Ecole Normale Supérieure Biology Master 1 Evolutionary Ecology
Fitness, Selection, Adaptation Dr. Regis Ferriere October 31, 2016
I. Adaptation: Concept And Evidence
Darwinian concept of adaptation • “One thing [the organism] is made to conform to another [the environment] in a large number of different respects.” (Fisher 1930)
• Any feature of an organism that has evolved through genetic response to some ecological agent of natural selection.
Evolution by natural selection 4 requirements for evolution by natural selection to take place in a population, as defined by Charles Darwin in On The Origin of Species: 0. Individuals live in a place where they can reproduce.
1. Individuals vary in some their traits (size, shape, color, behavior…) 2. Traits are inherited, at least in part, from parents to offspring. 3. Different traits cause individuals to have different reproductive success = fitness.
• Inherited variation in reproductive success Natural selection --- inevitably! • Natural selection = only evolutionary process that can lead to biological adaptations. “If it could be demonstrated that any complex organ existed, which could not have been formed by numerous, successive, slight modifications, my theory would absolutely break down.” (Darwin 1859)
• George C. Williams (1966) links adaptation to biological function which evolves by natural selection. • Alan Grafen (1988): Adaptation, in Williams’ sense of a feature of an organism that has a function and evolved by natural selection, should be distinguished from selection-in-progress = adaptation as a process, adaptive evolution • H. Kern Reeve and Paul Sherman (1994): close relation between adaptation as an outcome and a process “An adaptation is a phenotypic variant that results in the highest fitness among a specified set of variants in a given environment.”
Can we evidence adaptation in Darwin’s finches? For the medium-ground finch on Daphne Major island: • Food resources fluctuate from year to year. • Finch population seems to fluctuate in response. • Hypothesis: The finch population evolves and adapt rapidly to cope with ups and downs of food availability.
Testing Postulate 1: Is the Finch population variable for some relevant trait?
Testing Postulate 2: Is Some of the Variation among Individuals Heritable? In both years that measurements were taken, strong positive correlation between parents’ and offspring beak depth
Beware!! A graph like this could be due to non-genetic causes:
- if environment influences development of beak, and parents and offspring shared environments. - If a non-genetic signal (hormones…) is passed on by mom to egg = maternal effect.
Molecular evidence that strongly supports genetic inheritance
Compare different species: different beak depth caused by different levels of expression of gene Bmp4 during development
Hypothesis: Variation of beak depth among individuals of the study species (medium-ground finch) may be linked to Bmp4 and variation in its expression level.
Testing Postulate 3: Does the trait (beak depth) influence lifetime reproductive success (fitness)?
Conclusion: In this extremely dry season, individuals with deeper beak had a greater chance of survival – higher fitness
Testing Darwin's prediction altogether: Did the population evolve?
Comparing 2 cohorts of chicks…
Evolution by natural selection can be very fast! • Not just in microbes… Even in vertebrates!
Evolution by natural selection can be very fast! • Not just in microbes… Even in vertebrates! Evolution is very dynamic! • Different traits may or may not respond to a given selective agent (drought…)
Evolution by natural selection can be very fast! • Not just in microbes… Even in vertebrates! Evolution is very dynamic! • Different traits may or may not respond to a given selective agent (drought…) • The same traits may respond differently to different episodes of the same selective agent if other factors change in the environment
In early 2000s large-beak ‘medium ground’ finches had to compete with ‘large ground’ finches flying in from another island… The direction of selection was reverted compared to 1977!
Evolution by natural selection can be very fast! • Not just in microbes… Even in vertebrates! Evolution is very dynamic! • Different traits may or may not respond to a given selective agent (drought…) • The same traits may respond differently to different episodes of the same selective agent if other factors change in the environment
Experimental evidence for adaptation: David Reznick’s study of guppies in Trinidad
Effect on size and age at maturity after 33 generations?
• A change in mortality risk (predation) triggers rapid adaptive evolution of morphological, behavioral, ecophysiological and life-history traits
Outstanding questions about adaptation
Outstanding questions about adaptation • Is a phenotypic trait ‘adaptive’? • Can adaptation explain phenotypic differences between populations or species?
• Is a trait under selection? How strong is selection? • Can we predict the adaptive response of a trait to selection? • What is the genetic basis of an adaptation? • How does epistasis affect adaptation? [the effect of a mutation depends on previous mutations] • How can adaptation proceeds in spite of recombination? • How can genetic variation persist in spite of adaptation? • Where does selection come from? What are the ecological agents of selection?
II. Adaptation In The Modern Synthesis And Challenges
Fisherian Theory of adaptation in the Modern Synthesis 1) Adaptation conceptualized as hill-climbing on a fitness landscape Population Fitness
Mean trait 1
Mean trait 2
Fisherian theory of adaptation in the Modern Synthesis 2) The Fundamental Theorem of Natural Selection • What is the rate of adaptation?
VA (w) w* E(w) w* • How population mean ‘fitness’ w* changes from one generation to the next • VA(w) = additive genetic variance in ‘fitness’ • E(w) measures “deterioration of the environment” due to population’s feedback
In absence of population-environment feedback: E(w) = 0 • Rate of adaptation decreases as mean ‘fitness’ increases or additive genetic variance decreases • Adaptive evolution comes to a halt when additive genetic variance in fitness is exhausted
Theory of adaptation in the Modern Synthesis • Ronald Fisher (1930) 3) The Geometric Model of Adaptation • Does adaptation proceed by mutations of large effects or mutations of small effects? The population begins on the surface of the sphere and, by substituting beneficial mutations (red vectors), evolves towards the phenotypic optimum at the centre of the sphere. The mutations that are substituted become smaller on average as the population nears the optimum.
Testing the theory: Richard Lenski’s experimental evolution of E. coli
Post-Modern Synthesis Theory: Adaptation as an optimization process
Post-Modern Synthesis Theory: Adaptation as an optimization process Predicting adaptation as an optimum within a range of possible options. • Optimality implies ‘fitness maximization under constraints’ on what is possible. • ‘Constraints’ in evolutionary optimization come from tradeoffs. • The Darwinian Demon that would begin reproducing at birth and would convert all acquired energy into offspring during an indefinite lifetime is impossible! • Tradeoffs are difficult to document… • Silver spoon effect (Sultan 1995): the difficulty of detecting tradeoffs in favorable environments.
How to predict the adaptive optimum • Phenotypic (trait) space contains set of feasible phenotypes. • Evolutionary optimization is maximization of fitness within feasible set. • Trade-offs determine the boundary of the feasible set, and optimum located on boundary. • Feasibility set and fitness contours were popularized by Richard Levins (1966, 1969).
• Example: resource allocation strategy, trait x Fecundity F x
1-x
x=1
Seedling survival and germination probability, S
F
x=0
Adult survival P P Maximum fecundity x=1
Maximum survival
x=0
Expected Lifetime Reproductive Success: R0 = S F(x)/ [1 - P(x)]
An adaptive optimum may be stochastic • WhenEx.: the environment fluctuates unpredictably Toxic algal blooms • Flipping a coin… may be the optimal strategy in an uncertain environment
An adaptive optimum may be a reaction norm • When the environment fluctuates with some degree of predictability • The adaptive reaction norm is a decision rule - how to respond to the state of a fluctuating environment • Different types - phenotypic (developmental) plasticity - phenotypic flexibility - intra- or trans-generational
Critique of the ‘adaptationist program’: Gould & Lewontin (1978)
Critique of the ‘adaptationist program’: Gould & Lewontin (1978) • Functional performance may partly result from physics or chemistry, rather than evolved adaptation. Example : Different metabolic responses to different environmental temperatures
E. coli
Performance
• Adaptation is also constrained by evolutionary history glucose diet
maltose diet
Time
• Need to avoid just-so stories about how selection works by carefully testing multiple competing hypotheses.
Resources from which selection arises can be mates -> sexual selection
Selection can operate at multiple levels Ex.: Toxic algal blooms
An adaptation may be polymorphic -Ex.: Several phenotypes coexist in the population Toxic algal blooms • How can a polymorphism be the outcome of natural selection? - Heterozygote advantage - Negative frequency-dependence - Temporally or spatially fluctuating environment
Example of geographic variation in environmental conditions that maintain genetic and phenotypic variation: populations of white clover that produce cyanide as defense against herbivores. •
Plants that produce cyanide are more likely to be killed by frost.
•
Thus, different subpopulations are subjected to different environmental conditions and selection agents.
Experimental evolution of a polymorphism: Paul Rainey’s lab study of Pseudomonas fluorescens
O2
WS SM
food SM
SM
WS SM FS WS SM
FS
Interspecific interactions may matter (a lot)
Ex.: Toxic algal blooms
Insufficiencies of evolutionary optimization theory • An optimum may not be reachable by the evolutionary process. • An optimum may not be stable (and instability may occur in various ways). • An optimum may not exist • An optimum may not be ‘simple’ • How should we define fitness? Is there a universal measure of fitness?
Richard Lewontin (1978): The fitness landscape is “rubbery”
“Organisms are not independent of the environment they occupy; rather, they at least in part define and continuously modify that environment by the very acts of surviving and reproducing. Part of their environment, in other words, is the ensemble of other organisms that are present in the population and with which they interact. (…) Organisms and environments co-evolve (…) It is vanishingly unlikely that a population could climb to a fitness peak that remained unaltered during the climb. Instead of thinking of a fixed landscape and of populations moving on it, we need to visualize a rubbery landscape whose very features are altered by the entities (populations) that exist on it.”
III. Next-Generation Theory Of Adaptation
The eco-evolutionary feedback loop
Ecological interactions, phenotypic expression
Individual adaptive traits
E
Population, community, ecosystem dynamics
I Selective pressures on heritable variation
Dieckmann & Ferrière 2014 Evol. Conserv. Biol.
• Adaptive evolution is conceptualized as a trait-substitution sequence • The environment changes concomitantly
‘Blue back’ pictures, courtesy of Dr. Ulf Dieckmann, IIASA
• Adaptive evolution is conceptualized as a trait-substitution sequence • The environment changes concomitantly
• Fitness is defined as Invasion fitness f(x’, x) = population growth rate of mutant x’ initially rare in equilibrium environment set by resident x population.
• For one-dimensional traits, the adaptive landscape obtains by plotting the invasion fitness function f(x’, x) • For given resident trait value x, the slope (derivative) of f(x’,x) at x gives the selection gradient at x
Mutant altruistic trait
Fitness landscape of altruistic trait
Resident altruistic trait Meszena et al. 2001 Selection
Le Galliard et al. 200E Evolution Ferriere & Michod 2011 Nature
• Evolutionary singularities = Trait values x at which the selection gradient is zero
The adaptive landscape is rubbery indeed!
Example: Resource utilization trait, x
Adaptive trait
• Eco-evolutionary feedbacks can cause evolutionary suicide • In a gradually changing environment, risk of evolutionary trapping
Extinction Environmental condition
With coevolving host
With non-coevolving host
Zhang & Buckling 2011 Ecol. Lett.
Massot et al. 2008 Global Change Biology
Toxic algal blooms as eco-evo driven catastrophic shifts between alternate ecological equilibria
Algal density
3. Public good
2. Positive feedback 4. Tragedy of The Commons 1. Individual selection
Algal toxicity
Driscoll, Ferriere, Hackett (2015 Ecology Letters)
Eco-evolutionary theory provides a predictive framework for empirical analysis of selection How to evaluate alternate hypotheses: 1. Construct full model and estimate probability of data under this model. 2. Compare with alternate models in which certain links are turned off. Eco-evolutionary pathways linking population (N), ecosystem (E), selection gradient (S), phenotype (z)
H1: Only direct effect of environmental change on selection: turn off links 3a and 3b. H2: No effect of ecosystem on selection other than mediated by population density (link 4b): turn off link 3b.
Ferriere, Reznick et al.(in prep.)
H3: No feedback from phenotypic change to ecosystem: turn off link 2b.
What have we learned?
