ArticlePDF Available

Evolutionary adaptation of high‐diversity communities to changing environments


Abstract and Figures

We use adaptive dynamics models to study how changes in the abiotic environment affect patterns of evolutionary dynamics and diversity in evolving communities of organisms with complex phenotypes. The models are based on the logistic competition model, and environmental changes are implemented as a temporal change of the carrying capacity as a function of phenotype. In general, we observe that environmental changes cause a reduction in the number of species, in total population size, and in phenotypic diversity. The rate of environmental change is crucial for determining whether a community survives or undergoes extinction. Until some critical rate of environmental changes, species are able to follow evolutionarily the shifting phenotypic optimum of the carrying capacity, and many communities adapt to the changing conditions and converge to new stationary states. When environmental changes stop, such communities gradually restore their initial phenotypic diversity. We use adaptive dynamics models to study how changes in the abiotic environment, such as climate change, affect patterns of evolutionary dynamics and diversity in evolving communities of organisms with complex phenotypes. We observe that the rate of environmental change is a crucial factor that determines whether the community survives and adapts to the changing conditions or undergoes extinction. In the majority of surviving systems, we observe a reduction in the number of species, total population size, and phenotypic diversity.
Content may be subject to copyright.
Ecology and Evolution. 2020;10:11941–11953.
Over the past decades, the issue of the impact of changing envi-
ronmental conditions on species and ecosystems has gained in-
creasing prominence, particularly in the context of global warming
(Trisos, Merow, & Pigot, 2020). Recent estimates have shown that
at the current rate of global warming, one of six species will become
extinct (Urban, 2015), and empirical evidence support s this finding
(Maclean & Wilson, 2011). Already there are species whose extinc-
tion occurred as a result of climate change. For example, the sea
level rise has destroyed the habit at of mosaic-tailed rat (Melomys
rubicola) and individuals of this species have not been seen since
2009 (Gynther, Waller, & Leung, 2016). Many species, such as polar
bears (Ursus maritimus), experience ecological stress. For hunting,
this species relies on sea ice, where seals, their primary source of
food, rest, and breed. Reduction in ice surfaces forces polar bears
to overcome long distances by swimming and thus strongly affec ts
the balance between anabolism and catabolism (Lone et al., 2018).
Several studies detected muscle atrophy and weight loss in polar
bears because of starvation and changes in metabolism of lipids
(Griffen, 2018; Obbard et al., 2016; Pagano et al., 2018; Tartu et al.,
2017; Whiteman et al., 2017).
However, the majority of ecosystems are characterized by ex ten-
sive adaptability. While changing environmental factors of ten lead to
diversit y reduction, in general many ecosystems will likely survive. For
example, this is currently observed in coral reefs. Increasing tempera-
ture and acidification of the ocean water affect the symbiotic relation-
ships between corals and microalgae in such a way that corals expel
their end osymbionts and b leach (Pogoreutz et al., 2018). Without ben-
efits of symbiosis, corals experience higher mor tality, become more
Received: 10 March 2020 
  Revised: 4 July 2020 
  Accepted: 15 July 2020
DOI: 10.1002/ece3.6695
Evolutionary adaptation of high-diversity communities to
changing environments
Evgeniia Alekseeva1| Michael Doebeli2| Iaroslav Ispolatov3
This is an op en access article under t he terms of the Creat ive Commons Attributio n License, which permits use, dist ribution and reproduc tion in any medium,
provide d the orig inal work is proper ly cited .
© 2020 The Authors . Ecology and Evolution published by John Wiley & Sons Ltd
1Skoltech, Moscow, Russia
2University of Br itish Columbia, Vancouver,
British Columbia, Canada
3Universidad de Santiago de C hile (USACH),
Santiago, Chile
Evgeniia Al ekseeva, Skolte ch, Moscow,
Iaroslav Ispolatov, Universid ad de Sant iago
de Chile (U SACH), Sant iago, Chile.
Funding information
Skoltech Academic Mobility Program;
Natural Science s and Engineering Research
Council of Canada; Depar tamento de
Investigaciones Científicas y Tecnológicas,
Universidad de Santiago de Chile, Grant/
Award Number: 041931Y
We use adaptive dynamics models to study how changes in the abiotic environment
affect patterns of evolutionary dynamics and diversity in evolving communities of or-
ganisms with complex phenotypes. The models are based on the logistic competition
model, and environmental changes are implemented as a temporal change of the car-
rying capacity as a function of phenotype. In general, we observe that environmental
changes cause a reduction in the number of species, in total population size, and in
phenotypic diversity. The rate of environmental change is crucial for determining
whether a community survives or undergoes extinction. Until some critical rate of
environmental changes, species are able to follow evolutionarily the shifting pheno-
typic optimum of the carrying capacity, and many communities adapt to the chang-
ing conditions and converge to new stationary states. When environmental changes
stop, such communities gradually restore their initial phenotypic diversity.
adaptation, environmental changes, extinction
sensitive to diseases, and decline. However, because strains of both
host and endosymbiont vary in their sensitivity to higher temperatures,
some of them can form a thermo-tolerant symbiosis (Baker, 2003; Bay,
Rose, Logan, & Palumbi, 2017; Grottoli et al., 2018; Little, Van Oppen,
& Willis, 20 04; Smith, Hume, Delaney, Wiedenmann, & Burt, 2017).
Such examples already exist in zones with extreme temperatures,
such as the Arabian/Persian Gulf (PAG) (Baker, Starger, McClanahan, &
Glynn, 2004); hence, this symbiotic communit y can adapt, in principle,
to changing environments despite a decrease in diversity of its par-
ticipant s. In a recent study of time series of species composition from
various geographical areas, Blowes et al. (2019) made a prediction that
with time climate change will mostly cause large-scale reorganization
of biodiversity rather than its global decline.
The challenges imposed by changing environments vary widely
between ecosystems and between species within ecosystems, as,
for example, the warming climate induces a diverse spectrum of
interconnected changes in environmental conditions and weather
patterns. Besides global climatic changes, there are numerous other
examples of how anthropogenic activit y disturbs ecosystems locally
by environmental pollution, poaching, modification of geographical
landscapes and many others factors (Laskar, Mahata, & Liang, 2016;
Scheffers, Oliveira, Lamb, & Edwards, 2019). Adaptation to changing
environments is also an impor tant topic of research in the context of
preventing the development of antibiotic and drug resistance. So, in
one way or another, biological populations frequently face a chang-
ing environment, which is an important force in their evolution.
Adaptation to environmental changes has been the subject of
both experimental and theoretical research. A nice example of an ex-
perimental study of evolution in an artificially created changing en-
vironment is the work on gradual bacterial adaptation to increasing
doses of antibiotic on a giant Petri dish (Baym et al., 2016). Another
example of experimental adaptation to a changing environment was
observed in a study of phytoplankton biodiversity in increasingly
warm water ( Yvon-Durocher et al., 2015). However, the slowness of
evolutionary processes on human timescales sets strong restrictions
of what can be done in such experiments.
Free of this limitation, theoretical studies are by far more nu-
merous. For example in Botero, Weissing, Wright, and Rubenstein
(2015), the authors found that certain types of climatic changes
force populations to cross “tipping points” and to switch from one
adaptive strategy to a qualitatively different one, which often leads
to extinction despite the successful adaptation in the context of the
previously used strategy. Another theoretical study has revealed
the influence of genetic variance and spatial dispersal on the suc-
cess of a given number of competing species subjected to changing
conditions (Norberg, Urban, Vellend, Klausmeier, & Loeuille, 2012).
The combination of high genetic variance and low spatial dispersal is
the most conducive to adaptation and survival of the species under
the effect of climatic change. In Jones (2008) and Northfield and
Ives (2013), the authors have investigated how coevolution in pairs
of species with various types of ecological interactions affects the
process of adaptation to environmental changes. They argued that
types of coevolution with conflicting interests help species to adapt,
often counterintuitively, whereas types of coevolution with noncon-
flicting interests enhance the detrimental effect of climatic changes.
These conclusions are corroborated by the studies specifically fo-
cused on predator–prey interac tions (Mellard, de Mazancourt, &
Loreau, 2015; Osmond, Otto, & Klausmeier, 2017), which demon-
strate that under certain conditions, and especially when the
predator–prey interaction trait is aligned with the direction of en-
vironmental change, predation increases adaptability and resilience.
The adaptive dynamics and individual-based model described in
Johansson (2008) and later adapted to different scenarios in many
subsequent studies, predicts a decrease in diversity and possible
complete extinction in the system of 1–3 competing species sub-
ject to an environmental change at a constant or fluctuating rate. An
extension of the model (Johansson, 2008) was used to investigate
multi-faceted effects of asymmetric resource availability or com-
petition between two species on the response to a changing envi-
ronment ( Van Den Elzen, Courtney, Kleynhans, & Otto, 2017). More
complex scenarios of multi-patch environment with migration (De
Mazancourt, Johnson, & Barraclough, 2008) and cyclic changes in
the environmental gradient (De Mazancourt et al., 2008) have also
been investigated in the context of effects of the climate change on
species diversity and population.
In this work, we take a somewhat different look at the influence
of environmental changes on an evolving system. As is widely re-
ported, the problem with environmental changes is often not so
much the actual state of the environmental variable, such as the
global temperature or the CO2 concentration, but the high and
previously unseen rates at which these variables change. Thus, in
this work we investigate how an ecosystem, modeled as a commu-
nity of interacting and evolving species, reacts to environmental
changes of various rates. Similarly to (Johansson, 2008), we focus
on the par ticular case of competing species, ignoring for now other
ecological interactions, and consider a diversifying community de-
scribed by a logistic competition model. However, as an extension
of (Johansson, 2008), we consider a more realistic scenario where
the competition between individuals is controlled by more than one
phenotypic traits. Previously, it was shown that in such systems, the
number of traits or the dimension of phenotype space affec ts diver-
sification, with higher dimensions leading to higher diversity (Doebeli
& Ispolatov, 2017). As a communit y diversifies from low numbers
of species, the rates of evolution and diversification slow down as
the saturation level of diversit y that the environment can sustain is
reached. Higher rates of evolution and diversification can be reacti-
vated only when the level of saturation decreases, which can happen
with aromorphosis and an extension of the phenotypic space into
higher dimensions, or through c atastrophic events, leading to mass
extinction (Ispolatov, Alekseeva, Alekseeva, & Doebeli, 2019).
However, it is not known which ecological and evolutionary pro-
cesses unravel in such a system when external intervention, such as
ongoing climate changes, continues indefinitely. A naive qualitative
guess (which turns out to be correc t) would be that if the rate of
change associated with such an intervention is much smaller than
some intrinsic adaptation rate of all species, the relative phenotypic
distribution of the species would remain almost int act and all spe-
cies would synchronously follow the environmental change with a
certain lag. Yet it is hard to predict even qualitatively what happens
when the rate of environmental changes increases, apart from the
ultimate ex tinction of all species when environmental change is very
fast. We therefore perform a systematic study of various ecological
and evolutionary indicators of the evolving communities for a wide
range of rates of environmental changes. To consider the most gen-
eral case, we, similarly to Van Den Elzen et al. (2017), consider an
asymmetric competition kernel, albeit in a different from Van Den
Elzen et al. (2017) form. The evolutionary dynamics in multidimen-
sional phenotype space with generally asymmetric competition is
usually complicated (Doebeli & Ispolatov, 2017) and even unpredict-
able (Doebeli & Ispolatov, 2014). Thus, complimentar y to many exist-
ing studies, we use a statistical approach, averaging results for each
rate of environmental change over many simulated replicas.
2.1 | The model
Following Doebeli and Ispolatov (2017) and Ispolatov et al. (2019),
we study a system that can be populated by a var ying number of
phenotypic species, each defined by its phenotype
Here and in the following all vector notations should have bar
above the variable rather than below.
in D-dimensional phenotype space. In monomorphic communi-
ties consisting of a single species with phenotype
, the population
size of that species at ecological equilibrium is given by the carrying
capacity function
, which is assumed to have the following form:
determines the width of the carrying capacity. This function
has its maximum value of 1 at the point that we call the center of the
carrying capacity (CCC)
. Thus, the population size of
a monomorphic species is maximal if the phenotype of that species is
equal to
Competition between two species with distinct phenotypes
is described by the competition kernel
, so that the com-
petitive effect of
is given by
There are t wo terms in the exponent of the competition kernel.
Similarly to Van Den Elzen et al. (2017), we believe that in general
the competition is nonsymmetric. Therefore, the first term rep-
resents the simplest nonsymmetric contribution to the competition,
which may result in complex evolutionary dynamics, that is, cyclic or
chaotic evolutionar y trajectories. Since we expect the evolutionary
dynamic s to unravel around the CCC, we explicitly introduce the co-
ordinates of CCC
into the term
. In our previous studies
(Doebeli & Ispolatov, 2014, 2017; Ispolatov, Madhok, Madhok, &
Doebeli, 2016), the CCC was fixed and positioned at zero, so the first
term did not include its coordinates.
The second term in the exponent is the usual Gaussian compe-
tition kernel with width
, reflecting the fact that species that are
closer phenotypically compete more strongly with each other than
species that are far ther apart in phenotype space. In our simulations,
we used
to ensure that the system is able to diver-
sify from the initial state of one species to a communit y of coexisting
phenotypes (Doebeli & Ispolatov, 2017; Ispolatov et al., 2016). Also
as in Doebeli and Ispolatov (2017), the coefficients bij of the nonsym-
metric part of the competition kernel were chosen randomly from
a Gaussian distribution with width 1 and zero mean (see the end of
this section for how this was implemented to obtain the simulation
Assuming that a community comprises m species with pheno-
in D-dimensional phenotype space, where
r=1, ,m
is the species index, the ecological dynamics of the den-
sity Nr of species r is given by the logistic equation
As a result of the logistic dynamics with constant external con-
ditions, the population of each species converges to its equilibrium
. In the following, we call the state of the system, where all
species have reached their
, as the ecological equilibrium.
In the framework of adaptive dynamics (see, for example,
Diekmann, 2002; Geritz, Mesze, & Metz, 1998), evolution occurs
when species, each assumed to be monomorphic in its phenot ype
and at the ecological equilibrium, constantly generate initially rare
mutants with uniformly random phenotypes that are close to but
distinc t from the parental phenotype. The derivation of adaptive
dynamic s is based on the separation of much faster ecological and
normally slower evolutionary timesc ales, which normally holds very
well. Mutants compete with the resident community for resources
and try to invade it with a per capita growth rate defined by Equation
(3), where self-competition is neglec ted because of mutant's rarity,
is the phenotype of a mutant occurring in species r. The
function defined in Equation (4) is known as invasion fitness. If the
invasion fitness is positive, the mutant population will grow in the
environment set by the resident community. If the invasion fitness
is negative, the mutant population goes ex tinct. The selection gra-
with components
points in the direction of mutant with
b𝑖𝑗 (xiyi)(xjx𝑐𝑗)
the highest growth rate. It is obtained by differentiating the invasion
fitness with respect to the mutant phenotype and evaluating the de-
rivative at the resident phenotype.
In general, the adaptive dynamics of phenotypes of a species is de-
termined by the product of its selection gradient (5), which quantifies
the selection pressure, and its mutational variance–covariance matrix,
which describes the rate and size of mutations occurring in each spe-
cies and their effects on the phenot ypes. For simplicity, we assume
there is no mutational covariance and that all traits in all species have
the same mutation rate and mutational variance. To satisfy the last
assumption, we implicitly rescale each direction in phenotype space
such that all traits evolve at a universal rate equal to the population
size times the selection gradient. Then, the adaptive dynamics of each
phenotypic component
i=1, ,D
r=1, ,m
is then given by
Further details of the model, including the procedure allowing
species to diversif y, are presented in the nex t section.
So far, this model has been defined in the same way as the one
in (Doebeli & Ispolatov, 2017; Ispolatov et al., 2016, 2019). Here,
however, we introduce environmental change by assuming that
over time, new phenotypes become optimal for the current state of
the changing environment. The optimal phenotype in our model is
defined by the position of the CCC in the phenotypic space. Thus,
similarly to Johansson (2008), Van Den Elzen et al. (2017), and
Jones, 2008, the environmental changes are be implemented as the
motion of the CCC and the carrying capacity itself in the phenotypic
. Here, the vector
determines the magnitude
and direc tion of change of the CCC. While the maximum of the car-
rying c apacit y function moves in phenotype space at a constant rate
, the general shape of the carr ying capacity function, and, in par-
ticular, its width
, stay the same in the moving frame of reference
in phenot ype space. In the simulations, an environmental change
starts once an evolving community has reached a stationary state in
the evolutionary dynamics with constant environment (Figure 2), as
described in Doebeli and Ispolatov (2017) and Ispolatov et al. (2019).
In the following, we denote this time as t*.
2.2 | Simulation procedure
For every dimension
D=1, 2, 3
, we prepare 30 replicate simulations,
each with a distinct set of coefficient s bij and initial conditions,
randomly chosen from a D-dimensional Gaussian distribution with
width 1 and mean 0. Every replica then evolves with nonmoving CCC
positioned at zero for time t* to converge to its stationary states. The
evolutionary equilibration time t* was determined empirically and
was found to noticeably increase with the dimension of phenotypic
space, Figure 1. Thus, in the second stage of simulation, we “reset
the clock ” and consider the beginning of environment al change as
the new initial time t = 0.
A run for each replica consists of many cycles of successive
steps (Figure 2), where each cycle increments the evolutionar y time
by a small amount. An iteration starts with the ecological dynam-
ics, where all species reach their ecological equilibrium according to
the logistic dynamics (Equation 3). Evolutionary time stays constant
during th is step. If a species cro sses the low populat ion limit set equal
, it is assumed to be extinct and is dropped from the system.
In the next step, the phenotypes of all species evolve according to
the adaptive dynamics specified in Equation (6). Phenotypic changes
in a single evolutionar y time step
are small enough to keep
populations close to their ecological equilibrium (calculated in the
previous step).
To model diversification, each 10 time units we split a randomly
chosen species in halves separated by a very small distance
). The direction of splitting is chosen randomly from
the isotropic distribution. If conditions are favorable for evolutionary
branching, the distance between the halves grows as a result of phe-
notypic dynamics, and the two “halves” become t wo separate spe-
cies. Otherwise, that is, if the competitive interactions do not favor
diversification and the halves do not move apart phenotypically, we
𝑟𝑖 =
dt =N
rS𝑟𝑖 =N
b𝑖𝑗 (x𝑠𝑗 x𝑐𝑗)(x𝑠𝑖 x𝑟𝑖 )
(x𝑟𝑖 x𝑐𝑖)3
,r=1, ,m;i=1, ,D
FIGURE 1 The evolutionary time t* required for diversity
saturation starting from the initial condition of one randomly
located species. Around the time t*, the number of species m
Dimension of phenotypic space, D
Time of saturation of di
versity, t
merge them back right before the next round of splitting without any
consequences to system behavior.
Both procedures of merging and splitting happen regularly every
few iterations (in this order, so that split halves have time to diverge).
In principle, if any two “not closely related” species come close in
phenot ypic distance at th e time of merging, they would be merged as
well, however, we have not observed such events in our simulations.
In the principal par t of our simulations, we analyzed how sat-
urated systems adapt to environmental changes. For each steady
state replica, a new simulation run is starting with that replica as
the initial condition and under the changing environment with a
given rate VC and a random direction. The rates VC are selected to
cover the range between 0.05 and 2 with the step 0.05. Before the
beginnin g of environmental ch anges, we merge all species, se parated
by phenotypic distance marginally larger than the merging distance
used in the diversification procedure, into distinct species,
visible as circles in Figure 3 and corresponding videos. This is done
to ensure that the population density Nr, which controls the adaptive
dynamic s evolutionary speed in Equation (6), corresponds to the in-
tegral population of a species, rather than to the meta-populations
of many split “halves,” created by our diversification procedure. The
biological motivation behind this final merge is that each of those
individual phenotypically close “halves” can produce a mutant that
could take over the whole species. Using the usually smaller indi-
vidual populations of “halves” as the factor Nr in Equation (6) would
have reduced the evolutionary speed and resulted in more difficult
FIGURE 2 Successive steps that are
iterated in the simulations. Each iterative
cycle advances the evolutionary time by a
small increment. The merging and splitting
steps are performed once every 10 time
units during the evolutionar y saturation
of the system. Environmental change
(movement of CCC) start s only after the
system converges to a steady state with
saturated diversity and the formation of
new species ends
FIGURE 3 Snapshot of the saturated
diversification at the beginning of
environmental change t = 0 (a) and at
t = 32 after the initial stage of adaptation
to environmental changes of various rates
(b–d) in two-dimensional phenotypic
; The adapt ation
processes that led to these configurations
can be seen in corresponding videos
are.12827054. In all three cases (b–d), the
vector of environmental changes has the
same direction to the upper left corner
of the frame and is indicated by the red
arrow. Dark gray circles show the location
of different species in phenot ype space,
with the size of the circles representing
the populations size, the red rhombus
shows the location of the CCC. The
coefficients bij and the initial conditions
can be found in Appendix S1: Section 2
adaptation and earlier extinction. For the same reason, star ting from
the onset of environmental change, we switch of f the procedure of
formation of new species. In general, we do not expect the diversi-
fication of species under the pressure of environmental change. We
further comment on this assumption in the Section 4.
Once started, the environmental change continues at a constant
rate VC for the time t* or until all species become extinct. If a sys-
tem survives the ef fects of CCC motion for the time t*, it usually
means that it reaches a new evolutionary steady state adapted to
the const ant environmental change. The results for each D and VC
are averaged over survived species communities from those 30 runs,
producing statistical data shown in Figure 4 for the final state and in
Figure 5 for any arbitrary time t.
The range of relevant CCC speed VC can be capped using the
following simple analytical estimate for the maximum evolutionary
speed that a single species could sustain (Johansson, 2008). The
case of a single surviving species is often the final outcome of ad-
aptation to sufficiently rapid environmental changes. If we ignore
the asymmetric part of the competition kernel represented by the
random coefficients bij in Equation (6), which could either reduce or
increase the strength of competition in a generally unpredictable
way, the remaining adaptive dynamics becomes quite simple. The
evolutionary speed
has components
Here, we have taken into account that the ecologically equili-
brated single-species population is equal to the corresponding car-
rying c apacit y. Assuming for simplicity that the CCC moves along
the first phenot ypic coordinate, we look for the maximum of u1,
differentiating (7) with respect to
. The maximum is achieved
, and the corresponding maximum speed of species
motion in phenotypic space is
umax =
. This sets the
upper limit on the sustainable velocit y of CCC, which results from
a combination between two trends: A faster motion of CCC makes
the species trail further behind in phenotype space, thus generat-
ing a larger selection gradient. However, the further a species trails
behind the CCC, the lower is its population, which makes mutations
more rare. The combination of these two trends defines the maxi-
mum velocity at which the single species can evolve, which, in other
words, is the maximum velocity of CCC that a species can follow at a
FIGURE 4 The number of species m (a), the fraction of species that sur vived till t*
(b), the total community population
the average per species population Nav (d), the phenotypic variation across the community
(e), and the time to total extinction
(f) versus the speed of environmental change VC. For each value of VC, quantities in a–e were averaged only over systems that survived
under the changing environment till the final time t*. The communities that underwent extinction before t* were completely excluded
from the average. The extinction time in (f) was measured only in communities that went extinc t before t*. Colors indicate the dimensions
of phenotype space: blue for D = 1, red for D = 2, and green for D = 3. Values of t* for each dimensionality of the phenot ype space are
presented in Figure 1. To reveal the power-law-like nature of many dependencies, all figures are presented in log–log scale with shadows
around lines indicating standard deviations. Dotted lines represent extinction threshold
for each dimensionality, respectively
FIGURE 5 Dynamics of average number of species m (row a), total population
(row b), average population of a species
(row c), and
phenotypic variation across the community
(row d) as a function of time after the onset of environmental change for 4 different values of
For each value of
, the quantities in a–d were averaged only over systems that survived under the changing environment till the final time t*. The
communities that underwent extinction before t* were completely excluded from the average. All plots are presented in semilogarithmic scale
steady state. In reality, due to the action of the asymmetric terms in
the competition kernel, and due to interspecies competition, some
species go extinct below this maximum CCC velocity, while others
persist even in slightly faster changing environments. The latter hap-
pens when a particular combination of randomly chosen bij coeffi-
cients results in a stronger selection gradient in the direction of the
2.3 | Measuring properties of the system
To analyze the response of evolving communities to environmental
change, in each run we measured the following system properties:
1. number of species in the system m;
2. total population of the system
3. average population of a species
av =
4. phenotypic diversity of the system
, defined as the average
square distance of the phenot ypic coordinates of all species
weighted by their population sizes around the center of mass of
the system
, where
The quantity
reflects how widely the phenotypes of the var-
ious species are separated from the center of mass of the system. It
is a measure of phenotypic diversity in a community, but should not
be confused with the number of species, since the phenotype distri-
bution in systems with smaller numbers of species can nevertheless
have a higher variance if the fewer species are more spread out in
phenotype space;
5. fraction of survived species
𝜌surv =mmsat
(10), where
the number of species in a replica before the initiation of
environmental change;
6. time to extinction
defined for systems that have not persisted
till t* as the time when the last species dies out.
There are t wo basic ways in which the above quantities can be
calculated to illustrate system behavior. First, they can be evaluated
at the final time t* after the onset of environmental change for many
different systems with the same control parameters (e.g., the same
VC). For example, we calculate the average number of coexisting spe-
cies at time t*, which normally corresponds to the new evolutionary
steady state adapted to the environmental change, by averaging m
at t* for many different systems. Second, these quantities can be
studied as a function of time in any given simulation run, usually
starting from the initiation of the movement of CCC, that is, from the
beginning of the environmental change. For example, before starting
the movement of CCC,
, that is, the number of species at sat-
uration. Once CCC starts to move, m usually change, reflecting the
effect of environmental change on the previously saturated commu-
nity. The distribution of properties of prepared saturated systems is
summarized in Figure S1.
We analyzed the adaptation of 30 saturated systems, properties
of which were analyzed previously (Doebeli & Ispolatov, 2017;
Ispolatov et al., 2019), to environmental changes of various rates.
For all those systems, the environmental change in the form of a
moving CCC either forces the system to adapt and converge to a
new quasi-stationary state, or it leads to extinction of the whole
3.1 | Adaptation to the CCC motion and
convergence to a new quasi-stationary state
When the rate of environmental change is not too high, after a tran-
sitory adaptation the system usually converges to a new quasi-sta-
tionary state that follows the changing environment. Properties of
such a quasi-stationary state vary depending on the rate of environ-
mental changes. In Figure 3 there is a two-dimensional visualization
of newly formed stationary configurations of the same community,
adapting to different rates of VC. For this particular example in
Figure 3, we used one direction of environmental change for all pre-
sented rates of VC to make visual comparison easier, in other simula-
tions the direction was chosen randomly each time.
The final number of species m, the fraction of surviving spe-
and the total population size
(Figure 4a,c) generally
decrease with increasing speed of environmental changes, VC. For
a given value of VC, the number of species m and total population
begin to decrease shortly af ter the onset of environmental
change and af ter some time stabilize (despite the ongoing CCC
movement), as illustrated in Figure 5a,b. The higher the VC is, the
faster and larger this decrease occurs. Conversely, the average per
species population
in quasi-stationar y stage increases with
increasing VC: as more species go extinct during the process of
adaptation, the surviving ones become more ecologically success-
ful due to experiencing less competition (Figures 4d and 5c). For
large VC, we also observed scenarios, when during the adaptation
first rapidly increased because of release from the competition
and then fell down to a plateau due to increased distance between
phenotypes of sur viving species and CCC. Such a pattern can be
seen on cur ves of
, which correspond to
, and
(Figure 5c).
tot =
(x𝑟𝑖 x𝑚𝑐
The average value of phenotypic diversit y,
decreases for
larger VC as well (Figure 4e), since larger rates of environmental
change cause the loss of more species. Over the course of an indi-
vidual simulation run, after the onset of environmental change
first decreases with time and then approaches a stationary state. For
very small values of VC, phenotypic dispersion may stay at the level
of saturated system without environmental change or even exceed
it despite a reduction in the number of species m (Figure 5d). Such
behavior of
indicates that phenotypic spread of the surviving
species around the center of mass becomes wider.
Typically, the time required to reach a new evolutionarily sta-
tionary state is small relative to the saturation time t*. The higher
the rate of environment al change VC is, the less time is required.
The timescale of Figure 5 is two times shor ter than the timescale
required for initial equilibration (Figure 1).
Even for small values of VC, the entire evolving community can go
extinct. Naturally, such event s become more likely for larger rates of
environmental changes VC.
In our simulations in any dimensionality of the phenotypic space,
there is a value of VC, denoted by dotted lines on Figure 4a,e, for
which none of our replicas survived. We call this value the extinc-
tion threshold,
. The fraction of sur viving communities as a
function of VC and the extinction threshold value of VC are shown
in Figure S2b of the Appendix S1. Generally, the higher the rate of
environmental change, the more likely extinction occurs. Obviously,
the value
may vary depending on the number of replicates and
chosen t*; however, we still can compare values
between pheno-
type spaces of different dimensionalities, since they were obtained
using the same conditions.
Times to ex tinction text generally become shorter for larger values
of VC (Figure 4f). It varies largely among replicates and is affected by
particular properties of the system, such as bij coefficients or the di-
rection of VC vector relative to the system's phenotypic configuration.
3.2 | The influence of phenotypic complexity on
adaptation and extinction
Having more phenotypic dimensions complicates the adaptation to
environmental changes. The surviving species in low-dimensional
phenotype space have larger populations Nav (Figure 4d), and even
the overall population of the community Ntot of a low-dimensional
system is larger than that of a higher-dimensional one for high rates
of VC (Figure 4c). Furthermore, the time required for the communit y
to reach the quasi-stationar y state becomes longer for higher phe-
notypic complexity: the parameters of systems that survive the envi-
ronmental change usually reach the new steady state plateau faster
in low-dimensional systems than in high-dimensional ones (Figure 5).
However, on average, the communities with more “simple” pheno-
types go extinc t faster when subjected to changing environmental
conditions, and we observe this effect for small and for large values
of VC (Figure 4f).
Moreover, the extinction threshold becomes lower for increas-
ing phenotypic complexity, which means that low-dimensional com-
munities are able to withstand rates of environmental changes that
would lead to extinc tion of the whole community in high-dimen-
sional phenotype spaces (Appendix S1: Figure S2b). A mechanistic
explanation for the observed reduction of the ex tinction threshold
is that in higher dimensions, the population size Nr of each species in
the community is generally smaller. In higher-dimensional systems,
each species has on average more competitors due to larger number
of “nearest neighbors” with slightly different phenotypes. Smaller
population sizes mean lower per species mutation rates, and hence
slower adaptation to changing environments (Equation 6).
To check the robus tness of these results, we have repeated these
simulations for fewer than 30 replicas for two other values of the
width of the competition kernel,
. Even though
the saturated level of diversity varies strongly with
(Doebeli &
Ispolatov, 2017), the trends shown in Figures 4 and 5 remain quali-
tatively unchanged.
3.3 | Comparison to individual-based simulations
To verify the robustness of our Adaptive Dynamics procedure, we,
following advices from Reviewers, performed an Individual-based
simulation of the same evolving system subject to environmental
change. Because individual-based simulations are intrinsically slower
than the adaptive dynamics model and repeating the statistical anal-
ysis is well beyond our computational capacity, we limited our scope
to the single example shown in Figure 3. Specifically, we considered
an ensemb le of individuals w ith birth rate eq ual to one and death r ate
given by the logistic competition term in Equation (3). The carrying
capacity was multiplied by the factor
, which set the scale
for the total number of individuals in the community. Bir th and death
events were executed via the Gillespie algorithm, each offspring was
offset by a randomly distributed mutation sampled from a uniform
distribution with the standard deviation
. The magnitude
was rescaled to take into account the actual coefficient
that should have been present in Equation (6) if it were derived
from the corresponding individual-based process. Thus, to make
individual-based simulations similar to the adaptive dynamics model
, the individual-based CCC speed becomes
In Figure 6, we show the snapshots of distributions of individu-
als separately (panels a and b) and clustered into species (panels c
and d) immediately after equilibration (panels a and c) and after a
transitory period (roughly corresponding to that in Figure 3) after
the onset of environmental changes (panels b and d). A comparison
between Figure 3 (panels a and c) and 6 suppor ts the conclusion
that our adaptive dynamics model is satisfactory reproducing the
predictions of the “first principle” individual-based simulations. We
further comment on the correspondence between these two meth-
ods and the intrinsic limitations of our adaptive dynamics scheme in
the Section 4.
We have investigated the evolutionar y dynamics in logistic com-
petition models under the continual environmental change, which
was implemented by assuming that the optimal phenotype, defined
by the maximum of the carrying capacity func tion moves at a con-
stant speed in phenotype space. We have analyzed the effect of
such environmental change on various statistical ecological and
evolutionary properties of the adapting communities, depending
on the rate of environmental change and on the dimension of phe-
notype space.
Our model is based on several assumptions, which limit us in our
understanding of the full picture of adaptive evolution. For example,
we assume that there is no mutational covariance and that all traits
in all species have the same mutation rate. We do not distinguish
phenotypic plasticity and genetic evolution, which are both import-
ant for evolution in changing environment (Ho & Zhang, 2018). In our
simulations, interaction of species is limited by the competition and
we do not consider numerous factors, which play an important role
in evolutionary processes, such as population structure and genetic
drift (Orr, 2005). We also make assumptions, which seem intrinsic to
our adaptive dynamics simulation scheme, that populations of spe-
cies are in their ecological equilibrium and do not diversify under
the pressure of environmental changes. However, in our individu-
al-based simulations we observed a few times that such diversifica-
tion under environmental change does happen, at least transitory.
An example is shown in the videos associated with Figure 6.
These assumptions are partially justified by the fact that our
principal goal is to consider the behavior of species communities
on large macroevolutionary timescales. On such timescales, spe-
cies undergo major phenotypic changes, which exceed the scale of
phenotypic plasticity and require much more time than equilibration
of species populations. A few transitory diversification events af-
fect neither the long-term evolution dynamics, nor the final steady
state distribution of surviving species. Another justification is that
we aimed to investigate global effects of environmental changes on
species communities. For this goal, the implicitly performed by us
rescaling of phenot ypic space in a way that all traits evolve at a rate
equal to the population size times the selec tion gradient appears to
be safe and not to introduce any noticeable artifact s.
Another limitation, also intrinsic to the standard adaptive dy-
namics protocols, is the assumption of continuous evolution of phe-
notypes instead of discrete jumps caused by individual mutations.
Naturally, since real mutations occur stochastically in time and carry
phenotypic effects of various sizes, the adaptive dynamics methods
are not well-suited to model evolution on fine temporal and pheno-
typic scales. Yet the temporal and phenot ypical stochasticity of mu-
tations may also affect even the large-scale evolutionary responses.
In (Matuszewski, Hermisson, & Kopp, 2014) it was shown that an in-
terplay between the decrease in the fraction of mutations that go in
the direc tion “right” for adaptation with the increase of dimension-
ality D, and the increase in the “usefulness” of mutations with larger
phenotypic effect (which make bigger steps to catch the CCC which
moved farther because of the rarity of useful mutations) results in
FIGURE 6 Snapshot of the saturated
diversification (a, c) and adaptation to
environmental changes (b, d) in individual-
based simulation. The processes that led
to these configurations can be seen in
corresponding videos here https://doi.
org/10.6084/m9.figsh are.12827054. The
vector of environmental changes has the
direction to the upper left corner of the
frame and is indicated by the red arrow.
In a and b, blue circles show individuals
in phenot ype space, in c and d, blue
circles show clustered populations of
phenotypically close individuals, and with
the size of the circles representing the
number of individuals, the red rhombus
shows the location of the CCC. The
details of simulations are presented in the
text. The coefficients bij and the initial
conditions can be found in Appendix S1:
Section 2
lesser adaptability of more complex phenotypes to environmental
changes. This effect is complementary to our model and observa-
tions, yet works in the same directions.
In our models, the crucial factor determining whether an
evolving community will survive in the long run is the rate of
environmental change. We found that in each dimensionality
of phenotype space there is a threshold rate of environmental
change above which none of our replicate communities sur vived.
However, when the rate of change is below the threshold value,
many communities are able to adapt to environmental changes,
and the evolving community can find a new quasi-stationary state.
Such adaptation requires much less time compared with the time
it takes a community to reach the prechange level of diversity from
a single ancestral species. To adapt, the phenotype of a species
generally has to evolve in step with the movement of the center
of the carrying capacit y. Species that fail to track the carrying ca-
pacity and trail too far behind, where their carrying capacity falls
significantly, suffer big population reduction and produce too few
mutants to keep up with the changing environment and eventually
go extinct. Similarly, species that trail too far behind their initial
intraspecies phenotypic position, may experience too strong com-
petition, which also results in the similar reduction of population,
inabilit y to produce enough mutants, and subsequent extinction.
These effects were studied in great detail in simpler one-dimen-
sional models with 1–3 species (Johansson, 2008; Van Den Elzen
et al., 2017). Our studies thus confirm that such evolutionary pat-
terns play the essential role in the response of higher-dimensional
and higher-diversity systems to environmental change as well.
The effect of reduction in the number of species, pronounced in
our simulation, was also observed in Johansson (2008) and Van
Den Elzen et al. (2017), albeit on a lesser number (2–3) of initial
species. These and other similarities between our results and
those of Johansson (2008) and Van Den Elzen et al. (2017) sig-
nify broad universality of the obser ved dynamical evolutionary
pattern given the differences in methodologies: We used adaptive
dynamics with clonal reproduction and constant carrying capac-
ity amplitude
, rather than individual-based simulation with
sexual reproduction and adjustable K0 used by Johansson (2008).
The functional forms of the competition function and carrying ca-
pacity were different as well. Nevertheless, even though one-di-
mensional systems were also included in our study, because of the
difference in methodologies it is hard to go beyond a simple qual-
itative comparison between the presented and published results.
The rate of environmental change also affects the composition
of newly formed quasi-stationary states. The higher the rate of en-
vironmental change is, the lower is the average number of coexist-
ing species and the tot al population size of sur viving communities.
However, surviving species get an ecological advantage: because
of the reduction in the number of competitors, the populations of
each surviving species can become larger than in the case of sta-
ble environmental conditions. Thus, generally the fewer species are
left in the surviving community, the larger their population size be-
comes. This results are consistent with observations from studies of
experimental evolution. In famous experiment with MEGA-plate, the
bacterial clones, which were able to adapt to higher concentration of
antibiotics, became dominant in population, while their fitness, mea-
sured as the growth rate, was much smaller than the average fitness
of the initial community (Baym et al., 2016).
Interestingly, phenotypic complexity makes communities less re-
silient to changes in the environment. In high-dimensional phenotypic
spaces, surviving systems require more time to adapt, and hence are
less able to resist changes. Hence, the extinction threshold, below
which no systems sur vive, appears at lower rates of environmen-
tal change. According to our previous studies of macroevolutionary
processes (Doebeli & Ispolatov, 2014, 2017; Ispolatov et al., 2019),
“complex phenotypes” have smaller populations and lower rates of
evolution, which makes adaptation challenging. In these models, evo-
lution under constant environmental conditions results in gradual ex-
pansion of phenotypic space, and more complex, higher-dimensional
phenotypes evolve only once diversity in lower dimensions has satu-
rated (Ispolatov et al., 2019). Conversely, changing environments tend
to reduce the average number of phenotypic dimensions in biological
systems, since less complex species are more likely to survive.
The evolving communities that have become less diverse due to
the environmental change will rediversify and reach previous satura-
tion levels if the environment ceases to change. This happens via the
same scenario as the initial diversification, illustrated, for example,
in Figure 3 and corresponding videos. However, despite the similar-
ity bet ween the new phenot ypic composition and the one existing
before the onset of environmental change, the genealogical history
and composition of the rediversified community may be very differ-
ent from the composition of the community that existed before the
environmental change was initiated.
Thus, such periods of reduction of diversity caused by envi-
ronmental changes were probably followed by periods of resat-
uration and resulted in evolution of entirely novel phenotypes. In
terms of our model of evolution of complex phenot ypes (Doebeli &
Ispolatov, 2017; Ispolatov e t al., 2019), where species have a c hoice to
diversif y in the same phenotypic space or inhabit a new phenotypic
dimension, such environmental changes would then likely generate
new bouts of rapid diversification leading to saturated communities
occupying higher-dimensional phenotype spaces. We plan to include
such dramatic evolutionary events in our future models of evolution
of complex phenotypes (Ispolatov et al., 2019).
I.I. acknowledges support from DICYT project 0 41931Y. E.A. ac-
knowledges support from SkolTech Academic Mobility Program.
M.D. was supported by NSERC, Canada.
None declared.
Evgeniia Alekseeva: Conceptualization (supporting); Data cura-
tion (lead); Formal analysis (lead); Funding acquisition (equal);
Investigation (lead); Methodology (equal); Project administra-
tion (equal); Resources (equal); Sof tware (lead); Supervision
(supporting); Validation (equal); Visualization (lead); Writing-
original draft (lead); Writing-review & editing (lead). Michael
Doebeli: Conceptualization (equal); Data curation (supporting);
Formal analysis (equal); Funding acquisition (equal); Investigation
(equal); Methodology (equal); Project administration (equal);
Resources (equal); Software (supporting); Supervision (support-
ing); Validation (equal); Visualization (supporting); Writing-original
draft (equal); Writing-review & editing (equal). Iaroslav Ispolatov:
Conceptualization (equal); Data curation (supporting); Formal
analysis (equal); Funding acquisition (equal); Investigation (equal);
Methodology (equal); Project administration (equal); Resources
(equal); Software (equal); Supervision (equal); Validation (equal);
Visualization (supporting); Writing-original draft (equal); Writing-
review & editing (equal).
Codes and averaged results of simulations are available at https:// iiaAl eksee va/Climate.
Iaroslav Ispolatov
Baker, A. C. (2003). Flexibility and specificity in coral-algal symbio-
sis: Diversity, ecology, and biogeography of Symbiodinium. Annual
Review of Ecology, Evolution, and Systematics, 34(1), 661–689.
Baker, A. C., Starger, C. J., McClanahan, T. R ., & Glynn, P. W. (2004).
Coral reefs: Corals’ adaptive response to climate change. Nature,
430(70 01 ), 741.
Bay, R. A., Rose, N. H., Logan, C . A., & Palumbi, S. R. (2017). Genomic
models predict successful coral adaptation if future ocean warming
rates are reduced. Science Advances, 3(11), e1701413.
Baym, M., Lieberman, T. D., Kelsic, E. D., Chait , R., Gross, R., Yelin, I., &
Kishony, R. (2016). Spatiotemporal microbial evolution on antibiotic
landscapes. Science, 353(63 04), 1147–1151. ht tps: //doi. org/10 .1126/
scien ce.aag0822
Blowes, S . A., Supp, S. R., Antão, L. H., Bates, A., Bruelheide, H., Chase,
J. M., … Dornelas, M. (2019). The geography of biodiversity change
in marine and terrestrial assemblages. Science, 366(6463), 339–345. ce.aaw1620
Botero, C. A., Weissing, F. J., Wright, J., & Rubenstein, D. R. (2015).
Evolutionary tipping points in the c apacit y to adapt to environmen-
tal change. Proceedings of the National Academy of Sciences of the
United States of America, 112(1), 184–189.
pna s.14 085 89111
De Mazancourt, C., Johnson, E., & Barraclough, T. G. (2008).
Biodiversity inhibits species’ evolutionary responses to chang-
ing environments. Ecology Letters, 11(4), 38 0–388. https://doi.
org /10.1111/j.1461-0 248 .2 00 8.01152.x
Diekmann, O. (2002). A beginners guide to adaptive dynamics. In
Summer school on mathematical biology, Banach Center Publications
63, 63–100. Warsaw, Poland: Institute of Mathematics Polish
Academy of Sciences. https://dspace.libra ream/handl
e/1874/10648/ Diekm an_04_A-BEGIN NER'S-GUIDE -TO-ADAPT
IVE-DYNAM ICS.pdf?seque nce=2
Doebeli, M., & Ispolatov, I. (2014). Chaos and unpredict abilit y in evolu-
tion. Evolution, 68(5), 1365–1373.
Doebeli, M., & Ispolatov, I. (2017). Diversit y and coevolutionary dynam-
ics in high-dimensional phenotype spaces. The American Naturalist,
189(2), 105–120. 86/689891
Geritz, S. A. H., Kisdi, É., Mesze, G., & Metz, J. A. J. (1998). Evolutionarily
singular strategies and the adaptive growth and branching of the
evolutionary tree. Evolutionary Ecology, 12(1), 35–57. https://doi.
org /10.1023/A:10065 549066 81
Griffen, B. D. (2018). Modeling the metabolic costs of swimming in polar
bears (Ursus maritimus). Polar Biology, 41(3), 491–503. https://doi.
org/10.1007/s0030 0-017-2209-x
Grottoli, A. G., Martins, P. D., Wilkins, M. J., Johnston, M. D., Warner, M.
E., Cai, W. J., … Schoepf, V. (2018). Coral physiology and microbiome
dynamics under combined warming and ocean acidification. PLoS
One, 13(1), e0191156.
Gynther, I., Waller, N., & Leung, L. K. P. (2016). Confirmation of the ex-
tinction of the Bramble Cay melomys Melomys rubicola on Bramble Cay,
Torres Strait: Results and conclusions from a comprehensive survey in
August-September 2014. Queensland Government.
Ho, W .- C. , & Zh ang, J. ( 2018 ). Ev o lution ar y adapt at ions to n ew en vironme nt s
generally reverse plastic phenotypic changes. Nature Communications,
9(1), 1–11. https :// /10.10 38/s 4146 7-0 17-02724 -5
Ispolatov, I., Alekseeva, E., & Doebeli, M. (2019). Competition-driven
evolution of organismal complexity. PLoS Computational Biology,
15(10), e10 07388.
Ispolatov, I., Madhok , V., & Doebeli, M. (2016). Individual-based models
for adaptive diversification in high- dimensional phenotype spaces.
Journal of Theoretical Biology, 390, 97105.
Johansson, J. (2008). Evolutionary responses to environmental changes:
How does competition affect adapt ation? Evolution, 62(2 ), 421– 435 .
Jones, A. G. (2008). A theoretical quantitative genetic study of negative
ecological interactions and extinction times in changing environ-
ments. BMC Evolutionary Biology, 8(1), 119.
Laska r, A. H., M ahata, S., & Li ang, M.-C. (2 016). Identifica tion of anthrop o-
genic CO2 using triple oxygen and clumped isotopes . Environmental
Science & Technology, 50(21), 11806–11814.
Little, A. F., Van Oppen, M. J. H., & Willis, B. L. (2004). Flexibility in algal
endosymbioses shapes growth in reef corals. Science, 304(5 676) ,
14921494. ce.1095733
Lone, K., Kovacs, K. M., Lydersen, C., Fedak, M., Andersen, M ., Lovell, P.,
& Aars, J. (2018). Aquatic behavior of polar bears (Ursus maritimus) in
an increasingly ice-free Arctic. Scientific Reports, 8( 1), 1–12 .
Maclean, I. M. D. , & Wi lson , R. J. (2 011 ). Recent ecological responses to cli-
mate change suppor t predictions of high extinction risk. Proceedings
of the National Academy of Sciences of the United States of America,
108(30), 12337–12342. ht tps:// .1073/pnas.10173 52108
Matuszewski, S., Hermisson, J., & Kopp, M. (2014). Fisher’s geometric
model with a moving optimum. Evolution, 68(9), 25712588. ht tps://
doi .or g/10.1111/evo.12465
Mellard, J. P., de Mazancourt, C., & Loreau, M. (2015). Evolutionar y re-
sponses to environmental change: Trophic interactions affect adap-
tation and persistence. Proceedings of the Royal Society B: Biological
Sciences, 282(1 805 ), 20 1413 51.
Norberg, J., Urban, M. C., Vellend, M., Klausmeier, C. A., & Loeuille,
N. (2012). Eco-evolutionary responses of biodiversity to climate
change. Nature Climate Change, 2(10), 747.
Northfield, T. D., & Ives, A. R. (2013). Coevolution and the ef-
fects of climate change on interacting species. PLoS Biology, 11 (10),
e10016 85.
Obbard, M. E., Cattet, M. R . L., Howe, E. J., Middel, K . R., Newton, E. J.,
Kolenosky, G. B., … Greenwood, C. J. (2016). Trends in body condition
in polar bears (Ursus maritimus) from the Southern Hudson Bay sub-
population in relation to changes in sea ice. Arctic Science, 2(1), 15–32.
Orr, H. A. (2 005). The genet ic theory of ad aptation: A bri ef history. Nature
Reviews Genetics, 6(2), 119–127. https :// rg1523
Osmond, M. M., Ot to, S. P., & Klausmeier, C. A. (2017). When predators
help prey adapt and persist in a changing environment. The American
Naturalist, 190 (1), 8 3–98. https :// /10.10 86/691778
Pagano, A. M., Durner, G. M., Rode, K. D., Atwood, T. C., Atkinson, S.
N., Peacock, E., … Williams, T. M. (2018). High-energy, high-fat life-
style challenges an Arctic apex predator, the polar bear. Science,
359(6375), 568–572. ce.aan8677
Pogoreut z, C., R ädecker, N., Cárdenas, A., Gärdes, A., Wild, C., &
Voolstra , C. R. (2018). Dominance of Endozoicomonas bacteria
throughout coral bleaching and mortality suggests structural inflex-
ibility of the Pocillopora verrucosa microbiome. Ecology and Evolution,
8(4), 2240–2252.
Scheffers, B. R., Oliveira, B. F., Lamb, I., & Edwards, D. P. (2019). Global
wildlife trade across the tree of life. Science, 366 (6461), 71–76. ce.aav5327
Smith, E. G., Hume, B. C. C., Delaney, P., Wiedenmann, J., & Burt, J. A.
(2017). Genetic structure of coral-Symbiodinium symbioses on the
world’s warmest re efs. PLoS One, 12(6), e 0180169.
Tartu, S., Lille-Langøy, R., Størseth, T. R., Bourgeon, S., Brunsvik, A., Aars,
J., … Routti, H. (2017). Multiple-stressor effects in an apex preda-
tor: Combined influence of pollutants and sea ice decline on lipid
metabolism in polar bears. Scientific Reports, 7(1), 1–12. https://doi.
org /10.103 8/s4159 8-017-16820 -5
Trisos, C. H ., Merow, C., & Pigot, A . L. (2020). The p rojected timi ng of abrupt
ecological disruption from climate change. Nature, 580(780 4), 496–501.
Urban, M . C. (2015). Accelerating extinction risk from climate change.
Science, 348(6234), 571–573. ce.aaa4984
Van Den Elzen, C. L ., Kleynhans, E. J., & Otto, S. P. (2017). Asymmetric
competition impacts evolutionary rescue in a changing environment.
Proceedings of the Royal Society B: Biological Sciences, 284(1857),
Whiteman, J. P., Harlow, H. J., Durner, G. M., Regehr, E. V., Rourke, B. C.,
Robles, M., … Ben-David, M. (2017). Polar bears experience skeletal
muscle atrophy in response to food deprivation and reduced activity
in winter and summer. Conservation Physiology, 5(1) , c ox0 49.
Yvon-Durocher, G., Allen, A. P., Cellamare, M., Dossena, M., Gaston, K. J.,
Leitao, M. , … Tri mm er, M. (2015). Five ye ars of exp erimental war ming
increases the biodiversity and productivity of phytoplankton. PLoS
Biology, 13(12), e100 2324.
Additional suppor ting information may be found online in the
Supporting Information section.
How to cite this article: Alekseeva E, Doebeli M, Ispolatov I.
Evolutionary adaptation of high-diversity communities to
changing environments. Ecol Evol. 2020;10:1194111953.
... The composition of microbial communities is sensitive to the environment (Alekseeva et al., 2020;Goldford et al., 2018;, which changes growth of individual species (Bittleston et al., 2020;de Mazancourt et al., 2008) and the interaction with other community members Gibert & Brassil, 2014). Modifications of the environment can affect predator-prey systems (Gilpin, 1972), and a stable predator-prey community may be destabilized due to dwindling densities of a keystone species (Banerjee et al., 2018;Gilljam et al., 2015). ...
Full-text available
Changing environmental conditions can infer structural modifications of predator-prey communities. New conditions often increase mortality which reduces population sizes. Following this, predation pressure may decrease until populations are dense again. Dilution may thus have substantial impact not only on ecological but also on evolutionary dynamics because it amends population densities. Experimental studies, in which microbial populations are maintained by a repeated dilution into fresh conditions after a certain period, are extensively used approaches allowing us to obtain mechanistic insights into fundamental processes. By design, dilution, which depends on transfer volume (modifying mortality) and transfer interval (determining the time of interaction), is an inherent feature of these experiments, but often receives little attention. We further explore previously published data from a live predator-prey (bacteria and ciliates) system which investigated eco-evolutionary principles and apply a mathematical model to predict how various transfer volumes and transfer intervals would affect such an experiment. We find not only the ecological dynamics to be modified by both factors but also the evolutionary rates to be affected. Our work predicts that the evolution of the anti-predator defense in the bacteria, and the evolution of the predation efficiency in the ciliates, both slow down with lower transfer volume, but speed up with longer transfer intervals. Our results provide testable hypotheses for future studies of predator-prey systems, and we hope this work will help improve our understanding of how ecological and evolutionary processes together shape composition of microbial communities.
Full-text available
As anthropogenic climate change continues the risks to biodiversity will increase over time, with future projections indicating that a potentially catastrophic loss of global biodiversity is on the horizon1–3. However, our understanding of when and how abruptly this climate-driven disruption of biodiversity will occur is limited because biodiversity forecasts typically focus on individual snapshots of the future. Here we use annual projections (from 1850 to 2100) of temperature and precipitation across the ranges of more than 30,000 marine and terrestrial species to estimate the timing of their exposure to potentially dangerous climate conditions. We project that future disruption of ecological assemblages as a result of climate change will be abrupt, because within any given ecological assemblage the exposure of most species to climate conditions beyond their realized niche limits occurs almost simultaneously. Under a high-emissions scenario (representative concentration pathway (RCP) 8.5), such abrupt exposure events begin before 2030 in tropical oceans and spread to tropical forests and higher latitudes by 2050. If global warming is kept below 2 °C, less than 2% of assemblages globally are projected to undergo abrupt exposure events of more than 20% of their constituent species; however, the risk accelerates with the magnitude of warming, threatening 15% of assemblages at 4 °C, with similar levels of risk in protected and unprotected areas. These results highlight the impending risk of sudden and severe biodiversity losses from climate change and provide a framework for predicting both when and where these events may occur. Using annual projections of temperature and precipitation to estimate when species will be exposed to potentially harmful climate conditions reveals that disruption of ecological assemblages as a result of climate change will be abrupt and could start as early as the current decade.
Full-text available
Non-uniform rates of morphological evolution and evolutionary increases in organismal complexity, captured in metaphors like “adaptive zones”, “punctuated equilibrium” and “blunderbuss patterns”, require more elaborate explanations than a simple gradual accumulation of mutations. Here we argue that non-uniform evolutionary increases in phenotypic complexity can be caused by a threshold-like response to growing ecological pressures resulting from evolutionary diversification at a given level of complexity. Acquisition of a new phenotypic feature allows an evolving species to escape this pressure but can typically be expected to carry significant physiological costs. Therefore, the ecological pressure should exceed a certain level to make such an acquisition evolutionarily successful. We present a detailed quantitative description of this process using a microevolutionary competition model as an example. The model exhibits sequential increases in phenotypic complexity driven by diversification at existing levels of complexity and a resulting increase in competitive pressure, which can push an evolving species over the barrier of physiological costs of new phenotypic features.
Full-text available
Polar bears are ice-associated marine mammals that are known to swim and dive, yet their aquatic behaviour is poorly documented. Reductions in Arctic sea ice are clearly a major threat to this species, but understanding polar bears' potential behavioural plasticity with respect to the ongoing changes requires knowledge of their swimming and diving skills. This study quantified time spent in water by adult female polar bears (n = 57) via deployment of various instruments bearing saltwater switches, and in some case pressure sensors (79 deployments, 64.8 bear-years of data). There were marked seasonal patterns in aquatic behaviour, with more time spent in the water during summer, when 75% of the polar bears swam daily (May-July). Females with cubs-of-the-year spent less time in the water than other females from den emergence (April) until mid-summer, consistent with small cubs being vulnerable to hypothermia and drowning. Some bears undertook notable long-distance-swims. Dive depths up to 13.9 m were recorded, with dives ≥5 m being common. The considerable swimming and diving capacities of polar bears might provide them with tools to exploit aquatic environments previously not utilized. This is likely to be increasingly important to the species' survival in an Arctic with little or no persistent sea ice.
Full-text available
Regional declines in polar bear (Ursus maritimus) populations have been attributed to changing sea ice conditions, but with limited information on the causative mechanisms. By simultaneously measuring field metabolic rates, daily activity patterns, body condition, and foraging success of polar bears moving on the spring sea ice, we found that high metabolic rates (1.6 times greater than previously assumed) coupled with low intake of fat-rich marine mammal prey resulted in an energy deficit for more than half of the bears examined. Activity and movement on the sea ice strongly influenced metabolic demands. Consequently, increases in mobility resulting from ongoing and forecasted declines in and fragmentation of sea ice are likely to increase energy demands and may be an important factor explaining observed declines in body condition and survival.
Full-text available
The importance of Symbiodinium algal endosymbionts and a diverse suite of bacteria for coral holobiont health and functioning are widely acknowledged. Yet, we know surprisingly little about microbial community dynamics and the stability of host-microbe associations under adverse environmental conditions. To gain insight into the stability of coral host-microbe associations and holobiont structure, we assessed changes in the community structure of Symbiodinium and bacteria associated with the coral Pocillopora verrucosa under excess organic nutrient conditions. Pocillopora-associated microbial communities were monitored over 14 days in two independent experiments. We assessed the effect of excess dissolved organic nitrogen (DON) and excess dissolved organic carbon (DOC). Exposure to excess nutrients rapidly affected coral health, resulting in two distinct stress phenotypes: coral bleaching under excess DOC and severe tissue sloughing (>90% tissue loss resulting in host mortality) under excess DON. These phenotypes were accompanied by structural changes in the Symbiodinium community. In contrast, the associated bacterial community remained remarkably stable and was dominated by two Endozoicomonas phylotypes, comprising on average 90% of 16S rRNA gene sequences. This dominance of Endozoicomonas even under conditions of coral bleaching and mortality suggests the bacterial community of P. verrucosa may be rather inflexible and thereby unable to respond or acclimatize to rapid changes in the environment, contrary to what was previously observed in other corals. In this light, our results suggest that coral holobionts might occupy structural landscapes ranging from a highly flexible to a rather inflexible composition with consequences for their ability to respond to environmental change.
Full-text available
Organismal adaptation to a new environment may start with plastic phenotypic changes followed by genetic changes, but whether the plastic changes are stepping stones to genetic adaptation is debated. Here we address this question by investigating gene expression and metabolic flux changes in the two-phase adaptation process using transcriptomic data from multiple experimental evolution studies and computational metabolic network analysis, respectively. We discover that genetic changes more frequently reverse than reinforce plastic phenotypic changes in virtually every adaptation. Metabolic network analysis reveals that, even in the presence of plasticity, organismal fitness drops after environmental shifts, but largely recovers through subsequent evolution. Such fitness trajectories explain why plastic phenotypic changes are genetically compensated rather than strengthened. In conclusion, although phenotypic plasticity may serve as an emergency response to a new environment that is necessary for survival, it does not generally facilitate genetic adaptation by bringing the organismal phenotype closer to the new optimum.
Full-text available
Rising seawater temperature and ocean acidification threaten the survival of coral reefs. The relationship between coral physiology and its microbiome may reveal why some corals are more resilient to these global change conditions. Here, we conducted the first experiment to simultaneously investigate changes in the coral microbiome and coral physiology in response to the dual stress of elevated seawater temperature and ocean acidification expected by the end of this century. Two species of corals, Acropora millepora containing the thermally sensitive endosymbiont C21a and Turbinaria reniformis containing the thermally tolerant endosymbiont Symbiodinium trenchi, were exposed to control (26.5°C and pCO2 of 364 μatm) and treatment (29.0°C and pCO2 of 750 μatm) conditions for 24 days, after which we measured the microbial community composition. These microbial findings were interpreted within the context of previously published physiological measurements from the exact same corals in this study (calcification, organic carbon flux, ratio of photosynthesis to respiration, photosystem II maximal efficiency, total lipids, soluble animal protein, soluble animal carbohydrates, soluble algal protein, soluble algal carbohydrate, biomass, endosymbiotic algal density, and chlorophyll a). Overall, dually stressed A. millepora had reduced microbial diversity, experienced large changes in microbial community composition, and experienced dramatic physiological declines in calcification, photosystem II maximal efficiency, and algal carbohydrates. In contrast, the dually stressed coral T. reniformis experienced a stable and more diverse microbiome community with minimal physiological decline, coupled with very high total energy reserves and particulate organic carbon release rates. Thus, the microbiome changed and microbial diversity decreased in the physiologically sensitive coral with the thermally sensitive endosymbiotic algae but not in the physiologically tolerant coral with the thermally tolerant endosymbiont. Our results confirm recent findings that temperature-stress tolerant corals have a more stable microbiome, and demonstrate for the first time that this is also the case under the dual stresses of ocean warming and acidification. We propose that coral with a stable microbiome are also more physiologically resilient and thus more likely to persist in the future, and shape the coral species diversity of future reef ecosystems.
Full-text available
There is growing evidence from experimental and human epidemiological studies that many pollutants can disrupt lipid metabolism. In Arctic wildlife, the occurrence of such compounds could have serious consequences for seasonal feeders. We set out to study whether organohalogenated compounds (OHCs) could cause disruption of energy metabolism in female polar bears (Ursus maritimus) from Svalbard, Norway (n = 112). We analyzed biomarkers of energy metabolism including the abundance profiles of nine lipid-related genes, fatty acid (FA) synthesis and elongation indices in adipose tissue, and concentrations of lipid-related variables in plasma (cholesterol, high-density lipoprotein, triglycerides). Furthermore, the plasma metabolome and lipidome were characterized by low molecular weight metabolites and lipid fingerprinting, respectively. Polychlorinated biphenyls, chlordanes, brominated diphenyl ethers and perfluoroalkyl substances were significantly related to biomarkers involved in lipid accumulation, FA metabolism, insulin utilization, and cholesterol homeostasis. Moreover, the effects of pollutants were measurable at the metabolome and lipidome levels. Our results indicate that several OHCs affect lipid biosynthesis and catabolism in female polar bears. Furthermore, these effects were more pronounced when combined with reduced sea ice extent and thickness, suggesting that climate-driven sea ice decline and OHCs have synergistic negative effects on polar bears.
Human activities are fundamentally altering biodiversity. Projections of declines at the global scale are contrasted by highly variable trends at local scales, suggesting that biodiversity change may be spatially structured. Here, we examined spatial variation in species richness and composition change using more than 50,000 biodiversity time series from 239 studies and found clear geographic variation in biodiversity change. Rapid compositional change is prevalent, with marine biomes exceeding and terrestrial biomes trailing the overall trend. Assemblage richness is not changing on average, although locations exhibiting increasing and decreasing trends of up to about 20% per year were found in some marine studies. At local scales, widespread compositional reorganization is most often decoupled from richness change, and biodiversity change is strongest and most variable in the oceans.
Wildlife trade is a multibillion dollar industry that is driving species toward extinction. Of >31,500 terrestrial bird, mammal, amphibian, and squamate reptile species, ~18% ( N = 5579) are traded globally. Trade is strongly phylogenetically conserved, and the hotspots of this trade are concentrated in the biologically diverse tropics. Using different assessment approaches, we predict that, owing to their phylogenetic replacement and trait similarity to currently traded species, future trade will affect up to 3196 additional species—totaling 8775 species at risk of extinction from trade. Our assessment underscores the need for a strategic plan to combat trade with policies that are proactive rather than reactive, which is especially important because species can quickly transition from being safe to being endangered as humans continue to harvest and trade across the tree of life.