We applied this
calculation to the full set of species in each community (Fig. 3),
and to two subsets (conifer and broadleaf specialists). To test for
possible bias owing to different sample sizes in 2010 and 2011,
we repeated all tests and included only logs that were sampled in
both years. We also tested for influence of rare species by restricting
the analyses to species occurring in at least 10 samples. Neither
test changed the results; thus, we only show results based on
our full data set.