Identifying Areas Within Peanut Fields Which Are at High Risk for Aflatoxin Contamination

Identify areas within peanut fields which are at high risk for aflatoxin contamination

Rational and Economic Significance:
Aflatoxin contamination of peanut occurs when the common fungi Aspergillus flavus and Aspergillus parasiticus infect peanut kernels under drought stress prior to harvest. There is general acceptance that aflatoxin contamination is not uniformly distributed within any given peanut field. It is also accepted that areas within a field which are prone to drought stress are at higher risk for aflatoxin contamination in dry years. Despite this knowledge, there is no easy method for identifying and delineating these high risk areas prior to harvest. We propose to use remote sensing techniques to identify and delineate areas within peanut fields which may be at high risk for aflatoxin contamination.

Past and Ongoing Research:
Click to view larger image To understand the distribution of aflatoxin within a typical peanut field, we initiated a pilot study in a 30 acre rain-fed peanut field in Tift Co. during the 2006 growing season. We installed soil moisture and temperature sensors at 17 locations within the field and collected data during the growing season. Prior to harvest, we collected all the peanuts within a 325 ft2 area surrounding each of the 17 sensor locations and analyzed these peanuts for aflatoxin concentrations. Aflatoxin concentrations were high and ranged from 50 to 450 ppb but as is shown in the figure at right, there were three aflatoxin hot spots (>150 ppb) in the field with the largest hot-spot being in the southwest corner of the field and covering approximately 6 acres. During harvest, peanuts from the hot spots would be mixed with peanuts from the rest of the field jeopardizing the quality of entire wagon loads of peanuts. If such a map predicting the areas of the field likely to contain aflatoxin hot spots was available prior to harvest, the hot spots could be segregated and harvested separately.

During 2006, we also conducted a preliminary evaluation of techniques we can use for delineating high-risk areas. Prior to planting, we developed a soil electrical conductivity (EC) map of the field. We then used specialized software to develop 4 management zones with common soil EC values. We found an inverse relationship between soil EC and aflatoxin concentrations. The management zone with the lowest soil EC has the highest aflatoxin concentration and the management zone with the highest soil EC has the lowest aflatoxin concentration. Although soil EC may be a good predictor of high risk areas, it can not account for the effects of weather during a growing season. An extremely dry year may put at risk areas which have soils with relatively high moisture holding capacities. To identify all high risk areas requires a tool that can gauge the status of the growing crop as it approaches maturity. The tool which shows the greatest promise for achieving this goal is remote sensing.

Click to view larger image Images acquired from a variety of platforms including airborne, satellite and fixed mounting on the ground have been successfully used to identify drought stress areas in peanut plots and fields. In Australia, researchers have used vegetation indices created from airborne multispectral images to identify drought-stressed areas within fields. Used in conjunction with weather monitoring and modeling, they have been successful in identifying and predicting the risk and incidence of aflatoxin.

Vegetation indices, or VIs, created from multispectral images taken from cameras suspended over fields or from cameras in airplanes and satellites have been used successfully to identify drought stress areas in many crops, including peanuts. VIs are mathematical ratios of reflectance at specific wavelengths. In Tifton, Dr. Dana Sullivan has used a multispectral camera (captures reflectance at specific wavelengths) suspended over peanut plots to successfully assess the drought stress of peanuts. Dr. Sullivan used the multispecral images to calculate VIs of the growing peanuts. VIs are very good at measuring plant vigor, biomass, and canopy stress, and consequently can be used to delineate highly stressed areas from less stressed areas. She and Dr. Corley Holbrook also found that there is frequently a high correlation between plots identified by the VIs as drought-stressed and aflatoxin contamination in peanut pods from those plots.

Based on this earlier work, we have great confidence that we can successfully delineate stressed areas within peanut fields using multispectral images and vegetation indices. If we can extrapolate the methodology to airborne images so that we can cover large areas at low cost and fine-tune the procedures so that we can consistently identify areas within peanut fields which are at high risk for aflatoxin contamination, we will be providing a valuable tool to the peanut industry. Areas identified as high-risk can be segregated and harvested and processed separately until sampling verifies presence or absence of aflatoxin.

Planned Work:
Multispectral aerial images of three peanut fields will be taken at regular intervals during the growing season. Frequency will be increased to weekly intervals for the 40 days prior to harvest. The data from the images will be used to calculate several vegetation indices including the NDVI (Normalized Difference Vegetation Index) and the Green NDVI. Maps will be created from the calculated indices and used to delineate stressed areas in the fields as shown in the figure at right. The figure was created from a thermal infra red image taken of the field discussed on the previous page during Sep 2005 while the field was in cotton. Red indicates high drought stress areas while blue indicates low stress areas. During the project, stressed areas will be ground-truthed to identify the cause of the stress. Prior to harvest, peanut kernel samples within the stressed areas and selected unstressed areas will be collected and analyzed for the presence of aflatoxin using established analytical techniques. Statistical techniques, including multivariate analysis, will be used to evaluate the strength of our delineation technique.

Project Leader: George Vellidis
Contact Info:
Affiliation: University of Georgia
P.O. Box 748
Tifton, GA 31793
(229) 386-3170