Ecosystem Interactions Class Discussion Presentation in Blue Green Lined Styl...
François Massol - présentation MEE2013
1. Evolution of dispersal in
heterogeneous and
uncertain environments
François Massol
May 2013 – Models in Evolutionary Ecology, Montpellier
2. Pros and cons
kin competition
inbreeding avoidance
catastrophes
temporal variability
environmental
heterogeneity
cost of dispersal
Pros Cons
van Valen 1971; Balkau & Feldman 1973; Roff 1975;
Hamilton & May 1977; Bengtsson 1978
3. The larger picture
Patterns (direct):Processes:
kin competition
inbreeding
catastrophes
temp. variability
heterogeneity
cost of dispersal
dispersal ESS
4. The larger picture
Patterns (direct):Processes:
kin competition
inbreeding
catastrophes
temp. variability
heterogeneity
cost of dispersal
dispersal ESS
polymorphism?
5. The larger picture
Patterns (direct):Processes:
kin competition
inbreeding
catastrophes
temp. variability
heterogeneity
cost of dispersal
dispersal ESS
polymorphism?
association with
other traits
6. The larger picture
Patterns (direct):Processes:
Patterns (indirect):
kin competition
inbreeding
catastrophes
temp. variability
heterogeneity
cost of dispersal
dispersal ESS
polymorphism?
association with
other traits
diversity
gene flow
…
7. Heterogeneity and variability
• Environments are temporally variable
organisms experience temporally variable habitats
• Environments are spatially heterogeneous
dispersing allows for different habitats among siblings
10. Questions
1. How can we model the evolution of dispersal
in uncertain heterogeneous environments?
2. Case study: the joint evolution of selfing and
dispersal when pollinator occurrence is
random
3. What happens when dispersal can be
informed?
11. Adaptive dynamics
Assumptions:
– phenotypic gambit
“The phenotypic gambit is to examine the evolutionary basis of a character as if the
very simplest genetic system controlled it: as if there were a haploid locus at which
each distinct strategy was represented by a distinct allele, as if the payoff rule gave
the number of offspring for each allele, and as if enough mutation occurred to
allow each strategy the chance to invade.” A. Grafen, in Krebs & Davies 1984
– rare mutations of small effects
Tools:
– expression for fitness (using matrices)
– selection gradient → convergence stability
– Hessian of mutant fitness → evolutionary stability
17. Life cycles
What if the order of events is different?
– selection gradient on dispersal rate may change
direction when reproduction, dispersal and
regulation happen in a different order (Johst &
Brandl 1997)
– the order of reproduction, dispersal and regulation
directly impact the evolution of local adaptation
traits (Ravigné et al. 2004)
18. A general model
Ingredients:
– 2 patch types (1 & 2; affect fecundity), infinity of
patches
– 4 life cycle events: reproduction, dispersal, regulation
& environmental change
– discrete, non-overlapping generations
– reproduction: result of local adaptation, not limiting
– regulation: local (but large populations)
– dispersal: global (no limitation by distance)
Massol (2013)
19. Classification of life cycles
extended from Massol (2013)
Levene soft selection
regime
Ravigné
type 3
Ravigné
type 3
Ravigné et al.’s
classification
Order of
events
Dempster hard
selection regime
D D D D D
D E E E
EEE
R R
RR RR
Simple life cycles Complex life cycles
Modelling
complexity
Simple life cycles
20. Classification of life cycles
extended from Massol (2013)
Order of
events
D D D D D
D E E E
EEE
R R
RR RR
When dispersal is unconditional
Classes of equivalence for fitness
correspond to Ravigné et al.’s
To the really interested audience:
• E always commutes with regulation.
• With unconditional dispersal, E also commutes with dispersal.
Levene soft selection
regime
Ravigné
type 3
Ravigné
type 3
Ravigné et al.’s
classification
Dempster hard
selection regime
21. Evolution of dispersal
Massol & Débarre (in prep)
Order of
events
DE
R
DE
RD
E R
LeveneRavignéRavigné et al.’s
classification
Dempster
22. Evolution of dispersal
Massol & Débarre (in prep)
Order of
events
DE
R
DE
RD
E R
LeveneRavignéRavigné et al.’s
classification
Dempster
evolution towards
philopatry
23. Evolution of dispersal
Massol & Débarre (in prep)
Order of
events
DE
R
DE
RD
E R
LeveneRavignéRavigné et al.’s
classification
Dempster
0.5 0.0 0.5 1.0
0.2
0.4
0.6
0.8
1.0
dispersalrate(d)
temporal autocorrelation in patch state ( φ )
Branching
24. Evolution of dispersal
Massol & Débarre (in prep)
Order of
events
DE
R
DE
RD
E R
LeveneRavignéRavigné et al.’s
classification
Dempster
Zero dispersal when
Total dispersal when
Branching when
2
1/2
2 31
g
c
g g
2c g
0
25. Evolution of dispersal
Massol & Débarre (in prep)
Order of
events
DE
R
DE
RD
E R
LeveneRavignéRavigné et al.’s
classification
Dempster
dispersalrate(d)
temporal autocorrelation in patch state ( φ )
0.5 0.0 0.5 1.0
0.2
0.4
0.6
0.8
1.0
bistability
26. Evolution of dispersal
Massol & Débarre (in prep)
Order of
events
DE
R
DE
RD
E R
LeveneRavignéRavigné et al.’s
classification
Dempster
Zero dispersal when
or
Total dispersal when 2c g
21/2
2 3 2
21/2
2 2 3
2
1 1 4 1
2 2
c
1/2
2 2 31/2
2 3 2 2
2
21/2
2 3
4 1
2 1 1 1 1
1
2 1
c
c c
c
c
27. Evolution of dispersal
Massol & Débarre (in prep)
DE
R
Ravigné
D
E R
Levene
DE
R
Dempster
Evolution towards total philopatry
Intermediate dispersal rates are possible
Branching happens for negatively autocorrelated
environments
Either total philopatry or total dispersal
Bistability can happen
29. Words from Baker (1955)
“With a self-compatible individual a single
propagule is sufficient to start a sexually
reproducing colony”
“Self-compatible flowering plants are usually
able to form seeds in the absence of visits from
specialised pollinating insects, which may be
absent from the new situation”
30. Baker’s law
Principle: selfing provides reproductive
assurance without pollination
Expectation: island species should be selfers
(verbal model)
Corollary belief: selfers disperse more
31. Baker’s law revisited
Self-fertilization & dispersal: two major traits
Modifications of these traits are expected to
greatly affect fitness
Their evolutions are often considered separately
in theoretical models
34. Baker’s law revisited
Infinite island model with large populations
Fluctuations in pollination, e
Inbreeding depression, δ
Migrant survival, q
Two traits under selection (s,d)
δ
q
d
1-e
1 - s
e
Cheptou & Massol (2009)
35. Baker’s law revisited
Cheptou & Massol (2009)
In terms of life cycles…
… with zero autocorrelation
(with autocorrelation: see Massol & Cheptou 2011)
DE
R
Ravigné
36. Baker’s law revisited
(1 )s
(1 )
1
2
1
2
s
s
s
1 d
1 d
d
d
q
q
/mutant residentw w
/mutant residentw w
reproduction dispersal regulation
e
1 e
W
Cheptou & Massol (2009)
37. Baker’s law revisited
Scenario I
dispersalrate(d)
selfing rate (s)
“dispersal/outcrossing” syndrome
*
*
0
1 (1 )
s
e
d
q e
Cheptou & Massol (2009)
38. Baker’s law revisited
Scenario II
dispersalrate(d)
selfing rate (s)
“no dispersal/selfing” syndrome
(may be mixed selfing)*
*
2
2 1
0
e
s
d
e
Cheptou & Massol (2009)
39. Baker’s law revisited
Scenario III
dispersalrate(d)
Both syndromes exist (bistability)
2 ESS (evolutionary stable eq.)
1 evolutionary repellor (saddle)
selfing rate (s)
Cheptou & Massol (2009)
40. Baker’s law revisited
1. Three scenarios for evolutionary equilibria
2. Two syndromes for trait associations:
dispersal/outcrossing vs. no dispersal/partial selfing
goes against Baker’s law
3. Under scenario III (two ESS), bistability with
Ravigné life cycles
Cheptou & Massol (2009)
41. Baker’s law revisited
Understanding Baker’s law:
– need separating extrinsic (e.g. pollinators) vs.
intrinsic (e.g. mate density) determinants of pollen
limitation
– the asymmetry of dispersal (mainland/island vs.
metapopulation) dictates the reproductive value of
“island” populations
– with local regulation, evolution will not “optimize”
Massol & Cheptou (2011)
43. Informed dispersal and life cycles
• Conditioning dispersal decision on patch “quality”
may decrease the indirect cost of dispersing
• First theoretical argument using two-patch
models (McPeek & Holt 1992)
• With almost static environments and bad cues,
dispersal is not conditioned on current perceived
patch quality (McNamara & Dall 2011)
→ bang-bang dispersal (all or nothing), with no
polymorphism (informed dispersal vs. polymorphism)
45. Classification of life cycles
Ravigné et al.’s
classification
Ravigné RavignéLevene
E E
E ER
R R RD
D
D D
Order of events
46. Information & life cycles
Ravigné et al.’s
classification
Ravigné RavignéLevene
E E
E ER
R R RD
D
D D
Order of events
evolution towards
philopatry
Massol & Débarre (in prep)
47. Information & life cycles
Ravigné et al.’s
classification
Ravigné RavignéLevene
E E
E ER
R R RD
D
D D
Order of events
Massol & Débarre (in prep)
0.5 0.0 0.5 1.0
0.2
0.4
0.6
0.8
1.0
dispersalrate(ei)
temporal autocorrelation in patch state ( φ )
e1 (out of bad patches)
e2 (out of good patches)
Results with perfect cue
48. Information & life cycles
Ravigné et al.’s
classification
Ravigné RavignéLevene
E E
E ER
R R RD
D
D D
Order of events
Massol & Débarre (in prep)
dispersalrate(ei)
temporal autocorrelation in patch state ( φ )
Results with imperfect cue
0.5 0.0 0.5 1.0
0.2
0.4
0.6
0.8
1.0
e2 (out of good patches)
branching
e1 (out of bad patches)
49. Information & life cycles
Ravigné et al.’s
classification
Ravigné RavignéLevene
E E
E ER
R R RD
D
D D
Order of events
Massol & Débarre (in prep)
dispersalrate(ei)
temporal autocorrelation in patch state ( φ )
0.5 0.0 0.5 1.0
0.2
0.4
0.6
0.8
1.0
e2 (out of good patches)
e1 (out of bad patches)
branching
Results with
perfect cue
50. Information & life cycles
Ravigné et al.’s
classification
Ravigné RavignéLevene
E E
E ER
R R RD
D
D D
Order of events
Massol & Débarre (in prep)
dispersalrate(ei)
temporal autocorrelation in patch state ( φ )
Results with
imperfect cue
0.5 0.0 0.5 1.0
0.2
0.4
0.6
0.8
1.0
e1 (out of bad patches)
branching
e2 (out of good patches)
51. Making sense of all of that…
0.5 0.0 0.5 1.0
0.2
0.4
0.6
0.8
1.0
0.5 0.0 0.5 1.0
0.2
0.4
0.6
0.8
1.0
dispersalrate(ei)
temporal autocorrelation in patch state ( φ )temporal autocorrelation in patch state ( φ )
e1 (out of bad patches)
e2 (out of good patches)
e2 (out of good patches)
e1 (out of bad patches)
Massol & Débarre (in prep)
be regulated where it’s
easy…
… & go find a better place
to reproduce
E
ER
RD D
Results with
perfect cue
52. Making sense of all of that…
dispersalrate(ei)
temporal autocorrelation in patch state ( φ )temporal autocorrelation in patch state ( φ )
Massol & Débarre (in prep)
be regulated where it’s
easy…
E
ER
RD D
0.5 0.0 0.5 1.0
0.2
0.4
0.6
0.8
1.0
e2 (out of good patches)
e1 (out of bad patches)
0.5 0.0 0.5 1.0
0.2
0.4
0.6
0.8
1.0
e1 (out of bad patches)
e2 (out of good patches)
… & go find a better place
to reproduce
Results with
imperfect cue
53. Informed dispersal and life cycles
Take-home messages:
– informed dispersal follows different rationales with
different life cycles
– disruptive selection can happen in an informed
dispersal model
Not shown here but true:
– bang-bang dispersal strategies happen for some
parameter values only, and are not often robust to
the cost of dispersal
– bistability can occur with certain life cycles
Massol & Débarre (in prep)
54. Final take-home messages
1. Environmental variability can affect the
evolution of dispersal in a variety of ways
2. The joint evolution of dispersal and other
traits (e.g. selfing) yields interesting, bistable
patterns that deserve more attention
3. Informed dispersal and dispersal
polymorphisms are not mutually exclusive
55. Discussion
• Evolution of dispersal: “mature topic”
• Combination of factors determining selection
• Feeds back on other traits (e.g. local
adaptation, selfing)
→ understanding the evolution of dispersal
might be key to understanding patterns
(diversity, functioning, resilience…) in spatially
structured models
56. Acknowledgements
Ongoning collaborations on this topic:
P.-O. Cheptou, P. David, F. Débarre, A. Duputié, P. Jarne, F. Laroche, T. Perrot
Any kind of help & comments:
J. Auld, N. Barton, J. C. Brock, M. Burd, J. Busch, V. Calcagno,
J. Chave, M. Daufresne, C. Eckert, D. Fairbairn, S. Gandon, I. Hanski,
J. Hermisson, K. Holsinger, J. Kelly, M. Leibold, R. Morteo,
N. Mouquet, I. Olivieri, M. Palaseanu-Lovejoy, O. Ronce, D. Schoen
Usual suspects: