ArticlePDF Available

Abstract and Figures

Temperate bacterial viruses (phages) may enter a symbiosis with their host cell, forming a unit called a lysogen. Infection and viral replication are disassociated in lysogens until an induction event such as DNA damage occurs, triggering viral-mediated lysis. The lysogen–lytic viral reproduction switch is central to viral ecology, with diverse ecosystem impacts. It has been argued that lysogeny is favoured in phages at low host densities. This paradigm is based on the fraction of chemically inducible cells (FCIC) lysogeny proxy determined using DNA-damaging mitomycin C inductions. Contrary to the established paradigm, a survey of 39 inductions publications found FCIC to be highly variable and pervasively insensitive to bacterial host density at global, within-environment and within-study levels. Attempts to determine the source(s) of variability highlighted the inherent complications in using the FCIC proxy in mixed communities, including dissociation between rates of lysogeny and FCIC values. Ultimately, FCIC studies do not provide robust measures of lysogeny or consistent evidence of either positive or negative host density dependence to the lytic–lysogenic switch. Other metrics are therefore needed to understand the drivers of the lytic–lysogenic decision in viral communities and to test models of the host density-dependent viral lytic–lysogenic switch.
Content may be subject to copyright.
Variability and host density independence in
inductions-based estimates of
environmental lysogeny
Ben Knowles1*,BarbaraBailey
2, Lance Boling1, Mya Breitbart3, Ana Cobián-Güemes1,
Javier del Campo4,RobEdwards
2, Juris Grasis1, Andreas F. Haas1, Parag Katira8,
Linda Wegley Kelly1, Antoni Luque2,5,7, Jim Nulton2, Lauren Paul1, Gregory Peters1, Nate Robinett1,
Stuart Sandin9, Anca Segall1,7, Cynthia Silveira1, Merry Youle10 and Forest Rohwer1,7*
Temperate bacterial viruses (phages) may enter a symbiosis with their host cell, forming a unit called a lysogen. Infection
and viral replication are disassociated in lysogens until an induction event such as DNA damage occurs, triggering viral-
mediated lysis. The lysogenlytic viral reproduction switch is central to viral ecology, with diverse ecosystem impacts. It
has been argued that lysogeny is favoured in phages at low host densities. This paradigm is based on the fraction of
chemically inducible cells (FCIC) lysogeny proxy determined using DNA-damaging mitomycin C inductions. Contrary to the
established paradigm, a survey of 39 inductions publications found FCIC to be highly variable and pervasively insensitive
to bacterial host density at global, within-environment and within-study levels. Attempts to determine the source(s) of
variability highlighted the inherent complications in using the FCIC proxy in mixed communities, including dissociation
between rates of lysogeny and FCIC values. Ultimately, FCIC studies do not provide robust measures of lysogeny or
consistent evidence of either positive or negative host density dependence to the lyticlysogenic switch. Other metrics are
therefore needed to understand the drivers of the lyticlysogenic decision in viral communities and to test models of the
host density-dependent viral lyticlysogenic switch.
Lysogenic dynamics can disassociate viral infection and
production, leading to virushost predatorprey feedbacks, den-
sities and ecosystem impacts divergent from those predicted
under lytic dynamics13. Although the majority of cultured laboratory
and environmental bacterial strains are known lysogens4,5,quantify-
ing the fraction of lysogens in natural mixed communities remains
challenging. The prevalence of lysogeny is most commonly estimated
by using the DNA-damaging agent mitomycin C to induce
prophages (viruses that have established sustained intra- or extra-
chromosomal residence in their hosts) to enter the lytic cycle and
produce quantiable viral progeny6,7. Lysogeny has been diagnosed
using this technique in laboratory strains for half a century6,8.
However, treatment of lysogens can yield induction, unsuccessful
induction, or inhibition of host and viral production under different
mitomycin C concentrations (Table 1)6that vary on a strain-specic
basis9. Inductions are difcult to interpret, even under single-strain
laboratory conditions. Despite these challenges, in the 1990s Jiang
and Paul4,10 extended this induction method from laboratory strains
to bacterial isolates from mixed natural communities, showing that
lysogeny is a common viral strategy in the environment (2562.5%
of strains were lysogens). Higher percentages were commonly
observed under oligotrophic conditions than in eutrophic systems,
suggesting links between the rate of lysogeny, nutrient regime and
host density4,10.
Concurrent studies directly probed natural communities by
adding mitomycin C into samples of sea water11,12. In those
studies, the fraction of lysogenic cells (FLC, hereafter referred to
as the fraction of chemically inducible cells, FCIC, due to disassocia-
tion of lysogeny and the induced fraction, as a percent of total
cellular density) was estimated as
% FCIC =
C× 100 (1)
using viral densities in the induced (V
) and control (V
) treatments,
burst size Band host density C(cells per ml or g of sample) before
incubation1113. In contrast to earlier isolate-based studies, this mixed-
community approach showed that either FCIC was insensitive to
ecosystem nutrient status12 or was higher undereutrophic conditions11.
Subsequent research led to the consensus view that the frequency
of lysogeny was inversely related to host density and nutrient avail-
ability1417. This suggested that lysogeny provides a temporary
refuge for viruses when hosts are starving and scarce14, seemingly
providing a low-density lysogenic dynamic complementary to the
1Department of Biology, San Diego State University, 5500 Campanile Drive, San Diego, California 92182, USA. 2Department of Mathematics and Statistics,
San Diego State University, 5500 Campanile Drive, San Diego, California 92182, USA. 3College of Marine Science, University of South Florida, 140 Seventh
Avenue South, St Petersburg, Florida 33701, USA. 4Department of Botany, University of British Columbia, 3529-6270 University Boulevard, Vancouver,
British Columbia V6T 1Z4, Canada. 5Computational Science Research Center, San Diego State University, 5500 Campanile Drive, San Diego, California
92182, USA. 6Department of Computer Science, San Diego State University, 5500 Campanile Drive, San Diego, California 92182, USA. 7Viral Information
Institute, San Diego State University, 5500 Campanile Drive, San Diego, California 92182, USA. 8Department of Mechanical Engineering, San Diego State
University, 5500 Campanile Drive, San Diego, California 92182, USA. 9Scripps Institution of Oceanography, University of California San Diego, 950 Gilman
Drive, California 92903, USA. 10Rainbow Rock, Ocean View, Hawaii 96737, USA. Present address: Department of Marine and Coastal Sciences, Rutgers
University, 71 Dudley Road, New Brunswick, New Jersey 08901, USA. *e-mail:;
NATURE MICROBIOLOGY 2, 17064 (2017) | DOI: 10.1038/nmicrobiol.2017.64 | 1
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
modelled high-density lytic kill-the-winnerdynamic18. By contrast,
recent work using extensive eld direct counts, experiments and
viromes (c.f. chemical induction) suggests that lysogeny is favoured
when hosts are highly abundant and rapidly growing1.Wecallthis
model piggyback-the-winner. The association of high host densities
with high rates of lysogeny reects the benets of lysogeny to hosts,
including prophage-mediated immunity against further infec-
tion19,20, protection from protist predation via virulence factors,
and gain of metabolic functions2123.
Approximately 40 studies measuring rates of lysogeny in mixed
natural communities using mitomycin C induction have been pub-
lished in the twenty years since the pioneering work of Jiang and
Paul4,10 (Supplementary Table 1). We have compiled data from pub-
lished environmental induction studies and found that FCIC is highly
variable and seldom correlated with host density. Attempts to identify
the source(s) of FCIC variability revealed issues with estimating the
rates of lysogeny using mitomycin C induction. We propose that lyso-
geny and FCIC are not equivalent and that FCIC measurements
obscure relationships between host density and lysogeny.
Global distribution of FCIC studies. Clusters of saltwater studies
in the Gulf of Mexico (18% of all studies) and predominantly
freshwater studies in France (26%) account for 44% of all FCIC
studies (Fig. 1a). Of the 39 studies, most (90%, n= 35) were from
aquatic environments. Of the 352 published data points compiled,
91% (320 points) came from aquatic environments (compared to
32 points from sediment and soil combined).
Global host density versus published FCIC values. The full 352-
point global data set was analysed for a correlation between host
density and FCIC (Fig. 1b). This analysis showed high variability
and no signicant support for the established paradigm that
FCIC is inversely related to host density (P= 0.07, n= 352,
m(slope) = 0.08, R
= 0.01, 39 studies; linear regression with
FCIC and host density log-transformed). Type II regression
yielded similar results (95% condence interval (CI) [0.17, 0.01]
for the slope, m=0.08, n= 352, R
= 0.01).
Owing to the high variability between FCIC and host density,
468 further data points (that is, site averages) would be required
to nd any (positive or negative) signicant global-level relation-
ship between host density and FCIC (two-tailed power analysis,
power = 0.8, P< 0.05; Fig. 1b). This represents a 2.3-fold increase
in data over that generated in the two-decade history of the eld.
Host density versus published FCIC values by environment. A
high degree of variability and lack of signicant relationships was
also observed when FCIC was plotted against host density for
specic environments (Fig. 2 and Supplementary Fig. 1).
Regressions in freshwater, saltwater and sediment data sets
showed no signicant effect of host density on FCIC (Fig. 2;
freshwater: P= 0.16, n= 131, m=0.18, R
= 0.02, 15 studies;
saltwater: P= 0.76, n= 189, m=0.02, R
< 0.01, 22 studies;
sediment: P= 0.32, n= 19, m=0.10, R
= 0.06, 3 studies; linear
regressions with FCIC and host density log-transformed). The
only environment with a slope signicantly different from zero
was soil, where a positive relationship was observed (P< 0.01,
n= 13, m= 0.20, R
= 0.58, 2 studies; linear regressions with FCIC
and host density log-transformed). Global regression lines
explained 58, 6, 2 and <1% of the variability between FCIC and
host density in soil (2 studies), sediment (3 studies), freshwater
(15 studies) and saltwater (22 studies), respectively (Fig. 2). Type
II regressions conrm these results (freshwater: 95% CI [0.42,
0.07] for the slope, m=0.17, n= 131, R
= 0.02, 15 studies;
saltwater: 95% CI [0.14, 0.11] for the slope, m=0.02, n= 189,
< 0.01, 22 studies; sediment: 95% CI [0.29, 0.10] for the slope,
m=0.10, n= 19, R
= 0.06, 3 studies; soil: 95% CI [0.09, 0.32]
for the slope, m= 0.20, n= 13, R
= 0.58, 2 studies). Evaluation of
within-study variability expressed as median within-study ranges
in FCIC from each environment ranked sediment as least variable
(3.9% median range), followed by freshwater (12.60% median
range), seawater (26.62% median range) and soil (most variable,
36.55% median range) (Fig. 2).
Due to the high variability between FCIC and host density, 386,
15,450 and 110 further values (that is, site averages) are required to
nd any (positive or negative) signicant relationships between
FCIC and host density in freshwater, saltwater and sediment
environments, respectively (two-tailed power analysis, power = 0.8,
P< 0.05; no further values are required in soil) (Fig. 2). This rep-
resents approximately 4-fold, 83-fold and 7-fold increases in FCIC
sampling over extant data accumulated in the past two decades in
these environments.
Host density versus FCIC values within studies. Non-signicant
relationships (P> 0.05) were also the norm within studies (Fig. 3a,
Supplementary Table 1 and Supplementary Fig. 1). Of the
42 analyses (39 studies, three of which include samples from two
environments; Supplementary Table 1 and Fig. 3a), only ve
studies showed signicant relationships (11.90% of studies,
reported in refs 16, 17, 24 and 25 and the sediment subset of
ref. 26). While negative relationships (slopes) between FCIC and
host density are most common, 40% of freshwater, 32% of
saltwater, 33% of sediment and both soil studies showed positive
trends (linear regressions with FCIC and host density log-
transformed; Figs 2 and 3a and Supplementary Fig. 1). Aquatic
environments with more intensive sampling, either as individual
data points or as number of studies, have lower R
values than the
less-sampled sediment and soil environments (Supplementary
Table 1 and Fig. 2).
Distribution of published FCIC values. The most common
published FCIC values for global and freshwater, saltwater and
sediment environments are 05% FCIC. Pooled published FCIC
data show a truncated normal distribution centred approximately
around 0% FCIC. Commonly excluded FCIC values 0, possibly
generated by host inhibition via mitomycin C treatment (Table 1),
probably ll out the lower range (Fig. 3b).
Frequency of FCIC values 0. To assess the frequency and
distribution of FCIC values 0, FCIC was estimated in technical
Table 1 | Phenotypes observed when mitomycin C is added to a lysogen lineage in correct, under- and over-dose concentrations.
Change in host densities (relative to
Change in viral densities (relative to
range Conclusion
Decline (induction) Increase (induction) >0% Lysogeny found at correct
Under-dose No change (unsuccessful induction) No change (unsuccessful induction) 0% Lysogeny underestimated Type II
Over-dose Decline (inhibition) Decline (inhibition) <0% Lysogeny underestimated Type II
Under- and over-dosing leadsto unsuccessful and inhibitedinduction, respectively. Although thesedifferent dosage categories yield distinctive FCIC rangeswhen lysogens are probed, incorrectdosage of either
form consistently yields Type II error.
NATURE MICROBIOLOGY 2, 17064 (2017) | DOI: 10.1038/nmicrobiol.2017.64 |
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
replicates from three saltwater and three freshwater sites. Highly
variable FCIC values were observed within and between sites
(Fig. 4a). The within-site average FCIC values varied from 2.17
to 9.61% and the within-site ranges were up to 14.51%
(Supplementary Table 2 and Fig. 4a). The site with the highest
variability, Spanish Landing, showed both negative and positive
values (4.17 to 10.34%; coefcient of variation of 1.44), despite a
markedly consistent host density across replicates (5.09 × 10
6.55 × 10
cells per ml, coefcient of variation of 0.11; Fig. 4a).
Four sites showed higher variability between technical replicates
than was observed in the 25th percentile of published freshwater
and saltwater environmental studies (Fig. 5c; comparison of
ranges in each study or site). FCIC values 0 were observed in
four of six sites and spanned the full range of host densities,
indicating that unsuccessfulinduction events or host inhibition
are independent of host density and site (Fig. 4a).
Effects of excluding FCIC values 0. When all data points were
considered, half of all sites showed no signicant evidence of
lysogeny (that is, FCIC values with bootstrapped 95% CIs that
included zero: Famosa Slough, Lake Murray and Spanish Landing;
Fig. 4b). When values 0 were excluded, site means increased and
CIs constricted so as to give the appearance of signicant levels of
lysogeny at all sites (Fig. 4b). Similar to the published FCIC data
sets, most FCIC values from our survey fell between 0 and 5% in
a normal distribution centred around zero (Fig. 4c), indicating the
true distribution and extent of the excluded data in Fig. 3b.
Experimental manipulation and FCIC. To probe the potential
effects of host density and host growth rate on FCIC, we
estimated FCIC in communities in technical replicates diluted
either with nutrient-free buffers (low growth) or ltered site water
(high growth). It was expected that higher dilutions would lead to
Southern Ocean 12
Antarctica 15 16 17
1110 2824
31 32
Fraction of chemically inducible cells (%)
Cells per ml or g of sample
m = −0.08
P = 0.07
R2 = 0.01
n = 352
Studies = 39
Global regression summary statistics
Figure 1 | Locations and global linear regression of data from meta-analysis of 39 published studies based on chemical induction of lysogens.
a, Schematic map showing the global distribution of studies included in the meta-analysis, numbered by reference and coloured by environment
(Supplementary Table 1). Note that samples for study 1 were from both freshwater and sediment, for studies 17 and 21 were from freshwater and saltwater
(white circles), and for study 37 were from two distant marine environments. Blue circle with no number: freshwater and saltwater sampling sites for the
current study around San Diego, California, USA. b, Global synthesis of FCIC data across host density across environments (both axes log-transformed).
Black: global line of best t. Grey shading: global 95% prediction interval. Summary statistics (m,P,R
and n) are from linear regression across all data.
Map adapted from
NATURE MICROBIOLOGY 2, 17064 (2017) | DOI: 10.1038/nmicrobiol.2017.64 | 3
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
articially increased FCIC estimates due to a greater sensitivity to
mitomycin C (ref. 16) and increased growth rates27, especially in the
higher-nutrient-site water dilutions compared to buffer dilutions.
Neither dilution series yielded monotonic changes in FCIC;
instead, there was a dramatic and inconsistent increase in FCIC
variability (Fig. 5a,b) compared to the undiluted technical replicates
(Fig. 5c). Buffer dilutions of samples from Chollas Reservoir,
Famosa Slough, Lake Murray, Old Mission Dam, Spanish Landing
and Vacation Road sites showed ranges of 39.19, 16.17, 22.70,
19.58, 38.39 and 15.39% FCIC, respectively, and the corresponding
site water dilutions were 32.73, 24.03, 13.19, 16.88, 17.86 and 56.36%
FCIC (Fig. 5c). Variability was not consistent with either dilution or
diluent at most sites (Fig. 5a,b). Furthermore, variation (range)
within the diluted technical replicates from four of six sites
(buffer dilutions: Chollas Reservoir and Spanish Landing; site
water dilutions: Chollas Reservoir and Vacation Road) exceeded
that observed across sites in half of the published freshwater and
saltwater studies (Fig. 5c), suggesting that variability in local
dynamics equals the impact of broader ecological drivers on
FCIC. Compared to the correctundiluted sample, dilution can
drive up to 39.19% overestimates or 38.39% underestimates of lyso-
geny, with bias measured as deviation from the undiluted samples
(Fig. 5a,b; median change from undiluted samples ± 12.85% FCIC
with dilution). As above, FCIC values showed a normal distribution
during dilutions (Fig. 5d). A third of the FCIC values measured with
(Fig. 5a,b) and without (Fig. 4a) experimental manipulation were
0 (25 of 74 values, 33.78%) and thus would normally have been
excluded (Fig. 3b).
Host density as a driver of lysogeny in published data sets. The
relationship between FCIC and host density at the global
(Fig. 1b), within-environment (Fig. 2) and within-study (Fig. 3a)
levels was almost universally highly variable and non-signicant.
Only a few studies show increasing FCIC at low host
densities16,17,2426. In soils, the relationship between FCIC and host
density was constrained (R
= 0.58), signicant (P< 0.01) and
positive (m= 0.20), contrary to the current paradigm. Because
analysis of aggregated data sets is predisposed towards nding
signicant relationships28 and because we conducted regressions
without alpha values corrected for our legion analyses (for
example, 50 linear regressions), the lack of signicance here is
robust evidence against a global, environmental or within-study
level dependence of FCIC on host density. Increased sampling
was associated with increased variability at both the within-study
and across-study levels, even though the vast majority of studies
Fraction of chemically inducible cells (%) Fraction of chemically inducible cells (%)
Cells per ml or g of sample Cells per ml or g of sample
m = −0.18
P = 0.16
R2 = 0.02
n = 131
Studies = 15
m = −0.02
P = 0.76
R2 < 0.01
n = 189
Studies = 22
m = 0.20
P < 0.01
R2 = 0.58
n = 13
Studies = 2
m = −0.10
P = 0.32
R2 = 0.06
n = 19
Studies = 3
Figure 2 | Reported fractions of chemically inducible cells (FCIC, %) in freshwater, saltwater and sediment environments, showing a lack of negative host
density dependence, and soils, showing positive host density dependence (cells per ml or g of sample). Host density (cells per ml or g of sample) and
FCIC are both log-transformed. Black lines: across-study lines of best t (black). Grey shading: 95% prediction intervals. Summary statistics (m,P,R
and n)
are from linear regression across all data within each environment. Data points and associated linear regression lines of best t from individual studies are
colour-coded and do not show prediction intervals (see Supplementary Table 1 for details of each study). Boxplots show the distribution of ranges from
individual studies within each environment (stars colour-coded by study). Boxplots are shown in sediment and soil environments for consistency, despite the
low sample numbers. Boxes show the 25th, 50th (median) and 75th percentiles and whiskers range from minimum to maximum values. Data points from
each reference are plotted individually in Supplementary Fig. 1.
NATURE MICROBIOLOGY 2, 17064 (2017) | DOI: 10.1038/nmicrobiol.2017.64 |
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
focused on two geographic areas (Supplementary Table 1 and
Fig. 1). Ultimately, multiple decades to centuries of further
research would be required to obtain a reliable host density
dependence in FCIC (positive or negative), based on current rates
of data accumulation in the twenty-year history of the eld (as
shown by power analysis). This lack of support for the current
density-dependence paradigm invites the search for unidentied
drivers of environmental lysogeny.
The silent thirdof FCIC values 0. FCIC values 0 made up
one-third of the FCIC values (the silent third) we observed and
were (1) remarkably inconsistently distributed, with some
technical replicates showing inhibition while others did not, and
(2) ubiquitous across a range of host densities and sites (Figs 4a
and 5a). Because FCIC cannot logically be less than zero, these
frequent negative FCIC values at the community level show an
abiding disassociation of lysogeny and FCIC. Combined with the
capacity for excluding these values to skew estimates of lysogeny
(Fig. 4b), the distribution of FCIC values 0 suggests that there
may be pervasive bias in the characterization of lysogeny and its
drivers so far.
Origins of the silent thirdof FCIC values 0. Mitomycin C
induction is a highly dose-dependent approach. FCIC values 0
arise when induction treatment yields unchanged or lowered viral
densities (V
in equation (1), Table 1). While an unchanged
viral density (FCIC = 0) may accurately reect the absence of
lysogens if mitomycin C is correctly dosed, it may also indicate
unsuccessful induction (that is, an absence of induction whether
lysogeny is present or not) if mitomycin C is underdosed, yielding
a false negative (Type II error, Table 1). FCIC values <0 (V
are more consistent with mitomycin overdose causing inhibition
than unsuccessful induction6(Table 1), although these silent
thirdof values are commonly treated similarly to FCIC = 0 values
and excluded as unsuccessful induction13. Although mitomycin C
dosage may be tailored in single-strain systems, when dosing
mixed communities an unknown proportion of organisms or
lineages may be underdosed, overdosed or correctly dosed at any
given dose11. Although it is unclear what any FCIC value really
means, the observed variability in FCIC may be indicative of the
taxonomic or physiological state of organisms in a given sample29,30.
Induction in mixed communities versus isolates. Although
lysogeny appears to be a variable phenomenon, the majority of
sequenced bacterial genomes contain prophages5, and previous
investigations reported 43% (ref. 10), 2562.5% (ref. 4) and 71%
(ref. 27) of bacterial isolates to be inducible with mitomycin C.
Two-thirds of bacterial viruses with known lifestyles and
sequenced genomes are temperate31. In comparison with these
gures, the 05% mode of FCIC values (Fig. 3b) suggests that
mitomycin C induction underestimates community-level lysogeny
relative to strain-level analyses, an effect probably masked by
exclusion of the silent third(Fig. 4b). Inclusion of the silent
third, combined with the use of more diverse inducers10,32,33 and
further investigation of the linkages between lysogeny and
mitomycin C induction, as well as host density13, identity30 and
growth rate27, may reconcile FCIC and lysogeny.
Stochastic effects of dilution on FCIC. We induced lysogens in
diluted samples to probe the potential systematic impacts of host
density on FCIC estimation in slow- (buffer diluent) and
fast- (site water diluent) growing communities. Mitomycin C was
added simultaneously with diluent, precluding changes in
lysogeny during induction. Regardless of diluent, the most
marked effect of dilution was a large and inconsistent increase in
the variability in FCIC between replicates compared to undiluted
replicates or published studies (Fig. 5a,b versus Fig. 4a). Dilution
can result in lowered taxonomic diversity, but with variable
changes in functional capability3437, making stochastic functional
changes in microbial communities with dilution a candidate for
further investigation.
Dilution stochasticity and viral production assays. When dilution
is used to derive estimates of lytic and lysogenic viral production,
rates of lysogeny are assumed to remain unchanged (for example,
ref. 38). However, if the variation in FCIC with dilution observed
here (Fig. 5) is real, changing rates of lysogeny during dilution
will methodologically bias dilution-based estimates of viral
0 5 10 20150 5 10 2015
0 5 10 2015 0 5 10 2015
Studies showing relationship
Studies showing relationship
Relationship (slope)
Relationship (P < 0.1) Total studies in environment
Relationship in global analysis
Fraction of chemically inducible cells (%)
Figure 3 | Relationships between FCIC and host density at the within-study level and the truncated distribution of published FCIC values. a, Summary of
the number of positive, negative and at relationships observed in individual studies within freshwater (n= 15), saltwater (n= 22), sediment (n=3)and soil
(n= 2) environments. Black bars: signicant relationships. Insuff: insufcient data, studies with n< 3. Grey bars: slopes. Stars: 95% condence level drawn
from the environment-level linear regressions shown in Fig. 2. Dashed vertical lines: total number of studies in that environment. b, Stacked histogram of
FCIC values (in 5% bins) with the number of observations for each bin, colour-coded by environment (n= 352). Data points from each reference are plotted
individually in Supplementary Fig. 1.
NATURE MICROBIOLOGY 2, 17064 (2017) | DOI: 10.1038/nmicrobiol.2017.64 | 5
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
production. By contrast, if rates of lysogeny remain constant during
dilution, then the observed variability reveals a divergence of actual
and estimated lysogeny (FCIC). Pairing FCIC and dilution-based
viral production estimates in future studies may characterize the
conditions that drive disassociation of FCIC from lysogeny.
The lurking variablethat drives FCIC. Although FCIC was not
driven by host density, the FCIC distribution was constrained and
consistent across published studies (Fig. 3b), undiluted samples
(Fig. 4c) and diluted samples (Fig. 5c). FCIC thus appears to be a
non-random parameter possibly driven by environmental
condition(s). Observations in this study suggest that the lurking
variable(or variables) (1) varies signicantly between sites,
unrelated to abiotic conditions such as temperature and salinity;
(2) does not vary in soil but is variable in particulate sediment
and aquatic environments; (3) varies inconsistently with dilution;
(4) is conserved across a range of microbial densities; and
(5) does not correlate with host density or growth rate. Many
variables t this description, including community taxonomic,
metabolic and functional composition (for example, refs 29, 30),
possibly varying by environment. While techniques such as
metagenomics provide both community proling and the
identication of prophage elements39,40, observational approaches
alone cannot identify which variables determine the prevalence of
lysogeny. Here, we have combined observational and experimental
approaches to capture the variability of FCIC and prole its driver(s),
while eliminating canonical determinants such as host density, to
inform future studies using mitomycin C inductions.
We have examined the evidence of host density dependence in
inductions-based studies to determine whether the paradigm that
lysogeny is promoted at low host density is generally supported,
or whether the recently proposed piggyback-the-winnermodel of
high-density lysogeny1is actually consistent with inductions data
sets. Rather than supporting either model, the analysis showed
high variability and a lack of density dependence (with the excep-
tion of soils, where piggyback-the-winnerwas supported) in
FCIC estimates.
No transitionas suggested by earlier FCIC studies and lytic
modelling effortsfrom low-density lysogenic to high-density
lytic dominance is supported. The observed issues hinder the use
of FCIC in describing the lyticlysogenic switch and may have
impeded earlier attempts to understand patterns of environmental
lysogeny (for example, ref. 43) while aiming to resolve the broadly
observed sublinear power-law relationship between viral and host
densities1,4143. Despite high variability, FCIC values showed mark-
edly constrained distributions when plotted against host density,
Lake Murray
Chollas Reservoir
Old Mission Dam
Famosa Slough
Vacation Road
Spanish Landing
Cells per ml (×106)Cells per ml (×106)
5 7.5 10 25 505 7.5 10 25 50
Fraction of chemically inducible cells (%)
Fraction of chemically
inducible cells (%)
−10 0 10 20
Means ± 95% CIs with or
without FCIC values 0
Fraction of chemically inducible cells (%)
All sites pooled
Values > 0
All values
Figure 4 | FCIC variability between technical replicates, within sites and between sites, with signicant impacts of excluding values 0inmostsites.
a, FCIC from technical replicates (n= 4) across three freshwater and three saltwater sites (colour-coded by site; host density log-transformed). Grey shaded
areas: stars show ranges of FCIC values in each environment. b, Comparison of FCIC means and condence levels with (unshaded area; n=4for all sites) or
without (shaded area; n= 2 in Lake Murray samples; n= 3 in Chollas Reservoir, Spanish Landing and Famosa Slough sites; n=4 in Old Mission Dam and
Vacation Road sites) FCIC values 0. Squares: means. Whiskers: bootstrapped 95% CIs. Colour-coded by site, with global mean ± 95% CIs also shown
(black). Asterisks: sites where exclusion of FCIC values 0signicantly alters evidence of the presence or absence of lysogeny (P< 0.05; 95% CIs do or do
not cross zero). c, Histogram (5% bins) of the distribution of replicate FCIC values from these sites, showing an approximately normal distribution centred
around the 05% FCIC bin (n=24).
NATURE MICROBIOLOGY 2, 17064 (2017) | DOI: 10.1038/nmicrobiol.2017.64 |
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
indicative of a Simpson paradox in which unexamined variables are
driving FCIC (ref. 44). Probing of this variability revealed profound
artefacts resulting from exclusion of FCIC values 0 in published
studies and also allowed proling, but not identication, of the
possible driver. Although it is currently unclear what FCIC values
mean, the variability in FCIC reported here may guide future
studies elucidating what now appears to be a nascent driver of
viral dynamics.
Fraction of chemically inducible cells (%)
Buer diluent
Estimated cells per ml (×106) Estimated cells per ml (×106)
Estimated cells per ml (×106) Estimated cells per ml (×106)
Fraction of chemically inducible cells (%)
Famosa Slough
Lake Murray
Chollas Reservoir
Vacation Road
Spanish Landing
Old Mission Dam
−20 0 4020
Fraction of chemically inducible cells (%)
Published (across sites)
Range in fraction of chemically
inducible cells (%)
Buer site water
Within site (dilutions)
Within site (no treatment)
Site water diluent
Site water
Figure 5 | Effect of dilution on variability in FCIC. a,b, FCIC plotted against cell density following dilution with 0.02-µm-ltered buffer (a)or
0.02-µm-ltered site water (b). Dashed heuristic arrows: lines of best t across technical replicates, colour-coded by site, pointing towards more diluted
samples. Black rings: correctundiluted samples, n= 1 for each data point shown. c, FCIC ranges from published studies and both undiluted (no treatment)
and diluted samples from this study (n= 4 for each range from this study; sample numbers for each published range are provided in Supplementary Table 1).
Boxplot: FCIC ranges from individual published studies (grey stars) from freshwater and saltwater environments. Dashed lines: extensions of published median
value and 25th and 75th percentiles across the undiluted and diluted samples analysed here. The majority (66%) of undiluted technical replicates showranges
in the 25th to 50th percentiles of published studies using biological replicates and addressing FCIC across sites. d, Histogram (5% bins) of the distribution of
replicate FCIC values from the dilution experiments, showing an approximately normal distribution centred around zero (buffers, white; site water,black).
NATURE MICROBIOLOGY 2, 17064 (2017) | DOI: 10.1038/nmicrobiol.2017.64 | 7
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
Mining published values. Published values of the fraction of chemically inducible
cells (FCIC, %) were taken from 39 studies mined from Google Scholar that reported
host densities and corresponding FCIC values estimated using induction by the
addition of mitomycin C (Supplementary Table 2, Supplementary Fig. 1 and
Supplementary Source Data)11,12,1417,2426,29,30,32,4570. Google Scholar hits to the
search mitomycin C induction lysogenywere mined to the tenth page of the search
engine results. Supplementary probing of papers citing Jiang and Paul10 showed a
high degree of redundancy to the Google Scholar approach and was considered
evidence that the majority of papers satisfying the SYBR and mitomycin C
methodological criteria had been retrieved. Where necessary, values were extracted
from gures using the webplotdigitizer Chrome extension tool (
WebPlotDigitizer/app/). Most papers did not report FCIC if induced viral densities
did not exceed untreated viral densities, as this would yield FCIC estimates 0,
which are interpreted as unsuccessful induction13. Published FCIC data sets were
therefore analysed with values 0 excluded to ensure consistent analysis. All
within-study and global across-study linear regressions were conducted using data
at the greatest resolution possible. Individual measures were used whenever
accessible, and mean values otherwise (Supplementary Table 1). Where FCIC
estimates were provided using minimum and maximum burst sizes (for example,
refs 12,30), minimum values were used.
FCIC estimation. Mitomycin C from a stock less than 1 month old, suspended in
Sigma water (Sigma-Aldrich) was added to environmental samples to a nal
concentration of 1 µg ml
, consistent with most environmental induction studies
(Supplementary Source Data) and as prescribed for high-density near-shore
samples13. Sigma water was substituted for control samples to ensure a similar
dilution to induced samples. Samples, with corresponding mitomycin-C-negative
controls, were then incubated for 1824 h at room temperature in the dark, and viral
densities were then compared between mitomycin-C-positive treatments and
mitomycin-C-negative controls. FCIC values were then calculated from equation
(1)1113 using a burst size of 25 to approximate the median value of burst sizes used
in published studies that did (median burst size 22.5) and did not (median burst size
30) vary burst sizes for each sample, respectively (Supplementary Source Data).
In situ studies. The following freshwater and saltwater sites around San Diego,
California, USA (Supplementary Table 2) were sampled with sterile 50 ml
polypropylene tubes: Famosa Slough, Spanish Landing, Vacation Road, Chollas
Reservoir, Old Mission Dam and Lake Murray. Samples were stored in the dark at
room temperature and processed within 2 h of collection. Subsamples were
aliquoted into 2 ml technical replicates in 24-well plates (technical replicates are
ecological pseudoreplicates as they are intentionally not independent samples;
Corning). The three technical replicate fractions required to estimate FCIC (initial
cell counts, nal mitomycin-C-positive viral counts and nal mitomycin-C-negative
control viral counts) for each technical replicate were transferred within one syringe-
draw to keep technical replicate fractions as coupled as possible. Initial counts were
then conducted, and mitomycin-C-positive and mitomycin-C-negative samples
treated. Mitomycin-C-positive and mitomycin-C-negative samples were segregated
in different plates to preclude airborne antibiotic impacts on controls. After
incubation for 1820 h (ref. 13), 1 ml samples were drawn from each 2 ml well, xed
with paraformaldehyde (2% nal concentration) for 30 min, and then ash-frozen
in liquid nitrogen1,71. Samples were thawed at room temperature immediately before
staining with 2× SYBR Gold nucleic acid stain (Life Technologies) for 30 min and
ltered onto 0.02 µm Anodisc lters (Whatman)1,71. Filters were mounted on slides
and imaged on an Olympus ×100 object magnication oil-immersion microscope,
and counts were conducted using Image Pro software (Media Cybernetics), with
observers blind to the sample identity until statistical analysis.
Dilution experiments. Sites sampled in the observational study were resampled for
manipulative dilution experiments. All sampling and aliquoting procedures were the
same as above. However, rather than aliquoting 2 ml of subsampled site water into
each well of the 24-well plates, technical replicates were diluted to produce
undiluted, 25, 50 and 75% dilutions of unltered water samples (undiluted, 75, 50
and 25% unltered site water, respectively). This was done by adding either
0.02-µm-ltered buffer (buffer dilutions, Fig. 5a; articial seawater buffer (Tropic
Marin, Germany) at saltwater sites or Hydra Media buffer72 at freshwater sites) or
0.02-µm-ltered site water (site water dilutions, Fig. 5b). Host densities in each
dilution were estimated as host density in undiluted samples when mitomycin C was
added multiplied by the dilution factor. Both articial seawater and freshwater
diluents maintain chemical buffering of site water, but do not contain nutrients to
sustain or enhance host growth. As a result of this, host density was diluted, with
probable concomitant simplication of community composition and elevation of the
mass action dose of inducing agent per cell present, while the rates of lysogeny and
host growth rate were presumably unchanged. Dilution with site water dilutes host
densities but does not change nutrient availability, allowing host growth to increase in
proportion to dilution. Undiluted samples (0% dilutions) were considered correct
estimates of FCIC for comparison, because they are equivalent to samples typically
processed for published FCIC studies. Mitomycin C addition, incubation, sample
xation, storage and processing were thesame as above, but slides were imaged using
an Olympus ×60 objective magnication oil-immersion microscope.
Statistical analysis. All statistical analyses were conducted with a conventional
a priori alpha of 0.05 (95% condence). Despite the large number of analyses
conducted herein (for example, 50 linear regressions), no correction was applied to
the alpha to facilitate identifying signicant relationships between FCIC and host
density, even at risk of incurring Type I error. Linear regressions were conducted
using the lm() function in R (reported in the plots in Figs 13). These were
complemented by ordinary least squares (OLS) Type II linear regressions using the
lmodel2() function of the lmodel2 R package across all data and across freshwater,
saltwater, sediment and soil environments, to ensure variability in host density
measurements was not skewing our analyses. Published data sets showed a negative
relationship between nand R
, suggesting that the analysis of means, as the majority
of data available for this study, facilitated nding relationships between FCIC and
host density (Supplementary Table 1). Power analysis using a test for correlation
were conducted using published data sets (Supplementary Source Data) with the
pwr.r.test() function in R to estimate the sample sizes needed to obtain a power of
0.8 with a signicance level of 0.05 in a two-tailed test. As the FCIC values generated
here were intentionally not independent (that is, they were technical, not biological,
replicates), a conservative non-parametric bootstrapping approach was used to
generate means and 95% CIs (Fig. 4). These parameters were estimated using the
boot_out() function in R with 10,000 iterations on data with all FCIC values
included as well as with FCIC values 0 excluded, consistent with the literature.
Data availability. Published data subjected to meta-analysis are provided in the
Supplementary Source Data. These data and experimental data sets are also available
from the corresponding author upon request.
Received 15 November 2016; accepted 22 March 2017;
published 28 April 2017
1. Knowles, B. et al. Lytic to temperate switching of viral communities. Nature 531,
466470 (2016).
2. Wang, Z. & Goldenfeld, N. Fixed points and limit cycles in the population
dynamics of lysogenic viruses and their hosts. Phys. Rev. E 82, 011918 (2010).
3. Weitz, J. Quantitative Viral Ecology: Dynamics of Viruses and Their Microbial
Hosts (Princeton University Press, 2015).
4. Jiang, S. C. & Paul, J. H. Signicance of lysogeny in the marine environment:
studies with isolates and a model of lysogenic phage production. Microb. Ecol.
35, 235243 (1998).
5. Canchaya, C., Proux, C., Fournous, G., Bruttin, A. & Brussow, H. Prophage
genomics. Microbiol. Mol. Biol. Rev. 67, 238276 (2003).
6. Otsuji, N., Sekiguchi, M., Iijima, T. & Takagi, Y. Induction of phage formation in
the lysogenic Escherichia coli K-12 by mitomycin C. Nature 184,
10791080 (1959).
7. Roberts, J. W. & Roberts, C. W. Proteolytic cleavage of bacteriophage lambda
repressor in induction. Proc. Natl Acad. Sci. USA 72, 147151 (1975).
8. Ackermann, H.-W. & DuBow, M. S. (eds) Viruses of Prokaryotes (CRC Press, 1987).
9. Muschel, L. H. & Schmoker, K. Activity of mitomycin C, other antibiotics, and
serum against lysogenic bacteria. J. Bacteriol. 92, 967971 (1966).
10. Jiang, S. C. & Paul, J. H. Seasonal and diel abundance of viruses and occurrence
of lysogeny/bacteriocinogeny in the marine environment. Mar. Ecol. Prog. Ser.
104, 163172 (1994).
11. Jiang, S. C. & Paul, J. H. Occurrence of lysogenic bacteria in marine microbial
communities as determined by prophage induction. Mar. Ecol. Prog. Ser. 142,
2738 (1996).
12. Weinbauer, M. G. & Suttle, C. A. Potential signicance of lysogeny to
bacteriophage production and bacterial mortality in coastal waters of the Gulf of
Mexico. Appl. Environ. Microbiol. 62, 43744380 (1996).
13. Paul, J. H. & Weinbauer, M. Detection of lysogeny in marine environments.
Man. Aquat. Viral Ecol. 4, 3033 (2010).
14. Maurice, C. F., Bouvier, T., Comte, J., Guillemette, F. & Del Giorgio, P. A.
Seasonal variations of phage life strategies and bacterial physiological states in
three northern temperate lakes. Environ. Microbiol. 12, 628641 (2010).
15. Evans, C. & Brussaard, C. P. D. Regional variation in lytic and lysogenic viral
infection in the Southern Ocean and its contribution to biogeochemical cycling.
Appl. Environ. Microbiol. 78, 67416748 (2012).
16. Payet, J. & Suttle, C. A. To kill or not to kill: the balance between lytic and lysogenic
viral infection is driven by trophic status. Limnol. Oceanogr. 58, 465474 (2013).
17. Brum, J. R., Hurwitz, B. L., Schoeld, O., Ducklow, H. W. & Sullivan, M. B.
Seasonal time bombs: dominant temperate viruses affect Southern Ocean
microbial dynamics. ISME J. 10, 437449 (2015).
18. Weitz, J. S. et al. A multitrophic model to quantify the effects of marine viruses
on microbial food webs and ecosystem processes. ISME J. 9, 13521364 (2015).
19. Bondy-Denomy, J. et al. Prophages mediate defense against phage infection
through diverse mechanisms. ISME J. 10, 28542866 (2016).
20. Dedrick, R. M. et al. Prophage-mediated defence against viral attack and viral
counter-defence. Nat. Microbiol. 2, 16251 (2017).
21. van Houte, S., Buckling, A. & Westra, E. R. Evolutionary ecology of prokaryotic
immune mechanisms. Microbiol. Mol. Biol. Rev. 80, 745763 (2016).
NATURE MICROBIOLOGY 2, 17064 (2017) | DOI: 10.1038/nmicrobiol.2017.64 |
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
22. Breitbart, M. Marine viruses: truth or dare. Mar. Sci. Annu. Rev. 4,
425448 (2012).
23. Paul, J. H. Prophages in marine bacteria: dangerous molecular time bombs or the
key to survival in the seas? ISME J. 2, 579589 (2008).
24. Cochran, P. K. & Paul, J. H. Seasonal abundance of lysogenic bacteria in a
subtropical estuary. Appl. Environ. Microbiol. 64, 23082312 (1998).
25. Thomas, R., Berdjeb, L., Sime-Ngando, T. & Jacquet, S. Viral abundance,
production, decay rates and life strategies (lysogeny versus lysis) in Lake Bourget
(France). Environ. Microbiol. 13, 616630 (2011).
26. Bettarel, Y., Bouvy, M., Dumont, C. & Sime-Ngando, T. Virusbacterium
interactions in water and sediment of West African inland aquatic systems.
Appl. Environ. Microbiol. 72, 52745282 (2006).
27. Stopar, D., Černe, A., Žigman, M., Poljšak-Prijatelj, M. & Turk,V. Viral abundance
and a high proportion of lysogens suggest that viruses are important members of
the microbial community in the Gulf of Trieste. Microb. Ecol. 47, 18 (2004).
28. Robinson, W. S. Ecological correlations and the behavior of individuals.
Am. Sociol. Rev. 15, 351357 (1950).
29. Palesse, S., Colombet, J., Ram, A. S. P. & Sime-Ngando, T. Linking host
prokaryotic physiology to viral lifestyle dynamics in a temperate freshwater lake
(Lake Pavin, France). Microb. Ecol. 68, 740750 (2014).
30. Maurice, C. F. et al. Disentangling the relative inuence of bacterioplankton
phylogeny and metabolism on lysogeny in reservoirs and lagoons. ISME J. 5,
831842 (2011).
31. McNair, K., Bailey, B. A. & Edwards, R. A. PHACTS, a computational approach
to classifying the lifestyle of phages. Bioinformatics 28, 614618 (2012).
32. Cochran, P. K., Kellogg, C. A. & Paul, J. H. Prophage induction of indigenous
marine lysogenic bacteria by environmental pollutants. Mar. Ecol. Prog. Ser. 164,
125133 (1998).
33. Wommack, K. E. & Colwell, R. R. Virioplankton: viruses in aquatic ecosystems.
Microbiol. Mol. Biol. Rev. 64, 69114 (2000).
34. Franklin, R. B., Garland, J. L., Bolster, C. H. & Mills, A. L. Impact of dilution on
microbial community structure and functional potential: comparison of
numerical simulations and batch culture experiments. Appl. Environ. Microbiol.
67, 702712 (2001).
35. Philippot, L. et al. Loss in microbial diversity affects nitrogen cycling in soil.
ISME J. 7, 16091619 (2013).
36. Peter, H. et al. Function-specic response to depletion of microbial diversity.
ISME J. 5, 351361 (2011).
37. Roger, F., Bertilsson, S., Langenheder, S., Osman, O. A. & Gamfeldt, L. Effects of
multiple dimensions of bacterial diversity on functioning, stability and
multifunctionality. Ecology 97, 27162728 (2016).
38. Wilhelm, S. W., Brigden, S. M. & Suttle, C. A. A dilution technique for the direct
measurement of viral production: a comparison in stratied and tidally mixed
coastal waters. Microb. Ecol. 43, 168173 (2002).
39. Roux, S., Enault, F., Hurwitz, B. L. & Sullivan, M. B. Virsorter: mining viral signal
from microbial genomic data. PeerJ 3, e985 (2015).
40. Arndt, D. et al. PHASTER: a better, faster version of the PHAST phage search
tool. Nucleic Acids Res. 44, W16W21 (2016).
41. Clasen, J. L., Brigden, S. M., Payet, J. P. & Suttle, C. A. Evidence that viral
abundance across oceans and lakes is driven by different biological factors.
Freshwat. Biol. 53, 10901100 (2008).
42. Wigington, C. H. et al. Re-examination of the relationship between marine virus
and microbial cell abundances. Nat. Microbiol. 1, 15024 (2016).
43. Parikka, K. J., Le Romancer, M., Wauters, N. & Jacquet, S. Deciphering the virus-
to-prokaryote ratio (VPR): insights into virushost relationships in a variety of
ecosystems. Biol. Rev. Camb. Philos. Soc. 92, 10811100 (2017).
44. Simpson, E. H. The interpretation of interaction in contingency tables. J. R. Stat.
Soc. Ser. B 13, 238241 (1951).
45. Breitbart, M., Wegley, L., Leeds, S., Schoenfeld, T. & Rohwer, F. Phage community
dynamics in hot springs. Appl. Environ. Microbiol. 70, 16331640 (2004).
46. Colombet, J. et al. Depth-related gradients of viral activity in Lake Pavin.
Appl. Environ. Microbiol. 72, 44404445 (2006).
47. Drewes, F., Peter, H. & Sommaruga, R. Are viruses important in the plankton of
highly turbid glacier-fed lakes? Sci. Rep. 6, 24608 (2016).
48. Laybourn-Parry, J., Anesio, A. M., Madan, N. & Säwström, C. Virus dynamics in
a large epishelf lake (Beaver Lake, Antarctica). Freshwat. Biol. 58,
14841493 (2013).
49. Lisle, J. T. & Priscu, J. C. The occurrence of lysogenic bacteria and microbial
aggregates in the lakes of the McMurdo Dry Valleys, Antarctica. Microb. Ecol.
47, 427439 (2004).
50. Lymer, D. & Lindström, E. S. Changing phosphorus concentration and
subsequent prophage induction alter composition of a freshwater viral
assemblage. Freshwat. Biol. 55, 19841996 (2010).
51. Pradeep Ram, A. S. & Sime-Ngando, T. Resources drive trade-off between viral
lifestyles in the plankton: evidence from freshwater microbial microcosms.
Environ. Microbiol. 12, 467479 (2010).
52. Ram, A. S. P. et al. Variable viral and grazer control of prokaryotic growth
efciency in temperate freshwater lakes (French Massif Central). Microb. Ecol.
66, 906916 (2013).
53. Ram, A. S. P., Palesse, S., Colombet, J., Thouvenot, A. & Sime-Ngando, T.
The relative importance of viral lysis and nanoagellate grazing for prokaryote
mortality in temperate lakes. Freshwat. Biol. 59, 300311 (2014).
54. Tapper, M. A. & Hicks, R. E. Temperate viruses and lysogeny in Lake Superior
bacterioplankton. Limnol. Oceanogr. 43, 95103 (1998).
55. Bettarel, Y. et al. Virioplankton distribution and activity in a tropical
eutrophicated bay. Estuar. Coast. Shelf Sci. 80, 425429 (2008).
56. Bettarel, Y. et al. Ecological traits of planktonic viruses and prokaryotes along a
full-salinity gradient. FEMS Microbiol. Ecol. 76, 360372 (2011).
57. Bongiorni, L., Magagnini, M., Armeni, M., Noble, R. & Danovaro, R. Viral
production, decay rates, and life strategies along a trophic gradient in the North
Adriatic Sea. Appl. Environ. Microbiol. 71, 66446650 (2005).
58. Bouvy, M. et al. Uncoupled viral and bacterial distributions in coral reef waters of
Tuamotu Archipelago (French Polynesia). Mar. Pollut. Bull. 65, 506515 (2012).
59. Laybourn-Parry, J., Marshall, W. A. & Madan, N. J. Viral dynamics and patterns
of lysogeny in saline Antarctic lakes. Polar Biol. 30, 351358 (2006).
60. Long, A., McDaniel, L. D., Mobberley, J. & Paul, J. H. Comparison of lysogeny
(prophage induction) in heterotrophic bacterial and Synechococcus populations
in the Gulf of Mexico and Mississippi River plume. ISME J. 2, 132144 (2008).
61. Maurice, C. F., Bouvier, C., Wit, R. & Bouvier, T. Linking the lytic and lysogenic
bacteriophage cycles to environmental conditions, host physiology and their
variability in coastal lagoons. Environ. Microbiol. 15, 24632475 (2013).
62. Muck, S. et al. Fracture zones in the Mid Atlantic Ridge lead to alterations in
prokaryotic and viral parameters in deep-water masses. Front. Microbiol. 5,
264 (2014).
63. Nguyen-Kim, H. et al. Coral mucus is a hot spot for viral infections. Appl.
Environ. Microbiol. 81, 57735783 (2015).
64. Ortmann, A. C., Lawrence, J. E. & Suttle, C. A. Lysogeny and lytic viral
production during a bloom of the cyanobacterium Synechococcus spp. Microb.
Ecol. 43, 225231 (2002).
65. Weinbauer, M. G. & Suttle, C. A. Lysogeny and prophage induction in coastal
and offshore bacterial communities. Aquat. Microb. Ecol. 18, 217225 (1999).
66. Weinbauer, M. G., Brettar, I. & Höe, M. G. Lysogeny and virus-induced
mortality of bacterioplankton in surface, deep, and anoxic marine waters.
Limnol. Oceanogr. 48, 14571465 (2003).
67. Williamson, S. J., Houchin, L. A., McDaniel, L. & Paul, J. H. Seasonal variation in
lysogeny as depicted by prophage induction in Tampa Bay, Florida.
Appl. Environ. Microbiol. 68, 43074314 (2002).
68. Mei, M. L. & Danovaro, R. Virus production and life strategies in aquatic
sediments. Limnol. Oceanogr. 49, 459470 (2004).
69. Montanié, H. et al. Microbial interactions in marine water amended by eroded
benthic biolm: a case study from an intertidal mudat. J. Sea Res. 92,
7485 (2014).
70. Ghosh, D. et al. Prevalence of lysogeny among soil bacteria and presence of 16S
rRNA and trzN genes in viral-community DNA. Appl. Environ. Microbiol. 74,
495502 (2008).
71. Haas, A. F. et al. Unraveling the unseen players in the oceanaeld guide to
water chemistry and marine microbiology. J. Vis. Exp. 21, 116 (2014).
72. Grasis, J. A. et al. Species-specic viromes in the ancestral holobiont Hydra.
PLoS ONE 9, e109952 (2014).
The authors thank R. Young for microbiological insight and guidance. Canadian Institute
for Advanced Research Integrated Microbial Biodiversity Program Fellowship Award
141679, National Science Foundation grants OISE-1243541 and DEB-1046413, a Gordon
and Betty Moore Foundation Investigator Award GBMF-3781 (to F.R.) and National
Science Foundation grants OCE-1538567 (to L.W.K.), IOS-1456301 and DEB-1555854 (to
M.B.) funded this work. The authors thank G. Gueiros and K. Furby for critiquing
the manuscript.
Author contributions
B.K. and F.R. designed, conducted and wrote up the study. B.B., L.B., M.B., A.C.-G., J.d.C.,
R.E., B.F., J.G., A.H., P.K., L.W.K., A.L., J.N., G.P., L.P., N.R., S.S., A.S., C.S. and M.Y.
contributed data, analysis and manuscript preparation.
Additional information
Supplementary information is available for this paper.
Reprints and permissions information is available at
Correspondence and requests for materials should be addressed to B.K. and F.R.
How to cite this article: Knowles, B. et al. Variability and host density independence
in induct ions-based estimates of e nvironmental lysogeny. Nat. Microbiol. 2,
17064 (2017).
Publishersnote:Springer Nature remains neutral with regard to jurisdictional claims in
published maps and institutional afliations.
Competing interests
The authors declare no competing nancial interests.
NATURE MICROBIOLOGY 2, 17064 (2017) | DOI: 10.1038/nmicrobiol.2017.64 | 9
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
... A lytic or productive infection is characterized by an immediate replication and expression of the viral genome within the host cell, subsequently releasing new viral particles by cell lysis (5). In contrast, a lysogenic infection is characterized by integrating the viral genome into the host chromosome or as an episomal element, forming a new biological entity known as lysogen (5)(6)(7). This cycle does not produce viral particles immediately after infection; however, it can switch to a productive cycle depending on multiple factors, such as virus or host genetics, virus-host ratio, host physiological state, host density (quorum sensing), and environmental conditions (5)(6)(7)(8). ...
... In contrast, a lysogenic infection is characterized by integrating the viral genome into the host chromosome or as an episomal element, forming a new biological entity known as lysogen (5)(6)(7). This cycle does not produce viral particles immediately after infection; however, it can switch to a productive cycle depending on multiple factors, such as virus or host genetics, virus-host ratio, host physiological state, host density (quorum sensing), and environmental conditions (5)(6)(7)(8). ...
... Lytic viruses directly influence microbial community composition through predatorprey dynamics, in which the dominant or active taxa in the microbial community are selectively lysed, as described in the "kill-the-winner" ecological model (2,6,7,(9)(10)(11). Conversely, it has been proposed that in ecosystems with high nutrients and high microbial abundances, viruses will lysogenize the most metabolically active and sometimes dominant taxa in the microbial community following the "piggyback-the-winner" ecological model (6,7). ...
Full-text available
Viruses exert diverse ecosystem impacts by controlling their host community through lytic predator-prey dynamics. However, the mechanisms by which lysogenic viruses influence their host-microbial community are less clear. In hot springs, lysogeny is considered an active lifestyle, yet it has not been systematically studied in all habitats, with phototrophic microbial mats (PMMs) being particularly not studied. We carried out viral metagenomics following in situ mitomycin C induction experiments in PMMs from Porcelana hot spring (Northern Patagonia, Chile). The compositional changes of viral communities at two different sites were analyzed at the genomic and gene levels. Furthermore, the presence of integrated prophage sequences in environmental metagenome-assembled genomes from published Porcelana PMM metagenomes was analyzed. Our results suggest that virus-specific replicative cycles (lytic and lysogenic) were associated with specific host taxa with different metabolic capacities. One of the most abundant lytic viral groups corresponded to cyanophages, which would infect the cyanobacteria Fischerella, the most active and dominant primary producer in thermophilic PMMs. Likewise, lysogenic viruses were related exclusively to chemoheterotrophic bacteria from the phyla Proteobacteria, Firmicutes, and Actinobacteria. These temperate viruses possess accessory genes to sense or control stress-related processes in their hosts, such as sporulation and biofilm formation. Taken together, these observations suggest a nexus between the ecological role of the host (metabolism) and the type of viral lifestyle in thermophilic PMMs. This has direct implications in viral ecology, where the lysogenic- lytic switch is determined by nutrient abundance and microbial density but also by the metabolism type that prevails in the host community.
... For example, Ribbon-Helix-Helix (RHH) domains are common in MetJ/Arc-family transcription factors as well as in TAs, suggesting that the bacterial transcription factors of this family were originally derived from antitoxins (Aravind et al., 2005). Given that TAs and other MGEs are activated by various signals, such as DNA damage (Knowles et al., 2017) or the presence of other MGE (McKitterick and Seed, 2018;LeGault et al., 2021), recruitment of MGE-encoded transcription factors could be favorable for the host, enabling it to respond to the same stressors (Benler and Koonin, 2020). ...
Full-text available
Prokaryotic genomes are replete with mobile genetic elements (MGE) that span a continuum of replication autonomy. On numerous occasions during microbial evolution, diverse MGE lose their autonomy altogether but, rather than being quickly purged from the host genome, assume a new function that benefits the host, rendering the immobilized MGE subject to purifying selection, and resulting in its vertical inheritance. This mini-review highlights the diversity of the repurposed (exapted) MGE as well as the plethora of cellular functions that they perform. The principal contribution of the exaptation of MGE and their components is to the prokaryotic functional systems involved in biological conflicts, and in particular, defense against viruses and other MGE. This evolutionary entanglement between MGE and defense systems appears to stem both from mechanistic similarities and from similar evolutionary predicaments whereby both MGEs and defense systems tend to incur fitness costs to the hosts and thereby evolve mechanisms for survival including horizontal mobility, causing host addiction, and exaptation for functions beneficial to the host. The examples discussed demonstrate that the identity of an MGE, overall mobility and relationship with the host cell (mutualistic, symbiotic, commensal, or parasitic) are all factors that affect exaptation.
... Additionally, the relative abundance of hallmark genes encoded by temperate viruses increased with microbial density in a coral reef (Knowles et al. 2016). These findings suggest that lysogenic infection may become dominant at high-cell densities because proviruses can replicate quickly in a way that will keep pace with their fast-growing host and provirus-mediated superinfection resistance might become increasingly important at high cell densities (called "Piggyback-the-Winner" model 4 ) (Knowles et al. 2016(Knowles et al. , 2017Coutinho et al. 2017) (Fig. 1.2). ...
... Virus-mediated prokaryotic mortality was calculated by dividing the lytic VP by burst sizes of 25 determined from a previous global investigation (Knowles et al., 2017). The percentage of lysed cells (PLC; expressed as % L − 1 ) was then estimated from the ratio of the virusmediated prokaryotic mortality to the in situ prokaryotic abundance of the macrocosm. ...
Full-text available
Viruses saturate environments throughout the world and play key roles in microbial food webs, yet how viral activities affect dissolved organic matter (DOM) processing in natural environments remains elusive. We established a large-scale long-term macrocosm experiment to explore viral dynamics and their potential impacts on microbial mortality and DOM quantity and quality in starved and stratified ecosystems. High viral infection dynamics and the virus-induced cell lysis (6.23-64.68% d − 1) was found in the starved seawater macrocosm, which contributed to a significant transformation of microbial biomass into DOM (0.72-5.32 μg L − 1 d − 1). In the stratified macrocosm, a substantial amount of viral lysate DOM (2.43-17.87 μg L − 1 d − 1) was released into the upper riverine water, and viral lysis and DOM release (0.35-5.75 μg L − 1 d − 1) were reduced in the mixed water layer between riverine water and seawater. Viral lysis was stimulated at the bottom of stratified macrocosm, potentially fueled by the sinking of particulate organic carbon. Significant positive and negative associations between lytic viral production and different fluorescent DOM components were found in the starved and stratified macrocosm, indicating the potentially complex viral impacts on the production and utilization of DOM. Results also revealed the significant viral contribution to pools of both relatively higher molecular weight labile DOM and lower molecular weight recalcitrant DOM. Our study suggests that viruses have heterogeneous impact on the cycling and fate of DOM in aquatic environments.
... The phage to microbe ratio (PMR) is about 1:1,000 which is considerably low (Fig. 2b). In general, phages prefer a lysogenic life cycle over a lytic cycle so as to take advantage of the bacterial proliferation in order to effectively propagate themselves in many ecosystems 16 . We then attempted to isolate 50 different colonies of E. gallinarum from the feces of C57BL/6 mice in steady state as well as after the onset of DSS colitis and subjected them to PCR analysis for the phiEG37k genome. ...
Full-text available
Increase of the enteric bacteriophages (phage), components of the enteric virome, has been associated with the development of inflammatory bowel diseases. However, little is known about how a given phage contributes to the regulation of intestinal inflammation. In this study, we isolated a new phage associated with Enterococcus gallinarum, named phiEG37k, the level of which was increased in C57BL/6 mice with colitis development. We found that, irrespective of the state of inflammation, over 95% of the E. gallinarum population in the mice contained phiEG37k prophage within their genome and the phiEG37k titers were proportional to that of E. gallinarum in the gut. To explore whether phiEG37k impacts intestinal homeostasis and/or inflammation, we generated mice colonized either with E. gallinarum with or without the prophage phiEG37k. We found that the mice colonized with the bacteria with phiEG37k produced more Mucin 2 (MUC2) that serves to protect the intestinal epithelium, as compared to those colonized with the phage-free bacteria. Consistently, the former mice were less sensitive to experimental colitis than the latter mice. These results suggest that the newly isolated phage has the potential to protect the host by strengthening mucosal integrity. Our study may have clinical implication in further understanding of how bacteriophages contribute to the gut homeostasis and pathogenesis.
... Whereas this long-established concept is well supported by theoretical and experimental studies 31,32 , data from environmental studies are more ambiguous. Indeed, some studies have reported a positive (or negative) correlation between the frequency of lytic (or lysogenic) phages and bacterial density 33,34 , whereas other studies reported weak or no correlations 35 . Also contradicting this view, the nutrient-rich, bacterially dense ecosystem of the mammalian gut is dominated by lysogenic bacteria 36,37 , and copiotroph species tend to have more prophages 38 . ...
We commonly acknowledge that bacterial viruses (phages) shape the composition and evolution of bacterial communities in nature and therefore have important roles in ecosystem functioning. This view stems from studies in the 1990s to the first decade of the twenty-first century that revealed high viral abundance, high viral diversity and virus-induced microbial death in aquatic ecosystems as well as an association between collapses in bacterial density and peaks in phage abundance. The recent surge in metagenomic analyses has provided deeper insight into the abundance, genomic diversity and spatio-temporal dynamics of phages in a wide variety of ecosystems, ranging from deep oceans to soil and the mammalian digestive tract. However, the causes and consequences of variations in phage community compositions remain poorly understood. In this Review, we explore current knowledge of the composition and evolution of phage communities, as well as their roles in controlling the population and evolutionary dynamics of bacterial communities. We discuss the need for greater ecological realism in laboratory studies to capture the complexity of microbial communities that thrive in natural environments.
On 11 March 2011, a mega-earthquake followed by a huge tsunami hit a large area of the Pacific coast of Tohoku, Japan, and caused widespread disturbances in marine ecosystems; however, the impacts on virus-bacteria systems have not yet been elucidated. To investigate whether virus-bacteria interactions were affected by the earthquake and tsunami, we analyzed time series data of viral and bacterial abundances in Otsuchi Bay. Data were collected every 2 mo, from July 2011 to September 2015. Bacterial abundance exhibited a recurrent seasonal cycle with high abundance during the warm season. The seasonal trend of viruses generally followed that of the bacteria, yielding an average virus to bacteria ratio (VBR) of 10.8 ± 3.6 (mean ± SD; n = 432). A notable exception was found at the first 2 sampling times (July and September 2011) when the VBR was consistently low, with an average value of 5.9 ± 1.2 (n = 32). The average VBR during these time periods was substantially lower than the VBR observed in the same season of subsequent years. An analysis of the subset of data collected in the warm season of 2011 and 2012 revealed that the viral abundance and VBR were negatively correlated with turbidity. These results support the hypothesis that viruses were scavenged by non-host particles from the resuspended sediments and damaged catchment. The earthquake and tsunami thus exerted a prolonged impact, over several months, on the virus-bacteria dynamics in Otsuchi Bay.
This chapter aims to give a condensed overview on the occurrence and potential roles of prokaryotic viruses (phages) in groundwater. While abundant and steadily growing data on marine and limnic surface habitats underscores the significance of phages for biogeochemical cycling and microbial community dynamics, research tapping into the ecology of phages in groundwater is extremely scarce. Here, we summarize the findings of the few available studies and discuss potential driving factors of viral abundance and activity based on published and yet unpublished datasets. Further, based on established mechanisms of phage host interactions, such as the ‘viral shunt,’ the ‘Kill-the-Winner’ dynamics and the modulation of host metabolisms via auxiliary metabolic genes, we propose (largely hypothetical) roles of phages in groundwater. Due to the nutrient and energy-limitation of aquifers, we might expect a high frequency of stalled or slowly progressing viral infections in groundwater. Yet, recent metagenomic and metatranscriptomic evidence as well as flow cytometric quantification of virus-like particles indicate that viruses in subsurface environments are not just persisting but are indeed involved in major biogeochemical cycles, particularly in recently recharged groundwater. Finally, although the few studies available confirm a dominance of tailed phages as in surface waters, also other groups such as Microviridae and Inoviridae may play an important role in the subsurface. Upcoming research will allow to evaluate, extend and refine these exciting early evidence and hypotheses.
This chapter aims to give a comprehensive overview of the activities and consequences of prokaryotic viruses (phages) in lake ecosystems. In addition to known impacts of lytic phages also potential ecological roles of temperate phages will be discussed. General principles of phage-host relationships, quantitative and metagenomic data of viruses in lakes as well as theoretical considerations will be used to give an account for one of the currently most vibrant fields in aquatic ecology. Since the interplay between phages and prokaryotes is extremely complex and lake-based studies remain sparse, this chapter will also draw on concepts from better studied marine environments.
Full-text available
Daily environmental oscillations that follow Earth’s rotation around the Sun set a metronome for life, under which all organisms have evolved. Entrainment to these cues allow organisms to rhythmically set the pace of their own endogenous biological clocks with which the timings of diverse cellular activities are coordinated. In recent years, our knowledge of biological rhythms has extended across all domains of life. This includes both free-living and symbiotic life forms. With the insurgence of metagenomic sequencing tools, the field of holobiont chronobiomics (encompassing chronobiology of host and its associated microbiota) has recently opened and gained significant traction. Here, we review current knowledge regarding free-living prokaryote rhythmic regulation before exploring active areas of research that consider the coordinated rhythmic regulatory activities of hosts and their symbionts as a single entity, i.e., holobiont, and even the extent to which rhythmicity influences virus–host interactions. We describe rhythmicity within non-photosynthetic bacteria, cyanobacteria, and archaea, before investigating the effect of light, and, thus, diel cycle, on viral life cycles and host–virus population dynamics in marine planktonic ecosystems along with their potential to influence host cyanobacterial circadian clocks. We then explore current evidence outlining coordinated rhythmic regulation within marine holobionts and the significance of this for holobiont health and adaptive fitness that, in turn, optimizes their success within their local environments. Finally, we assess the critical role of circadian regulation for holobiont innate immunity and metabolism within well-studied non-marine mammalian systems, and, thus, assess how this can guide us within understudied marine chronobiomics research.
Full-text available
The ISME Journal: Multidisciplinary Journal of Microbial Ecology is the official Journal of the International Society for Microbial Ecology, publishing high-quality, original research papers, short communications, commentary articles and reviews in the rapidly expanding and diverse discipline of microbial ecology.
Full-text available
Bacteria are essential for many ecosystem services but our understanding of factors controlling their functioning is incomplete. While biodiversity has been identified as an important driver of ecosystem processes in macrobiotic communities, we know much less about bacterial communities. Due to the high diversity of bacterial communities, high functional redundancy is commonly proposed as explanation for a lack of clear effects of diversity. The generality of this claim has, however, been questioned. We present the results of an outdoor dilution-to-extinction experiment with four lake bacterial communities. The consequences of changes in bacterial diversity in terms of effective number of species, phylogenetic diversity and functional diversity were studied for (i) bacterial abundance, (ii) temporal stability of abundance, (iii) nitrogen concentration, and (iv) multifunctionality. We observed a richness gradient ranging from 15 to 280 operational taxonomic units. Individual relationships between diversity and functioning ranged from negative to positive depending on lake, diversity dimension and aspect of functioning. Only between phylogenetic diversity and abundance did we find a statistically consistent positive relationship across lakes. A literature review of 24 peer-reviewed studies that used dilution-to-extinction to manipulate bacterial diversity corroborated our findings: about 25% found positive relationships. Combined, these results suggest that bacteria-driven community functioning is relatively resistant to reductions in diversity. This article is protected by copyright. All rights reserved.
Full-text available
The activity of bacteriophages poses a major threat to bacterial survival. Upon infection, a temperate phage can either kill the host cell or be maintained as a prophage. In this state, the bacteria carrying the prophage is at risk of superinfection, where another phage injects its genetic material and competes for host cell resources. To avoid this, many phages have evolved mechanisms that alter the bacteria and make it resistant to phage superinfection. The mechanisms underlying these phentoypic conversions and the fitness consequences for the host are poorly understood, and systematic studies of superinfection exclusion mechanisms are lacking. In this study, we examined a wide range of Pseudomonas aeruginosa phages and found that they mediate superinfection exclusion through a variety of mechanisms, some of which affected the type IV pilus and O-antigen, and others that functioned inside the cell. The strongest resistance mechanism was a surface modification that we showed is cost-free for the bacterial host in a natural soil environment and in a Caenorhabditis. elegans infection model. This study represents the first systematic approach to address how a population of prophages influences phage resistance and bacterial behavior in P. aeruginosa.The ISME Journal advance online publication, 3 June 2016; doi:10.1038/ismej.2016.79.
Full-text available
PHASTER (PHAge Search Tool – Enhanced Release) is a significant upgrade to the popular PHAST web server for the rapid identification and annotation of prophage sequences within bacterial genomes and plasmids. Although the steps in the phage identification pipeline in PHASTER remain largely the same as in the original PHAST, numerous software improvements and significant hardware enhancements have now made PHASTER faster, more efficient, more visually appealing and much more user friendly. In particular, PHASTER is now 4.3× faster than PHAST when analyzing a typical bacterial genome. More specifically, software optimizations have made the backend of PHASTER 2.7X faster than PHAST, while the addition of 80 CPUs to the PHASTER compute cluster are responsible for the remaining speed-up. PHASTER can now process a typical bacterial genome in 3 min from the raw sequence alone, or in 1.5 min when given a pre-annotated GenBank file. A number of other optimizations have also been implemented, including automated algorithms to reduce the size and redundancy of PHASTER's databases, improvements in handling multiple (metagenomic) queries and higher user traffic, along with the ability to perform automated look-ups against 14 000 previously PHAST/PHASTER annotated bacterial genomes (which can lead to complete phage annotations in seconds as opposed to minutes). PHASTER's web interface has also been entirely rewritten. A new graphical genome browser has been added, gene/genome visualization tools have been improved, and the graphical interface is now more modern, robust and user-friendly. PHASTER is available online at
Full-text available
Viruses are ubiquitous in aquatic ecosystems where they significantly contribute to microbial mortality. In glacier-fed turbid lakes, however, viruses not only encounter low host abundances, but also a high number of suspended mineral particles introduced by glacier meltwaters. We hypothesized that these particles potentially lead to unspecific adsorption and removal of free virus from the plankton, and thus significantly reduce their abundance in this type of lake. We followed the distribution of free virus-like particles (VLP) during the ice-free season across a turbidity gradient in four alpine lakes including one adjacent clear system where hydrological connectivity to the receding glacier is already lost. In the glacier-fed turbid lakes, VLP abundance increased with distance to the glacier, but the highest numbers were observed in the clear lake by the end of August, coinciding with the maximum in prokaryotic abundance. Our results suggest that viral loss by attachment to particles is less important than expected. Nevertheless, the relatively lower variability in VLP abundance and the lower virus-toprokaryote ratio found in the turbid lakes than in the clear one point to a rather low temporal turnover and thus, to a reduced impact on microbial communities.
Full-text available
When we think about viruses we tend to consider ones that afflict humans—such as those that cause influenza, HIV, and Ebola. Yet, vastly more viruses infect single-celled microbes. Diverse and abundant, microbes and the viruses that infect them are found in oceans, lakes, plants, soil, and animal-associated microbiomes. Taking a vital look at the "microscopic" mode of disease dynamics, Quantitative Viral Ecology establishes a theoretical foundation from which to model and predict the ecological and evolutionary dynamics that result from the interaction between viruses and their microbial hosts. Joshua Weitz addresses three major questions: What are viruses of microbes and what do they do to their hosts? How do interactions of a single virus-host pair affect the number and traits of hosts and virus populations? How do virus-host dynamics emerge in natural environments when interactions take place between many viruses and many hosts? Emphasizing how theory and models can provide answers, Weitz offers a cohesive framework for tackling new challenges in the study of viruses and microbes and how they are connected to ecological processes—from the laboratory to the Earth system. Quantitative Viral Ecology is an innovative exploration of the influence of viruses in our complex natural world.
The definition of second order interaction in a (2 × 2 × 2) table given by Bartlett is accepted, but it is shown by an example that the vanishing of this second order interaction does not necessarily justify the mechanical procedure of forming the three component 2 × 2 tables and testing each of these for significance by standard methods.*
Temperate phages are common, and prophages are abundant residents of sequenced bacterial genomes. Mycobacteriophages are viruses that infect mycobacterial hosts including Mycobacterium tuberculosis and Mycobacterium smegmatis, encompass substantial genetic diversity and are commonly temperate. Characterization of ten Cluster N temperate mycobacteriophages revealed at least five distinct prophage-expressed viral defence systems that interfere with the infection of lytic and temperate phages that are either closely related (homotypic defence) or unrelated (heterotypic defence) to the prophage. Target specificity is unpredictable, ranging from a single target phage to one-third of those tested. The defence systems include a single-subunit restriction system, a heterotypic exclusion system and a predicted (p)ppGpp synthetase, which blocks lytic phage growth, promotes bacterial survival and enables efficient lysogeny. The predicted (p)ppGpp synthetase coded by the Phrann prophage defends against phage Tweety infection, but Tweety codes for a tetrapeptide repeat protein, gp54, which acts as a highly effective counter-defence system. Prophage-mediated viral defence offers an efficient mechanism for bacterial success in host–virus dynamics, and counter-defence promotes phage co-evolution.
SUMMARY Bacteria have a range of distinct immune strategies that provide protection against bacteriophage (phage) infections. While much has been learned about the mechanism of action of these defense strategies, it is less clear why such diversity in defense strategies has evolved. In this review, we discuss the short- and long-term costs and benefits of the different resistance strategies and, hence, the ecological conditions that are likely to favor the different strategies alone and in combination. Finally, we discuss some of the broader consequences, beyond resistance to phage and other genetic elements, resulting from the operation of different immune strategies.
The discovery of the numerical importance of viruses in a variety of (aquatic) ecosystems has changed our perception of their importance in microbial processes. Bacteria and Archaea undoubtedly represent the most abundant cellular life forms on Earth and past estimates of viral numbers (represented mainly by viruses infecting prokaryotes) have indicated abundances at least one order of magnitude higher than that of their cellular hosts. Such dominance has been reflected most often by the virus-to-prokaryote ratio (VPR), proposed as a proxy for the relationship between viral and prokaryotic communities. VPR values have been discussed in the literature to express viral numerical dominance (or absence of it) over their cellular hosts, but the ecological meaning and interpretation of this ratio has remained somewhat nebulous or contradictory. We gathered data from 210 publications (and additional unpublished data) on viral ecology with the aim of exploring VPR. The results are presented in three parts: the first consists of an overview of the minimal, maximal and calculated average VPR values in an extensive variety of different environments. Results indicate that VPR values fluctuate over six orders of magnitude, with variations observed within each ecosystem. The second part investigates the relationship between VPR and other indices, in order to assess whether VPR can provide insights into virus-host relationships. A positive relationship was found between VPR and viral abundance (VA), frequency of visibly infected cells (FVIC), burst size (BS), frequency of lysogenic cells (FLC) and chlorophyll a (Chl a) concentration. An inverse relationship was detected between VPR and prokaryotic abundance (PA) (in sediments), prokaryotic production (PP) and virus-host contact rates (VCR) as well as salinity and temperature. No significant relationship was found between VPR and viral production (VP), fraction of mortality from viral lysis (FMVL), viral decay rate (VDR), viral turnover (VT) or depth. Finally, we summarize our results by proposing two scenarios in two contrasting environments, based on current theories on viral ecology as well as the present results. We conclude that since VPR fluctuates in every habitat for different reasons, as it is linked to a multitude of factors related to virus-host dynamics, extreme caution should be used when inferring relationships between viruses and their hosts. Furthermore, we posit that the VPR is only useful in specific, controlled conditions, e.g. for the monitoring of fluctuations in viral and host abundance over time.