Supplement 1. Maximum-likelihood estimation procedure for diet composition based on the Dirichlet function: a simple example.

File List FitDietDirichlet.m - Matlab source code for maximum likelihood estimation Read me.pdf - Instructions for using Matlab code and description of routines sampledata.mat - Example stomach data samplenames.mat - Example prey names for labeling plots Dirichlet Example for distribution.zip - All files at once Description The Matlab code in FitDietDirichlet.m loads example diet composition data contained in sampledata.mat and prey names for labeling plots contained in samplenames.mat. It then normalizes and bootstraps the diet composition data forming distributions. The distributions are fit to a Dirichlet function using the dirichlet_fit() routine contained in the fastfit toolbox (this also requires the lightspeed toolbox; see Read me.pdf for download instructions). Parameters for the marginal beta distributions are returned, as well as variances, means, maximum likelihoods and confidence intervals. Two plots are created, one showing the marginal beta distributions and one showing the source data and bootstrapped data.