Conference PaperPDF Available

Patch-level Selection in Darwinian Daisyworld

Authors:
Patch-level Selection in Darwinian Daisyworld
Matthew Bardeen
Universidad de Talca, Talca, Chile
mbardeen@utalca.cl
Abstract
Scientists have used Richard Dawkins’ ideas of the extended
phenotype to postulate levels of selection higher than an indi-
vidual in evolution. Dawkins rejects thisextension and insists
that there must be a reproductive bottleneck for the extended
phenotype, and thus, higher levels of selection to exist. In
this research, a model is presented that shows levels of se-
lection higher than the individual, without the reproductive
bottleneck insisted upon by Dawkins. A 2-dimensional cellu-
lar automata Daisyworld model is extended with a gene that
controls the rate of albedo mutation. A large number of runs
of the model are performed with a variety of different pa-
rameters, and the statistics for the runs are analyzed. The
results show that contrary to expectations, the mutation rate
does not stay low but instead rises to high levels. The reasons
for this are analyzed and it is shown that patch level selec-
tion pressures are acting upon the individuals. It is concluded
that selection pressures higher than the individual can exist,
mimicking the extended phenotype, without the need for a
reproductive bottleneck.
Introduction
The existence of multiple levels of selection in evolution has
been under much debate (Okasha, 2007; Sober and Wilson,
1999). Traditionally many biologists believed that selec-
tion could operate on a group of individuals of one species.
The justification for this belief was the apparent willing-
ness of one individual to put itself in danger for the good
of the group. However, in 1964 Hamilton published two ar-
ticles showing this behavior could be explained by a process
called kin selection, where individuals aid relatives based
on the probability of having shared genetic code (Hamilton,
1964a,b). Based on this work and others (Trivers, 1971),
Dawkins (1976) postulated the existence of the selfish gene,
describing a view where the gene is the unit of selection.
Genes, he argued, are inherently selfish favoring behav-
iors that serve to help them reproduce. A gene that inspired
its carrier to commit suicide before reproduction, for exam-
ple, would not survive very long in the gene pool. Genes
together in the body of an individual are forced to work to-
gether by virtue of having to pass through the same repro-
duction event, and are the vehicle of selection. Dawkins’
“vehicle of selection” is analogous to the “level of selection”
used by other authors (Okasha, 2007), a phrasing I will use
in this paper.
Dawkins (1982) later recognized that some genes have in-
fluences outside their bodies, a concept which he called the
extended phenotype. Here genes in one individual can be
tied to genes existing in other bodies by way of environmen-
tal modifications the classic example is the beaver dam,
where the genes for building & maintaining dams enhance
the survival of the immediate organism and others within its
colony. The genes still remain as the unit of selection, but, in
Dawkins’ terms, the group becomes the vehicle of selection.
Recently there has been debate on how far these effects ex-
tend beyond the organism and under what conditions (Bier-
naskie and Tyerman, 2005; Dawkins, 2004; Laland, 2004;
Jablonka, 2004; Turner, 2004; Whitham et al., 2003, 2005).
In particular, Dawkins (2004) insisted that there must be a
single reproductive event (a bottleneck) for all the genes in-
volved in the extended phenotype to force the genes to work
together.
Swenson et al. (2000) showed that it is possible for real
ecosystems to respond to artificial selection. They theo-
rized that such selection could happen in the natural world,
suggesting that small scale “microecosystems” could be se-
lected upon given the differential survival of such systems.
Also, they noted that discrete boundaries are not necessary
for an ecosystem to be a level of selection. The key is “lo-
calized interactions, such that one patch fares better than an-
other on the basis of its properties, even when the boundaries
between patches are fuzzy” (Swenson et al., 2000). Penn
and Harvey (2004) showed a similar response to artificial
selection in non-evolving artificial ecosystems.
In this paper I use a cellular automata Daisyworld model
to show the existence of patch-level selection and I demon-
strate that it arises from the transfer of heat across the planet.
I introduce a heritable albedo mutation rate to the daisy
genotypeand show that although its variationcannot be seen
on the individual level, it is subject to selection pressure.
This is because variations in the albedo mutation rate can be
seen by looking at groups of individuals in a larger popula-
tion and because these groups compete among themselves
for space.
To give a brief view of how this paper is organized, in
the followingsection I describe the basic ideas of the Daisy-
world model. I describe the model in more mathematical
terms in the Model Description section, and then describe
the experiments and show the results in graphical and nu-
merical forms in the Results section. The Discussion section
deals with the explanations and implications of the results
and is followed by the concluding remarks of the paper.
Methodology
Watson and Lovelock (1983) presented the Daisyworld
model to address some of the more prevalent doubts about
the Gaia theory. This model has also proved useful in study-
ing evolution under the assumption that organisms affect
their own environment (Dyke et al., 2007). Recently the
idea of niche construction, where organisms exert influence
on environment has gained prominence in the discussion
of evolutionary theory (Odling-Smee et al., 2003). While
this influence on environment has been acknowledged pre-
viously (Dawkins, 1976), the implications for evolutionary
theory have not been obvious (Bardeen, 2009).
Daisyworld was a toy-world, intended as a proof of con-
cept of the Gaia hypothesis, rather than a model of a real
physical system. The idea behind it was simple localized
interactions can affect global dynamics and generate home-
ostatic behavior. The model consisted of a “planet”, heated
by the sun and populated by black and white daisies. The
black daisies have a lower albedo (reflectiveness) than the
white daisies, and they absorb a greater amount of solar radi-
ation and raise the local temperature. The growth rate of the
daisies is linked to the local temperature, which is directly
influenced by albedo. This difference in growth rate causes
the area covered by black and white daisies to vary, causing
the overall temperature of the planet to vary in turn. This
creates a homeostatic response to external forces, such as
increasing incoming solar radiation (insolation) and keeping
the temperature of the planet relatively constant. This pro-
cess is mainly due to the niche construction aspects of the
individual daisies on their local environment.
I use a variant of the 2D cellular automata Daisyworld
model first described by von Bloh and Schellnhuber (1999).
The growth patterns of the daisies are given by a cellular
automata model. There is heat transfer between neighbor-
ing cells, so a daisy can affect its local neighborhood. This
model is useful in that all the effects seen are, by definition,
local. Any global effects that are seen must be emergent
properties of local interactions.
Another attraction of this model is the ability to “tune”
the diffusion rate, which permits experimentation with how
quickly and strongly effects of local daisies are transmitted
to their neighbors, and by extension, the globalenvironment.
This will allow me to quantify the probability that group-
level selection will arise in the system based on the diffusion
rate of heat across the planet.
To this model I add a gene that affects the mutation rate
of the daisy albedo. This gene will not affect the fitness of
individual daisies immediately, but will allow the effects of
a selection pressure at levels higher than an individual daisy.
There is biological evidence of different mutation rates be-
tween species, and even evidence of differential mutation
rates on the same genome (Wolfe et al., 1989), so this ex-
tension is not pure fantasy. When asked once about mu-
tation rates in natural systems, the eminent biologist John
Maynard-Smith replied that he expected them to be set as
lowas possible (Bedau and Seymour,1994; Maynard Smith,
1989). The reasoning is that, according to Travis and Travis
(2004): “..in constant environments, most mutations are
deleterious, hence mutation occurs at a low rate, constrained
only by the costs of error avoidance and error repair”.
The expected role of evolution by natural selection is that
of optimization and adaptation, and this should be no differ-
ent in the context of the Daisyworld (Ackland et al., 2003;
Ackland, 2004; Bardeen, 2009; St¨ocker, 1995).
Model Description
The base model for this article is an extended version of the
Daisyworld model described in von Bloh et al. (1997).
The temperature field T (x, y, t) is represented by the en-
ergy balance equation:
C
T (x, y, t)
t
= D
T
(
2
x
2
+
2
y
2
)T (x, y, t) (1)
σ
B
T (x, y, t)
4
+ S(1 A(x, y, t)),
where D
T
is the heat diffusion constant and A(x, y, t) is
the space/time distribution of albedo. The diffusion uses the
von Neumann neighborhood (the four adjacent neighbors to
the cell). S is the current solar radiation, and at S = 917
the albedo which producesthe optimal temperature for daisy
growth is around 0.53.
Growth patterns for the daisies are generated using a cel-
lular automata (CA) model. If a cell is empty, then there is
a chance that a daisy in a neighboring cell (Moore neighbor-
hood) will produce offspring in the empty cell. This chance
is based upon the temperature of that cell and given by
β(T ) =
4
(T
max
T
min
)
2
(T T
min
)(T
max
T ) (2)
where T
max
and T
min
are the maximum and minimum tem-
peratures at which the daisies can grow and T is the cur-
rent temperature of the cell. T
opt
is equivalent to
1
2
(T
min
+
T
max
). In this paper T
max
= 313 Kelvin and T
min
= 278
Kelvin, meaning T
opt
= 295.5 Kelvin.
The chance a daisy will die is given by:
γ(T ) = 1
4ρ
(T
max
T
min
)
2
(T T
min
)(T
max
T ) (3)
where ρ [0, 1] and serves set the base mortality rate. If ρ
is large, then the base mortality rate will be low.
Each daisy genotype consists of two floating-point val-
ues. The first is the color of the daisies albedo, which is in
the range of 0 : 1 inclusive and has a 50% chance of be-
ing changed at birth by adding a random value to the parent
albedo, drawn from a Gaussian distribution with a standard
deviation of r. If the parent albedo plus mutation falls out-
side the range of the albedo, the mutation is redrawn from
the Gaussian distribution.
The second value, r, is essentially the mutation rate of the
albedo, which is also in the range of 0 : 1 inclusive and is
also mutated at birth by adding a random value drawn from
a Gaussian distribution with a standard deviation of 0.001
(the mutation rate of the albedo mutation rate). If the parent
mutation rate plus the delta falls outside the range of the
albedo, r is redrawn from the same Gaussian distribution.
Results
The principal set of experiments in this chapter are designed
to test the long-term stable solution of the Daisyworld with a
heritable albedo mutation rate. To this end, a 200
2
cell world
is populated randomly (the chance of daisy in a given cell
is 10%) with daisies having uniform albedos of 0.53 (near
optimal for the starting insolation) and initial albedo muta-
tion rates of 0.01. This world is allowed to evolve for one
million timesteps with a constant incoming solar radiation
(S = 917), at which point the simulation is stopped. This
process forms one evolutionary run. Each run is repeated 50
times for each variation in parameter values; Tested are dif-
ferent diffusion constants (D
T
= 50, 100, 500, 1000, 1500,
2000, 2400). These experiments will show the adoption of a
high albedo mutation rate by most of the daisies under high
diffusion regimes.
Figure 1(a) shows the average planetary temperature over
two separate runs. In one, the planetary temperature oscil-
lates closely around the optimum for life on the planet. In
the other, the average planetary temperature climbs past the
optimum. The evolutionary trajectory of the mutation rate
shows the reason for this difference (Figure 1(b)). In the
first run the mutation rate stays low, as is expected, while in
the second the mutation rate climbs past 0.2. As the muta-
tion rate climbs, the average albedo of the planet drops (seen
in Figure 1(c)). This has the effect of increasing the average
mortality rate and decreasing the average birth rate of the
daisies.
A closer look (Figure 2) reveals that the mutation rate is
not uniform over the entire planet, but rather is limited to
patches of daisies.
Figure 3 shows that the average final mutation rate is in-
fluenced both by the diffusion rate of heat between the cells
and the base mortality rate. Higher diffusion results in a
Figure 2: Snapshots of world state from one run, Insolation
L = 1.0, Diffusion D
T
= 1500, Mortality rate is 10% and
the mortality model is the variable mortality model. Black
is lower mutation, white is higher mutation, blue is dead.
World is mostly dominated by low mutation daisies with well
defined patches of high mutation daisies
0
0.1
0.2
0.3
0.4
0.5
0 500 1000 1500 2000 2500
Albedo mutation rate
Diffusion Rate
20% Mortality
15% Mortality
10% Mortality
5% Mortality
Figure 3: Average final mutation rate of 50 runs, each con-
sisting of 2 million timesteps.
290
292
294
296
298
300
302
Temperature
D
T
= 50
Planetary Temperature
Extremas
Optimum
290
292
294
296
298
300
0 2000 4000 6000 8000 10000
Timesteps (x100)
D
T
= 2400
(a) Steady state temperatures with a variable albedo mutation rate.
0
0.05
0.1
0.15
0.2
0.25
0.3
Mutation Rate
D
T
= 50
0
0.05
0.1
0.15
0.2
0 2000 4000 6000 8000 10000
Timesteps (x100)
D
T
= 2400
(b) Average albedo mutation rate.
0.48
0.5
0.52
0.54
0.56
Average Albedo
D
T
= 50
0.48
0.5
0.52
0.54
0 2000 4000 6000 8000 10000
Timesteps (x100)
D
T
= 2400
(c) Average albedo.
Figure 1: Evolution of planetary average temperature, average albedo mutation rate, and average albedo for 2 runs over 1
million timesteps. Mortality rate of 10%, Grid size is 200
2
.
0
0.5
1
1.5
2
2.5
3
3.5
4
100 150 200 250 300 350 400 450 500
Standard Deviation
Timestep (x100)
Low mutation, low diffusion
High mutation, low diffusion
Low mutation, high diffusion
High mutation, high diffusion
Figure 4: Standard deviation of temperature across the cells
of the planet. Under high diffusion, the high mutation rate
daisies present a much more uniform environment than the
low mutation rate daisies. Under low diffusion, the temper-
ature is more uniform for the high mutation rate daisies than
the low mutation rate daisies, but not to the same extent.
(High Diffusion D
T
= 2400, Low diffusion D
T
= 50, Base
mortality rate = 15%, High mutation = 0.4, Low mutation =
0.01)
Table 1: Average number of children over 25 runs, Low mu-
tation = 0.01, High mutation = 0.4
High Diffusion Low Diffusion
D
T
= 2400 D
T
= 50
Low Mutation 0.424 0.420
High Mutation 0.417 0.410
higher final mutation rate, as does a higher base mortality
rate.
Inspecting the average number of children for runs with
fixed low (0.01) and high (0.4) albedo mutation rates shows
that there is little difference between the average number of
children for both (Table 1). The major difference between
the two strategies is that the temperature across the planet
has less variance under the high mutation daisies than un-
der the low mutation daisies (Figure 4). Under low diffu-
sion rates the difference between the variance in tempera-
ture caused by the two strategies is much less than under
high diffusion rates.
Discussion
The results of the experiments leave some questions:
What is the cause of the increase in temperature?
What are the implications of a high mutation rate?
Why would a high mutation rate be a selective advantage
under certain circumstances and not under others?
Is the selective advantage caused by an individual level
selection pressure or a higher level selection pressure?
In this section, I will answer these questions in turn, then
discuss the wider implications of the answers.
What is the cause of the increase in temperature?
The plots of temperature and the mutation rate (Figures 1(a)
and 1(b)) showthat the increase in temperatureis linkedwith
that of mutation rate, however it does not reveal the cause.
Inspecting the snapshot of the planet state shows that, un-
surprisingly, the albedos are very diverse when the mutation
rate is high.
In low diffusion environments, the heat is mainly retained
within a single daisy cell and there is little transfer to other
cells. In high diffusion environments the heat flows freely
across the cells, and a group of daisies with random albe-
dos will appear, at a higher level, to have the temperature
of a single gray daisy with an albedo of around 0.5. Thus,
as more daisies adopt the high mutation rate strategy, the
average albedo of the planet becomes closer to 0.5 and the
temperature rises away from the optimum.
What are the effects of a high mutation rate?
A high mutation rate causes a number of changes to the sys-
tem. Comparing two planets, one with a fixed high muta-
tion rate and one with a fixed low mutation rate, shows that
the high mutation rate planet has a lower average growth
rate, a higher average death rate, and lower number average
number of children per daisy. Another notable difference is
that the standard deviation of the cell temperature across the
planet is lower on the high mutation planet.
The high mutation rate also affects the heredity of the
daisy albedo. Lewontin (1978) gives the necessary condi-
tions of natural selection as: individuals within a species
differ, this variation is heritable, different variants leave dif-
fering amounts of offspring, and variations that favor an in-
dividual’s reproductivesuccess will be preserved. In the sys-
tems with a high albedo mutation rate, the selection of indi-
vidual daisies seems to fail on the second of these principles
the variation in albedo does not seem heritable from parent
to offspring.
Furthermore, it is unclear how the albedo mutation rate is
being selected upon. Identifying the variation between the
albedos of two individuals is easy. Identifying the variation
between the albedo mutation rate of two individuals is much
harder. The only way to measure this variation would be
to look at the range of variability in the albedos of the re-
spective offspring. However, with only an average of 0.4
offspring per parent, the quality of this measure for natural
selection is limited. Thus the variation must be seen either
above the level of the individual or over a large period of
time.
Why would a high mutation rate be a selective
advantage?
The previous subsection highlights a crucial question – why
would a high albedo mutation rate be a selective advantage
for the daisies bearing it, but only under certain circum-
stances? From the fact that the high mutation rate daisies
are only seen primarily under high diffusion rate worlds, we
can discard the idea that there is an unintended systemic ef-
fect from, for example, bias in the mutation operator (Bul-
lock, 2001). If there was such a systemic effect, it would
be seen in all parameter ranges, and not only under certain
conditions.
This leaves non-systemic causes to blame. Figure 2 shows
the existence of patches of daisies with similar mutation
rates. With low mutation rates, patches of daisies with sim-
ilar albedos are likewise seen. This is due to the cellular
automata rules, since a new daisy can only be born next to
another living daisy, they tend to clump into patches of sim-
ilar daisies.
This phenomenon is a hindrance when the albedos are
near identical patches that have albedos higher or lower
than the optimal are inherently unstable. They become too
hot or too cold and die off, replaced by daisies bearing albe-
dos which are more suitable to the changed environment.
When they die, not only is their albedo gene lost, but their
albedo mutation rate gene is lost too.
Having patches of daisies with highly variable albedos
means that the patch temperature stays relatively constant,
though not optimal. Less environmental change in the daisy
patch signifies less change is needed by the genome. In this
case, the gene for albedo mutation rate “uses” the albedo
gene as a buffer between it and the environment. Its repro-
ductive environment becomes more stable and, as a result,
the high albedo mutation rate gene lasts longer – unchanged
– within the gene pool.
The experiments with fixed mutation rates support this
conclusion – the variability of the temperature on the planet
populated with a fixed high mutation rate is much less than
that of a planet with fixed low mutation rate daisies (Figure
4). It can be assumed that this same phenomena is seen on a
smaller scale within the patches.
Is the selective advantage an individual level or a
higher level selection pressure?
Now the question becomes on what level is the selective ad-
vantage operating – is it an individual level pressure or some
pressure operating on a higher level? Williams (1966) gives
the following guide: “Do these processes show an effective
design for maximizing the number of descendants of the in-
dividual, or do they show an effective design for maximizing
the number, rate of growth, or numerical stability of the pop-
ulation or larger system?”.
The unit of selection here is surely an individual daisy.
Daisies do not reproduce at the same time, nor do they share
genetic information with one another. However, the level at
which the selection pressure is operating is not clear.
If it was an individual level pressure, we would expect
to see the maximization of birth rate, the minimization of
the death rate, or a higher number of children born per in-
dividual. For this to happen, they should be at their optimal
albedo, since that will maximize their chances of producing
offspring and minimize their chances of dying. Likewise,
the mutation rate should be very low. As seen, this is indeed
the case under low diffusion environments.
However under high diffusion rates, we see the average
mutation rate start to rise, for the reasons discussed prior.
Patches
However, the previous explanation leaves a conundrum: If
having highly variablealbedos is a such a wise strategy, then
why do patches of high and low mutation daisies appear on
the planet at the same time, as seen in Figure 2? Why don’t
all the daisies convert to high mutation rates?
The answer is that patches of daisies compete among
themselves those that are more successful at maintain-
ing the high mutation rate gene have slower growth rates,
but higher gene stability. Thus the incidence of daisies with
high albedo mutation rates tends to increase within the pop-
ulation. But low mutation rate daisies with near optimal
albedos occasionally find purchase with their higher repro-
ductive rates and lower death rates, creating patches of their
own.
Thus there is competition between the two strategies on
the basis of their effect on the environment. Individual
daisies are linked to others by means of their geographical
vicinity. When the diffusion is high, those links are stronger
than when it is low. In low diffusion environments, daisies
with high albedo mutation rates are not competitive with
those that have low mutation rates.
Frank (1996) pointed out that in parasitism, we often see
selection between kin groups at high levels, but competi-
tion between individuals in lower levels. He says that “In
the population of parasites within the host, a mutant parasite
with a faster growth rate will usually increase in frequency.
However if this growth rate causes host death before trans-
mission to neighboring hosts, the effective long-term fitness
of the mutant is non-existent. Thus there is a balance be-
tween exploitation of resources (individual level selection)
and cooperation (kin group selection).
This is essentially what is happening in the Daisyworld
model described here competition between kin groups
leads to cooperation within the group in some cases. This
immediately calls to mind the example most used for the ex-
tended phenotype – beaver dams. Beaver dams are typically
shared between kin groups. Beaver families that build better
dams in more advantageous places are more successful than
those that do not. The shared phenotype in this example is
the dam. But can the beaver kin group be thought of as an
organism or vehicle of selection?
Central to the idea of the organism in the extended phe-
notype was the insistence that there be a reproductive bot-
tleneck (Dawkins, 1982, 2004) to force cooperation. This
insistence can be seen again in Frank (1996). In the beaver
example, new dams are often created by a single breeding
pair – a reproductive bottleneck.
However, here the daisies reproduce in a random fashion
- there is no bottleneck. So what forces the cooperation be-
tween the daisies and causes high mutation rates? It can
be nothing more than the shared environment of the daisies.
The high diffusion rate links the fate of one daisy to the fate
of its neighbors, forcingcooperation. This is why high muta-
tion rates are only seen in high diffusion rate environments.
Furthermore. this is why the extended phenotype is
not limited by the reproductive bottleneck described by
Dawkins. If there is a tight enough coupling between differ-
ent organisms, such that the the fate of one is linked to the
fate of another, they will evolve as a group, rather than in-
dividuals. Further work is necessary to quantify how strong
the linkage needs to be and under what conditions this link-
age can come about.
Conclusion
The results of this study can be generalized relatively easily
– the “mutation rate” here really refers to the rate of pheno-
typic change in the daisies in comparison to the change in
environment. The diffusion rate is analogous to the impact
an individual has on its neighbors and competitors. With
high diffusion rates, the influence of individual daisies on
their local temperatureis minimal. The variation part of evo-
lution as seen from the planetary perspective is no longer
one daisy, but clumps of daisies, since that is where most
of the phenotypic variation lies. Conversely when the dif-
fusion rate is low, the focus of evolution is on individual
daisies, since the selection method (birth rate/death rate) is
very much dependent on the individual daisy phenotype.
This idea has important consequences in evolution. One
can imagine how natural selection works on all levels si-
multaneously. Micro-organisms (like soil fungi) would be
subject to individual level selection at their own level since
their effects are more immediate and diffuse slowly in com-
parison to their reproductive speed. From higher levels (i.e.,
from a forest level) they could be selected upon as groups
since their effects appear to diffuse rapidly in relation to the
reproductive speed of other organisms at the higher level.
In this work I have demonstrated the existence of patch-
level selection upon individuals in a model world. Neces-
sary conditions for this development were: a spatial struc-
ture, the modification of local environment by individuals,
the transmission of local effects to neighboring organisms,
and a gene that controls the rate of change in the phenotypic
property that modifies the local environment. These nec-
essary conditions can be found, without great difficulty, in
nature.
Furthermore, it shows that the reproductive bottleneck in
Dawkins ideas of the extended phenotype is more strict than
it needs to be. All that is really needed is the existence of
some force linking the fate of the genes in one organism to
the fate of genes in another. And if this is the case, then
the arguments presented by Laland (2004), Turner (2004),
and Whitham et al. (2003) for the extension of the extended
phenotype do indeed hold merit.
Acknowledgements
I would like to thank Narciso Cerpa and the three anony-
mous reviewers for their comments and suggestions they
have resulted in a vast improvement in the arguments of the
paper.
References
Ackland, G. (2004). Maximization principles and Daisy-
world. Journal of Theoretical Biology, 227:121–128.
Ackland, G., Clark, M., and Lenton, T. (2003). Catastrophic
desert formation in Daisyworld. Journal of Theoretical
Biology, 223:39–44.
Bardeen, M. (2009). Lessons from Daisyworld: Survival of
the Stable. Dphil, University of Sussex.
Bedau, M. A. and Seymour, R. (1994). Adaptation of Muta-
tion Rates in a Simple Model of Evolution. In Stonier,
R. J. and Yu, X. H., editors, Complex Systems: Mech-
anism of Adaptation, pages 37–44. IOS Press, Amster-
dam.
Biernaskie, J. and Tyerman, J. (2005). The overextended
phenotype.
´
Ecoscience, 12:3–4.
Bullock, S. (2001). Smooth Operator? Understanding
and Visualising Mutation Bias, volume 2159 of Lec-
ture Notes in Computer Science. Springer Berlin Hei-
delberg, Berlin, Heidelberg.
Dawkins, R. (1976). The Selfish Gene. Oxford University
Press.
Dawkins, R. (1982). The Extended Phenotype. Oxford Uni-
versity Press.
Dawkins, R. (2004). Extended Phenotype - But Not Too Ex-
tended. A Reply to Laland, Turner, Jablonka. Biology
and Philosophy, 19:377–396.
Dyke, J., McDonald-Gibson, J., Di Paolo, E., and Harvey, I.
(2007). Increasing complexity can increase stability in
a self-regulatingecosystem. LectureNotes in Computer
Science, 4648:133.
Frank, S. A. (1996). Models of Parasite Virulence. The
Quarterly Review of Biology, 71(1):37.
Hamilton, W. D. (1964a). The genetical evolution of social
behaviour. I. Journal of Theoretical Biology, 7(1):1–
16.
Hamilton, W. D. (1964b). The genetical evolution of social
behaviour. II. Journal of Theoretical Biology, 7(1):17–
52.
Jablonka, E. (2004). From Replicators to Heritably Varying
Phenotypic Traits: The Extended Phenotype Revisited.
Biology & Philosophy, 19(3):353–375.
Laland, K. N. (2004). Extending the Extended Phenotype.
Biology & Philosophy, 19(3):313–325.
Lewontin, R. C. (1978). Adaptation. Scientific American,
239:212–228.
Maynard Smith, J. (1989). The Limitations of Evolution-
ary Theory. In Maynard Smith, J., editor, Did Darwin
Get It Right?, pages 180–191. Chapman and Hall, New
York.
Odling-Smee, J., Laland, K., and Feldman, M. (2003). Niche
Construction: The Neglected Process in Evolution.
Princeton University Press.
Okasha, S. (2007). Evolution and the Levels of Selection.
Oxford University Press, New York, NY, USA.
Penn, A. and Harvey, I. (2004). The Role of Non-Genetic
Change in the Heritability, Variation, and Response to
Selection of Artificially Selected Ecosystems. In Pol-
lack, J., Bedau, M., Husbands, P., Ikegami, T., and Wat-
son, R. A., editors, Proceedings of the Ninth Interna-
tional Conference on the Simulation and Synthesis of
Living Systems, Alife IX. MIT Press.
Sober, E. and Wilson, D. S. (1999). Unto others: The evo-
lution and psychology of unselfish behavior. Harvard
University Press, Harvard, MA, first pape edition.
St¨ocker, S. (1995). Regarding Mutations in Daisyworld
Models. Journal of Theoretical Biology, 175:495–501.
Swenson, W., Wilson, D. S., and Elias, R. (2000). Artifi-
cial ecosystem selection. Proceedings of the National
Academy of Sciences of the United States of America,
97(16):9110–4.
Travis, E. R. and Travis, J. M. J. (2004). Mutators in space:
the dynamics of high-mutability clones in a two-patch
model. Genetics, 167(1):513–22.
Trivers, R. (1971). The evolution of reciprocal altruism. The
Quarterly Review of Biology, 46(1):35.
Turner, J. S. (2004). Extended Phenotypes and Ex-
tended Organisms. Biology and Philosophy, (Dawkins
1982):327–352.
von Bloh, W., Block, A., and Schellnhuber, H. (1997). Self
stabilization of the biosphere under global change: A
tutorial geophysiological approach. Tellus, 49B:249–
262.
von Bloh, W. and Schellnhuber, H. (1999). Tutorial Mod-
elling of geosphere-biosphere interactions: the effect
of percolation-type habitat fragmentation. Physica A,
266:186–196.
Watson, A. and Lovelock, J. (1983). Biological homeostasis
of the global environment: The parable of Daisyworld.
Tellus, 35B:286–289.
Whitham, T., Lonsdorf, E., Schweitzer, J., Bailey, J., Fis-
cher, D., Shuster, S., Lindroth, R., Hart, S., Allan, G.,
Gehring, C., Keim, P., Potts, B., Marks, J., Rehill, B.,
DiFazio, S., LeRoy, C., Wimp, G., and Woolbright,
S. (2005). All effects of gene on the world”: Ex-
tended phenotypes, feedbacks, and multi-level selec-
tion.
´
Ecoscience, 12:5–7.
Whitham, T., Young, W., Martinsen G. Gehring, C.,
Schweitzer, J., Shuster, S., Wimp, G., Fischer, D.,
Bailey, J., Lindroth, R., Woolbright, S., and Kuske,
C. (2003). Community and Ecosystem Genetics: A
Consequence of the Extended Phenotype. Ecology,
84:559–573.
Williams, G. C. (1966). Adaptation and Natural Selection.
Princeton: Princeton University Press.
Wolfe, K., Sharp, P., and Li, W. (1989). Mutation rates dif-
fer among regions of the mammalian genome. Nature,
337:283–285.
ResearchGate has not been able to resolve any citations for this publication.
Article
Full-text available
Several evolutionary processes influence virulence, the amount of damage a parasite causes to its host. For example, parasites are favored to exploit their hosts prudently to prolong infection and avoid killing the host. Parasites also need to use some host resources to reproduce and transmit infections to new hosts. Thus parasites face a tradeoff between prudent exploitation and rapid reproduction-a life history tradeoff between longevity and fecundity. Other tradeoffs among components of parasite fitness also influence virulence. For example, competition among parasite genotypes favors rapid growth to achieve greater relative success within the host. Rapid growth may, however, lower the total productivity of the local group by overexploiting the host, which is a potentially renewable food supply. This is a problem of kin selection and group selection. I summarize models of parasite virulence with the theoretical tools of life history analysis, kin selection, and epidemiology. I then apply the theory to recent empirical studies and models of virulence. These applications, to nematodes, to the extreme virulence of hospital epidemics, and to bacterial meningitis, show the power of simple life history theory to highlight interesting questions and to provide a rich array of hypotheses. These examples also show the kinds of conceptual mistakes that commonly arise when only a few components of parasite fitness are analysed in isolation. The last part of the article connects standard models of parasite virulence to diverse topics, such as the virulence of bacterial plasmids, the evolution of genomes, and the processes that influenced conflict and cooperation among the earliest replicators near the origin of life.
Article
Full-text available
A model is presented to account for the natural selection of what is termed reciprocally altruistic behavior. The model shows how selection can operate against the cheater (non-reciprocator) in the system. Three instances of altruistic behavior are discussed, the evolution of which the model can explain: (1) behavior involved in cleaning symbioses; (2) warning cries in birds; and (3) human reciprocal altruism. Regarding human reciprocal altruism, it is shown that the details of the psychological system that regulates this altruism can be explained by the model. Specifically, friendship, dislike, moralistic aggression, gratitude, sympathy, trust, suspicion, trustworthiness, aspects of guilt, and some forms of dishonesty and hypocrisy can be explained as important adaptations to regulate the altruistic system. Each individual human is seen as possessing altruistic and cheating tendencies, the expression of which is sensitive to developmental variables that were selected to set the tendencies at a balance ap...
Chapter
There are a lot of things we do not know about evolution, but they are not the things that non-biologists think we do not know. If I admit to a non-biological colleague that evolution theory is inadequate, he is likely to assume at once that Darwinism is about to be replaced by Lamarckism and natural selection by the inheritance of acquired characters. In fact, nothing seems to me less likely. In common with almost everyone working in the field, I am an unrepentant neo-Darwinist. That is, I think that the origin of evolutionary novelty is a process of gene mutation which is non-adaptive, and that the direction of evolution is largely determined by natural selection. I am enough of a Popperian to know that this is a hypothesis, not a fact, and that observations may one day oblige me to abandon it, but I do not expect to have to. Indeed, everything that has happened during my working life as a biologist, and in particular the development of molecular biology, has strengthened rather than weakened the neo-Darwinist position.
Book
The seemingly innocent observation that the activities of organisms bring about changes in environments is so obvious that it seems an unlikely focus for a new line of thinking about evolution. Yet niche construction--as this process of organism-driven environmental modification is known--has hidden complexities. By transforming biotic and abiotic sources of natural selection in external environments, niche construction generates feedback in evolution on a scale hitherto underestimated--and in a manner that transforms the evolutionary dynamic. It also plays a critical role in ecology, supporting ecosystem engineering and influencing the flow of energy and nutrients through ecosystems. Despite this, niche construction has been given short shrift in theoretical biology, in part because it cannot be fully understood within the framework of standard evolutionary theory. Wedding evolution and ecology, this book extends evolutionary theory by formally including niche construction and ecological inheritance as additional evolutionary processes. The authors support their historic move with empirical data, theoretical population genetics, and conceptual models. They also describe new research methods capable of testing the theory. They demonstrate how their theory can resolve long-standing problems in ecology, particularly by advancing the sorely needed synthesis of ecology and evolution, and how it offers an evolutionary basis for the human sciences. Already hailed as a pioneering work by some of the world's most influential biologists, this is a rare, potentially field-changing contribution to the biological sciences.
Article
Does natural selection act primarily on individual organisms, on groups, on genes, or on whole species? This book provides a comprehensive analysis of the long-standing controversy in evolutionary biology over the levels of selection, focusing on conceptual, philosophical, and foundational questions. In the first half of the book, a systematic framework is developed for thinking about natural selection acting at multiple levels of the biological hierarchy; the framework is then used to help resolve outstanding issues. Considerable attention is paid to the concept of causality as it relates to the levels of selection, particularly the idea that natural selection at one hierarchical level can have effects that 'filter' up or down to other levels. Full account is taken of the recent biological literature on 'major evolutionary transitions' and the recent resurgence of interest in multi-level selection theory among biologists. Other biological topics discussed include Price's equation, kin and group selection, the gene's eye view, evolutionary game theory, selfish genetic elements, species and clade selection, and the evolution of individuality. Philosophical topics discussed include reductionism and holism, causation and correlation, the nature of hierarchical organization, and realism and pluralism about the levels of selection.