The planting season is starting to sneak up on us already, and it is now less than 2 months away for crops like spring wheat. Let me assume that you still have not selected or made 100% of your seed purchases for all your crops or that your intended acres for each crop may change due to weather conditions. I think most farmers will agree that they want to buy the variety that returns the highest profit per-acre.

Most people justify a buying decision after they have already made it based on emotion. Unfortunately, seed purchases often do not escape this blight. Things that influence our selection and purchase of varieties include brand reputation, loyalty and tradition, friends and family members, advertising, and company representatives.

It seems obvious, but I always encourage growers to utilize yield data in this variety selection process. Yield data can be collected from side-by-side comparisons on the farm, company variety trials, and third-party variety trials. Reliability of this yield data is not equal. I am not talking about who (farmer, company, university) does the work, but the methodology.

There are three key methods we can use to increase our confidence that one variety does truly outperform another:

  1. Blocking (splitting the test plot into similar environments)
  2. Randomization (random placement of varieties within the test plot)
  3. Replication (same variety appears in the test plot several times)

Another key term I need to describe is called experimental error, which is simply variation in yield measured in the same variety that was tested independently several times within a test plot. The source of this variation can be soil difference in the test plot or even inability to reproduce the exact same conditions with equipment operations and measurements (easier to control than the former).

Let me explain why these three methods (blocking, randomization, and replication) are important when making yield data comparisons among varieties. For example, Figure 1. below may be a 2011 yield monitor data from a portion of the field where a test plot will be located in 2013 (it was all the same variety).

Figure 1. Yield monitor data from one variety planted in 2011.

Crop variety selection: Eliminate emotion, increase profitability

What happens if we planted four varieties so the rows ran left to right this spring? Would that be a fair comparison based on what we know from 2011?  No. However, blocking or splitting this area into four similar environments helps us reduce the amount of experimental error by comparing the four varieties within smaller areas with more similar conditions (Figure 2).

Figure 2. Blocking or splitting the area into four similar environments.

Crop variety selection: Eliminate emotion, increase profitability

Another component of this process is replication or having the variety appear once in each of the four blocks at random within each block (Figure 3.). Replication is a means of estimating the amount of variation in experimental error within the same variety while randomization provides a valid estimation of that experimental error variance. This is a lot of words and just means that replication and randomization are needed to get reliable data.

Figure 3. Randomization of four varieties (Var.) within each block.

Crop variety selection: Eliminate emotion, increase profitability

The use of blocking, replication, and randomization helps managers of crop testing performance trials like the ones we conduct at South Dakota State University (SDSU) determine whether varieties perform differently at a location, the margin of difference between varieties, and confidence that the differences measured are attributed to the variety, not experimental error.

The differences between varieties for yield and other important agronomic traits are posted annually for all major crops in South Dakota on the iGrow Variety Trial Results pages. These results provide an opportunity for growers to make informed purchasing decisions when seed is ordered for their farm. The magnitude of performance differences between varieties or hybrids can be significant enough to create a $250 profit per-acre swing. For example, within a company the cost difference per-bag (140,000 seeds) of soybeans is only about $9, but the yield differences may be 7 bushels-per-acre between varieties. Farmers generally have not even finished harvest before the earliest season seed discounts are available. Waiting until yield data is available to aid decision making on variety selection will pay off. Also, written descriptions for one variety may sound better than another, but the yield may not be.

So before you buy any more seed, please consider more data-driven decisions that increase your farm's profitability by using crop yield results determined by using blocking, randomization, and replication.

Source: Nathan Mueller