• Rachel

Learning about On-Farm Research

Updated: Mar 7, 2019

I recently attended a meeting hosted by the Nebraska On Farm Research Network. At the meeting, growers from across the state shared results from studies that they conducted on their farms during 2018. There were over 80 studies in total. I was impressed by the number of farmers who took the time to set up well thought out studies testing a wide variety of topics ranging from plant populations, to growth promoters, to downforce systems, to fertility rates. I spent some time after the meeting thinking about how inquisitive farmers are in Brown and surrounding Counties. As a whole, you all are very willing to try new things to test that you are making the right decisions for your operation. With that in mind, I wanted to share with you some of the steps and tips presented at the meeting as well as resources from leaders in On Farm Research.


Motivation

So, what is the motivation behind setting up an on-farm study on your farm? For many producers, it is the relevancy of taking university or industry research and moving it to a level that is applicable to them. Not only are these results now on a scale relatable to the farmer, but also more impactful by doing it in conditions that are familiar to the producer with his or her own equipment and resources. Many farmers who conduct on farm research appreciate the fact that it allows them to test specific topics they are interested in. Have you been thinking about seeding rates? Or planting depth? Maybe even some new planting attachments? Fertilizer? In season Nitrogen applications? Multiple fungicide applications? Cover crops? You can test any of these topics yourself! And in the most relevant setting: your farm. Ultimately, this will let you make educated management decisions. Instead of guessing if a new management practice will make a difference, you can test it out first. This could potentially save you a lot of money. Instead of converting every acre to a new, unproven method, you are able to test the efficacy, productivity and financial viability of a practice on a smaller number of acres. With our access to the latest precision agriculture equipment such as GPS and autosteer, implementing and harvesting strips is easier than ever. Autosteer in particular allows you to plant all the locations of one treatment and then come back and fill in with other treatments easily.


Chose A Topic

Like I mentioned before, On Farm Research lets you test any topic you want. The possibilities are nearly endless. But don’t bite off more than you can chew. Too complicated of a study will take up too much space, be difficult to analyze, and ultimately be frustrating to you as a producer. Say you are interested in learning what planting population, depth, and timing is the best. If we were to create one study including all those factors, by the time we included two or three different treatments for each and then each of the replications needed for getting us accurate results, our study would take up an entire field and be very complicated to analyze and keep straight! Even if we just did two planting populations, two planting dates and two planting depths, that would be 8 treatments in every replication, and likely over 40 planted treatments total once we get all the replications included. Just too complicated for an on-farm research study. We would suggest focusing on one factor at a time. Perhaps just start with planting depth and create a study to test that single factor first.


Setting Up A Trial


Figure 1: Split Field Layout. Treatment results can be affected by field variability from things like soil type, as seen in this figure. Splitting a field in half by treatment will not give you accurate results.


How do we go about setting up an on-farm trial? One of the most important aspects of setting up a trial is accounting for variability. Even the most uniform appearing fields have factors that affect the soil differently, or historical management practices that affect how the soil physical property or soil fertility levels change across the field. Keeping in mind that variability will still exist, we want to aim for a field site where variability is low. Consequently, prior knowledge of the field is very important.


When setting up a study, it is very tempting to set up a side by side, say for instance, fungicide vs no fungicide. This could be two passes next to each other in the field, or maybe you split the field in half and place one treatment on one side, and the other treatment on the other side, like in Figure 1. While this would take less time, we would have misleading or inaccurate results because of the variability that we see across the field. Perhaps one side of the field has a more productive soil. If the fungicide treatment is placed on the more productive side of the field, that treatment will automatically have the advantage simply because of where it is placed. To eliminate any bias, we need to place the treatments multiple times across the field. This means that the more times we are able to include treatments across the field, the more confident we can be that the treatment results aren’t based on where they are placed in the field. This is called Replication. Figure 2 shows an example of replication.


Figure 2: Replicated Study in area with multiple soil types. This helps dealing with variability, but one more step is needed to adjust for inequalities in productivity across the field.


While this goes a long way towards correcting for variability, there is one more step that we need to take. Lets consider a field that is increasing in productivity from West to East. (The field is more productive on the east side.) Imagine again that we are trying to test a fungicide treatment out. If we add replications in this field with the fungicide treatment always to the East of the control treatment, the fungicide treatment always has a little higher productivity in those paired comparison. This again biases the treatments towards the fungicide treatment. So how do we compensate for this? Within each paired strips (replication), we need to randomize the treatments. This means that sometimes the fungicide treatment is on the left side of the treatment, and sometimes on the right side, as shown in Figure 3. This will help correct for that variation across the field that I just mentioned. We’ve learned to get the most accurate results, we must randomize and replicate our treatments. Generally, we like to see somewhere around 5 randomized and replicated strips in our field study in order to get a representative average for the field.


Figure 3: A Randomized and replicated field study accounts for variability in the field and removes directional bias between paired strips.


Analyze Results

So how do we know our results are accurate? How confident can we be? When we look at results from research studies, we will get averages of the treatments across all replications. We will also look at something called a confidence level. This tells us how sure we can be that the results we are seeing are due to the treatment and not from variation in the field from soil type, topography, residue amounts, pest or disease pressures, weather events, drainage, or any other factor that impacts the field disproportionately. Ordinarily, we use an 85, 90, or 95% confidence level. This gives us confidence in the difference between treatments.


Common Problems

Setting up a good on farm research project does take time and effort. Committing to the extra time will provide excellent quality results that you can help inform the decisions you make on your farm. However, there are some common problems that we should address in on farm research projects.


One of those is selecting the correct strip width for the study. Because these are field studies and not small block research, we will be using full sized equipment that you have on your farm. Its important to make sure that your planting passes are in multiples of your harvester header widths. For instance, you may plan a strip as two passes with your 30 foot planter. You might then harvest with your 15 foot header. When we start getting mismatched size equipment we need to monitor closer to make sure we are only taking one treatment per harvest pass and not a mixed treatment pass. This may also require some extra time to come back and harvest a small strip remaining from differences in planter and header sizes.


Another factor we need to consider is something called an edge effect. Will the edge of one treatment affect another treatment? This can be common in research on nitrogen. Plants will pull nitrogen from several rows over impacting the treatment next to it. Another example is fungicide treatment. Spray drift can move several rows into the next treatment. When this is the case, we often like to leave a buffer strip between treatments that won’t be included in the study, or just take the center passes out of each treatment to ensure we aren’t contaminating our results.


Finally, a successful on farm research project often is marked by detailed notes. A lack of note taking can lead to confusion in treatment location, background management practices, or weather events affecting the crop.


Steps for Success

The Nebraska On-Farm Research Program outlined 10 steps to a successful on farm research program. I wanted to share a couple here that I thought were important.


1. Consider field history. You may take soil factors such as slope and soil types into consideration when setting up a field trial, but don’t forget to consider man made variation. Was there a location where an old farmstead or barn was that would have different fertility levels? Did the field used to be split into multiple smaller fields? Has drainage been implemented in only part of the field? Think through these factors before choosing a location for a plot.


2. Collect Data. It is possible you are already collecting the main components of data that we need, for example, yield monitor data. However, there is additional data that can be collected that might be beneficial, such as aerial images, plant populations, geo-referenced strip locations, stalk strength, and weed pressures.


3. Use Precision Ag Tools. This greatly reduces the time needed for On-Farm Research projects. The vast majority of our studies can be set up to be harvested with a yield monitor only, greatly reducing the time that would have been spent weighing various trials. Additionally, at planting time, variable rate equipment allows for quick changes between rates or treatments.


4. Get your neighbors involved. We all learn more when we can share what we are learning on our farms. Get a neighbor involved to do the same study as you. More data will help you make better decisions for you farm. Attend meetings with the Kansas or Nebraska On Farm Research Networks or with Pederson Seed to share you results.


Get help from Pederson Seed

We’d love to help you get set up for in study you may be interested in conducting in 2019. Let us know what topics you may be interest in and we can help you find a good field location, set up the trial, collect data, and analyze results. Let us know soon—planting will be here before we know it!


Resources

https://cropwatch.unl.edu/2017/field-studies-replicationed-comparisions-vs-side-side-comparisons

https://wsaregrants.usu.edu/grants/docs/FarmResearch.pdf

https://cropwatch.unl.edu/farmresearch/getting-started

https://www.sare.org/Learning-Center/Bulletins/How-to-Conduct-Research-on-Your-Farm-or-Ranch/Text-Version/Basics-of-Experimental-Design/Common-Research-Designs-for-Farmers

https://cropwatch.unl.edu/2016/10-steps-farm-research-success

https://cropwatch.unl.edu/2017/field-studies-setting-trial

https://extension.psu.edu/on-farm-research-ii-summarization-and-analysis-of-data

https://www.sare.org/Learning-Center/Bulletins/How-to-Conduct-Research-on-Your-Farm-or-Ranch/Text-Version/Basic-Statistical-Analysis-for-On-Farm-Research

https://cropwatch.unl.edu/on-farm-research

29 views

© 2020 by Pederson Seed and Services

  • Black Twitter Icon
  • Black Facebook Icon
  • Black Instagram Icon

900 S. 1st St

Hiawatha, KS 66434

785-742-3241

pedersonseed@gmail.com