Source code to fit the ZSM

This page provides brief documentation for the source code used to simulate and estimate parameters for the modified ZSM (zero sum multinomial) described in the book the The Unified Neutral Theory of Biogeography by Stephen Hubbell 2001

The methods are described in more detail in a paper in Nature and in supplementary materials. This paper should be cited if you publish results using the source code on this page.

Five main files and four ancillary files are provided:

hubbell.c: is a C program file which generates one random sample from the ZSM. It is based on pp. 290-293 and figure 9.1 in Hubbell's book
hubbell.dll: is a binary executable compilation of hubbell.c for Win32 platforms. For other platforms users must use MATLAB's MEX utility to build this file
esthub.m: is a Matlab program file which repeatedly calls myhubeng.m to compare the goodness of fit for different permutations of parameters
myhubeng.m: is a Matlab program file which repeatedly calls hubbell.c to generate an average ZSM for one given set of parameters
hubplot.m: is a Matlab program which will plot an estimated and observed SAD for visual comparison
Miscellaneous: Four MATLAB utility routines which I have written and which are used by esthub.m: sadhist.m, myhistc.m, isvector.m and getglobal

Calling syntax for these routines

randint=hubbell(randseed) - seeds the random # generator in hubbell. Use this to give repeatable results

ab=hubbell(theta,Jm) or hubbell(theta,Jm,1.0) - generate a random metacommunity, returns a vector of observed abundances

ab=hubbell(theta,J,m) - generates a local community (uses Jm=1,000,000 to first generate metacommunity)

ab=hubbell(theta,J,m,ts) - generates a local community, using ts timesteps to approach equilibrium in the local community (ts defaults to 10,000)

ab=hubbell(theta,J,m,tsary) - as above but returns the community for each timestep in tsary

[thetaest,mest]=esthub(abundvect,nruns,bDraw) - estimates theta and m for the given abundances. nreps is optional and describes how many communities to generate for each set of potential parameters. bDraw is an optional boolean argument that indicates whether to draw the fitted curve. A third output argument can be added to give the likelihood surfaces created. Abundances should be integers greater than zero.

[mn,var]=myhubeng(theta,Jm,m,nreps,bDraw,ts) - calls the hubbell generator routine nreps times and returns the mean and variance in abunce for species of rank 1...N. The last four arguments are optional

hubplot(estabund,variance,obsab,hPlot,lognormalparams) - plots a rank-abundance SAD plot overlaying estimated and observed curves. All but the first parameter is optional

Downloading

Anyone may download and use this code. If you publish results using this code you must cite the above mentioned paper. You must also fill out a brief form giving your name, insitution and email address before downloading he code.

I will not provide this information to any other person or institution.

Your name: Your institution: Your email:

Go to Brian McGill's homepage