Blog Post 9: Field Research Reflections

I had originally envisioned random sampling across the entire park. In the end, it was much easier for me to implement my sampling via stratification. For one, the predictor variable I was working with (tree species composition) was fairly well divided into subsections. If I had relied on recording tree species composition for each individual sampling point, I would have had to employ a second sampling unit and a whole secondary methodology to determine which predictor class a given sample fell under. Given my relatively large sample size for the scope of the project (n= 60), it would have taken much longer to collect field data had this been my strategy.

QGIS was instrumental in automating my randomization. I had a few setbacks while trying to transfer data from QGIS to the limited software available for my GPS unit, but overall I think it was worthwhile to employ this strategy. I have used QGIS for a number of applications, including mapping species distributions using herbarium data, but never to implement sampling. It was nice to have an excuse to expand my GIS skillset.

One thing which was challenging about sampling was taking things from the digital realm to the field. From a satellite image or a shapefile it’s impossible to predict which areas will be too dense with brush to reach to sample or where there is standing water (although I didn’t run into the second problem in my data collection). I had a hard time trying not to incorporate subjectivity when I was forced to slightly move my sample site due to unforeseen obstacles. In the end, I decided to move 2m away in a random direction, but its hard to say how random that direction actually is when I have to consciously make the decision to choose a direction. It goes to show that even if you go into the field with fully randomized predetermined sample points, there is always some margin of human subjectivity that gets incorporated into your data.

Lastly, I definitely have a deepened understanding of the development of ecological theory. The pitfalls of trying to observe patterns in nature without accidentally incorporating your own bias toward patterning are prominent and hard to avoid. Like in all science, in ecological theory the importance of building upon previous knowledge and peer review is indispensable in rendering theoretical assertions universally applicable. Without multiple viewpoints, bias cannot be diminished to acceptable levels.