![Manitoba Agriculture, Food and Rural Initiatives](/web/20061121033538im_/http://www.gov.mb.ca/agriculture/images/mafri1.gif)
![Factsheets & Publications](/web/20061121033538im_/http://www.gov.mb.ca/agriculture/images/header/factsheets_hdr.jpg)
October 2005 |
![back button](/web/20061121033538im_/http://www.gov.mb.ca/agriculture/images/back_button.gif) |
On-Farm Testing in Manitoba
|
What is an On-Farm Test?
An on-farm test (OFT) is a grower-directed, evaluation of new production techniques or
products using field-scale equipment. On-farm testing utilizes the principles of modern
scientific methods to evaluate a new practice, product, or innovative idea before applying
the new system to the whole farm. |
|
Table of Contents:
|
An OFT is NOT an unreplicated field demonstration. A demonstration only allows a local
comparison of how a practice/product "looks". An example of a demonstration
would be if you were to plant half a field with one variety, and the other half to a
different variety. A demonstration of this type is useful to observe the different
physical characteristics of the varieties but not yield. The reason why unreplicated
demonstrations are not acceptable for measuring yield differences is because there is no
measure of the variability within the field. Even in the most uniform field you will never
get exactly the same yield from any 2 harvested areas. Therefore, if you compared the
yields between the variety demonstration plots, you would not know if the results were due
to variety differences or soil characteristics. A properly designed OFT will be able to
separate the effects of natural field variability from the effects of treatments being
compared, and will provide accurate, reliable information upon which a grower can base
sound management decisions. |
Replication and Randomization Are the Key
Statistically valid OFTs will have true replication and randomization. Replication,
meaning repetition, allows you to determine if observed differences are due to treatments
or natural field variability. Replication is based on the theory that if one
practice/product is superior to another, it will become evident if you give it several
chances. This is similar to a Formula-One racecar driver who may not win every race (and
therefore receives less points), but will win more often than his/her competitors. By the
end of the year, the superior driver (the one with the most points) is obvious.
Replications can be next to each other in one field (see Figure A), in different areas
of a field, or even in neighboring fields (see Figure B).
|
|
Figure A
![fbc10s00a.gif (20145 bytes)](/web/20061121033538im_/http://www.gov.mb.ca/agriculture/facts/images/fbc10s00a_small.gif)
Click to expand
Figure B
![fbc10s00b.gif (26852 bytes)](/web/20061121033538im_/http://www.gov.mb.ca/agriculture/facts/images/fbc10s00b_small.gif)
Click to expand |
Repeating the whole test for several years is considered replication over time. By
replicating over years, the new practice/product is subject to a variety of growing
conditions and as such increases the reliability of the results. The minimum number of
replications is three. However, the most accuracy for the least number of plots occurs
with four replications. By having four replications, if you lose a replication for
whatever reason, you will still have the minimum number of acceptable replications for a
statistically valid test. By increasing the number of replications, you increase the
precision of the test and you will be able to detect smaller differences between
treatments. There usually is little benefit to having more than six replications. Harvesting
several samples within field strips is not true replication because there is no
randomization of treatments within the area where the samples were taken.
Randomization
of the treatments assigned to field strips ensures that any one treatment is not favoured
or biased in any way. Each product or practice is given an equal chance to perform to its
full potential. If only two treatments are being compared, a simple coin toss can be used
to assign treatments to plots (e.g. treatment A for heads, treatment B for tails). If
there is more than two treatments, assign numbers to treatments and write the numbers on
pieces of paper. Draw the numbers from a bag or hat. Repeat for each replication.
Seven Steps to a Successful On-Farm
Test
The success of your OFT will depend on how well it is planned!
ESTABLISH YOUR GOAL AND OBJECTIVES
Why do you want to do an OFT? What idea do you want to test or
question do you want answered? An OFT is suited to questions like what works better on my
farm, not why does it work better. Objectives are usually met by measuring something and
are used to obtain your goal. For example, your goal may be to improve crop yields on your
farm. A possible objective may be to compare your current fertility practices with a new
practice to improve yield. You would evaluate the differences by collecting yield data.
Some questions are better answered by replicated small plot tests as conducted by
researchers. It is important to evaluate current information that is available regarding a
new product or practice. Someone else may have already answered your question for you.
SELECTION OF TREATMENT
An effective way to choose treatments is to select two or three treatments
that represent significantly different production techniques or products. Remember that it
is more difficult to compare products or practices where the difference is expected to be
small. The smaller the differences expected between treatments, the more replication you
will need to establish differences. An appropriate check or control plot must be included;
the check may be the current practice or product. The check must be treated exactly the
same as the other treatments in terms of crop management factors. On-farm tests can be
used to evaluate single factors (different rates of herbicides) or systems (conventional
vs. zero-tillage). Plot management will be different between treatments when evaluating a
system approach because you will be comparing all aspects of crop production. For example,
you may still use crop yield to measure differences between two crop production systems,
however you may have to use different seeding equipment or methods of fertilizer
application.
Note: When selecting fertilizer treatment rates:
a) use rates that differ by
equal intervals (e.g. 40, 80, 120 lb. N/acre) and
b) ensure the full range of
rates, minimum to optimum to excess, is used.
|
- SITE SELECTION
When selecting a site you must consider the uniformity of the field. It is very
important to choose an area where all treatments have an equal opportunity to perform (see
Figure C). When choosing a site consider the following:
- previous crop management (fertilizer and herbicide rates, tillage)
- drainage
- soil texture and depth
- topography (see Figure D)
- bordering influences such as trees
- run-off from adjacent fields
- fencing
Avoid field corners and headlands. Access to the site should also be considered.
|
|
Figure C
![fbc10s00c.gif (131265 bytes)](/web/20061121033538im_/http://www.gov.mb.ca/agriculture/facts/images/fbc10s00c_small.gif)
Click to expand
Figure D
![fbc10s00d.gif (22380 bytes)](/web/20061121033538im_/http://www.gov.mb.ca/agriculture/facts/images/fbc10s00d_small.gif)
Click to expand |
- EXPERIMENTAL DESIGN AND PLOT LAYOUT
The use of a randomized complete block design for OFTs is considered to be the
standard. With this experimental design, all treatments are represented within a
"block". Treatments are randomly applied to the plots (test strip) within the
block (like a grouping). The blocks are then replicated three or more times. Blocks are
often referred to as replicates. The area chosen for a block should be as uniform as
possible, but the conditions between blocks do not have to be similar. For example, plots
within the block should receive the same amount of sunlight, have similar soil
characteristics such as soil type, depth, fertility, slope, and should have been treated
in a similar fashion in the previous year (e.g. tillage, manure application). Different
blocks may have a different soil type (see
Figure C). Using
this design allows all treatments to have an equal potential to perform while
"blocking out" the variability between replications. Blocks can be next to each
other, in different areas of the field, or even in different fields.
Plot Size
Plot size is determined by field size, uniformity of the field, equipment used, and
area needed to carry out a particular treatment. On-farm test plots will be more reliable
as the plot length is increased. There is no "optimum" plot length because each
field varies in uniformity. The more uniform the field, the shorter the acceptable plot
length. It is recommended that plots be as long as is practical. The width of the plots
should be two to three times the width of the swather or combine straight cut header
width. This will make it possible to harvest a full uniform width. Successful tests have
been performed in strips as short as 300 feet (very uniform fields), but strips 1,000 feet
or more will ensure that you will be able to detect true differences.
- DATA COLLECTION AND RECORD KEEPING
Who, what, when, why, and how measurements are taken should be decided before
establishing the OFT. What you will measure depends on the objectives of the OFT. A map of
field and plot locations is the first step, followed by a description of each of the
treatments. Yield estimates are required to make production and economic comparisons
between treatments and is almost always collected. To be valid, yield estimates should be
measured from comparable areas within each treatment and the size of the harvested area
must be determined. By having the width of the plot at least two to three times the width
of the header, you will be able to harvest a full header width from the center of each
treatment plot. This will ensure that yields are not affected by conditions bordering the
treatment. All plots should be swathed and harvested in the same direction. Yields must be
measured separately for each plot using a local truck scale, weigh wagon, or yield
monitor. Record moisture content of the grain at harvest for each treatment plot.
Collecting and recording data throughout the growing season is also an important part
of an OFT. Sometimes unexpected results can be explained by the conditions of the growing
season. The following type of information should be collected in order to help in the
interpretation of the results: field history, soil test and fertility program, seeding
information (e.g. variety, rate, depth) and planting conditions (e.g. depth of soil
moisture at seeding), all field operations and equipment used, weather, insect, weed, and
disease problems, pesticide applications, and crop growth and development.
- DATA EVALUATION AND CONCLUSION
Although this step is one of the last, the method of evaluating your results
should be planned before you start your OFT. This is one step where you may not have much
experience. Request help from a professional agronomist or researcher. Analysis of the
results will not be time-consuming or complicated once you know what to do. It is most
likely that you will be able to do your data analysis using a calculator. The hardest part
of the analysis will be to learn the statistical lingo and know what the numbers are
telling you so you can make appropriate conclusions.
- SHARING THE RESULTS WITH OTHERS
Sharing the results and conclusions from your OFT is an important part of the
OFT process. By sharing your information with others, others will likely share theirs with
you. You may develop new OFT research partners, which could reduce the number of plots you
would have to maintain in the future. Keeping the flow of practical information and
experience is an essential part of moving agricultural production and management ahead.
|
On-Farm Testing: A grower's Guide. Washington State University Cooperative
Extension Bulletin EB 1706, 1997. A guide to designing and carrying out On-farm testing.
Order from WSU Cooperative Extension Bulletin Office (509-335-2999).
On-Farm Research Guidebook. Dan Anderson, Department of Agricultural Economics,
University of Illinois. (217-333-1588)
Field Experimentation in Agriculture. Contact Dr. John C. Kons, Alberta
Agriculture, Food and Rural Development, Edmonton, Alberta. (403-422-4385) |
|