in 3.8 s. Simulation tools were able to swap 33,316 and 44,550 parameters values to complete 405 spatial simulations with the AquaCrop and DSSAT models in 112 s and 398 s, respectively. These results demonstrate the utility of GeoSim for summarizing large geospatial data sets and transferring the data to the file formats of multiple models. Simulation duration was increased as compared to stand-alone model simulations without parameter swapping, which may be problematic for applications requiring large numbers of simulations. The flexible design of GeoSim is intended to support spatial modeling exercises for a variety of models and environmental applications.