ABGG Core Course Genetics and Behavior

Report
ABGG Core Course
Genetics and Behavior
Andy Sih
4-7243
[email protected]
Why study genetics of behavior?
•
•
Adaptive evolution requires heritable variation and selection.
Behavioral ecologists use the ‘phenotypic gambit’ (Grafen 1984)
to explain behavior based on current utility (optimality) – assume
no genetic constraints.
What do we need to know about genetics?
Genetic Architecture
• Number of genes – a few vs. polygenic - faster evolution
with fewer genes – but not if optimum outside of current
genetic variation
• Additive vs. non-additive (dominance, epistasis) –
evolution easier to project with additive genetic variation
• Genetic correlations – linking traits – pleiotropy vs.
linkage
• Coadapted gene complexes (epistasis) –
associated with integrated phenotypes
• Location of genes –understanding (sex)
linkage
• Candidate genes – linking genes to
neuroendocrine mechanisms to behavior –
e.g., DRD4 dopamine receptor
• Genomics
Quantitative genetics
•
Does not study genes themselves, but relies on
statistical similarity of phenotypes in related individuals
•
Suitable for continuous traits controlled by many genes
– e.g., behavior
Key books: Falconer & Mackay 1996, Lynch & Walsh 1998
Phenotypic value (P) = genetic value (G) + environmental effect (E) +
interaction (G x E)
G x E is genetic variation in response to the environment
Genetic value (G) = additive (A) + non-additive – including dominance,
epistasis
Heritability
Var(P) = var(G) + var(E) + 2Cov(G,E)
Broad sense heritability = var(G)/ var(P) = H2
Var(G) = var(A) + non-additive genetic variation
Narrow sense heritability = var(A)/ var(P) = h2
Var(ind) = between-individual differences in behavior = personality
= var(G) + var(PE)
Var (PE) = permanent environment effects
Repeatability = var(ind)/var(P)
Which is larger – heritability or repeatability?
Genetic correlation between two traits x, y = rg =
cov(x,y)/(var(x) var(y))0.5
Need genetic variation to have a genetic correlation
Why are
2
h
and rg important?
Predicting evolution! (Lande and Arnold 1983)
1) For one trait:
Δz = h2 s change in trait = heritability x selection
2) For 2 correlated traits:
Δz(1) = h1 β1 + r(1,2) β2
Δz(2) = h2 β2 + r(2,1) β1
Evolution of each trait depends on correlation with the other
and selection on the other.
3) For multiple traits
Δz(i) = G β
G = G matrix of genetic variances and covariances
β = selection on each trait
Step 1 – is the behavior heritable?
No need to estimate heritability precisely?
•
Common garden experiments – contrasting different
populations
Preferably in 2 or more environments. Why?
•
Study the F2, not just F1. Why?
1)
Estimating h2
1) Resemblance between relatives: parent/offspring,
siblings (full or half), twins
•
Controlled breeding designs, constant environment
•
Pedigrees – the ‘animal’ model – can be applied to
field populations – Kruuk 2003
2) Selection experiments: Δz = h s, so h = Δz/s
•
Select for x  evolution of y – estimates genetic r(x,y)
3) Can’t do 1 or 2?
How heritable is behavior?
Literature reviews: Mousseau and Roff 1987, Meffert 2002,
Van Oers and Sinn 2013
Mean heritability =
• Domestic
• Wild
In wild populations:
Activity
Aggression
Boldness
Exploratory tendency
Some examples
e.g., aggression – Drosophila, lab rodents, dogs, pigs,
bluebirds, great tits, stickleback
boldness – dogs, cows, squid, bighorn sheep
coping style – lab rodents, paradise fish, great tits
courtship – crickets, Drosophila, housefly, molly
antipredator – garter snakes
foraging – Drosophila, honeybees, garter snakes
learning – blowfly, Drosophila, honeybee, mouse, pig, rat
migration – armyworm, milkweed bug, mite, vole, warbler
How about humans?
(Kendler and Greenspan 2006; Hatemi and McDermott 2012
Anxiety disorders
Major depression
Alcoholism
Schizophrenia, autism
Personality
Sexual conservatism
Military defense attitudes
Economic attitudes
Racial attitudes
Social trust
Punishment attitudes
Liberal/conservative
Political knowledge
Selection and genetic variation
• Strong selection should reduce genetic variation
YET – lots of genetic variation maintained! HOW?
Genetic correlations
1) Resemblance between relatives - Roff (1996) – literature
review; N = 166 genetic r for behavior, 6 species
Mean genetic r =
2) Selection line experiments – Drosophila, lab rodents,
domesticated animals, great tits – YES – selection on one
behavior drives correlated evolution of a suite of
behaviors. (“Behavioral syndromes”)
3) Mutations – have numerous behavioral effects –
demonstrates pleiotropy
e.g., dunce gene in Drosophila – poor learning, also affects
biorhythms, female mate receptivity, fertility, longevity
Phenotypic r is a good estimate of genetic r?
• Phenotypic r ~ genetic r + environmental r
• Roff (1996) – review – phenotypic
correlations often ARE a good estimate of
genetic correlations.
• Lynch (1998) – theory – yes, if large
sample sizes (e.g., 500) and at least 20%
of the sample are relatives
Key issue: Constancy of the G matrix
1) Projecting evolution (future or past) assumes constancy
of the G matrix.
• Some evidence of constancy- comparing G matrices for
similar species.
2) BUT selection should:
– reduce h – select out deleterious alleles
– alter r – differential epistasis – genes that modify pleiotropy
• Lab expts show evolution of G in response to selection.
3) AND – simple evolutionary equations assume:
• No G x E interaction = genetic variation in response to
environmental variation
• No G x E correlation = genetic variation in
environments experienced
Reaction norms and G x E
Phenotype
Phenotype
Reaction norm = the pattern of phenotypic expression of a
single genotype across a range of environments
Phenotypic plasticity = change in phenotype in response to
change in the environment
Environments
No G x E
Environments
Significant G x E
Character state approach to plasticity
P in each environment is a separate trait
Quantify genetic correlations across
environments (Via and Lande 1985)
1
P
P(E2)
1
2
3
E1
E2
P
2
3
E1
E2
Plasticity as a trait (Scheiner 1993)
P = P + slope x E
Slope = degree of plasticity
Quantify QG of mean and slope
(and polynomial terms for non-linear
reaction norm)
3
1) Genotypes differ in mean P
2) Do not differ in plasticity
P(E1)
P(E2)
1
2
Reaction norm approach to plasticity
3
2
1
P(E1)
1) Genotypes differ in plasticity
2) Do not differ in mean P
Behavioral reaction norms
Dingemanse et al. 2010 TREE
Novel environment OUTSIDE of the ‘normal range’:
human-altered ‘natural’ environments, captive environments:
Extended reaction norms
(West-Eberhard 2003; Ghalambor et al. 2007; Lande 2009)
NEW OPTIMUM
B
TRAIT
A
Evolutionary
mismatch
X
Past
Environments
New
Environment
Baldwin effect (1896)
1) Plasticity (behavior, learning) allows some
individuals to cope with a novel environment.
• Facilitates evolution: allows NS to shape
evolution of genetic adaptation to the new
environment
•  later LOSS of plasticity (genetic assimilation)
• OR – Reduces evolution
Evolution of
adaptive plasticity
Phenotype
F1
P
F2
E1
E2
Plasticity is not necessarily favored!
WHY NOT?
But behavior is infinitely reversible with very short
Maybe NOT!
Alternative idea: individuals have
consistent behavioral types (BTs)
• Human personalities
• Coping styles: proactive vs. reactive
• Shy-bold continuum (D.S. Wilson
1994)
• Behavioral syndromes (Sih et al.
2004) – aggressiveness, boldness,
activity
Growing literature that BTs are
heritable in humans and other
animals.
Behavior
time lags? So, expect optimal plastic behavior?
E1
E2
Evolutionary questions about behavioral genetic correlations
Q: Why should individuals have a BT? Why not exhibit
optimal behavioral plasticity?
Q: Why are boldness and aggressiveness sometimes
positively correlated, but other times not?
E.g., stickleback fish (Bell 2005; Bell and Sih, unpubl data)
• Population surveys – bold and aggressive are correlated
under one predation regime, but not the other.
Q: When are they correlated?
Q: Why?
G x E correlations
Different genotypes experience different environments
Implications of behavioral G x E correlation
1) Alters selection on each genotype
2) Increases positive assortative interactions
3) Allows specialization
4) Divergence and maintenance of variation
Genetics of social behavior
Indirect genetic effects (IGEs)
I. maternal effects
A (mom)
A (off)
M
Trait (mom)
Maternal
Environment
Env (mom)
Trait (off)
Env (off)
M = maternal effects coefficient
1) Part of offspring environment is provided by mom:
parental care, oviposition/nest site
2) This environmental effect can have a genetic component.
3) So, offspring traits (and fitness) affected by mom’s genes in 2 ways:
• Direct genetic effects – inheritance of her genes
• Indirect genetic effects – genetic component of her maternal effect.
IGEs – II. unrelated individuals
A (indiv 1)
Behavior (1)
E (1)
ψ
Social E(1)
Social E(2)
A (2)
E (2)
ψ
Behavior (2)
E.g., aggressiveness
Implications
1) Alters the genotypephenotype relationship
2) Positive feedback loop
increases rate of
evolution
QTL = Quantitative trait locus
A region of DNA that is statistically associated with a
particular phenotypic trait.
1) Number of QTLs, distribution of QTL effects, interactions
among QTLs (correlations, epistasis etc) – major parts
of genetic architecture.
Note: QG assumes many genes of small effect. QTL studies
often show a few QTLs of moderate effect along with
many other QTLs of small effect.
2) In principle, QTLs could lead to candidate genes – e.g.,
sequence the DNA at a QTL and compare to database
for known gene sequences.
QTL mapping
1) Need a set of genetic markers (identified regions of
variable DNA) spread across the genome.
GOAL will be to identify statistical associations between a
phenotype and one or more of these markers.
2) Use controlled crosses to create population with lots of
variation in phenotypes and marker alleles, retaining
tight linkage between marker and QTL.
e.g., RIL = recombinant inbred lines
•
•
•
•
Cross 2 inbred strains – generates lots of genetic variation in F1
Use the F1 to generate multiple inbred lines, little variation within
each inbred line, lots of variation among lines.
Genotype the markers and measure phenotypes for each line.
Use statistical programs to assess statistical associations
Henderson et al (2004). QTL analysis of multiple behavioral
measures of anxiety in mice. Behavior Genetics 34: 267-293
(prize winning paper)
1)
2)
3)
4)
5)
6)
4 long-term selection lines of mice – 2 selected for high activity
and 2 for low activity in the open-field assay.
H1 x L1 and H2 x L2 to generate F1s. F1 x F1  F2s, about 820
mice for each F2. Genotype markers assessed.
Each mouse run through battery of tests: open-field, light-dark
box, elevated plus maze, mirror chamber, elevated square mazewith numerous behaviors measured in each assay.
Also, monitor home cage activity
Clustered behaviors into several main categories:
Safe activity, risk activity, rearing, latency to act, defecation
Calculate LOD scores (log10 of odds of linked/odds of unlinked) of
each marker with each behavioral category. LOD > 4.3
considered to be good evidence for a QTL
Genetic results
5 QTLs are associated with most of the measured
behaviors– on 5 different chromosomes: 1, 4, 7, 15, 18
•
Safe activity: chr 1 strong; 4, 7, 15, 18 significant
•
Risk activity: chr 7, 15 strong; 1, 4, 18 significant
•
Rearing: chr 15 strong, 4, 7, 18 significant
•
Latency: only chr 15 strong and significant
•
Defecation: chr 1 strong, otherwise, different QTLs than
the other behaviors.
1)
2)
3)
A few QTLs of moderate influence
Most of these influence multiple aspects of behavior.
Behavioral correlations explained by shared QTLs.
Pros and cons
Which questions to study with which approach?
• h2 and genetic r
• Reaction norms and G x E
• G x E correlations
• Indirect genetic effects
• QTLs
MOLECULAR APPROACHES

similar documents