On-Farm Research Basics

If you’ve looked at your yield monitor read out and thought, “You know, I could use this to do some field research,” you’re on the right track. “Anyone can do on-farm research,” says Dan Anderson, on-farm research coordinator at the University of Illinois. “Yield monitors make it even easier.”

But while it may be easy to measure yield differences with the monitor, do these differences have anything to do with the question you’re researching? Maybe…maybe not. It all depends on whether your study is designed to yield valid results.

Advertisement

If you’re going to make input decisions based on the results — your results must be valid. That is, results that point to the input factors that really made a difference. And that takes some understanding of field research methods and statistics.

But don’t turn the page yet. While a full description of on-farm research takes more explanation than this article can provide, it’s not like trying to master brain surgery. There are practical books and accessible experts ready to help you.

MORE BY MATT HOPKINS

What we can do in this article is give you a general idea of what constitutes good on-farm research — and how to formulate an action plan for your own research efforts.

Top Articles
Integrating Data-Driven Solutions for Fine-Tuned Climate Control in Indoor Farming

The effort will be worth it, says Rick Exner, farming systems coordinator for the Practical Farmers of Iowa and an extension associate at Iowa State University.

“While you don’t want to make serious management decisions based on one year’s worth of data, with your own on-farm research you’ll have information you can’t get from any other source.”

What Makes Good On-Farm Research

Joe Gednalske, manager of product development for Cenex/Land O’Lakes Agronomy, describes a basic on-farm “pseudo-research” dilemma: “Farmer used one herbicide on one field, used a different herbicide on another field,” Gednalske explains. “Got different yields. Was it the herbicide? Who knows? He got a difference in yield, but what other things were different? You just don’t know.”

So the key to good research is isolating what you’re testing from other factors that can affect the outcome. In a crop field, where conditions vary from foot to foot, that’s no simple task. But it’s possible.

“Even if you have a very uniform field and do everything exactly the same throughout the field, you’ll still get variation,” explains Anderson. “Like on a yield map — every 2 feet it’s a different yield. So if you see differences, you don’t know if that’s because of natural variations in the field or the treatments.”

There are ways to compensate for natural variations. If this is your first try at on-farm research, make it easy on yourself, and look for a fairly uniform field, suggets John Oolman, director of precision farming research for Agri-Growth Inc., Hollandale, Minn.

Look For Uniformity

“Pick out a field that doesn’t have a history of uneven manure application, doesn’t have any old field boundaries in the middle and doesn’t have large elevation changes,” Oolman says. “Don’t use the boundaries of the field — the headlands — because they’re not typical of the rest of the field, either.”

Look at a yield map or search your memory (and maybe the memories of your elders) to assure that the field is fairly uniform. “We’ve had some surprises,” he says. “You’ll find spots in a field that are real high in fertility and soil quality, and the 40-year-old farmers won’t know why. But Grandpa will say, ‘Happy Jack had an old hog farm there.’ You’ll still see high fertility leels around old homesteads and barns that have been gone for 50 years.”

To further factor out variation, do each treatment more than one time. “This is called replication,” Anderson says. “Replication is the key. There’s a mathematical way to account for the natural variations — the background noise — and separate that from the treatment effect, if there is one.

“For on-farm work, six replications of each plot generally provides the optimum results,” Anderson explains. “Any accuracy you gain with more than six reps is probably not worth the extra expense.”

Replication Filters Out Variations

According to Anderson, to make at least six replications of each plot, or treatment alternative, a commonly used experimental design for on-farm research is what’s called the randomized complete block.

Each block contains one set of all the plots, or treatments being compared, in random order. A block represents one replication, and blocks are placed side-by-side. A diagram for a simple two-treatment experiment is illustrated in “The Big Picture.”

The replicated block design is useful in accounting for and filtering out variation in agricultural soils, Anderson notes. That’s why splitting a 40-acre field, treating one side one way and the other side another way, yields dubious research information. However, this can provide a memorable visual comparison.

“We call that a demonstration research plot,” Anderson says. “Sure, you can tell obvious differences, like if one side of the field is dead and the other is growing, or if there’s a 50-bushel yield difference. But most studies don’t generate differences that dramatic.”

Gednalske makes another practical suggestion for your first ventures into on-farm research. “Look at as few variables as possible — just compare A to B, don’t go further and try to compare C, D and E,” he recommends. “The analysis is too complex. And when you limit the variation in the inputs, you’ll be more likely to get what the universities would consider good replication.”

Keep Equipment In Mind

Plot size is determined by your equipment size. “If a farmer has an eight-row planter and a four-row combine, for example, I’ll recommend that each plot be eight rows wide. It’s a convenient size to plant and harvest,” Anderson says.

Interestingly, Anderson recommends that only the center rows of the plots be harvested and this data used for comparison. So in the example above, he would direct the farmer to harvest and measure the center four rows and leave two rows on each side of the plot as buffers.

“This is because treatments on the edge rows of each plot can bleed into each other and confuse the effects,” Anderson explains. “This is particularly true for herbicide comparisons.”

Plot length is typically determined for convenience’s sake, usually to the end of the field to avoid having to turn around in the middle of the field. But plots don’t have to be this long, Anderson says. “Smaller plots will contain less natural variation.” he explains. “They can be 100 feet long and provide reliable data and cut down considerably on the amount of land used.”

Oolman notes an alternative that farmers sometimes use to compare seed selection is to stirp-plant the field, loading one type of seed in half the planter, another type in the other side. Although this does not randomize the plots, it can be an acceptable method “if you do 25 or 30 strips like this,” Oolman says.

Just take care not to favor one side over the other. “Make sure you’re planting at the same depth,” he notes. Also ensure that your planter is calibrated for the correct seed size and shape on each side.

Your Action Plan

With diligence and knowledge, good on-farm research is well within your reach.

If you’d like to pursue this topic further, here are some good resources to tap into: “On-Farm Research Guidebook” by Dan Anderson, available by writing: Department of Agricultural Economiccs, University of Illinois, 305 Mumford Hall, 1301 W. Gregory Drive, Urbana, IL 61801.

On-farm research coordinators are available through the agricultural economics department at the University of Illinois and the agronomy department at Iowa State University.

Many extension ag agents can provide research guidance and direct you to other resources at your land grant university.

Contact your farm supply dealer or the company rep from your herbicide or seed supplier. They can put you in touch with the research departments of your input suppliers, who may be able to provide advice, guidance and possibly some tangible support.

Editor’s note: This article first appeared in the March-April 1997 issue of PrecisionAg Illustrated.

0

Leave a Reply