Recently I posted a question to Ecolog about an interpretation of Levin's simpe 
model.  Here is the orignal question and the responses I got.

Alan Griffith

__________________________

I've been struggling to come up with a simple and intuitive explanation for the 
equilibrium result of Levin's simple metapopulation model.  This is the model 
with internal colonization only.  No rescue is possible.  That result is

p = 1 - e/c or the equilibrium proportion of patches occupied is determined 
solely by the ratio of extinction chance and colonization chance of individual 
patches. 

The particular situation I wish to describe is when e = c and the 
metapopulation goes extinct.  This is always puzzling to students because they 
reason that if extinction probability just balances colonization probability, 
the metapopulation could just hang on as patches are just replaced as they go 
extinct.

Does anyone have an easy explanation for this clash between the math and the 
mind?

Alan Griffith

Responses

1. Would it help to remind you students that the model is a differential 
equation, which means that everything happens instantaneously and at the same 
time? They cannot balance each other because at the instant that dispersal 
occurs, the population goes extinct, so you have colonization from a zero 
population.  It sounds like they are looking at it from the destination patch's 
point of view, and forgetting that the source is also changing - I had a 
similar difficulty with the concept when I first encountered it.
 
This would also be a good chance to discuss the difference between models and 
reality.  They'll never encounter a real world example where e and c match 
exactly.
 
Hope this helps,
 
Nate
 
-- 
Nathan Lichti 
Graduate Research Assistant 
Forestry and Natural Resources 
Purdue University 
715 W State Street 
West Lafayette IN 47907-2061 

2. This is a bit convoluted to explain, but basically is a consequence  
of the fact that the actual colonization rate of vacant patches  
within the landscape is a function of a constant colonization rate  
and the number of occupied patches.  So colonization is dependent on  
occupancy.  Extinction of occupied patches is independent though.  So  
increased extinction (relative to c) reduces the number of occupied  
patch (all else equal) which reduces the 'production' of propagules  
that can colonize vacant patches.  When e = c there simply aren't  
enough (any, actually) occupied patches left that can generate  
propagules fast enough to keep up.
Maybe it helps to reassess the actual model in different terms:   
Levins model just describes the change in patch occupancy over time,  
which can be written as:
Occupied-patches * patch-colonization rate - Occupied-patches * patch  
extinction, or dp/dt = p * m - p * e
The equilibrium is where patch colonization equals patch extinction:
m = e
This is the equilibrium you and your students are thinking of.  But  
the colonization rate depends on the number of occupied patches:
m = c * (1 - p)
So at the equilibrium a certain proportion (p) of patches is occupied:
e = c * (1 - p)
When e = c, that equilibrium proportion of occupied patches is 0.   
Extinction catches up with 'production'.

I'll put a model on the web if you're interested.  It is a simulation  
that might make things clearer.  Or not...
http://life.uiuc.edu/~sstoddard/Levins.html 

Steven T. Stoddard
Program in Ecology and Evolutionary Biology
University of Illinois at Urbana-Champaign

www-u.life.uiuc.edu/~sstoddar
[EMAIL PROTECTED] 
(217) 333 - 2235

3. My students have struggled with this as well (as have I). They 
invariably assume that at equilibrium e=c. I'm not a modeler, but I 
think one way to explain it is to remind the students that Levins' 
model is really just another way of expressing the logistic growth 
model dN/dt = rN(1-N/K), where ln(c/e) is analogous to r and N/K is 
analogous to P. We don't assume that r is equal to 0 at equilibrium. 
In fact, r > 0, but dN/dt approaches 0 as N approaches K. r 
represents sort of a biotic potential for growth in the absence of 
density dependence. Likewise, I think you could argue that c/e 
represents a biotic potential for population expansion throughout a 
region.

Another point that might be relevant is that in order for the 
equilibrium metapopulation to persist when perturbed (i.e., when P is 
reduced below the equilibrium value), it must be able to increase 
when rare and return to that equilibrium. If e/c = 1, there is no way 
for that to happen.

Steve

Steve Brewer
Associate Professor
Department of Biology
PO Box 1848
University, MS 38677-1848
telephone: (662) 915-1077
FAX: (662) 915-5144

4. 

    /   \     
 //      \\    Alan B. Griffith, Ph.D.       
 \\ |  | //    Department of Biological Sciences
\ \`|()|'/ /   University of Mary Washington
\_`(  )'_/    1301 College Avenue
   /(  )\      Fredericksburg, VA  22401-5300
 /  ^^  \     540-654-1422
   |      |      [EMAIL PROTECTED]
   |      |     

Reply via email to