Results You Can Use - Tips for Improving Plant Trials
Ever tried something new on your farm and weren’t sure whether or not it worked? Conducting plant trials in a way that yields useful and informative results is not an easy task. Even in a controlled environment setting there is still a wide range of variables that can easily impact plant growth and muddy results. Thankfully, by implementing a few simple practices you can reduce the variability in a plant trial and improve the quality of your data.
Accounting For Environmental Variability
Environmental factors such as temperature, lighting, and humidity can all have a significant impact on plant growth and yields. When conducting a plant trial you want to pick regions of the greenhouse that experience similar environmental conditions to eliminate variability. It is best to avoid conducting plan trials around the edges of your greenhouse as these regions experience “edge effects”, which are environmental conditions that are not reflective of the main area of the greenhouse. Additionally, data generated from continuous or semi-continuous environmental monitoring of each trial area can be used to determine if there is a correlation between an increase in yields and environmental variability.
Our team recommends conducting a baseline test to establish the normal variability between testing regions before conducting any plant trials. Establishing baseline levels for the normal variability within and between trial groups enables your team to more accurately assess the effect of any treatment tested.
Trial Tip: Bimodal growth is a key indicator of overcrowding as larger plants overshadow their smaller neighbors. Bimodal results will appear as two distinct peaks on a yield histogram.
The Human Factor
Whether it be differences in measuring techniques or choosing which plants to harvest, humans can have a big impact on trial results. To help account for the human factor ensure your team has clear instructions on how and when results are measured. Measurement devices should be calibrated and checked to confirm they are outputting the same results. Additionally, team members should agree upon how many decimal places they are recording and where on each plant they are taking measurements for traits such as internodal distance. Whenever possible it is best to have the same individual measure all samples of one variable type for the entire experiment to improve consistency.
Trial Tip: Have team members transfer their data to a shared digital file at the end of each day to ensure trial results don’t get lost.
Experimental Design
While the optimal experimental design from a statistical point of view may not always be practical to implement, there are four key questions you should keep in mind when designing a trial.
Does each trial group contain enough plants to account for the natural variability that is inherent when growing plants? Biological replicates allow you to determine how well the results of a smaller trial will translate to the entire farm.
Are you collecting data in a manner where statistical significance can be calculated? Data must be collected on a per-plant basis to calculate the significance of the difference between groups.
Does the layout of the trial groups minimize the impact of environmental variability (for example, randomized block design)?
Are you measuring the data in a way that your downstream analysis of the results will yield a conclusion that provides actionable information?
Sometimes for practical reasons certain types of variability cannot be avoided. In these instances our team implements a strategy we call “consistent inconsistencies” to minimize variability between trial groups. Basically the theory is that as long as every trial group receives initial plants with the same size distribution then you will still be able to determine differences between the trial groups. For example, germinated seedlings for a trial are often not identically sized. To account for the size variation your team would want to make sure each trial group received the same number of small and large seedlings.
Conducting trials in a commercial setting requires balancing cost and practicality with thoroughness and optimization. Understanding the limitations of your trial design and the impact that will have on your ability to interpret the results is key. To do so we recommend starting with the statistical analysis you want to run and working backwards to ensure the trial will yield the data you need. Stay tuned for our blog post on statistical analysis for plant trials for more details on this topic.
Trial Tip: Germinating / ordering extra seedlings then removing outliers allows you to reduce biological variation and improve the statistical significance of your results.
Conclusion
While eliminating variability in plant trials is impossible, minimizing the impact of said variability is a worthwhile task. Thoughtful experimental design, good communication, and using environmental monitoring devices can go a long way to ensuring the success of any plant trial.
