Building Calibrated Isoscapes

[Scott Clem]

Hello again Alex,

I have another quick question. I apologize if there is something I am missing here, but is it possible to build an isoscape of the mean and variance of the calibrated precipitation data? On the workflow and in the manual, I have only been able to find how to do this with raw precipitation. I ask this because I would like to report an isoscape of this sort in my publication.

Thank you,

Scott

[Alexandre Courtiol]

Hi Scott,

I did not code this in IsoriX because this does not reflect how IsoriX works.

While alternative assignment workflows do calibrate the precipitation isoscape and then assign the sample in such landscape, IsoriX does not.

IsoriX instead internally transforms each assignment sample into precipitation equivalent and builds different map for each sample to assign.

Those maps directly reflect the difference between the isotopic composition of precipitation at the (unknown) origin of the sample and that of each cell.

We do this because of how our null hypothesis testing scheme works: each cell is in turn considered as a potential location of origin to which the isotopic signature at the unknown origin is compared to.

Yet, it is easy to produce such maps based on the calibration object (irrespective of the method for calibration being used) and the isoscape.

So, I could indeed add a function in IsoriX creating such maps.

In the meantime, here is n example of how you could produce such a map by "hand" using the latest CRAN  version of IsoriX:

(note: I am using examples from the package as a basis to generate the isoscape and calibration fit before getting to it)

 

```r

library(IsoriX)

## Building the precipitation isoscape
GNIPDataDEagg <- prepsources(data = GNIPDataDE)
GermanFit <- isofit(data = GNIPDataDEagg, mean_model_fix = list(elev = TRUE, lat_abs = TRUE))
GermanScape <- isoscape(raster = ElevRasterDE, isofit = GermanFit)

## Calibration using lab method (here, on made up data)
set.seed(123)
CalibDataAlien2 <- create_aliens(calib_fn = list(intercept = 3, slope = 0.5,
                                                 resid_var = 5),
                                 isoscape = GermanScape,
                                 raster = ElevRasterDE,
                                 n_sites = 25,
                                 min_n_samples = 5,
                                 max_n_samples = 5)

CalibDataAlien2 <- CalibDataAlien2[, c("site_ID", "sample_ID", "source_value",
                                       "sample_value")]
CalibAlien2 <- calibfit(data = CalibDataAlien2, method = "lab")

## plotting mean isoscape for precipitation
plot(GermanScape)

## plotting mean isoscape for predicted animal tissue
TissueScape <- CalibAlien2$param["intercept"] + CalibAlien2$param["slope"] * GermanScape$isoscapes$mean
plot(TissueScape)

## hack for plotting using IsoriX
TissueScape2 <- list(isoscapes = list(mean = TissueScape))
class(TissueScape2) <- c("ISOSCAPE", "list")
plot(TissueScape2)

## hack to change the title posthoc
plot_obj <- lattice::trellis.last.object() # recover last plotting object
plot_obj$main <- bquote(.("Animal")~.(bquote(delta**2*H))) # new title, call also be just character string
print(plot_obj) # replot with updated title

```

That should work, but let me know otherwise.

Best,

Alex

[Scott Clem]

Thank you, once again, for the help! This code seems to work fine.

Cheers,

Scott