ArticlePDF Available


Theories that engender fundamental transformations in our world view seldom come perfect from the outset for two reasons. First, the empirical discoveries and theoretical framework necessary for their full explanatory efficacy are often not yet in place. Secondly, as a consequence of the first, some of the auxiliary theories and assumptions they rely upon are often antiquated and erroneous. For these reasons, anomalies are frequent in scientific theories. In this thesis, I discuss some of the major scientific anomalies, including particularly, the paradox of altruism. I suggest that the paradox of altruism arises because one of the most fundamental Mendelian genetic principles is misapplied. I show that today’s explanatory models err in supposing altruism and selfishness to be genetic allelomorphs. The supposition is inconsistent with the field data on altruism, and entails a logical inconsistency in accounting for the evolution of altruism. Largely, the models that purport to resolve the paradox hinge on the conditional expression of the altruistic gene, a move which I argue contradicts the theoretical assumption that engenders the paradox in the first place. I demonstrate from the empirical data that altruism and selfishness are rather plastic phenotypic expressions of a single genotype. And by supplanting the standard neo-Darwinian assumptions with the principle of phenotypic plasticity, I provide a parsimonious account of the evolution and maintenance of altruism which entails no paradox.
The Altruism Paradox: A Consequence of Mistaken Genetic
Yussif Yakubu
Received: 15 February 2013 / Accepted: 4 March 2013
Konrad Lorenz Institute for Evolution and Cognition Research 2013
Abstract The theoretical heuristic of assuming distinct
alleles (or genotypes) for alternative phenotypes is the
foundation of the paradigm of evolutionary explanation we
call the Modern Synthesis. In modeling the evolution of
sociality, the heuristic has been to set altruism and self-
ishness as alternative phenotypes under distinct genotypes,
which has been dubbed the ‘‘phenotypic gambit.’’ The
prevalence of the altruistic genotype that is of lower evo-
lutionary fitness relative to the alternative genotype for
non-altruistic behavior in populations is the basis of the
‘paradox of altruism.’’ I show in this article that the
assumption of contrasting genotypes for altruism and
selfishness in our ‘‘phenotypic gambit’’ is inconsistent with
the empirical data when viewed in the light of today’s post-
Mendelian understanding of gene expression. I demon-
strate that however nuanced and sophisticated the models
may have become today, they are still rooted in that fun-
damentally problematic assumption. I then offer a genetic
conception of altruism that best fits the field data.
Keywords Altruism Phenotypic gambit Phenotypic
plasticity Social evolution
Alger and Weibull (2012, p. 42) have suggested that ‘‘when
evolution operates at the level of behavior rules rather than
directly on acts, as is usually assumed, the level of coop-
eration generally violates Hamilton’s rule at the behavioral
level.’’ By ‘‘evolution operating directly on acts’’ they mean
‘when a trait is an action always to be taken,’’ as opposed to
the situation in which the trait expresses one behavior or
another contingent upon some exogenous factor. My argu-
ment in this article is that, whereas our current models
generally fall under the former category, as Alger and
Weibull (2012) concur, the empirical data overwhelmingly
suggest the latter to be the real state of affairs in nature. In
their recent paper, ‘‘The Evolution of Eusociality,’’ Nowak
et al. (2010) proposed a model consistent with the latter
conception, and were severely critical of Hamilton’s theory.
In the rancorous debate that ensued, van Veelen et al. (2010)
urged a return to basics, rigor in analysis, and empirical
testing of theories and assumptions. We see all the more
reason why, in the following discussion.
The ‘‘paradox of altruism’’ is arguably the most endur-
ing riddle in evolutionary biology. In extant conceptuali-
zation, it fits what William (1981, p. 164) describes as ‘‘the
classic problem of a mechanism by which a behavior
can evolve (genetically) even though it lowers the fitness of
the individual engaging in this behavior.’’ But why is that a
problem? Darwin’s ([1859] 1998) theory of natural selec-
tion predicts, via mathematical models pursuant to the
Modern Synthesis, that a fitter (genetic) trait would
increase in frequency in a population while the less fit
alternative would be eliminated. Thus, the paradox arises
because we suppose distinct genetic factors for altruism
and selfishness in a situation where the less fit ‘‘altruistic
allele’’ is persistent, contrary to the Darwinian theoretical
prediction that it should perish. It is none other than this
conception that gives occasion to theories such as kin
selection, group selection, evolutionarily stable strategies
(ESS), etc., all of which came about on account of the
perceived paradox of a flourishing maladaptive altruistic
trait. In this article I point out that the most fundamental
Y. Yakubu (&)
Department of Philosophy, McMaster University, Hamilton,
ON, Canada
Biological Theory. Vol 7, November 2017
DOI 10.1007/s13752-013-0120-4
Author's personal copy
The final publication is available at
assumption under which we model the evolution of altru-
ism is itself problematic; correcting it could dissipate the
paradox and consequently obviate the need for special
mechanisms for the evolution of altruism.
The Phenotypic Gambit
Grafen (1984) has noted that our modeling of the evolution
of behavioral phenotypes involves a ‘‘phenotypic gam-
bit’’—that there is a haploid locus at which each distinct
phenotype is represented by a distinct allele. Grafen raises
some concerns as well as some hopes about the phenotypic
gambit. He explains that due to the dearth of our knowl-
edge regarding the exact relationships between genotypes
and phenotypes of behavioral traits, building population-
genetic models for such traits would be quite impossible
without making such a gambit. Grafen indicates however,
that ‘‘taken literally the gambit is usually false’’ (1984,
p. 64), and that ‘‘it is a leap of faith’’ we take (p. 65).
However, the phenotypic gambit as I see it is likely false
only if we think of it as a phenotypic ‘‘gamble,’’ as Gra-
fen’s discussion seems to suggest. If under our phenotypic
gambit we want to imagine the single haploid locus as
representing the totality of a complex behavior such as
altruism, then it is most certainly false. Grafen (1984)
explains, however, that we nevertheless need to concep-
tualize it in that manner on grounds of theoretical expe-
dience. I think not, for the following reason.
Darwin ([1859] 1998, pp. 112, 160) describes evolution
by natural selection as happening through a cumulative
series of small, imperceptible changes. If every one of
these small changes has to be heritable, as Darwin insists,
then they will each have to be inscribed by some heritable
factor, which in today’s understanding will be a single
mutation or nucleotide substitution. When we adopted
Mendelian particulate genetics under the Modern Synthe-
sis, as opposed to blending inheritance, we essentially
quantized heredity, and consequently, the evolutionary
process. Thus, our phenotypic gambit should be seen as a
‘zoom-in’’ on just a single one of the hundreds or perhaps
thousands of discrete but imperceptible heritable steps by
which a trait evolves. Each discrete step should coincide
with a single nucleotide substitution. Think of this evolu-
tionary step as one of the series of infinitesimal changes
that we imagine in calculus to collectively constitute a
curve. In trying to understand how a curve of any shape
comes about, we isolate one of the infinitesimal steps that
make up the curve and describe it to represent the manner
in which the curve is formed. Similarly, an evolutionary
model describes the manner of evolutionary change, and it
necessarily has to describe the character of the steps that
constitute the change.
The models in which we make the phenotypic gambit
therefore describe the conditions under which such an
evolutionary step in a particular direction might occur. Our
models should be conceived as suggesting, consistent with
Darwin ([1859] 1998, pp. 112–113), that if such conditions
persist, more of such steps would occur, and the cumulative
effect would be a visible phenotypic change. Viewed in
this sense—i.e., a phenotype as a collection of advanta-
geous heritable variations, one of which our evolutionary
model describes—rather than the totality of the phenotype,
our phenotypic gambit would more likely be true than
false. In fact, any trait whose evolution cannot be simpli-
fied into this format is likely non-Darwinian. Therefore,
when we make our phenotypic gambit that depicts a single
haploid locus, we are actually dealing with just a ‘‘quantum
bit’’ of evolutionary change, a long series of which ulti-
mately yields the phenotype whose evolution we are
Unfortunately, our population-genetic models are often
not clear on this point; in fact, they are generally mis-
leading. Evolution by natural selection is actually a two-
dimensional progression. The first dimension is the ‘‘cal-
culus analogy’’ I have discussed above. It is the gradual
accumulation of favorable heritable variations (Darwin
[1859] 1998, pp. 137, 141), which leads to the intensifi-
cation, magnification, actualization or ultimate manifesta-
tion of the trait. The other dimension of evolutionary
change is the spread or penetration of the advantageous
variation in the population, which is the gradual sup-
planting of the individuals without the favorable variation
by those who have that variation (p. 154), leading to the
ultimate fixation of the trait. However, when we define and
model evolution under the Modern Synthesis simply as a
change in gene frequencies, where the genes/alleles are
assumed to represent complete behavioral phenotypes/
strategies, we reduce a two-dimensional progression into
one. For example, we generally purport to model the
evolution of cooperation by describing the conditions
under which one full-fledged and well-formed complex
strategy may successfully invade another. The models thus
collapse the numerous steps involved in the intensification
process into a single step or a single mutation and then
describe how it spreads though the population to eventual
fixation. As Grafen (1984, p. 64) puts it, we proceed ‘‘as if
enough mutation occurred to allow each strategy to
However, Darwin ([1859] 1998, p. 160) placed the
essence of natural selection in the intensification process,
saying that ‘‘natural selection acts exclusively by the
preservation and accumulation of variations which are
beneficial.’’ He adds further that ‘‘it seems as improbable
that any part should have been suddenly produced perfect’
(p. 66), and that we should expect ‘‘monstrosities’’ (sudden
Y. Yakubu
Author's personal copy
major deviations in structure) to be rare (p. 121). This is
why any evolutionary model that supposes an entire com-
plex trait to be represented by a single haploid locus is
quite un-Darwinian. It would be as if the sequential
nucleotide substitutions that occurred at numerous loci all
happened in one flash to create the phenotype. Grafen
(1984) is thus right in saying that the phenotypic gambit is
likely false, but I do not agree with him that it is only by
such a false simplification (what he calls a caricature) that
we are able to model the evolution of behavior by natural
selection. As the discussion above has shown, this partic-
ular problem is fixed if we re-cast our conception of the
phenotypic gambit according to the ‘‘calculus analogy’’ I
have described above.
The Phenotypic Gambit and Altruism
An additional difficulty arises from the way we frame the
phenotypic gambit for modeling the evolution of sociality.
It is abundantly clear from the literature (see the next
section) that in modeling the evolution of altruism, most
models take selfishness to be the alternative phenotype
with which the altruistic phenotype vies for evolutionary
ascendency. As a consequence, our ‘‘phenotypic gambit’’
has been to assume corresponding competing alleles for the
two phenotypes. This is another problematic aspect of our
modeling assumption, as it is starkly contrary to the field
data on altruism, properly interpreted in the light of today’s
understanding of gene expression. There have been con-
cerns raised over this state of affairs (West-Eberhard 2003).
In fact, the empirical data presented here will show
consistently that the altruistic and selfish phenotypes are
plastic expressions of a single genotype under alternative
environmental circumstances. If the altruistic and selfish
phenotypes are thus tied to a single genotype, they cannot
be competing alternatives in populations as today’s models
cast them. Had we recognized this empirical fact as the
fundamental genetic principle in altruism, and had we had
that empirical reality reflected in our genetic models rather
than the contrasting allele assumption, the theoretical
landscape of the evolution of altruism would look much
different, and certainly much less complicated and prob-
lematic than it is today.
Some behavioral ecologists tend to be quite sensitive to
discussions of genes and behavior because of the associated
issue of genetic determinism. However, supposing separate
genes for two contrasting phenotypes does not necessarily
entail genetic determinism (Okasha 2009). It is one thing to
acknowledge that we make an assumption that is ‘‘literally
false,’’ and defend it on grounds of theoretical expedience
(Grafen 1984), or explain what it should not be miscon-
strued for (Okasha 2009). It is another to deny altogether,
as some are inclined to do, that our standard models make
such an assumption, or to charge that this kind of criticism
sets up a ‘‘straw man.’’ People who hold this latter view
contradict the preeminent scholars on this subject, most of
whom will be quoted below as clearly espousing the con-
trasting allele view for altruism and selfishness. In fact, the
next section is devoted to proving the prevalence of the
genotypic dichotomy assumption, which would not have
been necessary here, were it not for the tendency in some
quarters to deny that our standard evolutionary models
hinge on such an assumption, or to suggest that there are
‘nuances’’ in the assumption that make it reasonable.
Extant Genetic View
In his review of Sober and Wilson (1998), Maynard-Smith
(1998, p. 639) notes that ‘‘there are two kinds of individ-
ual[s]: altruists, who benefit others at a cost to themselves;
and non-altruists who do not. A field example of these two
distinct individuals is given by Okasha (2009), who writes:
‘To see this, imagine that some members of a group of
Vervet monkeys give alarm calls when they see predators,
but others do not.’’ In this example, those individuals who
call the alarm are the altruists, and those who do not call the
alarm are selfish. This binary conception of phenotype as
well as genotype (Van Veelen et al. 2012, p. 68) is standard
in the modeling of altruism. The preferred terms in game-
theoretic models are ‘‘cooperators’’ and ‘‘defectors.’
The critical question is: what is the genetic relationship
between such altruistic and selfish individuals (or strate-
gies) in a population? Is it as described by William (1981):
‘differences among phenotypes are causally associated
with genotypic differences (in other words) genetic dif-
ferences underlie phenotypic differences’’? This is one of
the least explored empirical and theoretical questions of
evolutionary research. Grafen (1984, p. 65), suggests that
even though the behavioral ecologist relies on population
genetics, ‘‘our method is designed to avoid doing genet-
ics.’’ Hence, according to him, the behavioral ecologist
takes a leap of faith, and goes with a ‘‘phenotypic gambit’
that there is an allele for one phenotype and a contrasting
allele for the other. Knowing the exact genetic details may
indeed be impossible, but we should not altogether ignore
even the low-hanging fruits of empirical research that
suggest what the broad genetic relationships may be. Our
phenotypic gambit becomes a true ‘‘caricature of reality’’
(Grafen 1984; Gardner et al. 2011) only if it contains
assumptions that are contrary to the empirical facts at hand.
It is the source of what West-Eberhard (2003,p.3)
observed as the ‘‘gap between the conclusions of the
genetical theory of the origin and spread of a new trait and
the observed nature of the trait being explained.’’ The
The Altruism Paradox
Author's personal copy
gambit of assuming contrasting alleles for contrasting
phenotypes as applied to altruism implies that the altruist
and non-altruist are distinguishable genetically by the
possession (or lack thereof), of ‘‘a gene for altruism.’’ In
other words, we assume a genetic basis for the contrasting
behaviors of altruism and selfishness.
In explaining kin selection, Okasha (2009), for one,
invites us to ‘‘imagine a gene which causes its bearer to
behave altruistically towards other organisms.’’ He sug-
gests that ‘‘organisms without the gene are selfish,’’ and
then goes on to say that ‘‘the altruists will be at a fitness
disadvantage, so we should expect the altruistic gene to be
eliminated from the population.’’ This renders most con-
cisely the problem of altruism as conceived by evolution-
ary biologists today, and is posted under ‘‘Biological
Altruism’’ in the Stanford Encyclopedia of Philosophy.
That is, an altruistic gene that codes for the weaker altru-
istic phenotype thrives in competition against a selfish gene
that codes for the fitter selfish phenotype—hence, a para-
dox. It would be hard to imagine any nuance or semantic
spin that could be put on such statements to mean anything
other than that we suppose a gene for altruism and a sep-
arate gene for selfishness. I am aware of no other basis ever
adduced for the paradox of altruism. From the angle of
group selection, Wilson and Wilson (2007, p. 329) render
the problem exactly the same way, as they write: ‘‘a heri-
table trait that increases the fitness of others in a group (or
the group as a whole) at the expense of the individual
possessing the trait will decline in frequency within the
Okasha (2009) vehemently denies that ‘‘genetic deter-
minism’’ is entailed in this kind of genetic supposition in
evolutionary models. He writes: ‘‘Kin selection theory does
not deny the truism that all traits are affected by both genes
and environment.’’ I make no charge of the kind of genetic
determinism Okasha denies. In fact, it is true that some
models do offer a scenario in which ‘‘having the cooper-
ative genotype only implies a certain probability of
expressing it’’ (van Veelen et al. 2012, p. 68). However,
such models (often called conditional altruism models) do
not set this scenario as something contrary to the con-
trasting genotypes view. In fact, they often do so while at
the same time maintaining separate genotypes for the two
phenotypes, which is actually my worry.
Consider, for example, Trivers’ (1971, p. 36) reciprocal
altruism model, which hinges upon conditional expression
of the altruistic genotype. Yet, he sets the following as the
genetic assumption in the model: ‘‘Assume that the altru-
istic behavior of an altruist is controlled by an allele
(dominant or recessive), a
, at a given locus and that (for
simplicity) there is only one alternative allele, a
, at that
locus and that it does not lead to altruistic behavior.’’
(Other examples of models that claim an altruistic gene
that is non-deterministic but distinct from the selfish gene
include Alger and Weibull 2012; Hamilton 1964; Queller
1985; van Veelen et al. 2012.)
Those models aside, straightforward declarations of the
‘contrasting genotypes’’ supposition are also ubiquitous in
the literature. On how kin selection explains altruism,
Curry (2006, p. 683) writes: ‘‘Well, genes for altruism can
spread if they help copies of themselves that reside in other
individuals.’’ Bowles (2006, p. 1569) supposes in his group
selection model that ‘‘(A) individuals are bearers of a
hypothetical ‘‘altruistic allele’’; those without the allele
(Ns) do not behave altruistically.’’ In Haldane’s (1932,
p. 208) model, aa is the recessive character that causes
altruistic behavior. Similarly, Rousset and Roze (2007,
p. 2321) engage in a very elaborate mathematical analysis
of the possible evolutionary outcome of a ‘‘helping allele
(H0)’’ versus a cheating allele (H1).’’ Sober (1984, p. 184)
supposes an altruistic trait ‘‘A–one that causes individuals
with the trait to benefit others at their own expense.’’ In one
of the most recent papers on the subject, Gardner et al.
(2011, pp. 1029–1030) give one of the most sophisticated
analyses of the major altruistic models to date; in setting up
the assumptions for the analysis, they write:
We assume an infinite population of haploid indi-
viduals engaged in two-player games. A single locus
controls the cooperation phenotype, with a proportion
p of individuals carrying an allele A which encodes
the cooperator strategy, and the remaining 1 -p
carrying an allele a, which encodes the non-cooper-
ator strategy.
This common supposition of altruism and selfishness as
genetic allelomorphs in extant models emanates from the
fundamental population-genetic template for modeling
evolution through gene frequency changes as shown in
Table 1(from Halliburton 2004, p. 133).
In modeling the evolution of altruism, extant models
build upon this template, in which they usually assume
altruism and selfishness to be the contrasting alleles A
—the phenotypic gambit. This is why I have referred to
this approach by extant evolutionary models of altruism as
the Altruism Selfishness Allelomorphism (ASA) models.
The alternative I propose is what the empirical data sug-
gest, i.e., altruism and selfishness as alternative phenotypes
of a single plastic genotype, hence, the Altruism Selfish-
ness Plasticity (ASP) model. In the latter, we cannot
Table 1 Allele frequencies chart (from Halliburton 2004, p. 133)
Genotype A
Frequency P
2pq q
Fitness w
Y. Yakubu
Author's personal copy
represent altruism and selfishness separately as A
and A
in Table 1, since they are of one and the same genotype, as
the empirical data suggest. The consequence of ASA is that
the distinct ‘‘altruistic allele’’ would be of lower evolu-
tionary fitness and therefore ought to decline in frequency.
This was what led Hamilton (1964, p. 16), who clearly held
that ASA conception of altruism, to declare that the
existing mathematical models did not allow for the evo-
lution of sociality (p. 1), and consequently some mecha-
nism was needed that would offset the apparent decline of
the altruism allele.
Like Hamilton and his inclusive fitness hypothesis, most
of our extant models of altruism (including those which
cautiously talk only of phenotypes) are driven by the need
to provide some countervailing mechanism to this theo-
retically predicted attenuation of altruistic allele frequen-
cies in populations. From the group-selectionist camp,
Wilson and Wilson (2007, p. 329) assert that ‘‘something
more than natural selection within single groups is required
to explain how altruism and other group-advantageous
traits evolve by natural selection’’; the group selection
answer is that the ‘‘within-group’’ disadvantage of the
altruist is counteracted by the between-group advantage of
the group with altruists. In the parlance of game-theoretic/
ESS models, we may read statements such as this from
Taylor and Nowak (2007, p. 2281) that ‘‘Cooperation is
always vulnerable to exploitation by defectors; hence, the
evolution of cooperation requires specific mechanisms,
which allow natural selection to favor cooperation over
defection’’ (see also Allen et al. 2012 and Nowak 2012).
This is exactly the problem Hamilton (1964) pointed out
five decades earlier, and such statements are driven by the
thought of a distinct altruistic allele (gene/genotype) that is
in danger of being overrun by a distinct selfish allele. In
Dawkins’ (1976/1989, p. 184) metaphor, ‘‘cheat genes’’ are
spreading through the population while ‘‘sucker genes’’ are
driven to extinction. Wilson (2005, p. 159) summarizes the
problem thus: ‘‘How might such a behavior evolve if the
genes promoting it are at such a disadvantage in competi-
tion with genes that oppose it?’
Nevertheless, the less-fit altruistic allele is prevalent in
natural populations, and this presents an anomaly to
explain. Suggested mechanisms of how this could have
come about include: (1) altruists associating exclusively
with other altruists (Maynard-Smith 1998; Sober and
Wilson 1998); and (2) conditional deployment of the
altruistic behavior, i.e., only towards genetic relatives
(Hamilton 1964), or towards other altruists (Trivers 1971).
Dawkins (1976/1989, p. 89) simplifies the concept for a
popular science audience with the ‘‘green beard’’ metaphor,
in which we are to imagine altruists identifying other
altruists by a characteristic green beard. For group selection
models, Godfrey-Smith (2009, p. 174) explains that if
social groups are formed randomly, ‘‘the A(altruistic) type
is lost regardless of the details.’’ However, the altruist can
be maintained, he explains further, ‘‘if groups are formed in
a way that ‘clumps’ the two types, so like tends to interact
with like [and] the benefits of having ‘As’ around tend to
fall mainly on other As’’ (p. 174). For ESS game-theoretic
models, Burton et al. (2012, p. 55) state: ‘‘one general
answer is that interactions need to be assortative, so that
individuals carrying genes coding for cooperation interact,
on average, more often with cooperating individuals than
individuals carrying genes coding for defection.’’ Simi-
larly, ‘‘clustering’’ is called for in the latest ESS models
using ‘‘evolutionary graph theory’’ so that ‘‘the benefits of
cooperation are received mostly by other cooperators’’
(Allen et al. 2012 and references therein). If these are the
‘nuances’’ suggested in the latest models, they do not
renounce the contrasting genotypes assumption. Rather,
they suggest how cooperation may still occur under that
All the safeguards that are sought by extant (ASA based)
models with a view to exclude the selfish individuals in the
population from benefiting from altruistic acts stem from
the supposition that the two phenotypes are of rival geno-
types. However, all these exclusionary mechanisms would
not be necessary if such models had a genetic conception of
altruism that is consistent with the field data, in which the
selfish individuals also carry and transmit the altruistic
gene, as the ASP model will show to be the case. This is
why we cannot take the matter of the genetic relationship
between the contrasting phenotypes in such populations
lightly. For if it is indeed the case that such phenotypes are
plastic expressions of a single genotype, then the exclu-
sionary mechanisms which have been the focus of the
evolutionary modeling of altruism in the last five decades
are actually not necessary.
If we subscribe to phenotypic plasticity as the genetic
basis of altruism and selfishness, then the phenotypically
selfish individuals in the population can also transmit the
altruistic gene and need not be excluded from receiving
altruistic acts. This is the case with the selfish queens in
eusocial organisms, as we shall see later. As I shall also
show later, even in non-eusocial altruistic societies, such as
human, baboon, or ground squirrel societies, selfish indi-
viduals have the capacity to reproduce altruistic individu-
als, and can even perform altruism themselves given the
right social circumstances. Conditional altruism models do
not see it this way, because they generally consider altru-
istic and selfish individuals to be genotypically different.
It is therefore very critical to evolutionary modeling to
ascertain whether it is indeed by virtue of a genotypic
difference that some individuals in a population behave
altruistically while others behave selfishly. If we say the
worker bee is altruistic and the queen bee is selfish; if a
The Altruism Paradox
Author's personal copy
ground squirrel that sounds the alarm is altruistic and the
adjacent squirrel that does not is selfish; and if also by
altruism we mean the vampire bat that shares blood with a
roost mate, while the bat that refuses to share we call
selfish; if these are the criteria by which we designate
altruists and non-altruists in populations—and in fact, these
are some of the most compelling cases of the altruism we
model—then it would be quite wrong to adopt a modeling
assumption that there is an allele for altruism and a con-
trasting one for selfishness. For as we shall see below, there
is enough in the field data to establish that the altruistic and
selfish traits are of the same genotype rather than separate.
Our ‘‘phenotypic gambit’’ in modeling the evolution of
altruism right now is blatantly contrary to this empirical
reality. In the sections that follow, I present an analysis of
the empirical data, which shows that the altruistic and
selfish phenotypes are not of separate genotypes.
Genes and Environment
The interplay of genes and the environment in shaping
phenotypes is not new in science today. We now know also
that there are many genes that would not express pheno-
typically unless triggered by some environmental cue. In
such cases, two individuals carrying the same gene could
nevertheless differ phenotypically with respect to that
genotype due to differences in their environmental expe-
riences. Figure 1(from Agrawal 2001) shows two clones of
the water flea Daphnia lumholtzi. The individual on the left
with the spiny helmet and longer tail spine was raised in an
environment in which chemical cues from a predacious fish
were introduced. The other clone (on the right) was the
control. The experiment demonstrates that any individual
from this species can assume either phenotype depending
on whether it is growing in an environment with predators
or in one without predators. Hence, the two phenotypes,
even though once thought to be separate species, actually
do not differ with respect to the genotype for helmet.
The property of a given genotype to produce different
phenotypes in response to distinct environmental condi-
tions defines phenotypic plasticity (Pigliucci 2001, p. 1). In
the case of Daphnia it is helmets that are expressed in
response to the presence of a predatory fish. There are
many other examples of phenotypic features that are
expressed only under certain environmental cues. Other
such cues include: parasites (Moore 1995), diet (Greene
1989; Pfennig and Murphy 2000); predators (Lively 1986;
Agrawal 2001); competition (Harvey et al. 2000); popu-
lation density (Deno and Roderick 1992); temperature
(Morreale et al. 1982; Roff 1986). In all of these examples,
as in Daphnia, a single genotype expresses one phenotype
or another, depending on the presence or absence of spe-
cific environmental cues. Clearly, any evolutionary model
that assumes a genotypic difference between such dimor-
phic phenotypes would simply be incorrect. For example,
in Table 1it would be wrong to designate helmets as allele
and non-helmets as allele A
. It would also be funda-
mentally wrong to even conceive of the two phenotypes as
competing evolutionary alternatives. In the sections that
follow, I will try to persuade the reader that the field data
suggest that we model the evolution of the two phenotypes
of altruism and selfishness as we would do for helmets and
non-helmets in Daphnia.
How could we talk of a declining altruistic allele against
a fitter selfish allele in the polyembryonic wasp (Copido-
soma floridanun), for example, in which clones from a
single embryo differentiate into altruistic soldiers who do
not reproduce but defend the selfish ones who reproduce
(Donnell et al. 2004)? Other such cases of clones differ-
entiating into altruistic and selfish individuals have been
reported in gall aphids (Ito 1989; Abbot et al. 2001). It is
clear in these cases that altruism and selfishness are indeed
plastic expressions of a single genotype, since the two
phenotypes are expressed by different individuals of the
same clone. Should there be any inclination to think that
these are obscure anecdotal examples, consider some of the
best-known examples of altruism in the sections below.
The Hymenoptera/Daphnia Parallel
Let us start by examining the detailed empirical observa-
tions of altruism as expressed in the social hymenoptera. In
a honey bee colony, for example, there are three castes
consisting of a queen who does nothing but reproduce; a
few hundred males called drones who also do not do much
other than wait for an opportunity to mate with a queen;
Fig. 1 Clones of D. lumholtzi (from Agrawal 2001)
Y. Yakubu
Author's personal copy
and thousands of non-reproductive females called workers,
who toil all their lives taking care of the colony, including
foraging and possibly laying down their lives when this is
necessary in order to defend the colony. The reproductive
queen has been designated as selfish, while the non-
reproductive workers are traditionally viewed as the
altruists. In fact, Shanahan (2004) regards the behavior of
the worker castes of the social insects as the epitome of
So, are there distinct alleles/genotypes for altruism and
selfishness in eusocial populations as we suppose in our
models of altruism? It has been known and documented
since the 1830s that ‘‘a fertilized honeybee egg, which
would normally yield a worker bee, will give rise to a
queen bee if the ensuing larva is fed ‘royal jelly’’’ (Prete
1990, p. 273). Detailed modern studies have revealed fur-
ther that whether a bee larva is raised a queen or a worker
begins with the type of honeycomb cell into which the egg
is laid (Winston 1987). The workers will rear a larva as a
queen if it is in a queen cell, by feeding it royal jelly. On
the other hand, they will rear it as a worker if it is in a
worker cell, by feeding it worker food. The eggs and early
larval stages are totipotent (i.e., can develop into different
functional entities). According to Winston (1987, p. 66), an
egg or larva less than 3 days old that is moved from a
worker cell into a queen cell will be fed royal jelly by the
nursing workers and it will consequently develop into a
queen. Conversely, an egg or larva transferred from a
queen cell into a worker cell will be fed worker food and
will consequently develop into a worker. This is a very
powerful indication that there is no genetic basis for initial
placement of an egg in a queen cell or a worker cell, and
whether a bee becomes a reproductive (selfish) queen or a
non-reproductive worker (altruist) is determined by an
environmental stimulus (i.e., diet) rather than genotype.
We now know the specific genes in the honeybee whose
differential expression results in the selfish queen and the
altruistic workers (Evans and Wheeler 1999); they are
plastic genes that are common to both the selfish and
altruistic castes. Patel et al. (2007) have detailed the sig-
naling pathways by which different diet regimes activate or
depress generic genes to yield different honeybee castes. In
fact, inter-caste individuals (i.e., individuals with both
queen and worker features) have been artificially created
by the experimental manipulation of larval diet (Wilkinson
1984, p. 68). Therefore, the observation that each female
honeybee has the potential to develop into a queen or a
worker suggests both phenotypes are expressed by the
same genotype in response to different environmental
stimuli (e.g., diet regimes) rather than separate genotypes
coding for the two phenotypes.
In other examples, experimental studies indicate that in
the eusocial wasps (Vespidae) differences in nutrition
during larval development are often the basis of caste
determination (O’Donnell 1998). In other species of social
insects, it has been demonstrated that individuals can make
a transition between altruistic and selfish behavior through
experimental manipulation of their environments (Field
et al. 2006). Thus, the genetics of altruism in these cases is
very much like that of helmets and non-helmets in Daph-
nia, and therefore an unmistakable case of phenotypic
plasticity, which is defined as ‘‘the environmentally sen-
sitive production of alternative phenotypes by given
genotypes’’ (DeWitt and Scheiner 2004, p. 2). This should
be quite obvious in the social insects. West-Eberhard
(1986) lists the queen-worker dimorphism in the social
insects as one of the examples of alternative phenotypes
that are produced by genes borne by all individuals of the
population. Wilson (2008, p. 18) has also come to the
understanding that ‘‘the different roles of the reproductive
mother and her non-reproductive offspring are not geneti-
cally determined.’’ Rather, ‘‘as the evidence from primi-
tively eusocial species has shown, they represent different
phenotypes of the same recently modified genome.’’
A genetic switching mechanism triggers such alternative
phenotypes depending on the developmental stage or some
environmental stimulus. Unless one rejects this entire cat-
alogue of empirical data, it would be wrong to designate
altruism and selfishness as the allelomorphs A
and A
Table 1, as the ASA models do. Also, it is clear that, with
this kind of genetics, the altruistic behavior is at no fitness
disadvantage relative to the selfish behavior in the social
insects. If so, why are evolutionary biologists, as cited
above, jumping through hoops to provide circuitous
explanations as to how an altruistic gene of lower evolu-
tionary fitness can be sustainable against a selfish alterna-
tive, when that is totally not the issue?
It all stems, as I explained above, from the modeling
assumption that is inherent in our population genetic
models of evolution. Consider the genetic assumptions in
the most widely accepted explanations of altruism today,
which emanated from the icons of evolutionary biology in
the 20th century. Maynard-Smith (1964), like Haldane
(1932, p. 208) and Trivers (1971) above, assumed altruism
to be caused by a Mendelian recessive character aa as
opposed to the characters AA and Aa for the non-altruistic
condition. Hamilton (1963, p. 354) supposes ‘‘a pair of
genes g and G such that G tends to cause some kind of
altruistic behavior and g is null. He also points out (1964,
p. 16) that in modeling his inclusive fitness hypothesis, he
imagined ‘‘model organisms, whose (altruistic) behavior is
determined strictly by genotype.’’ These seminal works
clearly assume a genotypic dichotomy between the altru-
istic and selfish phenotypes, which in turn precipitates the
concern over a declining altruistic allele and consequently
the paradox of altruism and a scramble for explanations
The Altruism Paradox
Author's personal copy
and auxiliary hypotheses. From these scholars emanated
inclusive fitness, kin selection, reciprocal altruism, and the
ESS models. Of course, we have seen the same concern
raised by the proponents of group selection cited above.
These models have gotten ever more complex over the
years, but the underlying genetic assumption remains.
Altruistic Expression and Social Cues
In the above section, honeybee society was used to dem-
onstrate how the altruistic and selfish phenotypes in the
social insects are determined by different environmental
cues rather than genetic differences. Now I turn to the non-
eusocial social organisms. Starting with vampire bats, let
us examine the relative efficacies of the ASA and ASP
models in explaining reciprocal altruism, for which Trivers
(1971) provided an explanation based on the ASA
assumption. Vampire bats roost in dark places by day and
go out at night to feed. For some species the diet is
exclusively blood, usually from other mammals. There are
occasions when some individuals will find very little to eat
while others will be more fortunate. Researchers have
observed that the hungry individuals would often solicit
some food from the individuals that are better fed. Some-
times an individual would oblige and regurgitate some
blood to a soliciting individual, while on other occasions
individuals have also been observed to steadfastly refuse to
share food with a soliciting individual. It is traditionally
held that those individuals observed to obligingly share
their food with soliciting individuals are altruistic while
those that refuse to share are selfish. Then under the
altruism/selfishness allelomorphism (ASA) models we
have to assume that the individuals that share blood carry
the altruistic allele whilst those that refuse to share are
under the influence of the selfish allele.
Closer observations reveal, however, that whether a
vampire bat shares blood or not in any situation would be
determined largely by the circumstances at the time, such as
whether the solicitor has given the actor blood before
(Wilkinson 1984), or whether the solicitor is judged likely to
give blood to the donor when he is in need. If so, we could
suppose that the vampire bat that is seen today sharing blood
with a neighbor and judged to be doing so under the
expression of an ‘‘altruistic allele,’’ could on another occa-
sion be seen steadfastly refusing to give blood to a bat that is
starving, possibly because the then solicitor may have
refused to share previously. Hence, the bat that is charac-
terized as the altruist today would be the selfish individual
on some other occasion. Since organisms are not known to
change genotypes in that manner, the difference between
sharing then (altruism) and refusing to share now (selfish-
ness) is not a matter of genes but largely the circumstances
of the (social) environment. Thus the social environment,
like chemical cues in Daphnia and diet in the honeybee,
serves as a cue for the conditional expression of altruism and
selfishness as dimorphic behavioral phenotypes.
Among Belding’s ground squirrels, mostly adult females
make alarm calls, and the frequency of the calls has been
observed to correlate with the presence of relatives (Sher-
man 1977). Thus, the alarm-calling behavior seems to be
conditional, depending upon the presence of relatives. That
is exactly what Hamilton (1964) suggested would enhance
inclusive fitness, and many would celebrate this as a tri-
umph for kin selection. Let me just point out here before I
proceed with the current chain of thought, that evidence of
kin-motivated altruism here and there does not establish
kinship as necessary or sufficient for altruism. The point
with this example is that the fact indicating that the same
individual can behave altruistically (i.e., call the alarm) at
one instance and selfishly (i.e., refuse to call the alarm) at
another, undermines the underlying genetic assumption
upon which kin selection is established. Rather than two
separate genotypes causing the two phenotypes as the
architects of kin selection, Hamilton (1964) and Maynard-
Smith (1964), suppose, the behavior suggests a dimorphic
phenotypic expression of a single genotype. Hence, it is
consistent with the ASP model and contrary to the ASA
Among olive baboons (Papio anubis), Packer (1977)
reports that an adult male will give aid to a soliciting troupe
member based on whether he has received help from the
solicitor before or whether the solicitor is deemed capable
of giving meaningful help when it is needed. Thus an adult
male is more likely to deny aid to soliciting juveniles and
females during fights. It has similarly been reported in
vervet monkeys (Cercopithecus aethiops) that whether an
individual responds to a solicitation or not depends on
whether it has previously received grooming (or aid) from
the solicitor, in addition to other social considerations
(Seyfarth and Cheney 1984). If giving aid is ‘‘altruism’
and refusing to help is ‘‘selfishness,’’ then it is evident here
that external factors, rather than genotype, determine
whether an individual behaves altruistically or selfishly.
A pattern thus seems to emerge from the key examples
of altruism analyzed here: that an individual will respond
altruistically only when certain environmental circum-
stances are present, and would respond selfishly if those
environmental cues were lacking. It is no different from the
arctic fox expressing white fur in the winter and brown in
the summer. It is important to note that no evidence has yet
been presented to date that demonstrates that under the
same set of environmental circumstances only certain
individuals (i.e., those who carry the altruistic allele) are
capable of reacting altruistically, while others will always
refuse to assist because they lack the altruistic gene. In
Y. Yakubu
Author's personal copy
other words, there is no empirical evidence of such an
altruistic allele that serves as the underlying distinction
between the altruistic and selfish phenotypes. To prove
genotypic dichotomy we need to demonstrate, for example,
that certain honeybees will always mature into workers
(altruists) irrespective of diet or any other external factor;
that certain members of a vampire bat colony will always
share food even when the solicitor is one who has con-
sistently refused to share; that from one external circum-
stance to another, only certain individuals will consistently
call the alarm while others would never call the alarm
under any circumstance. The ASA models presume these
tests to be met. In reality there is no basis for such a
What is clear and consistent from the studies cited here
is the association between certain environmental cues and
the expression of the altruistic phenotype, while other
circumstances trigger a selfish response. For example, a
baboon gives aid (altruistic) under one circumstance and
denies aid (selfish) under another. That strongly suggests a
plastic behavioral response of a single genotype to different
(social) environmental circumstances. Thus, as in the
Daphnia example, there cannot be separate genotypes for
altruism and selfishness, since each individual in the pop-
ulation has the capacity to express both phenotypes.
Reproductive Altruism and the Social Environment
Reduced fecundity in deference to others has often been
cited among the examples of altruistic expression (Shana-
han 2004; Okasha 2009). Reproductive suppression (or
even exclusion) of subordinate females and males is a
common feature of animal social organizations. In these
cases of altruism it becomes ever more preposterous to
imagine that a genotypic difference could be causing the
behavioral difference between the individuals who repro-
duce and those who do not. Observations indicate very
strongly that in such societies the ‘‘altruistic’’ behavior is
imposed by external circumstances rather than by specific
genotypes. In social mammals for example, it is often the
dominant female or male that prevents the others from
breeding, through a variety of schemes, including physical
deterrence from mating. In the naked mole rat (Hetero-
cephalus glaber), pheromones given off by the dominant
female act on the hormonal systems of subordinate females
to render them infertile (Faulks et al. 1991). Those pher-
omones are analogous to chemical cues from predacious
fish in the case of Daphnia, and it is they rather than
genotype that elicit the non-reproductive altruistic behavior
in the mole rat. In meerkat societies the reproductive
efforts of subordinate females are deterred and disrupted by
the dominant female (Young and Clutton-Brock 2006). In
the case of helper birds, individuals are forced to assume
the non-reproductive (helper) position by external cir-
cumstances such as demography, rank, and availability of
nest cites (Rabenold 1985) rather than the dictates of some
‘altruistic gene’’ in the helper. Yet these are all frequently
cited examples of altruism, whose sustainability we are
confounded by as we attempt to explain it by making a
‘phenotypic gambit’’ that the two phenotypes are sup-
ported by contrasting genotypes.
In most social situations, it is where an individual ranks
in the social structure that determines whether it reproduces
or not. In hyena and wolf packs for example, only the alpha
male and female breed and the rest of the pack we must call
altruists. However, upon the death of the alpha female, as
observed in the naked mole rat (Heterocephalus glaber)by
Lacey and Sherman (1991), any of the non-reproductive
(altruistic) females can undergo some hormonal changes
and ascend to the role of the reproductive (selfish) female.
A similar observation has been made with the termite,
Zootermopsis nevadensis, in which a replacement is drawn
from amongst the workers upon the death of the king or
queen (Johns et al. 2009). This means a phenotypic trans-
formation of an altruistic worker into a selfish king or
queen. Recall the sex change behavior of the marine goby
Coryphopterus personatus (Allsop and West 2004) from
the literature of phenotypic plasticity. In this case also, the
same individual can be non-reproductive (altruistic) in one
social circumstance and become reproductive (selfish)
when the circumstances change. Such transitions between
the altruistic and selfish phenotypes by individuals belie the
assumption of an underlying genotypic dichotomy between
the phenotypes in current models.
Gadagkar (1997, p. 28) notes that a social organism
would assume a subordinate role not because of any
altruistic reasons but because it is the best of the available
alternatives. In the social wasp Ropalidia marginata, Ga-
dagkar (1997, p. 72) reports that individual wasps can act
as queens or workers in response to the opportunities
available. He observed further that often a worker would
later drive its mother (the queen) out and become the
queen. The change in status or phenotypic behavior from
worker to queen has also been reported in other social
insects (Field et al. 2006). As Queller (2006, p. 42)
observes in the eusocial insects, ‘‘workers are not leaping at
every opportunity to be altruistic, they are coerced.’
Coercion as a trigger of altruism, in the absence of which
an individual would rather remain selfish, is indicative of
the plastic phenotypic deployment of a common genotype.
Wenseleers and Ratnieks (2006) also conclude from studies
of ten social insect species that ‘‘it is mainly social sanc-
tions’’ that keep individuals altruistic where they would
otherwise have behaved selfishly. Emlen and Wrege (1992)
report that in the white-fronted bee-eater (Meropsis
The Altruism Paradox
Author's personal copy
bullockoides), young males are forced by older nest-own-
ing males into helper status by harassment and disruptions
of their attempts to set up their own nests. Such ‘‘altruistic’
helpers can change their status to ‘‘selfish’’ reproductive
nest owners whenever the opportunities arise in the future.
In fact, in meerkat societies, as Young and Clutton-Brock
(2006) report, not only are our designated altruists (the
subordinate females) able to express the selfish phenotype
by reproducing when they get the opportunity, they are
able to match the selfishness of the dominant female by
murdering the infants of other mothers.
In all these examples, it is remarkably consistent that the
altruistic and selfish phenotypes are determined by envi-
ronmental circumstances rather than genotype. One very
crucial observation is that individuals are often able to
make transitions between the two phenotypes in response
to changes in their social environment. These facts are
clearly, inconsistent with the notion of two separate
genotypes for altruistic and selfish individuals as assumed
by extant genetic models of altruism.
One possible objection, which space does not allow me
to discuss in detail here, is that some may think conditional
altruism extenuates this criticism. In the usual conception
of conditional altruism, an individual who is genetically an
‘altruist’’ is able to withhold altruistic behavior under
certain circumstances. But the models often do not char-
acterize such situations in which an altruist withholds
altruism as selfish behavior, i.e., they do not see the altruist
to be expressing the same phenotype as the selfish indi-
viduals. Also, they certainly do not think the genetically
selfish individuals exhibit conditional selfishness. So those
models still try to maintain two genotypically distinct
individuals in the population, which is contrary to the
empirical evidence.
Concluding Remarks
In the foregoing discussion, I have tried to bring the con-
ceptions and assumptions in our evolutionary modeling of
sociality closer to reality. No evolutionary biologist thinks
that a single mutation underlies a complex trait such as
altruism. Yet, we feel compelled to model it as if that were
the case. I have shown a way out of that. Second, we have
generally assumed in our evolutionary models that altruism
and selfishness are competing evolutionary alternatives
under distinct genotypes. This I have also shown to be
blatantly contrary to the empirical data. It is another
‘caricature of reality’’ we need to wean our models off. We
have an epistemic responsibility to keep our theoretical
assumptions consistent with well-confirmed empirical
evidence. Theoretical expediency should never supersede
the empirical evidence.
Acknowledgments I would like to thank Professor Richard Arthur
of the Department of Philosophy, McMaster University for reading
the manuscript and making some valuable suggestions.
Abbot P, Withgott JH, Moran NA (2001) Genetic conflict and
conditional altruism in social aphid colonies. Proc Natl Acad Sci
USA 98:12068–12071
Agrawal AA (2001) Phenotypic plasticity in the interactions and
evolution of species. Science 294:321–326
Alger I, Weibull JW (2012) A generalization of Hamilton’s rule-love
others how much? J Theor Biol 299:42–54
Allen B, Traulsen A, Tarnita CE, Nowak MA (2012) How mutation
affects evolutionary games on graphs. J Theor Biol 299:97–105
Allsop DJ, West SA (2004) Sex allocation in the sex-changing marine
goby (Coryphopterus personatus) on Atoll-fringing reefs. Evol
Ecol Res 6:843–855
Bowles S (2006) Group selection, reproductive leveling, and the
evolution of human altruism. Science 314:1569–1572
Burton S, Fletcher JA, Doebeli M (2012) Hamilton’s rule in multi-
level selection models. J Theor Biol 299:55–63
Curry O (2006) One good deed: can a simple equation explain the
development of altruism? Nature 444:683
Darwin C ([1859] 1998) The origin of species. Reprint: Modern
Library, New York
Dawkins R (1976/1989) The selfish gene. Oxford University Press,
Deno RF, Roderick GK (1992) Density related dispersal in planthop-
pers: effects of interspecific crowding. Ecology 73:1323–1334
DeWitt TJ, Scheiner SM (2004) Phenotypic variation from single
genotypes. In: DeWitt TJ, Scheiner SM (eds) Phenotypic
plasticity: functional and conceptual approaches. Oxford Uni-
versity Press, Oxford
Donnell DM, Corley LS, Chen G, Strand MR (2004) Caste
determination in a polyembryonic wasp involves inheritance of
germ cells. Proc Natl Acad Sci USA 101(27):10095–10100
Emlen ST, Wrege PH (1992) Parent–offspring conflict and the
recruitment of helpers among bee-eaters. Nature 356:331–333
Evans JD, Wheeler DE (1999) Differential gene expression between
developing queens and workers in the honey bee, Apis mellifera.
Proc Natl Acad Sci USA 96:5575–5580
Faulks CG, Abbott DH, Liddell CE, George LM, Jarvis JUM (1991)
Hormonal and behavioural aspects of reproductive suppression
in female naked mole-rats. In: Sherman PW, Jarvis JUM,
Alexander RD (eds) The biology of the naked mole-rat.
Princeton University Press, Princeton, pp 426–445
Field J, Cronin A, Bridge C (2006) Future fitness and helping in social
queues. Nature 441:214–217
Gadagkar R (1997) Survival strategies: cooperation and conflict in
animal societies. Harvard University Press, Cambridge, MA
Gardner A, West SA, Wild G (2011) The genetical theory of kin
selection. J Evol Biol 24:1020–1043
Godfrey-Smith P (2009) Darwinian populations and natural selection.
Oxford University Press, Oxford
Grafen A (1984) Natural selection, kin selection and group selection.
In: Krebs JR, Davies NB (eds) Behavioural ecology: an
evolutionary approach, 2nd edn. Blackwell, Oxford, pp 62–84
Greene E (1989) A diet-induced developmental polymorphism in a
caterpillar. Science 243:643–646
Haldane JBS (1932) The causes of evolution. Longmans, Green and
Co., London
Halliburton R (2004) Introduction to population genetics. Pearson
Prentice Hall, Upper Saddle River
Y. Yakubu
Author's personal copy
Hamilton WD (1963) Evolution of altruistic behaviour. Am Nat
Hamilton WD (1964) Genetical evolution of social behaviour, I and
II. J Theor Biol 7:1–52
Harvey JA, Corley LS, Strand MR (2000) Competition induces
adaptive shifts in caste ratios of polyembryonic wasps. Nature
Ito Y (1989) The evolutionary biology of sterile soldiers in aphid.
Trends Ecol Evol 4(3):69–73
Johns PM, Howard KJ, Breisch NL, Rivera A, Thorne BL (2009)
Nonrelatives inherit colony resources in a primitive termite. Proc
Natl Acad Sci 106(41):17452–17456
Lacey AH, Sherman P (1991) Social organization of mole-rat
colonies: evidence of for division of labour. In: Sherman PW,
Jarvis JUM, Alexander RD (eds) The biology of the naked mole-
rat. Princeton University Press, Princeton, pp 275–336
Lively CM (1986) Predator-induced shell dimorphism in the acorn
barnacle (Chthamalus anisopoma). Evolution 40(2):232–242
Maynard-Smith J (1964) Group selection and kin selection. Nature
Maynard-Smith J (1998) The origin of altruism. Nature 393:639–640
Moore J (1995) Behaviour of parasitized animals. Bioscience 45(2):
Morreale SJ, Ruiz GJ, Spotila JR, Sandora EA (1982) Temperature-
dependent sex determination: current practices threaten conser-
vation of sea turtles. Science 216(4551):1245–1247
Nowak MA (2012) Evolving cooperation. J Theor Biol 299:1–8
Nowak MA, Tarnita CE, Wilson EO (2010) The evolution of
eusociality. Nature 466:1057–1062
O’Donnell S (1998) Reproductive caste determination in eusocial wasps
(Hymenoptera: Vespidae). Annu Rev Entomol 43(1):323–346
Okasha S (2009) Biological altruism. In: Zalta EN (ed) The Stanford
encyclopedia of philosophy, winter 2009 edn. http://plato.
Packer C (1977) Reciprocal altruism in olive baboons. Nature
Patel A, Fondrk MK, Kaftanoglu O, Emore C, Hunt G, Frederick K,
Amdam GV (2007) The making of a queen: TOR pathway is a
key player in diphenic caste development. PLoS ONE 2(6):509
Pfennig DW, Murphy PJ (2000) Character displacement in poly-
phenic tadpoles. Evolution 54:1738–1749
Pigliucci M (2001) Phenotypic plasticity: beyond nature and nurture.
Johns Hopkins University Press, Baltimore
Prete FR (1990) The conundrum of the honey bees: one impediment
to the publication of Darwin’s theory. J Hist Biol 23:271–290
Queller DC (1985) Kinship, reciprocity and synergism in the
evolution of social behavior. Nature 318:366–367
Queller DC (2006) To work or not to work. Nature 444:42–43
Rabenold KN (1985) Cooperation in breeding by non-reproductive
wrens: kinship, reciprocity, and demography. Behav Ecol
Sociobiol 17:1–17
Roff DA (1986) The evolution of wing dimorphism in insects.
Evolution 40:1009–1040
Rousset F, Ruze D (2007) Constraints on the origin and maintenance
of genetic kin recognition. Evolution 61(10):2320–2330
Seyfarth RM, Cheney DL (1984) Grooming, alliances and reciprocal
altruism in velvet monkeys. Nature 308:541–543
Shanahan T (2004) The evolution of Darwinism: selection, adapta-
tion, and progress in evolutionary biology. Cambridge Univer-
sity Press, Cambridge
Sherman PW (1977) Nepotism and the evolution of alarm calls.
Science 197:1246–1253
Sober E (1984) The nature of selection. Princeton University Press,
Sober E, Wilson DS (1998) Unto others: the evolution and psychology
of unselfish behavior. Harvard University Press, Cambridge, MA
Taylor C, Nowak MA (2007) Transforming the dilemma. Evolution
Trivers RL (1971) The evolution of reciprocal altruism. Q Rev Biol
van Veelen M, Garcı
´a J, Sabelis MW, Egas M (2010) Call for a return
to rigour in models. Nature 467:661
van Veelen M, Garcı
´a J, Sabelis MW, Egas M (2012) Group selection
and inclusive fitness are not equivalent: the Price equation vs.
models and statistics. J Theor Biol 299:64–80
Wenseleers T, Ratnieks FLW (2006) Enforced altruism in insect
societies. Nature 444:50
West-Eberhard MJ (1986) Alternative adaptations, speciation, and
phylogeny. Proc Natl Acad Sci USA 83:1388–1392
West-Eberhard MJ (2003) Developmental plasticity and evolution.
Oxford University Press, Oxford
Wilkinson GS (1984) Reciprocal food sharing in the vampire bat.
Nature 308:181–184
William BJ (1981) A critical review of models in sociobiology. Annu
Rev Anthropol 10:163–192
Wilson EO (2005) Kin selection as the key to altruism: its rise and
fall. Soc Res 72:159–166
Wilson EO (2008) One giant leap: how insects achieved altruism and
colonial life. Bioscience 58:17–25
Wilson DS, Wilson EO (2007) Rethinking the theoretical foundation
of sociobiology. Q Rev Biol 82:327–348
Winston ML (1987) The biology of the honeybee. Harvard University
Press, Cambridge, MA
Young JA, Clutton-Brock T (2006) Infanticide by subordinates
influences reproductive sharing in cooperatively breeding
meerkats. Biol Lett 2:385–387
The Altruism Paradox
Author's personal copy
ResearchGate has not been able to resolve any citations for this publication.
Full-text available
Sex allocation theory applied to sex-changing animals predicts that the amount of sex change and the proportion of individuals that mature early as the second sex depend upon the mating system of the species or population in question. In turn, theory suggests that the mating system is governed by the size and distribution of resources critical to reproduction, and by population density. Here we investigate the social and ecological factors that govern the amount of selection for sex change and the production of alternative male strategies in a protogynous (female first) goby, Coryphopterus personatus, on atoll-fringing reefs in Belize. We found that: (1) increasing population density lead to an increase in the proportion of early maturing males on leeward-facing reefs, as predicted, but not on windward reefs; (2) contrary to predictions, the proportion of early maturing males was higher on continuously distributed coral gardens than on isolated patches of reef in windward locations, with no difference in leeward locations; and (3) the proportion of early maturing males can be used as a predictor of the population sex ratio, with less biased sex ratios occurring with a higher proportion of early maturing males, as predicted by theory. We discuss these conflicting results in terms of the differences between windward and leeward reefs that might lead to differing selective regimes acting in these locations.
Full-text available
No other scientific theory has had as great an impact on our understanding of the world as Darwin's theory outlined in his Origin of Species. Yet the theory has been the subject of controversy from its very beginning. This book focuses on three issues of debate in Darwin's theory of evolution--the nature of selection, the nature and scope of adaptation, and the question of evolutionary progress. It traces the varying interpretations to which these issues were subjected historically through the fierce contemporary debates continuing to rage.
In 1859 Charles Darwin described a deceptively simple mechanism that he called "natural selection," a combination of variation, inheritance, and reproductive success. He argued that this mechanism was the key to explaining the most puzzling features of the natural world, and science and philosophy were changed forever as a result. The exact nature of the Darwinian process has been controversial ever since, however. The author draws on new developments in biology, philosophy of science, and other fields to give a new analysis and extension of Darwin's idea. The central concept used is that of a "Darwinian population," a collection of things with the capacity to undergo change by natural selection. From this starting point, new analyses of the role of genes in evolution, the application of Darwinian ideas to cultural change, and "evolutionary transitions" that produce complex organisms and societies are developed.
Wing-dimorphic insects are excellent subjects for a study of the evolution of dispersal since the nondispersing brachypterous morph is easily recognized. The purpose of this paper is to develop a framework within which the evolution of wing dimorphism can be understood. A review of the literature indicates that the presence or absence of wings may be controlled by a single locus, two-allele genetic system or a polygenic system. Both types of inheritance can be subsumed within a general threshold model. An increase in the frequency of a brachypterous morph in a population may result from an increased relative fitness of this morph or the emigration of the macropterous type. The abundance of wing-polymorphic species argues for an increased fitness of the brachypterous form. An analysis of the life-history characteristics of 22 species of insects indicates that the brachypterous morph is both more fecund and reproduces earlier that the macropterous morph. Unfortunately, data on males are generally lacking. It is suggested that suppression of wing production results when some hormone, perhaps juvenile hormone, exceeds a threshold value during a critical stage of development. Further, it is known that in the monomorphically winged species Oncopeltus fasciatus both flight and oviposition are regulated by the titer of juvenile hormone. These observations are used to construct a possible pathway for the evolution of wing dimorphism. This suggests that evolution to a dimorphic species requires both an increase in the rate of production of the wing suppressing hormone and a change in the threshold level at which wing and wing-muscle production are suppressed. The stage in this evolutionary sequence that an organism will reach depends on the stability of the habitat.
In dense populations, most planthoppers (Homoptera:Delphacidae) produce fully winged migratory forms that can escape to new habitats. Under less crowded conditions, flightless morphs with reduced wings result. For two sympatric species, Prokelisia marginata and Prokelisia dolus, interspecific crowding was found to be as strong a stimulus for the production of migrants as intraspecific crowding. The effects were reciprocal for both species and were demonstrated both in the laboratory and field. However, because migratory forms were triggered at a much lower density in P. marginata than P. dolus, interspecific interactions were asymmetric, with P. dolus having a far greater influence on the wing form of P. marginata. Interspecific interactions did not directly influence survival, development time, or body size, but reproductive capability was indirectly affected because migratory females are less fecund than their flightless counterparts. Intraspecific crowding adversely affected the survivorship, development time, and body size of P. marginata, but had no influence on the fitness of P. dolus. P. marginata is a much more migratory species than P. dolus. Consequently, mobility is associated with the extent that each species' fitness is reduced by intraspecific crowding. Our results demonstrate that for wing dimorphic insects, the effects of interspecific competition can be extended to include altered wing form, a character that directly influences dispersal capability as well as reproductive potential. We suggest that when wing-dimorphic species share a common habitat, interspecific interactions can influence population dynamics by increasing emigration and decreasing population growth.