Designing Field Trials - A Very Brief Introduction

This year the WCTA research committee has made a significant addition to its goals. The new goal is "to define and contract out projects to better meet members specific needs." Basically the WCTA has allocated money for members to conduct their own trials. This article outlines some of the points that anybody designing a trial needs to consider before the test begins. The approach will be to refer to an example project to show how some the general points would apply to a specific trial.

Example Project

Assume I manage a turfgrass facility and that one of my biggest problems is that the turf quickly goes under water stress whenever the weather turns warm and sunny. The soil is sandy and I believe that the problem is that the water holding capacity of the soil is low. I am interested in trying some new soil amendments that are claimed to improve the water holding capacity of the soil and hopefully reduce drought stress. I have investigated many products and have narrowed my choice to three. Since I am trying to remedy an existing turf I will incorporate each material by topdressing it into the turf after core cultivation.

Defining the Problem. What do I want to know?

Defining the problem is one of the biggest challenges when designing a trial. Consider the example above. There are actually two questions being considered here. They are:

Will any of the products improve the water holding capacity of the soil?

Will any of the products reduce the amount of water stress the turf is suffering?

For the sake of this article I will focus on the second question.

Data Collection. What kind of data will meet my objectives?

My goal is to reduce the amount of water stress and more generally improving turf quality. In the case of the example I want to determine how long it takes for the turf to show signs of water stress (foot printing or wilting) after an irrigation. Therefore from an irrigation I will be collecting data on incidence of foot printing and visual quality as the soil dries. Because I am testing more than one product I may decide to hand water plots that show stress first in order to see when water stress occurs in all of the plots.

In trials of this kind it is important that I have control over the irrigation of the test area. Therefore I will assume responsibility for irrigating the test area. This kind of control is especially important when testing different fertilizers.

One note about assessing visual quality. It usually works best if one person collects the visual quality data. Individuals can have remarkably different assessments of the quality of the same turf.

At the end of the experiment I may want to sample the soil to determine the water holding capacity of the soil and to determine the root system of the turf under each plot.

Deciding on the Treatments. Remember the controls.

How many treatments will my trial include? I would have five treatments in this test. Each product would be a treatment plus two controls. One control would be to do nothing to an area. A second control would be to core the turf and not add any amendment.

It may be that the coring is the most important factor and the amendments really have no benefit. Or it may be that the year I do the experiment none of the turf experiences any drought stress. The only way to eliminate both of those options is by having control treatments in the trial.

Locating the trial. Where will I do my testing?

Where will I locate the trial? There are a number of factors to consider:

Select an area that suffers from the problem. In my example I need an area where I reasonably expect drought stress to occur

Select a uniform test area. Results from a uniform test area are easy to collect and interpret. Areas that are not uniform can be used but get some help from someone who has experience setting up field trials before you start.

Select areas where different results among plots will not be a problem for the use of the turf. For example if I was setting this trial out on a golf course I would use a practice green rather than the 18th green. In most cases a checkerboard of treatment effects is not acceptable to the users of the turf.

Select an area that is convenient to your office or yard. This minimizes the time needed to treat the plots and collect data.

Almost inevitability, the site chosen will not be ideal. Make sure that the compromises you make still allow you to answer your research question.

Replications.

First, why replicate the treatments? Consider the example experiment again. In spite of my best efforts to select a uniform area I may place one treatment where no drought stress will occur. The data from that treatment is not representative of the effect of the treatment. The only way to account for this variability is to replicate treatments. The number of replicates may vary from two to five for most field trials. The exact number will depend upon:

The number of treatments. The fewer the treatments the greater the number of replicates required. (If only two treatments are being considered, a control and a product, it may be practicable to have a large number of replications in different areas.)

The size of the expected difference. If I expect that the treatments will have a large effect then I will need fewer replicates. If I expect that the treatments will have a small effect then I need a greater number of replicates.

The variability of the test area. The more variable the test area the greater the number of replicates needed.

The time you have to manage the trial.

Plot Size. How big should a treatment area be?

Plot sizes can vary from 1 m2 to 100 m2 or more. The plot needs to be big enough so that you can measure the results but big plots require more work. A practical way to set plot width is to use the width of any application equipment. In my example I would use the width of the core cultivator (or two passes) as one plot dimension. I may use the width of a drop spreader (or two passes) for the other plot dimension.

If you are interested in looking at a factor like disease incidence make the plots as large as possible. You probably are not going to infect each plot and the level of inoculum may vary greatly over your trial area.

If the data collected requires a lot of core sampling then the experiment will require larger plots than if the data collected was your estimate of turf colour.

Experimental Design. How do I arrange the treatments?

When I apply the treatments to the turf how do I assign treatments to specific plots. The treatments will be assigned to the plots at random. Randomly assigning treatments to plots ensures that treatment differences are due to the treatments not to the plot assignment.

I want to make sure that I can find the corners of the plots later in the year when I establish the plots. Paint only lasts a week or two so a more permanent method is required. Use a pair of fixed points (irrigation heads or trees or fence posts) to locate the corners of the plots. Also make a map of the experimental area and record the treatment plan and distances on it.

Conclusion

So in my example trial will have five treatments and be replicated three times for a total of fifteen plots. Each plot will be five feet long by three feet wide. The total area required is 225 square feet. During the dry down cycles I will collect data daily and will be hand-watering plots until the most drought tolerant plot shows foot printing. I expect that this will take 45 minutes to an hour per day or from five to seven hours per week. I would expect that the original layout of the trial would take two people about five hours. Soil and root sampling at the end of the trial may take four more hours.

Conducting trials can be of great benefit to you. It gives you the opportunity for your own independent assessment of new products or techniques. If you choose to participate in the new WCTA program your experiences will also benefit other members of the association. In any even taking some care in designing your trial will help to ensure that the trial will be a success.

David Davidson is the Chair of the Turf Management Department of Kwantlen University College, Langley Campus.

Reprinted from the February / March 1997 issue of the Turf Line News.