Jason Lems

Document Type

Thesis - University Access Only

Award Date

1998

Degree Name

Master of Science (MS)

Department / School

Plant Science

Abstract

Smaller profit margins and increased environmental concern are motivating farmers to reduce broadcast herbicide treatments. Increases in profits and lowered environmental risks can be achieved by reallocating herbicide inputs only to areas where weeds exceed economic thresholds. With advances in precision farming technology, site specific or variable rate herbicide applications are now becoming a feasible option in production agriculture. The objectives of this thesis is to describe the spatial variability of weeds on a field-wide scale using geostatistics and to evaluate GWM (General Weed Management Model) a bioeconomic crop/weed competition model. Two 65 ha fields in eastern South Dakota were grid sampled for weed seedlings in the springs of 1995 and 1996. These fields were grid sampled on a 15- by 30- m and 30- by 30- m grids, respectively. All weed seedlings at each grid point were enumerated in a 10- by 50-cm (0.1 m2) quadrant. Skewness, kurtosis, semivariance, and semivariograms were calculated using GEO-EAS. Kriged maps were generated using Surfer: 6.0 (Golden Software, Inc.; Golden CO). The research showed that weeds were clustered into small patches. The clustered distributions resulted in distributions that were highly skewed toward smaller numbers. The research showed that weeds were clustered into small patches. The clustered distributions resulted in distributions that were highly skewed toward smaller numbers with a heavy tail. Semivariogram analysis revealed that weed populations sampled on a 15- by 30- m grid were spatial dependent. However, weed populations sampled on a 30- by 30- m grid did not show spatial dependency. These results suggest weed density maps on a field-scale can be produced by kriging if the grid sampling scale is small enough to account for within field variation. GWM was evaluated in two 65 ha fields in eastern South Dakota in 1996. Three sites within a no-till corn-soybean cropping practice and four sites within a conventional no-till continuous corn cropping regime were selected for study. Plot location sites were selected on landscape position and prior weed history. All plots were a RCBD with 5 treatments replicated 4 times. The treatments included an untreated control, a GWM recommended post emergence only, pre- plus post emergence, and pre-emergence only treatments, and the producer's treatment. GWM pre-emergence only and pre- plus post emergence treatments were generated from seed bank data while the GWM post emergence only treatment was generated from weed seedling counts. The producer's treatments used consisted of the herbicide program used by the producer for that particular field.

The GWM recommendations at the soybean trial performed similar to or better than the producer's treatment in all plots. GWM pre-emergence only treatments had higher net profits in areas with high densities of Ambrosia artemisiifolia and Setaria spp. while GWM post emergence only treatment had a higher net profit in an area with low density of Setaria 5pp. In the continuous corn field, the producer treatment had higher yields and net profits than all GWM treatments in areas with high and low densities of Setaria spp. In areas with high densities of Abutilon theophrasti and medium densities of Chenopodium album, all GWM treatments had higher yields and net profits than the producer's treatments. Kriged maps of the two most prevalent weeds revealed about 49 and 13 ha of Setaria spp. and Ambrosia artemisiifolia, respectively, were present. Calculated herbicide cost on a field scale, using a site specific application of the two most profitable GWM recommended treatments for this weed regime, showed a net return increase of about $4,500 for the entire field. Therefore, using model recommended weed treatments in conjunction with field-scale weed maps may increase profitability and obtain satisfactory weed control.

Library of Congress Subject Headings

Weeds -- Geographical distribution Weeds -- Control

Format

application/pdf

Number of Pages

118

Publisher

South Dakota State University

Download

Share

COinS