GENOME model, the composition of the emergent communities, the rate of biochemical transformations that the microbes mediate, and the
resultant metagenome and transcriptome can
freely adapt to a changing environment.
The gene for ammonium transport into cells,
amtB, was broadly distributed in the model
(Fig. 5A), with a spatial pattern similar to that
of ecosystem biomass (Fig. 3), because this func-
tion is ubiquitous. Observations of amtB gene
concentrations along a gradient in the Amazon
River plume (17, 38) broadly matched the model;
however, the spatial scale of the observations
was smaller than the gradients achieved in the
model. amtB was also actively expressed in the
model (Fig. 5B), because recycling food webs
fueled by ammonium dominated in the modeled
subtropical and tropical Atlantic.
In contrast, a gene for aromatic compound
metabolism (pcaH, indicating aromatic ring cleav-
age) was observed at lower absolute abundance
than amtB, although it remained finite across
the sampling domain (Fig. 5C), and the model
projection was consistent with this pattern. Ob-
served pcaH gene expression is generally higher
in water with high organic matter content (Fig. 2D).
This resulted in higher transcript concentrations
in the model near river mouths, such as the
Amazon, Orinoco, and Mississippi, where aro-
matic compounds are introduced (Fig. 5D). Each
of the GENOME model runs developed a dif-
ferent community with diverse individual gene-
tic capabilities, but for which the community
transcriptomes and genomes were consistent
with the limited observations (Fig. 2 and figs.
S4 and S5).
At present, we lack mechanistic understanding of the cost-benefit trade-offs for marine microbial communities when carrying, expressing,
and regulating genes or synthesizing proteins,
and therefore we inferred these indirectly. Furthermore, some processes were included in the
model for which we have no identified genetic
pathway or marker genes (e.g., genes associated
with grazing). As quantitative observations of
gene abundance and expression expand, data-assimilative techniques to minimize the difference between modeled and observed gene copy
Fig. 6. Comparison
of metagenomic and
model runs along a
along the Amazon
River plume salinity
gradient in each
of four model runs in
June, clustered by
per cubic meter).
the Amazon River
gradient in each of
four model runs
in June, clustered
by similarity. (C and
D) Model run and
with each sample.
Genes are described
in table S3.