I am amazed how often I visit a grower’s trial to pass comment and find that some fundamental error in the design does not allow for an accurate answer to be given.
When designing an experiment, it does not hurt to ask yourself, "Why am I doing this?" I know this sounds obvious, but it is so obvious that we often forget to ask it. The usual answer is, to see if a certain activity works as well as or better than something I am currently doing. Some examples are:
Does reflective cloth color fruit as well as summer pruning? or
Does this new powdery mildew spray work as well as the one I am –currently using?
Sounds simple, but by thinking about these examples we realize that we need to include at least three treatments: the new spray, the old spray, and doing nothing.
This final treatment—doing nothing—is the treatment most often left out of grower trials, but it is critical for an accurate answer to the –question: Does this work as well as what I am currently using?
For example, if we assess a trial and find that trees treated with the new spray had 1.0 percent of leaves with powdery mildew, compared with the current spray where 0.5 percent of the leaves were affected, does this mean that the new material did not work as well as the current spray? To answer this, we need to know how much mildew was in the orchard anyway. If the untreated trees had 0.5 percent of leaves with mildew, then there was insufficient mildew to draw any reliable conclusions from this trial. If the untreated trees had 40 percent of leaves with mildew, then both sprays are extremely effective and probably similar. Finally, if the untreated trees had 4 percent of leaves with mildew, then there is indication that the new spray might not be as good as the current spray, and statistics will be needed to verify if this is true. To know for sure, it is essential that there be a small collection of untreated trees within the trial.
So, now we have decided on our three (or more, if desired) treatments. The next usual problem is where to put them. This will be affected by the nature of the treatment and our desires, and may be single-tree treatments or large field trials. We will start by looking at small trials, where treatments are applied to a small number of trees, and then look at some practical ways of applying treatments to larger areas.
It is easy when starting experimentation to decide to place our three treatments near the edge of an orchard as shown in Figure 1 (below). This is because the trees can be easily identified and the treatments will not get in the way of everyday management practices. However, let’s stop to think about this. Where is codling moth damage worst? Where does the windbreak’s shade fall? Which trees are wind damaged?
Placing an experiment in an outside row is a bad idea and will usually lead to a doomed experiment or inconsistent results. Always place –experiments inside the orchard, well away from the edge, to avoid these problems and to ensure that the treatments are subjected to normal orchard practices and environment. If it is necessary to use outside rows, then leave at least three to four rows from the edge and three to four trees (20 to 25 feet) from the end of a row as buffers where normal management practices are employed.
So, now we have an experiment design as outlined in Figure 2 (below). Let’s think about this. What happens if there is a slight change in something as we go down the row? This might be in soil type, soil depth, irrigation output, cold air drainage, and so forth. Hence, the new treatment may be placed on trees that are more prone to the problem or on trees in a superior growing position. The way around this is to repeat the treatments on a second set of trees with the treatments in a different order. This repetition may be further up the row or in another row, as in Figure 3 (below). This design is fine if the results are easily apparent. However, if you are looking for a small difference—say a 10 percent improvement—then you need at least four replicates of the treatments (or seven replicates for two treatments).
Now that the trial is established, it is imperative that the site be marked so whoever is working in the block does not accidentally damage the trial. Not only should the whole site be marked, but the individual trees, so you know exactly which trees received which treatments, and, if repeat applications are needed,they are made to the same trees.
For larger trials, to simplify application, it is possible to use whole rows as a treatment. If it is a spray treatment, it is wise to spray three consecutive rows with the first product to avoid overspray and then the following three rows are sprayed with the alternative material across the entire block. Remember that an untreated control is essential, but it does not need to be entire rows. For this purpose, it is easy to turn the sprayer off for the final 30 to 50 meters (roughly 100-150 feet) for one of the treatments. This untreated group of trees only needs to occur on four of the rows to make our four entire replicates of the treatments. An example is shown in Figure 4 (below).
Finally comes the assessment. It is easy to grab a piece of paper and a pen and jot down some notes, so why not do it?
The information recorded will depend on the question being asked, but, often, visual assessments are all that are needed. It is best to try to keep assessments to numbers, for example a 0 to 10 visual score of tree performance, where 0 is terrible, 5 is just acceptable, and 10 is extreme pleasure with the performance of the tree.
These numbers need to be collected from each of the treatments within each replicate, so that, if necessary, we can determine the confidence that the results are real. Statistics using all the data can tell us if we are 90, 95, or 99 percent confident that the treatments are different, and this is useful when treatment means are similar. Do not hesitate to talk to a consultant about getting the numbers analyzed. He or she should be able to get this done for you.
After assessment, don’t forget to write up what has happened and the results of the treatments so you can refer back to them in later years.
Dr. Gordon Brown is director of Scientific Horticulture, Tasmania, Australia.