Farmers are constantly being bombarded with recommendations and advertisements for new inputs or practices that claim to increase crop production and farm profits. Examples include various speciality fertilizers or crop protection products that claim to increase crop production. In some cases, excellent information from replicated trials at agricultural research centers or applied research associations may be available.
But some new products are promoted based only on testimonials with little scientific testing to show yield benefits. On-farm research trialing can be very useful to farmers for answering crop production questions and solving production problems. On-farm research allows a farmer to examine a new product or practice in test strips to verify potential benefits and examine how local factors, such as differences in soil types and environmental conditions, may affect the performance of a new practice or input.
On-farm trialing can provide very good information that is specific to your farm and local conditions. From an economic
standpoint, if a new product costs $10 per acre and you plan to use it on 2,000 acres, the annual cost is $20,000. Spending a bit of time to test the product on your farm would help determine if it would be financially beneficial to use in future years. If the product works, that’s great information; but if it doesn’t work as claimed, that is equally good information, as it will save you the extra expense in future years. Regardless of the result, on-farm trialing provides excellent information for managing your farm.
Doing it right
On-farm research must be done right. Splitting a 160-acre field in half, to compare 80 acres conducted with a normal farming practices versus the other 80 acres conducted with a new practice, is not on-farm research. Any yield differences between the two sides of the field could be due to natural field variation, which is not taken into account in this type of demonstration. Statistical analysis cannot be done with this type of demonstration and, therefore, you have no confidence that any yield difference is due to the new treatment or simply the result of natural field variability.
The greatest challenge for farmers when conducting on-farm trials is that it is time consuming to design, set up and manage the trial plots. For many farmers, the busy day-to-day farm activities during the hectic growing season take priority over conducting and monitoring field research trials. One alternative to dedicating the time required is to collaborate with a private agronomist who has the knowledge and time to make the necessary in-field observations, work with the farmer to collect yield data, statistically analyze the data and assist with interpreting the results.
When conducting on-farm research, be sure any differences measured are the result of treatment differences and not soil, topography or other variability within your field. Set the trial up so you can statistically analyze and evaluate the results. Statistics are simply used to calculate the odds that any yield difference measured has a very good probability of being repeatable. Speak to a government crop specialist/research scientist or applied research association in your area for advice on setting up an on-farm research trial that would be statistically sound.
The first step in conducting an on-farm trial to develop a testable statement or question that you want to answer with your field trial – this is called the hypothesis. For example, you are interested in claims about a new foliar-applied fertilizer being promoted that will increase crop yield. You want to test the product with several crops on your farm. Develop a good, detailed plan and protocol for conducting the on-farm experiment. This includes the treatments to be used, how you will lay out the trial in the field, what needs to be measured, how to determine crop yield and how to analyze and interpret the data collected.
Factors to keep in mind
Keep treatments simple: Two or three treatments is often best – one treatment is the control, which is your normal practice, and the second and/or third treatments are the new products/practice you want to test.
Decide on field location: Select a field with the most uniform soil and topography to minimize field variation. Make sure to document crops, fertilizers and pesticides applied in the previous three years to ensure there won’t be interaction factors that might negatively affect the on-farm field trial.
Decide on size: Treatments should run the length of the field and be long, narrow and parallel to each other. In terms of treatment width, make each treatment the width of your seeder or width of your sprayer, depending on the product you are applying and testing. Long, narrow treatments that are parallel to each other help to overcome field variability.
Replicate: Replicate each set of treatments at least four times within the field and do not have the order of the treatments the same in each replicate.
Control: Ensure a control, which is your normal practice, is in each replicate as this is essential for comparison.
Carefully draw the field layout of the on-farm test strips and treatments so you know exactly what you are going to do and where the strips will be in the field. Make sure you write down the protocols you will follow to conduct the experiment. It is a good idea to have one field book for all your notes and document everything you do from seeding to harvest.
Prior to establishing the on-farm trial, take a set of soil samples in each strip where the proposed trials will be laid out. Ensure good seed quality by testing for per cent germination, vigour and seed-borne disease.
When establishing the strips, ensure that each treatment is carefully and uniformly laid out. Seed all strips the same day. In each treatment strip, make regular observations and notes about seedling emergence, plant counts after emergence (e.g. number of plants in two parallel rows for one metre), weekly growth stages, weed pressure, crop height, disease levels and lodging. Always make observations at paired locations across treatment strips, and repeat the observation at multiple locations across each set of treatments to account for field variability. Put up a good rain gauge and record rainfall throughout the growing season.
Manage the field and the treatment strips as uniformly and consistently as possible. For example, when applying a herbicide or fungicide that is not part of the experiment, treat the entire field with the same product, at the same rate in one day, ideally applying the product at a 90 degree angle to the test strips, so the strips are treated as equally as possible.
At harvest, combine and weigh each treatment strip separately. Also, take two to four grain samples per treatment to determine the test weight, grain protein and grain moisture content. For oilseed crops you may want to check oil content.
The final critical step is to work with a specialist to statistically analyze the yield results from the research trial. This will tell whether or not there were statistically significant differences among the treatments in your trial. Having a minimum of four replicates for each treatment is important to determine statistical probability. Probability is an estimation of likelihood of an occurrence. Probability ranges between zero and 100 per cent. A zero per cent chance will not happen and 100 per cent chance will happen. The higher the degree of probability, the more likely the input or practice has of giving positive results. Typically in agricultural research, a 95 per cent confidence level in statistical analysis is used, which simply means that 19 times out of 20 the difference in crop yield is due to treatment differences.
Alberta Agriculture and Rural Development has some good information on how to set up and conduct on-farm research trials. Visit http://www1.agric.gov.ab.ca/$department/deptdocs.nsf/all/sag3023 for more information. Another good place to start is to chat with a crop specialist/research scientist in your region to assist you in developing your on-farm research plan.
April 23, 2014 By Ross H. McKenzie PhD P. Ag.