Mathematical Biosciences 152 (1998) 165177

Optimal harvesting policy for single

population with periodic coecients 1
Meng Fan *, Ke Wang
Department of Mathematics, Northeast Normal University, Changchun 130024, People's Republic of
China
Received 23 December 1997; received in revised form 24 April 1998

Abstract
In this paper, we examine the exploitation of single population modeled by timedependent Logistic equation with periodic coecients. First, it is shown that the timedependent periodic Logistic equation has a unique positive periodic solution, which is
globally asymptotically stable for positive solutions, and we obtain its explicit representation. Further, we choose the maximum annual-sustainable yield as the management objective, and investigate the optimal harvesting policies for constant harvest and
periodic harvest. The optimal harvest eort that maximizes the annual-sustainable yield,
the corresponding optimal population level, the corresponding harvesting time-spectrum, and the maximum annual-sustainable yield are determined, and their explicit
expressions are obtained in terms of the intrinsic growth rate and the carrying capacity
of the considered population. Our interesting and brief results generalize the classical
results of Clark for a population described by the autonomous logistic equation in
renewable resources management. 1998 Elsevier Science Inc. All rights reserved.
Keywords: Optimal harvesting policy; Time-dependent logistic equation with periodic
coecients; EulerLagrange equation; Harvesting time-spectrum

Corresponding author.
Project Supported by Natural Sciences Foundation of People's Republic of China and Natural
Sciences Foundation of Liaoning Province of People's Republic of China.
1

0025-5564/98/$19.00 1998 Elsevier Science Inc. All rights reserved.

PII: S 0 0 2 5 - 5 5 6 4 ( 9 8 ) 1 0 0 2 4 - X

166

M. Fan, K. Wang / Mathematical Biosciences 152 (1998) 165177

1. Introduction
The optimal management of renewable resources, which has a direct relationship to sustainable development, has been studied extensively by many
authors. Economic and biological aspects of renewable resources management
have been considered by Clark [1,2] and other authors [36].
Suppose that x(t), the density of the resource population at time t, satises
the famous Logistic equation:

dx
x
rx 1
;
1
dt
K
where r, assumed to be positive constant, is called the intrinsic growth rate, and
the positive constant K is usually referred to as the environmental carrying
capacity, or saturation level. Now suppose that the resource population described by the logistic Eq. (1) is subject to exploitation, under the catch-perunit-eort hypothesis
h Ex:
The equation of the harvested population reads

dx
x
rx 1
Ex;
dt
K
where E denotes the harvesting eort. As is well known, in order to gain the
maximum sustainable yield, the optimal harvesting eort should be
1
E r;
2
and the maximum sustainable yield, corresponding to E is
1
h Kr;
4
while the corresponding optimal population level is given by
1
x K:
2
Almost all known models assume that the intrinsic growth rate and the
carrying capacity of the population are independent of time t, but in the real
world the natural growth rates of many populations vary with t; for example,
due to seasonality. Few results on non-autonomous biological systems are
found in the literature, which is much harder to deal with than autonomous
systems. To our knowledge, there have been no results on harvest and the
continuously harvesting time-spectrum for non-autonomous harvested systems
in the literature.
In this paper, we discuss the harvesting problems of population resources
under the assumption that r and K are both periodic functions with respect to t.

M. Fan, K. Wang / Mathematical Biosciences 152 (1998) 165177

167

Qualitative methods and optimal control methods are applied to determine the
optimal harvesting policy, including the optimal harvesting time-spectrum. We
choose annual-sustainable harvesting yield as the management objective.
The organization of the paper is as follows: to begin with, a theorem on the
existence and uniqueness of periodic solution for the logistic equation with
periodic coecients is proved in Section 2. In subsequent portions of this
paper, the optimal harvesting policies are determined for both constant eort
harvest and periodic eort harvest.
2. Periodicity of the logistic equation with periodic coecients
Consider the generalized logistic equation


dx
x
rtx 1
:
2
dt
Kt
Here r(t) and K(t) are both continuous period 1 functions. Clearly, (Eq. (2)) is a
Bernoulli type equation. If Kt 6 0, the solution of (Eq. (2)) with initial value
xt0 x0 x0 6 0, is given by
0
8
9
8
9 11
Zt
Zt
Zt
<
=
<
=
rs
B1
C
exp rs ds dsA : 3
xt @ exp rsds
:
;
:
;
x0
Ks
t0

t0

Clearly, the null solution xt 0 exists. By the uniqueness of solutions, we

can see that solutions with positive initial values remain positive. Hence, x(t),
which is represented by (Eq. (3)) is positive for all t P t0 if xt0 x0 > 0. For
biological considerations, we are only interested in positive solutions. In this
paper, we always need xt0 x0 > 0.
R1
Theorem 2.1. If Kt > 0; 0 rsds > 0, then there exists a unique positive period
1 solution xP t of (Eq. (2)), which is represented by
81
9
8
9 11
0
10t1
Z
<Z
=
< Zt
=
rs
xP t @ exp
rsds 1A@
exp rs ds dsA :
:
;
:
;
Ks
0

4
In addition, xP t is globally asymptotically stable for xt with positive initial
value xt0 x0 > 0 in the sense that
lim jxt xP tj 0:

t!1

Proof. It is easy to verify that Eq. (4) is a period 1 solution of Eq. (2). We only
need to show that every solution of Eq. (2) with positive initial value tends to
Eq. (4) as t ! 1. In Eq. (3),

168

M. Fan, K. Wang / Mathematical Biosciences 152 (1998) 165177

8
<

1
exp
:
x0

Zt
rs ds
t0

9
=
;

! 0;

t ! 1:

So it is only necessary to verify that

8
9
9
8 t
t1
Zt
Z
< Zt
=
=
< Z
rs
1
rs
exp rs ds ds
exp rs ds
:
;
;
:
Ks
A 1 Ks
t0

ds ! 0;

t ! 1;
6
R1
where A exp 0 rs ds. We can rewrite Eq. (6) as
8 t
8
9
9
t1
Zt
Zt
Z
Z
<
<
=
=
rs
1
rs
exp rs ds ds
exp rs ds ds
:
:
;
;
Ks
A 1 Ks
t0

90 t
8s
9
Zt
= Z rs
<Z
=
B
rs ds
exp
exp rs ds @
:
;
:
;
Ks

8
<

t0

A1
8
<

tZ
1

t0

t0

8s
9
<Z
=
rs
C
exp
rs ds dsA
:
;
Ks
t0

9
Zt
=
: exp rs ds F t;
:
;

t0

where

8 t
9
<Z
=
rt
exp
rs ds
F 0 t
:
;
Kt
0

t0

8t1
8 t
9
91
Z
Z
<
=
<
=
1 B rt 1
rt
C
exp
exp
rs ds
rs ds A

@
:
; Kt
:
;
A 1 Kt 1

8 t
<Z

rt
exp
:
Kt

t0

rs ds

9
=

t0

8 t
<Z

1 rt
exp
:
A 1 Kt

t0

t0

90
8t1
9
1
=
<Z
=
rs ds @ exp
rs ds 1A 0;
;
:
;
t

M. Fan, K. Wang / Mathematical Biosciences 152 (1998) 165177

169

since rs is periodic with period 1. Hence, F t constant. So Eqs. (5) and (7)
imply Eq. (6).The uniqueness of the periodic solution follows from the globally
asymptotical stability. So the proof is completed.
3. Optimal harvesting policy for constant eort harvesting
Now, we consider a population of size x(t), which obeys the generalized
logistic growth law, is harvested by means of a constant eort. Equation of the
harvested population reads


dx
x
rtx 1
Ex;
8
dt
Kt
where rt and Kt are both continuous period 1 functions, and Kt is positive
and
Z1
rt dt > E > 0:

Under the above assumptions, the solution of Eq. (8) with initial value
xE t0 x0 > 0 is given by
0
8
9
< Zt
=
B1
xE t @ exp rs E ds
:
;
x0
t0

Zt

t0

8
<

rs
exp
:
Ks

Zt
rsE ds
s

9
=
;

11
C
dsA :

10

Since the null solution xt 0 exists, the uniqueness of solutions imply

xE t > 0.
In analogy to Theorem 2.1, we can easily prove the following theorem:
Theorem 3.1. Under the above assumptions on Eq. (8), there exists a unique
positive period 1 solution xPE t of Eq. (8), which is represented by
81
9
0
1
<Z
=
rs E ds 1A
xPE t @ exp
:
;
0

8
9 11
0t1
Zt
Z
<
=
@ rs exp rsE ds dsA :
:
;
Ks
t

11

170

M. Fan, K. Wang / Mathematical Biosciences 152 (1998) 165177

In addition, xPE t is globally asymptotically stable for xE t with positive initial

value xE t0 x0 > 0 in the sense that
lim jxE t xPE tj 0:

t!1

So the annual-sustainable yield is

Z1
Y E ExPE t dt
0

Z1

o
E ds E
dt
 Rt

R t1 rs
exp s rs E ds ds
t
Ks
E exp

Z1

R t1
0

nR

1
rs
0

AEeE E
dt : AEeE Ef E;


Rt
rs
exp

rs

E
ds
ds
s
Ks
12

where
Z1
f E

R t1
0

Since

rs
Ks

1
dt:

Rt
exp s rs E ds ds


xPE t

denoted by Eq. (11) is a positive period 1 solution, we have

Z1
13
Y E ExPE t dt > 0:
0

Using the assumption Eq. (9), one can obtains

81
9
<Z
=
rs ds E 1 > 0:
AEeE E E exp
:
;

14

Eqs. (12)(14) imply f E > 0.

Our objective is to nd an E such that Y E reaches its maximum at E E .
This is an optimization of a function. We have
Y 0 E A exp fEg AE exp fEg 1f E
AE exp fEg Ef 0 E; Y 0 0
A 1f 0 > 0; Y 0 ln A ln Af ln A < 0:
By continuity, there exists an E 2 0; ln A such that Y 0 E 0.
Let
X fE: E 2 0; ln A; Y 0 E 0g
be called the admissible set. For practical considerations, the desired optimal
harvesting eort must lie in X.

M. Fan, K. Wang / Mathematical Biosciences 152 (1998) 165177

171

4. Optimal harvesting policy for periodic eort harvesting

The assumption in Section 3 that the harvesting eort is a constant is too
special. In this section, we assume the instantaneous harvesting eort is a
function of t, say E(t), which is a non-negative, piece-wise continuous period 1
function. The set of such functions is denoted by E; and the element in E, is
called a harvesting eort function.
Let
8
9
Z1
<
=
 E 2 E: Et dt 1 :
E
:
;
0

 is called a harvesting time-spectrum, and is very im 2E

The element Et
portant in practial management. Obviously, for any Et 2 E and Et 6 0,
 and a unique n such that Et nEt.
 2E

there exists an Et
R1

In fact, we only need to set n 0 Et dt; Et Et=n. We call n the
harvesting eort coecient corresponding to the harvesting time-spectrum

Et.
Now we investigate the optimal harvesting policy, namely, the optimal
harvesting eort and the optimal harvesting time-spectrum.
Suppose Et 2 E, consider the equation of the harvested population


dx
x
rtx 1
Etx;
15
dt
Kt
where r(t) and K(t) are both continuous period 1 functions, Kt is positive and
Z1
rt Et dt > 0:

16

Since E(t) is non-negative, by Eq. (16), we have

Z1

Z1
rt dt >

Et dt > 0:
0

The solution of Eq. (15) with xE t0 x0 > 0 reads

0
8
9
Zt
<
=
B1
xE t @ exp rs Es ds
:
;
x0
t0

Zt

t0

9 11
8 t
Z
=
<
rs
C
exp rs Es ds dsA :
;
:
Ks
s

17

172

M. Fan, K. Wang / Mathematical Biosciences 152 (1998) 165177

By the existence of the null solution x(t) 0 and the uniqueness of solutions,
we can conclude that xE t > 0.
Analogous to the proof of Theorem 2.1, we can easily reach the following
theorem:
Theorem 4.1. Under the above assumptions on Eq. (15), there exists a unique
positive period 1 solution xPE t of Eq. (15), which is represented by
81
9
0
1
<Z
=
rs Es ds 1A
xPE t @ exp
:
;
0

8 t
9 11
0t1
Z
Z
<
=
rs
exp rs Es ds dsA :
@
:
;
Ks
t

18

In addition, xPE t is globally asymptotically stable for xE t with positive initial

value xE t0 x0 > 0 in the sense that
lim jxE t xPE tj 0:

t!1

So the annual-sustainable harvesting yield can be expressed as

Z1
Y Et

EtxPE tdt


n
o

Z1 Et exp R 1 rs Es ds 1
0
dt:
R t1 rs
 Rt

exp s rs Es ds ds
t
Ks

19

Let
D

8
<
:

Z1
E 2 E:
0

9
=

rt Et dt > 0 :
;

20

D is called the set of admissible curve. What we wish to do is to nd E t 2 D

such that Y Et reaches its maximum, namely to nd the optimal harvesting
eort, which is a dynamic optimization problem of a functional. It is only
necessary to solve
Y E t max Y Et:
Et2D

21

Theorem 4.2 (Main theorem). If rt P 2K 0 t=Kt, then there exists a unique

optimal harvesting eort, namely E t 12 rt K 0 t=Kt, which is the unique
solution of Eq. (21). The optimal harvesting time-spectrum corresponding to
E t is given by

M. Fan, K. Wang / Mathematical Biosciences 152 (1998) 165177

E t
E t R 1
:
E t dt
0

173

22

The corresponding population level is

1
x t Kt:
2
The maximum annual-sustainable harvesting yield reads
1
Y E t
4


Proof. Let
yt exp

Z1
rtKt dt:

24

8 t
<Z
:

23

9
=
Es ds :
;

25

Then
0

Et ln yt :

26

Eq. (19) can then be written

Y yt
Z1

R t1
t

Z1

Z1

exp

Rt
s

A=y1 1 lnyt0
dt

 Rs

 Rt

rs ds exp 0 Es ds exp 0 Es ds ds

A=y1 1 lnyt0
dt
 Rt
1
R t1 rs
yt t Ks exp s rs ds ys
ds
 0
1
A=y1 1 yt
dt
 Rt
1
R t1 rs
exp s rs ds ys
ds
t
Ks

Z1 A=y1 1

1
yt

0
dt;

Zt

where
tZ
1

Zt
t

Then

rs
Ks

8
9
< Zt
= 1
rs
exp rs ds
ds:
:
; ys
Ks
s

27

28

174

M. Fan, K. Wang / Mathematical Biosciences 152 (1998) 165177

8t1
<Z

1
rt 1

; yt 1 Kt yt
t
8
9
t1
Zt
Z
<
= 1
rs
exp rs ds
ds
rt
:
; ys
Ks
t
s
81
8
8 t1
9
9
9
<Z
=
< Zt
=
< Z
=
rt
rs ds exp Es ds exp Es ds
exp

:
;
:
;
:
;
Kt

Z 0 t

rt 1
exp
:
Kt 1

9
=

rs ds

rt 1
rt 1

rtZt
A=y1 1 rtZt:
Kt yt
Kt yt
On rearranging, we obtain
A=y1 1=yt

29

fZ 0 t rtZtgKt
:
rt

30

Hence Eq. (27) becomes


0
Z1 fZ 0 trtZtgKt
rt
dt:
Y^ Zt
Zt

31

On integrating by parts, we obtain



Kt
Y^ Zt
ln Zt0 Kt j10
rt
1

Z 
2
Kt 
0
0
lnZt Kt ln Zt dt:

rt

32

Let
W t ln Zt0 :

33

Then Eq. (32) becomes




Z1 
Kt
Kt 2
1
~
W t Kt j0
W tKtW t dt
Y W t
rt
rt
0

Z1

Kt 2
W t KtW t
rt

Z1
: F t; W ; W 0 dt:
0

Kt
W tKt
rt

0 !
dt

34

M. Fan, K. Wang / Mathematical Biosciences 152 (1998) 165177

175

To complete the proof, we do it in three steps: (1) nd W  t which maximize

Y~ W t in Eq. (34); (2) determine E t corresponding to W  t; (3) prove

E t 2 D.
First by the optimal control theory, a W  t that maximizes Y~ W t in
Eq. (34) must satisfy the EulerLagrange Equation in Ref. [7]:


oF
d oF

0:
35
oW dt oW 0
Substitution of F in Eq. (34) into Eq. (35) and simplication produces
1
W  t rt;
2
which is unique.
Second, determine the corresponding E t.
By Eq. (33), one obtains
1
ln Z  t0 rt:
2
Integrating Eq. (37) from 0 to t on both sides and rearranging gives
8
9
< 1 Zt
=
rs ds :
Z  t Z  0 exp
: 2
;

36

37

38

Substituting Eq. (38) into Eq. (30), and simplifying, we get

 t

R

1
2A=y 1 1 exp 2 rs ds
0
:
y  t
Z  0Kt

39

Hence
lny  t ln 2A=y  1 1

1
2

Zt

rs ds ln Z  0

ln Kt:

40

By Eq. (26), dierentiation of the above equation with respect to t produces

1
K 0 t
:
E t rt
2
Kt

41

What remains is only to show that E t 2 D which is dened by Eq. (20). The
assumption that rt P 2K 0 t=Kt implies
E t P 0:
By Eq. (41), Eq. (17) and the periodicity of K(t), we have

42

176

M. Fan, K. Wang / Mathematical Biosciences 152 (1998) 165177

Z1

Z1 

rt E t dt
0

2
1

1
K 0 t
rt
2
Kt

rt dt
0

dt

Z1

Z1

Z1

ln Kt dt
0

1
rt dt ln K1 ln K0
2

Z1
rt dt > 0:
0

43
From Eqs. (42) and (43), we can conclude that E t 2 D. So by the uniqueness
of solution of Eq. (35), E t is the unique solution of Eq. (21), which is the
optimal harvesting eort.
Substituting Eq. (41) into Eq. (18), we have
81
9
0
1
=
<Z  1
0
K
s
x t @ exp
rs
1A
:
2
Ks ;
0
8 t
9 11
0t1
 =
Z 
Z
<
0
rs
1
K s
exp
rs
ds dsA
@
:
;
Ks
2
Ks
9s
81
9
0 t8 1
1
<1Z
=
<Z
=
@ exp
rs ds exp
d ln Ks 1A
:2
;
:
;
0
0
8
9
8s
9 11
0t1
Z
< 1 Zt
=
<Z
=
rs
exp
rs ds exp
d ln Ks dsA
@
: 2
;
:
;
Ks
t
s
t
8
911
0t1

 Z
< 1 Zt
=
rs
1 K1
A2
rs ds A
1 @
exp
: 2
;
K0
Kt
t
s
8 s
9111
0
0
tZ
1
<1Z
=
 1 
2
A2 1 @
d @ exp
rs ds AA
:2
;
Kt
t
t
8 t1
9
0
11
<1Z
=
Kt  1 @
A2 1

rs ds 1A
exp
:2
;
2
t

1 Kt
Kt 1=2
A 1 A1=2 1
:

2
2

44
Substituting Eq. (41) into Eq. (19), we can get the maximum annual-sustainable yield Y E t:

M. Fan, K. Wang / Mathematical Biosciences 152 (1998) 165177

Z1

Z1 

1
K 0 t
rt
2
Kt

177

Kt
dt
2
0
0
0 1
1
Z
Z1
1@1
1

rtKt dt K1 K0A
rtKt dt: 45
2 2
4

Y E t

E tx t dt

Now we obtain the corresponding population level x t and the maximum

annual-sustainable harvesting yield Y E t, which are denoted by Eqs.
R 1 (23) and
(24) respectively, and x t > 0. It is easy to verify that E t E t= 0 E t dt
 this is to say that E is the optimal harvesting time-spectrum correis in E,
sponding to E t. So we have completed the proof of the main theorem.
Remark. In the main theorem, if rt r; Kt K; r; K are positive constants,
then
1
1
1
x K;
Y  Kr;
E r;
2
2
4
which are results obtained by Clark [1,2] for autonomous harvested system. So
the results in this paper generalize the classical results of Clark for renewable
resources management.
Acknowledgements
The authors would like to express their gratitude to Professors Qichang
Yang, Yujuan Zhang, Shu wen Zhang, Huimin Liu in Anshan Teacher College
for their helpful discussions and suggestions. The authors are grateful to referees for their careful reading of the manuscript and a number of excellent
criticisms and suggestions.
References
[1] C.W. Clark, Mathematical Bioeconomics: The Optimal Management of Renewal Resources,
Wiley, New York, 1976.
[2] C.W. Clark, Mathematical Bioeconomics: The Optimal Management of Renewable Resources, 2nd ed., Wiley, New York, 1990.
[3] G.V. Tsretkva, Construction of an optimal policy taking into account ecological constraints
(Russian), Modelling of natural system and optimal contral problems (Russian). (Chita),
Vo``Naukce'', Novosibirsk, 1995, pp. 6574.
[4] A.W. Leung, Optimal harvesting-coecient control of steady-state preypredator diusive
VolterraLotka systems, Appl. Math. Optim. 31 (2) (1995) 219.
[5] Eio'lko, Mariusz, Kozlowski, Some optimization models of growth in biology, IEEE Trans.
Automat. Cont. 40 (10) (1995) 1779.
[6] D.K. Bhatta charya, S. Begum, Bionomic equilibrium of two-species System I, Math. Biosci.
135 (2) (1996) 111.
[7] Trowtman, L. John, Variational Calculus and Optimal Control, Springer, New York, 1996.