1/73
Study of the Yoccoz-Birkeland population model
Sylvain Arlot
CNRS, ´ Ecole Normale Sup´ erieure
Universit` a degli studi di Siena, Nov, 3rd 2009
2/73
Plan
1 The model
Voles in the Arctic Description of the model Mathematical analysis
2 Simulation study Principle
Exploring the parameter space
Study of an attractor: (0.15 ; 0.30 ; 8.25)
3/73
Clethrionomys glaerolus Microtus epiroticus on Svalbard:
4/73
Microtus epiroticus on Svalbard
arctic environment high fertility rate
few breeding places (permafrost)
no reproduction in winter (snow)
(N. Yoccoz, Ims & Steen 1993)
5/73
Between-year variation (N. Yoccoz et al )
0 10 20 30 40 50 60 70 80 90 100
6/73
Seasonal variation (N. Yoccoz et al )
0 200 400 600
800
CORE
0 50 100 150 200
250
RIDGE L
0 50 100 150
200
LOWER
0 10 20 30 40 50
60
RIDGE R
0 50 100 150 200
250
BORDER
0 20 40 60 80
100
REM
90 91 92 93 94 95 96 90 91 92 93 94 95 96 90 91 92 93 94 95 96
90 91 92 93 94 95 96 90 91 92 93 94 95 96
90 91 92 93 94 95 96
7/73
Seasonal variation (N. Yoccoz et al )
0 50 100 150 200
250 RIDGE L
90 91 92 93 94 95 96
8/73
Possible mechanisms
High fertility rate, strongly density-dependent, and:
available food quantity (N. Yoccoz et al 2001)
predation by arctic fox (Turchin et al 2000; Frajford 2002) environmental variability (N. Yoccoz & Ims 1999)
plasticity of the maturation age (Lambin & N. Yoccoz 2001)
Are such variations still possible if the maturation age and the
environment are constant?
8/73
Possible mechanisms
High fertility rate, strongly density-dependent, and:
available food quantity (N. Yoccoz et al 2001)
predation by arctic fox (Turchin et al 2000; Frajford 2002) environmental variability (N. Yoccoz & Ims 1999)
plasticity of the maturation age (Lambin & N. Yoccoz 2001)
Are such variations still possible if the maturation age and the
environment are constant?
8/73
Possible mechanisms
High fertility rate, strongly density-dependent, and:
available food quantity (N. Yoccoz et al 2001)
predation by arctic fox (Turchin et al 2000; Frajford 2002) environmental variability (N. Yoccoz & Ims 1999)
plasticity of the maturation age (Lambin & N. Yoccoz 2001)
Are such variations still possible if the maturation age and the
environment are constant?
8/73
Possible mechanisms
High fertility rate, strongly density-dependent, and:
available food quantity (N. Yoccoz et al 2001)
predation by arctic fox (Turchin et al 2000; Frajford 2002) environmental variability (N. Yoccoz & Ims 1999)
plasticity of the maturation age (Lambin & N. Yoccoz 2001)
Are such variations still possible if the maturation age and the
environment are constant?
8/73
Possible mechanisms
High fertility rate, strongly density-dependent, and:
available food quantity (N. Yoccoz et al 2001)
predation by arctic fox (Turchin et al 2000; Frajford 2002) environmental variability (N. Yoccoz & Ims 1999)
plasticity of the maturation age (Lambin & N. Yoccoz 2001)
Are such variations still possible if the maturation age and the
environment are constant?
8/73
Possible mechanisms
High fertility rate, strongly density-dependent, and:
available food quantity (N. Yoccoz et al 2001)
predation by arctic fox (Turchin et al 2000; Frajford 2002) environmental variability (N. Yoccoz & Ims 1999)
plasticity of the maturation age (Lambin & N. Yoccoz 2001)
Are such variations still possible if the maturation age and the
environment are constant?
9/73
The model (J.-C. Yoccoz & Birkeland 1998)
N(t ) = Z A
1A
0S(a)m ρ (t − a)N(t − a)m(N(t − a))da (female and mature population; A 0 : maturation age)
Fertility: m(N) =
( m 0 if N ≤ 1 m 0 N −γ otherwise Seasonal parameter: m ρ (t) =
( 0 if 0 ≤ t ≤ ρ mod. 1 1 if ρ ≤ t ≤ 1
Survival rate: S(a) = 1 − a
A 1
9/73
The model (J.-C. Yoccoz & Birkeland 1998)
N(t ) = Z A
1A
0S(a)m ρ (t − a)N(t − a)m(N(t − a))da (female and mature population; A 0 : maturation age)
Fertility: m(N) =
( m 0 if N ≤ 1 m 0 N −γ otherwise Seasonal parameter: m ρ (t) =
( 0 if 0 ≤ t ≤ ρ mod. 1 1 if ρ ≤ t ≤ 1
Survival rate: S(a) = 1 − a
A 1
9/73
The model (J.-C. Yoccoz & Birkeland 1998)
N(t ) = Z A
1A
0S(a)m ρ (t − a)N(t − a)m(N(t − a))da (female and mature population; A 0 : maturation age)
Fertility: m(N) =
( m 0 if N ≤ 1 m 0 N −γ otherwise Seasonal parameter: m ρ (t) =
( 0 if 0 ≤ t ≤ ρ mod. 1 1 if ρ ≤ t ≤ 1
Survival rate: S(a) = 1 − a
A 1
10/73
Parameters (N. Yoccoz, Ims, Steen 1993)
fertility m 0 ≈ 54 .
maturation age A 0 ∈ [0.14 ; 0.20] , minimum = 0.10 . maximal age A 1 ≈ 2 .
winter length ρ ∈ [0.35 ; 0.45] (temperate climate) ou ρ ≈ 0.7 (arctic climate).
winter survival rate ≈ 0.1 summer survival rate ≈ 0.85 . density-dependence γ
spring length ε .
10/73
Parameters (N. Yoccoz, Ims, Steen 1993)
fertility m 0 ≈ 54 .
maturation age A 0 ∈ [0.14 ; 0.20] , minimum = 0.10 . maximal age A 1 ≈ 2 .
winter length ρ ∈ [0.35 ; 0.45] (temperate climate) ou ρ ≈ 0.7 (arctic climate).
winter survival rate ≈ 0.1 summer survival rate ≈ 0.85 .
density-dependence γ
spring length ε .
10/73
Parameters (N. Yoccoz, Ims, Steen 1993)
fertility m 0 ≈ 54 .
maturation age A 0 ∈ [0.14 ; 0.20] , minimum = 0.10 . maximal age A 1 ≈ 2 .
winter length ρ ∈ [0.35 ; 0.45] (temperate climate) ou ρ ≈ 0.7 (arctic climate).
winter survival rate ≈ 0.1 summer survival rate ≈ 0.85 . density-dependence γ
spring length ε .
11/73
Mathematical analysis: dynamical system
∀t 0 ∈ [0, 1), Y t
0is the set of continuous functions N:
[−A 1 , 0] 7→ [0, +∞) such that
N(0) = Z A
1A
0S(a)m ρ (t 0 − a)N(−a)m(N(−a))da
Phase space:
Y ] = {(t, N) / t ∈ R / Z , N ∈ Y t } Semi-group (T s ) s≥0 : T s (t, N) = (t + s mod. 1, N t s ) N
ts(−a) =
( N(s − a) if 0 ≤ s ≤ a ≤ A
1R
A1A0
S (b)N(s − a − b)m(N(s − a − b))m
ρ(t + s − a − b)db otherwise
⇒ well-defined for 0 ≤ s ≤ A 0 ,
⇒ extended to s ≥ 0 by the semi-group property.
11/73
Mathematical analysis: dynamical system
∀t 0 ∈ [0, 1), Y t
0is the set of continuous functions N:
[−A 1 , 0] 7→ [0, +∞) such that N(0) =
Z A
1A
0S(a)m ρ (t 0 − a)N(−a)m(N(−a))da Phase space:
Y ] = {(t, N) / t ∈ R / Z , N ∈ Y t }
Semi-group (T s ) s≥0 : T s (t, N) = (t + s mod. 1, N t s ) N
ts(−a) =
( N(s − a) if 0 ≤ s ≤ a ≤ A
1R
A1A0
S (b)N(s − a − b)m(N(s − a − b))m
ρ(t + s − a − b)db otherwise
⇒ well-defined for 0 ≤ s ≤ A 0 ,
⇒ extended to s ≥ 0 by the semi-group property.
11/73
Mathematical analysis: dynamical system
∀t 0 ∈ [0, 1), Y t
0is the set of continuous functions N:
[−A 1 , 0] 7→ [0, +∞) such that N(0) =
Z A
1A
0S(a)m ρ (t 0 − a)N(−a)m(N(−a))da Phase space:
Y ] = {(t, N) / t ∈ R / Z , N ∈ Y t } Semi-group (T s ) s≥0 : T s (t, N) = (t + s mod. 1, N t s )
N
ts(−a) =
( N(s − a) if 0 ≤ s ≤ a ≤ A
1R
A1A0
S (b)N(s − a − b)m(N(s − a − b))m
ρ(t + s − a − b)db otherwise
⇒ well-defined for 0 ≤ s ≤ A 0 ,
⇒ extended to s ≥ 0 by the semi-group property.
12/73
Mathematical analysis: assumptions
N 7→ Nm(N) is K f Lipschitz m 0 ≥ m(N) ≥ m 0 /2 if N ≤ 1
m 0 N −γ ≥ m(N) ≥ min {1/2, N −γ } m 0 1 ≥ m ρ (t) ≥ 0 for every t ∈ [0, 1]
m ρ (t) = 1 on some interval of length 1 − ρ − ε > 0
large domain of the parameter space:
γ ≥ 1
A
1≥ max {2A
0, A
0+ 1}
m
0R
A0+1A0+ρ+ε
S (a)da > 2
⇒ T s : Y ] 7→ Y ] is max {1, K f (A 1 − A 0 )} Lipschitz w.r.t.
| · | + k·k ∞
12/73
Mathematical analysis: assumptions
N 7→ Nm(N) is K f Lipschitz m 0 ≥ m(N) ≥ m 0 /2 if N ≤ 1
m 0 N −γ ≥ m(N) ≥ min {1/2, N −γ } m 0 1 ≥ m ρ (t) ≥ 0 for every t ∈ [0, 1]
m ρ (t) = 1 on some interval of length 1 − ρ − ε > 0 large domain of the parameter space:
γ ≥ 1
A
1≥ max {2A
0, A
0+ 1}
m
0R
A0+1A0+ρ+ε
S (a)da > 2
⇒ T s : Y ] 7→ Y ] is max {1, K f (A 1 − A 0 )} Lipschitz w.r.t.
| · | + k·k ∞
12/73
Mathematical analysis: assumptions
N 7→ Nm(N) is K f Lipschitz m 0 ≥ m(N) ≥ m 0 /2 if N ≤ 1
m 0 N −γ ≥ m(N) ≥ min {1/2, N −γ } m 0 1 ≥ m ρ (t) ≥ 0 for every t ∈ [0, 1]
m ρ (t) = 1 on some interval of length 1 − ρ − ε > 0 large domain of the parameter space:
γ ≥ 1
A
1≥ max {2A
0, A
0+ 1}
m
0R
A0+1A0+ρ+ε
S (a)da > 2
⇒ T s : Y ] 7→ Y ] is max {1, K f (A 1 − A 0 )} Lipschitz w.r.t.
| · | + k·k ∞
13/73
Mathematical analysis: attractor
K t
0= {t 0 }×{N ∈ Y t
0/ N min ≤ N(t) ≤ N max and N is L-Lipschitz}
(N min , N max and L are explicit functions of m 0 , A 0 , A 1 and γ )
∀t 0 ∈ [0, 1), K t
0is compact w.r.t. the uniform convergence
∀(0, N) ∈ K 0 , ∀s ≥ 0, T s (0, N) ∈ K s
∀N ∈ Y 0 , ∃s 0 (N) ≥ 0, ∀s ≥ s 0 , T s (0, N) ∈ K s Λ = {(t, N) / t ∈ R / Z , N ∈ Λ t } where Λ t = T
n∈ N T n (K t ) Λ is compact, stable by T s for all s ≥ 0
For any neighborhood U of Λ and any N ∈ Y 0 , T s (0, N) ∈ U
for every s ≥ s 0 (U, N)
13/73
Mathematical analysis: attractor
K t
0= {t 0 }×{N ∈ Y t
0/ N min ≤ N(t) ≤ N max and N is L-Lipschitz}
(N min , N max and L are explicit functions of m 0 , A 0 , A 1 and γ )
∀t 0 ∈ [0, 1), K t
0is compact w.r.t. the uniform convergence
∀(0, N) ∈ K 0 , ∀s ≥ 0, T s (0, N) ∈ K s
∀N ∈ Y 0 , ∃s 0 (N) ≥ 0, ∀s ≥ s 0 , T s (0, N) ∈ K s Λ = {(t, N) / t ∈ R / Z , N ∈ Λ t } where Λ t = T
n∈ N T n (K t ) Λ is compact, stable by T s for all s ≥ 0
For any neighborhood U of Λ and any N ∈ Y 0 , T s (0, N) ∈ U
for every s ≥ s 0 (U, N)
13/73
Mathematical analysis: attractor
K t
0= {t 0 }×{N ∈ Y t
0/ N min ≤ N(t) ≤ N max and N is L-Lipschitz}
(N min , N max and L are explicit functions of m 0 , A 0 , A 1 and γ )
∀t 0 ∈ [0, 1), K t
0is compact w.r.t. the uniform convergence
∀(0, N) ∈ K 0 , ∀s ≥ 0, T s (0, N) ∈ K s
∀N ∈ Y 0 , ∃s 0 (N) ≥ 0, ∀s ≥ s 0 , T s (0, N) ∈ K s Λ = {(t, N) / t ∈ R / Z , N ∈ Λ t } where Λ t = T
n∈ N T n (K t ) Λ is compact, stable by T s for all s ≥ 0
For any neighborhood U of Λ and any N ∈ Y 0 , T s (0, N) ∈ U
for every s ≥ s 0 (U, N)
13/73
Mathematical analysis: attractor
K t
0= {t 0 }×{N ∈ Y t
0/ N min ≤ N(t) ≤ N max and N is L-Lipschitz}
(N min , N max and L are explicit functions of m 0 , A 0 , A 1 and γ )
∀t 0 ∈ [0, 1), K t
0is compact w.r.t. the uniform convergence
∀(0, N) ∈ K 0 , ∀s ≥ 0, T s (0, N) ∈ K s
∀N ∈ Y 0 , ∃s 0 (N) ≥ 0, ∀s ≥ s 0 , T s (0, N) ∈ K s Λ = {(t, N) / t ∈ R / Z , N ∈ Λ t } where Λ t = T
n∈ N T n (K t ) Λ is compact, stable by T s for all s ≥ 0
For any neighborhood U of Λ and any N ∈ Y 0 , T s (0, N) ∈ U
for every s ≥ s 0 (U, N)
13/73
Mathematical analysis: attractor
K t
0= {t 0 }×{N ∈ Y t
0/ N min ≤ N(t) ≤ N max and N is L-Lipschitz}
(N min , N max and L are explicit functions of m 0 , A 0 , A 1 and γ )
∀t 0 ∈ [0, 1), K t
0is compact w.r.t. the uniform convergence
∀(0, N) ∈ K 0 , ∀s ≥ 0, T s (0, N) ∈ K s
∀N ∈ Y 0 , ∃s 0 (N) ≥ 0, ∀s ≥ s 0 , T s (0, N) ∈ K s Λ = {(t, N) / t ∈ R / Z , N ∈ Λ t } where Λ t = T
n∈ N T n (K t ) Λ is compact, stable by T s for all s ≥ 0
For any neighborhood U of Λ and any N ∈ Y 0 , T s (0, N) ∈ U
for every s ≥ s 0 (U, N)
13/73
Mathematical analysis: attractor
K t
0= {t 0 }×{N ∈ Y t
0/ N min ≤ N(t) ≤ N max and N is L-Lipschitz}
(N min , N max and L are explicit functions of m 0 , A 0 , A 1 and γ )
∀t 0 ∈ [0, 1), K t
0is compact w.r.t. the uniform convergence
∀(0, N) ∈ K 0 , ∀s ≥ 0, T s (0, N) ∈ K s
∀N ∈ Y 0 , ∃s 0 (N) ≥ 0, ∀s ≥ s 0 , T s (0, N) ∈ K s Λ = {(t, N) / t ∈ R / Z , N ∈ Λ t } where Λ t = T
n∈ N T n (K t ) Λ is compact, stable by T s for all s ≥ 0
For any neighborhood U of Λ and any N ∈ Y 0 , T s (0, N) ∈ U
for every s ≥ s 0 (U, N)
14/73
Mathematical results (J.-C. Yoccoz & Birkeland 1998)
Theorem: Λ is an attractor for (T s ) s≥0 , which attracts any initial condition (0, N) with N ∈ Y 0 .
Unseasonal model: an equilibrium exists, that is stable when
1 ≤ γ ≤ γ 0 ≈ 6.2 . Hopf bifurcation when γ crosses γ 0 ?
14/73
Mathematical results (J.-C. Yoccoz & Birkeland 1998)
Theorem: Λ is an attractor for (T s ) s≥0 , which attracts any initial condition (0, N) with N ∈ Y 0 .
Unseasonal model: an equilibrium exists, that is stable when
1 ≤ γ ≤ γ 0 ≈ 6.2 . Hopf bifurcation when γ crosses γ 0 ?
15/73
Plan
1 The model
Voles in the Arctic Description of the model Mathematical analysis
2 Simulation study Principle
Exploring the parameter space
Study of an attractor: (0.15 ; 0.30 ; 8.25)
16/73
Discretization of the dynamical system
Each year is divided into p = 100 intervals
Define I k = [(k − 1)/p; k /p) for k ≥ −pA 1 , k ∈ N n k : number of births in I k
N k : average number of mature females during I k
e k : average of m ρ,ε over I k
s k : fraction of surviving mature females among females born k time intervals before
n i = m(N i ) × N i × e i p N i =
A
1p
X
k =1
(s k × n i−k )
16/73
Discretization of the dynamical system
Each year is divided into p = 100 intervals
Define I k = [(k − 1)/p; k /p) for k ≥ −pA 1 , k ∈ N n k : number of births in I k
N k : average number of mature females during I k
e k : average of m ρ,ε over I k
s k : fraction of surviving mature females among females born k time intervals before
n i = m(N i ) × N i × e i p N i =
A
1p
X
k =1
(s k × n i−k )
16/73
Discretization of the dynamical system
Each year is divided into p = 100 intervals
Define I k = [(k − 1)/p; k /p) for k ≥ −pA 1 , k ∈ N n k : number of births in I k
N k : average number of mature females during I k
e k : average of m ρ,ε over I k
s k : fraction of surviving mature females among females born k time intervals before
n i = m(N i ) × N i × e i p N i =
A
1p
X
k =1
(s k × n i−k )
17/73
Principle of the simulation experiments
A 1 = 2 m 0 = 50 ε = 0.1 (A 0 ; ρ ; γ) varying
Stationary behaviour: t ≥ t 0 = 10000 (for one arbitrary initial condition).
t = 0 at the end of summer Bifurcation diagram: (x, N(t)) .
Projection on R 3 : (N(t ), N(t + 1), N(t + 2)) (Whitney
theorem, assuming a fractal dimension D < 3/2 for ω(N)).
17/73
Principle of the simulation experiments
A 1 = 2 m 0 = 50 ε = 0.1 (A 0 ; ρ ; γ) varying
Stationary behaviour: t ≥ t 0 = 10000 (for one arbitrary initial condition).
t = 0 at the end of summer Bifurcation diagram: (x, N(t)) .
Projection on R 3 : (N(t ), N(t + 1), N(t + 2)) (Whitney
theorem, assuming a fractal dimension D < 3/2 for ω(N)).
17/73
Principle of the simulation experiments
A 1 = 2 m 0 = 50 ε = 0.1 (A 0 ; ρ ; γ) varying
Stationary behaviour: t ≥ t 0 = 10000 (for one arbitrary initial condition).
t = 0 at the end of summer Bifurcation diagram: (x, N(t)) .
Projection on R 3 : (N(t ), N(t + 1), N(t + 2)) (Whitney
theorem, assuming a fractal dimension D < 3/2 for ω(N)).
18/73
(A 0 = 0.18 ; ρ = 0.41 ; γ ∈ [2; 16])
19/73
(A 0 = 0.18 ; ρ = 0.41 ; γ ∈ [2; 16]) : Fractal dimension
2 4 6 8 10 12 14 16
−0.2 0 0.2 0.4 0.6 0.8 1 1.2
γ df(Λγ)
Evolution de la dimension fractale : A0 = 0.18 ; ρ = 0.41 ; ε = 0
20/73
(A 0 = 0.18 ; ρ = 0.41 ; γ = 8.61) : d f ≈ 1.06
1 2
3 4
5 6
1 2 3 4 5 6 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5
N(t) A0 = 0.18 A1 = 2 ; ρ = 0.410 ε = 0 ; m0 = 50 γ = 8.61 m(N) C1 10001 ≤ t ≤ 19998 ; 1.2935 ≤ N ≤ 5.1125 ; 100 pas de temps par an
N(t+1)
N(t+2)
21/73
(A 0 = 0.18 ; ρ = 0.41 ; γ = 9.89) : d f ≈ 0.92
1 2
3 4
5
1 2 3 4 5 1 1.5 2 2.5 3 3.5 4 4.5 5
6
2 10 8 4
N(t) A0 = 0.18 A1 = 2 ; ρ = 0.410 ε = 0 ; m0 = 50 γ = 9.89 m(N) C1 10001 ≤ t ≤ 19998 ; 1.2334 ≤ N ≤ 4.6427 ; 100 pas de temps par an
9
N(t+1) 3 7 1 5
N(t+2)
22/73
(A 0 = 0.18 ; ρ ∈ [0 ; 0.5] ; γ = 8.25) [see video]
23/73
(A 0 = 0.18 ; ρ = 0.16 ; γ = 8.25) : d f ≈ 0.99
1.6 1.8 2 2.2 2.4 2.6 2.8 3
1.5 2 2.5 3 1.6 1.8 2 2.2 2.4 2.6 2.8 3
N(t) A0 = 0.18 A1 = 2 ; ρ = 0.160 ε = 0.1 ; m0 = 50 γ = 8.25 m(N) C1 10002 ≤ t ≤ 19998 ; 1.6909 ≤ N ≤ 2.6275 ; 100 pas de temps par an
N(t+1)
N(t+2)
24/73
(A 0 = 0.15 ; ρ = 0.78; γ = 8.25) : d f ≈ 1.25
25/73
(A 0 ∈ [0 ; 0.4] ; ρ = 0.30 ; γ = 8.25)
26/73
(A 0 = 0.18 ; ρ = 0.30 ; γ = 8.25) : d f ≈ 1.19
27/73
(A 0 = 0.15 ; ρ = 0.30 ; γ = 8.25) : d f ≈ 1.33
28/73
One connex component (physical measure)
29/73
Injectivity of the projection R 201 7→ R 3
0 10 20 30 40 50 60 70 80
0 10 20 30
j = no sur la carte sup des rapports des normes LInf(R201) / LInf(R3) sur la boule B(xj,0.1) de LInf(R201)
0 10 20 30 40 50 60 70 800
500 1000 1500
nombre de points dans la boule B(xj,0.1) de LInf(R201) Test de l’injectivité de la projection LInf(R201) → LInf(R3) ; δt = 0
30/73
Dynamics (1): equilibrium, period-2, period-3
2 2.5
3 3.5
4 4.5
5
1.2 1.3 1.4 1.5 1.6 1.7 1.8
2 2.5 3 3.5 4 4.5
A B
N(2t+1)
C2
N(2t+2)
C3 C1
N(2t+3)
31/73
Dynamics (2)
2 2.5
3 3.5
4 4.5
5
1.2 1.3 1.4 1.5 1.6 1.7 1.8
2 2.5 3 3.5 4 4.5
2 eq 1
0
N(2t+1)
6
3
4
N(2t+2)
5
N(2t+3)
32/73
Dynamics (3) [+ see video]
2 2.5 3 3.5 4 4.5 5
1.2 1.4 1.6 1.8
2 2.5 3 3.5 4 4.5
eq D3 B3
B0 A0 C0 A1 D0 C3
B1
N(2t+1)
C1 D1 A2 B2 A3 D2 C2
N(2t+2)
N(2t+3)
33/73
Remarkable points: local unstable manifolds
2 2.5
3 3.5
4 4.5
5
1.2 1.3 1.4 1.5 1.6 1.7 1.8
2 2.5 3 3.5 4 4.5
A B
N(2t+1)
C2
N(2t+2)
C3 C1
N(2t+3)
34/73
Equilibrium: expansion of the unstable manifold (1)
2 2.5 3 3.5 4 4.5 5 5.5
1.2 1.4 1.6 1.8 2 2 3 4 5 6
N(2t+1) Variéte instable : f
12(W
u(x
eq
))
N(2t+2)
N(2t+3)
35/73
Equilibrium: expansion of the unstable manifold (2)
2 2.5 3 3.5 4 4.5 5 5.5
1.2 1.4 1.6 1.8 2 2 3 4 5 6
N(2t+1) Variéte instable : f
15(W
u(x
eq
))
N(2t+2)
N(2t+3)
36/73
Equilibrium: expansion of the unstable manifold (3)
2 2.5 3 3.5 4 4.5 5 5.5
1.2 1.4 1.6 1.8 2 2 3 4 5 6
N(2t+1) Variéte instable : f
16(W
u(x
eq
))
N(2t+2)
N(2t+3)
37/73
Equilibrium: expansion of the unstable manifold (4)
2 2.5 3 3.5 4 4.5 5 5.5
1.2 1.4 1.6 1.8 2 2 3 4 5 6
N(2t+1) Variéte instable : f
17(W
u(x
eq
))
N(2t+2)
N(2t+3)
38/73
Equilibrium: global unstable manifold
2 3
4 5
6
1.2 1.4 1.6 1.8 2 2 2.5 3 3.5 4 4.5 5 5.5
N(2t+1) Variéte instable : f
18(W
u(x
eq))
N(2t+2)
N(2t+3)
39/73
Period 2: expansion of the unstable manifold (1)
2 2.5
3 3.5
4 4.5
5
1.2 1.3 1.4 1.5 1.6 1.7 1.8
2 2.5 3 3.5 4 4.5
N(2t+1) Varu(eq. num 2) − step 07
N(2t+2)
N(2t+3)
40/73
Period 2: expansion of the unstable manifold (2)
2 2.5
3 3.5
4 4.5
5
1.2 1.3 1.4 1.5 1.6 1.7 1.8
2 2.5 3 3.5 4 4.5
N(2t+1) Varu(eq. num 2) − step 08
N(2t+2)
N(2t+3)
41/73
Period 2: expansion of the unstable manifold (3)
2 2.5
3 3.5
4 4.5
5
1.2 1.3 1.4 1.5 1.6 1.7 1.8
2 2.5 3 3.5 4 4.5
N(2t+1) Varu(eq. num 2) − step 09
N(2t+2)
N(2t+3)
42/73
Period 2: expansion of the unstable manifold (4)
2 2.5
3 3.5
4 4.5
5
1.2 1.3 1.4 1.5 1.6 1.7 1.8
2 2.5 3 3.5 4 4.5
N(2t+1) Varu(eq. num 2) − step 10
N(2t+2)
N(2t+3)
43/73
Period 3: expansion of the unstable manifold (1)
2 2.5
3 3.5
4 4.5
5
1.2 1.3 1.4 1.5 1.6 1.7 1.8
2 2.5 3 3.5 4 4.5
N(2t+1) Varu(per. 3) − step 15
N(2t+2)
N(2t+3)
44/73
Period 3: expansion of the unstable manifold (2)
2 2.5
3 3.5
4 4.5
5
1.2 1.3 1.4 1.5 1.6 1.7 1.8
2 2.5 3 3.5 4 4.5
N(2t+1) Varu(per. 3) − step 16
N(2t+2)
N(2t+3)
45/73
Period 3: expansion of the unstable manifold (3)
2 2.5
3 3.5
4 4.5
5
1.2 1.3 1.4 1.5 1.6 1.7 1.8
2 2.5 3 3.5 4 4.5
N(2t+1) Varu(per. 3) − step 17
N(2t+2)
N(2t+3)
46/73
Period 3: expansion of the unstable manifold (4)
2 2.5
3 3.5
4 4.5
5
1.2 1.3 1.4 1.5 1.6 1.7 1.8
2 2.5 3 3.5 4 4.5
N(2t+1) Varu(per. 3) − step 18
N(2t+2)
N(2t+3)
47/73
Period 3: expansion of the unstable manifold (5)
2 2.5
3 3.5
4 4.5
5
1.2 1.3 1.4 1.5 1.6 1.7 1.8
2 2.5 3 3.5 4 4.5
N(2t+1) Varu(per. 3) − step 19
N(2t+2)
N(2t+3)
48/73
Period 3: expansion of the unstable manifold (6)
2 2.5
3 3.5
4 4.5
5
1.2 1.3 1.4 1.5 1.6 1.7 1.8
2 2.5 3 3.5 4 4.5
N(2t+1) Varu(per. 3) − step 20
N(2t+2)
N(2t+3)
49/73
Period 3: expansion of the unstable manifold (7)
2 2.5
3 3.5
4 4.5
5
1.2 1.3 1.4 1.5 1.6 1.7 1.8
2 2.5 3 3.5 4 4.5
N(2t+1) Varu(per. 3) − step 21
N(2t+2)
N(2t+3)
50/73
Period 3: expansion of the unstable manifold (8)
2 2.5
3 3.5
4 4.5
5
1.2 1.3 1.4 1.5 1.6 1.7 1.8
2 2.5 3 3.5 4 4.5
N(2t+1) Varu(per. 3) − step 22
N(2t+2)
N(2t+3)
51/73
Period 3: global unstable manifold
2 2.5
3 3.5
4 4.5
5
1.2 1.3 1.4 1.5 1.6 1.7 1.8
2 2.5 3 3.5 4 4.5
N(2t+1) Varu(per. 3) − step 23
N(2t+2)
N(2t+3)
52/73
Dynamics in continuous time: equilibrium
0 0.5 1 1.5 2 2.5 3 3.5 4
0 1 2 3 4 5 6 7
L’équilibre : (N(t),N(t+1),N(t+2))=(4.5381,1.2352,4.5381)
t
N(t)
53/73
Dynamics in continuous time: birth at equilibrium
−2 −1.8 −1.6 −1.4 −1.2 −1 −0.8 −0.6 −0.4 −0.2 0
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45
Naissances à l’équilibre : (N(t),N(t+1),N(t+2))=(4.5381,1.2352,4.5381)
t
naiss(t)
54/73
Dynamics in continuous time: period-3 orbit (triangle)
−5 −4 −3 −2 −1 0 1 2 3 4 5
0 1 2 3 4 5 6 7
A0 = 0.15 ; point 78 [ 73 → 78 → 60 ]
t−197440
N(t)
55/73
Attractor: there is a “fold”
2 3
4 5
6
1.2 1.4 1.6 1.8 2 2 2.5 3 3.5 4 4.5 5 5.5
N(2t+1) Variéte instable : f
18(W
u(x
eq))
N(2t+2)
N(2t+3)
56/73
Folding (1): before the fold
57/73
Folding (2)
58/73
Folding (3): curvature
0 0.5 1 1.5 2 2.5 3 3.5 4
−0.5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5
t log10(Maximum de courbure a l’instant t)
Formation du pli sur 4 ans
59/73
Fold: dynamics in continuous time
−20 −1 0 1 2 3 4
1 2 3 4 5 6 7
t N
j(t)
Population mature au niveau du pli
j = 1
j = 690
j = 999
60/73
Global dynamics: summary
shrinking stretching
folding (stable in the parameter space, see video)
“branching” (from equilibrium/period-2 to period-3)
60/73
Global dynamics: summary
shrinking stretching
folding (stable in the parameter space, see video)
“branching” (from equilibrium/period-2 to period-3)
60/73
Global dynamics: summary
shrinking stretching
folding (stable in the parameter space, see video)
“branching” (from equilibrium/period-2 to period-3)
61/73
Local geometry: zoom on a “filament”
23
62/73
Local geometry: zoom on a peak
63/73
Local geometry: zoom on a peak
54
64/73
Dynamics in continuous time: in a “peak”
−5 −4 −3 −2 −1 0 1 2 3 4 5
0 1 2 3 4 5 6 7
A0 = 0.15 ; point 54 [ 69 → 54 → 01−05 ]
t−182848
N(t)
65/73
Sensitivity to initial conditions around the equilibrium
66/73
Conclusions: conjectures
For some of the parameter values:
periodic orbits, doubling period bifurcations, sub-harmonic cascade.
Hopf bifurcations.
Chaotic dynamics, Henon-type attractors.
For the attractor (0.15 ; 0.30 ; 8.25) :
Strange attractor. Persistent chaotic dynamics.
Fractal dimension in (1 , 3/2) . Quasi-hyperbolicity.
Spectral decomposition.
66/73
Conclusions: conjectures
For some of the parameter values:
periodic orbits, doubling period bifurcations, sub-harmonic cascade.
Hopf bifurcations.
Chaotic dynamics, Henon-type attractors.
For the attractor (0.15 ; 0.30 ; 8.25) :
Strange attractor. Persistent chaotic dynamics.
Fractal dimension in (1 , 3/2) . Quasi-hyperbolicity.
Spectral decomposition.
67/73
Biological conclusions
Simple model, unrealistic.
Sufficient to induce chaotic dynamics, not well understood mathematically.
Qualitatively looks like real data.
68/73
69/73
Smoothing of the fertility rate
0 0.5 1 1.5 2 2.5
0 10 20 30 40 50 60
N = effectif mature
m(N) = fécondite individuelle
Comparaison des fonctions de fécondité ; m
0
= 50 ; γ = 8.25
fécondité continue fécondité C170/73
Smoothing of the seasonal factor
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0 0.2 0.4 0.6 0.8 1
t en années m
ρ,ε ete(N) = facteur saisonnier
Régularisation du facteur saisonnier (ρ = 0.41)
Saisons non−continues : εete = 0 Saisons C1 : εete = 0.1
71/73
Estimation of the fractal dimension
−51 −4.5 −4 −3.5 −3 −2.5 −2 −1.5 −1 −0.5 0
1.5 2 2.5 3 3.5 4 4.5 5
log10(r) log10(N(r))
A0=0.15 ; ρ=0.3 ; γ=8.25 ; dimension fractale de l’attracteur
log10(N(r)) = −1.3289× log10(r)+1.1915
72/73
Dynamics on the attractor: expansion areas
73/73
Fold: unstable direction
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
−3
−2
−1 0 1 2 3 4 5 6
t
dT
2 x(t)
Premier vecteur propre de la différentielle de T
2en x=x
j