Ex 3: Lake

In MixSIAR: Bayesian Mixing Models in R

Here we step through the Lake Example using the script version of MixSIAR. For a demonstration using the GUI version, see the MixSIAR Manual. For a thorough walkthrough of how to use MixSIAR in a script, see the Wolves Example, which provides more commentary and explanation.

For a clean, runnable .R script, look at mixsiar_script_lake.R in the example_scripts folder of the MixSIAR package install:

library(MixSIAR)
mixsiar.dir <- find.package("MixSIAR")
paste0(mixsiar.dir,"/example_scripts")

You can run the lake example script directly with:

source(paste0(mixsiar.dir,"/example_scripts/mixsiar_script_lake.R"))

Lake Example

The Lake Example data is simulated based on Francis et al. 2011 and we include it as a example of how MixSIAR can include a continuous effect. Here we examine the diet of zooplankton in 21 lakes using:

• 2 biotracers ($\delta^{13}$C, $\delta^{15}$N)
• 1 continuous effect (Secchi Depth : Mixed Layer Depth)
• Raw source data

MixSIAR fits a continuous covariate as a linear regression in ILR/transform-space. Two terms are fit for the proportion of each source: an intercept and a slope. The plot uses the posterior median estimates of the intercept and slope, and the lines are curved because of the ILR-transform back into p-space. For details, or if you would like to make modifications, see ?plot_continous_var.R.

Fitting a model with a continuous effect is more complex than the categorical fixed/random effects and can be a bit finicky.

Load MixSIAR package

library(MixSIAR)

Load mixture data

See ?load_mix_data for details.

The lake consumer data has 1 covariate, which we fit as a continuous effect (cont_effects="Secchi.Mixed"). There are no fixed/random effects (factors=NULL, fac_random=NULL, fac_nested=NULL).

# Replace the system.file call with the path to your file
mix.filename <- system.file("extdata", "lake_consumer.csv", package = "MixSIAR")

mix <- load_mix_data(filename=mix.filename,
                     iso_names=c("d13C","d15N"),
                     factors=NULL,
                     fac_random=NULL,
                     fac_nested=NULL,
                     cont_effects="Secchi.Mixed")

Load source data

See ?load_source_data for details.

We have no fixed/random effects in the model (source_factors=NULL), and we do not have concentration dependence data (conc_dep=FALSE). We have the original "raw" source data, not means and SDs (data_type="raw").

# Replace the system.file call with the path to your file
source.filename <- system.file("extdata", "lake_sources.csv", package = "MixSIAR")

source <- load_source_data(filename=source.filename,
                           source_factors=NULL,
                           conc_dep=FALSE,
                           data_type="raw",
                           mix)

Load discrimination data

See ?load_discr_data for details.

# Replace the system.file call with the path to your file
discr.filename <- system.file("extdata", "lake_discrimination.csv", package = "MixSIAR")

discr <- load_discr_data(filename=discr.filename, mix)

Plot data

This is your chance to check:

• Are the data loaded correctly?
• Is your mixture data in the source polygon?
• Are one or more of your sources confounded/hidden?

# Make an isospace plot
plot_data(filename="isospace_plot", plot_save_pdf=TRUE, plot_save_png=FALSE, mix,source,discr)

Calculate convex hull area

Calculate normalized surface area of the convex hull polygon(s) as in Brett (2014).

Note 1: discrimination SD is added to the source SD (see ?calc_area for details)

# Calculate the convex hull area, standardized by source variance
calc_area(source=source,mix=mix,discr=discr)

Plot prior

Define your prior, and then plot using "plot_prior"

• RED = your prior
• DARK GREY = "uninformative"/generalist (alpha = 1)
• LIGHT GREY = "uninformative" Jeffrey's prior (alpha = 1/n.sources)

# default "UNINFORMATIVE" / GENERALIST prior (alpha = 1)
plot_prior(alpha.prior=1,source)

Write JAGS model file

In the Lake Example we demo the "Residual only" error option. The differences between "Residual * Process", "Residual only", and "Process only" are explained in Stock and Semmens (2016).

# Write the JAGS model file
model_filename <- "MixSIAR_model.txt"
resid_err <- TRUE
process_err <- FALSE
write_JAGS_model(model_filename, resid_err, process_err, mix, source)

Run model

Choose one of the MCMC run options:

| run == | Chain Length | Burn-in | Thin | # Chains | | ------------- | ------------- | ------------- | ------------- | ------------- | | "test" | 1,000 | 500 | 1 | 3 | | "very short" | 10,000 | 5,000 | 5 | 3 | | "short" | 50,000 | 25,000 | 25 | 3 | | "normal" | 100,000 | 50,000 | 50 | 3 | | "long" | 300,000 | 200,000 | 100 | 3 | | "very long" | 1,000,000 | 500,000 | 500 | 3 | | "extreme" | 3,000,000 | 1,500,000 | 500 | 3 |

First use run = "test" to check if 1) the data are loaded correctly and 2) the model is specified correctly:

jags.1 <- run_model(run="test", mix, source, discr, model_filename)

After a test run works, increase the MCMC run to a value that may converge:

jags.1 <- run_model(run="normal", mix, source, discr, model_filename)

Analyze diagnostics and output

MixSIAR fits a continuous covariate as a linear regression in ILR/transform-space. Two terms are fit for the proportion of each source: an intercept and a slope. The plot uses the posterior median estimates of the intercept and slope, and the lines are curved because of the ILR-transform back into p-space. For details, or if you would like to make modifications, see ?plot_continous_var. See ?output_JAGS for output options.

The other posterior plots MixSIAR produces for a continuous effect show the estimated diet for the minimum, median, and maximum individuals.

output_JAGS(jags.1, mix, source, output_options)

MixSIAR documentation built on Oct. 23, 2020, 6:45 p.m.