ProbDensFunc: Probability Density Function

Description

Using a variety of parameters in modelling will inevitably bring variability in predictions, especially when it comes to making future predictions. This function enables an overall viewing of the future predictions range per species and gives the likelihood of range shift estimations. It will calculate the optimal way for condensing a difined proportion (50, 75, 90 and 95% per default) of the data.

Usage

ProbDensFunc(initial,
             projections,
             groups = NULL,
             plothist = TRUE,
             cvsn = TRUE,
             resolution = 5,
             filename = NULL, ...)

Arguments

initial

a vector (resp. a SpatialPointsDataFrame) in a binary format (ones and zeros) representing the current distribution of a species which will be used as a reference for the range change calculations

projections

a matrix (resp; a rasterStack) grouping all the predictions where each column is a single prediction. Make sure you keep projections in the same order as the initial vector (line1=site1, line2=site2, etc.).

plothist

set to TRUE to plot the range change histogram

cvsn

stands for "current vs new". If true, the range change calculations will be of two types: the percentage of cells currently occupied by the species to be lost, and the relative percentage of cells currently unoccupied but projected to be, namely 'new' cel

groups

an option for ungrouping the projections enabling a separated visualisation of the prediction range per given group. A matrix is expected where each column is a single prediction and each line is giving details of one parameter (See the examples section).

resolution

the step used for classes of prediction in graphics. The default value is 5

filename

the name of file (with extension) where plots will be stored. If not NULL, no ploting windows will be open

...

futher arguments:

lim: ordered numeric vector indicating the proportion of data to consider for histogramme representation (by default : c(0.5,0.75,0.9,0.95) )

nb.points.max: the maximum number o

Value

This is a plotting function, no objects are returned or created.

Details

The future range changes are calculated as a percentage of the species' present state. For example, if a species currently occupies 100 cells and is estimated by a model to cover 120 cells in the future, the range change will be + 20%.

Resolution : Note that modifying the resolution will directly influence the probability scale. Bigger classes will cumulate a greater number of predictions and therefore represent a greater fraction of the total predictions. The probability is in fact that of the class and not of isolated events.

Examples

Run this code

DataSpecies <- read.csv(system.file("external/species/mammals_table.csv",
                                      package="biomod2"), row.names = 1)
  head(DataSpecies)
  
  # the name of studied species
  myRespName <- 'GuloGulo'
  
  # the presence/absences data for our species 
  myResp <- as.numeric(DataSpecies[,myRespName])
  
  # remove all 0 from response vector to work with
  # presence only data (Pseudo Absences selections)
  rm_id <- which(myResp==0)
  myResp <- myResp[-rm_id]
  
  
  # the XY coordinates of species data
  myRespXY <- DataSpecies[-rm_id,c("X_WGS84","Y_WGS84")]
  
  
  # Environmental variables extracted from BIOCLIM 
  myExpl = stack( system.file( "external/bioclim/current/bio3.grd", 
                       package="biomod2"),
                  system.file( "external/bioclim/current/bio4.grd", 
                               package="biomod2"), 
                  system.file( "external/bioclim/current/bio7.grd", 
                               package="biomod2"),  
                  system.file( "external/bioclim/current/bio11.grd", 
                               package="biomod2"), 
                  system.file( "external/bioclim/current/bio12.grd", 
                             package="biomod2"))
    
  # 1. Formatting Data
  myBiomodData <- BIOMOD_FormatingData(resp.var = myResp,
                                       expl.var = myExpl,
                                       resp.xy = myRespXY,
                                       resp.name = myRespName,
                                       PA.nb.rep=3)
  
  # 2. Defining Models Options using default options.
  myBiomodOption <- BIOMOD_ModelingOptions()
  
  # 3. Doing Modelisation
  myBiomodModelOut <- BIOMOD_Modeling( myBiomodData, 
                       models = c('CTA','RF','GLM','GAM','ANN','MARS'), 
                       models.options = myBiomodOption, 
                       NbRunEval=5, 
                       DataSplit=70, 
                       Prevalence=0.5,
                       models.eval.meth = c('TSS'),
                       do.full.models = FALSE,
                       rescal.all.models=T,
                       modeling.id='test')
  
  # 4. Build ensemble-models that will be taken as reference
  myBiomodEM <- BIOMOD_EnsembleModeling( modeling.output = myBiomodModelOut,
                                         chosen.models = 'all',
                                         em.by = 'all',
                                         eval.metric = c('TSS'),
                                         eval.metric.quality.threshold = c(0.7),
                                         prob.mean = TRUE,
                                         prob.median = TRUE)
  
  # 5. Projection on current environemental conditions
  myBiomodProjection <- BIOMOD_Projection(modeling.output = myBiomodModelOut,
                                          new.env = myExpl,
                                          proj.name = 'current',
                                          selected.models = 'all',
                                          binary.meth = 'TSS',
                                          compress = FALSE,
                                          build.clamping.mask = TRUE)
  
  BIOMOD_EnsembleForecasting(projection.output=myBiomodProjection,
                             EM.output=myBiomodEM,
                             binary.meth='TSS')
  
  # 6. load binary projections
  consensusBin <- stack('GuloGulo/proj_current/
  proj_current_GuloGulo_TotalConsensus_EMbyTSS_TSSbin.grd')
  projectionsBin <- stack('GuloGulo/proj_current/
  proj_current_GuloGulo_TSSbin.grd')
  
  # 7. build a ref state based on ensemble-models
  ref <- sampleRandom(subset(consensusBin, 1, drop=T), size=5000, sp=T, na.rm=T)
  
  # 8. autoatic creation of groups matrix
  find_groups <- function(diff_by_pix){
    data.set <- sapply(names(diff_by_pix),.extractModelNamesInfo,info='data.set')
    run.eval <- sapply(names(diff_by_pix),.extractModelNamesInfo,info='run.eval')
    models <- sapply(names(diff_by_pix),.extractModelNamesInfo,info='models')
    return(rbind(data.set,run.eval,models))  
  }
  
  groups <- find_groups(projectionsBin)
  
  # 9. plot ProbDensFunct graphs
  ProbDensFunc(initial = ref,
               projections = projectionsBin, 
               plothist=TRUE, 
               cvsn=TRUE, 
               groups=groups, 
               resolution=2, 
               filename=NULL, 
               lim=c(0.5,0.8,0.95))