Statistical Analysis
Proportional representation of OTUs across experimental groups was
calculated by means of two-way contingency tables for each comparison of
interest. Because DNA sequences derived from mid-succession control
soils were systematically shorter than sequences from all other groups,
these soils were excluded from our cross-succession comparison. This
left us with four comparisons of interest: 1) early vs. late succession,
2) alder vs. non-alder soils in late succession, 3&4) mid-succession
control vs. mid-succession fertilized soils (both N and P). Each
comparison was made using a separate contingency table, with statistical
significance of each bivariate relationship determined via likelihood
ratio tests using the ‘contingency analysis’ platform in JMP Pro v.
16.0. Statistical comparison of proportional differences in OTU
representation across comparison groups was conducted using the
‘categorical analysis’ platform in JMP Pro v. 16.0. Briefly, this
platform allows a user to define a set of pair-wise comparisons between
levels of a categorical explanatory variable. Differences in levels of
the categorical response variable are calculated, and statistical
significance determined, via likelihood ratio, Pearson chi-square, and
Fisher’s exact tests (JMP 2021). We based our comparisons on the latter,
as it consistently provided the most conservative results.