ArticlePDF Available

Abstract and Figures

Phenotypic variation is suggested to facilitate the persistence of environmentally growing pathogens under environmental change. Here we hypothesized that the intensive farming environment induces higher phenotypic variation in microbial pathogens than natural environment, because of high stochasticity for growth and stronger survival selection compared to the natural environment. We tested the hypothesis with an opportunistic fish pathogen Flavobacterium columnare isolated either from fish farms or from natural waters. We measured growth parameters of two morphotypes from all isolates in different resource concentrations and two temperatures relevant for the occurrence of disease epidemics at farms and tested their virulence using a zebrafish (Danio rerio) infection model. According to our hypothesis, isolates originating from the fish farms had higher phenotypic variation in growth between the morphotypes than the isolates from natural waters. The difference was more pronounced in higher resource concentrations and the higher temperature, suggesting that phenotypic variation is driven by the exploitation of increased outside‐host resources at farms. Phenotypic variation of virulence was not observed based on isolate origin but only based on morphotype. However, when in contact with the larger fish, the less virulent morphotype of some of the isolates also had high virulence. As the less virulent morphotype also had higher growth rate in outside‐host resources, the results suggest that both morphotypes can contribute to F. columnare epidemics at fish farms, especially with current prospects of warming temperatures. Our results suggest that higher phenotypic variation per se does not lead to higher virulence, but that environmental conditions at fish farms could select isolates with high phenotypic variation in bacterial population and hence affect evolution in F. columnare at fish farms. Our results highlight the multifaceted effects of human‐induced environmental alterations in shaping epidemiology and evolution in microbial pathogens.
This content is subject to copyright. Terms and conditions apply.
Evolutionary Applications. 2022;15:417–428.
Received: 12 June 2020 
Revised: 27 Januar y 2022 
Accepted: 1 February 2022
DOI : 10.1111/eva .13355
Rich resource environment of fish farms facilitates phenotypic
variation and virulence in an opportunistic fish pathogen
Katja Pulkkinen1| Tarmo Ketola1| Jouni Laakso2| Johanna Mappes1,2|
Lotta- Riina Sundberg1,3
This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium,
provided the original work is properly cited.
© 2022 The Authors. Evolutionary Applications published by John Wiley & Sons Ltd.
1Department of Biological and
Environmental Science, University of
Jyväskylä, Jyväskylä, Finland
2Research Programme in Organismal and
Evolutionary Biology, Faculty of Biologic al
and Environmental Sciences, University of
Helsinki, Helsinki, Finland
3Nanoscience Center, University of
Jyväskylä, Jyväskylä, Finland
Katja Pulkkinen, Department of Biological
and Environmental Science, P.O. Box
35, FI- 40014 University of Jy väskylä,
Jyväskylä, Finland.
Funding information
Jane ja Aatos Erkon Säätiö; Academy of
Finland, Grant/Award Number: 304615,
252411, 266879, 314939 and 7128888;
Suomen Kulttuurirahasto
Phenotypic variation is suggested to facilitate the persistence of environmentally
growing pathogens under environmental change. Here, we hypothesized that the in-
tensive farming environment induces higher phenotypic variation in microbial patho-
gens than natural environment, because of high stochasticity for growth and stronger
survival selection compared to the natural environment. We tested the hypothesis
with an opportunistic fish pathogen Flavobacterium columnare isolated either from
fish farms or from natural waters. We measured growth parameters of two morpho-
types from all isolates in different resource concentrations and two temperatures
relevant for the occurrence of disease epidemics at farms and tested their virulence
using a zebrafish (Danio rerio) infection model. According to our hypothesis, isolates
originating from the fish farms had higher phenotypic variation in growth between
the morphot ypes than the isolates from natural waters. The difference was more pro-
nounced in higher resource concentrations and the higher temperature, suggesting
that phenotypic variation is driven by the exploitation of increased outside- host re-
sources at farms. Phenotypic variation of virulence was not observed based on isolate
origin but only based on morphotype. However, when in contact with the larger fish,
the less virulent morphotype of some of the isolates also had high virulence. As the
less virulent morphotype also had higher growth rate in outside- host resources, the
results suggest that both morphotypes can contribute to F. columnare epidemics at
fish farms, especially with current prospects of warming temperatures. Our results
suggest that higher phenotypic variation per se does not lead to higher virulence, but
that environmental conditions at fish farms could select isolates with high phenotypic
variation in bacterial population and hence affect evolution in F. columnare at fish
farms. Our results highlight the multifaceted effects of human- induced environmental
alterations in shaping epidemiology and evolution in microbial pathogens.
aquaculture, bacterium, colony type, fish diseases, phenotypic variation
    PU LKKIN EN Et aL.
Anthropogenic modifications in the environment, such as changes in
temperature, precipitation, or nutrients, affect species in many ways
and cause adaptations suited to the prevailing regime of changes
(Kristensen et al., 2020). For pathogens, changes in host availability
and survival of environmental stages of pathogens have the poten-
tial to drive evolution of pathogen traits via anthropogenic change
(Mideo & Reece, 2012; Wolinska & King, 2009). One example of
human- induced change in the selective environment for pathogens
is intensive farming (Mennerat et al., 2010; Pulkkinen et al., 2010).
The epidemiological conditions at farms that select for fast growth
and high transmission between hosts in pathogens are likely to se-
lect for different genotypic and/or phenotypic properties than the
conditions outside the farming environment (Mennerat et al., 2010;
Wolinska & King, 2009). For example, intensive farming has been as-
sociated with increases in virulence (harm to the host) for pathogens
such as Marek´s disease virus in poultry farming (Atkins et al., 2013)
as well as salmon lice (Mennerat et al., 2010, 2012; Ugelvik et al.,
2017) and Flavobacterium columnare in aquaculture (Pulkkinen et al.,
2010; Sundberg et al., 2016).
Phenotypic plasticity is one of the mechanisms that allow or-
ganisms to adjust to changing environments (Ackermann, 2015;
Nussey et al., 2007; Rainey & Travisano, 1998; Scheiner, 1993; Via
et al., 1995). The ability of a single genotype to produce different
phenotypes increases the genotype's fitness under environmen-
tal change and ensures persistence in fluctuating environments
(Ackermann, 2015; Via et al., 1995). Energetic costs associated with
maintaining the metabolic machinery or trade- offs between differ-
ent phenotypic properties in different environments may limit the
adaptive value of phenotypic variation (DeWitt et al., 1998; Koch
& Guillaume, 2020). However, these trade- offs may be relaxed if a
resource- rich environment enables stronger allocation to several
costly functions simultaneously (van Noordwijk & de Jong, 1986).
The costs of phenotypic responses depend also strongly on ambi-
ent temperature due to temperature dependence of metabolic rates
(Huey & Kingsolver, 2011). In addition, predictability and the speed
of the environmental change determine the profitability of pheno-
typic change (DeWitt et al., 1998; Kristensen et al., 2008; Kussell &
Leibler, 2005; Padilla & Adolph, 1996). A randomly changing envi-
ronment increases the cost of plasticity due to a mismatch between
the signal for phenotypic change and the realized environmental
conditions (Reed et al., 2010; Tonsor et al., 2013). In such unpredict-
able environments, risk- spreading strategies, like bet- hedging, are
expected to evolve (Botero et al., 2015; DeWitt & Langerhans, 2004;
Levins, 1968). In this paper, we use the term phenotypic variation to
describe a genotype's ability to change phenotype, irrespective of
whether it is due to an environmental cue, or due to random switch-
ing of phenotypes (bet- hedging).
Phenotypic variation might be especially important for opportu-
nistic pathogens that survive and replicate not only within the host
but also in the outside- host environment (Brown et al., 2012; Ketola
et al., 2016). The available ecological opportunities outside the host
can differ vastly from those encountered by the pathogen inside the
host (Anttila et al., 2016; Brown et al., 2012). Within the host, the
greatest challenges are posed by the host immune system (Schmid-
Hempel, 2009). In the outside- host environment, low availability
of resources, presence of predators, parasites and competitors,
and abiotic factors, such as temperature, often restrict the growth
(Adiba et al., 2010; Friman et al., 2009, 2011; Hibbing et al., 2010;
Ketola et al., 2013; Zhang et al., 2014). The ability to change from
one alternative phenotype to another with different characteristics,
for example, for competitive ability or immune evasion might be cru-
cial for the expression of pathogenicity for opportunistic pathogens
(Holland et al., 2014; Ketola et al., 2016; Kreibich & Hardt, 2015).
In bacteria, phenotypic variation is often visible in the form of
different types of colony morphologies, with differences in growth
characteristics in the bacterial population forming the colony
(Friman et al., 2009; Koh et al., 2007; Kunttu, Suomalainen, et al.,
2009; Rainey & Travisano, 1998). Phenotypic variation in bacteria
may rise as a response to change in environmental conditions or as
a consequence of stochastic switching between alternative pheno-
types produced by the same genotype (Ackermann, 2015).
Intensive farming represents an environment that has been heav-
ily modified by human actions. For pathogens, intensive farming rep-
resents an environment with ample possibilities but also challenges
for growth and survival. On one hand, the high density of genetically
homogeneous hosts and excess feed offers abundant resources for
growth, but on the other hand, medical treatments during outbreak
season pose a recurrent and strong threat for pathogen survival
(Atkins et al., 2013; Mennerat et al., 2010; Pulkkinen et al., 2010).
For most of the time in the natural environment, the microbes per-
sist outside of the hosts with few opportunities for fast growth. At
farms, high host availability may favor phenotypes capable of fast
exploitation of hosts, that is, high virulence (Ebert, 1998; Frank,
1996). On the other hand, periodical medical treatments targeted to
diseased hosts may favor phenotypes that can escape treatments by
growing on organic material in the outside- host environment. The
ability of environmentally growing pathogens to persist and repli-
cate outside the host is expected to lead to increased virulence as
pathogen reproductive success is not restricted by host death; that
is, there is no trade- off between virulence and transmission (Day,
2002; Ewald, 1994).
Our study focused on an opportunistic fish pathogen
Flavobacterium columnare, a cause of severe economic losses to fish
production worldwide (Declercq et al., 2013; Wagner et al., 2002).
The virulence of the pathogen has been suggested to have increased
during the last few decades at fish farms in Finland (Ashrafi et al.,
2018; Pulkkinen et al., 2010; Sundberg et al., 2016). At Finnish fish
farms, the disease outbreaks occur only during summer, when water
temperatures reach c.a. 18°C (Pulkkinen et al., 2010). Disease out-
breaks start when waterborne bacterial cells proliferate in alive or
dead fish tissue, or other organic material in the water (Kunttu et al.,
2012; Kunttu, Valtonen, et al., 2009). Increased resource levels in the
outside- host environment increase virulence in the bacterium via in-
creased doses and virulence factor activation (Kinnula et al., 2017;
Penttinen et al., 2016) and could increase the contribution of less
virulent strains in outbreaks (Pulkkinen et al., 2018).
Flavobacterium columnare changes colony morphotype from a
spreading rhizoid (Rz) type to a nonspreading rough (R) type either
spontaneously or after exposure to virulent phages, with a con-
comitant decrease in virulence (Laanto et al., 2012; Sundberg et al.,
2014). The two colony types also differ in growth parameters and
responses to protozoan predators, with suggested costs of express-
ing R type in the outside- host environment (Zhang et al., 2014). The
spontaneous change of colony types expressed by F. columnare is
reversible (Laanto et al., 2012). While the exact mechanisms leading
to colony morphology changes are not known, and the abundance
and the possible role of the R type in the environment are unclear,
the different colony types are suggested to serve different functions
in invasion and replication in fish and the outside- host environment
(Kunttu, Suomalainen, et al., 2009; Kunttu, Valtonen, et al., 2009;
Laanto et al., 2014; Zhang et al., 2014).
We hypothesized that large changes in environments associ-
ated with intensive farming facilitate higher phenotypic variation
in environmentally growing microbial pathogens than the more
stable natural environment. Increased resource availability in in-
tensive farming environments promotes large population sizes and
could also decrease the costs associated with phenotypic variation
(DeWitt et al., 1998; Friman et al., 2008) and increase the probabil-
ity of phenotypic variation. We hypothesized that F. columnare iso-
lates originating from fish farms would present higher phenotypic
variation in growth and virulence than isolates from natural waters.
We also hypothesized that due to resource- and temperature- driven
growth, these differences would be more pronounced in higher re-
sources and higher temperatures. To test the hypothesis, we mea-
sured the growth of two morphotypes (ancestral rhizoid, Rz, and its
rough, R, derivative, Figure S1) of each of five isolates originating
from both environments. We took measurements in five different
resource concentrations and two temperatures, relevant for occur-
rence of disease epidemics at farms (Ashrafi et al., 2018; Pulkkinen
et al., 2010). In addition to maximal growth rate, we also analyzed
bacterial yield as a potentially relevant trait for bacterial persistence
at farms. Due to potential costs in expressing the different morpho-
ty pes (Zha ng et al. , 2014), we also tested the stabilit y of the morpho-
types in plate cultivation under the same resource concentrations
that were used for the growth measurements. Finally, we tested
phenotypic differences between the natural and the fish farm iso-
lates in bacterial virulence in fish challenge tests in vivo.
2.1  |Flavobacterium columnare colony
Flavobacterium columnare exhibits three different colony mor-
phologies, of which the rhizoid type (Rz) is associated with high
virulence (Kunttu, Suomalainen, et al., 2009). The other two types,
rough (R) and soft (S), are associated with lower virulence (Kunttu,
Suomalainen, et al., 2009; Laanto et al., 2012). Primary isolations
from natural samples on selective agar plates (Decostere et al., 1997)
produce generally a rhizoid colony morphology, while the two other
types appear during continued re- cultivation (Kunttu, Suomalainen,
et al., 2009; Sundberg et al., 2014), or as a response to infection by
phages (Laanto et al., 2012). In this study, we compared the ancestral
Rz type and its R derivative formed spontaneously in subcultivation.
2.2  |  Isolation and cultivation of bacteria
Fish farm isolates were obtained from diseased fish or tank water
during outbreaks at fish farms as part of disease surveillance. The
isolates from nature were collected from water or a diseased wild
fish upstream of a fish farm and they represent the variation present
in natural waters (Ashrafi et al., 2015; Kunttu et al., 2012) (Table 1).
Bacteria isolations were performed with standard culture methods,
using modified Shieh medium (from now on: Shieh medium) (Song
et al., 1988) supplemented with tobramycin (Decostere et al., 1997).
Pure cultures were stored at −80°C with 10% glycerol and 10% fetal
calf serum.
All isolates exhibited originally a rhizoid morphotype. Rough
colonies that appeared spontaneously during plate cultivation were
collected singly and cultured in Shieh medium at room temperature
with constant shaking (210 rpm) and plated to assure the loss of pa-
rental Rz growth before storing at −80°C (Figure S1).
Before the experiments, bacteria were revived from frozen stocks
by inoculation to Shieh medium (1× concentration). Bacteria were
cultured at room temperature with constant agitation (210 rpm) for
24 h and subsequently enriched in 1/10 dilution in fresh Shieh me-
dium overnight in the same conditions. For growth measurements
at different resource concentrations, 10 µl of the enriched culture
was applied into 100- well Bioscreen C® plates containing 400 µl of
fresh Shieh medium at different concentrations (2×, 1×, 0.5×, 0.1×,
and 0.05× of Shieh medium). Maximum growth rates and yield were
determined from biomass growth data recorded at 5 min intervals
with 420– 580 nm optical density for 6 days at 25°C and 10 days
at 15°C. Each isolate– morphotype combination was included twice
into separate measurements on two subsequent weeks, totaling four
replicate measurements. In addition, after liquid culture, bacterial
samples were plate cultured on Shieh agar with the five different re-
source concentrations mentioned above, to check the morphotype
stability under different resource levels.
2.3  |  Virulence experiments
The fish experiments were conducted according to the Finnish Act
of the Use of Animals for Experimental Purposes, under permis-
sion ESAVI- 2010- 05569/Ym- 23 granted by the National Animal
Experiment Board at the Regional State Administrative Agency for
Southern Finland for L- RS.
    PU LKKIN EN Et aL.
For testing virulence, we used zebrafish (Danio rerio), which
have been established as a reliable model system for revealing dif-
ferences in virulence among strains in F. columnare (Kinnula et al.,
2015). Zebrafish were obtained from Core Facilities (COFA) and
research services of Tampere University, Finland. The fish used in
the experiment were disease- free, adult, and unsexed. The weight
range of fish challenged with fish farm isolates and natural isolates
were 0.06– 0.88 g and 0.08– 0.87 g, respectively. For each isolate–
morphotype combination, 10 individual fish were challenged, with
10 additional fish used as controls, totaling 210 individual fish. The
fish were infected using continuous immersion challenge (Kinnula
et al., 2015). Fish were placed individually in 1- L plastic containers
filled with 500 ml of aerated well water at 25°C with freshly grown
bacteria added in 4 ml of fresh Shieh medium to reach a final bacte-
rial concentration of 1 × 105 CFU ml−1 in the container. For control
fish, 4 ml of Shieh was added without bacteria. Fish were monitored
for 4 days for disease symptoms and morbidity. Morbid fish that did
not respond to external stimuli were considered dead, removed, put
down by cutting the cordial spine with scissors, and weighed. At the
end of the experiment, all remaining fish were terminally euthanized
with MS- 222. The presence of F. columnare infection on fish was
checked by plating a primary culture from the gills onto modified
Shieh agar plates.
2.4  |  Data analysis
The maximum growth rates of the bacteria were assessed from the
OD data. For calculating the growth parameters, we used all the data
points concerning 25°C data as there were no apparent lag phases,
whereas with the 15°C data the estimation of the maximum growth
rate was started after 2 h to exclude the lag phase, that is, no visible
growth after inoculation. First, we log- transformed OD data, which
linearizes the exponential growth. This allows finding the point of
fastest growth rate using linear regression, by sequential fitting of
25 (or other desired) data point sliding windows to the data. From all
subsets of 25 data points, we sought the subset with the steepest
regression slope that equals the maximum growth rate. Maximum
biomass, that is, yield was determined as the largest found aver-
aged OD value over the subset. The MATLAB code to perform these
analyses is described in Ketola et al. (2013). The point estimates of
maximum growth rate and yield were further analyzed with linear
As the dataset on bacterial growth contained several factors
that could interact in many ways, we utilized model selection to re-
duce the complexity of the models explaining growth rate or yield.
We started model building from the model containing all possible
interactions between measurement temperature (T), resource con-
centrations (R), origin of the isolate (O), and morphotype (M) (all
fixed factors). In all models, we included “maternal” clone identity
as a random factor to control for nonindependence resulting from
four measurement replicates, and from the fact that colony types
originated from the same bacterial culture. Moreover, the effect of
measurement day was fitted as a fixed factor in all models. Starting
from the highest order interactions, each of the interactions was re-
moved from the model, if the interaction was not statistically signif-
icant (based on p- value > 0.10) and its removal improved the model
fit (AIC). The same selection procedure was used to test how exper-
imental factors explain biomass yield. Since yield is often found to
trade- off with growth rate (Velicer & Lenski, 1999), we tested also
TAB LE 1  The Flavobacterium columnare isolates used in this study
Isolate identity Genotype Location
isolated Source
B392 NDaOff- farm Ve06/2010 Fish Abramis brama, natural waters
B394 A Off- farm Ve06/2010 Fish Abram is brama, natural waters
B399 G/ST7bOff- farm Ve06/2010 Natural waters
B408 C Off- farm Ve08/2010 Natural waters
Tul o2 A Off- farm Ve08/2010 Natural waters
B402 C Farm Ve2010 Fish (Coregonus lavaretus)
B425 NAcFarm Ve20 07 Fish (Oncorhynchus mykiss)
B067 A Farm Le2 007 Fish (Salmo trutta)
Jip39/87 (ATCC 49513) GenIdType strain, France 1987 Fish (Ictalurus melas)
E E Farm Oe2002 Fish (Salmo salar)
Note: Genotyping of Finnish strains is based on ARISA (Suomalainen et al., 2006), or MLSA analysis, which associates uniformly with ARISA typing
(Ashrafi et al., 2015).
aND not determined, falls outside previously determined genetic groups of Finnish strains, see (Ashrafi et al., 2015), additional file; isolate B392 has
only allele 4.
bSee Ashrafi et al. (2015), additional files 1 and 6.
cNA not analysed.
dSee Ashrafi et al. (2015), Fig. 2.
eSee Laanto et al. (2012).
if our results concerning growth rate could be explained by the in-
clusion of yield as a covariate. To explore whether yield has similar
effects in all treatment groups, we fitted also interactions of the co-
variate (standardized to mean of zero) with all factors (Hendrix et al.,
1982). The same model selection as above was utilized to drop out
nonsignificant factors to covariate interactions. Since covariate and
factor by covariate interactions did not affect the significance of our
results or conclusions, it is clear that changes between treatments
are not caused by life history trade- offs between maximal growth
rate and yield. Thus, these results are not presented or considered
further. Statistical testing was performed with REML mixed models
with IBM SPSS v. 19 (IBM).
In fish challenge experiments, the fish were monitored for 96 h
and the last moribund fish was encountered and euthanized at 54 h.
Thus, the remaining challenged fish were considered as true sur-
vivors, and not as censored cases. Therefore, we used generalized
linear mixed models for binomial distribution to examine the morbid-
ity caused by different morphotypes of replicate strains originating
from nature or fish farms. We analyzed the morbidity risk of a host
in an hour with a model including all possible interactions between
the origin of the isolate (O; nature, fish farm), colony type (T; rhizoid,
rough), and fish weight (W; as a continuous covariate). Strain iden-
tity (S) was included as a random factor. Model reduction based on
AIC criteria was performed with drop1- function in package MASS
(Venables & Ripley, 20 02) starting from the full model. The analy-
sis was conducted using R software (version 3.3.2) and the Lme4
package (R Development Core Team, 2015). To be able to visual-
ize data fitting accuracy in Figure 5, we rerun the best model using
equivalent model but using Bayesian methods (Stan Development
Team, 2020). Posterior values of estimates were used to calculate
estimates of mortality risks given the fish weight and the colony
morphology. To formally test the region of fish weight where rhi-
zoid and rough colony types differ in their mortality risk, we uti-
lized a method analogous to Johnson- Neyman procedure (Hayes &
Matthes, 2009), where the posterior distribution of mortality risks at
each fish size (i.e., predicted values for 0.01- g intervals), for rhizoid
and rough, were compared for posterior distribution overlap. This
way we approximated the critical weight, below which the strain
morphology causes differences in mortality (p < 0.05).
3.1  |  Growth rate
The maximal bacterial growth rate was found to be affected by sev-
eral factors (Table 2, Table S1). The growth rate was affected the
most by the temperature, with the higher temperature supporting
higher growth rates (Table 2). Growth rate also increased with re-
source concentration (0.05×, 0.1×, 0.5×, 1×, and 2× growth me-
dium), although there were no statistical differences between the
two lowest resource concentrations (p < 0.3 in pairwise tests). The
rest of the pairwise tests indicated differences between resource
concentrations (p < 0.001). The origin of isolate or morphotype af-
fected growth rate only in interaction with other factors. Within
fish farm isolates, rough morphotype (R) replicated faster than the
rhizoid (Rz) (Figure 1a, F1,620 = 10.048, p = 0.002), whereas this
was not evident within isolates from natural water (F1,620 = 1.902,
p = 0.168), supporting our hypothesis of higher phenotypic varia-
tion in the fish farm isolates. Moreover, the growth rate of fish farm
TAB LE 2  Results of mixed model analyses on determinants of growth rate and yield in Flavobacterium columnare isolates
Growth rate Yield
df1df2F p df1df2F p
Temperature 1 772 1298.417 <0.001 1775 70 9.919 <0.001
Resource concentration 4 772 168.453 <0.001 4775 1 527. 139 <0.001
Morphotype 1 772 4.411 0.069 1775 6.621 0.010
Origin of isolate 1 8 1.603 0.206 1 8 0.017 0.899
Temperature × Resource 4 772 62.830 <0.001 4775 74 .572 <0.0 01
Temperature × Morphotype 1 772 6.863 0.009
Temperature × Origin of isolate 1 772 1 7. 8 8 6 <0.001
Resource × Morphotype 4 775 6.984 <0.001
Resource × Origin of isolate 4 772 2 .788 0.026
Morphotype × Origin of isolate 1 772 10.346 <0.001
Block 1 772 34.581 <0.001 1775 2.958 0.086
Isolate’s ID σ2 = 3.43 × 10−5, Wald Z = 1.695, p = 0.090 σ2 = 5.47 × 10 −3, Wald Z = 1.931, p = 0.053
Note: Clones were either isolated from fish farms or natural waters. From each clone, we measured two morphotypes (rough and rhizoid) in high and
low temperatures (25 and 15°C) and five different resource concentrations (0.05×, 0.1×, 0.5×, 1×, and 2× Shieh medium). In statistical analyses, we
also included effects of measurement block (identical measurements were done in two subsequent weeks), and maternal isolate's identity to control
for nonindependence of observations arising from the shared genetic background. Excluded factor interactions are denoted with – .
    PU LKKIN EN Et aL.
isolates was higher than the growth rate of isolates from natural wa-
ters only in the R morphotype (F1,620 = 9. 7 72, p = 0.010), but not in
the Rz morphotype (F1,620 = 0,620, p = 0.448). Growth advant age of
R morphotype was found only at high temperature (F1,772 = 7.551,
p = 0.0 06, Fi gure 1b) , wher eas at low te mper ature , there we re no di f-
ferences between the rough and rhizoid morphotype (F1,772 = 0.916,
p = 0.339).
Farm isolates had higher growth rates than isolates from natural
water in high temperature (Figure 1c, F1,10.620 = 12.20 9, p = 0.005),
but not in low temperature (F1,10. 620 = 0.176, p = 0.684). Farm iso-
lates had significantly higher growth rates than isolates from nat-
ural waters in the two highest resource concentrations (Figure 2a,
0.05× medium: F1,20.6 37 = 0.826, p = 0.374; 0.1×: F1,20.6 37 = 0.906,
p = 0.352; 0.5×: F1,20. 637 = 0.630, p = 0.436; 1×: F1,20.637 = 11.882,
p = 0.002, 2×: F1,20. 637 = 4.740, p = 0.041).
Differences in growth rates between resource concentrations
were larger in high temperature (Figure 2b): In high temperature,
all pairwise tests between different resource concentrations were
clearly significant (p < 0.006) except 0.05× vs. 0.1× (p = 0.203) and
1× vs. 2× (p = 0.999). In low tem per atu re, di f fer ence s bet wee n 0.05×
and 0.1× (p = 0.999), 0.5× vs. 1× (p = 0.999), 0.5× vs. 2× (p = 0.999)
were not found, whereas other pairwise comparisons yielded signif-
icant differences.
FIGURE 1 Effects of morphotype (Rz, rhizoid, R, rough) and
origin (Natural, natural waters, Farm, fish farm) (a), morphotype
and temperature (b), and origin of isolation and temperature (c) on
maximal growth rate h−1 of Flavobacterium columnare (estimated
marginal means ± SE). The growth rate was measured as a change in
OD at 420– 580 nm. Significant differences between treatments are
denoted with *, **, ***; p < 0.05, p < 0.01, p < 0.001, respectively
FIGURE 2 Effects of resource concentration (0.05×, 0.1×, 0.5×,
1×, and 2× Shieh medium) and origin of isolate (a) and temperature
(b) on maximal growth rate h−1 of Flavobacterium columnare
(estimated marginal means ± SE). The growth rate was measured
as a change in OD at 420– 580 nm. Significant differences between
treatments are denoted with *, **, and ***; p < 0.05, p < 0.01, and
p < 0.001 respectively
FIGURE 3 Effects of resource concentration (0.05×, 0.1×,
0.5×, 1× and 2× Shieh medium) and morphotype (a), and resource
concentration and temperature (b) on maximal biomass yield of
Flavobacterium columnare (estimated marginal means ± SE). The
yield was measured as maximal OD at 420– 580 nm. Significant
differences between treatments are denoted with *, **, ***;
p < 0.05, p < 0.01, p < 0.001 respectively
3.2  |  Yield
We found that biomass yield was best explained by a model with
significant main effects of temperature, colony morphology and
medium, and interactions between medium and temperature, and
medium and colony morphology (Table 2, Table S1). Higher tem-
perature led to higher biomass yield than lower temperature (yield
at 15°C = 0.492, SE = 0.024; 25°C = 0.728, SE = 0.024, Table 2).
In addition, Rz colony morphology (yield = 0.621, SE = 0.024) had
higher biomass yield than R colony morphology (yield = 0 . 599,
SE = 0.024, Table 2). The biomass yield was higher the richer the
medium (p < 0.001 for all pairwise tests between different resource
Rz morphotypes produced the highest biomass yield in inter-
mediate resource concentrations (0.5×: F1,775 = 8.795, p = 0.003;
1×: F1,775 = 22.882, p < 0.001, Figure 3a). In the smallest and in the
largest concentrations, colony types did not differ in their biomass
yield (0.05×: F1,775 = 0.105, p = 0.746, 0.1×: F1,775 = 2.773, p = 0.096,
2×: F1,775 = 0.001, p = 0.995). Increase in resource concentration in-
creased biomass yield, and the increase was larger in high tempera-
tures (Figure 3b, Table 2. 0.05× resource: F1,775 = 0.688, p = 0.407,
0.1×: F1,775 = 22.458, p < 0.001, 0.5×: F1,775 = 273.742, p < 0.001,
1×: F1,775 = 437.327, p < 0.001, 2×: F1,775 = 273.993, p < 0.001).
3.3  |  Morphotype stability
When plated on agar plates containing the nutrient concentra-
tions used in the growth experiment (0.1×, 0.5×, 1×, and 2× Shieh
medium), some of the original rough colony morphotypes grew in
a rhizoid form in concentrations from 0.1× to 1×. At the highest
resource concentration (2×), the colony spreading decreased. The
rhizoid colonies contracted and thickened, but some retained rhizoid
edges (Figure S2). This change was more pronounced for isolates
from natural waters (Fisher's exact test, p = 0.043) than for the fish
farm isolates (p = 0.090; Figure 4). Plating on 0.05 × Shieh agar did
not result in visible colonies for most bacterial cultures; hence, this
treatment was omitted from statistical analysis.
3.4  |  Virulence test
In contrast to our hypothesis, isolate origin did not affect virulence.
The best model explaining fish morbidity in the challenge experi-
ment included the effects of colony morphology, fish weight, and
their interaction (Table 3, Table S2). The virulence induced by the
rhizoid morphotype was higher than virulence induced by the rough
morphotype and increased with fish weight. For rhizoid colony type,
the mor bidity risk increased steadil y with fish weight, while the mor-
bidity risk induced by the rough colony morphology remained lower
than that induced by the rhizoid type until a sharp rise at the larg-
est fish weights (Table 3, Figure 5). Based on the modified Johnson-
Neyman procedure (see Materials and Methods), the mortality risk
did not differ between morphotypes for fish weighing over 0.55 g.
The isolates from gills of diseased fish were always rhizoid, even
when the fish were challenged with the rough morphotype. All con-
trol fish survived until the end of the experiment. No F. columnare
could be isolated at the end of the experiment from those fish who
did not exhibit disease symptoms or from control fish. Rough mor-
photypes of the isolates B392 and B067 did not cause disease in
exposed fish.
Human- induced environmental change that favors different geno-
typic and/or phenotypic properties than the natural environment
can have strong effects on organismal evolution (Baltazar- Soares
et al., 2021), including parasites and pathogens (Wolinska & King,
2009). Environmental changes particularly impact environmentally
growing or transmitted pathogens because changes affect them
directly and not only via effects on the hosts (Ashrafi et al., 2018;
Brown et al., 2012). Phenotypic plasticity, the ability of a genotype
to produce different phenotypes, has been considered important
for adaptation to alternating environments for all kingdoms of life
(DeWitt & Scheiner, 2004; Kristensen et al., 2020; Nussey et al.,
2007; Scheiner, 1993). The phenotypic variation could be expected
to be especially beneficial for environmentally growing opportunis-
tic microbial pathogens for coping with selection pressures in two
different environments, within and outside the hosts (Brown et al.,
2012; Ketola et al., 2016). Here, we show increased phenotypic vari-
ation in an opportunistic bacterial pathogen in intensive fish farming,
an environment that has been heavily modified by human activities
compared to the natural environment. Contrary to our expectations,
we did not find increased phenotypic variation in virulence in iso-
lates from fish farms. Rather, high phenotypic variation in growth of
the bacterial population together with high availability of outside-
host resources at high temperatures and simultaneous high availabil-
ity of hosts could facilitate disease outbreaks at farms. The favorable
environmental conditions for growth at fish farms along with antimi-
crobial treatments could therefore select high phenotypic variation
in bacterial population and hence affect their evolution. Our results
highlight the multifaceted effects of human- induced environmen-
tal alterations in shaping epidemiology and evolution in microbial
In clonally replicating bacteria, the heterogeneity of the phe-
notype can be achieved by inducible responses to environmental
cues and by stochastic switching between alternative phenotypes
produced by the same genotype (Ackermann, 2015; Balaban et al.,
2004). In F. columnare, a plastic response to environmental change
related to, for example, outside- host resources or presence of fish
hosts, is likely regulated by gene expression (Declercq et al., 2016,
2019; Laanto et al., 2012; Penttinen et al., 2018). Nutrient con-
centration has been found to affect expression of putative viru-
lence genes and genes associated with gliding motility differently
for the rhizoid and rough morphotypes of F. columnare (Penttinen
    PU LKKIN EN Et aL.
et al., 2016, 2018). Loss of spreading and gliding motility in F. co-
lumnare as a response to adding cortisol, the stress hormone ex-
creted by teleost fish hosts, has been suggested to be connected to
biofilm- forming ability on fish gills and hence infectivity (Declercq
et al., 2016, 2019). We found that individual isolates varied in mor-
photype stability when plated on agar at different resource con-
centrations. Our findings support the hypothesis that morphotype
change is inducible and related to nutrient searching. Variation in
morphotype stability, altering resource and host availability at fish
farms and environment are expected to affect the evolution of phe-
notypic change in F. columnare.
According to our results, the isolates originating from fish farms
responded to higher temperature with increased growth as com-
pared to isolates from natural waters. Temperature is one of the
key factors affecting the occurrence of F. columnare epidemics at
fish farms (Ashrafi et al., 2018; Pulkkinen et al., 2010). However,
the difference in growth between fish farm and natural isolates is
most likely not driven by a temperature difference between these
two environments, because the farms take their water from natural
water systems (Karvonen et al., 2010). Rather, our results point to
a response of fish farm isolates to growth conditions driven by the
interaction between temperature and other environmental factors,
such as resource availability or medical treatments. Indeed, fish farm
isolates responded to increasing resources more intensively than the
natural isolates. Results pointing to this direction were found also
in a previous study, which showed that F. columnare isolates in out-
let water of a fish farm responded more to differences in resource
concentration than isolates from inlet water (Sundberg et al., 2016).
These results show the importance of considering interactions or
correlations between the environmental variables when studying
the effects of human- induced environmental change in causing phe-
notypic change in microbial pathogens.
Increased nutrient supply in the outside- host environment can
directly affect the growth of opportunistic environmental microbes
(Brown et al., 2012; Kinnula et al., 2017; Penttinen et al., 2016;
van Elsas et al., 2011). At fish farms, the outside- host environment
can have high concentrations of fish feed and feces with protein-
rich substances 23 orders of magnitude higher than in the Shieh
medium used in our growth experiments (Naylor et al., 1999). The
accumulation of uneaten fish feed and excreta in the water is par-
ticularly evident in increased temperatures when the feeding ac-
tivity of the fish is impaired (Ellis et al., 2002; Wedemeyer, 1996).
The increased phenotypic variation in growth at farms was due to
better growth of the less virulent rough morphotype of the fish
farm isolates, suggesting that they are adapted to exploit the higher
resource concentrations in water at farms in combination with in-
creased temperatures. The resource- rich environment at fish farms
could facilitate increased phenotypic variation in growth by relaxing
FIGURE 4 Stability of the rhizoid and rough colony morphotype
on agar plates containing different concentrations of Shieh growth
medium (0.1×, 0.5×, 1×, and 2×) after plating an equal number
(five) of both colony types for isolates from natural waters (a) and
fish farm isolates (b). Some of the rough morphotypes changed into
rhizoid type in the three lowest resource concentrations, while Rz
colonies changed into R type in the highest resource concentration.
Note that one fish farm isolate did not grow on 0.1× Shieh
TAB LE 3  The effect of colony morphology (type) and fish weight
on the morbidity risk of the host fish in the virulence experiment
Source Estimate SE df χ2p
Intercept −3.05 0.3
Type (rough) −2.1 0.41 137. 67 <0.001
Weight 1.08 0.62 120.99 <0.001
Typ e × weight 3.04 1.02 18.83 0.003
FIGURE 5 The effect of colony morphology of Flavobacterium
columnare (blue: rhizoid, red: rough) and weight of the fish on fish
mortality. Thin dashed lines indicate 80% credible intervals. The
vertical line denotes critical weight (0.55 g) where, and below
which, the mortality caused by rhizoid and rough colonies differ
statistically significantly. Critical weight was approximated with
the Johnson- Neyman procedure (see Materials and Methods)
implemented on predicted mortalities based on fish size, calculated
using posterior distribution of estimates
the costs related to maintaining multiple phenotypes (Friman et al.,
2008; van Noordwijk & de Jong, 1986) and hence increase the adap-
tive value of phenotypic variation at farms (DeWitt et al., 1998; Koch
& Guillaume, 2020).
The epidemiological conditions in intensive farming that select
for fast growth and high transmission between hosts (Brown et al.,
2012; Ebert, 1998; Frank, 1996) could be expected to select for
both higher virulence and higher phenotypic variation (in virulence)
between the morphotypes in farms. We did not find difference in
virulence based on the origin of the isolates. Interestingly, our re-
sults suggest that the less virulent rough morphotype can revert
into rhizoid type expressing high virulence when in contact with
the fish. In the virulence experiment, rough morphotypes of several
isolates induced host death, and all fish exposed to bacteria from
rough morphotypes exhibited only rhizoid morphotypes in bacterial
isolates from gills of diseased fish. The virulence induced by the bac-
terial population of the rough morphotype increased sharply with
the weight of fish. At farms, growth on the organic waste from fish
feed and feces that are nutritionally close to fish tissue could pre-
adapt bacteria for infecting fish (Ketola et al., 2016; Pulkkinen et al.,
2018). A similar effect can be caused by mucins, the glycoprotein
components of the mucosa on fish skin (Almeida et al., 2019). Rough
type could be responding to the higher nutrient levels or mucins re-
leased by the larger fish hosts in the water or to a larger surface area
for bacterial attachment and growth provided by the larger fish. The
zebrafish model system used here has been shown to give a quali-
tatively similar response to bacterial doses and strains as the natu-
ral host rainbow trout (Oncorhynchus mykiss) (Kinnula et al., 2015,
2017), and therefore, we are confident that the results apply also
to natural host settings. These findings suggest that conditions in
farms do not directly select for phenotypic variation in virulence,
but phenotypic variation in growth and morphotype reversibility in
combination with environmental conditions favoring host invasion
could select for higher phenotypic variation at fish farms.
In addition to maximal growth rate, bacterial performance and
survi val are af fec ted also by yield, the amount of biomas s tha t bac te-
ria can accumulate with given resource quantity (Novak et al., 2006).
Growth rate and yield are opposing metabolic strategies that are
thought to be a trade- off in microbes (Novak et al., 2006) and hence
increased growth rate could indicate selection for the lower yield.
For F. columnare, the high growth rate in the outside- host environ-
ment is connected to high virulence in fish (Pulkkinen et al., 2010),
and thus, the ability for fast utilization of outside- host resources at
fish farms could facilitate epidemics. On the other hand, high yield
could be beneficial for environmental persistence outside fish hosts,
both in natural environment with scarcity of fish hosts but also at
fish farms during antibiotic treatments. The interaction between
temperature and resource level for yield in our experiments sug-
gests that given sufficient resources, F. columnare could continue to
build biomass even in a temperature below its optimum, while the
growth rate was strongly restricted by temperature reg ardless of re-
source level. Our results suggest that human- induced environmen-
tal nutrient enrichment could sustain the environmentally growing
opportunistic pathogens in the outside- host environment also in
suboptimal temperatures and increase the risk of epidemics when
temperature increases.
In conclusion, our study suggests that intensive farming could
act as an environment where the potential costs related to expres-
sion and maintenance of phenotypic variation are relaxed due to
the availability of high resource concentration in the outside- host
environment (Friman et al., 2008; van Noordwijk & de Jong, 1986).
In combination with this, the simultaneous high availability of hosts
facilitates host invasion. Fish farms could therefore select for higher
phenotypic variation and drive evolution in F. columnare. Our results
contribute to the growing evidence of the role of intensive farming
in inducing changes in the epidemiology of parasites and pathogens
(Atkins et al., 2013; Mennerat et al., 2010, 2017; Pulkkinen et al.,
2010; Sundberg et al., 2016) and of the potential of human- induced
environmental alterations in driving pathogen evolution. In addition,
our results pinpoint the importance of considering determinants of
not only the mean of the traits, but the variation in them, in depicting
the selective role of environments.
This study was supported by The Centre of Excellence in Biological
Interactions (research themes led by Prof. Jaana K. Bamford and
Prof. Johanna Mappes, #252411) and by Finnish Cultural Foundation
(KP), by Academy of Finland (grants #266879, #304615, #7128888,
#314939), and by Jane and Aatos Erkko foundation. The authors
want to thank J.- F. Bernardet for providing Fc isolate JIP39/87 and
H. Kunttu and P. Rintamäki for isolates from natural waters and fish
farms used in the study, and M. Nicolini, H. Kinnula and R. Penttinen
for assistance in the laboratory. This paper is dedicated to the
memory of our dear colleague and co- author Dr. Jouni Laakso, who
passed away while this paper was under revision.
The authors do not declare competing interests.
Data for this study are available at the Dryad Digital Repository t7cr.
Katja Pulkkinen
Lotta- Riina Sundberg
Ackermann, M. (2015). A functional perspective on phenotypic hetero-
geneity in microorganisms. Nature Reviews Microbiology, 13(8), 497–
508. ro3491
Adiba, S., Nizak, C., van Baalen, M., Denamur, E., & Depaulis, F. (2010).
From grazing resistance to pathogenesis: The coincidental evo-
lution of virulence factors. PLoS One, 5(8), e11882. https://doi.
org/10.1371/journ al.pone.0011882
Almeida, G. M. F., Laanto, E., A shrafi, R., & Sundberg, L. R. (2019).
Bacteriophage adherence to mucus mediates preventive
    PU LKKIN EN Et aL.
protection against pathogenic bacteria. MBio, 10 (6), 1– 12. https:// - 19
Anttila, J., Laakso, J., Kaitala, V., & Ruokolainen, L. (2016). Environmental
variation enables invasions of environmental opportunist patho-
gens. Oikos, 125(8), 1144 1152.
Ashrafi, R., Bruneaux, M., Sundberg, L. R., Pulkkinen, K., Valkonen, J., &
Ketola, T. (2018). Broad thermal tolerance is negatively correlated
with virulence in an opportunistic bacterial pathogen. Evolutionar y
Applications, 11(9), 1700– 1714.
Ashrafi, R., Pulkkinen, K., Sundberg, L. R., Pekkala, N., & Ketola, T. (2015).
A multilocus sequence analysis scheme for characterization of
Flavobacterium columnare isolates. BMC Microbiology, 15(243), 243.
h t t p s : // d o i . o r g / 1 0 . 1 1 8 6 / s 1 2 8 6 6 - 0 1 5 - 0 5 76 - 4
Atkins, K. E., Read, A. F., Savill, N. J., Renz, K. G., Islam, A. F. M. F.,
Walkden- Brown, S. W., & Woolhouse, M. E. J. (2013). Vaccination
and reduced cohort duration can drive virulence evolution: Marek's
disease virus and industrialized agriculture. Evolution, 67(3), 851
860. 5646.2012.01803.x
Balaban, N. Q., Merrin, J., Chait, R., Kowalik, L., & Leibler, S. (2004).
Bacterial persistence as a phenotypic switch. Science, 305(5690),
1622 1625. ce.1099390
Baltazar- Soares, M., Brans, K. I., & Eizaguirre, C. (2021). Human- induced
evolution: Signatures, processes and mechanisms underneath an-
thropogenic footprints on natural systems. Evolutionary Applications,
14(10), 2335– 2341.
Botero, C. A., Weissing, F. J., Wright, J., & Rubenstein, D. R. (2015).
Evol utionary tipping po ints in the capacit y to adapt to environmen-
tal change. Proceedings of the National Academy of Science s of the
United States of America, 112(1), 184– 189. /10.1073/
pnas.14 085 89111
Brown, S. P., Cornforth, D. M., & Mideo, N. (2012). Evolution of virulence
in opportunistic pathogens: Generalism, plasticity, and control.
Trends in Microbiology, 20(7 ), 336– 342. https://doi.o rg/10.1016/j.
Day, T. (2002). Virulence evolution via host exploitation and toxin pro-
duction in spore- producing pathogens. Ecology Letters, 5, 471– 476. 0248.2002.00342.x
Declercq, A. M., Aerts, J., Ampe, B., Haesebrouck, F., De Saeger, S., &
Decostere, A. (2016). Cortisol directly impacts Flavobacterium co-
lumnare in vitro growth characteristics. Veterinary Research, 47(1),
8 4 . h t t p s : / / d o i . o r g / 1 0 . 1 1 8 6 / s 1 3 5 6 7 - 0 1 6 - 0 3 7 0 - 9
Declercq, A. M., Cai, W., Naranjo, E., Thongda, W., Eeckhaut, V., Bauwens,
E., Arias, C., De La Fuente, L., Beck, B. H., Lange, M. D., Peatman,
E., Haesebrouck, F., Aerts, J., & Decostere, A. (2019). Evidence that
the stress hormone cortisol regulates biofilm formation differently
among Flavobacterium columnare isolates. Veterinary Research, 50,
2 4 . h t t p s : / / d o i . o r g / 1 0 . 1 1 8 6 / s 1 3 5 6 7 - 0 1 9 - 0 6 4 1 - 3
Declercq, A. M., Haesebrouck, F., Van den Broeck, W., Bossier, P., &
Decostere, A. (2013). Columnaris disease in fish: A review with em-
phasis on bacterium- host interactions. Veterinary Research, 44, 27.
h t t p s : // d o i . o r g / 1 0 . 1 1 8 6 / 1 2 9 7 - 9 7 1 6 - 4 4 - 2 7
Decostere, A., Haesebrouck, F., & Devriese, L. A. (1997). Shieh me-
dium supplemented with tobramycin for selective isolation of
Flavobacterium columnare (Flexibacter columnaris) from diseased
fish. Journal of Clinical Microbiology, 35(1), 322– 324. https://doi.
org/10.1128/jcm.35.1.322- 324.1997
DeW itt , T. J. , & Lange rh ans, R. B. (20 04). Integr ated solutions to envir on -
mental heterogeneity: theory of multimoment reaction norms. In T.
J. DeWitt & S. M. Scheiner (Eds.), Phenotypic plasticity: Functional
and conceptual approaches (pp. 98– 111). Oxford University Press.
DeWitt, T. J., & Scheiner, S. M. (Eds.) (20 04). Phenotypic plasticity:
Functional and conceptual approaches. Oxford University Press.
DeWitt, T. J., Sih, A ., & Wilson, D. S. (1998). Costs and limits of pheno-
typic plasticity. Trends in Ecology & Evol ution, 13(2), 77– 81. https://
d o i . o r g / 1 0 . 1 0 1 6 / S 0 1 6 9 - 5 3 4 7 ( 9 7 ) 0 1 2 7 4 - 3
Ebert, D. (1998). Experimental evolution of parasites. Science, 282, 1432–
1435. ce.282.5393.1432
Ellis, T., North, B., Scott, A. P., Bromage, N. R., Porter, M., & Gadd, D.
(2002). The relationships between stocking density and welfare
in farmed rainbow trout. Journal of Fish Biolog y, 61(3), 493– 531.
Ewald, P. W. (1994). Evolution of infectious disease. Oxford University
Frank, S. (1996). Models of parasite virulence. The Quarterly Review of
Biology, 71(1), 37– 78.
Friman, V.- P., Hiltunen, T., Jalasvuori, M., Lindstedt, C., Laanto, E.,
Örmälä, A.- M., Laakso, J., Mappes, J., & Bamford, J. K. H. (2011).
High temperature and bacteriophages can indirectly select for bac-
terial pathogenicity in environmental reservoirs. PLoS One, 6(3),
e17651. al.pone.0017651
Friman, V. P., Hiltunen, T., Laakso, J., & Kaitala, V. (20 08). Availability
of prey resources drives evolution of predator- prey interaction.
Proceedi ngs of the Royal Society B: Biological Sciences, 275(16 43),
1625– 1633.
Friman, V. P., Lindstedt, C., Hiltunen, T., Laakso, J., & Mappes, J. (2009).
Predation on multiple trophic levels shapes the evolution of patho-
gen virulence. PLoS One, 4(8), e6761.
Hayes, A. F., & Matthes, J. (2009). Computational procedures for probing
interactions in OLS and logistic regression: SPSS and SAS imple-
mentations. Behavior Research Methods, 41(3), 924– 936. https:// RM.41.3.924
Hendrix, L. J., Carter, M. W., & Scott, D. T. (1982). Covariance analyses
with heterogeneit y of slopes in fixed models. Biometrics, 38(3),
Hibbing, M. E., Fuqua, C., Parsek, M. R., & Peterson, S. B. (2010). Bacterial
competition: Surviving and thriving in the microbial jungle. Nature
Reviews Microbiology, 8(1), 15– 25.
Holland, S. L., Reader, T., Dyer, P. S., & Avery, S. V. (2014). Phenotypic
heterogeneity is a selected trait in natural yeast populations sub-
ject to environmental stress. Environmental Microbiology, 16 (6),
1729– 1740. 2920.12243
Huey, R. B., & Kingsolver, J. G. (2011). Variation in universal temperature
dependence of biological rates. Proceedings of th e National Academy
of Science s of the United States of America, 108 (26), 10377– 10378. 30108
Karvonen, A ., Rintamäki, P., Jokela, J., & Valtonen, E. T. (2010).
Increasing water temperature and disease risks in aquatic systems:
Climate change increases the risk of some, but not all, diseases.
International Journal for Parasitology, 40(13), 1483– 1488 . https://
Ketola, T., Mikonranta, L., Laakso, J., & Mappes, J. (2016). Different food
sources elicit fast changes to bacterial virulence. Biology Letters,
12(1), 20150660.
Ketola, T., Mikonranta, L., Zhang, J. I., Saarinen, K., Örmälä, A.- M., Friman, V.- P.,
Mappes, J., & Laakso, J. (2013). Fluctuating temperature leads to evo-
lution of thermal generalism and preadaptation to novel environments.
Evolution, 67(10), 2936– 2944.
Kinnula, H., Mappes, J., Valkonen, J. K., Pulkkinen, K., & Sundberg, L. R.
(2017). Higher resource level promotes virulence in an environmen-
tally transmitted bacterial fish pathogen. Evolutionary Applications,
10(5), 462– 470.
Kinnula, H., Mappes, J., Valkonen, J. K., & Sundberg, L. R. (2015). The
influence of infective dose on the virulence of a generalist patho-
gen in rainbow trout (Oncorhynchus mykiss) and zebra fish (Danio
rerio). PLoS One, 10 (9), e0139378.
Koch, E. L., & Guillaume, F. (2020). Restoring ancestral phenotypes is
a general pattern in gene expression evolution during adaptation
to new environments in Tribolium castaneum. Molecular Ecology,
29(20), 3938– 3953.
Koh, K. S., Lam, K. W., Alhede, M., Queck, S. Y., Labbate, M., Kjelleberg,
S., & Rice, S. A. (2007). Phenotypic diversification and adaptation
of Serratia marcescens MG1 biofilm- derived morphotypes. Journal
of Bacteriology, 189(1), 119– 130.
- 0 6
Kreibich, S., & Hardt, W. D. (2015). Experimental approaches to pheno-
typic diversity in infection. Current Opinion in Microbiology, 27, 25–
Kristensen, T. N., Hoffmann, A . A., Overgaard, J., Sorensen, J. G., Hallas,
R., & Loeschcke, V. (2008). Costs and benefits of cold acclimation
in field- released Drosophila. Proceedings of the National Academy of
Sciences of the United States of America, 105(1), 216– 221. https:// 74105
Kristensen, T. N., Ketola, T., & Kronholm, I. (2020). Adaptation to en-
vironmental stress at different timescales. Annals of the New
York Academy of Sciences, 1476 (1), 5– 22. .1111/
nyas .13974
Kunttu, H. M., Sundberg, L. R., Pulkkinen, K., & Valtonen, E. T. (2012).
Environment may be the source of Flavobacterium columnare out-
breaks at fish farms. Environmental Microbiology Reports, 4(4), 398–
402. 2229.2012.00342.x
Kunttu, H. M., Suomalainen, L. R., Jokinen, E. I., & Valtonen, E. T. (2009).
Flavobacterium columnare colony types: connection to adhesion
and virulence? Microbial Pathogenesis, 46(1), 21– 27. https://doi.
org/10.1016/j.micpa th.2008.10.0 01
Kunttu, H. M., Valtonen, E. T., Jokinen, E. I., & Suomalainen, L . R.
(2009). Saprophytism of a fish pathogen as a transmission
strategy. Epidemics, 1(2), 96– 100.
Kussell, E., & Leibler, S. (2005). Phenotypic diversity, population growth,
and information in fluctuating environments. Science, 309(574 3),
2075– 2078. ce.1114383
Laanto, E., Bamford, J. K., Laakso, J., & Sundberg, L. R. (2012). Phage-
driven loss of virulence in a fish pathogenic bacterium. PLoS One,
7(12), e53157. al.pone.0053157
Laanto, E., Penttinen, R. K., Bamford, J. K., & Sundberg, L. R. (2014).
Comparing the different morphotypes of a fish pathogen–
implications for key virulence factors in Flavobacterium
columnare. BMC Microbiology, 14 (170), 170. https://doi.
org /10.1186/1471- 218 0- 14- 170
Levins, R. (1968). Evolution in changing environments: Some theoretical ex-
plorations. Princeton University Press.
Mennerat, A., Hamre, L., Ebert, D., Nilsen, F., Davidova, M., &
Skorping, A. (2012). Life history and virulence are linked
in the ectoparasitic salmon louse Lepeophtheirus salmonis.
Journal of Evolutionary Biology, 25(5), 856– 861. https://doi.
org/10.1111/j.1420- 9101.2012.02474.x
Mennerat, A., Nilsen, F., Ebert, D., & Skorping, A . (2010). Intensive
farming: Evolutionary implications for parasites and pathogens.
Evolutionary Biology, 37(2– 3), 59– 67.
2 - 0 1 0 - 9 0 8 9 - 0
Mennerat, A., Ugelvik, M. S., Hakonsrud Jensen, C., & Skorping, A.
(2017). Invest more and die faster: The life history of a parasite on
intensive farms. Evolutionary Applications, 10(9), 890– 896. https://
Mideo, N., & Reece, S. E. (2012). Plasticity in parasite phenotypes:
Evolutionary and ecological implications for disease. Future
Microbiology, 7(1), 17– 24. ht tp s://
Naylor, S. J., Moccia, R. D., & Durant, G. M. (1999). The chemical com-
position of settleable solid fish waste (manure) from commercial
rainbow trout farms in Ontario, Canada. North American Journal of
Aquaculture, 61(1), 2126.
Novak, M., Pfeif fer, T., Lenski, R. E., Sauer, U., & Bonhoeffer, S. (2006).
Experimental tests for an evolutionary trade- off between growth
rate and yield in E. coli. The American Naturalist, 168(2), 242– 251. 86/506527
Nussey, D. H., Wilson, A. J., & Brommer, J. E. (2007). The evolution-
ary ecology of individual phenotypic plasticity in wild popula-
tions. Journal of Evolutionary Biology, 20(3), 831– 844. https://doi.
org /10.1111/j .1420 - 9101.20 07.013 00.x
Padilla, D. K., & Adolph, S. C. (1996). Plastic inducible morphologies are
not always adaptive: The importance of time delays in a stochas-
tic environment. Evolutiona ry Ecology, 10 (1), 105– 117. ht tps://doi.
org/10.1007/BF012 39351
Penttinen, R., Hoikkala, V., & Sundberg, L. R. (2018). Gliding motility and
expression of motility- related genes in spreading and non- spreading
colonies of Flavobacterium columnare. Frontiers in Microbiology, 9,
Penttinen, R., Kinnula, H., Lipponen, A., Bamford, J. K., & Sundberg, L.
R. (2016). High nutrient concentration can induce virulence factor
expression and cause higher virulence in an environmentally trans-
mitted pathogen. Microbial Ecology, 72(4), 955– 964. https://doi.
o r g / 1 0 . 1 0 0 7 / s 0 0 2 4 8 - 0 1 6 - 0 7 8 1 - 1
Pulkkinen, K., Pekkala, N., Ashrafi, R., Hämäläinen, D. M., Nkembeng, A. N.,
Lipponen, A., Hiltunen, T., Valkonen, J. K., & Taskinen, J. (2018). Effect
of resource availability on evolution of virulence and competition in
an environmentally transmitted pathogen. FEMS Microbiology Ecology,
94(5), fiy060. c/fiy060
Pulkkinen, K., Suomalainen, L. R., Read, A. F., Ebert, D., Rintamäki, P.,
& Valtonen, E. T. (2010). Intensive fish farming and the evolution
of pathogen virulence: The case of columnaris disease in Finland.
Proceedi ngs of the Royal Society B: Biological Sciences, 277(1681),
593– 6 00. ht tps://
R Development Core Team. (2015). R: A lan guage and environment for sta-
tistical computing. R Foundation for Statistical Computing. http://
w w w . R - p r o j e c t . o r g
Rainey, P. B., & Travisano, M. (1998). Adaptive radiation in a hetero-
geneous environment. Nature, 394(6688), 69– 72. https://doi.
org/10.103 8/27900
Reed, T. E., Waples, R. S., Schindler, D. E., Hard, J. J., & Kinnison, M. T.
(2010). Phenotypic plasticity and population viability: The impor-
tance of environmental predictability. Proceedings of the Royal
Society B: Biological Sciences, 277(1699), 3391– 3400. https://doi.
org/10.1098/rspb.2010. 0771
Scheiner, S. M. (1993). Genetics and evolution of phenotypic plasticity.
Annual Review of Ecology and Systematics, 24(1), 35– 68. https://doi.
Schmid- Hempel, P. (2009). Immune defence, parasite evasion strategies
and their relevance for ‘macroscopic phenomena’ such as virulence.
Philosophical Transac tions of the Royal Societ y of London. Serie s B,
Biological Sciences, 364 (1513), 85– 98.
Song, Y. L., Fryer, J. L., & Rohovec, J. S. (1988). Comparison of six media
for the cultivation of Flexibacter columnaris. Fish Pathology, 23, 91–
Stan Development Team. (2020). RStan: The R inter face to Stan. R package
version 2.21. 2. ht tp://mc-
Sundberg, L. R., Ketola, T., Laanto, E., Kinnula, H., Bamford, J. K.,
Penttinen, R., & Mappes, J. (2016). Intensive aquaculture selects
for increased virulence and interference competition in bacteria.
Proceedi ngs of the Royal Society B: Biological Sciences, 283(1826),
Sundberg, L. R., Kunttu, H. M., & Valtonen, E. T. (2014). Starvation can
dive rsify th e populat ion stru ctu re and viru len ce st rateg ies of an en-
vironmentally transmitting fish pathogen. BMC Microbiology, 14, 67.
h t t p s : // d o i . o r g / 1 0 . 1 1 8 6 / 1 4 7 1 - 2 1 8 0 - 1 4 - 6 7
Suomalainen, L. R., Kunttu, H., Valtonen, E. T., Hir vela- Koski, V., & Tiirola,
M. (2006). Molecular diversity and growth features of Flavobacterium
columnare strains isolated in Finland. Diseases of Aquatic Organisms,
70(1– 2), 55– 61. 0055
    PU LKKIN EN Et aL.
Tonsor, S. J., Elnaccash, T. W., & Scheiner, S. M. (2013). Developmental
instability is genetically correlated with phenotypic plasticity, con-
straining heritability, and fitness. Evolution, 67(10), 2923– 2935.
Ugelvik, M. S., Skorping, A., Moberg, O., & Mennerat, A. (2017).
Evolution of virulence under intensive farming: salmon lice increase
skin lesions and reduce host growth in salmon farms. Journal of
Evolutionary Biology, 30(6), 1136– 1142. https://doi .org/10.1111/
van Elsas, J. D., Semenov, A. V., Costa, R., & Trevors, J. T. (2011). Survival
of Escherichia coli in the environment: Fundamental and pub-
lic health aspects. The ISME Journal, 5(2), 173– 183. https://doi.
org/10.103 8/ismej.2010.80
van Noordwijk, A . J., & de Jong, G . (1986). Acquisition and allocation of
resources: Their influence on variation in life history tactics. The
American Naturalist, 128, 137– 1 42 . ht tp s://d oi .or g /10 .10 86 /28 45 47
Velicer, G. J., & Lenski, R . E. (1999). Evolutionar y trade- offs under condi-
tions of resource abundance and scarcity: Experiments with bacte-
ria. Ecology, 80(4), 1168– 1179.
Venables, W. N., & Ripley, B. D. (2002). Modern a pplied s tatistic s with S
(4th ed.). Springer- Verlag.
Via, S., Gomulkiewicz, R., De Jong, G., Scheiner, S. M., Schlichting, C. D.,
& Van Tienderen, P. H. (1995). Adaptive phenotypic plasticity: con-
sensus and controversy. Trends in Ecology & Evolution, 10(5), 212–
2 1 7 . h t t p s : / / d o i . o r g / 1 0 . 1 0 1 6 / S 0 1 6 9 - 5 3 4 7 ( 0 0 ) 8 9 0 6 1 - 8
Wagner, B. A., Wise, D. J., Khoo, L. H., & Terhune, J. S. (20 02). The epide-
miology of bacterial diseases in food- size channel catfish. Journ al of
Aquatic Animal Health, 14(4), 263– 272.
Wedemeyer, G. A. (1996). Physiology of fish in intensive culture systems.
Chapman & Hall.
Wolinska, J., & King, K. C. (2009). Environment can alter selection in
host- parasite interactions. Trends in Parasitology, 25(5), 236– 244.
Zhang, J., Laakso, J., Mappes, J., Laanto, E., Ketola, T., Bamford, J. K.,
Kunttu, H., & Sundberg, L. R. (2014). Association of colony mor-
photypes with virulence, growth and resistance against protozoan
predation in the fish pathogen Flavobacterium columnare. FEMS
Microbiology Ecology, 89(3), 553562 . ht tps:// /10.1111/157
4- 6941.12356
Additional supporting information may be found in the online
version of the article at the publisher’s website.
How to cite this article: Pulkkinen, K., Ketola, T., Laakso, J.,
Mappes, J., & Sundberg, L.- R. (2022). Rich resource
environment of fish farms facilitates phenotypic variation
and virulence in an opportunistic fish pathogen. Evolutionary
Applications, 15, 417428. htt ps:// .13355
Shrimp diseases frequently occur during pre-adult to adult lifestages, when the rearing water temperature and nutrient level are high. This phenomenon provides clues that the virulence of pathogens might be induced by elevated temperature and nutrient, whereas the underlying molecular mechanism is lacking. To address this pressing knowledge, we explored how elevated temperature and nutrient stimulate the expression profiles of Vibrio parahaemolyticus, a notorious pathogen causing shrimp acute hepatopancreatic necrosis disease (AHPND). The stimulated pathogenicity was verified by pretreatment of V. parahaemolyticus under elevated temperature or nutrient condition, which caused significantly faster and higher rate of shrimp mortality. RNA-Seq analysis revealed that elevated temperature and nutrient significantly altered the transcriptional patterns of V. parahaemolyticus, resulting in 232 and 82 differential genes, respectively. Weighted Gene Co-Expression Network Analysis depicted that the two conditions induced different gene expression patterns. Specifically, elevated temperature enriched genes involved in arginine biosynthesis and flagellar assembly, whereas elevated nutrition stimulated genes facilitating citrate cycle, cysteine and methionine metabolism pathways. Based on the profiles of differential genes, we inferred that high temperature negatively regulated OpaR and ToxR, thereby activating type VI secretion system 1 (T6SS1). Concurrently, the induced expressions of adhesion Multivalent Adhesion Molecule 7 (MAM7) and flagella-related genes enable V. parahaemolyticus to actively attack host under elevated temperature. By contrast, elevated nutrient activates the T6SS1 through TfoY. These findings provide comprehensive insights into the molecular mechanism implicating in the pathogenicity of V. parahaemolyticus, and partially explain why disease frequency is increased under the scenarios of increasing temperature and eutrophication in aquaculture and natural ecosystems.
Full-text available
Significance Incipient species are at an intermediate stage of speciation where reproductive isolation is counteracted by the homogenizing effects of gene flow. Human activity sometimes leads such species to reunite, as seen in the Orange Sulphur butterfly, which forms large hybridizing populations with the Clouded Sulphur in alfalfa fields. Here we show that sex chromosomes maintain these species as distinct, while the rest of their genome is admixed. Sex chromosomes notably determine which males display to females a bright, iridescent UV signal on their wings. Genetic mapping, antibody stainings, and CRISPR knockouts collectively indicate that the gene bric a brac controls whether UV-iridescent nanostructures develop in each species, illustrating how a master switch gene modulates a male courtship signal.
Full-text available
Abstract The impact of human activities on the global environment has increased to such an extent that the current geological era has been coined the Anthropocene. Studies dedicated to understanding the evolutionary consequences of human‐induced selection on all levels of diversity (species, populations, traits, genes) provide direct knowledge about the mechanisms underlying species' responses and their evolutionary potential. A better understanding of the effects of human‐induced selection is needed to leverage evolved mechanisms to develop appropriate conservation programmes to guarantee the maintenance of healthy systems. In this special issue, we focus on different types of human‐mediated selection pressures, from the direct harvesting of individuals (e.g. hunting, fishing), to the more pervasive effects of climate change. Contributions highlight the diversity of human‐induced selection pressures ranging from fisheries, trophy‐hunting, poaching and domestication to climate change, and pollution. With those, we question whether there are parallel evolutionary solutions across fisheries systems, whether hunting pressures alter population dynamics and population structure, and whether climate change is an evolutionary dead‐end. The contributions reflect the direction of travel of the field and the solutions to mitigate the impact of human activities.
Full-text available
A big question in behavioral ecology is what drives diversity of color signals. One possible explanation is that environmental conditions, such as light environment, may alter visual signaling of prey, which could affect predator decision-making. Here, we tested the context-dependent predator selection on prey coloration. In the first experiment, we tested detectability of artificial visual stimuli to blue tits (Cyanistes caeruleus) by manipulating stimulus luminance and chromatic context of the background. We expected the presence of the chromatic context to facilitate faster target detection. As expected, blue tits found targets on chromatic yellow background faster than on achromatic grey background whereas in the latter, targets were found with smaller contrast differences to the background. In the second experiment, we tested the effect of two light environments on the survival of aposematic, color polymorphic wood tiger moth (Arctia plantaginis). As luminance contrast should be more detectable than chromatic contrast in low light intensities, we expected birds, if they find the moths aversive, to avoid the white morph which is more conspicuous than the yellow morph in low light (and vice versa in bright light). Alternatively, birds may attack first moths that are more detectable. We found birds to attack yellow moths first in low light conditions, whereas white moths were attacked first more frequently in bright light conditions. Our results show that light environments affect predator foraging decisions, which may facilitate context-dependent selection on visual signals and diversity of prey phenotypes in the wild.
Full-text available
Sexually dimorphic development is responsible for some of the most remarkable phenotypic variation found in nature. Alternative splicing of the transcription factor gene doublesex is a highly conserved developmental switch controlling the expression of sex specific pathways. Here, we leverage sex-specific differences in butterfly wing color pattern to characterize the genetic basis of sexually dimorphic development. We use RNA-seq, immunolocalization, and motif binding site analysis to test specific predictions about the role of Doublesex in the development of structurally-based ultraviolet (UV) wing patterns in Zerene cesonia (Southern Dogface). Unexpectedly, we discover a novel duplication of Doublesex that shows a sex-specific burst of expression associated with the sexually dimorphic UV coloration. The derived copy consists of a single exon that encodes a DNA binding but no protein binding domain, and has experienced rapid amino-acid divergence. We propose the novel dsx paralog may suppress UV scale differentiation in females, which is supported by an excess of Dsx binding sites at cytoskeletal and chitin-related genes with sex-biased expression. These findings illustrate the molecular flexibility of the dsx gene in mediating the differentiation of secondary sexual characteristics.
Full-text available
Coloration is perhaps one of the most prominent adaptations for survival and reproduction of many taxa. Coloration is of particular importance for aposematic species, which rely on their coloring and patterning acting as a warning signal to deter predators. Most research has focused on the evolution of warning coloration by natural selection. However, little information is available for color mutants of aposematic species, particularly at the genomic level. Here, I compare the transcriptomes of albino mutant caterpillars of the aposematic wood tiger moth (Arctia plantaginis) to those of their full sibs having their distinctive orange‐black warning coloration. The results showed >290 differentially expressed genes genome‐wide. Genes involved in the immune system, structural constituents of cuticular, and immunity were mostly downregulated in the albino caterpillars. Surprisingly, higher expression was observed in core melanin genes from albino caterpillars, suggesting that melanin synthesis may be disrupted in terminal ends of the pathway during its final conversion. Taken together, these results suggest that caterpillar albinism may not be due to a depletion of melanin precursor genes. In contrast, the albino condition may result from the combination of faulty melanin conversion late in its synthesis and structural deficiencies in the cuticular preventing its deposition. The results are discussed in the context of how albinism may impact individuals of aposematic species in the wild. Albinism in warningly colored caterpillars.
Full-text available
Warning signals are predicted to develop signal monomorphism via positive frequency‐dependent selection (+FDS) albeit many aposematic systems exhibit signal polymorphism. To understand this mismatch, we conducted a large‐scale predation experiment in four countries, among which the frequencies of hindwing warning coloration of the aposematic moth, Arctia plantaginis, differ. Here we show that selection by avian predators on warning colour is predicted by local morph frequency and predator community composition. We found +FDS to be the strongest in monomorphic Scotland and lowest in polymorphic Finland, where the attack risk of moth morphs depended on the local avian community. +FDS was also found where the predator community was the least diverse (Georgia), whereas in the most diverse avian community (Estonia), hardly any models were attacked. Our results support the idea that spatial variation in predator communities alters the strength or direction of selection on warning signals, thus facilitating a geographic mosaic of selection. A geographic mosaic of selection by predators could explain the paradoxical maintenance of warning signal variation, but direct ecological evidence is scarce and focused on tropical systems. We monitored local avian predators and attacks on 4000 + moth models representing red, yellow or white warning colour morphs in a temperate moth system with natural variation in local morph frequencies. We found positive frequency‐dependent selection to be strongest in monomorphic populations and the direction and strength of selection to be significantly associated with local predator community composition and diversification, which can explain not only geographic variation (polytypism) but also local polymorphism when coupled with gene flow.
Most animal species on Earth are insects, and recent reports suggest that their abundance is in drastic decline. Although these reports come from a wide range of insect taxa and regions, the evidence to assess the extent of the phenomenon is sparse. Insect populations are challenging to study, and most monitoring methods are labor intensive and inefficient. Advances in computer vision and deep learning provide potential new solutions to this global challenge. Cameras and other sensors can effectively, continuously, and noninvasively perform entomological observations throughout diurnal and seasonal cycles. The physical appearance of specimens can also be captured by automated imaging in the laboratory. When trained on these data, deep learning models can provide estimates of insect abundance, biomass, and diversity. Further, deep learning models can quantify variation in phenotypic traits, behavior, and interactions. Here, we connect recent developments in deep learning and computer vision to the urgent demand for more cost-efficient monitoring of insects and other invertebrates. We present examples of sensor-based monitoring of insects. We show how deep learning tools can be applied to exceptionally large datasets to derive ecological information and discuss the challenges that lie ahead for the implementation of such solutions in entomology. We identify four focal areas, which will facilitate this transformation: 1) validation of image-based taxonomic identification; 2) generation of sufficient training data; 3) development of public, curated reference databases; and 4) solutions to integrate deep learning and molecular tools.
Plasticity and evolution are two processes allowing populations to respond to environmental changes, but how both are related and impact each other is still controversial. We studied plastic and evolutionary responses in gene expression of Tribolium castaneum after beetles’ exposure to new environments that differed from ancestral conditions in temperature, humidity or both. Using experimental evolution with ten replicated lines per condition, we were able to demonstrate adaptation after 20 generations. We measured whole‐transcriptome gene expression with RNA‐seq to infer evolutionary and plastic changes. We found more evidence for changes in mean expression (shift in the intercept of reaction norms) in adapted lines than for changes in plasticity (shifts in slopes). Plasticity was mainly preserved in selected lines and was responsible for a large part of the phenotypic divergence in expression between ancestral and new conditions. However, we found that genes with the largest evolutionary changes in expression also evolved reduced plasticity and often showed expression levels closer to the ancestral stage. Results obtained in the three different conditions were similar, suggesting that restoration of ancestral expression levels during adaptation is a general evolutionary pattern. With a larger sample in the most stressful condition, we were able to detect a positive correlation between proportion of genes with reversion of the ancestral plastic response and mean fitness per selection line.
Canaries changing colors Many animals are sexually dimorphic, with different phenotypes in males and females. To identify the genetic basis of sexual differences in bird coloration, Gazda et al. investigated red coloration in mosaic canaries and related species (see the Perspective by Chen). Using a combination of genetic crosses, genomic mapping, transcriptomics, and comparative analyses, the authors show that trans-regulation of the carotenoid-processing gene BCO2 is involved in sexual dichromatism. Although such variation in coloration among the sexes is common, particularly in birds, there are few candidate genes known to be involved. This study helps to elucidate the molecular mechanisms that underlie the evolution of dichromatism and may aid in uncovering sexually selected traits. Science , this issue p. 1270 ; see also p. 1185