what is evolutionary fitness for?
TRANSCRIPT
WHAT IS EVOLUTIONARYFITNESS FOR?
Mississippi State University, 11/6/2015
Charles H. Pence
Department of Philosophyand Religious Studies
NATURALSELECTION
t = 0
t = 1 t = 2
t = 0 t = 1
t = 2
t = 0 t = 1 t = 2
FITNESS
>
>
>
Orange organismsleave more offspringthan teal organisms.
A circle: the tautology problem
Orange organismsleave more offspringthan teal organisms.
A circle: the tautology problem
Orange organisms willprobably (are disposed to)leave more offspring than
teal organisms.
THE PROPENSITYINTERPRETATIONOF FITNESS
TWONOTIONSOF FITNESS
Matthen and Ariew (2002)
[F]or many this notion of an organism’s overallcompetitive advantage traceable to heritable traits isat the heart of the theory of natural selection.Recognizing this, we shall call this measure of anorganism’s selective advantage its vernacularfitness. According to one standard way ofunderstanding natural selection, vernacular fitness –or rather the variation thereof – is a cause ofevolutionary change. (56)
Matthen and Ariew (2002)
Fitness occurs also in equations of populationgenetics which predict, with some level ofprobability, the frequency with which a gene occursin a population in generation n+ 1 given itsfrequency in generation n. In population genetics,predictive fitness (as we shall call it) is a statisticalmeasure of evolutionary change, the expected rateof increase (normalized relative to others) of agene … in future generations…. (56)
Causal (vernacular) fitness: general(causal) notion in natural selection
Predictive (mathematical) fitness:predict future representation fromcentral tendency/expected value
THECLAIM
Causal fitness can be made tosurvive counterexamples against it,
but at a cost.
It’s not clear just what predictivefitness is supposed to predict.
CAUSALFITNESS
The basic idea: Define the propensityinterpretation in terms of facts aboutthe possible lives an organism (with
a given genotype, in a givenenvironment) could have lived.
F(G, E) = exp( limt→∞
1t ∫ω∈Ω
Pr(ω) ⋅ ln(ϕ(ω, t)) dω)
• Multi-generational life histories
• Changing genotypes and environmentsover time
• Disposition (propensity) defined overmodal facts about other possible livesof organisms
F(G, E) = exp( limt→∞
1t ∫ω∈Ω
Pr(ω) ⋅ ln(ϕ(ω, t)) dω)
• Multi-generational life histories
• Changing genotypes and environmentsover time
• Disposition (propensity) defined overmodal facts about other possible livesof organisms
F(G, E) = exp( limt→∞
1t ∫ω∈Ω
Pr(ω) ⋅ ln(ϕ(ω, t)) dω)
• Multi-generational life histories
• Changing genotypes and environmentsover time
• Disposition (propensity) defined overmodal facts about other possible livesof organisms
F(G, E) = exp( limt→∞
1t ∫ω∈Ω
Pr(ω) ⋅ ln(ϕ(ω, t)) dω)
• Multi-generational life histories
• Changing genotypes and environmentsover time
• Disposition (propensity) defined overmodal facts about other possible livesof organisms
Can plausibly be saved fromcounterexamples
Results in a potentialmetaphysical mess
PREDICTIVEFITNESS
What is our inferential basisfor determining the values of
predictive fitness?
“Darwinian fitness” in basic populationgenetics:ptqt = wt ⋅ p0q0
Expected number of offspring:
A(O, E) =∑ P(QOEi )QOE
i
Fitness Property Inferential Basis Sample
Individual fitness,relativizing toenvironmentalconditions
One individuallife-history
Very small, un-representative
Individual fit-ness, includingsimilar/clonalorganisms
A small numberof life-histories insimilar environ-mental conditions
Small, likely un-representative
Fitness Property Inferential Basis Sample
Trait fitness,including envi-ronmental andpleiotropic effects
One trait-history Very small, un-representative
Trait fitness, in-cluding similartraits
A small number oftrait-histories insimilar environ-mental conditions
Small, likely un-representative
Fitness Property Inferential Basis Sample
Type fitness, natu-ral populations
A moderatenumber of type-histories in similarenvironmentalconditions
Moderately-sized, possiblyrepresentative
Type fitness,experimentalevolution
A huge number oftype-histories innearly identicalenvironmentalconditions
Large and rep-resentative,high-qualitypredictions
Best-case (long-termexperimental evolution): great
inferential basis
Almost all natural populations:poor inferential basis
Another test case: chaoticpopulation dynamics
Assumption of most models offitness: non-chaotic population
dynamics
Question: How common isnon-chaotic dynamics in evolving
systems?
Approach of Doebeli & Ispolatov (2014):Investigate by simulating populations with
two features:
1. Density-dependent selection pressures
2. High-dimensional phenotype space
“Our main result is that the probability ofchaos increases with the dimensionality dof the evolving system, approaching 1 ford ∼ 75. Moreover, our simulations indicatethat already for d ≳ 15, the majority ofchaotic trajectories essentially fill out the
available phenotype space overevolutionary time….” (D&I, 1368)
-100 0 100
-100
0
100
x1
x2
Doebeli and Ispolatov (2014), fig. 5
Surely there’s no way to definepredictive fitness in these
scenarios?
“The invasion is exponential, but nonlineardynamics of the resident type produce
fluctuations around this trend. [Fitness] cantherefore be most accurately estimated bythe slope of the least squares regression of[daughter population size] on t.” (Grant
1997)
Chaotic population dynamics:Common, and render
predictive fitness meaningless
Predictive fitness isn’t verypredictive after all
THE MORAL
Causal fitness can be savedfrom counterexamples at the
cost of being mademetaphysically problematic
Predictive fitness … isn’t
Many uses of fitness:
• Mathematical parameter in models
• Causal property
• Proxies for strength of selection inpopulations
• Statistical estimator for any of theabove
Fitness concepts are farmore complex than a
dichotomy between twosimple roles for fitness.
[email protected]://charlespence.net
@pencechp