Legends Supplementary Material
Table S1
This table contains a simulated community containing 3 taxa (taxon 1-3) in two
different treatment (treatment x,y) in order to demonstrates how PP, TP, community
functional change and theoretical community functional change were calculated. We
constructed communities for 2 different scenarios: 1) no replacement of functions
among individual taxa, which results in identical values for the community functional
change and theoretical community functional change, and 2) replacement of functions
among individual taxa, which results in a lower value for the community functional
change compared to the theoretical community functional change and therefore
indicates functional stability. The degree of functional stability in the second scenario
is determined by the difference between the theoretical community functional change
and the community functional change.
Table S2
List of taxa used for calculations in the manuscript. For each taxon, information on
the phylogeny and abundance within treatments, average abundance of both
treatments within an experiment, fold-difference, PP (Bray-Curtis dissimilarities of
gene composition of size normalized taxa), BCfd (Bray-Curtis dissimilarities of gene
composition of non size-normalized taxa) and TP (BCfd corrected for the change of
sample size in non-size-normalized samples) is given. We further provide results from
a Blastn search of 16S rRNA gene sequences of the representative genomes against
16S rRNA gene OTUs sequenced in previous studies from the same samples (184 and
284 OTUs available for Exp_SwDi and Exp_CyDi; Landa et al. 2013a; Landa et al.
2013b). The 16S rRNA OTUs (410 bp) contain the variable regions V2/V3 and
reported similarity values to 16S rRNA gene sequences of reference genomes
underestimate similarities for the full length 16S rRNA gene. (Query ID: Genbank ID
for DNA sequence containing the 16S rRNA gene of the reference genomes; Subject
ID: ID for 16S rRNA OTUs from our samples). We also show the bin-specific
fractions of transcripts coding for ribosomal proteins (KEGG pathway ko03010),
proteins involved in oxidative phosphorylation (KEGG pathway ko00190), and RNA
polymerase rpoB (KEGG number K03043). These housekeeping genes should be
present and transcribed in all members of a community actively growing in
continuous culture via aerobic metabolism. A single taxon bin that did not express
transcripts coding for oxidative phosphorylation may be the exception, potentially
suggesting an alternate niche for anaerobic growth (e.g., embedded in particulate
material). Missing transcripts in some taxon bins for the rpoB gene may be caused by
low abundance of the taxon or binning errors.
Figure S1
To check the consistency with which metatranscriptomic data was assigned to
genome bins, we compared the transcriptome composition of 5 highly abundant taxa
(minimum 5% of all transcripts in at least one biological replicate) after normalizing
to the same number of reads by subsampling. Transcripts were subsequently clustered
using Bray-Curtis dissimilarity and Ward’s linkage. (Alc: Alcanivorax sp. DG881,
Col: Colwellia psychrerythraea, Gla: Glaciecola sp. HTCC2999, Nep: Neptuniibacter
caesariensis, IMC: gammaproteobacterium IMCC1989; for abbreviation for cultures,
see Table 1). Transcriptome bins from Exp_SwDi are labeled with filled circles
whereas those from Exp_CyDi are labeled with open circles. The dendrogram
demonstrates that each reference genome bin was more similar in transcript
composition across different replicates and treatments than when compared to the
other genome bins. The results provide evidence that at least among treatments within
one experiment, transcripts binning to a taxon were derived from genetically related
organisms.
Figure S2
Percent protein identity of transcripts to the reference genome for universal singlecopy genes rpoB (green) and recA (blue) for the three most abundant taxa in
communities of the control/cyano treatment (Exp_SwDi/Exp_CyDi, dashed line) and
diatom treatment (Exp_SwDi and Exp_CyDi, solid line). A single peak indicates that
a single organism was sampled for the gene and overlapping peaks for the same gene
across treatments indicates that sequences originating from highly related populations
were being compared.
Figure S3
Plot of PP versus the average abundance of taxa within each experiment. Regression
analysis suggests that the magnitude of PP is independent of taxon abundance (r2:
0.007, p:0.319). Introducing a weighting factor for taxon abundance therefore did not
introduce a bias when calculating the average weighted PP/TP.
Figure S4
Distribution of taxon bin-specific fractions of transcripts coding for ribosomal
proteins (KEGG pathway ko03010), oxidative phosphorylation (KEGG pathway
ko00190), and RNA polymerase rpoB (KEGG number K03043). Individual taxa with
high fractions of the targeted genes may reflect unique transcription patterns in this
taxon, but could also indicate binning errors.