[Marxan] Marxan and simple problems
ayse.turak at dkm.org.tr
Tue Mar 25 19:38:44 EST 2008
Questions asked by both Daniel Brockington and Daniel Beaver, seem to have
gone unanswered in the list.
Daniel(s), did you, by any chance receive any answer to your private
adresses? If not, lets hope this mail will induce more experienced users
will think about the problems?
Ayşe S. Turak, Ph.D.
Conservation GIS Expert
Nature Conservation Centre
320. Sok. No:4, Ortadoğu Sitesi,
Yüzüncüyıl, Ankara, Turkey
tel:+90 312 287 8144 fax:+90 312 287 4067
e-mail: ayse.turak at dkm.org.tr
From: marxan-bounces at sib.uq.edu.au [mailto:marxan-bounces at sib.uq.edu.au] On
Behalf Of Daniel Brockington
Sent: Monday, February 18, 2008 7:00 PM
To: marxan at sib.uq.edu.au
Subject: Re: [Marxan] Marxan and simple problems
Am resending this email. I saw a reply to it in my Inbox which then
disappeared. This is an recurrent problem. Would the person who replied to
it be so kind as to resend their email.
I am a new user of Marxan and am familiarising myself with the software. I
will shortly be conducting a Marxan analysis to explore how existing
conservation NGO revenues might be used to support as many protected areas
in sub-Saharan Africa as possible.
To familiarise myself with the software I have set Marxan some simple
1. I gave it a list of 6 reserves representing species A-E. Reserves 1
contains species A alone, reserve 2 species B alone etc. There are equal
abundances of each species in each reserve, and the target (10) is the same
as the abundance of the species (10) in each reserve. Reserve 5 and no 6
conserving species E in the same abundance (10) but reserve 6 was more
expensive. SPF was set to 1 for all species
The optimal solution is to conserve reserves 1-5 and leave out 6.
After 100 repeat runs Marxan never choses 6. But, using an adaptive
annealing schedule, it was content to conserve between 1-4 reserves many
times. It did not carry on looking until it found all 5 and a zero penalty
factor. Why was it content to do this? How could the adaptive annealing
schedule be made more responsive to the task being set? I thought that it
mighe be due to adaptive annealing samples only once (number of repeat
runs/100= 1), but 10,000 repeat runs still produces the same pattern.
Setting the spf to 100 resulted in it always chosing 4 or 5 reserves, 50% of
the time each. Interestingly the reserve that was always only chosen 50% of
the time was reserve 5, regardless of whether it was the cheapest, or the
most expensive (not including reserve 6). This suggests that Marxan was
working through the reserves sequentially.
Setting the spf to 1000 resulted in Marxan always chosing all 5 reserves.
2. I then tried to incorporate the influence of boundaries. I had 5 species,
and 8 reserves. As before reserve 1 contained species A in its target
abundance (10). Reserve 2 contained species B in its target abundance etc.
Reserves 5,6,7 and 8 each contained half the target abundance of species E
(5 each). Reserve 5 and 6 share a common boundary. 7 and 8 are isolates.
Reserves 5-8 all cost the same amount.
The optimal solution is to chose reserves 1-6 and leave 7 and 8.
On 100 repeat runs Marxan never selected reserves 7 and 8. But, even with an
spf of 1000, and the boundary modifer at 1 Marxan was content not to select
reserve 6 50% of the time. It always selected reserve 5. Why was it content
to have a reserve which only met half the target? Again its incomplete use
of reserve 6 suggests it is going through the list sequentially.
I then made reserves 5 and 6 the isolates and gave 7 and 8 a common
boundary. The optimal solution should be reserves 1-4 and 7-8.
But this time, on 100 iterations, Marxan never selected reserves 6, 7 or 8.
It was always content just to chose reserve 5 and not to hit the target
number of species E.
I then increased the Boundary Modifer to 100. This resulted in Marxan always
chosing reserve 5 but adding reserves 7 and 8 half of the time. It never
chose the optimal solution. 50% of the time it chose a more epensive
solution which included a larger reserve size for species E than was
Why did it not drop reserve 5 in its iterations? Why is it content to run
through the sequence of reserves in the order given? Should we be changing
the order in which reserves are listed in the input files in order to
improve the flexibility of the programme?
I suspect that there is probably a silly careless mistake at work here.
Please tell me what it is!
Institute for Development Policy and Management,
School of Environment and Development,
Arthur Lewis Building,
The University of Manchester,
daniel.brockington at manchester.ac.uk
<mailto:Pdaniel.brockington at manchester.ac.uk>
Phone: +44 161 275 3233
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Marxan