Ecological Archives C006-098-A3
J. Stephen Brewer. 2015. Competitive effects of fire-resistant saplings on their fire-sensitive neighbors are greater than the reverse. Ecosphere 6:255. http://dx.doi.org/10.1890/es15-00116.1
Appendix C. Reasons for the unbalanced design.
The experimental design was not balanced for several reasons. First, I excluded a small number of pairs within each plot because they were damaged by subsequent treefalls or, in a couple of cases, deer browsing. Second, in one instance at the burned site, I neglected to clip a pair that I had randomly assigned to the clipping treatment. Therefore, I considered it to be an unclipped pair in the analyses. Likewise, one oak-non oak pair within the unburned plot that I had assigned to the no-clip control was accidentally clipped initially. I therefore switched it to the clipping treatment. Third, there was initial mortality of target individuals or neighbors in some pairs. I excluded these from analyses of growth rate (but not topkill analyses). Subsequent mortality resulted in lower error degrees of freedom for year 2 analyses than for year 1 analyses. Finally, because I included all pairs of previously marked appropriately-sized saplings in the 10 x 30 subplots established by Cannon and Brewer (2013), there was no assurance of a set ratio of oak:oak, oak:non-oak, and non-oak:non-oak pairs. It just so happened that non-oak:non-oak pairs were underrepresented in these subplots.