I attempted to use Excel Version 2404 and RStudio software, with limited software knowledge. Some of the comparisons I hoped to make weren’t possible just because I couldn’t figure out the coding for the specific hypothesis tests. I ended up going with non-parametric methods, because assumptions for ANOVA parametric tests could not be met, as a check of the frequency distribution and data variance showed suspicious outlier data (ie. early seral site, for example a single 110cm DBH remnant cavity stump found in one early plot; one old plot encompassing 6 cavities) which could skew results. I also determined that median values for density were likely more representative of actual conditions, considering the outlier data.
I ended up creating multiple graphs to communicate the results of median and mean cavity differences between the class sites. I used a box & whiskers graph to show the variation and outlier data and then described my interpretation of the graphed results.
I did have trouble making multi-variable graphs and computing the hypothesis tests in RStudio. I didn’t find that the data revealed anything too unexpected. I was surprised at just how high median density was in the old class (37.5 nest cavities/ha), which was higher than the global mean density of 16/ha reported by a paper I cited. But I also realize my methods of surveying from ground were likely to cause over estimation of numbers, being that I could not confirm nest occupancy or internal conditions.
I went further into comparing the diversity of cavity features between the different classes by analyzing differences in height on tree and cavity size (cm2), and plotting the range of data points in another box and whisker plot. I liked how this graph portrayed the much broader choices of nests available in the mature and old class stands.