How to Conduct a Field Test the Right Way
Instructions from Usevalad Henin — precision farming professional, agricultural chemistry expert, and OneSoil co-founder.
Usevalad Henin
Usevalad is an expert in GIS and agricultural chemistry. He has been developing precision farming tools since 2013. He is also the co-founder of OneSoil.
A field test or experiment connects theoretical studies and agricultural practice. These experiments aren't limited to scientific institutions alone. Agricultural holdings and farmsteads use field experiments to compare different ways of increasing yields and make decisions about incorporating them. A field experiment helps answer questions like 'should new fertilizers be bought?' 'should certain kinds of seeds be used?' or 'should new equipment be purchased now or later?'
Since 2011, I've been conducting field tests in which I study variable-rate fertilizer application, variable-rate seeding, and how crop hybrids react to different seeding rates. I work with farmers to compare tilling methods and to study the difference between grain and row crop planters, herbicides, fungicides, desiccants, and much more.
From my experience, I've learned that conducting the test the right way is only half the battle. It's also essential to analyze the test results correctly. Misinterpreting the conclusions can lead to significant losses in yield and money. To help you avoid these mistakes, I'd like to share how to conduct a field experiment correctly, how to prepare for it, and how to analyze its results.
Since 2011, I've been conducting field tests in which I study variable-rate fertilizer application, variable-rate seeding, and how crop hybrids react to different seeding rates. I work with farmers to compare tilling methods and to study the difference between grain and row crop planters, herbicides, fungicides, desiccants, and much more.
From my experience, I've learned that conducting the test the right way is only half the battle. It's also essential to analyze the test results correctly. Misinterpreting the conclusions can lead to significant losses in yield and money. To help you avoid these mistakes, I'd like to share how to conduct a field experiment correctly, how to prepare for it, and how to analyze its results.
Get all new articles straight to your inbox!
A field test or experiment connects theoretical studies and agricultural practice. These experiments aren't limited to scientific institutions alone. Agricultural holdings and farmsteads use field experiments to compare different ways of increasing yields and make decisions about incorporating them. A field experiment helps answer questions like 'should new fertilizers be bought?' 'should certain kinds of seeds be used?' or 'should new equipment be purchased now or later?'
Since 2011, I've been conducting field tests in which I study variable-rate fertilizer application, variable-rate seeding, and how crop hybrids react to different seeding rates. I work with farmers to compare tilling methods and to study the difference between grain and row crop planters, herbicides, fungicides, desiccants, and much more.
From my experience, I've learned that conducting the test the right way is only half the battle. It's also essential to analyze the test results correctly. Misinterpreting the conclusions can lead to significant losses in yield and money. To help you avoid these mistakes, I'd like to share how to conduct a field experiment correctly, how to prepare for it, and how to analyze its results.
Since 2011, I've been conducting field tests in which I study variable-rate fertilizer application, variable-rate seeding, and how crop hybrids react to different seeding rates. I work with farmers to compare tilling methods and to study the difference between grain and row crop planters, herbicides, fungicides, desiccants, and much more.
From my experience, I've learned that conducting the test the right way is only half the battle. It's also essential to analyze the test results correctly. Misinterpreting the conclusions can lead to significant losses in yield and money. To help you avoid these mistakes, I'd like to share how to conduct a field experiment correctly, how to prepare for it, and how to analyze its results.
Content
Glossary of foreign terms
Well, they’re not actually foreign. But when you see them for the first time, they may look like Ancient Greek to you.
A field test consists of researching the experiment variations in field conditions.
The experiment variation is the object of your experiment. These could be the fertilizer rate, tilling method, crop varieties, growing conditions, or something else, for example.
The control is what you want to compare with the experiment variations. The standard sowing rate is the control for experiment variations.
Replication is the number of areas in the field that have the same experiment variations and the same soil fertility conditions.
A test strip is a part of the field that contains one experiment variation. Fields may have different soil fertility conditions.
The experiment variation is the object of your experiment. These could be the fertilizer rate, tilling method, crop varieties, growing conditions, or something else, for example.
The control is what you want to compare with the experiment variations. The standard sowing rate is the control for experiment variations.
Replication is the number of areas in the field that have the same experiment variations and the same soil fertility conditions.
A test strip is a part of the field that contains one experiment variation. Fields may have different soil fertility conditions.
Recommendations on conducting field tests from scientific institutions
Select a homogeneous field for your experiment. It would be better to suspend this field from crop rotation.
Perform broadcast seeding in the selected field. You’ll need it to even out the soil fertility. A good precursor option would be grain crops for herbage production.
Establish 3−4 replications to avoid mistakes when analyzing the result.
Allocate one test strip as a control if you want to compare something in the field.
These are the basic recommendations for setting up a field experiment. My experience shows that scientific institutions are the only ones who follow them. That’s probably because it’s not always possible to suspend a field from crop rotation, and it doesn’t always make sense. Secondly, farmers and agronomists want to know if this technology will work on their fields under usual conditions outside of the experiment. As you can see, it’s challenging to follow these instructions.
Perform broadcast seeding in the selected field. You’ll need it to even out the soil fertility. A good precursor option would be grain crops for herbage production.
Establish 3−4 replications to avoid mistakes when analyzing the result.
Allocate one test strip as a control if you want to compare something in the field.
These are the basic recommendations for setting up a field experiment. My experience shows that scientific institutions are the only ones who follow them. That’s probably because it’s not always possible to suspend a field from crop rotation, and it doesn’t always make sense. Secondly, farmers and agronomists want to know if this technology will work on their fields under usual conditions outside of the experiment. As you can see, it’s challenging to follow these instructions.
Why these recommendations are hard to follow
Homogeneous fields are very rare in nature. In most cases, productivity varies within a single field. There's nothing we can do about it. As a result, the only thing that's left to do is to work with heterogeneous fields.
Productivity zones in the same field may vary from year to year. I usually divide the fields into two types of fields: ones with stable and unstable productivity zones. The productivity zone features in the first kind of fields are not affected by either weather conditions or cultivated crop features. In this case, the field itself has a factor or factors that limit the yield. For example, their influence exceeds the influence of humid or dry weather.
For the second type of field, productivity zone features vary from year to year. During one variable-rate sowing experiment I conducted last year, I saw how a moderate productivity zone became a high productivity zone after harvest. This is an example of the second type of field. In unstable fields, the yield is primarily affected by weather conditions or by cultivated crop features instead of internal factors.
It is essential to determine whether the field is stable or unstable to be able to interpret the results well.
It is very easy to make a mistake when identifying productivity zones. The field's heterogeneity is characterized by its productivity zones. There are several ways to identify them. You can use soil analysis data, vegetation data, or yield data over multiple years. All methods are good if you know the factor that limits the yield on your field. But that's not always the case.
The most surefire way to identify productivity zones is to use yield maps over at least three last years. Doing so will help you ensure you know whether or not the productivity zones are stable and be able to spot these zones in the field. Important! The yield data should be precise. The combines are calibrated to extract this data.
Productivity zones in the same field may vary from year to year. I usually divide the fields into two types of fields: ones with stable and unstable productivity zones. The productivity zone features in the first kind of fields are not affected by either weather conditions or cultivated crop features. In this case, the field itself has a factor or factors that limit the yield. For example, their influence exceeds the influence of humid or dry weather.
For the second type of field, productivity zone features vary from year to year. During one variable-rate sowing experiment I conducted last year, I saw how a moderate productivity zone became a high productivity zone after harvest. This is an example of the second type of field. In unstable fields, the yield is primarily affected by weather conditions or by cultivated crop features instead of internal factors.
It is essential to determine whether the field is stable or unstable to be able to interpret the results well.
It is very easy to make a mistake when identifying productivity zones. The field's heterogeneity is characterized by its productivity zones. There are several ways to identify them. You can use soil analysis data, vegetation data, or yield data over multiple years. All methods are good if you know the factor that limits the yield on your field. But that's not always the case.
The most surefire way to identify productivity zones is to use yield maps over at least three last years. Doing so will help you ensure you know whether or not the productivity zones are stable and be able to spot these zones in the field. Important! The yield data should be precise. The combines are calibrated to extract this data.
If you don't have access to several years of data or you doubt its accuracy, you can use productivity zones from the OneSoil web app. We build productivity zones based on vegetation data for the last four years. We use this data to model the relative yield. The algorithms allow us to calculate how much one part of the field is more productive than another.
My tips for designing on-farm trial
Highlight the productivity zones on the field and analyze them. If you’re using the OneSoil web app, estimate how well the productivity zones in the app match your estimate of annual yield distribution. Study the vegetation data for the last four years in the app. Note the vegetation maps when crops begin to ripen. They usually correlate to the yield distribution on the field. If you see anomalous vegetation index values for some years, exclude these years when building the productivity zones.
Try to identify the limiting factor. Figure out what exactly keeps the yields low in low productivity zones and what leads to high yield in high productivity zones. Here are brief instructions on how to do it.
Buy a yield mapping system. A combine with a yield monitoring system increases the number of replications several-dozen-fold. This kind of combine also makes it easier to collect and analyze the data. The combine makes it possible to measure yield anywhere in the field and compare it to productivity at the same point. If you don’t have a combine with a yield monitoring system, exclude the productivity’s influence on the experiment’s results. There are two ways you can do it. The first option is to conduct the experiment only in one productivity zone. The other option is to draw the test strips in such a way that each of them has a plot with different fertility levels (low, moderate, and high) and approximately similar areas.
Try to identify the limiting factor. Figure out what exactly keeps the yields low in low productivity zones and what leads to high yield in high productivity zones. Here are brief instructions on how to do it.
Buy a yield mapping system. A combine with a yield monitoring system increases the number of replications several-dozen-fold. This kind of combine also makes it easier to collect and analyze the data. The combine makes it possible to measure yield anywhere in the field and compare it to productivity at the same point. If you don’t have a combine with a yield monitoring system, exclude the productivity’s influence on the experiment’s results. There are two ways you can do it. The first option is to conduct the experiment only in one productivity zone. The other option is to draw the test strips in such a way that each of them has a plot with different fertility levels (low, moderate, and high) and approximately similar areas.
The first option is to conduct a test only on a high-productivity plot. You should collect the harvest in the test strips to analyze the experiment’s results.
The second option is to conduct a test in three plots with relatively equal areas and various productivity levels. This approach lowers the probability of making a mistake while analyzing the results of the field experiment.
You’ve identified the productivity zones, discovered the reason for the field’s heterogeneity, and selected the plot where you will conduct the experiment. Now it’s time to set up the experiment.
How to set up a field test by productivity zones
Let’s imagine that we have to compare three kinds of tilling: traditional tilling, Strip-Till, and No-Till.
The traditional method calls for plowing the entire field. The soil is tilled several times to keep weeds under control.
Strip-Till is a strip method of tilling. Only the strip that will be sown is plowed. The rest of the field remains fallow.
No-Till stands for zero tilling. The field is left unplowed and covered with mulch, then a new crop is sown in stover.
Strip-Till is a strip method of tilling. Only the strip that will be sown is plowed. The rest of the field remains fallow.
No-Till stands for zero tilling. The field is left unplowed and covered with mulch, then a new crop is sown in stover.
The traditional method calls for plowing the entire field. The soil is tilled several times to keep weeds under control.
Strip-Till is a strip method of tilling. Only the strip that will be sown is plowed. The rest of the field remains fallow.
No-Till stands for zero tilling. The field is left unplowed and covered with mulch, then a new crop is sown in stover.
Select which field to experiment in. The ideal option is a field with productivity zones located close to each other. In other words, we need a field that has a large high-productivity zone, a large moderate-productivity zone, and a low-productivity zone of the same size.
An example of a good field for conducting a test. Despite the field’s heterogeneity, the productivity zones have approximately equal area and are located next to each other.
Check the direction the equipment should move in the experiment field. The machines should pass through all three productivity zones.
This is what passing through all zones looks like
Divide the field in test strips. To compare three different tilling methods, the plots where they’ll be applied should border each other. We end up with four strips.
We can use this map to compare the traditional tilling with Strip-Till, Strip-Till with No-Till, and No-Till with traditional tilling.
Allocate plots to analyze the yield after the harvest. First, remove outliers from the yield data collected by the combines. Outliers are anomalous yield levels. After that, we select which plots to analyze. The yield within a plot shouldn’t vary a great deal. Plots are allocated at the borders between test strips. Why? The yield may vary even within the same productivity zone. If the soil fertility factor turns out to be stronger than the tilling factor, we might make false conclusions for the entire experiment. Nevertheless, the closer the testing plots are to each other, the more likely it is that they’ll have similar productivity levels and soil physical and chemical properties. That’s why, when comparing methods, we only need to look at the nearest plots of neighboring test strips.
These are the plots you should compare
These are the test strips you shouldn't
Another thing I recommend is doing two passes on the combine: in one direction in one row and in the opposite direction in the next row. This will help you reduce the chance of making a mistake due to an inaccurately calibrated combine.
Analyze the yield data. To do so, compare the average yield in each plot and compare the values of the plots opposite them.
Draw conclusions!
Analyze the yield data. To do so, compare the average yield in each plot and compare the values of the plots opposite them.
Draw conclusions!
Things to remember
A field experiment is better off on a field where different productivity zones are located next to each other. They should be big, and their area should be roughly equal.
To avoid false conclusions, it is important to analyze the productivity zones and identify the limiting factor before the experiment.
During the experiment, the work should be executed in all productivity zones.
The test strips should be allocated in such a way that every variation borders another.
The onboard computer data should be free of outliers. In other words, outlying data shouldn't be considered in the analysis.
When analyzing the yield, I suggest comparing the values in plots derived from different test strips that are similar in their physical and chemical properties. This will make the experiment's conclusions more objective.
To avoid false conclusions, it is important to analyze the productivity zones and identify the limiting factor before the experiment.
During the experiment, the work should be executed in all productivity zones.
The test strips should be allocated in such a way that every variation borders another.
The onboard computer data should be free of outliers. In other words, outlying data shouldn't be considered in the analysis.
When analyzing the yield, I suggest comparing the values in plots derived from different test strips that are similar in their physical and chemical properties. This will make the experiment's conclusions more objective.
Things to remember
A field experiment is better off on a field where different productivity zones are located next to each other. They should be big, and their area should be roughly equal.
To avoid false conclusions, it is important to analyze the productivity zones and identify the limiting factor before the experiment.
During the experiment, the work should be executed in all productivity zones.
The test strips should be allocated in such a way that every variation borders another.
The onboard computer data should be free of outliers. In other words, outlying data shouldn't be considered in the analysis.
When analyzing the yield, I suggest comparing the values in plots derived from different test strips that are similar in their physical and chemical properties. This will make the experiment's conclusions more objective.
To avoid false conclusions, it is important to analyze the productivity zones and identify the limiting factor before the experiment.
During the experiment, the work should be executed in all productivity zones.
The test strips should be allocated in such a way that every variation borders another.
The onboard computer data should be free of outliers. In other words, outlying data shouldn't be considered in the analysis.
When analyzing the yield, I suggest comparing the values in plots derived from different test strips that are similar in their physical and chemical properties. This will make the experiment's conclusions more objective.
Mistakes to avoid when conducting on-farm trial
Situation: a heterogeneous field where fungicides A and B were compared. The productivity of the plot where fungicide A was tested is 34% higher than the productivity of the plot using fungicide B.
If this field looks like an experiment you've recently done, don't tell anybody about it
Experiment conclusion: applying fungicide A results in a higher yield increase than applying fungicide B.
Do you think the results from this kind of experiment can be trusted? Share your thoughts in the comments.
Do you think the results from this kind of experiment can be trusted? Share your thoughts in the comments.
Can you trust the results of an experiment that doesn’t take into consideration productivity zones?
Usevalad Henin
Our experiments
People, fields and technology
Get all new articles straight to your inbox! You can read them even in the field.