Farms can run their own research trials to test how a particular practice or product performs in the local environment. On-farm trials could test, for example, the difference between 60 pounds and 120 pounds of nitrogen fertilizer per acre, or between acres treated or untreated with fungicide. They can also evaluate the return on investment for novel products.
Here are the basic steps for effective on-farm trials:
Start with a simple plan. A two-treatment trial (new versus normal) is probably a good realistic objective for an on-farm trial. More treatments make the trial more complicated to plan and properly manage.
Select the test area. Longer strips produce more reliable results, so aim for 750 feet or more. For strip width, have a treated width that is wider than the harvester width. This will ensure that the area harvested keeps fully within the treatment area. Put strips of treated versus untreated crop in a uniform part of the field. If a uniform area is not possible, choose an area of the field that reflects the field as a whole. Note that completely uniformity is not possible, which is why the next step is important.
Replicate the strips. Try a few untreated and treated strips in the same block (four strips of each are ideal) within the same field. For a more accurate comparison, replicate these blocks in different areas of the farm and, ideally, in different years. This increases the confidence that differences between treatments are because of the treatments, and not because of chance variation is caused by differences in weather, soil and other factors.
Control other variables. If comparing a fungicide, for example, make sure the treated and untreated strips are the same variety, seeded the same day with the same tool, and follow the same practices for fertilizer, weed control and harvest. If it’s a fertilizer trial, have details soil sample results for the treatment area to rule out natural variation. That is another reason why multiple strips within a relatively uniform part of the field are important. Another consideration: If applying fungicide to a test area, spray perpendicular to the direction of seeding to ensure the same amount of wheel tracks throughout the trial. Apply to entire trial on the same day.
Weigh the results. Harvest all treatments on the same day. Cut through the middle of each strip to avoid edge effects and combine each row separately. Use a weigh wagon to get the most accurate yield data for each strip. Measure the exact length and width of the strips. Make sure hopper is empty after each treatment.
Keep notes. Record weather conditions, soil moisture, seeding date, pest pressures, harvest date, harvest quality and anything else you can think of. That will help create scenarios where a product may or may not work.
Get advice. Agronomists with experience in trials could provide help on set up, harvest and evaluation. Agronomists may also have weigh wagons or know how to get one.
Evaluate the results. A higher fertilizer rate or a fungicide treatment may produce higher yield, but it also needs to increase profitability, make life easier or reduce risk. To evaluate a return on an investment, one needs to know the average treatment response, expected crop price, and the cost of the treatment. Accuracy of the result will increase with more replications and site-years. Check out some result summaries from previous on-farm (field scale) research trials carried out through the Ultimate Canola Challenge program.
For a couple of simple data evalution tools, enter weights for each strip into the Paired T-test calculator or try the Indian Head Agricultural Research Foundation (IHARF) data analysis tool. Terms to know:
—Mean. This is the average of the five yields for each treatment. The mean difference between Treatments A and B is 1.7 bu./ac.
—Probability of this result. Due to the fairly consistent yield results, the calculator shows a low probability — 5.6% — that this result could have occurred by chance alone. The lower the better. Alberta Agriculture oilseed specialist Murray Hartman says: “When analyzing field data, we generally have more confidence in a result when the probability is 5% or lower.”
—Least significant difference (LSD). As the IHARF tool says: “A p-value of 0.05 is chosen most frequently in scientific experiments, however 0.10 is sometimes recommended for large-scale field trials to account for increased overall variability relative to small-plot studies. At p=0.05, there is a 5% probability of either (1) concluding there is a difference between two treatments when, in actuality, there is no difference, or (2) concluding there is no difference between the two treatments when a difference actually existed. At p=0.10, the probability that one of the previous two errors will occur is 10%.
The Canola Council of Canada website has more detailed tips for on-farm trials, including suggested protocols and corresponding data collection sheets for nitrogen, seeding rate, seeding speed and variety (cultivar) comparisons.