web-header_Peterson-test-plot.png

What is Replicated Testing?

At Peterson Farms Seed, testing isn't just a talking point - it's the way we do business. We simply believe in testing - and lots of it.

Our testing program consists of a network of replicated testing plots, as well as several small variety increases and hybrid and inbred evaluations. Our research is administered by our own research manager - as opposed to relying on data from strip trials and hear-say. We utilize multiple years of data in multiple locations - the only way selections should be made.

We talk a lot about the importance of replicated testing. But, why do we think it is so important?

Replicated testing takes into account variation in fields and locations. Because conditions can be so variable, it is important to minimize variation to get accurate results. Using replicated testing in different testing locations helps to minimize land and environmental variation.

In our testing, each variety appears in the each plot 3 times (3 Reps) and averages are calculated to determine actual yield. The more replications and testing locations, the more accurate the results are.

In test plots, yields of certain varieties will be numerically greater than other varieties. However, they may not be statistically different. Statistical formulas are used in replicated trials to determine if there are true yield differences when comparing varieties against each other.

In non-replicated trials or "strip trials", varieties are usually entered only once. The location that a variety is placed within a strip trial can give a huge advantage or disadvantage. That is why it is extremely difficult to find the best variety utilizing strip plot data. Making planting decisions utilizing strip plot data may be worse than using no data at all.

The example below is a strip plot we conducted in 2010. There were 5 side-by-side 760 foot-long strips of 56J86VT3. The results showed a 58.6 bu/acre difference in yield and a 2.5% difference in moisture from the high to the low. Since there was no genetic variation, there must have been environmental variation.

The chart below on the right shows the chance of finding the "best" entry in a strip trial. In a simple plot with only two varieties, the chance of finding the best variety is a little more than a flip of the coin and the odds drastically go down as you add more entries.

2010 Strip Plot

CompanyHybrid% Harvest MoistureTest WeightYield/Acre
PFS 56J86 16.8 54.7 222.2
PFS 56J86 18.2 53.8 175.3
PFS 56J86 18.8 54.2 163.6
PFS 56J86 19.3 54.7 184.6
PFS 56J86 18.2 54.8 172.7

 

Chance of Finding the "Best"

# VarietiesChance of Finding the "Best"
2 52%
3 42%
4 33%
10 13%
20 7%
30 6%

 

How to Find a Winning Variety

YearSourceLocationPFS 11R08PFS 0509Difference
2010 PFS Harwood H 35.2 7.8 27.3
2010 PFS Harwood L 52.8 47.4 5.4
2010 PFS Kindred 22.9 14.5 8.4
2010 PFS Mayville 43.1 38.9 4.2
2010 PFS Oakes 29.8 39.9 -10.1
2010 PFS Valley City 47.3 48.8 -1.5
2010 PFS Wyndmere 50.7 53.6 -2.9
2010 PFS Groton 58.3 45.3 13
2010 PFS Harwood L 50.9 48.9 2
2010 PFS Kindred 45.8 29.1 16.6
2010 PFS Morris 55.3 51.1 4.3
2010 PFS Oakes 34.7 38.6 -3.9
2010 PFS Olivia 35.6 50 -14.4
2010 PFS Wheaton 51.4 57.9 -6.5
2010 PFS Wyndmere 46.7 49.6 -2.9
2010 IND Harwood H 35.2 29 6.2
2010 IND Harwood L 50.7 45.4 5.3
2010 IND Harwood L 43.4 47.7 -4.2
2010 IND Kindred 28.7 17.6 11.1
2010 IND Mayville 40.8 37.6 3.2
2010 IND Oakes 40.3 41.2 -0.9
2010 IND Valley City 51.8 47 4.8
2010 IND Wyndmere 52.4 48 4.4
2011 PFS Harwood H 14.9 14 0.9
2011 PFS Kindred 30.3 25 5.3
2011 PFS Mayville 45.3 41.4 3.9
2011 PFS Perley 40.2 35.7 4.5
2011 PFS Valley City 40.1 35.1 5.1
2011 PFS Verona 39.4 37.3 2
2011 PFS Breckenridge 67.4 60.7 6.8
2011 PFS Harwood L 55.3 53.8 1.5
2011 PFS Kindred 39 23.9 15.1
2011 PFS Verona 37.9 35.7 2.3
2011 PFS Wendell 56.5 49.4 7.1
2011 PFS Wyndmere 30.8 37.6 -6.8
2011 PFS Harwood H 21.1 21.5 -0.5
2011 PFS Harwood L 58.1 59.4 -1.3
2011 IND Groton 58.4 44.6 13.8
2011 IND Morris 31.7 13.2 18.5
2011 IND Olivia 58.4 51 7.4
2011 IND Wheaton 69.2 56.8 12.4 
      43.8 39.8 4.1
          28/40 70%

When a variety consistently outperforms another in replicated testing over multiple years and over a wide variety of testing locations, you can be fairly certain that it is the correct variety to select. In this case, 11R08 RR2Y did not “win” every plot every year. Statistically, 11R08 RR2Y out yielded 0509 RR by 4.1 bushels over two years, winning 70% of the time over the 40 plots.