 The paper introduces MIXSIR, an open source or package for Bayesian tracer, stable isotope, mixing model analysis that offers flexibility and classivity and guidance for implementation. It discusses practical differences between mixture data error structure formulations and their relation to common mixing model study designs in ecology. The paper also outlines methods for establishing prior distributions, source data inputs, fixed and random effects as covariates, and calculating relative support for multiple models via information criteria. A case study of alligator Mississippian's diet partitioning demonstrates the power of this approach, while limitations to mixing model applications are discussed. This article was authored by Brian C. Stock, Andrew L. Jackson, Eric J. Ward, and others.