Evolutionary significance of stochastic forces and small

populations

Coyne JA, Barton NH and Turelli M. 1997. A critique of Sewall Wright’s shifting balance theory of evolution. Evolution 51:643-671

Genetic differentiation

• Evidence for population differentiation in plants is indisputable.– Deterministic forces (Natural selection)– Stochastic processes (Genetic drift)

Drift causes random changes in allele frequencies

Simulated population; N = 10

Determinants of drift

• small population size (N)

• restricted dispersal (m)N

N

N

N

N

population

neighbourhood

Effective population size, Ne

- a standardized measure of population size - size of an ‘idealized’ population with the same

strength of genetic drift as the target population. - the census number (N), adjusted for skewed sex

ratio, perenniality, selfing, persistent seed bank, ploidy, non-random variation in fecundity etc.

- most cases, Ne is less than the actual count of individuals in the population (N)

How important is chance?

• Darwin (1859): acknowledged that historical accidents and chance could oppose the forces of natural selection

• Gulick (1872): Hawaiian land snails• Hagedoorn, A. L. and Hagedoom, A.

C. The Relative Value of the Processes Causing Evolution. Pp. 294. Martinus Nijhoff. The Hague, 1921.

Wright and Fisher

Fisher: adaptive evolution results simply from Darwinian mass selection.

Wright: adaptation cannot be explained by selection alone. Stochastic processes such as genetic drift often play an important role.

Shifting Balance Theory

Genotype/phenotype

Fitn

ess

selection

drift

Fitness landscape

selection

Coyne, Barton and Turelli 1997

“….it seems unreasonable to consider the shifting balance process as an important explanation for the evolution of adaptation”

Role of small populations and genetic drift in the evolution of mating systems

in Eichhornia paniculata

Eichhornia paniculata

•Pontederiaceae•short-lived perennial/annual•insect pollinated

Ephemeral water bodies in Brazil, Cuba, Jamaica, parts of Central America

•3 mating types• mating is disassortative and outcrossing• stable state: frequency-dependent selection maintains equal morphs frequencies

Tristyly

N = 167 populations

Estimate mating type frequencies

Trimorphic = 118

Dimorphic = 42

Monomorphic = 7

Mating type structure

•Trimorphic populations near 1:1:1, or low on S•Most dimorphic pops missing the S morph;•All monomorphic pops are M

How is mating system measured?

1. Need 8-10 half sib offspring from each of 20-30 mothers

2. Genotype mothers and offspring using genetic markers (allozymes, microsatellites, AFLPs)

3. Infer the genetic contribution of the paternal parent

ABAA?ABAB

Mother = AA

4. Estimate the rate of outcrossing (t) that produces the distribution of offspring observed. S = 1-t

Population outcrossing rate varies with mating type diversity

Self-fertilizing1 mating type

Cross-fertilizing3-mating types

Selfing variant of the M morph

• Natural selection against the S morph, perhaps related to pollinator x mating type interactions

• Stochastic events associated with small, short-lived populations

What evolutionary forces have caused the the loss of mating types and the transition from a stable outcrossing breeding system to self-fertilization?

Hypotheses

Selection

• Pollinator limitation: long-tongued solitary bees; may be unpredictable in small pops; S morphs may be most vulnerable

Fertility in the field

0

10

20

30

40

50

60

70

80

90

100

L M S

Style Morph

Fruit set (% of flowers)

but S < M,L in 3 of 6 popsF = 0.31, p > 0.50

Effective Population Size (Ne)

•Individual-based simulations of tristylous populations• When Ne < 40, drift can overcome selection and cause the loss of mating types.• Ne < 10, more likely to lose two mating types.

Mating types not lost equally

S morph - most likely to be lost

ssmm ssMm

ssMM

SsMm

SsMM

SSMm

SSMM

• frequency-dependent selection resists loss of morphs•if 1:1:1, all morphs equally likely to disappear due to sampling error• however, S allele is only carried by S morphs and thus cannot segregate from remaining L and M.

Effective population size in 10 populations of E. paniculata

Genetic methodSample allele frequencies over at least 2 years

Var

ian

ce i

n a

llel

e fr

eq.

Ne

V(p) =

€

p(1− p)[1− 1−1

2Ne

⎛

⎝ ⎜

⎞

⎠ ⎟

t

]

Ne - Demographic method

Five estimates1. Estimate # of individuals2. N, corrected for variation in among

years3. N, corrected for variance in flower

production4. N, corrected for mating type frequency5. N, corrected for self-fertilization

Ne

Mean N = 763 (range 30.5 - 5040)Mean Ne = 15.8 (range 3.4 - 70.6) Mean Ne / N = 0.106

Ne < 40 in 120 of 167 pops

0

20

40

60

80

100

120

140

160

0 20 40 60 80

Ne-genetic

Ne-demographic

Ne/N DemographyTemporal var = 0.47Reprod effort = 0.42Selfing rate = 0.98Morph freq = 0.95

Effect of drift onSpatial variation in morph structure

Predictions

Effective population size

Dimorphic/monomorphicTrimorphic

Spatial variation in mating type structure

Temporal variation in frequency of S mating type

S morph lost from pops

0

0.1

0.2

0.3

0.4

0.5

1-25 26-50 51-150 151-500 501-1500

Population size (N)

Change in morph diversity / yr

Obs Exp

Temporal variation in S as a function of N

What accounts for the loss of the L morph?

• Reproductive assurance: ability to self-fertilize in the absence of pollinators favours selfing M morph

0

10

20

30

40

50

60

70

80

90

100

L M

Style Morph

Fruit set (% of flowers)

F=2.8, p = 0.13

Why doesn’t the M morph spread in trimorphic populations?

• pollinators not scarce in large pops• siring advantage doesn’t exist when S is present

Genotype/phenotype

Fitn

ess

selection

drift

Fitness landscape

selectionoutcrossing

selfing