I am currently working with both SIAR and MixSIR to analyze food web data using three food sources and dual stable isotopes (C/N). Does anyone know about the restrictions of each of the two mixing models when the isotope data of the food sources are quite close to each other? I would like to hear some opinions about the differences in the models' function. And generally, would the restrictions of the number of food sources and the range of data be the same for both mixing models?
Following reference might give you an idea,
"Jackson AL, Inger R, Bearhop S, Parnell A (2009) Erroneous behaviour of MixSIR, a recently published Bayesian isotope mixing model: a discussion of Moore & Semmens (2008). Ecology Letters 12:E1-E5"
These two modelling approaches have now been combined by there authors into a new model called MixSiar: check out the facebook page (https://www.facebook.com/pages/SIAR-Stable-Isotope-Analysis-in-R/148501811896914) and the project page (http://conserver.iugo-cafe.org/user/brice.semmens/MixSIAR)
Thank you both very much for your reply!
Asanka I was aware of this publication but it seems there is a big discussion on which of the two models is better and what are the errors in each, Actually this occurs at the period when Siar was first out and most discussions are between the groups developing these programs, so I guess there was still field for improvements.
Thank you Claire for your suggestion, I was not aware of this combination, I will check out its perfofmance with my data.
SIAR has on additional residual error term per isotope axis used in the model. If you have consumer (mixture) data that is highly variable compared to your source (prey) data, you should use this added residual error term (SIAR). If your consumer data is relatively invariant compared to your sources, you should not. MixSIAR allows you to choose. Note, however, that we are in the process of publishing a new error structure (and will incorporate into MixSIAR) that can be used in both instances, and is thus to be used in nearly all cases.
Regarding the "erroneous" behavior of MixSIR ... it was actually the simulation code used in the critique, not the model, that was erroneous. See http://www.ncbi.nlm.nih.gov/pubmed/19245585.
- Badges
- Science method
- Similar topics
- Biological Science
- Ecology