isoscape: Predicts the spatial distribution of source isotopic values
In IsoriX: Isoscape Computation and Inference of Spatial Origins using Mixed Models

isoscape

R Documentation

Predicts the spatial distribution of source isotopic values

Description

This function produces the set of isoscapes, i.e. the spatial prediction (i.e. maps) of the distribution of source isotopic values, as well as several variances around such predictions. The predictions are computed using the fitted geostatistical models for each raster cell of a structural raster. All shape files can be exported and loaded into any Geographic Information System (GIS) if needed (see online tutorials).

Usage

isoscape(raster, isofit, verbose = interactive())

Arguments

`raster`	The structural raster (SpatRaster) such as an elevation raster created using prepelev
`isofit`	The fitted isoscape created by isofit
`verbose`	A logical indicating whether information about the progress of the procedure should be displayed or not while the function is running. By default verbose is `TRUE` if users use an interactive R session and `FALSE` otherwise.

Details

This function computes the predictions (mean), prediction variances (mean_predVar), residual variances (mean_residVar) and response variances (mean_respVar) for the isotopic values at a resolution equal to the one of the structural raster. It also computes the same information for the residual dispersion variance (disp_pred, disp_predVar, disp_residVar, or disp_respVar).

The predictions of isotopic values across the landscape are performed by calling the function spaMM::predict from the package spaMM on the fitted isoscape produced by isofit.

Let us summarize the meaning of mean, mean_predVar, mean_residVar and mean_respVar (see Courtiol & Rousset 2017 and Courtiol et al. 2019 for more details):

Our model assumes that that there is a single true unknown isoscape, which is fixed but which is represented by the mixed-effect model as a random draw from possible realizations of isoscapes (random draws of the Matérn-correlated process and of the uncorrelated random effects if considered). We infer this realized isoscape by fitting the model to a limited amount of data, with some uncertainty since different random draws of the unknown isoscape may give the same observed data. There is thus a conditional distribution of possible true isoscapes given the data. For linear mixed-effects models, the mean prediction is the mean of this conditional distribution. The prediction variance is ideally the mean square difference between the true unknown value of the linear predictor and the mean prediction at a given location. The residual variance is simply the prediction of the variance in isotopic value at a given location. Its exact meaning depends on the aggregation scheme used in prepsources, but by default, it would correspond to the temporal variation between months and across years. The response variance estimates the variance of new observations drawn from the true unknown isoscape at a given location. The response variance is simply equal to the sum of the prediction variance and the residual variance (note that the residual variance considered assume that a single observation is being observed per location).

The isoscape can be plotted using the function plot.ISOSCAPE (see examples).

Value

This function returns a list of class ISOSCAPE containing a set of all 8 raster layers mentioned above (all being of class SpatRaster), and the location of the sources as spatial points.

References

Courtiol, A., Rousset, F. (2017). Modelling isoscapes using mixed models. https://www.biorxiv.org/content/10.1101/207662v1

Courtiol A, Rousset F, Rohwäder M, Soto DX, Lehnert L, Voigt CC, Hobson KA, Wassenaar LI, Kramer-Schadt S (2019). Isoscape computation and inference of spatial origins with mixed models using the R package IsoriX. In Hobson KA, Wassenaar LI (eds.), Tracking Animal Migration with Stable Isotopes, second edition. Academic Press, London.

Examples


## The examples below will only be run if sufficient time is allowed
## You can change that by typing e.g. options_IsoriX(example_maxtime = XX)
## if you want to allow for examples taking up to ca. XX seconds to run
## (so don't write XX but put a number instead!)

if (getOption_IsoriX("example_maxtime") > 30) {
  ## We prepare the data
  GNIPDataDEagg <- prepsources(data = GNIPDataDE)

  ## We fit the models
  GermanFit <- isofit(
    data = GNIPDataDEagg,
    mean_model_fix = list(elev = TRUE, lat_abs = TRUE)
  )

  ## We build the isoscapes
  GermanScape <- isoscape(raster = ElevRasterDE, isofit = GermanFit)

  GermanScape
  plot(GermanScape)

  ## We build more plots
  PlotMean <- plot(x = GermanScape, which = "mean", plot = FALSE)

  PlotMeanPredVar <- plot(x = GermanScape, which = "mean_predVar", plot = FALSE)

  PlotMeanResidVar <- plot(x = GermanScape, which = "mean_residVar", plot = FALSE)

  PlotMeanRespVar <- plot(x = GermanScape, which = "mean_respVar", plot = FALSE)

  ## We display the plots
  print(PlotMean, split = c(1, 1, 2, 2), more = TRUE)
  print(PlotMeanPredVar, split = c(2, 1, 2, 2), more = TRUE)
  print(PlotMeanResidVar, split = c(1, 2, 2, 2), more = TRUE)
  print(PlotMeanRespVar, split = c(2, 2, 2, 2), more = FALSE)

  ## We build a sphere with our isoscape
  plot(x = GermanScape, which = "mean", plot = FALSE, sphere = list(build = TRUE))

  ## We can save a rotating sphere with the isoscape as a .gif-file.
  ## This file will be located inside your working directory.
  ## Make sure your current rgl device (from the previous step) is still open
  ## and that you have both the packages 'rgl' and 'magick' installed.
  ## The building of the .gif implies to create temporarily many .png
  ## but those will be removed automatically once the .gif is done.
  ## Uncomment to proceed (after making sure you have rgl, magick & webshot2 installed)
  # if(require("rgl") && require("magick") && require("webshot2")) {
  #   movie3d(spin3d(axis = c(0, 0, 1), rpm = 2), duration = 30, dir = getwd())
  # }
}

IsoriX documentation built on Nov. 14, 2023, 5:09 p.m.