MixSIAR is a comprehensive software package for mixing models, integrating advancements from previous software such as IsoSource, MixSIR, SIAR, and IsotopeR. It constructs a JAGS model file using the `write_JAGS_model` function, which is crucial due to the interdependence of model components. The software supports hierarchical fitting of source data, allowing for covariance between tracers if raw data are available, or assuming independence if only summary statistics are provided. Users can also choose to turn off source fitting by setting the sample size to a large number, which is useful for poorly resolved mixing systems. MixSIAR fits source data in three preference levels: 1. **Raw source data**: Covariance between tracers is allowed. 2. **Summary statistics**: Tracers are assumed to be independent. 3. **No source fitting**: Source means are fixed at their sample means. The software handles various data types, including fatty acid profiles, and normalizes the data before fitting. It supports both fixed and random effects, with practical distinctions based on the number of parameters and independence of factor levels. Continuous effects are added as linear terms in ILR-space. MixSIAR calculates the mixture mean and variance in three ways: 1. **Process x Residual error (with covariance)**: Covariance between sources and time-dependent dilution factors (TDF) is considered. 2. **Residual error**: Suitable for situations where true variation in source values is not reflected in the mixture data. 3. **Process error**: Used when there is only one mixture datapoint or per fixed/random effect level. The likelihood for the data is calculated based on the mixture mean and covariances, ensuring a robust and flexible approach to mixing model analysis.MixSIAR is a comprehensive software package for mixing models, integrating advancements from previous software such as IsoSource, MixSIR, SIAR, and IsotopeR. It constructs a JAGS model file using the `write_JAGS_model` function, which is crucial due to the interdependence of model components. The software supports hierarchical fitting of source data, allowing for covariance between tracers if raw data are available, or assuming independence if only summary statistics are provided. Users can also choose to turn off source fitting by setting the sample size to a large number, which is useful for poorly resolved mixing systems. MixSIAR fits source data in three preference levels: 1. **Raw source data**: Covariance between tracers is allowed. 2. **Summary statistics**: Tracers are assumed to be independent. 3. **No source fitting**: Source means are fixed at their sample means. The software handles various data types, including fatty acid profiles, and normalizes the data before fitting. It supports both fixed and random effects, with practical distinctions based on the number of parameters and independence of factor levels. Continuous effects are added as linear terms in ILR-space. MixSIAR calculates the mixture mean and variance in three ways: 1. **Process x Residual error (with covariance)**: Covariance between sources and time-dependent dilution factors (TDF) is considered. 2. **Residual error**: Suitable for situations where true variation in source values is not reflected in the mixture data. 3. **Process error**: Used when there is only one mixture datapoint or per fixed/random effect level. The likelihood for the data is calculated based on the mixture mean and covariances, ensuring a robust and flexible approach to mixing model analysis.