I am taking measurements with Multispeq in a quite wide barley field where plants present erect leaves. I would like to know in which side of the plot should I stand to take the measurements, as one side is shaded and the other faces the sun, even if I try to measure in central hours when the sun is high. What would be more accurate?
I understand the part about not moving the leaf from its original position, but the values will be different if I take measures from the right side (all leaves exposed to the sun) or if I go through the plots with the sun on my face (all leaves shaded). Would be enough taken into account in a mix model with sqrtPAR and timeofday?
Thanks in advance
Accuracy is not really an issue. It is really a question of what you want to know. Yes, the shading will play an important role and therefore taking observation under a variety of light conditions might be the best way for revealing differences across lines, treatments or field position. Keep in mind the values of phi2, ecst and others will change not only depending on plot shading but as a function of how cloudy/sunny it is on a given day.
I guess this is a long way of saying if you are going to take lots of measurements it is best to get them under the full range of conditions in your plot. You could aways add a question in your project about shade position to the project.
Hope that helps
This is a really great question, and we have been looking into the effects of this and other types of "user bias" in PhotosynQ experiments.
When used as a "big data" field instrument, the most powerful approach is to take as unbiased sampling of photosynthesis under the entire gamut of conditions the plant experiences, but also record these conditions so that analyses can figure out what happened. In this case, I would argue that you should randomize the direction of the measurements to avoid bias. Note that, in fields with north-south oriented rows, the effects of "row sidedness" will be compounded by such factors as time of day!
(Of course, you can always use the MultispeQ and PhotosynQ on "reductionist" experiments, in which lighting etc are highly controlled, e.g. in a growth chamber or greenhouse, and if the conductions are sufficiently simplified or even-ized it might be possible to use one side of the plant as representative. However, these environments may not really be relevant to the field!)
We are working on statistical methods to automatically detect these sorts of bias in experiments by analyzing factors that should not be biologically-determined. Some of the experiments we have analyzed even in controlled greenhouses, show unexpected biases. For example, in one experiments we found that the greenhouse lighting favored the control over the experimental plants, i.e. the light was higher in one part of the greenhouse room compared to the rest! This led us to suggest some statistical sub-sampling for the existing data, and recommend a different layout for the plots in future experiments. (If you have some good field data sets relevant to detecting such bias, and are willing/interested in participating in our efforts to detect and address these, please let me know!)
In any case, it is always better to avoid bias as much as possible during the measurements. To facilitate less biased data collection, we introduced the ability fo PhotosynQ App to guide the user using randomized sampling instructions. These can be uploaded using the .csv feature during the generation of projects.