Abstract: While dual isotope plots provide a qualitative view of 
producer and consumer interactions, the challenge of multiple source 
mixing models is to quantitatively examine those relationships.  Here I 
present a multisource mixing model that uses linear programming 
techniques.  This approach finds an algebraic solution for the unknown 
diet fractions of different food sources using matrices of known producer 
and consumer isotope ratios, and finds unique solutions for successive 
combinations of food sources.  Each solution that falls within a 
specified fractional range is a vertex of the solution space.  The full 
set of vertices forms a cloud of points, and the ``center of mass'' of 
those points is an estimate of the direct or indirect diet and trophic 
level of the consumer.  The size of the combination of sources tested 
is limited by the number of known isotope ratios.  Additional indicators 
could also be included in this technique to improve its resolution, 
as long as those indicators, like the stable isotope ratios currently 
in use, are linearly independent.  We will use isotopic data from Willapa 
Bay as an example.