Adapting to Environmental Heterogeneity: Selection and Radiation

Abstract

Environmental heterogeneity is invoked as a key explanatory factor in the adaptive evolution of a surprisingly wide range of phenomena. This article aims to analyze this explanatory scheme of categorizing traits or properties as adaptations to environmental heterogeneity ("heterogeneity adaptations"). First it is suggested that this scheme can be understood as a reaction to how heterogeneity adaptations were discounted or ignored in the modern synthesis. Then a positive account is proposed, distinguishing between two broad categories of adaptation to environmental heterogeneity: properties selected for by well-defined patterns of environmental heterogeneity, and properties that help organisms exploit novel patterns of environmental heterogeneity.

Full text

Vol.:(0123456789)
1 3
Biological Theory
https://doi.org/10.1007/s13752-021-00373-y
THEMATIC ISSUE ARTICLE: CONCEPTUALIZING THEENVIRONMENT INNATURAL
SCIENCES
Adapting toEnvironmental Heterogeneity: Selection andRadiation
HughDesmond1,2
Received: 1 July 2019 / Accepted: 4 December 2020
© Konrad Lorenz Institute for Evolution and Cognition Research 2021
Abstract
Environmental heterogeneity is invoked as a key explanatory factor in the adaptive evolution of a surprisingly wide range
of phenomena. This article aims to analyze this explanatory scheme of categorizing traits or properties as adaptations to
environmental heterogeneity ("heterogeneity adaptations"). First it is suggested that this scheme can be understood as a
reaction to how heterogeneity adaptations were discounted or ignored in the modern synthesis. Then a positive account is
proposed, distinguishing between two broad categories of adaptation to environmental heterogeneity: properties selected
for by well-defined patterns of environmental heterogeneity, and properties that help organisms exploit novel patterns of
environmental heterogeneity.
Keywords Adaptation· Adaptive radiation· Environmental heterogeneity· Environmental novelty· Selective environment
Introduction
Paradigmatic cases of evolution by natural selection are typi-
cally represented as processes of adaptation to fixed environ-
mental states. For instance, natural selection on Kettlewell’s
moths occurred in an environment defined by one of two
environmental states, sooted or non-sooted trees (Kettlewell
1955), and individual moths were assumed to be either adap-
tive to sooted environments, or non-sooted environments,
but not a mixture of both. Alternatively, consider how Dar-
win illustrated the theory of natural selection by describing
how different varieties of wolf might adapt to one of two
environmental states: mountainous habitats or lowland habi-
tats (Darwin 2008, p. 70). Such paradigmatic examples of
adaptation to environments defined by fixed states form the
backbone of contemporary introductory textbooks on the
subject (see, for instance: Ridley 2004, p. 75; Futuyma and
Kirkpatrick 2017, pp. 55–56).
While no biologist would deny that natural environments
are much more complicated than such textbook models,
such paradigmatic models can exert undue influence on
how observations are characterized. For instance, observa-
tions of phenotypic plasticity were for a long time typically
characterized as developmental noise. In 1965 Bradshaw
complained about how phenotypic plasticity was too often
interpreted as a “lack of stability” in development (Brad-
shaw 1965, pp. 115–116). More recently, Pigliucci similarly
complained that plasticity was still too often considered an
empirical “nuisance” and understood to be “simply a fancy
word to indicate the old ‘environment component of the phe-
notype’” (Pigliucci 2010, p. 355).
The work of biologists such as Bradshaw or Pigliucci
serves to underscore a general insight: how environmen-
tal heterogeneity (change over time/space) is not only a
source of evolutionary contingency (compare with Beatty
1995), but can also be a source of selection pressures, for
instance favoring plasticity in a particular trait over more
canalized development. Properties or traits are not necessar-
ily only adaptive to fixed environmental states, but can also
be “heterogeneity adaptations”: adaptive to environmental
heterogeneity.
This article aims to problematize heterogeneity adapta-
tions, for it is less clear than may seem at first what is meant
by categorizing some property or trait to be adaptive to
“environmental heterogeneity” (EH). Some have understood
EH very broadly, especially in the context of the evolution of
cognition: general cognitive properties and even cognition
* Hugh Desmond
hugh.desmond@uantwerpen.be
1 KU Leuven, Leuven, Belgium
2 University ofAntwerp, Leuven, Belgium

H.Desmond
1 3
itself have been hypothesized to be adaptive to EH as such.1
Others, such as Pigliucci, have been at pains to emphasize
how heterogeneity adaptations such as phenotypic plasticity
cannot be adaptive to EH as such, but only to certain spati-
otemporal patterns of EH (Pigliucci 2001).
If one broadens one’s view, literatures on evolutionary
transitions (eukaryotes, multicellularity, sex) or on the evo-
lution of fundamental ecological interactions (metabolic
exchange, motility, predation, cooperation, communica-
tion) seem to often (implicitly) partially explain traits or
properties as adaptations to environments that are termed
“heterogeneous”—or near-synonyms such as “variable,”
“complex,” “structured,” “noisy,” “unpredictable.” Table1
gives a short overview of how a vast array of traits and prop-
erties is hypothesized as adaptive to EH.
The challenge today, unlike for Bradshaw in 1965, is not
to rehabilitate the significance of heterogeneity adaptations,
but rather to gain more conceptual clarity on what hetero-
geneity adaptations should mean. While Table1 should not
be overinterpreted—it says nothing about how widespread
the references to EH are, nor about the context in which it
occurs—it does, at the very least, suggest that explanations
that categorize traits/properties as "heterogeneity adapta-
tions" should receive some more analysis. What does it
mean for a trait or organism to adapt to environmental het-
erogeneity? Hence, one can ask how precisely the following
explanatory scheme should be understood:
The evolution of property P ("heterogeneity adapta-
tion") can be partially explained by P being adaptive
to EH. (CG)
The puzzle here is that, on the one hand, the coarse-
grained scheme (CG) seems to pick out a prima facie plau-
sible way of thinking about adaptations to EH. Yet, on the
other, environmental heterogeneity is a broad, amorphous
category, strictly speaking referring to all natural environ-
ments. Since explanatory schemes often leave out signifi-
cant information (cf. Van Fraassen 1980), the question then
becomes: where does the explanatoriness of (CG) lie?
In bringing attention to heterogeneity adaptations, this
article aims to, on the one hand, situate their role within the
larger development of the extended evolutionary synthesis
(EES), and on the other hand, to offer an analysis of the
explanatory logic of explanations as schematized by (CG).
In particular, a distinction will be made between two types
of adaptation to EH: a process of natural selection in a well-
defined selective environment, and a radiative process where
the adaptative trait allows a novel resource to be exploited.
The article is structured as follows. In the next section,
further motivation and background is given to the cen-
tral question: how precisely are heterogeneity adaptations
hypothesized across literatures, and why should one enquire
as to the logic of the associated explanations? In the third
Table 1 The range of properties (both of individuals and populations) currently hypothesized to be adaptive to environmental heterogeneity*
*Not all explanations are explicitly labeled as “adaptive” by the authors: when authors have claimed that EH (or some related concept) is the
“ecological” or “environmental” condition that explains the evolution of the explanandum, I have interpreted such explanations as adaptive
explanations
Trait/Property Adaptive to
Homeostasis “Changing internal or external environment” (Cohen etal. 2012); noisy environments (Woods and Wilson 2013)
Motility Environments that are “physically structured in such a way that nutrients are not equally available to all cells” (Wei
etal. 2011, p. 4047); “heterogeneous environments” (Fenchel 2002)
Intraspecific predation “Environments characterized by large fluctuations in food resources,” boom and bust dynamics with abundance and
depletion (Polis 1981, p. 234)
Cooperation “Unpredictable environments” (Jarvis etal. 1994; Avilés 1999); “Spatially structured environments” (Chao and Levin
1981; Nowak and May 1992)
Multicellularity “Spatially structured” environments (Pfeiffer and Bonhoeffer 2003)
Polymorphism Pattern of the environment in space and time (Levins 1968, p. 10)
Sex “Overall environmental heterogeneity” (i.e., both biotic and abiotic, spatial and temporal) (Toman and Flegr 2018)
Plasticity “Environmental variation in time or space” (Schmalhausen 1949; Bradshaw 1965; Lively (1986), building on Levins
(1968))
Quorum sensing “Fluctuations in cell-population density” (Miller and Bassler 2001)
Cognition “Environmental complexity” (Godfrey-Smith 1996)
Brains “Environmental variation” (Allman 1999, p. 2)
General intelligence (g) “Evolutionary novelty” (Byrne and Byrne 1995; Kanazawa 2004)
Cultural learning “Fluctuating environments”; “moderate autocorrelation” (Boyd and Richerson 1985)
1 Consider, for instance, Godfrey-Smith’s environmental complex-
ity thesis: “The function of cognition is to enable the agent to deal
with environmental complexity” (Godfrey-Smith 1996). Or con-
siderKanazawa’s hypothesis that general intelligence (g) is an adapta-
tion to "novel" environmental conditions (Kanazawa 2004).

Adapting toEnvironmental Heterogeneity: Selection andRadiation
1 3
section, the scheme (CG) is contrasted with how EH was
treated by adaptive explanations in the modern synthesis,
and it is suggested that a different approach to heterogeneity
adaptations is an important part of the innovativeness of the
EES. The fourth section builds on the pioneering work of
Levins (1968), and sketches how precisely defined patterns
of EH constitute the selective environment. The subsequent
section argues that this does not exhaust the concept of het-
erogeneity adaptation: some traits or properties are adaptive
to EH without a process of selection being responsible for
that adaptiveness.
Adaptation andtheUbiquity
ofEnvironmental Heterogeneity
To give a sense of how EH features in a broad range of eco-
evolutionary explanations, consider some leading hypoth-
eses of how motility, predation, or multicellularity evolved.
For instance, studies and simulations explicitly refer to heter-
ogeneity in nutrients (over time and/or space) as a condition
for motile organisms having a selective advantage over ses-
sile organisms (e.g., Wei etal. 2011). The reasoning behind
such studies is, roughly, that motility represents a spreading
of the risk associated with each location. Even when motility
is undirected (i.e., the organism moves randomly), on aver-
age organisms can avoid low-nutrient areas: this strategy
outperforms sessile strategies when there is sufficient change
in nutrient density over time. A second, related process that
is sometimes cited as favoring motility is that motility can
be a means to avoid competition (Hibbing etal. 2010): thus
motility is not only a response to heterogeneity in abiotic
variables such as nutrients, but can also be a response to
specifically biotic variables.
Motility is one of the oldest adaptations to EH, but the
ecological strategy of predation is also very old, having
evolved numerous times among bacteria (Jurkevitch and
Davidov 2007; Pérez etal. 2016). Moreover, while pre-
dation is a paradigmatically noncooperative interaction,
some forms of predation are not easily distinguishable from
symbiotic interactions: for instance, when a protist grazer
becomes enveloped within a larger host, it can be difficult
to say whether this is an instance of epibiotic predation
(i.e., predation by enveloping the prey) or of endosymbio-
sis (López-García etal. 2017). Intraspecific predation (i.e.,
cannibalism) in particular has explicitly been hypothesized
to be an adaptation to boom-bust dynamics abundance and
depletion (Polis 1981, p. 234).
Further illustrating the sometimes indistinguishability of
predation and cooperation, the two leading hypotheses on
the origin of eukaryotes are that they occurred either through
phagy (i.e., predation) or through symbiosis of plasmids or
mitochondria (see Blackstone 2016 or O’Malley 2010).2
Regardless of which hypothesis is the correct one, in either
case Blackstone notes in a review of the subject (Black-
stone 2013) that eukaryotes may have enjoyed a selective
advantage over prokaryotes in part due to their much greater
mobility, partially due to their much larger body size allow-
ing for lower Reynolds numbers and thus less encumbrance
by viscosity. Eukaryotes are also hypothesized to be able to
exploit “more or different” nutrients compared to prokary-
otes (Blackstone 2013, p. 2): another instance of how preda-
tion and/or cooperation are adaptive to EH.
Greater body size and motility are hypothesized as key
explanatory factors for why multicellularity arose (Schir-
rmeister etal. 2011). Greater cooperation (symbiosis) is
another: multicellularity allows for cooperative hunting,
cooperation for excretion of enzymes, or cooperation to
produce antipredator toxins (Pfeiffer and Bonhoeffer 2003).
For instance, one of the most basic transitions to multicel-
lularity involves cooperating predatory Myxococcus bacteria
(Grosberg and Strathmann 2007; Berleman and Kirby 2009).
As these hypotheses are advanced—either formally via
experiment and simulation, or informally as a background
explanation to help make sense of why some trait/property
evolved—an explanatory structure emerges similar to the
one familiar in the evolution of plasticity literature, where
one of two rival traits/properties (plastic versus fixed, motile
versus sessile, predatory versus grazing, multicellular versus
unicellular) is favored in conditions of environmental hetero-
geneity. The question naturally emerges why this explana-
tory structure seems to be so versatile, especially given how
heterogeneity is a quasi-metaphysical and ubiquitous prop-
erty of natural environments.
To emphasize why this question should be posed, con-
sider how adaptation is commonly construed as an adap-
tation to a specific "factor" in the environment (see, for
instance, the "feature-factor" relationships in Bock 1980).
When adaptation to EH is then considered (e.g., in the lit-
erature on phenotypic plasticity), it is analyzed as a simple
variation (e.g., on/off) on a single factor. This common way
of representing adaptation—as a feature-factor relation-
ship—gives the misleading impression that there are two
types of adaptation: adaptations to fixed factors, and adapta-
tions to EH. Yet, as will now be argued, such "factors" are
2 An alternative hypothesis is that prokaryotes and eukaryotes have
a common ancestor that was neither a prokaryote nor an eukaryote
(Forterre 2013). Forterre proposes that, “the ancestors of archaea
(and bacteria) escaped protoeukaryotic predators by invading high
temperature biotopes, triggering their reductive evolution toward the
’prokaryotic’ phenotype” (Forterre 2013, p. 1).

H.Desmond
1 3
themselves idealizations of patterns of EH. Every adaptation
is an adaptation to EH.
For instance, in Darwin’s distinction between "mountain-
ous" and "lowland" habitats, the label "mountainous" in real-
ity refers to a pattern of fluctuation of altitude, vegetation,
competitors, prey, and so on. In other words, a species of
wolf adapted to mountainous habitats is adapted to a par-
ticular pattern of heterogeneity across a number of different
environmental variables. Thus, strictly speaking, every trait
or property that is adaptive to any environmental "factor" is
in fact adaptive to a particular pattern of EH.
It is unclear what, if any, "feature-factor" relationship
could not be analyzed in this way. In fact, the very existence
of an organism entails a heterogeneity between the environ-
ment within the organism and the environmental external
to the organism. Thermodynamic equilibrium is by defini-
tion incompatible with life: not only does the existence of
organisms entail a gradient, but to persist organisms need
gradients in various variables. Even the basic (and likely
essential) property of metabolism involves the organism
feeding on gradients in chemical elements (iron, zinc, phos-
phorus, etc.) and creating novel gradients. For instance,
bacteria consume both oxygen and H2S, and thus can create
steep gradients in these compounds over short spatial scales
(Fenchel 2002). Interestingly, these waste products can, in
turn, be consumed by other species as metabolic inputs
(Fenchel 2002).
Of course, it may be justified to abstract away from pat-
terns of EH: the precise fine-grained patterns of fluctuation
within mountainous habitats may not matter when explain-
ing why a certain variety or species of wolf is adapted to
those habitats (as opposed to another variety/species). Simi-
larly, iron is central to the metabolism of almost every liv-
ing creature, and in fact iron metabolic pathways are almost
ubiquitous across all phylogenies (Frausto da Silva and
Williams 2001, pp. 512ff). So any model of the adaptive
evolution of two wolf varieties will not need to take into
consideration the micro-fluctuations in iron density: the two
varieties both have robust iron metabolism so that patterns in
iron fluctuation will likely not cause fitness differences. (By
contrast, micro-fluctuations in iron density may be important
for the evolution of some prokaryote lineages.)
The lesson here is not the complexity of natural environ-
ments cannot be idealized for certain explanatory ends, but
that EH is ubiquitous in natural environments. There are no
"factors" in natural environments. In a sense this lesson is
simply a restatement of the core insight of process ontology
(Dupré 2012): heterogeneity is a quasi-metaphysical prop-
erty, reflecting the processual nature of environments. The
processual nature may be safely ignored in many adaptive
explanations, but only because the adaptations to those pro-
cesses are very widespread and hence do not affect fitness
differences between organisms.
In sum, adaptation to heterogeneity seems to play an
intriguing role in explanations of the evolution of basic
ecological strategies (motility, predation, etc.) and of the
evolution of evolutionary transitions (eukaryotes, multicel-
lularity). Yet, upon closer reflection, a "fixed environmental
factor" simply is an idealization of a dynamic process, and
so every trait is adaptive to some pattern of EH. So what
does it mean to explain a trait or property as adaptive to EH,
and why is it explanatory?
Adaptations toEH andtheModern Synthesis
Part of the story here is how the adaptive significance of EH
was discounted in the modern synthesis. One instance of this
discounting, as mentioned in the introduction, is well-docu-
mented in how the adaptiveness of plasticity was relatively
ignored in the modern synthesis (see, e.g., Pigliucci 2006 or
Pigliucci and Müller 2010). However, the discounting can
be generalized to how the environmental heterogeneity was
approached in the modern synthesis.
There is much controversy about the meanings of the
terms "modern synthesis" and "Extended Evolutionary Syn-
thesis," but for purposes here the former will be understood
as a vision of biological evolution, where evolution is con-
ceived as the change in allele frequencies through mutation,
selection, drift, and migration. In other words, it is a vision
of evolution where the paradigm is the theoretical frame-
work of population genetics.
To understand the role of EH in this vision, it is instruc-
tive to recall one of the central explanatory interests of popu-
lation genetics: the mapping of patterns of heredity between
generations. Historically this explanatory interest was tied
up with one of the core goals of eugenics, namely, the “the
conscious social direction of human biological evolution”
(Muller 1935, p. 41; cited in Kevles 1985, p. 176). Thus,
for instance, it is no coincidence that more than a third
of Fisher’s monumental The Genetical Theory of Natural
Selection (Fisher 1930) was dedicated to the socioeconomic
and eugenicist implications of his mathematical synthesis
of Mendelism with Darwin’s theory of natural selection.
The hope here was that, once the patterns of heredity were
understood, one could shape the phenotypes of future gen-
erations by selecting which individuals of the current gen-
eration would reproduce. In other words, understanding
heredity would allow for “better” artificial selection, through
measures ranging from anticonception and marriage laws, to
forced sterilization and other more extreme measures.
This explanatory focus influenced how environmen-
tal heterogeneity was conceived in adaptive explanations.
Altering the environment (i.e., creating EH) was viewed
as important to control selection pressures, and thus direct
human evolution. However, here the patterns of EH were not

Adapting toEnvironmental Heterogeneity: Selection andRadiation
1 3
viewed as occasioning specific selection pressures, but rather
as creating different selection pressures over time. In other
words, heterogeneity in the external environment (variation
in the factors affecting fitness) was viewed as corresponding
to heterogeneity in the selective environment (variation in
selection pressures).
This explanatory interest in heredity helps explain why
EH was often approached as an interfering factor to be mini-
mized. One of the main types of experimental investigation
into patterns of heredity of traits is known as "common gar-
den experiments" where trait distributions in a parent and
offspring populations inhabiting a common garden are com-
pared. Such experiments aim at minimizing environmental
heterogeneity (
VE,VG×E
), so that phenotypic variation could
be attributed to genetic variation:
where the rest term r refers to all residual unexplained varia-
tions. Mapping the genetic basis of phenotypic variation (
VG)
then allows for the classic questions of population genetics
to be addressed: e.g., how alleles at different loci interact to
produce phenotypes (e.g., epistasis), or how alleles at the
same locus interact to produce a phenotype (dominance,
codominance, incomplete dominance). Without eliminating
the environmental component of phenotypic variation, these
questions cannot be answered. For instance, if a phenotypic
trait is not transmitted from one generation to the next, it
becomes unclear whether this is due to dominance effects,
or due to the offspring developing in a different environment.
These difficulties remain present in the genomic era (de Vil-
lemereuil etal. 2016).
One of the first departures from this theoretical treatment
of EH was Bradshaw’s assertion that phenotypic variation
due to environmental heterogeneity was too often being
explained as developmental noise. Instead, some phenotypic
variation was claimed to be adaptive to EH:
Much of the evidence has taken the viewpoint that
stability and adaptation are correlated, and that lack
of stability indicates lack of adaptation. […] But […]
it seems that plasticity, or lack of stability, can be of
positive adaptive value in many circumstances. (Brad-
shaw 1965, pp. 115–116)
Considering this historical background of the explanatory
goals and conceptual framework of the modern synthesis,
this suggests that environmental heterogeneity is primarily
conceptualized in the modern synthesis as a source of con-
tingency, producing variable selection pressures.
In sum, if one considers how EH was approached
within the modern synthesis, the coarse-grained explana-
tory scheme—i.e., explaining a trait as adaptive to EH as
such—can be interpreted as signaling a departure from the
VP=VG+VE+VG×E+r=VG+r
modern synthesis. Hypothesizing a trait or property to be
a heterogeneity adaptation is, in a sense, to categorize it as
a non-textbook (NT) adaptation: the role of EH cannot be
abstracted away, and the environment cannot be conceptual-
ized in terms of feature-factor relationships. This suggests
the following contrastive reading of the CG-scheme:
The evolution of property P ("heterogeneity adapta-
tion") can be partially explained by P being adaptive
to EH, and not to any fixed environmental state. (NT)
Selective Heterogeneity
The scheme (NT) can help situate heterogeneity adaptations
with respect to the EES, but as such it does not do much to
help clarify the explanatory logic of categorizing proper-
ties/traits as heterogeneity adaptations. In this section I will
recapitulate some of the core insights of Levins (1968), and
sketch one major analysis of adaptation to EH.
Models ofSelection forHeterogeneity Adaptations
Consider a continuous range of phenotypes with fitnesses
W1
when the environment is in state
E1
and fitnesses
W2
when
the environment is in state
E2
. These states can represent
different states in condition variables (e.g., temperature), or
in resource variables (high or low food availability, absence
or presence of predators). A core assumption is that no sin-
gle phenotype will maximize fitness in both environments:
every phenotype involves some trade-offs in fitness. The
mathematical construct that encapsulates this idea is the
"fitness-set"3 (see Fig.1).
The figure includes three dashed lines: each represents
the set of (theoretically) possible fitness trade-offs that maxi-
mize fitness when environments
E1
and
E2
occur with rela-
tive probabilities
p
and
(1−p)
. The fitness set represents a
constraint on what fitness trade-offs are actually possible, so
the optimal fitness trade-off is given by the intersection of a
dashed line with the fitness set. When only
E1
(or only
E2
)
occurs, the optimum is the fitness trade-off that maximizes
W1
(or
W2
): a specialized phenotype. When
E1
and
E2
occur
with relative probabilities
p
and
(1−p)
, then the optimal
tradeoff is determined by the intersection of the fitness set
with the line
pW
1
+(1−p)W
2
=wmax
,
p
for maximal
wmax
,
p.
This is an intermediate phenotype.
3 This is a reformulation of Levins (1968), but without discussion of
the case of “fine-grained heterogeneity” (i.e., temporal heterogeneity
shorter than generation time, and spatial heterogeneity smaller than
habitat spatial scale). This is sufficient for purposes of this article; for
more details see Levins (1968, pp. 18ff).

H.Desmond
1 3
In general, the optimal fitness trade-off is determined by
the geometry of the fitness set, and in particular, its convex-
ity or concavity. For a very convex fitness set (left graph in
Fig.2), even if the environment shifts from an
E1
-dominated
state (
p≈1
) to an
E2
-dominated state (
p≈0
), the optimal
fitness trade-off does not change much. In other words, spe-
cialized and intermediate phenotypes (or fitness trade-offs)
do not differ much.
By contrast, in a concave fitness set, such as the right
graph in Fig.2, there are no optimal intermediary states.
Instead, as the environment shifts from an
E1
-dominated
state (
p≈1
) to an
E2
-dominated state (
p≈0
), the optimal
phenotype jumps discontinuously.
It is in a concave fitness set that environment-tracking can
be optimal. Typical models of, for instance, the evolution of
plasticity (e.g., Lively 1986; Moran 1992; Godfrey-Smith
1996) all take the relative frequency
p
of two environment
types as a key parameter. Another important parameter is
the probability of developing the adaptive phenotype in
E1
and
E2
(i.e., the probability of “making the right choice”).
All such models identify cutoffs between situations where
monomorphic specialization is optimal, and where flexible
tracking is optimal. For instance, in Lively’s model,
E1
and
E2
respectively represent benign and harsh patches. Here
are clear conditions for the environment-tracking strategy
being optimal (see Fig.3, box a): the probability of making
the right choice has to be large enough, and the probability
of the benign patch occurring cannot be too high (because
then it would be optimal to just ignore the occurrence of
harsh patches).
In Lively’s model, intermediate phenotypes are never
optimal, regardless of how often benign patches occur (i.e.,
the value of p). Instead, the choice is between monomor-
phic specialization (A or B), environmental tracking (C),
Fig. 1 The fitness-set, given by the curve, for a range of phenotypes. Each point on this set represents a phenotype with a specific fitness trade-
off
Fig. 2 A very convex fitness set (left) and a concave fitness set
(right). In a very convex fitness set, large changes in the pattern of
heterogeneity do not lead to large changes in the optimal fitness trade-
off. In a concave fitness set, small changes in the pattern of heteroge-
neity can lead to discontinuous changes in the optimal fitness trade-
off

Adapting toEnvironmental Heterogeneity: Selection andRadiation
1 3
or polymorphism (e.g., A and B, or A and C). The latter
occurs in Fig.3,boxes b and c, when competition between
strategies is allowed.
The convexity or concavity of a set is a representation of
the tolerance of specialized phenotypes (Levins 1968, pp.
17–18). The quantities
(w
1,
A−w
2,
A)
and
(w
1,
B−w
2,
B)
meas-
ure “how much worse” the specialized phenotypes do in the
environment in which they are not specialized (see Fig.4).
When a specialized phenotype performs very poorly in an
adverse environment, then tracking may be optimal, even
though the latter may not do as well in benign environments.
The tolerance determines whether monomorphic or track-
ing/polymorphic strategies maximize fitness (e.g., Godfrey-
Smith 1996, p. 211).
What determines whether polymorphism or environ-
mental tracking is selectively favored depends on the spati-
otemporal scale of heterogeneity, and specifically, whether
the temporal scale exceeds generation time or the spatial
scale exceeds the typical habitat range of an individual.4
This means that an individual in the course of a lifetime
will typically not encounter any EH, but a population will
encounter EH. In the latter case, a mix of specialized phe-
notypes (i.e., polymorphism) is optimal; in the former, evo-
lutionary tracking will be favored.
Note that models of selection for plasticity (e.g., Lively
1986; Moran 1992; Godfrey-Smith 1996) typically do not
explicitly represent the tolerance of phenotypes or the scale
of heterogeneity. Instead, they tend to focus on the reliabil-
ity of cues: the probability that tracking will be accurate in
various environments. However, the reliability of the cue
is sometimes related to the temporal scale of EH: when the
latter is shorter than the reaction time of the phenotype, the
phenotype can never adapt on time. Thus, for instance, if
the density of predators varies on a timescale that is shorter
than the reaction time of the defenses of bryozoans, by the
time the defenses are set up, the predators will be gone. The
pattern of EH will effectively be just noise for the bryozoan.
Two lessons can be drawn from this technical discussion.
The first lesson is that a main distinction among adapta-
tions to EH is to be made between monomorphic phenotypes
(either intermediate or specialized phenotypes) and tracking
phenotypes. The second is that the concept of a “pattern of
EH” can be spelled out in detail, in terms of dimensions such
Fig. 3 Environment-tracking is optimal when the environment turns harsh sufficiently frequently, and when the organism is sufficiently accurate
in choosing the right patch.Reproduced from Lively (1986, p. 658)
Fig. 4 Tolerance curves for
a range of phenotypes in
environments
E1
and
E2
, with
monomorphic specializations
A and B
4 Levins terms this "coarse-grained" heterogeneity (Levins 1968, p.
18).

H.Desmond
1 3
as: (1) the geometry of fitness sets (containing information
about environmental tolerance), (2) relative frequency of
environment types, (3) spatiotemporal scale of EH, and (4)
reliability of cues. To this could be added: (5) autocorrela-
tion between environments across generations (how similar
the parent environment is to the offspring environment: this
adds a multigenerational timescale), and (6) the cross cor-
relations between environmental variables. For purposes of
this article, a “specific pattern of EH” can be understood to
be a pattern definable by such dimensions.
Adapting toSelective Heterogeneity
The following analysis of the explanatory scheme (CG) can
be proposed: when a property is explained as a heterogeneity
adaptation, this can mean that the property was selectively
favored in a selective environment characterized by a par-
ticular pattern of EH.
The evolution of property P can be partially explained
by P selectively favored by a pattern of EH. (SH)
Here it is the pattern of EH, and not the brute fact of
EH, that is deemed explanatorily significant. The selective
heterogeneity scheme (SH) also covers the previous scheme
(NT), since adaptation to fixed environmental states can be
analyzed as a special case of an adaptation to a pattern of
EH. When the pattern matters, parameters such as the rela-
tive frequency of environmental states, or the spatiotemporal
scale of variation matter. Other factors, such as autocorrela-
tion, or cross-correlation between environmental variables
also define the pattern of EH.
As an illustration, consider two environments: one is
more or less stable, characterized by slow climate changes,
and the second environment varies at seasonal rates. Type
O1 is adaptive to the first environment, while the second type
O2 is adaptive to the second. Figure5 illustrates this situa-
tion, where X is some environmental variable (temperature,
humidity, nutrition density). The average value of X is the
same in both E1 and E2, but, from the perspective of the
organism, in E2 there is more uncertainty due to intragen-
erational change. The right graph in Fig.5 represents E1
and E2 as two selective environments with different selective
regimes: O1 is fitter in E1, and O2 is fitter in E2.
When one would claim O2 as adaptive to EH—for
instance, migratory behavior among birds is viewed as adap-
tive to EH (Rappole 2013)—then this claim is explanatory
when a specific pattern of EH is intended (e.g., seasonal
change).
The scheme (SH) can be further illustrated by applying it
to some representative examples listed in Table1:
Sexual reproduction is adaptive to EH insofar as it is a
mixed strategy at the level of populations.5 Unlike polymor-
phism in a single trait, sexual reproduction can be explained
as adaptive to heterogeneity in a large number of environ-
mental variables.
Motility is adaptive to EH insofar as it is a mixed strat-
egy at the level of the individual, whereby location is varied
randomly over time in order to maximize the probability of
exposure to nutrient-rich locations (see Wei etal. 2011).
Most mechanisms of motility involve some sensory system
Fig. 5 Patterns of EH individuate selective environments
5 Adaptive strategies at the level of the individual are adaptive to the
environment of individuals; whereas those at the level of the popula-
tion are adaptive to the environment of the population. For our pur-
poses here, the environment of the population can be understood to
consist of the sum total of the environments of the individual.

Adapting toEnvironmental Heterogeneity: Selection andRadiation
1 3
as well (e.g., chemotaxis), and suchsensorimotor systems
are strategies of environment-tracking (see, e.g., Fenchel
2002).
Phenotypic plasticity is adaptive to EH insofar as it is
a strategy of environment-tracking: as documented above,
plasticity is adaptive to patterns of EH that exceed the tol-
erance of fixed phenotypes and with scales of fluctuation
that are smaller than generation time or typical habitat size.
Depending on further details of the pattern (autocorrelation
of environments across generations) and interaction between
individual and pattern (accuracy of cue, time lag of cue),
different types of plasticity are optimal (for reference, see,
e.g., Pigliucci 2001, p. 200).
Cognitive and proto-cognitive properties are adaptive
to EH insofar as they constitute a strategy of environment-
tracking at the level of individual. Many analyses of the
evolution of cognition draw heavily on the adaptive evolu-
tion of plasticity and/or sensorimotor systems. For instance,
Godfrey-Smith’s claim that cognition is adaptive to envi-
ronmental complexity is supported by a model of the selec-
tion for single-trait phenotypic plasticity (Godfrey-Smith
1996). Accounts that analyze proto-cognition as sensorimo-
tor capacities point to the latter as allowing an organism to
“better adapt to rapidly changing environmental conditions”
(van Duijn etal. 2006).
Cultural learning is adaptive to patterned EH insofar it is
a strategy of environment-tracking at the level of population.
Adaptation via cultural learning is favored over adaptation
via genetic change when the magnitude of EH at the time-
scales of multiple generations is sufficiently high, such that
tolerances are exceeded (see Boyd and Richerson 1985, p.
111).
Radiative Heterogeneity
This final section identifies and explores a second kind of
adaptation to environmental heterogeneity. Here the process
of adaptation occurs without being caused by fitness differ-
ences or natural selection, for instance, when an organism
encounters a novel exploitable resource and is the first to
take advantage. The organism cannot, by definition, share a
common selective environment with its rivals, because its
fitness (but not theirs) is influenced by the novel resource.
Hence the fitness differences are not commensurable: a pre-
requisite for a selective environment to be shared (Brandon
1990). The adaptive evolutionary process that ensues is that
of adaptive radiation. Hence the type of EH that gives rise
to this process is termed "radiative heterogeneity."
Novelty andtheLimits ofSelective Environments
As a concrete illustration of radiative heterogeneity, consider
the following passage:
A question that remains, however, is whether harsh
demographic or ecological conditions are necessary
for sociality to evolve. It is possible that sociality may
also have arisen in species sitting comfortably in a
region of sustainable growth, if, by cooperating, indi-
viduals could gain access to rich and otherwise inac-
cessible resources […]. Potential examples of this sec-
ond alternative are the conifer-bark beetles, which, by
attacking en masse, are able to overcome the defenses
of live trees […] and some of the cooperative hunters
such as wolves […] and social spiders […], which by
hunting in groups are able to gain access to a range of
prey sizes unavailable to non-social species of a simi-
lar body size […]. (Avilés 1999, p. 470)
In other words, here it is hypothesized (and for similar
pictures of the evolution of cooperating beetles, see, e.g.,
Berryman etal. 1989 or Birch 1984) that the typical selec-
tion pressures associated with patterns of EH (harsh con-
ditions, with patchy resources) were not necessary for the
evolution of sociality. Instead, an environmental variable
that had not previously impacted fitness values, namely the
presence of live trees,6 now is part of the external environ-
ment of groups of bark beetles due to a behavioral innova-
tion. Sociality is adaptive to this novel pattern of EH, and
the cooperating conifer-bark beetles will continue to evolve
in this novel environment (and start competing with each
other). Thus the exploitation of the resource leads to adap-
tive radiation.
Paradigmatic cases of adaptive radiation, like the radia-
tion of Darwin’s finches to different islands, involves physi-
cal migration to spatially distinct environments with differ-
ent characteristics. However, a more general characterization
of radiation is: “invading underexploited environments”
(Grant etal. 2013). In this way radiation can involve pro-
cesses without any physical displacement, but can be caused
by developmental, behavioral innovation, or evolutionary
change, such as new sensory capacities being evolved, or
larger body size allowing previously inaccessible resources
to be exploited. While radiation involves evolutionary diver-
gence of populations (e.g., speciation of an ancestral spe-
cies into descendant species), radiation can sometimes be
considered a process of “adaptation” in the sense that the
6 More precisely: the fitness impact of live trees is “screened off”
(sensu Salmon 1984) by the fitness impact of dead trees. So live trees
qua live trees (and not just potential dead trees) were not part of the
external environment.

H.Desmond
1 3
radiating populations avail themselves of different “ecologi-
cal opportunities” (Grant etal. 2013, p. 561).
Inspired by this, the concept of radiative heterogeneity
can be defined in the following way:
Radiative heterogeneity: a pattern of EH in the external
environment of an individual or population is radiative if:
(1) it is novel, and (2) it does not constitute any selective
environment of the individual or population.
Let us unpack this definition. First, stipulating that a
radiative pattern of EH does not constitute any selective
environment means that the pattern does not occasion any
new selection pressures or give rise to novel selection pres-
sures. It does affect fitness (it can represent, for instance, an
underexploited resource), but because the radiating popula-
tion inhabits a different external environment, the fitness
differences between radiating and non-radiating populations
are not commensurable and hence do not constitute selection
(on the short-term at least).
To give this some formal detail(Fig.6): suppose that
organism type
O3
gains sensitivity to fast variations in envi-
ronmental variable X (e.g., intraday variation as opposed to
slower seasonal variation). Assume for the sake of simplicity
that these fast variations occur in a spatially distinct environ-
ment
E3
. After evolving or developing the novel capacity,
O3
migrates to this (uninhabited) environment
E3
.
Organism
O3
is adapted to
E3
, even though there was not
any selection process that led to this adaptiveness. The novel
pattern was not part of
O3
’s previous selective environment
(shared with
O1
). In
O3
’s novel external environment (
E3
),
the concept of selective environment is not well-defined,
since there are no fitness differences to speak of. Unlike the
case of selective heterogeneity, radiative heterogeneity does
not generate any specific selection pressures: it does not even
make sense to ask what selective environment is associated
with a pattern of radiative EH.
In general, adapting to radiative heterogeneity need not
involve a preexisting calibrated response to a well-defined
pattern of heterogeneity. Simply escaping selective com-
petition, for instance by migrating outside of one’s usual
habitat, can be adaptive, and in this sense can be considered
as an instance of adapting to radiative heterogeneity. Escap-
ing a selective environment may be adaptive if the potential
opportunity of the unencountered outweighs the potential
risk, despite the precise pattern of EH being unknown. As
previously mentioned, this is precisely how motility is some-
times considered to be an adaptation to EH.
This contrast between radiative and selective heteroge-
neity implicitly draws on the fact that the Lewontin con-
ditions—heritable fitness differences in phenotype—are,
strictly speaking, insufficient to characterize natural selec-
tion. To give an exaggerated example: fitness differences
between peppered and black moths constitute natural selec-
tion, but fitness differences between orca and krill do not.
The concept of selective environment is one way to account
for why fitness differences between orca and krill are not
commensurable: very different environmental processes
determine their respective fitnesses. A pattern of radiative
EH is a novel pattern that only part of the ancestral popula-
tion can exploit, and hence the fitness of individuals in the
ancestral population are not affected by the same environ-
mental variables. In this way, the novel pattern cannot be
considered as part of a common selective environment.
The other definitional element of radiative heterogeneity
needing unpacking is the concept of environmental novelty.
A full treatment of this concept would warrant a separate
paper; it is sufficient for the purposes here to understand a
novel pattern as a pattern that is part of the current exter-
nal environment, but was not part of a previous external
environment and so did not previously affect fitness. This is
continuous with other definitions of environmental novelty,
for instance, that of Sol etal. (2005) where novelty refers to
“areas outside the natural geographic range of the species,”
and specifically to novel resources or unknown enemies in
those areas.
The concept of novelty is relevant for understanding the
example of cooperating and noncooperating beetles (Avilés
1999). The property of "cooperation" allows the cooperat-
ing beetles to exploit a factor in the environment with which
Fig. 6 An illustration of radiative heterogeneity (as a fast variation in X), and how organism
O3
is adapted to
E3
despite not having been selected
for in environment
E3

Adapting toEnvironmental Heterogeneity: Selection andRadiation
1 3
there had not been any previous direct interaction. This
means that they do not entirely share a selective environment
with the non-cooperating beetles. However, this is not to say
that at a later stage there may be some selective competition:
if the population size of cooperating beetles grows, then this
could lead to subsequent displacement of noncooperating
beetles. Nonetheless, initially, the adaption to environmental
heterogeneity occurs without natural selection in a common
selective environment. In this way the distribution of live
trees represents an instance of radiative heterogeneity.
Adapting toRadiative Heterogeneity
Intuitively, adaptations to radiative heterogeneity simply
allow for novel ecological opportunities to be exploited.
However, the traits/properties discussed here go beyond
traits such as the different beak sizes of Darwinian finches,
but concern fundamental properties like motility or mul-
ticellularity: properties that characterize a broad swathe
of phylogenies and that have colonized many habitats and
niches. To account for the fundamental nature of radiative
heterogeneity—and associated adaptations—three different
types of radiative heterogeneity can be distinguished.
One type of radiative heterogeneity reflects the spatiotem-
poral extension of resources. This means that properties that
allow organisms and populations to take advantage of the
extension of resources—venturing outside a species’ normal
habitat range, migration, and so on—are adaptations. Habitat
expansion means that the organism can avail itself of more,
and more varied, resources (food types, mating opportuni-
ties, etc.). As mentioned, this is why motility is considered
adaptive, and thus motility is not only adaptive to selective
heterogeneity (giving motile organisms a selective advan-
tage), but also to radiative heterogeneity.
A second type of radiative heterogeneity reflects the
multidimensionality in resources (the various types of food,
shelter, potential cooperators, etc.), where not all potentially
exploitable resources are actually exploitable by extant
organisms. To revisit the external environment of bacterial
life (second section): initially resources such as light and
many of the chemical elements were not exploited by bac-
terial life forms. As life evolved more patterns of EH were
exploited. For instance, the evolution of chlorophyll allowed
for high-density regions of solar radiation to be exploited.
Moreover, as life evolved, novel patterns of EH were created,
as organisms created metabolic inputs for other organisms,
either by their waste products or by their existence. The evo-
lution of predation allowed for the exploitation of the latter.
A third type of radiative heterogeneity reflects the multi-
tude of existing patterns of EH. Not only are environments
characterized by a large number of resource types that are
unevenly spread across space and time, but the distribution
of these resource types follows intricate patterns that are
often imperfectly regular and involve the superposition of
many simple (i.e., sinusoidal) patterns. For instance, resident
birds may pick up on complex patterns in resource distri-
bution in their regular habitat; however, these distributions
may change over time (for instance, due to seasonal change)
and space (for instance, due to latitude). Migratory birds
exploit these latter spatiotemporal patterns of heterogene-
ity, and in this way avian migration can be considered as an
"adaptation" (e.g., Rappole 2013, p. 162). However, inter-
estingly, migrant species usually do not directly compete
with resident avian species, and are primarily in competition
with other migrant species (cf. Rappole 2013, p. 160): the
spatiotemporal pattern of seasonal and latitudinal variation
defines a novel selective environment in which different vari-
ants of migrant behavior are selected for.
Nonetheless, migrant and resident variants can still enter
into competition, especially when resources are scarce. The
reason for this is that seasonal variation is only adaptively
significant because it is a cue for an already existing selec-
tive pattern of EH, namely (underexploited) food resources.
In this way, this third type of radiative heterogeneity can be
considered to refer to additional dimensions in exploitable
resources: subtle undetected patterns represent an "infor-
mational resource" that can be exploited. This third type of
radiative heterogeneity also reflects how the spatiotemporal
distribution of resources involves the superposition of many
simple patterns, and hence can be predictable to varying
degrees. One organism type may be able to exploit slow,
seasonal variation, whereas another organism type may be
able to additionally exploit faster, diurnal variation.
More work would need to be done to formally analyze the
category of "radiative heterogeneity" with the same preci-
sion as that with which selective heterogeneity was analyzed.
However, even this short analysis is sufficient to suggest why
so many very fundamental evolutionary innovations seem to
be adaptive to radiative heterogeneity: radiative heteroge-
neity is ubiquitous and itself reflects fundamental physical
properties of environmental resources, such as multidimen-
sionality and spatiotemporal extension. An adaptation like
motility directly takes advantage of spatiotemporal exten-
sion; an "adaptation" like cooperation allows access to novel
resources and thus to take advantage of multidimensionality
in the environment.
In sum, the coarse-grained explanatory scheme (CG) can
be understood to refer not just to adaptations to selective het-
erogeneity, but also to adaptations to radiative heterogeneity:
The evolution of property P ("heterogeneity adapta-
tion") can be partially explained by P allowing for
exploiting radiative patterns of EH. (RH)
The selective heterogeneity (SH) and radiative heteroge-
neity (RH) explanatory schemes are not mutually exclusive

H.Desmond
1 3
and can be applied to the same property; they just refer to
distinct adaptive processes. In general, an adaptation to
radiative EH may be used to gain a selective advantage, and
adaptations to selective EH may be used to exploit novel
resources. Motility, in that it allows bacteria to both avoid
selective competition as well as occupy the most desirable
positions, is a case in point.
The scheme (RH) can be further illustrated by applying it
to some representative examples listed in Table1:
Motility is adaptive to radiative heterogeneity insofar as
it allows escaping selective competition and/or colonizing
novel patterns of EH. Bacterial motility is often used to gain
a fitness advantage in a common selective environment—
for instance, to gain access to more oxygenated or nutrient
rich regions in biofilms (Hibbing etal. 2010, p. 7). This
would be an instance of motility as an adaptation to selective
heterogeneity. However, some bacterial strains have been
noted to use motility to actively evade competition, and to
disperse more rapidly into less occupied regions (Hibbing
etal. 2010, p. 7). Here motility would be an adaptation to
radiative heterogeneity.
Cooperation is adaptive to radiative heterogeneity insofar
as it allows colonizing novel patterns of EH. We already
noted the example of cooperating bark beetles; cooperation
is of course a widespread and biologically fundamental phe-
nomenon. It lies at the basis of multicellularity, which has
been hypothesized to have allowed for the exploration of
new niches (Schirrmeister etal. 2011). Wolfpacks of coop-
erating predatory bacteria (Myxococcus) are considered to
be a form of rudimentary multicellularity (e.g., Grosberg
and Strathmann 2007), and exemplify in another way how
motility also is adaptive to radiative heterogeneity.
Cultural learning is adaptive to radiative heterogene-
ity insofar as it has allowed significant niche expansion in
human evolutionary history. Culturally transmitted knowl-
edge (e.g., about which roots are nutritious and which are
toxic) as well as culturally mediated cooperative hunting
allowed for dietary expansion, and this dietary expansion
was one of the factors that allowed the geographic expan-
sion of the human metapopulation (Teaford and Ungar 2000;
Ulijaszek 2002). This dietary expansion also involved access
to previously inaccessible food sources, such as big game
through cooperative hunting. Innovations such as structured
hearths, more sophisticated artificial shelters, and clothing
allowed for buffering against periods of severe cold dur-
ing glacial periods as well as migration to areas with cold
climates (Gilligan 2007). Such properties, allowing popula-
tions to undergo niche evolution, were clearly not primar-
ily a means to selectively compete with other populations,
although they have been linked with some interspecific com-
petition with, e.g., Neanderthals (cf. Gilligan 2007).
Endothermy (a homeostatic process) is adaptive to radia-
tive heterogeneity insofar as it allows colonizing novel
patterns of EH. Endothermy allows mammals and birds to be
less affected by diurnal cycles in temperature than ectother-
mic organisms: they can keep on feeding without having to
resort to "basking" behaviors or microhabitat selection (cf.
Ruben 1995, p. 70). Hence endothermy is thought to have
allowed for organisms to expand into new niches – areas
with sparse resources requiring extended feeding periods, or
areas with cold temperatures (Ruben 1995, p. 71).
Discussion andConclusion
The main aim of the article was to draw attention to a rela-
tively common explanatory scheme where traits/properties
are explained as adaptations to environmental heterogene-
ity. Where the explanatory import lies is not obvious when
one considers the ubiquity of environmental heterogeneity.
Environmental heterogeneity is not special, so why invoke
it in an adaptive explanation?
Part of the story, it was suggested, lies in the contrast
with how adaptation is represented in textbooks as "feature-
factor" relationships and with how EH was treated in the
modern synthesis, namely as a source of contingency. It is
now widely recognized that patterns of EH can occasion
specific selection pressures, but perhaps it is still necessary
to emphasize the contrast with depictions of adaptation in
textbooks or in the modern synthesis.
However, the main positive contribution of the article
was to propose the distinction between two types of envi-
ronmental heterogeneity: selective heterogeneity, which
serves to define specific selection pressures, and radiative
heterogeneity, where EH functions as a resource that can be
exploited by individuals and populations. Radiative hetero-
geneity, in particular, seems to be ubiquitous, and this helps
to account for why the explanatory scheme of explaining
traits/properties as heterogeneity adaptations is so versatile
and widespread.
The notion of an adaptation to radiative heterogeneity
raises interesting further questions that cannot be treated
here. Adapting to radiative heterogeneity is, in the first
instance, an ecological process; calling it an "adaptation"
raises deeper questions as to how we should understand both
the concept of adaptation and the link between adaptation
and selection. On the one hand, adaptations such as coopera-
tive behavior can “fit” factors in the external environment,
such as the presence of live conifers. Evolutionary ecolo-
gists speak of "adaptations" when referring to a trait that
allows for resource exploitation or niche colonization. Yet
it is clear that adaptations to radiative heterogeneity cannot
be analyzed as adaptations according to perhaps the most
influential analysis of that concept, namely as a property that
gives (or has given during some selective history) a fitness
advantage in a given selective environment (Williams 1966;

Adapting toEnvironmental Heterogeneity: Selection andRadiation
1 3
Brandon 1978). Thus the question arises how this explana-
tory practice should be squared with canonical accounts of
adaptation in the philosophy of biology.
Other questions could also be raised about how natural
selection interacts with adaptations to radiative heterogene-
ity. It is clear that an adaptation to radiative heterogene-
ity can be used to gain a selective advantage. Motility, for
instance, allows bacteria to explore novel niches but also
to gain a selective advantage over nonmotile strains. Endo-
thermy can allow for niche expansion, but can also allow for
a selective advantage to be gained over ectotherm types in
a common selective environment. Sometimes the selective
advantage follows the niche expansion: does this mean that
some selective environment was shared all along? While
this is unclear, one can conclude that taking the default het-
erogeneity of environments into account does seem to cast
concepts such as "adaptation" and "natural selection" in a
different light.
Acknowledgments The author wishes to thank two anonymous review-
ers for useful feedback.
References
Allman JM (1999) Evolving brains. Scientific American Library. Dis-
tributed by W.H. Freeman and Co, New York
Avilés L (1999) Cooperation and non-linear dynamics: an ecological
perspective on the evolution of sociality. Evol Ecol Res 1:459–477
Beatty J (1995) The evolutionary contingency thesis. In: Wolters G,
Lennox JG (eds) Concepts, theories, and rationality in the biologi-
cal sciences. Universitätsverlag Konstanz-University of Pittsburgh
Press, Konstanz
Berleman JE, Kirby JR (2009) Deciphering the hunting strategy of a
bacterial wolfpack. FEMS Microbiol Rev 33:942–957. https ://doi.
org/10.1111/j.1574-6976.2009.00185 .x
Berryman AA, Raffa KF, Millstein JA, Stenseth NC (1989) Interaction
dynamics of bark beetle aggregation and conifer defense rates.
Oikos 56:256. https ://doi.org/10.2307/35653 45
Birch MC (1984) Aggregation in bark beetles. In: Bell WJ, Cardé RT
(eds) Chemical ecology of insects. Springer, Boston, pp 331–353
Blackstone NW (2016) An evolutionary framework for understanding
the origin of eukaryotes. Biology. https ://doi.org/10.3390/biolo
gy502 0018
Blackstone NW (2013) Why did eukaryotes evolve only once? Genetic
and energetic aspects of conflict and conflict mediation. Philos
Trans R Soc B Biol Sci 368:20120266. https ://doi.org/10.1098/
rstb.2012.0266
Bock WJ (1980) The definition and recognition of biological adapta-
tion. Am Zool 20:217–227. https ://doi.org/10.1093/icb/20.1.217
Boyd R, Richerson P (1985) Culture and the evolutionary process.
University of Chicago Press, Chicago
Bradshaw AD (1965) Evolutionary significance of phenotypic plas-
ticity in plants. Adv Genet 13:115–155. https ://doi.org/10.1016/
S0065 -2660(08)60048 -6
Brandon RN (1990) Adaptation and environment. Princeton University
Press, Princeton
Brandon RN (1978) Adaptation and evolutionary theory. Stud Hist
Philos Sci Part A 9:181–206. https ://doi.org/10.1016/0039-
3681(78)90005 -5
Byrne R, Byrne RW (1995) The thinking ape: evolutionary origins of
intelligence. Oxford University Press, Oxford
Chao L, Levin BR (1981) Structured habitats and the evolution of anti-
competitor toxins in bacteria. Proc Natl Acad Sci USA 78:6324–
6328. https ://doi.org/10.1073/pnas.78.10.6324
Cohen AA, Martin LB, Wingfield JC, McWilliams SR, Dunne JA
(2012) Physiological regulatory networks: ecological roles and
evolutionary constraints. Trends Ecol Evol 27:428–435. https ://
doi.org/10.1016/j.tree.2012.04.008
Darwin C (2008) On the origin of species. Oxford University Press,
New York
de Villemereuil P, Gaggiotti OE, Mouterde M, Till-Bottraud I (2016)
Common garden experiments in the genomic era: new perspec-
tives and opportunities. Heredity 116:249–254. https ://doi.
org/10.1038/hdy.2015.93
Dupré J (2012) Processes of life: essays in the philosophy of biology.
Oxford University Press, Oxford
Fenchel T (2002) Microbial behavior in a heterogeneous world. Science
296:1068–1071. https ://doi.org/10.1126/scien ce.10701 18
Fisher RA (1930) Thegenetical theory of natural selection. Oxford
University Press, Oxford
Forterre P (2013) The common ancestor of archaea and eukarya was
not an archaeon. Archaea. https ://doi.org/10.1155/2013/37239 6
Frausto da Silva JJR, Williams RJP (2001) The biological chemistry of
the elements: the inorganic chemistry of life. Oxford University
Press, Oxford
Futuyma DJ, Kirkpatrick M (2017) Evolution. Sinauer, Sunderland
Gilligan I (2007) Neanderthal extinction and modern human behav-
iour: the role of climate change and clothing. World Archaeol
39:499–514. https ://doi.org/10.1080/00438 24070 16804 92
Godfrey-Smith P (1996) Complexity and the function of mind in
nature. Cambridge University Press, Cambridge
Grant PR (2013) Adaptive radiation. In: Losos JB, Baum DA, Futuyma
DJ, Hoekstra HE, Lenski RE, Moore AJ etal (eds) The Prince-
ton guide to evolution. Princeton University Press, Princeton, pp
559–566
Grosberg RK, Strathmann RR (2007) The evolution of multicellularity:
a minor major transition? Annu Rev Ecol Evol Syst 38:621–654.
https ://doi.org/10.1146/annur ev.ecols ys.36.10240 3.11473 5
Hibbing ME, Fuqua C, Parsek MR, Peterson SB (2010) Bacterial com-
petition: surviving and thriving in the microbial jungle. Nat Rev
Microbiol 8:15–25. https ://doi.org/10.1038/nrmic ro225 9
Jarvis JUM, O’Riain MJ, Bennett NC, Sherman PW (1994) Mamma-
lian eusociality: a family affair. Trends Ecol Evol 9:47–51. https
://doi.org/10.1016/0169-5347(94)90267 -4
Jurkevitch E, Davidov Y (2007) Phylogenetic diversity and evolution
of predatory prokaryotes. In: Jurkevitch E (ed) Predatory prokary-
otes: biology, ecology and evolution. Springer, Berlin, Heidelberg,
pp 11–56
Kanazawa S (2004) General intelligence as a domain-spe-
cific adaptation. Psychol Rev 111:512–523. https ://doi.
org/10.1037/0033-295X.111.2.512
Kettlewell HBD (1955) Selection experiments on industrial mela-
nism in the Lepidoptera. Heredity 9:323. https ://doi.org/10.1038/
hdy.1955.36
Levins R (1968) Evolution in changing environments: some theoretical
explorations. Princeton University Press, Princeton
Lively CM (1986) Canalization versus developmental conversion in a
spatially variable environment. Am Nat 128:561–572
López-García P, Eme L, Moreira D (2017) Symbiosis in eukaryotic
evolution. J Theor Biol 434:20–33. https ://doi.org/10.1016/j.
jtbi.2017.02.031
Miller MB, Bassler BL (2001) Quorum sensing in bacteria. Annu Rev
Microbiol 55:165–199
Moran NA (1992) The evolutionary maintenance of alternative phe-
notypes. Am Nat 139:971–989. https ://doi.org/10.1086/28536 9

H.Desmond
1 3
Muller HJ (1935) Out of the night; a biologist’s view of the future.
Vanguard Press, New York
Nowak MA, May RM (1992) Evolutionary games and spatial chaos.
Nature 359:826–829. https ://doi.org/10.1038/35982 6a0
O’Malley MA (2010) The first eukaryote cell: an unfinished history of
contestation. Stud Hist Philos Biol Biomed Sci 41:212–224. https
://doi.org/10.1016/j.shpsc .2010.07.010
Pérez J, Moraleda-Muñoz A, Marcos-Torres FJ, Muñoz-Dorado J
(2016) Bacterial predation: 75 years and counting! Environ Micro-
biol 18:766–779. https ://doi.org/10.1111/1462-2920.13171
Pfeiffer T, Bonhoeffer S (2003) An evolutionary scenario for the transi-
tion to undifferentiated multicellularity. Proc Natl Acad Sci USA
100:1095–1098. https ://doi.org/10.1073/pnas.03354 20100
Pigliucci M (2006) Phenotypic plasticity and evolution by genetic
assimilation. J Exp Biol 209:2362–2367. https ://doi.org/10.1242/
jeb.02070
Pigliucci M (2010) Phenotypic plasticity. In: Pigliucci M, Müller G
(eds) Evolution: the extended synthesis. MIT Press, Cambridge
Pigliucci M (2001) Phenotypic plasticity: beyond nature and nurture.
Johns Hopkins University Press, Baltimore and London
Pigliucci M, Müller GB (2010) Elements of an extended evolution-
ary synthesis. In: Pigliucci M, Müller GB (eds) Evolution: the
extended synthesis. MIT Press, Cambridge
Polis GA (1981) The evolution and dynamics of intraspecific predation.
Annu Rev Ecol Syst 12:225–251. https ://doi.org/10.1146/annur
ev.es.12.11018 1.00130 1
Rappole JH (2013) The avian migrant: the biology of bird migration.
Columbia University Press, New York
Ridley M (2004) Evolution, 3rd edn. Blackwell Publishers, Malden
Ruben J (1995) The evolution of endothermy in mammals and birds:
from physiology to fossils. Annu Rev Physiol 57:69–95. https ://
doi.org/10.1146/annur ev.ph.57.03019 5.00044 1
Salmon WC (1984) Scientific explanation and the causal structure of
the world. Princeton University Press, Princeton
Schirrmeister BE, Antonelli A, Bagheri HC (2011) The origin of mul-
ticellularity in cyanobacteria. BMC Evol Biol 11:45. https ://doi.
org/10.1186/1471-2148-11-45
Schmalhausen II (1949) Factors of evolution: the theory of stabilizing
selection. Blakiston Company, Philadelphia
Sol D, Duncan RP, Blackburn TM, Cassey P, Lefebvre L (2005) Big
brains, enhanced cognition, and response of birds to novel envi-
ronments. Proc Natl Acad Sci USA 102:5460–5465. https ://doi.
org/10.1073/pnas.04081 45102
Teaford MF, Ungar PS (2000) Diet and the evolution of the earliest
human ancestors. Proc Natl Acad Sci USA 97:13506–13511. https
://doi.org/10.1073/pnas.26036 8897
Toman J, Flegr J (2018) General environmental heterogeneity as the
explanation of sexuality? Comparative study shows that ancient
asexual taxa are associated with both biotically and abiotically
homogeneous environments. Ecol Evol 8:973–991. https ://doi.
org/10.1002/ece3.3716
Ulijaszek SJ (2002) Human eating behaviour in an evolutionary ecolog-
ical context. Proc Nutr Soc 61:517–526. https ://doi.org/10.1079/
PNS20 02180
van Duijn M, Keijzer F, Franken D (2006) Principles of minimal cogni-
tion: casting cognition as sensorimotor coordination. Adapt Behav
14:157–170. https ://doi.org/10.1177/10597 12306 01400 207
Van Fraassen BC (1980) The scientific image. Oxford University Press,
Oxford
Wei Y, Wang X, Liu J, Nememan I, Singh AH, Weiss H, Levin BR
(2011) The population dynamics of bacteria in physically struc-
tured habitats and the adaptive virtue of random motility. Proc
Natl Acad Sci USA 108:4047–4052. https ://doi.org/10.1073/
pnas.10134 99108
Williams GC (1966) Adaptation and natural selection: a critique of
some current evolutionary thought. Princeton University Press,
Princeton
Woods HA, Wilson JK (2013) An information hypothesis for the evo-
lution of homeostasis. Trends Ecol Evol 28:283–289. https ://doi.
org/10.1016/j.tree.2012.10.021
Publisher’s Note Springer Nature remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.