In cmerow/occOutliers: Identify Outlying Species Occurrence Records

knitr::opts_chunk$set(echo = TRUE)
knitr::opts_chunk$set(cache = F)
knitr::opts_chunk$set(eval = T,fig.width=4, fig.height=4, dpi=200,out.width='80%')
knitr::opts_knit$set(root.dir = rprojroot::find_rstudio_root_file())
mc.cores=1

library(occOutliers)
library(sp)

We start by finding n obvious spatial outlier.

myPres=read.csv(system.file('extdata/SpeciesCSVs/Astraea_macroura.csv',
                           package='occOutliers'))
myPres=myPres[complete.cases(myPres),]
sp::coordinates(myPres)=c(1,2)
myEnv=raster::stack(system.file('extdata/AllEnv.tif',package='occOutliers'))
names(myEnv)=read.table(system.file('extdata/layerNames.csv',package='occOutliers'))[,1]
myPresDF=sp::SpatialPointsDataFrame(myPres,data.frame(raster::extract(myEnv,myPres)))

presOut=findOutlyingPoints(pres=myPresDF,
                           spOutliers=TRUE,
                           pval=1e-5)
world.shp=readRDS(system.file('extdata/worldShpMollwide.rds',package='occOutliers'))
plotOutliers(presOut,shpToPlot = world.shp)

Using the IQR test

presOut=findOutlyingPoints(pres=myPresDF,
                           spOutliers=TRUE,
                           envOutliers = FALSE,
                           method='iqr')
plotOutliers(presOut,shpToPlot = world.shp,outlierNames = 'spOutlier')

Using dixon

presOut=findOutlyingPoints(pres=myPresDF,
                           spOutliers=TRUE,
                           envOutliers = FALSE,
                           method='dixon')
plotOutliers(presOut,shpToPlot = world.shp,outlierNames = 'spOutlier')

Using Rosner

presOut=findOutlyingPoints(pres=myPresDF,
                           spOutliers=TRUE,
                           envOutliers = FALSE,
                           method='rosner',
                           kRosner=3)
plotOutliers(presOut,shpToPlot = world.shp,outlierNames = 'spOutlier')

Next, we find outliers in evironmental space.

myPres2=read.csv(system.file('extdata/SpeciesCSVs/Anarthria_scabra.csv',package='occOutliers'))
myPres2=myPres2[complete.cases(myPres2),]
sp::coordinates(myPres2)=c(1,2)
myPres2DF=sp::SpatialPointsDataFrame(myPres2,data.frame(raster::extract(myEnv,myPres2)))
presOut2=findOutlyingPoints(pres=myPres2DF,
                           spOutliers=TRUE,
                           envOutliers=TRUE,
                           pval=1e-5)
plotOutliers(presOut2,shpToPlot = world.shp,legLoc='topleft')

Here's an example finding both spatial and environmental outliers at the same time.

#should test for normality shapiro.test(dat_hist value)
myPres3=read.csv(system.file('extdata/SpeciesCSVs/Camissonia_tanacetifolia.csv',
                             package='occOutliers'))
myPres3=myPres3[complete.cases(myPres3),]
sp::coordinates(myPres3)=c(1,2)
myPres3DF=sp::SpatialPointsDataFrame(myPres3,data.frame(raster::extract(myEnv,myPres3)))
presOut3=findOutlyingPoints(pres=myPres3DF,
                           spOutliers=TRUE,
                           envOutliers=TRUE,
                           pval=1e-3)
plotOutliers(presOut3,shpToPlot = world.shp)