Bacterial networks
The association network analyses highlighted differences in community
complexity and interactions across sampling times and land cover types,
defined by node degree (the average number of edges connecting the
vertices), network size, and edge weight (i.e., whether the interactions
were positive or negative) (Figure 4). At the phylum level, bacterial
ASVs in the cleared plots had a higher node degree (total x̅ node degree
= 45.2) than the native plots (total x̅ node degree = 36) (t = -3.18, df
= 148, p = <0.01).
There were also clear differences in the node degree across sampling
times. The Mark Oliphant CP cleared plot had a steadily increasing node
degree from 00:00 through to 18:00. Moreover, the mean node degree was
typically higher for native plots between 00:00 and 06:00 (combined x̅
node degree =19.9) than between 12:00 and 18:00 (combined x̅ node degree
=16.1) (t = 2.22, df = 80, p = 0.02) but was generally lower for
cleared plots between 00:00 and 06:00 (combined x̅ node degree =18.4)
than between 12:00 and 18:00 (combined x̅ node degree =26.8) (t = -4.36,
df = 83, p = <0.01).
Evaluation of either positive or negative edge types highlighted
differences across land cover types with more negative associations
among ASVs in the native plots. One striking finding was that the mean
edge weight between 00:00 and 06:00 in the native Mark Oliphant CP plot
dropped from 0.88 (a high level of positive interactions) to 0.10 (a
high level of negative interactions).
At the genus level (Figure 5), bacterial ASVs in the cleared plots had a
higher node degree (total x̅ node degree = 27.6) than the native plots
(total x̅ node degree = 20.3), as expected following the phylum-level
analysis. There were also clear differences in the node degree across
sampling times; however, the patterns were variable. The native Kenneth
Stirling CP plot had a steadily decreasing node degree from 00:00
through to 18:00. The cleared Kenneth Stirling CP plot had a steadily
increasing node degree from 18:00 to 12:00. There were also differences
in edge weights across site types with more negative associations among
ASVs in the native sites (combined x̅ edge weight = 4.78) than cleared
plots (combined x̅ edge weight = 5.47). The striking difference in mean
edge weight between 00:00 and 06:00 in the native Mark Oliphant CP plot
was also visible in the genus-level network plots. Conversely, the mean
edge weight increased between native Kenneth Stirling CP plot at
00:00 and 06:00.
The evaluation of hub taxa (bacterial groups with the highest degree of
either positive or negative associations) showed that the keystone
bacteria at the genus level were different at each sampling time for
both the native and cleared plots, with positive degree ranging from 4
to 19 and negative degree ranging from 0 to 15 (Table 2). At the phylum
level, Acidobacteriota were the most common taxa with the highest number
of positive associations, occurring in native Kenneth Stirling and Mark
Oliphant CP plots at 18:00 and the cleared Mark Oliphant CP plot at
06:00 and 12:00. Bdellovibrionota were the top negative hub taxa,
featuring in native Kenneth Stirling CP 00:00, native Mark Oliphant CP
12:00, and cleared Mark Oliphant CP 12:00.