Subscribe and stay up-to-date with the latest news and great offers from DEKALB, Asgrow and Deltapine.
Don't miss out on the latest agronomic news.
Local agronomic alerts.Delivered straight to your inbox.
A successful on-farm research trial begins with thorough designing and planning before planting, which includes:
1. Field Topography and Planter. If a yield monitor will be used to map results, the plot should be planted perpendicular to changes in soil type, slope, and other field conditions so that yield can be compared as the products transition from one environment to another. If the right equipment is not available or you are more interested in the general differences, fields with variable soil types, slopes, irregular boundaries, and tile lines running parallel with the rows should be avoided. Longer, field-length strips are preferred to reduce variability in the test. Border rows should be included on each side of the plot to avoid edge effects. The split-planter comparison method works well for a side-by-side plot.
A split-planter comparison can be easily established by randomly placing a different seed product in each half of the planter. Care should be taken to ensure that seeding rates and planter adjustments are appropriate to achieve the same stand for each product on both sides of the planter. In addition, the planter box should be thoroughly cleaned before loading the next crop product.
2. Product Selections. When selecting corn or soybean products, try to keep the comparison within plus or minus two days relative maturity (RM) for corn and 0.2 RM for soybean.
3. Harvest Equipment. For corn, the width of the combine harvest header should be one-half the planter width. For soybean, plant strips twice as wide as the soybean header so that the center of the plots can be harvested.
4. Randomization and Replications. The key to a successful an on-farm test is that it must be repeated in unbiased side-by-side comparisons of the products in question. Randomizing and replicating are important for laying out a scientifically valid plot. This is what separates a demonstration plot from one that can be used to make valid recommendations/conclusions.
To adequately compensate for field variations, each pair of the compared products should be replicated at least four to six times, although the latter is better. In addition, treatments should be randomly located within the pair. Where applicable, the appropriate refuges should be planted as outlined in the IRM guide.
Throughout The Growing Season. Plots should be monitored during the growing season and all crop inputs should be recorded. Notes taken during the growing season can be very useful during data analysis to provide insight on variances that may be present. Some important observations that should be recorded include:
Before Harvest. As the season progresses, it is important to walk a few strip plots to observe corn/soybean product differences:
For corn, ear size uniformity should be checked for length and girth; if it is big and girthy or long and slender. Since kernel number per row is controlled by the environment, a corn product producing a large number of kernels per row under dry conditions is an indication that this product has the ability to tolerate stress conditions. A simple push test should be performed to evaluate stalk quality in both products. Any root lodging should be noted as well.
For soybean, evaluate the standability of each product and note any significant differences. Also, evaluate each product for any late season diseases that may have affected one product more adversely than another. Look for seed shattering and green stems while scouting.
Harvest. All equipment used in harvest data collection should be accurate and well calibrated:
Plot yield mapping can provide even more information as products may differ within strips due to soil type, topography, or past management practices.
Many on-farm comparisons are required to determine with confidence that a statistical difference exists between two products. The more data replicated and analyzed, the better the confidence level in picking a winning product.
In addition, comparing data with results from similar trials in other locations can be beneficial to help substantiate the data collected. The best conclusions are usually drawn from trial results that are conducted in more than one location and include more than one year.
When comparing two crop products in a yield trial, statistical analysis will help determine if differences are due to product superiority or field variability such as weather conditions, insect pressure, soil conditions, and/or management.
Statistical t-test. Plot yields in a side-by-side trial are compared by using a paired t-test. The test is used to show if the yields of two sets of treatments (i.e. corn products) are significantly different. Yield differences between the two treatments are used to compute a t-value, which is declared significant only if it is greater than a critical value at the 5% probability level. The critical value is the minimum expected difference between the two products, and the probability level indicates a confidence of 95% that the observed difference is not happening by chance. In other words, we are 95% confident that the observed difference occurred due to product superiority. For example, product A yielded 100 bu/acre and product B yielded 60 bu/acre; if the calculated t-value was higher than the critical value then product B is statistically different than product A and the experimental treatment (i.e. corn product selection) has an effect on the yield.
Computed t-values and probabilities are sometimes listed at the bottom of a yield table or in the plot results. Yield differences that are not statistically significant indicate that the differences are more likely due to experimental variations in the field, rather than differences between products.
Keep in mind that the performance of any seed product is the result of the combined effects of genetics and respective environment. No product can win every yield plot or test. Industry-leading products have a head-to-head winning percentage of 60 to 65%, over many locations.