Accessing and Interpreting Yield Data

• Yield results can be used to compare products in similar growing environments and can be used to help growers make selection decisions for the next season.
• When comparing yield data, consider multi-year and multi-location data, multiple growing scenarios, and seek head-to-head comparisons.
• Yield data can be presented differently based on the data source. When examining university or independent trials, be sure to look at the LSD and understand its meaning.

When selecting seed products for next season, understanding the potential productivity of a product for your situation may involve using yield data from several sources. In addition to Channel® harvest data, data from several reliable, independent yield trials, such as university sponsored trials, can also be used as supporting information for your decisions. Also, results from yield trials can be used to compare seed products in similar growing environments. All farmers should understand that no trial can ever make exact predictions and that the best insight into a product’s potential performance involves data across multiple geographies and environments.

To access Channel yield data, you may begin by looking online at www.channel.com for some of the data highlighted for your area. However, in order to get the best picture of how specific products are performing in your area, contact your Channel Seedsman to access more in-depth reports and learn how specific products performed. As you view Channel product summary reports, it is helpful to understand what you are viewing. The summary reports consist of data from Channel and third party head-to-head comparisons of Channel products to competitor products. The report shows the products, applicable traits, and maturity group used in each head-to-head comparison. Also, yield values (adjusted to 15% moisture content for corn; 13% moisture content for soybean), harvest moisture content, and gross income for each product is included. For each product report, there is also a summary presented that outlines the averages across competitor product comparisons.

In addition to Channel data, you may also consider looking at other independent and university performance trials to determine how certain products performed in a range of different locations and scenarios. These tests are generally conducted by maturity range with seed from a broad group of sources and tested across a variety of growing conditions within a state or region. Performance data for each test within a region is usually reported in tables that list yield in descending order. Comparisons between products should only be made within a table. Comparing two products from different tables (i.e. different growing conditions and/or production levels) would generally not be useful.

Whether or not products will perform similarly to trial results is dependent upon having growing conditions similar to the trial conditions. Use of multi-location and multi-year data is a more predictive source of information for making selection decisions. It is best to review data from trials that use randomization, replication, and multiple locations within a region.

Multi-Year Data. When possible, consider selecting products that are proven to be high-yielding for two or more consecutive years. Single year data has a lower predictive probability.

Multi-Location Data. In addition to multi-year data, another factor to consider for performance consistency is geography. Some products may provide great yield in one location and yield poorly in another. Numerous variables such as weather, fertility, and insect pressure can affect product performance across locations. Consider products that have a stable performance over a range of environmental conditions. If there is inconsistency across locations, predictive ability of the data is compromised. The best data are those that show consistency in performance over years and across locations.

Multiple Scenarios. There are factors aside from the geography and growing season that can influence a product’s performance. Other items to look at include field management history such as planting timeframe, crop rotation, tillage type, and how insects and weeds were controlled. If traits were present in the seed, consider how they contributed to the performance.

Statistical Difference. Most university and independent yield data that you access includes a measure of experimental variability, which basically shows what amount of yield variation can be attributed to the product itself versus influence from outside factors. Two common statistics are Least Significant Difference (LSD) and Coefficient of Variation (CV). An example using LSD and CV values to compare products is given in Figure 1.

LSD represents in bu/acre the amount of yield variation that can be attributed to variations in soil types, soil fertility, moisture availability, insect infestation, diseases, and differences due to planting and harvesting techniques. Yield differences greater than the LSD can be attributed to actual differences in genetic yield potential of products. Yield differences less than the LSD are not considered statistically different and are likely due to other factors such as those mentioned above.

Another measurement used to determine statistical differences is the coefficient of variation (CV). It expresses the extent of variability in relation to the mean (average) of the plot data. In order to determine the CV, you must know the standard deviation, which is a measure of how “spread out” the data numbers are, and the mean, which is the average of all the numbers in the data set. CV can be calculated as a ratio of the standard deviation to the mean. A higher CV indicates that there is greater experimental variability, which results in the yield data having a lower predictive value. A low CV generally results from a more uniform plot location. A CV of 15% or less is desirable for field test results and the closer it is to zero, the higher the data quality.

 ¹ Rouse, J. 2014. Iowa crop performance tests corn 2014. PM 660. Iowa State University. www.croptesting.iastate.edu

² DeVillez, P. and Foster W.D. 2014 Purdue corn & soybean performance trials. Department of Agronomy Purdue University. http://ag.purdue.edu/agry/

³ Mullen, R. and Florence D.C. 2010. Agricultural statistics, least significant differences (LSD). CORN Newsletter 2010-40. The Ohio State University Extension. http://corn.osu.edu/newsletters/2010/2010-40/agricultural-statistics-least-significant-differences-lsd

⁴ Nafziger, E. 2012. Variety trials and choosing seed for 2013. University of Illinois. http://cropsci.illinois.edu/news/variety-trials-and-choosing-seed-2013

⁵ Bowman, D.T. 2001. Common use of the CV: A statistical aberration in crop performance trials. Journal of Cotton Science 5:137-141. http://journal.cotton.org/ Web sources verified 08/17/15.