gene content (15). Except for these few pairs,
each cell carries at least one gene cassette not
found in any other. In some cases, a few closely
related cells (a subclade) within backbones share
a distinct gene cassette. Among these genes are,
again, glycosyltransferase genes, as well as transporters and genes involved in nucleotide binding
and processing. In a few cases, cells from different backbone subpopulations carry similar flexible gene cassettes [e.g., high-light–related genes
(Table 1) and phosphonate related genes], demonstrating the combinatorial nature of backbones
and flexible genes.
If backbone subpopulations have differential
fitness, we would expect their relative abundance
to change with changing environmental conditions (Fig. 1). Accordingly, the majority of the
largest subpopulations exhibited significant seasonal abundance variation (Fig. 4A), higher than
expected by chance (15), consistent with the
hypothesis that this reflects selection, but more
data are needed to draw that conclusion. Backbone
subpopulations maintain their genomic composition between seasons (tested for C1) (15), which
we would expect, as the establishment of new
mutations and the acquisition and loss of genes are
not likely to be in play on these time scales (15).
The congruency of genomic and ITS phylog-enies in Prochlorococcus at both coarse (4, 19)
and fine resolution (Fig. 2) suggests that ITS-ribotype clusters coincide, in most cases, with
distinct genomic backbones (15). This allowed us
to estimate the number of coexisting backbone
subpopulations in our samples through rarefaction analysis, revealing at least hundreds of coexisting subpopulations with distinct backbones
(Fig. 4B) in each sample. These backbone subpopulations are estimated to have diverged at least a
few million years ago (15), suggesting ancient,
stable niche partitioning. That they have different
alleles of genes associated with environmental
interactions, carry a distinct set of flexible genes,
and differ in relative abundance profiles as the
environment changes suggests strongly that they
are ecologically distinct.
Enormous population sizes and immense physical mixing probably played a role in the evolution
of diverse genomic backbones in Prochlorococcus.
A simple fluid mechanics model bridging the
micrometer and kilometer scales for a typical
ocean suggests that just-divided cells will be
centimeters apart within minutes, tens of meters
apart within an hour, and a few kilometers apart
within a week (15). Thus, Prochlorococcus pop-
ulations are expected to be well mixed over large
water parcels (~10 km2 area by 3 m depth) on ec-
ologically relevant time scales (~1 week) (15). This
mixing and a stable collective Prochlorococcus
population density of 107 to 108 cells liter−1 (17)
make the size of each backbone subpopulation in
such parcels enormous (>1013 cells) (15). The
effective population size is arguably close to
this census population size (15), implying that
Prochlorococcus evolution is governed by selec-
tion, not genetic drift [based on population gen-
etics theory (26)]. Consistent with this argument,
the difference in the observed FST distribution
from that estimated for no selection (Fig. 3B)
provides further evidence that the differentiation
of genomic backbones in Prochlorococcus is a
product of selection (15).
The correlation between phylogeny and flexible gene content (Table 1, tables S1 and S13, and
fig. S5) leads us to propose that the emergence of
a genomic backbone is initiated by the acquisition of a beneficial flexible gene cassette, followed by slow fine-adjustment of the core gene
alleles to the new niche dimension afforded by
the acquired cassette. Given the huge effective
population size, even extremely weak fitness differentials among alleles (27) can facilitate fine-adjustment of core genes (15) over the millions of
years of evolution after divergence.
The diverse set of hundreds of subpopulations
with distinct genomic backbones probably plays
an important role in the dynamic stability of the
Prochlorococcus “collective” in the global oceans
(fig. S6). Small fitness differentials, niche differentiation, and selective phage and grazer predation, in the context of temporal and spatial
environmental variation, help to explain their
coexistence (28, 29). On seasonal time scales, the
Prochlorococcus collective maintains a relatively
stable population size through temporal and local
adjustments in the relative abundance of backbone subpopulations (Figs. 1C and 4A and fig.
S6D). On longer time scales (decades to millions
of years), the collective may respond to shifting
selective pressures through the exchange of gene
cassettes between and within backbone subpop-
ulations, and through the evolution of the back-
bones themselves. The coherence of the collective
population holds as long as subpopulations do
not diverge to the point where they are no longer
able to exchange flexible genes and backbone
extinction and emergence rates are relatively bal-
anced. If Prochlorococcus backbone subpopula-
tions were designated as distinct species (30), it
would imply that the global collective is an as-
sortment of thousands of species. It is likely that
such a large set of coexisting subpopulations
with distinct genomic backbones is a character-
istic feature of free-living bacterial species with
very large population sizes living in highly mixed
References and Notes
1. F. Partensky, W. R. Hess, D. Vaulot, Microbiol. Mol.
Biol. Rev. 63, 106–127 (1999).
2. L. R. Moore, G. Rocap, S. W. Chisholm, Nature 393,
3. Z. I. Johnson et al., Science 311, 1737–1740 (2006).
4. G. C. Kettler et al., PLOS Genet. 3, e231 (2007).
5. J. Grote et al., MBio. 3, e00252-12 (2012).
6. D. E. Hunt et al., Science 320, 1081–1085 (2008).
7. S. L. Simmons et al., PLOS Biol. 6, e177 (2008).
8. H. Cadillo-Quiroz et al., PLOS Biol. 10, e1001265 (2012).
9. A. Gonzaga et al., Genome Biol. Evol. 4, 1360–1374 (2012).
10. R. T. Papke et al., Proc. Natl. Acad. Sci. U.S.A. 104,
11. S. Rodrigue et al., PLOS ONE 4, e6864 (2009).
12. T. Kalisky, P. Blainey, S. R. Quake, Annu. Rev. Genet. 45,
13. R. Stepanauskas, Curr. Opin. Microbiol. 15, 613–620 (2012).
14. R. S. Lasken, Nat. Rev. Microbiol. 10, 631–640 (2012).
15. Materials and methods are available as supplementary
materials on Science Online.
16. A. F. Michaels et al., Deep Sea Res. Part I 41, 1013–1038
17. R. R. Malmstrom et al., ISME J. 4, 1252–1264 (2010).
18. O. Wurtzel, M. Dori-Bachash, S. Pietrokovski,
E. Jurkevitch, R. Sorek, PLOS ONE 5, e15628 (2010).
19. M. Mühling, Environ. Microbiol. 14, 567–579
20. M. Nei, in Human Genetics, Part A: The Unfolding
Genome, B. Bonné-Tamir, T. Cohen, R. M. Goodman, Eds.
(Alan R. Liss, New York, 1982), p. 167.
21. R. Mehra-Chaudhary, J. Mick, L. J. Beamer, J. Bacteriol.
193, 4081–4087 (2011).
22. S. Avrani, O. Wurtzel, I. Sharon, R. Sorek, D. Lindell,
Nature 474, 604–608 (2011).
23. J. Pernthaler, Nat. Rev. Microbiol. 3, 537–546 (2005).
24. F. Malfatti, F. Azam, Aquat. Microb. Ecol. 58, 1–14 (2009).
25. U. Dobrindt, B. Hochhut, U. Hentschel, J. Hacker,
Nat. Rev. Microbiol. 2, 414–424 (2004).
26. J. F. Crow, M. Kimura, An Introduction to Population
Genetics Theory (Harper & Row, New York, 1970).
Table 1. Flexible gene cassettes associated with different cN2 backbone
subpopulations highlighting gene content that may contribute to eco-
logical differentiation. GT, glycosyltransferase; ABC-T, adenosine triphosphate–
binding cassette (ABC) transporter; HLIP, high-light–inducible protein; CO,
Cytochrome oxidase c subunit VIb; HlpA, histone-like protein; CpsL, poly-
saccharide biosynthesis protein. In the “Selected gene annotations” column,
numbers before gene annotation refer to number of that type of gene. A
complete list of the genes in each cassette is described in table S1 (15).
Clade Cassette ID Position No. of genes in cassette Selected gene annotations Cassette function
cN2-C1 CST_I Island 2.1 4 HLIP, CO Redox stress response
CST_II Island 4 7 3GT, ABC-T Outer membrane modification
cN2-C2 CST_II Island 4 7 3GT, ABC-T Outer membrane modification
cN2-C3 CST_III Island 1 2 2GT Outer membrane modification
cN2-C4 CST_I Island 2.1 4 HLIP, CO Redox stress response
CST_IV Island 4 14 3GT, HlpA, CpsL Outer membrane modification
cN2-C5 CST_V Island 4 5 2GT Outer membrane modification