Environmental Stochasticity and Minimum Viable Populations

[Doug Bashford]

� Area: Sci.Environm
���������������������������������������������������������
Msg#: 52 Date: 03-12-94 16:41
From: Alan Mcgowen Read: Yes Replied:
No
To: All Mark:

Subj: Ecocentral 9
������������������������������������������������������������������������������
From: Alan McGowen <al...@igc.apc.org>

/* Written 12:47 am Apr 12, 1992 by alanm in cdp:sci.environmen */
/* -!!!!!!!!- "ECO CENTRAL 9" -!!!!!!!!- */
ECO CENTRAL 9

Environmental Stochasticity
and Minimum Viable Populations

Real environments fluctuate, and this induces fluctuation in the
parameters of population models such as intrinsic growth rate r
and carrying capacity K. To incorporate environmental
fluctuation, we replace the deterministic models of a population
size as a function of time, N(t), with a probability distribution
f(n,t), the probability that the population is of size n at time
t. Just as the population models we have looked at previously had
equilibrium solutions Ni*, so in stochastic models there are
equilibrium probability distributions f*(n). The strength of the
environmental fluctuation is measured by the coefficient of
variance, CV, of the distribution f*, also called sigma^2.

In the deterministic case, we saw in ECO CENTRAL 7 that the
stability of the equilibrium was controlled by the real part of
the eigenvalues of the community matrix: if the largest of these
was negative the equilibrium is neighborhood stable:

1) lambda > 0 where -lambda = max[real(Li)],
the Li given by solving det|A -LI| = 0, A is the community matrix.

The stability of the stochastic models requires that lambda > CV.
Otherwise, if CV > lambda, the fluctuations bring the model into
regions of instability, and the population will eventually become
extinct. If lambda < CV, we say that environmental stochasticity
is strong, and if lambda > CV, we say that it is weak. The
requirement

2) lambda > CV

for stability of a population model in the face of environmental
fluctuation is that we have only weak environmental
stochasticity. When environmental stochasticity is strong

3) lambda <= CV

the population will eventually become extinct.

Stochastic models are produced from deterministic ones by way of
the diffusion equation:

4) Df(n,t)/Dt = -D/Dn (M(n)f(n,t)) + (1/2)D^2/Dn^2 (V(n)f(n,t))

[here D is partial differentiation]

where M(n) = <F(n)> is the expected value of F, and
V(n) = <(F(n) - M)^2> is the variance. F is the dynamical
function of the deterministic model dN(t)/dt = F(N(t)).

The equilibrium probability distribution (if it exists) is found
by setting 4) to zero:

5) 0 = -d/dn(M(n)f*(n)) + (1/2)d^2/dn^2 (V(n)f*(n))

The generalization to m species is straightforward. For the
deterministic model dNi(t)/dt = Fi(N1,N2,...,Nm), define
Mi(n1,n2,...nm) = <Fi>, and Vij(n1,n2,...nm) = <(Fi - Mi)(Fj -
Mj)>, and the diffusion equation becomes

6) Df/Dt = - SUM(i=1,m) D/Dni (Mi f)
+ (1/2) SUM(i,j=1,m)D^2/DniDnj(Vij f).

Again, the equilibrium distribution, if it exists, is found by
setting 6) to 0.

The stability condition is

7) lambda > V(n)

For biologically realistic models, V(n) is often a decreasing
function of K, or lambda an increasing function of K. If so, there is
a value of K called MVP such that

8) lambda <= V(n) when K < MVP
lambda > V(n) when K > MVP

i.e. the population will become extinct if it is smaller than MVP
and will not if it is greater. MVP is the *minimum viable
population*. Determining an MVP is known as *population
vulnerability analysis.* The MVP is attained in a habitat with a
carrying capacity equal to the MVP. Other things being equal
(i.e. assuming constant habitat quality), carrying capacity is
proportional to the area of the habitat -- to increase K,
increase the area. Ensuring that a population does not fall
beneath MVP means choosing a preserve size large enough that its
carrying capacity is >= MVP. Such an analysis, for example, has
been done for the Northern Spotted Owl [in the Jack Ward Thomas
report]. Of course, *protecting* the required preserve size is a
political problem. When it cannot be done, eventually human
pressure on the habitat will cause the K to reduce below MVP, and
extinction is almost inevitable.

-!!!!!!!!!-
Alan McGowen

-- Transfer complete, hit <RETURN> to continue --

-!-
! Origin: ONE WORLD Usenet<->Fidonet (1:102/129.1)

- If you scratch a cynic,
- you'll find a defeated idealist.