Viscous populations evolve altruistic programmed
ageing in ability conflict in a changing environment
School of Life Sciences, Nanjing University, Jiangsu, China
Questions: Is ageing evolutionarily adaptive? Can programmed ageing widely evolve as
altruism in viscous populations (i.e. widely distributed populations with limited offspring
dispersal) in a changing environment?
Features of model: The model is individual-based. The probabilities of survival and repro-
duction are determined by abilities, and abilities increase with both inherited abilities and
age-related abilities, so the old can survive and reproduce even if they are genetically less
adapted to the environment (termed ‘ability conflict’). Inherited traits are determined by
multiple independent loci; thus active ageing can enhance the local accumulation of adaptive
inherited abilities in viscous populations.
Ranges of key variables: Dispersal varied from 0 (no dispersal) to 1 (global). The probability
of environment-change during each calculation cycle varied from 0 to 1.
Conclusions: Altruistic ageing evolves in structured viscous biological populations with
ability conflict in a changing environment to allow the survival of genetically fitter young
progenies. To evolve altruistic ageing requires no more environmental change than does sex,
suggesting that the generality of altruistic ageing should be no less than sex in viscous
populations. If selfish mutants appear only at low rates, higher-level selection would be
stabilized even if the environment changes slowly. More extrinsic death can decrease ageing rate
(intrinsic death rate) to ensure the same expected lifespan in altruistic ageing, providing testable
predictions against traditional ageing theories. My individual-based model also shows how
traditional mathematical population genetics largely underestimated the prevalence of group
Keywords: evolvability, genetic creativity, kin selection, longevity, population viscosity,
Originally, evolutionists argued that ageing is an adaptation to each individual accumulat-
ing damage from small injuries during its lifetime, an adaptation that can ‘make room
for the young’ (Weismann, 1889). However, most evolutionary theorists abandoned this group
Correspondence: J.-N. Yang, School of Life Sciences, Nanjing University, Jiangsu 210093, China. e-mail:
Consult the copyright statement on the inside front cover for non-commercial copying policies.
Evolutionary Ecology Research, 2013, 15: 527–543
© 2013 Jiang-Nan Yang
selection argument and came to regard ageing as mainly a non-adaptive consequence of the
decreased power of natural selection at advanced age, so mutations that are deleterious only
at old age accumulate in the population (Medawar, 1952). To counter the hypothesis of weak
natural selection in this mutation accumulation theory, antagonistic pleiotropy theory
suggests that genes deleterious only at old age can be fixed in the population if they are
beneficial earlier in life (Williams, 1957; Hamilton, 1966; Bourke, 2007; Rose et al., 2007; Kirkwood and Melov, 2011).
Although antagonistic pleiotropy theory describes some adaptive life-history trade-offs,
ageing itself is described as non-adaptive, and thus should not have conserved genetic
Increasing empirical evidence, however, shows that lots of age-related diseases have
conserved genetic mechanisms to ‘activate’ them (Bowles, 2000a) without any apparent
beneficial effect earlier in life (Linnane et al., 1990; Mitteldorf, 2004; Khaidakov et al., 2006; Brack et al., 2007;
Thum et al., 2008; Goldsmith, 2010; Kenyon, 2010; Pishel et al., 2012). Experimental support for the
traditional ageing theories can also be explained by adaptive ageing theories (Mitteldorf, 2004,
2010; Goldsmith, 2010). More important, the traditional ageing theories predict that faster ageing
evolves with higher extrinsic death rate, but guppies living in regions with higher predation
actually have a slower ageing rate (and higher fecundity) (Reznick et al., 2004).
Thus, some theorists return to Weismann (1889) and argue that ageing/programmed death
can benefit the population or kin by making room for the young (Bowles, 1998; Skulachev, 2001;
Goldsmith, 2004, 2008; Travis, 2004; Mitteldorf, 2006; Woodberry et al., 2007; Mitteldorf and Pepper, 2009; Martins, 2011),
especially in a changing environment where faster ageing can result in faster local evolution.
However, these theories assume asexual reproduction and/or group selection and that
makes many biologists sceptical about their validity or generality (Bourke, 2007). The
assumption of asexual reproduction is used to argue against these theories (Bourke, 2007)
because novel adaptive genes generated in faster ageing groups can be recombined into
selfishly long-lived individuals in a sexual population. However, the assumption of asexual
reproduction is not necessarily a problem for altruistic ageing because fitness can involve
multiple genes, and it will be very difficult to recombine all the un-linked novel adaptive
genes into a selfishly long-lived individual just by chance.
The rejection of group selection by most biologists is based on mathematical analyses
from Price’s equation (Price, 1970), evolutionarily stable strategy theory (Smith, 1976), and kin
selection theory (Hamilton, 1964). Although powerful, these mathematical analyses ignored
important factors in evolution that are extremely difficult to reduce to mathematical
equations, because the mathematics has assumed the identity/similarity of individuals when
counting individual/offspring/allele numbers or fitness, but the theory of evolution from the
beginning stresses mutations, variations, and individual differences for natural selection
to work on. Thus the way these mathematical analyses are constructed loses important
information, such as exact individual genomic differences, and population–environment
interactions in long-term evolution where novel adaptations differentially accumulate with
time among different local groups. In addition, the Price equation ignores continuous
(multiple) novel mutations, and thus does not apply to long-term evolution. Although
population viscosity (i.e. limited offspring dispersal) has been proposed as a potential
general kin/group selection mechanism for the evolution of altruism (Hamilton, 1964), existing
models pursuing mathematical simplicity fail to show the evolution of altruism in viscous
populations with constant carrying capacity because the benefit of additional offspring
resulting from kin cooperation is exactly cancelled out by the cost of kin competition
because offspring cannot be easily exported (Taylor, 1992; Wilson et al., 1992).
These viscous population models, however, also imply that any additional benefit ignored
by these models should be sufficient for the evolution of altruism. For example, altruism
and ageing can evolve if population size is allowed to fluctuate to store temporarily the
additional offspring resulting from kin cooperation (Mitteldorf and Wilson, 2000; Mitteldorf, 2006).
More important, since these models only count offspring number (quantity) as fitness, they
ignore the important fact that selection among progenies produced with mutation and
sexual recombination in a changing environment will increase the average genetic quality of
surviving progenies by eliminating the weaker progenies and helping to store the benefit
of kin cooperation in the form of progeny quality, and that can be modelled only with
individual-based models incorporating genetic information for each individual (emphasiz-
ing individual differences). These fitter progenies can invade other groups with inheritable
long-term fitness. So, the generality of altruism has been severely underestimated in existing
theories. Evidence from social insects also suggests that altruism evolves beyond the
explanation of Hamilton’s kin selection theory and inclusive fitness (Wilson, 2005; Gadagkar, 2011).
To explore the benefit of inheritable progeny quality in viscous populations and apply it
to the evolution of altruistic ageing, I built individual-based computer-simulation models
of viscous populations to show the evolution of programmed death (here called active/
adaptive/altruistic ageing) without early-life benefit (i.e. without antagonistic pleiotropy).
I base the models on assumptions that were not previously considered by adaptive models
and were believed to be most adverse to the evolution of active ageing: sexual reproduction
with possible selfish mutations (intra-group selfish mutants in adaptive groups are most
adverse for altruism); constant carrying capacity (no room for additional individuals to
store fitness); high cost of mutations; and relatively slow environmental change (against
genetic creativity, which many call ‘evolvability’). Individual-based models can more easily
capture population structure, individual–individual interactions, individual–environment
interactions, and genome structures in long-term evolution, each of which is important for
altruistic ageing but typically ignored by pure mathematical models.
The model has a number (N) of regions arranged in a circle. Each region can contain tens
of diploid individuals, and offspring can be stochastically dispersed to other regions
depending on parameters of dispersal probability and dispersal distance. Each individual in
individual-based computer-simulation models contains individual-specific information
that is independent of other individuals. In my model, groups do not split or become extinct
as whole groups, and higher-level selection has to be based on individual-level selection.
All individual traits are determined by multiple genes (binary numbers, with stochastic
mutation and sexual recombination), and trait values are proportional to the proportion of
1s in these binary numbers. The multiple-gene nature of inherited abilities can make them
difficult to segregate completely with their associated phenotypes (such as altruistic ageing)
by chance during sex. Actual ability A of an individual is calculated from age a and
inherited ability Ain:
A = (a + 0.5)(Ain− Amin+ ε)D, (1)
where Amin is the minimum inherited ability among all individuals in the same region, ε is a
small positive number used to avoid A becoming 0 when all individuals have the same
inherited ability, and D (ability distinctness, or selection strength) is a parameter used to set
The evolution of general altruistic ageing529
the selection sensitivity to different values of inherited ability. When only individuals with
the fittest phenotype can survive or reproduce, D equals infinity; otherwise, D can be as
small as 0 for drift only. Individuals with larger A will proportionately compete more
resources to survive and reproduce in a region. Thus continuous selection among progenies
will lead to the accumulation of positive ability genes, which will be important for both
short-term individual-level and long-term group-level competition. Individuals may die
either because they (if any) have reached the end of their inherited maximum lifespans (L),
or (mainly) because of failure in survival competition in a region with constant carrying
capacity (individual selection). Both survival and sperm competition are stochastic with
success probabilities proportional to resource acquired. Egg number is also proportional to
resource acquired. I use the final evolved L compared to its initial value and its random
walk under neutral evolution to determine whether active programmed death has evolved.
During an environmental change, the binary site of ability genes that have evolved (without
being reset) for the longest time is reset to 0 for all individuals to allow a novel ability gene
I discuss the model in more detail in evolutionary-ecology.com/data/2825Appendix.pdf.
I chose parameter values close to the conditions of most animals (wide population
distribution but limited offspring dispersal, some environmental change, heavy mutational
costs, etc.) as the default parameters (2825Appendix.pdf, Table S1). Such parameter values
improve our ability to understand the effects of different parameters on the evolution of
altruistic ageing. I present a comprehensive exploration of the parameter space in a simpler
model in the final subsection of the results. The default dispersal parameters are consistent
with most terrestrial non-flying animals, as well as flying and aquatic animals that do not
migrate continentally or globally. Parameters of environmental change and mutational
costs are difficult to determine, but I will show that they were either set consistently to the
requirements of the evolution of sex or else they were non-essential. The increase of actual
abilities with age is also widely true for animals with body growth, learning and experience,
and acquired immunities. I set sexual maturation age to 2 years (calculation cycles of
reproduction, resource competition, death, and possible environmental change), although
this did not qualitatively affect the results. The maximum L allowed is 100 years. The
evolution of L without selection was a random walk around 50 ± 15 years (mean ± S.D.)
Programmed death actively evolves in viscous populations with kin
competition and group competition
L responded to dispersal conditions (Fig. 1). When dispersal distance dN/2 and dispersal
probability pd were both 0, different resource regions evolved independently and genes of
longer L (72.2 ± 1.1 years) accumulated. When dispersal increased a little (d = pd= 0.04) so
that there existed both kin competition and group competition, L was ‘attracted’ from an
initial value of 50 years to low levels with small standard deviation (17.5 ± 1.8 years),
implying a benefit for low L. When dispersal was unrestricted (d = pd= 1), L evolved to be
very large (86.2 ± 5.8 years). Therefore, kin competition led individuals to have altruistically
short L in interconnected viscous sub-populations. In accordance with L, more senescent
deaths evolved when dispersal was neither 0 nor too large (2825Appendix.pdf, Fig. S1).
Programmed death evolves to overcome ability conflicts and increase genetic
creativity in local populations in a changing environment
Environmental change was required for the evolution of active ageing (Fig. 2a), as it
ensured the opportunity for the evolution of new ability genes. Similarly, setting inherited
ability to a fixed value eliminated the evolution of active ageing. By default, the number of
ability genes in a gamete was 31 for computational efficiency. This was a relatively small
number, and all ability genes were quickly removed if the environment-changing rate was
high, making novel adaptive ability genes no longer useful for long-term group competition
(Fig. 2a). Individuals in the model with small-ability genome size were also very likely to
have the same novel mutations, making the creation of genetically fitter progenies by sexual
Fig. 1. Short inherited maximum lifespan (L) can actively evolve in viscous populations. dN/2 is the
maximum distance offspring could be dispersed to (both sides), and pd is dispersal probability.
(a) Population average L actively and stably evolved to altruistically low levels from an initial value of
about 50 years when kin competition (small dispersal) and group competition (non-zero dispersal)
both existed. (b) More detailed response of the final evolved L (±S.D.) to the change of d and pd,
averaged from 100,000 to 300,000 years. The simulations were insensitive to initial values, so replicates
were done by extension of time.
The evolution of general altruistic ageing 531
recombination inefficient. However, by allocating 310 ability genes in a gamete, strong
altruistic ageing (L = 13.8 ± 1.4 years) evolved for the default parameters, and evolved L
decreased even with the fastest allowed environment-changing rate of 1 (default 0.1) (to
10.4 ± 1.2 years).
Higher ability distinctness D (selection efficiency/strength) should increase the import-
ance of inherited abilities, and could promote altruistic ageing (Fig. 2b). An example of
ability distinctness of a trait is running speed and survival. Individuals that can run slightly
faster may have a much greater chance of survival than others if they are chased and preyed
upon by carnivores (high ability distinctness). However, if those same individuals are easily
Fig. 2. Programmed ageing evolves to improve progeny inherited abilities in a changing environment
under ability conflicts. (a) Environmental change was required for the evolution of short L.
Environment-changing rate, the probability for environmental change to happen in a year. Extremely
fast change of environment eliminated newly evolved ability genes so that they could no longer
promote altruistic ageing. (b) Higher ability distinctness D (which measures selection efficiency)
promoted the evolution of short L. (c) Old individuals’ viable and reproductive (valid) offspring
number continuously decreased with age (because of decreased inherited ability). (d) Little active
ageing was observed when age did not contribute to actual ability A or A was recalculated every year
according to inherited ability genes (constant renewal).
preyed upon by eagles (extremely fast) or snakes (stealthy), running faster will probably
make little difference (low ability distinctness) unless they still have a fair chance to escape
after being targeted. An ability distinctness of 1 was sufficient for obvious altruistic ageing
to evolve (Fig. 2b), suggesting that altruistic ageing does not require extremely high
Active intrinsic death requires old individuals to have strong actual abilities to survive
extrinsic death even though they are genetically inferior to their progenies. This condition
is termed ‘ability conflict’ and is achieved by the contribution of age (acquired ability) to
A. With ability conflicts, old individuals had fewer viable and reproductive offspring even
though their total offspring increased with age (Fig. 2c). Active ageing could not evolve if
actual ability was not affected by age or was constantly renewed by germ-line inherited
ability (Fig. 2d). The reason for that result is that intrinsic active death is no longer
necessary where there are no ability conflicts, and the poorly adapted old individuals can be
eliminated by extrinsic death from local survival competition.
Dynamically, altruistically shorter L could be stochastically generated in a few, although
not many, local regions and then spread to other regions because of its association with
higher inherited ability (http://evolutionary-ecology.com/data/2825Video.avi). Eventually,
the population achieved equilibrium when the random generation and selected spreading of
short L were cancelled out by the invasion of selfish variants. Any factor affecting this
equilibrium should also affect L.
The evolution of programmed death does not require more
environmental change than does sex
The evolution of active ageing requires environmental change. But how quickly does that
change have to occur? To answer this, I incorporated the evolution of sexual reproduction,
which also requires environmental change, as a control. The model was modified by adding
another trait, the probability of infant sexual development. Infants could either develop
into sexually reproducing individuals or develop into asexual individuals [which do not pay
the two-fold cost of sex (Otto and Lenormand, 2002)]. When environment-change rate increased
above a certain level so that sex probability evolved above 0.5, active ageing also evolved in
viscous populations (Fig. 3). So, the evolution of active ageing does not require more
environmental change than the evolution of sex does. The evolution of sex requires genetic
diversity, which also forms the basis for the increase of offspring genetic fitness after
selection and promotes the evolution of altruistic ageing. Thus, the generality of sex also
suggests the generality of altruistic ageing. Comparing Figs. 2a and 3b, we see that the
evolved altruistic L was lower given the possibility of asexuality, which suggests that sex
itself reduces altruistic ageing and increases L. However, sex suggests the existence of
environmental change, which promotes altruistic ageing.
Slower (or biased) mutation rates of genes for ageing and larger population
ranges promote programmed ageing by higher-level selection
Traditional theories often assumed arbitrary mutations and argued that higher-level
selection is too slow to be common in biology. However, both mutation rate and manner
are important. The emergence of selfish mutants also takes time and their predominance in
The evolution of general altruistic ageing 533
the population is not necessarily faster than higher-level selection. It is unreasonable to
assume a selfish mutant whenever there is a possibility of such. In fact, a slower mutation
rate of L genes could evolve to promote the evolution of limited L (Fig. 4a,b).
The model of Fig. 4 differs from the model of Fig. 1 by having lifespan genes with
mutation coolness r so that the mutation rate of L genes is 1002r − 1 times (ranging from
0.01 to 100) that of other genes. Smaller r could enhance selection at higher levels by
protecting altruistic groups from selfish invasion for a longer period of time, allowing them
to accumulate larger inter-group differences, and by reducing the chance for selfish
mutations to occur simultaneously in different regions. If dispersal increases, making
selection for low r and L less efficient, r and L could still be selected to low levels if more
resource regions existed (Fig. 4c). A large number of resource regions can reduce the
probability that all regions become occupied by selfish individuals and thus can also
enhance selection at higher levels. When I allowed only 20% L mutations to increase L while
the other 80% decreased L (by default, L mutations had an equal chance to increase and
decrease L), L also decreased consistently to low levels (7.1 ± 0.6 years) from the model of
Fig. 1. Biased mutation could happen if immortality requires perfect functions/sequences in
some genes, since it would be easier for mutations to make a sequence imperfect rather than
perfect (only individual selection promotes perfection).
To further illustrate higher-level selection, I allowed dispersal probability pd and the level
of dispersal distance d to change independently (Fig. 4d) in the model of Fig. 1. A small
value of either of them was sufficient for the evolution of active ageing. Active ageing could
evolve even if all offspring were dispersed, as long as they were dispersed only to nearby
regions. This result also strongly supports the generality of altruistic ageing; it can evolve
even if genes can flow freely among nearby regions.
Fig. 3. Altruistic ageing does not require more environmental change to evolve than sex does. (a) The
evolution and maintenance of sexual reproduction required a relatively high environment-changing
rate (ECR). (b) Altruistically short L had already evolved under such a rate of environmental change
in viscous populations.
Higher extrinsic death rate decreases ageing rate under altruistic ageing
I tested the effect of death rate on the evolution of L. I imposed an additional random death
at a certain rate on each individual each year before survival competition (Fig. 5). In
accordance with the traditional theory, higher death rate reduced selection forces at old ages
and decreased L when conditions did not favour altruism. However, when dispersal
(d = pd= 0.04) and environmental change favoured a short L, high death rate actually
increased L. This happened because old individuals were actively dying, so higher extrinsic
death rate reduced the need for active intrinsic death, as seen in guppies (Reznick et al., 2004). So,
the relationship between death rate and L depends on the degree of altruistic ageing. If
altruistic ageing is general, higher random death rate should decrease ageing rate instead of
Fig. 4. Evolved low mutation rate of ageing genes and wide population distribution can promote
altruistic ageing by higher-level selection. (a) The evolution began with mutation coolness (r) near 1 so
that lifespan genes mutated 1002r − 1= 100 times as fast as other genes, so selfish invasions became
more frequent and individuals gave up altruistic ageing. However, r was quickly selected to low levels
and promoted the evolution of short L. (b) r evolved to low levels only in viscous populations where
altruistic ageing could evolve. (c) A large number (N) of resource regions (i.e. a wide population
distribution) restored low levels of r and L when dispersal became less favourable for altruistic ageing.
(d) A low level of pd or d alone could result in altruistic ageing. The default value of N was 100.
The evolution of general altruistic ageing535
The parameter spaces that evolved altruistic ageing
To further understand how altruistic ageing could evolve, it would be helpful to identify the
parameter spaces where altruistic ageing could evolve. To achieve this, the model needed to
be simplified to comprehensively simulate different parameter combinations and to show
the key determinants of altruistic ageing.
To simplify the model, I set carrying capacity (final local survival each year) to 1, thus
reducing regions to virtual points. I set dispersal probability to an even distribution within
the dispersal distance. I set ability distinctness D to 1 to achieve completely linear selection
on inherited abilities. I assumed no mutational costs and used mutation rate as a parameter
instead of an evolvable trait. Individuals reproduced every calculation cycle to avoid the use
of maturation age (i.e. setting it to 1). I set the maximum possible value of L to 40 years;
thus a random walk of L would be around L = 20 years.
After these simplifications, only five parameters remained: the number of resource
regions/points (N); dispersal distance; environment-changing rate (ECR); mutation rate
(MutRt); and resource at each region/point (QT). I used three or four values spanning
at least two orders of magnitude for each parameter and I simulated all their possible
Fig. 5. Extrinsic death decreases non-altruistic inherited maximum lifespan (L) but also increases
altruistic L. All individuals were subjected to random death every year with random death rate values
displayed along the x-axis. Whether the environment-changing rate ECR = 0, or pd= d = 0 or
pd= d = 1, the model always evolved non-altruistic (long) L. But if pd= d = 0.04, the model evolved
altruistic short L for death rate < 0.20 and non-altruistic long L for death rate ≥ 0.25.
combinations (Table 1). Small changes of the parameters did not qualitatively affect the
results since the results showed continuity with the change of parameters. I used values at
the last two-thirds of the calculation time. I made calculation times long enough (600,000
to 100 million years) so that the absolute values of Spearman’s rho (rank correlation)
between the values of L and time were smaller than 0.1. I used an evolved average L smaller
than 17 years as the cut-off value for altruistic ageing.
Generally, population viscosity was required for the evolution of altruistic ageing, i.e.
altruistic ageing evolved where population range (N) was large and dispersal was neither
0 nor global. Altruistic ageing was still possible even though the other three parameters
changed over at least two orders of magnitude, suggesting the robustness of altruistic
Mutation rate could change over three orders of magnitude without determining whether
altruistic ageing could evolve. But smaller mutation rates stabilized stronger altruistic
ageing. However, when mutation rate was below 3.3 × 10–5, the occurrence and spread of
longevity and ability mutations could easily become independent, and altruistic ageing
could evolve only if resource in each region/reproduction rate was high (which speeds up
the occurrence and accumulation of novel ability genes). High mutation rate diminished
altruistic ageing, because higher-level selection cannot work when selfish mutants easily
Altruistic ageing requires environmental change. However, its effect on altruistic ageing
depended on mutation rate. When mutation rate was low, contrary to popular belief, slower
environmental change actually promoted altruistic ageing, because it provided more time
for the rare novel ability genes to remain adaptive in higher-level selection. Thus, slow
environmental change is not necessarily a problem for the evolution of altruistic ageing as
long as mutation rate is low. Since the model had only a small number of sites (31 sites) for
ability genes, environmental change created opportunity for novel ability genes by removing
older ones (though not always true in nature). So, if it occurred too fast for different regions
Table 1. Parameter combinations where short inherited maximum lifespan (L <17) evolved
3.3 × 10–5
0 ∼ 40
3.3 × 10–5
3.3 × 10–5
3.3 × 10–5
The evolution of general altruistic ageing 537
to maintain variability, it diminished altruistic ageing. On the other hand, at higher muta-
tion rates (0.001 or 0.033), altruistic ageing became weaker with slow environmental change
(0.01), since the sites for ability genes could easily be filled and few novel ability mutations
could co-occur with selfish mutations.
The effect of resources in each region/point depended on other factors; by itself, it did not
strongly affect altruistic ageing. At low mutation rate, more resource (thus more repro-
duction) promoted altruistic ageing because it accelerated the local accumulation rate of
novel ability genes more than that of long-lived selfish mutations (a longevity mutation
does not affect fitness in earlier life but an ability mutation does). However, by the same
mechanism, it diminished altruistic ageing at higher mutation rate (half of the mutations
were beneficial) coupled with slow environmental change (because ability sites could easily
So, the regional co-occurrence of novel ability genes at the rise and spread of selfish
mutations is the key for the evolution of altruistic ageing and is required for understanding
the complexity of the above factors.
I built individual-based models to study the evolution of programmed death (i.e. ageing)
and longevity in sexual populations with selfish mutations, heavy costs of beneficial
mutations, and constant carrying capacity. For simplicity, I did not consider age-
associated gradual fitness decline in these models. Nor was model longevity associated with
Originally, sexual reproduction with selfish mutations, costly beneficial mutations,
and constant carrying capacity were believed to work against the evolution of active
ageing, and few of them were included in previous models of active ageing. However,
I have demonstrated here that active ageing can evolve even if these conditions are
present as long as the species has limited offspring dispersal, a wide distribution, and
faces an environment that keeps changing, thus driving ability conflicts between
generations. Limited offspring dispersal maintains kinship and altruism among local
individuals. The demand for high creativity and evolutionary responsiveness to keep up
with environmental changes drives the selection of programmed ageing. Although old
individuals tend to have superior survival abilities associated with age if they lack ageing
and active death, their inherited abilities tend to be inferior (ability conflicts), because
variation and natural selection among progenies increase the average fitness of survived
progenies. Active ageing can evolve even when high cost of mutations is introduced to
work against the improvement of progenies (2825Appendix.pdf, Fig. S2), as ‘excessive’
reproduction and selection enable local populations to tolerate the costs of lethal and
detrimental mutations. Increasing the importance of inherited abilities in survival and
reproduction also increases the rate of ageing. For the evolution of active ageing, the
environment is not required to change any faster than it must for sex to evolve. Lower
mutation rate of ageing mechanisms can evolve to reduce the generation rate of selfish
mutations and to promote and stabilize the evolution of active ageing, and this also
decreases the requirement for fast environmental change. More resource regions (larger
population distribution) can similarly promote active ageing by preventing selfishness from
winning in all local regions. Higher extrinsic death rate can decrease the rate of ageing
(intrinsic death) if conditions allow altruistic/active ageing to evolve and if extrinsic
death has low ability distinctness (i.e. more random for different individuals and capable of
eliminating strong old individuals).
The model and improvement of population genetics
The success of altruism through stronger ability genes is based on a mechanism known as
genetic hitchhiking or genetic drift (Barton, 2000), although traditional hitchhiking requires the
neutral/harmful gene to be directly linked to the beneficial gene in sexual populations, which
is not the case here, as I assumed no genetic linkages at all. It works because both ageing and
abilities involve multiple genes, so they are much harder to separate by chance than just two
genes. It will take many generations for the progenies from selfishly long-lived groups to
gain all the novel adaptive genes from altruistic groups by sexual recombination. However,
without competitive inherited abilities, these progenies can hardly survive and reproduce for
so many generations once they come into competition with altruistic strong individuals
(slower ageing is of no help if one cannot survive to an old age).
Slower generation rate of selfish mutants can evolve to promote altruistic ageing. Because
of mutation, Hamilton’s rule of kin selection does not apply. For example, if there are no
mutations, individuals in a population will eventually become identical because of selection
and genetic drift. However, if mutation rate is high, even ‘identical twins’ cannot be viewed
as relatives. This is why mutation rate of altruistic genes (and population range), which is
not considered in kin selection theory, is so important for the evolution of altruistic ageing.
The key to understanding the evolution of altruism is the maintenance of altruism against
selfish variants in the population dynamics of the stochastic emergence of altruism, the
spreading of altruism by higher-level selection, and the invasion by selfish variants.
Kin selection based on Hamilton relatedness is only a specific means for individuals in
un-mutated lineages to defend against selfish invaders from mutated lineages.
Higher-level selection is often argued to be too slow to be common for altruistic
evolution. However, this claim is also based on unjustified assumptions of selfish mutations.
It ignores that the generation rate of selfish mutants also takes time and it is not necessarily
faster than the accumulation of adaptive genes. By assuming that lifespan genes and ability
genes had the same mutation rate and number of loci, the present model showed that
altruistic ageing could stably evolve. Slower mutation rate of altruistic genes could also
evolve to promote higher-level selection and altruism. If mutations were more likely to
decrease than increase longevity, higher-level selection and altruistic ageing could also be
more predominant. So, even if the generation rate of novel adaptations is slow and the
young are sometimes genetically no better than the old, altruistic ageing can still evolve,
because the generation rate of selfish mutants is also limited. Besides, fitness is
multi-dimensional in the complex ecological interactions but longevity is only one factor.
Genes affecting longevity is generally conceived to be encoded by only a small proportion
of the genome. This also promotes the evolution of altruistic ageing by promoting the faster
overall occurrence of novel ability genes than that of selfish mutations.
How altruistic ageing theories explain empirical evidence better than the
theory of mutation accumulation and antagonistic pleiotropy
In the natural world, most animal species have limited offspring dispersal but wide
distribution, so kin group selection and altruistically limited lifespan are the usual predicted
The evolution of general altruistic ageing539
cases in my model. Species that suffer less or no kin competition usually have longer
inherited maximum lifespan (IML). For example, as birds and bats can fly (Crow, 1997; Arnheim
and Calabrese, 2009), they usually live in a wider range of areas and thus suffer less kin com-
petition than terrestrial non-flying mammals, and they usually have longer lifespan than
terrestrial non-flying mammals with the same body size and similar life cycle. An alternative
explanation from the traditional ageing theory of extrinsic mortality (Medawar, 1952) is
that flight enables animals to escape predators and leaves more older individuals alive to
promote the evolution of longer lifespan (Drake et al., 1998; Dytham and Travis, 2006; Arnheim and Calabrese,
2009). Although the present study does not exclude the effect of extremely high death rate on
IML evolution if conditions favour longer lifespan, it shows that higher extrinsic random
death rate can also result in longer IML if altruism can evolve. This complementarity
between extrinsic and intrinsic death rates has been observed in natural guppy populations
(Reznick et al., 2004). Some counterarguments based on highly unrelated species, such as com-
paring porcupines and elephants with some short-lived mammals, may be irrelevant,
because the long lifespan of porcupines and elephants may simply be because of their life-
history differences (such as slower sexual maturation and yearly reproduction) compared
with short-lived animals, rather than because of their good protection from predators.
Animals living individually in vast oceans (and maybe hunting different prey at different
ages), such as sea turtles (other turtles usually have much shorter lifespans) and some fishes
(Vaupel et al., 2004), or corals (Vaupel et al., 2004) and most plants that do not move freely but
disperse offspring relatively far away, are all species that suffer little kin competition and
thus have long lifespan and negligible senescence. The immortal hydras (Solomon et al., 2002)
also suffer little kin competition, as their adhesive lifestyle is similar to that of plants; their
constant tissue renewal (Martinez, 1998) may also imply that they do not have obvious ability
conflicts between young and elder ages.
Most longevity villages in humans are found in nearly isolated mountain areas (Poulain et al.,
2004) such as Bama in China or islands (e.g. Okinawa in Japan) where higher-level selection
does not exist. The longevity of residents in Rugao County in China may possibly evolve
from the other extreme condition of kin competition, as the county is not isolated but
characterized by many immigrations throughout its history and dense population [it is the
most populated agricultural area in China (Government of Rugao County, 2001)], and thus low level
of relatedness and kin competition. Evidence from birds is similar: generally, island birds
live longer than mainland birds, and colonial birds live longer than solitary birds (Arnheim and
Alleged experimental support (Sinclair, 2005; Charmantier et al., 2006; Rose et al., 2007) for the
antagonistic pleiotropy theory does not solely support this theory either. For example,
experiments on caloric restriction and life-history ‘trade-offs’ are also consistent with
adaptive ageing theories. The main point of adaptive ageing is to avoid competition with
and leave resource to genetically fitter progenies, so individuals should have enough
offspring before ageing to ensure the existence of fitter progenies to inherit their resource.
On the other hand, adaptive/active ageing itself means that individuals should not live to
too old an age, i.e. individuals should have an optimized number of offspring given a certain
degree of altruism. Thus, of course, when food is restricted and reproduction rate has to be
reduced, individuals should live longer.
The proposition in the theory of mutation accumulation that there are too few old
individuals in natural populations to allow the evolution of active ageing is a false
assumption. The theory often gives examples of highly preyed upon species with high
mortality rate, but does not mention carnivores or even top predators that also show
obvious senescence. Bowles (2000b) clearly pointed out that natural immortal populations can
be old-dominated because of stronger acquired abilities of the old. There is sufficient
evidence that survival and fecundity decrease with old age in natural populations (Ricklefs,
1998, 2008; Loison et al., 1999; Bonduriansky and Brassil, 2002; Libertini, 2008; Nussey et al., 2009). So, ageing does
exist in natural populations, and the alleged non-existence of very old individuals in natural
populations may simply be because of active ageing itself.
This paper strongly supports an adaptive theory that ageing and age-related diseases
are actively evolved altruistic characteristics. It is possible this altruism can evolve in
most animal species with wide distribution but limited offspring dispersal. I predict that
species with less random death and stronger selection should have a stronger active ageing
mechanism (relative to their wild life history such as sexual maturation age and repro-
duction rate). I also predict that the molecular mechanism of ageing and age-related
diseases should lie in mechanisms where it is difficult for selfish mutants to arise or spread,
such as sequence imperfection of genes involved in DNA repair, DNA replication, cell cycle
control and other body maintenance [longevity gene sequences of short-lived species should
evolve more imperfections and some have been found (Semeiks and Grishin, 2012) to diverge
further from long-lived species such as humans], as it is much easier to mutate from than to
a perfect sequence because of entropic reasons.
I thank Jorge Azpurua, Alexis I. Stein, and Michael Rosenzweig for their advice and English editing.
I thank Josh Mitteldorf and anonymous reviewers for comments.
Arnheim, N. and Calabrese, P. 2009. Understanding what determines the frequency and pattern of
human germline mutations. Nature Rev. Genet., 10: 478–488.
Barton, N.H. 2000. Genetic hitchhiking. Phil. Trans. R. Soc. Lond. B, 355: 1553–1562.
Bonduriansky, R. and Brassil, C.E. 2002. Senescence: rapid and costly ageing in wild male flies.
Nature, 420: 377.
Bourke, A.F.G. 2007. Kin selection and the evolutionary theory of aging. Annu. Rev. Ecol. Evol.
Syst., 38: 103–128.
Bowles, J. 1998. The evolution of aging: a new approach to an old problem of biology. Medical
Hypotheses, 51: 179–221.
Bowles, J. 2000a. Sex, kings and serial killers and other group-selected human traits. Medical
Hypotheses, 54: 864–894.
Bowles, J. 2000b. Shattered: Medawar’s test tubes and their enduring legacy of chaos. Medical
Hypotheses, 54: 326–339.
Brack, A.S., Conboy, M.J., Roy, S., Lee, M., Kuo, C.J., Keller, C. et al. 2007. Increased Wnt signaling
during aging alters muscle stem cell fate and increases fibrosis. Science, 317: 807–810.
Charmantier, A., Perrins, C., McCleery, R.H. and Sheldon, B.C. 2006. Quantitative genetics of age
at reproduction in wild swans: support for antagonistic pleiotropy models of senescence. Proc.
Natl. Acad. Sci. USA, 103: 6587–6592.
The evolution of general altruistic ageing541
Crow, J.F. 1997. The high spontaneous mutation rate: is it a health risk? Proc. Natl. Acad. Sci. USA,
Drake, J., Charlesworth, B., Charlesworth, D. and Crow, J. 1998. Rates of spontaneous mutation.
Genetics, 148: 1667–1686.
Dytham, C. and Travis, J.M.J. 2006. Evolving dispersal and age at death. Oikos, 113: 530–538.
Gadagkar, R. 2011. Science as a hobby: how and why I came to study the social life of an Indian
primitively eusocial wasp. Curr. Sci. India, 100: 845–858.
Goldsmith, T. 2004. Aging as an evolved characteristic – Weismann’s theory reconsidered. Medical
Hypotheses, 62: 304–308.
Goldsmith, T. 2008. Aging, evolvability, and the individual benefit requirement: medical implications
of aging theory controversies. J. Theor. Biol., 252: 764–768.
Goldsmith, T.C. 2010. The Evolution of Aging: How New Theories Will Change the Future of
Medicine. Crownsville, MD: Azinet Press.
Government of Rugao County. 2001. Yearbook of Rugao County in 2000. Rugao: Central
Hamilton, W. 1964. The genetical evolution of social behaviour. J. Theor. Biol., 7: 1–52.
Hamilton, W. 1966. The moulding of senescence by natural selection. J. Theor. Biol., 12: 12–45.
Kenyon, C.J. 2010. The genetics of ageing. Nature, 464: 504–512.
Khaidakov, M., Siegel, E.R. and Reis, R.J.S. 2006. Direct repeats in mitochondrial DNA and
mammalian lifespan. Mech. Ageing Develop., 127: 808–812.
Kirkwood, T.B.L. and Melov, S. 2011. On the programmed/non-programmed nature of ageing
within the life history. Curr. Biol., 21: R701–R707.
Libertini, G. 2008. Empirical evidence for various evolutionary hypotheses on species demonstrating
increasing mortality with increasing chronological age in the wild. Sci. World J., 8: 182–193.
Linnane, A.W., Baumer, A., Maxwell, R.J., Preston, H., Zhang, C.F. and Marzuki, S. 1990.
Mitochondrial gene mutation: the ageing process and degenerative diseases. Biochem. Int., 22:
Loison, A., Festa-Bianchet, M., Gaillard, J., Jorgenson, J. and Jullien, J. 1999. Age-specific survival
in five populations of ungulates: evidence of senescence. Ecology, 80: 2539–2554.
Martinez, D. 1998. Mortality patterns suggest lack of senescence in hydra. Exp. Gerontol., 33:
Martins, A.C. 2011. Change and aging senescence as an adaptation. PLoS One, 6: e24328.
Medawar, P. 1952. An Unsolved Problem of Biology. London: H.K. Lewis.
Mitteldorf, J. 2004. Aging selected for its own sake. Evol. Ecol. Res., 6: 1–17.
Mitteldorf, J. 2006. Chaotic population dynamics and the evolution of ageing. Evol. Ecol. Res.,
Mitteldorf, J. 2010. Evolutionary origins of aging. In The Future of Aging: Pathways to Human Life
Extension (G.M. Fahy, M.D. West, L.S. Coles and S.B. Harris, eds.), pp. 87–126. New York:
Mitteldorf, J. and Pepper, J. 2009. Senescence as an adaptation to limit the spread of disease.
J. Theor. Biol., 260: 186–195.
Mitteldorf, J. and Wilson, D.S. 2000. Population viscosity and the evolution of altruism. J. Theor.
Biol., 204: 481–496.
Nussey, D.H., Kruuk, L.E.B., Morris, A., Clements, M.N., Pemberton, J.M. and Clutton-Brock,
T.H. 2009. Inter- and intrasexual variation in aging patterns across reproductive traits in a wild
red deer population. Am. Nat., 174: 342–357.
Otto, S.P. and Lenormand, T. 2002. Resolving the paradox of sex and recombination. Nature Rev.
Genet., 3: 252–261.
Pishel, I., Shytikov, D., Orlova, T., Peregudov, A., Artyuhov, I. and Butenko, G. 2012. Accelerated
aging versus rejuvenation of the immune system in heterochronic parabiosis. Rejuv. Res., 15:
Poulain, M., Pes, G.M., Grasland, C., Carru, C., Ferrucci, L., Baggio, G. et al. 2004. Identification
of a geographic area characterized by extreme longevity in the Sardinia island: the AKEA study.
Exp. Gerontol., 39: 1423–1429.
Price, G.R. 1970. Selection and covariance. Nature, 227: 520–521.
Reznick, D.N., Bryant, M.J., Roff, D., Ghalambor, C.K. and Ghalambor, D.E. 2004. Effect of
extrinsic mortality on the evolution of senescence in guppies. Nature, 431: 1095–1099.
Ricklefs, R.E. 1998. Evolutionary theories of aging: confirmation of a fundamental prediction, with
implications for the genetic basis and evolution of life span. Am. Nat., 152: 24–44.
Ricklefs, R.E. 2008. The evolution of senescence from a comparative perspective. Funct. Ecol.,
Rose, M.R., Rauser, C.L., Benford, G., Matos, M. and Mueller, L.D. 2007. Hamilton’s forces of
natural selection after forty years. Evolution, 61: 1265–1276.
Semeiks, J. and Grishin, N.V. 2012. A method to find longevity-selected positions in the mammalian
proteome. PLoS One, 7(6): e38595.
Sinclair, D.A. 2005. Toward a unified theory of caloric restriction and longevity regulation. Mech.
Ageing Develop., 126: 987–1002.
Skulachev, V.P. 2001. The programmed death phenomena, aging, and the Samurai law of biology.
Exp. Gerontol., 36: 995–1024.
Smith, J.M. 1976. Commentary: Group selection. Q. Rev. Biol., 51: 277–283.
Solomon, E., Berg, L. and Martin, D. 2002. Biology. Belmont, CA: Brooks-Cole Thompson
Taylor, P.D. 1992. Altruism in viscous populations – an inclusive fitness model. Evol. Ecol., 6:
Thum, T., Gross, C., Fiedler, J., Fischer, T., Kissler, S., Bussen, M. et al. 2008. MicroRNA-21
contributes to myocardial disease by stimulating MAP kinase signalling in fibroblasts. Nature,
Travis, J.M.J. 2004. The evolution of programmed death in a spatially structured population.
J. Gerontol. A: Biol. Sci. Med. Sci., 59: 301–305.
Vaupel, J.W., Baudisch, A., Dolling, M., Roach, D.A. and Gampe, J. 2004. The case for negative
senescence. Theor. Popul. Biol., 65: 339–351.
Weismann, A. 1889. Essays Upon Heredity and Kindred Biological Problems. Oxford: Clarendon
Williams, G.C. 1957. Pleiotropy, natural selection, and the evolution of senescence. Evolution, 11:
Wilson, D.S., Pollock, G.B. and Dugatkin, L.A. 1992. Can altruism evolve in purely viscous
populations. Evol. Ecol., 6: 331–341.
Wilson, E.O. 2005. Kin selection as the key to altruism: its rise and fall. Soc. Res., 72: 159–166.
Woodberry, O.G., Korb, K.B. and Nicholson, A.E. 2007. A simulation study of the evolution of
ageing. Evol. Ecol. Res., 9: 1077–1096.
The evolution of general altruistic ageing543
Page 18 Download full-text