Community overlap between sampling strategies
We examined the community overlap, that is, the total number and
identity of individual SVs irrespective of their relative abundance,
that were recovered from the milkweed population using both sampling
strategies. AMF had the lowest total SV richness and the highest
community overlap at 61% (Figure 3). Even when rarefying AMF at 700 SVs
as opposed to 7820 SVs, to more closely match rarefaction levels of
other groups, overlap was maintained at 61% between sampling
strategies. This indicated that it was not the higher rarefaction
numbers that caused the greater overlap observed in this group.
Sequencing rarefaction curves also showed sufficient sequence numbers to
capture a majority of sample richness, even at the lower rarefaction
levels. Bacteria and non-AM fungi in roots had a moderate overlap (34%
each) and foliar fungal endophytes had the highest SV richness and the
least overlap with just 10% of total SVs recovered by both sampling
strategies.
To explore the extent to which low-abundant SVs influenced the
differences between each sampling strategy, we gradually removed SVs
that were represented by <0.01% and then <0.05% of
total sequences (Supporting Information Table 1). This resulted in the
removal of SVs represented by fewer than 16, 2, 40, and 47 sequences
(<0.01%); and 78, 11, 201, and 235 sequences
(<0.05), for non-AM root fungi, bacteria, FFE and AMF,
respectively. At each removal step, the overlap of microbial communities
gradually increased (Table 1). After removing SVs that were represented
by <0.05% of all sequences, the AMF community overlapped 92%
between the two sampling strategies. Non-AM fungi in roots overlapped
63%. Bacterial and FFE SV composition, however, remained more different
than alike with just a 47% and 35% overlap, respectively. Since foliar
fungi showed the greatest differences, we then compared SV overlap after
clustering at 98.5% similarity and 90% co-occurrence. This coarser
clustering did not increase the overlap between sampling strategies (7%
as opposed to 10%), suggesting that at least for this group, the fact
that we chose to cluster at 100% similarity using DADA2 did not explain
the general lack of overlap.
Within individual plants, the number of SVs overlapping in both sampling
strategies was highly variable, particularly for all non-AM fungal
communities (Table 2). For foliar fungal endophytes, the percentage SV
overlap within a single plant ranged between 0% and 36%. Similarly,
for root bacteria, the percentage SV overlap ranged from 0% to 33%
among individual plants. Three A. speciosa individuals yielded
completely different communities between sampling strategies (0%
overlap in SVs) for foliar fungi. Root bacteria and non-AM root fungi
each had one plant with non-overlapping communities between sampling
strategies (Supporting Information, Fig. S3). Three and four additional
plants yielded just a single common SV between the two sampling
strategies for FFE and non-AM fungi, respectively, whereas 2 plants had
only 1 overlapping SV in root bacteria.