After reading through the sampling design tutorial and sampling the virtual forest, I had a better idea of what it was that I was sampling and how to go about doing it. I chose to study the woody plant species found on St. Patrick’s Island, using a 3m x 3m quadrat and a random sampling methodology. In order to select random plots, I used ArcGIS to create a grid over the whole sampling area (the island) and then the random function in Excel to choose a subset of sites. I then sampled the first five of those random polygons, on the west end of the Island. My predictor variables are Beaver herbivory (Y/N), flood frequency (2yr, 5yr, 8yr, 10yr, 20yr, 50yr, 100yr) and disturbance type (trampling, grazing, pathway, paved, altered). My response variables were % cover (by each species), and total woody spp cover. My hypothesis is that non-native woody species have a competitive advantage in habitats with lower flood frequencies.
I was hoping to catch some invasive woody species in my study area, but I will clearly need to sample a lot more sites to do that, as the invasives tend to occupy only select (altered) habitats where they have either been planted, or have invaded following some disturbance. I also found that several species were only present in one or two of my sites, while others were common at many. I will need to return and sample more quadrats to increase the sample size for this study. I think that I also need to revisit my random sampling methodology, and employ stratified random sampling so that I can have equal effort in each of the flood frequency strata. Looking at the sample points that I generated randomly, the following breakdown is what I had originally planned. As you can see, there are a disproportionate number of sites (14) in the 20 yr flood strata, with only 1 in each of the 50 and 100 yr strata.
Distributing sampling effort more evenly throughout the different flood strata and increasing sample size should improve my ability to make sound inferences on any potential relationship between this variable and my response variables.
very interesting! Nancy