2 results
Site-to-site and year-to-year variation in Triticum aestivum–Aegilops cylindrica interference relationships
- Marie Jasieniuk, Bruce D. Maxwell, Randy L. Anderson, John O. Evans, Drew J. Lyon, Stephen D. Miller, Don W. Morishita, Alex G. Ogg, Jr., Steven Seefeldt, Phillip W. Stahlman, Francis E. Northam, Philip Westra, Zewdu Kebede, Gail A. Wicks
-
- Journal:
- Weed Science / Volume 47 / Issue 5 / October 1999
- Published online by Cambridge University Press:
- 12 June 2017, pp. 529-537
-
- Article
- Export citation
-
Crop yield loss–weed density relationships critically influence calculation of economic thresholds and the resulting management recommendations made by a bioeconomic model. To examine site-to-site and year-to-year variation in winter Triticum aestivum L. (winter wheat)–Aegilops cylindrica Host. (jointed goatgrass) interference relationships, the rectangular hyperbolic yield loss function was fit to data sets from multiyear field experiments conducted at Colorado, Idaho, Kansas, Montana, Nebraska, Utah, Washington, and Wyoming. The model was fit to three measures of A. cylindrica density: fall seedling, spring seedling, and reproductive tiller densities. Two parameters: i, the slope of the yield loss curve as A. cylindrica density approaches zero, and a, the maximum percentage yield loss as A. cylindrica density becomes very large, were estimated for each data set using nonlinear regression. Fit of the model to the data was better using spring seedling densities than fall seedling densities, but it was similar for spring seedling and reproductive tiller densities based on the residual mean square (RMS) values. Yield loss functions were less variable among years within a site than among sites for all measures of weed density. For the one site where year-to-year variation was observed (Archer, WY), parameter a varied significantly among years, but parameter i did not. Yield loss functions differed significantly among sites for 7 of 10 comparisons. Site-to-site statistical differences were generally due to variation in estimates of parameter i. Site-to-site and year-to-year variation in winter T. aestivum–A. cylindrica yield loss parameter estimates indicated that management recommendations made by a bioeconomic model cannot be based on a single yield loss function with the same parameter values for the winter T. aestivum-producing region. The predictive ability of a bioeconomic model is likely to be improved when yield loss functions incorporating time of emergence and crop density are built into the model's structure.
Evaluation of models predicting winter wheat yield as a function of winter wheat and jointed goatgrass densities
- Marie Jasieniuk, Bruce D. Maxwell, Randy L. Anderson, John O. Evans, Drew J. Lyon, Stephen D. Miller, Don W. Morishita, Alex G. Ogg, Jr., Steven S. Seefeldt, Phillip W. Stahlman, Francis E. Northam, Philip Westra, Zewdu Kebede, Gail A. Wicks
-
- Journal:
- Weed Science / Volume 49 / Issue 1 / February 2001
- Published online by Cambridge University Press:
- 20 January 2017, pp. 48-60
-
- Article
- Export citation
-
Three models that empirically predict crop yield from crop and weed density were evaluated for their fit to 30 data sets from multistate, multiyear winter wheat–jointed goatgrass interference experiments. The purpose of the evaluation was to identify which model would generally perform best for the prediction of yield (damage function) in a bioeconomic model and which model would best fulfill criteria for hypothesis testing with limited amounts of data. Seven criteria were used to assess the fit of the models to the data. Overall, Model 2 provided the best statistical description of the data. Model 2 regressions were most often statistically significant, as indicated by approximate F tests, explained the largest proportion of total variation about the mean, gave the smallest residual sum of squares, and returned residuals with random distribution more often than Models 1 and 3. Model 2 performed less well based on the remaining criteria. Model 3 outperformed Models 1 and 2 in the number of parameters estimated that were statistically significant. Model 1 outperformed Models 2 and 3 in the proportion of regressions that converged on a solution and more readily exhibited an asymptotic relationship between winter wheat yield and both winter wheat and jointed goatgrass density under the constraint of limited data. In contrast, Model 2 exhibited a relatively linear relationship between yield and crop density and little effect of increasing jointed goatgrass density on yield, thus overpredicting yield at high weed densities when data were scarce. Model 2 had statistical properties that made it superior for hypothesis testing; however, Model 1's properties were determined superior for the damage function in the winter wheat–jointed goatgrass bioeconomic model because it was less likely to cause bias in yield predictions based on data sets of minimum size.