Evolution by natural selection
Natural selection requires phenotypic variation
Selection occurs if variation in fitness is associated with variation in the phenotype (i.e., there is a covariance between fitness and phenotype)
Evolution by natural selection can occur if part of the phenotypic variance is heritable (i.e., there is a covariance between phenotype and genotype)
1 Evolution by natural selection
Directional Selection (change in population mean and reduction in VA)
Non-linear Selection (stabilizing)
(reduction in VA)
Non-linear Selection (disruptive)
(increase in VA)
Prediction: Selection reduces the standing genetic variation in populations
Collared flycatcher (Ficedula albicollis)
Gustaffson (1986)
2 Prediction: Selection reduces the standing genetic variation in populations
Across species comparison
Roff (1997)
Maintenance of heritable variation
EquilibriumEquilibrium betweenbetween newlynewly emergingemerging geneticgenetic variabilityvariability andand selectionselection removingremoving geneticgenetic variantsvariants fromfrom thethe population.population.
Population-,Population-, interindividualinterindividual variability variability
Mutation Selection Migration
Selection differential: Covariance between fitness and phenotype
The „Price Equation“ (Price, 1970) („Robertson-Price Identity“: Lynch and Walsh,1998)
2 Δzi = cov(wi , zi ) = h S = βVa = βG
Δz = evolutionary change from generation t to t+1 (equivalent to R in breeders equation) i = index-subscript for a particular phenotype/trait β = selection gradient (slightly different from selection differential)
G = additive genetic variance (equivalent to VA) ⎛W ⎞ w = relative fitness ⎜ i ⎟ i ⎝W⎠
Directional Selection
Linear Regression Directional Selection differential (S): within-generational difference y = a + bx between mean phenotype after episode of selection (but before reproduction) and the mean before the selection episode (sometimes: Linear Fitness Equation “total selection”). Equivalent to w regression of relative fitness on phenotype. Only for directional selection. w = a + βz Selection gradient (β, γ): partial regression coefficients of relative z fitness on a given trait (sometimes: a: baseline fitness “direct selection”). For directional, β: directional selection gradient non-linear and correlational z: phenotypic trait selection.
4 Directional Selection
V G S R = h2S = a S = S = G = Gβ univariate Vp P P
Δz = GP−1S = Gβ multivariate hence −1 S ⎯⎯→⎯ β = P S = (the bold upright font P indicates matrix notation)
Non-linear Selection
Quadratic Regression y = a + bx + cx2
1
Non-linear fitness equation 0.8 1 0.6 w = a + βx + γx2 0.4 2 0.2 2 4 6 8 10 Linear component Non-linear (quadratic) (gradient) component (gradient)
The non-linear selection gradient does not directly affect the change in the phenotypic mean from one generation to the next (Δz ). But it affect the genetic co-/variances. Lande & Arnold 1983
5 Evolution by natural selection
Of selective targets PhenotypicPhenotypic traitraitt underunder selectionselection
and agents Ecological/socialEcological/social factorfactor thatthat generatesgenerates aa covariancecovariance betweenbetween fitnessfitness andand phenotypephenotype
Evolution by natural selection
Agent 001 = Agent 002 = parasitoid Bird predation
Goldenrod = Goldrute
(Target)
Eurosta gall fly
6 Example non-linear Selection
Stabilizing selection?
Goldenrod = Goldrute
Eurosta fly Eurosta gall
Eurosta gall fly
Directional and Stabilizing Selection
Fitness function Phenotype distribution in the population
StabilizingStabilizing andand directionaldirectional selectionselection areare notnot mutuallymutually exclusiveexclusive
7 Directional and Stabilizing Selection
Fitness function Phenotype distributions in the population β = 0 β > 0 β > 0 β = 0 β < 0 γ = 0 γ = 0 γ < 0 γ < 0 γ ≥ 0 w
Phenotypic trait value
Natural/Sexual Selection
Experimental Design
8 Definitions of Fitness
• Strict definition: A genotype‘s rate of increase in a population (e.g. Futuyma 1986)
• Working definitions: • Fitness components 1. Individual survival (longevity) 2. Fecundity (clutch size, hatching success, number of surviving offspring, number or reproducing offspring, fedundity of offspring, number of grand-children, ...) 3. Insemination rate 4. .... Often specific to the biology of the study organism.
Example: Estimating selection differentials and gradients Fitness (lifetime nr. of nr. of (lifetime inseminations)
Horn length (mm) Elytra length (mm)
Fungus beetle Bolithotherus cornutus
9 OnOn thethe „+“„+“ side side •• Selection Selection isis observedobserved asas itit isis actuallyactually „happening“„happening“ in in thethe populationpopulation •• Estimation Estimation ofof indirectindirect selectionselection
OnOn thethe „-“„-“ side side •• It It isis notnot possiblepossible toto bebe completelycompletely suresure toto havehave causallycausally isolatedisolated thethe „target„target ofof selection“selection“** •• It It isis difficultdifficult toto inferinfer thethe „agents„agents ofof selection“selection“ causally causally** *The problem of third variable causation (e.g., Ruxton and Colgrave, 2006)
Example: Experimentally testing targets of selection
Long-tailed widow bird S C L (Euplectes progne)
Anderson 1982. Nature 299, 818-820
10 OnOn thethe „+“„+“ side side •• Causal Causal identificationidentification ofof targettarget ofof selectionselection (manipulation(manipulation ofof targettarget trait)trait) •• Causal Causal identificationidentification ofof agentagent ofof selectionselection (manipulation(manipulation ofof ecologicalecological factor)factor)
OnOn thethe „-“„-“ side side •• Univariate Univariate approach,approach, usuallyusually oneone traittrait atat aa timetime •• Not Not allall traitstraits cancan bebe easilyeasily andand directlydirectly manipulatedmanipulated •• Risk Risk ofof generatinggenerating exaggeratedexaggerated („unnatural“)(„unnatural“) variancevariance inin thethe traittrait →→ CAREFULCAREFUL experimentationexperimentation requiredrequired
It can be interesting to experimentally create phenotypes that are outside the natural range
Experimental extreme values
w
Phenotypic trait value
Example: sensory biases in the evolution of communication/signalling
11 Zebra finch (Taenopygia guttata)
Male Attractiveness
Female Attractiveness
Of 49 offspring reaching adulthood, 29 were reared by males with red colour bands, and 36 had mothers with orange bands. Burley 1981, Science
Note on fitness components
•• In In evolutionaryevolutionary andand behaviouralbehavioural ecology,ecology, fitnessfitness componentscomponents areare oftenoften partitionedpartitioned intointo a benefit part andand aa costcost partpart
•• Benefits Benefits andand costscosts areare oftenoften investigatedinvestigated separatelyseparately
•• But: But: whatwhat mattersmatters isis thethe life-timelife-time valuevalue forfor thethe componentcomponent (trade-offs(trade-offs involved)involved)
Does this result mean that there is ongoing directional selection on male tail length in these widowbirds?
12 Note on fitness components
0.5
0.4
0.3
: Fitness 0.2
W 0.1
S2 CL4 6 8 10 Phenotype Benefit part of fitness component Cost part of fitness component (e.g., mating success, nr nests) (e.g., survival probability) 1 1 0.8 0.8 0.6 0.6
0.4 0.4
0.2 0.2
SCL2 4 6 8 10 SCL2 4 6 8 10
13