Wednesday, August 29, 2012

On how I chose my samples - Part 2


So last I left off on the topic of choosing samples I was discussing how I realized that I really needed to have a good look around me and think about what I was trying to accomplish in order to help me come up with a sampling scheme.  And for the most part, that's just what I did.  During my field visits in the Summer and Fall of 2009, and even somewhat during field visits in 2010, I spent several hours standing in patches of Glossy Buckthorn thinking about how best to select plants to tag and follow.  Sometimes I would call my friend Adam (also studying plant ecology at the time) and bounce ideas off of him (or ask him to generate random numbers for me).  Sometimes I would call my wife and complain about the dense blackberry bushes, multi-flora rose, ticks, etc - ya know, sometimes you just need to whine a bit.

Here are a few of the standout observations that helped me develop my sampling scheme:

  • Glossy buckthorn patches vary in density and I wanted to avoid over-representing plants in areas of high density or under-representing plants in areas of low density
  • For the purposes of my study I want to follow individuals, therefore areas that are bare of Glossy buckthorn are not good places to look for plants to follow (yeah - this one is obvious)
  • Glossy buckthorn grows in varying ecological conditions (under story, gaps, riparian zones, uplands, etc.) and my samples should come from as many of these conditions as is feasible
  • The ecological conditions mentioned above are generally clumped. That is, the ecological conditions I'm interested in usually make up their own patch (e.g. a large patch of understory or wetland)
With these observations and thoughts in mind, I decide that the most important aspect of my sampling scheme should be that I select individual plants to follow across different plant densities and in different ecological conditions.  So here's the final sampling scheme I chose:
  • Ecological conditions - I selected a specific combination of ecological conditions (e.g. upland-understory forest) and using a GPS outlined the borders of a Glossy buckthorn infestation within the bounds of the borders of the ecological condition.
  • Individual plants - Once I had the border of an infestation, I selected 15 to 20 random points within these bounds where I set up 2x2m plots, or quadrats.  Within these plots, I tagged and measured all Glossy buckthorn plants that were taller than 10cm. Plants smaller than 10cm were considered saplings, which were sampled differently.
This type of sample scheme is essentially a stratified-random design, stratified across ecological condition and random within that condition.  So far I think it has worked out pretty well.

Random 2x2m plot in a large forest gap. Orange flags mark the corners of the quadrat.

No comments:

Post a Comment