how does cooperation evolve?

cooperation => group

evolution => natural selection

=> mechanism of evolution of cooperation is group selection

factors determining strength of group selection

● local fitness effects

genes which give the individual higher fitness are selected

● genetic structure groups are defined by the sharing genetic structure, i.e. cooperation

investigate the effects of

● varying ecology

● group selection + kin interaction VS group selection – kin interaction

● alarm calling VS restrained feeding

evolution of altruism by group selection(Pepper & Smuts 2000)

agent-based model

world

● 2D wrap around lattice

agents

● plant

● forager

model continued

plant behaviour

● grow

●linear

●logistic

● be consumed

linear

logistic

model continued

forager behaviour

● movement

same as sugarscape with vision = 1 and can move into any of 8 cells

● death

same as sugarscape with forager lifetime = infinity

● reproduction

reproduce asexually when energy >= fertility threshold

parent energy -= child initial energy

child born in cell closest to parent

model continued

cooperation

● alarm calling

● feeding restraint

model continued

targeted individual

Range around it in which foragers will give alarm calls

model continued

forager has 0.02 probability of being targeted

alarm callers will respond if within 5 cells of targeted forager

probability of kill = 1 / ( n + 1 )where n is the number of alarm callers

targeted forager can not make an alarm call

kill population = alarm callers + targeted forager

a random forager is chosen from the kill population

model continued

50%

99%

plantsize

Restrained feeders consumption = 0.5 * plant energy

Unrestrained feeders consumption = 0.99 * plant energy

model continued

model continued

patch width

patch gap width

pure population mixed population

uniform environment(one plant per cell)

patch width = 529

gap width = 0

alarm-caller

non-caller

uniform environment(one plant per cell)

patch width = 529

gap width = 0

pure population mixed population

restraint feeding

non-restraint feeding

discussion of results(pure population)

who cares

tells us nothing about between-group selection since there is only one group

discussion of results(mixed population)

local fitness effects

● group selection ignores suboptimisation problem within cooperative group (Heylighen 1997) fitness(non-cooperators) > fitness(cooperators)

genetic structure

● cooperative systems eroded from within by genetic competition (Campbell 1983)

mixed population => non-cooperative genes selected => local fitness and genetic structure effects not strong enough for group selection to occur

variable environment(mixed population)

population = 0.5 * alarm caller + 0.5 * non-alarm caller

variable environment(mixed population)

population = 0.5 * restraint feeder + 0.5 * non-restraint feeder

discussion of results(mixed population)

local fitness effects

● population size must be small (Futuyma 1986)

small patch width + high gap width => many small population groups

groups a

#(cooperators) >> #(non-cooperators)

groups b

all other groups fit into groups b

discussion of results(mixed population)

local fitness effects continued

● altruistic group has higher fitness due to synergy of cooperation (Heylighen 1997)

fitness(groups a) > fitness(group b)

discussion of results continued(mixed population)

genetic structure

● there can not be significant gene flow (Futuyma 1986, Goldstein & Zimmerman 2000)

● migration rates must be implausibly low (Ridley 1993)

low patch size + high gap width + low vision

=> low probability of migration => gene flow

=> reduced probability of non-cooperator infiltration of groups a

discussion of results continued(mixed population)

genetic variance continued

● successful groups must be able to export their local productivity from the local area (Wilson et al 1992)

patch full => steady emigration

fitness(cooperator) > (non-cooperator) => higher probability of successful colonisation for cooperators than non-cooperations

difficulty of migration => infiltration of non-cooperators low

=> local fitness and genetic structure effects are strong enough in some scenarios for group-selection => cooperation evolves

variable environment(mixed population + absence of kin assortment)

alarm calling never evolved in any of the 100 runs BUTrestraint feeding did

discussion of results(mixed population + absence of kin assortment)

local fitness

● alarm calling can only spread if foragers are heavily recompensated by others increasing their fitness relative to themselves (Wilson 1979, 1980)

recompensation comes through spatial association to cooperators

cooperators <=> kin

spatial association was removed largely by randomising birth locations

fitness(alarm callers) < fitness(population)

discussion of results(mixed population + absence of kin assortment)

local fitness continued

however,

feeding restraint conferred benefits as well as costs on the bearer

=> fitness(restraint feeders) > fitness(alarm-callers)

discussion of results continued(mixed population + absence of kin assortment)

genetic structure

● kin selection increases genetic selection between-groups and decreases it within-groups (Smith 1964)

spatial association <=> kin discrimination

randomised birth starting location

=> kin selection was not operating

=> selection between-groups was reduced

discussion of results continued(mixed population + absence of kin assortment)

genetic structure continued

migration rates must be implausibly low (Ridley 1993)

there can not be significant gene flow (Futuyma 1986, Goldstein & Zimmerman 2000)

random birth locations => mixed population => gene flow

=> non-cooperators selected over cooperatots

=> local fitness effects and genetic structure are not enough for between-group selection to occur for alarm callers

discussion of results continued(mixed population + absence of kin assortment)

genetic structure continued

however restraint feeders were selected when patch width low and gap-width high

small group size => restraint feeder becomes an increasing proportion of the acts recipients

=> kin selection was not needed

=> local fitness effects and genetic structure were strong enough for the evolution of feeding restraint

summary

evolution of cooperation

● favored by group-selection

diminshed by within-group selection

● evolution of cooperation is dependent on

ecological patchiness

small patches and large gaps stabilise

degree of migration

strong vs weak altruism

critique

● kin selection

there was no kin discrimination rule but the rule is defined in biology

● reproduction

reproduction was asexual and the offspring were the genetic clones of their parents whereas the rules of genetics are well established

● movement

movement rule had vision of 1 which made migration difficult if not impossible

critique continued

● model parameters

the starting population size was 40 which is small

the size of the world was not given, the assumption is x = y = 527 which is small

● death

foragers lived forever, a more realistic life expectancy was given in sugarscape

● simple

not a very sophisticated model

references

d. j. Futuyma, evolution biology, 1986

t. h. Goldsmith, w. f. Zimmerman, biology, evolution, and human nature, 2000

f. heylighen, http://pespmc1.vub.ac.be/COOPGEVO.html, genetic scenarios for evolving cooperation, 1997

j. w. Pepper, b. b. Smuts, the evolution of cooperation in an ecological context: an agent-based model, 2000

m. Ridley, evolution, 1993