Biocrust community resistance to climate disturbance
The final stress of our two-dimensional NMDS analysis was 0.157,
indicating a good representation of the data. Community composition of
biocrusts in control and treatment plots did not show clear differences
(Figure 2). Neither distance nor direction of change from control to
paired treatment plot were consistent among pairs of treated and control
plots. We tested the magnitude and direction by using shifts in axis
coordinates on both axis 1 and 2. A one sample t-test in axis 1 did not
show a significant difference (p = 0.916). The same test for Axis 2
showed a significant shift toward the positive pole of the axis (p =
0.012), which correlates with the direction of the aridity gradient. Our
combined test of these two results (Fisher’s C = 9.04, p = 0.06)
suggested only a mild overall shift in direction and magnitude.
No site had significant TBI indices at the p<0.05 significance
level after correction for 25 simultaneous tests (Supporting
Information, Table S1). The overall dissimilarity was partitioned in the
component indices, where species abundance losses (B) represented 49.6%
of the dissimilarity and the species abundance gains (C) 50.6% (Table
S2, Table S3). The mean of the differences between gains and losses was
positive but not significant (p = 0.905, Table S4).
We observed positive linear slopes between control diversity and the
percentage difference dissimilarity index (R2 = 0.072, p = 0.110),
species abundance losses (R2 = 0.095, p = 0.078), and species abundance
gains (R2 = 0.091, p = 0.083), after excluding one outlier in every
model. However, multiple regression models showed that neither control
plot diversity, aridity, nor their interaction had a significant effect
on the overall dissimilarity of the biocrust communities or their
components (Table 2, Figure 3).