Ecospat Manual

User Manual:

Open the PDF directly: View PDF PDF.
Page Count: 98

Package ‘ecospat’
June 14, 2018
Version 3.0
Date 2018-06-013
Title Spatial Ecology Miscellaneous Methods
Author Olivier Broennimann [cre,aut,ctb],
Valeria Di Cola [aut,ctb],
Blaise Petitpierre [ctb],
Frank Breiner [ctb],
Manuela D Amen [ctb],
Christophe Randin [ctb],
Robin Engler [ctb],
Wim Hordijk [ctb],
Julien Pottier [ctb],
Mirko Di Febbraro [ctb],
Loic Pellissier [ctb],
Dorothea Pio [ctb],
Ruben Garcia Mateo [ctb],
Anne Dubuis [ctb],
Daniel Scherrer [ctb],
Luigi Maiorano [ctb],
Achilleas Psomas [ctb],
Charlotte Ndiribe [ctb],
Nicolas Salamin [ctb],
Niklaus Zimmermann [ctb],
Antoine Guisan [aut]
Maintainer Olivier Broennimann <olivier.broennimann@unil.ch>
VignetteBuilder knitr
Depends ade4 (¿= 1.6-2), ape (¿= 3.2), gbm (¿= 2.1.1), sp (¿= 1.0-15)
Imports adehabitatHR (¿= 0.4.11), adehabitatMA (¿= 0.3.8), biomod2 (¿=
3.1-64), dismo (¿= 0.9-3), ecodist (¿= 1.2.9), maptools (¿=
0.8-39), randomForest (¿= 4.6-7), spatstat (¿= 1.37-0), raster
(¿= 2.5-8), rms (¿= 4.5-0), MigClim (¿= 1.6), gtools (¿=
3.4.1), PresenceAbsence (¿= 1.1.9), methods (¿= 3.1.1),
doParallel (¿= 1.0.10), foreach (¿= 1.4.3), iterators (¿=
1.0.8), parallel, classInt (¿= 0.1-23), vegan (¿= 2.4-1),
poibin (¿= 1.3), snowfall (¿= 1.61), snow
Suggests rgdal (¿= 1.2-15), rJava (¿= 0.9-6), XML (¿= 3.98-1.1), knitr
(¿= 1.14)
LazyData true
1
2Rtopics documented:
URL http://www.unil.ch/ecospat/home/menuguid/ecospat-resources/tools.html
Description Collection of R functions and data sets for the support of spatial ecology analy-
ses with a focus on pre-, core and post- modelling analyses of species distribu-
tion, niche quantification and community assembly. Written by current and former mem-
bers and collaborators of the ecospat group of Antoine Guisan, Department of Ecol-
ogy and Evolution (DEE) & Institute of Earth Surface Dynamics (IDYST), Univer-
sity of Lausanne, Switzerland.
License GPL
BugReports https://github.com/ecospat/ecospat
NeedsCompilation no
Rtopics documented:
ecospat-package..................................... 3
ecospat.adj.D2.glm................................... 5
ecospat.binary.model.................................. 6
ecospat.boyce...................................... 7
ecospat.calculate.pd .................................. 8
ecospat.caleval ..................................... 10
ecospat.CCV.communityEvaluation.bin . . . . . . . . . . . . . . . . . . . . . . . 11
ecospat.CCV.communityEvaluation.prob . . . . . . . . . . . . . . . . . . . . . . . 13
ecospat.CCV.createDataSplitTable . . . . . . . . . . . . . . . . . . . . . . . . . . 15
ecospat.CCV.modeling................................. 17
ecospat.climan ..................................... 19
ecospat.cohen.kappa .................................. 20
ecospat.CommunityEval ................................ 21
ecospat.consCscore .................................. 23
ecospat.cor.plot..................................... 24
ecospat.cooccurrences................................. 25
ecospat.Cscore ..................................... 26
ecospat.cv.example................................... 27
ecospat.cv.gbm ..................................... 28
ecospat.cv.glm ..................................... 29
ecospat.cv.me...................................... 30
ecospat.cv.rf....................................... 31
ecospat.env ....................................... 32
ecospat.Epred...................................... 32
ecospat.ESM.EnsembleModeling . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
ecospat.ESM.EnsembleProjection . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
ecospat.ESM.Modeling................................. 38
ecospat.ESM.Projection ................................ 41
ecospat.grid.clim.dyn.................................. 43
ecospat.makeDataFrame................................ 45
ecospat.mantel.correlogram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
ecospat.max.kappa ................................... 48
ecospat.max.tss..................................... 49
ecospat.maxentvarimport ............................... 50
ecospat.mdr....................................... 51
ecospat.mess ...................................... 53
ecospat.meva.table ................................... 54
ecospat-package 3
ecospat.migclim..................................... 55
ecospat.mpa....................................... 56
ecospat.niche.dyn.index ................................ 57
ecospat.niche.dynIndexProjGeo . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
ecospat.niche.equivalency.test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
ecospat.niche.overlap.................................. 61
ecospat.niche.similarity.test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
ecospat.niche.zProjGeo................................. 63
ecospat.npred...................................... 65
ecospat.occ.desaggregation............................... 65
ecospat.occupied.patch................................. 66
ecospat.permut.glm................................... 68
ecospat.plot.contrib................................... 69
ecospat.plot.kappa ................................... 70
ecospat.plot.mess.................................... 71
ecospat.plot.niche.................................... 72
ecospat.plot.niche.dyn ................................. 73
ecospat.plot.overlap.test ................................ 73
ecospat.plot.tss ..................................... 74
ecospat.rand.pseudoabsences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
ecospat.rangesize .................................... 76
ecospat.rcls.grd..................................... 79
ecospat.recstratprop.................................. 80
ecospat.recstratregl .................................. 81
ecospat.sample.envar.................................. 82
ecospat.SESAM.prr................................... 83
ecospat.shift.centroids ................................. 85
ecospat.testData .................................... 86
ecospat.testEnvRaster ................................. 89
ecospat.testMdr..................................... 90
ecospat.testNiche.................................... 91
ecospat.testNiche.inv.................................. 92
ecospat.testNiche.nat.................................. 93
ecospat.testTree..................................... 94
ecospat.varpart ..................................... 95
Index 97
ecospat-package Spatial Ecology Miscellaneous Methods
Description
Collection of methods, utilities and data sets for the support of spatial ecology analyses
with a focus on pre-, core and post- modelling analyses of species distribution, niche quan-
tification and community assembly. Specifically,
-Pre-modelling:
Spatial autocorrelation –¿ ecospat.mantel.correlogram;
Variable selection –¿ ecospat.npred;
Climate Analalogy –¿ ecospat.climan, ecospat.mess and ecospat.plot.mess;
4ecospat-package
Phylogenetic diversity measures –¿ ecospat.calculate.pd;
Biotic Interactions –¿ ecospat.co-occurrences and ecospat.Cscore;
Minimum Dispersal routes –¿ ecospat.mdr;
Niche Quantification –¿ ecospat.grid.clim.dyn, ecospat.niche.equivalency.test,
ecospat.niche.similarity.test, ecospat.plot.niche, ecospat.plot.niche.dyn,
ecospat.plot.contrib, ecospat.niche.overlap, ecospat.plot.overlap.test,
ecospat.niche.dyn.index and ecospat.shift.centroids;
Data Preparation –¿ ecospat.caleval, ecospat.cor.plot, ecospat.makeDataFrame,
ecospat.occ.desaggregation, ecospat.rand.pseudoabsences, ecospat.rcls.grd,
ecospat.recstrat prop, ecospat.recstrat regl and ecospat.sample.envar;
-Core Niche Modelling:
Model evaluation –¿ ecospat.cv.glm, ecospat.permut.glm, ecospat.cv.gbm,
ecospat.cv.me, ecospat.cv.rf, ecospat.boyce, ecospat.CommunityEval,
ecospat.cohen.kappa, ecospat.max.kappa, ecospat.max.tss, ecospat.meva.table,
ecospat.plot.kappa, ecospat.plot.tss and ecospat.adj.D2.glm;
Spatial predictions and projections –¿ ecospat.ESM.Modeling,
ecospat.ESM.EnsembleModeling, ecospat.ESM.Projection, ecospat.ESM.EnsembleProjection,
ecospat.SESAM.prr, ecospat.migclim, ecospat.binary.model, ecospat.Epred
and ecospat.mpa;
Variable Importance –¿ ecospat.maxentvarimport;
-Post Modelling:
Variance Partition –¿ ecospat.varpart;
Spatial predictions of species assemblages –¿ ecospat.cons Cscore;
Range size quantification –¿ ecospat.rangesize and
ecospat.occupied.patch;
The ecospat package was written by current and former members and collaborators of the
ecospat group of Antoine Guisan, Department of Ecology and Evolution (DEE) & Institute
of Earth Surface Dynamics (IDYST), University of Lausanne, Switzerland.
Details
Package: ecospat
Type: Package
Version: 2.2.0
Date: 2017-11-22
License: GPL
Author(s)
Olivier Broennimann [aut], Valeria Di Cola [cre, aut], Blaise Petitpierre [ctb], Frank Breiner
[ctb], Manuela D‘Amen [ctb], Christophe Randin [ctb], Robin Engler [ctb], Wim Hordijk
ecospat.adj.D2.glm 5
[ctb], Julien Pottier [ctb], Mirko Di Febbraro [ctb], Loic Pellissier [ctb], Dorothea Pio [ctb],
Ruben Garcia Mateo [ctb], Anne Dubuis [ctb], Daniel Scherrer [ctb], Luigi Maiorano [ctb],
Achilleas Psomas [ctb], Charlotte Ndiribe [ctb] Nicolas Salamin [ctb], Niklaus Zimmermann
[ctb], Antoine Guisan [aut]
ecospat.adj.D2.glm Calculate An Adjusted D2
Description
This function is used for calculating an adjusted D2 from a calibrated GLM object
Usage
ecospat.adj.D2.glm(glm.obj)
Arguments
glm.obj Any calibrated GLM object with a binomial error distribution
Details
This function takes a calibrated GLM object with a binomial error distribution and returns
an evaluation of the model fit. The measure of the fit of the models is expressed as the
percentage of explained deviance adjusted by the number of degrees of freedom used (similar
to the adjusted-R2 in the case of Least-Square regression; see Weisberg 1980) and is called
the adjusted-D2 (see guisan and Zimmermann 2000 for details on its calculation).
Value
Returns an adjusted D square value (proportion of deviance accounted for by the model).
Author(s)
Christophe Randin <christophe.randin@unibas.ch>and Antoine Guisan <antoine.guisan@unil.ch>
References
Weisberg, S. 1980. Applied linear regression. Wiley.
Guisan, A., S.B. Weiss and A.D. Weiss. 1999. GLM versus CCA spatial modeling of plant
species distribution. Plant Ecology,143, 107-122.
Guisan, A. and N.E. Zimmermann. 2000. Predictive habitat distribution models in ecology.
Ecol. Model.,135, 147-186.
Examples
glm.obj<-glm(Achillea_millefolium˜ddeg+mind+srad+slp+topo,
family = binomial, data=ecospat.testData)
ecospat.adj.D2.glm(glm.obj)
6ecospat.binary.model
ecospat.binary.model Generate Binary Models
Description
Generate a binary map from a continuous model prediction.
Usage
ecospat.binary.model (Pred, Threshold)
Arguments
Pred RasterLayer predicted suitabilities from a SDM prediction.
Threshold A threshold to convert continous maps into binary maps (e.g. the out-
put of the function ecospat.mpa() or use the optimal.thresholds from
PresenceAbsence R package.
Details
This function generates a binary model prediction (presence/absence) from an original
model applying a threshold. The threshold could be arbitrary, or be based on the maximum
acceptable error of false negatives (i.e. percentage of the presence predicted as absences,
omission error).
Value
The binary model prediction (presence/absence).
Author(s)
Ruben G. Mateo <rubeng.mateo@gmail.com>with contributions of Frank Breiner <frank.breiner@wsl.ch>
References
Fielding, A.H. and J.F. Bell. 1997. A review of methods for the assessment of prediction
errors in conservation presence/absence models. Environmental Conservation,24: 38-49.
Engler, R., A Guisan and L. Rechsteiner. 2004. An improved approach for predicting
the distribution of rare and endangered species from occurrence and pseudo-absence data.
Journal of Applied Ecology,41, 263-274.
Liu, C., Berry, P. M., Dawson, T. P. and R. G. Pearson. 2005. Selecting thresholds of
occurrence in the prediction of species distributions. Ecography,28, 385-393.
Jimenez-Valverde, A. and J.M.Lobo. 2007. Threshold criteria for conversion of probability
of species presence to either-or presence-absence. Acta oecologica,31, 361-369.
Liu, C., White, M. and G. Newell. 2013. Selecting thresholds for the prediction of species
occurrence with presence-only data. J. Biogeogr.,40, 778-789.
Freeman, E.A. and G.G. Moisen. 2008. A comparison of the performance of threshold
criteria for binary classification in terms of predicted prevalence and kappa. Ecological
Modelling,217, 48-58.
ecospat.boyce 7
See Also
ecospat.mpa,optimal.thresholds
Examples
library(dismo)
# get predictor variables
fnames <- list.files(path=paste(system.file(package="dismo"), '/ex', sep=''),
pattern='grd', full.names=TRUE )
predictors <- stack(fnames)
# file with presence points
occurence <- paste(system.file(package="dismo"), '/ex/bradypus.csv', sep='')
occ <- read.table(occurence, header=TRUE, sep=',')[,-1]
colnames(occ) <- c("x","y")
# fit a domain model, biome is a categorical variable
do <- domain(predictors, occ, factors='biome')
# predict to entire dataset
pred <- predict(do, predictors)
plot(pred)
points(occ)
# use MPA to convert suitability to binary map (90% of occurrences encompass by binary map)
mpa.cutoff <- ecospat.mpa(pred,occ)
pred.bin.mpa <- ecospat.binary.model(pred,mpa.cutoff)
plot(pred.bin.mpa)
points(occ)
ecospat.boyce Calculate Boyce Index
Description
Calculate the Boyce index as in Hirzel et al. (2006). The Boyce index is used to assess
model performance.
Usage
ecospat.boyce (fit, obs, nclass=0, window.w="default", res=100, PEplot = TRUE)
Arguments
fit A vector or Raster-Layer containing the predicted suitability values
obs A vector containing the predicted suitability values or xy-coordinates (if
”fit” is a Raster-Layer) of the validation points (presence records)
8ecospat.calculate.pd
nclass The number of classes or vector with class thresholds. If nclass=0, the
Boyce index is calculated with a moving window (see next parameters)
window.w The width of the moving window (by default 1/10 of the suitability range)
res The resolution of the moving window (by default 100 focals)
PEplot If true, plot the predicted to expected ratio along the suitability class
Details
The Boyce index only requires presences and measures how much model predictions differ
from random distribution of the observed presences across the prediction gradients (Boyce
et al. 2002). It is thus the most appropriate metric in the case of presence-only models. It
is continuous and varies between -1 and +1. Positive values indicate a model which present
predictions are consistent with the distribution of presences in the evaluation dataset, values
close to zero mean that the model is not different from a random model, negative values
indicate counter predictions, i.e., predicting poor quality areas where presences are more
frequent (Hirzel et al. 2006).
Value
Returns the predicted-to-expected ratio for each class-interval: F.ratio
Returns the Boyce index value: Spearman.cor
Creates a graphical plot of the predicted to expected ratio along the suitability class
Author(s)
Blaise Petitpierre <bpetitpierre@gmail.com>and Frank Breiner <frank.breiner@unil.ch>
References
Boyce, M.S., P.R. Vernier, S.E. Nielsen and F.K.A. Schmiegelow. 2002. Evaluating resource
selection functions. Ecol. Model.,157, 281-300.
Hirzel, A.H., G. Le Lay, V. Helfer, C. Randin and A. Guisan. 2006. Evaluating the ability
of habitat suitability models to predict species presences. Ecol. Model.,199, 142-152.
Examples
obs <- (ecospat.testData$glm_Saxifraga_oppositifolia
[which(ecospat.testData$Saxifraga_oppositifolia==1)])
ecospat.boyce (fit = ecospat.testData$glm_Saxifraga_oppositifolia , obs, nclass=0,
window.w="default", res=100, PEplot = TRUE)
ecospat.calculate.pd Calculate Phylogenetic Diversity Measures
Description
Calculate all phylogenetic diversity measures listed in Schweiger et al., 2008 (see full refer-
ence below).
ecospat.calculate.pd 9
Usage
ecospat.calculate.pd (tree, data, method="spanning", type="clade", root=FALSE,
average=FALSE, verbose=TRUE)
Arguments
tree The phylogenetic tree
data A presence or absence (binary) matrix for each species (columns) in each
location or grid cell (rows)
method The method to use. Options are ”pairwise”, ”topology”, and ”spanning”.
Default is ”spanning”.
type Phylogenetic measure from those listed in Schweiger et al 2008. Options
are ”Q”, ”P”, ”W”, ”clade”, ”species”, ”J”, ”F”, ”AvTD”,”TTD”, ”Dd”.
Default is ”clade”.
root Phylogenetic diversity can either be rooted or unrooted. Details in Schweiger
et al 2008. Default is FALSE.
average Phylogenetic diversity can either be averaged or not averaged. Details in
Schweiger et al 2008. Default is FALSE.
verbose Boolean indicating whether to print progress output during calculation.
Default is TRUE.
Details
Given a phylogenetic tree and a presence/absence matrix this script calculates phylogenetic
diversity of a group of species across a given set of grid cells or locations. The library ”ape”
is required to read the tree in R. Command is ”read.tree” or ”read.nexus”. Options of
type: ”P” is a normalized mearure of ”Q”. ”clade” is ”PDnode” when root= FALSE, and
is ”PDroot” ehn root =TRUE. ”species” is ”AvPD”.
Value
This function returns a list of phylogenetic diversity values for each of the grid cells in the
presence/absence matrix
Author(s)
Nicolas Salamin <nicolas.salamin@unil.ch>and Dorothea Pio <Dorothea.Pio@fauna-flora.org>
References
Schweiger, O., S. Klotz, W. Durka and I. Kuhn. 2008. A comparative test of phylogenetic
diversity indices. Oecologia,157, 485-495.
Pio, D.V., O. Broennimann, T.G. Barraclough, G. Reeves, A.G. Rebelo, W. Thuiller, A.
Guisan and N. Salamin. 2011. Spatial predictions of phylogenetic diversity in conservation
decision making. Conservation Biology,25, 1229-1239.
Pio, D.V., R. Engler, H.P. Linder, A. Monadjem, F.P.D. Cotterill, P.J. Taylor, M.C. Schoe-
man, B.W. Price, M.H. Villet, G. Eick, N. Salamin and A. Guisan. 2014. Climate change
effects on animal and plant phylogenetic diversity in southern Africa. Global Change Biol-
ogy,20, 1538-1549.
10 ecospat.caleval
Examples
fpath <- system.file("extdata", "ecospat.testTree.tre", package="ecospat")
tree <-read.tree(fpath)
data <- ecospat.testData[9:52]
pd <- ecospat.calculate.pd(tree, data, method = "spanning", type = "species", root = FALSE,
average = FALSE, verbose = TRUE )
plot(pd)
ecospat.caleval Calibration And Evaluation Dataset
Description
Generate an evaluation and calibration dataset with a desired ratio of disaggregation.
Usage
ecospat.caleval (data, xy, row.num=1:nrow(data), nrep=1, ratio=0.7,
disaggregate=0, pseudoabs=0, npres=0, replace=FALSE)
Arguments
data A vector with presence-absence (0-1) data for one species.
xy The x and y coordinates of the projection dataset.
row.num Row original number
nrep Number of repetitions
ratio Ratio of disaggregation
disaggregate Minimum distance of disaggregation (has to be in the same scale as xy)
pseudoabs Number of pseudoabsences
npres To select a smaller number of presences from the dataset to be subsetted.
The maximum number is the total number of presences
replace F to replace de pseudoabsences
Details
This functions generates two list, one with the calibration or training dataset and other list
with the evaluation or testing dataset disaggregated with a minimum distance.
Value
list(”eval”=eval,”cal”=cal))
Author(s)
Blaise Petitpierre <bpetitpierre@gmail.com>
ecospat.CCV.communityEvaluation.bin 11
Examples
data <- ecospat.testData
caleval <- ecospat.caleval (data = ecospat.testData[53], xy = data[2:3], row.num = 1:nrow(data),
nrep = 2, ratio = 0.7, disaggregate = 0.2, pseudoabs = 100, npres = 10, replace = FALSE)
caleval
ecospat.CCV.communityEvaluation.bin
Calculates a range of community evaluation metrics based on
different thresholding techniques.
Description
The function uses the output of ecospat.CCV.modeling to calculate a range of community
evaluation metrics based on a selection of thresholding techniques both for the calibration
data and independent evaluation data.
Usage
ecospat.CCV.communityEvaluation.bin(ccv.modeling.data,
thresholds= c('MAX.KAPPA','MAX.ROC','PS_SDM'),
community.metrics=c('SR.deviation','Sorensen'),
parallel=TRUE,
cpus=4,
fix.threshold=0.5,
MCE=5,
MEM=NULL)
Arguments
ccv.modeling.data
a'ccv.modeling.data'object returned by ecospat.CCV.modeling
thresholds a selection of thresholds ('FIXED','MAX.KAPPA','MAX.ACCURACY','MAX.TSS','SENS SPEC','MAX.ROC','OBS.PREVALENCE','AVG.PROBABILITY','MCE','PS SDM, MEM')
to be calculated and applied for the model evaluation.
community.metrics
a selection of community evaluation metrics ('SR.deviation','community.AUC','community.overprediction','community.underprediction','community.accuracy','community.sensitivity','community.specificity','community.kappa','community.tss','Sorensen','Jaccard','Simpson')
to be calculated for each seleted thresholding technique.
parallel should parallel computing be allowed (TRUE/FALSE)
cpus number of cpus to use in parallel computing
fix.threshold fixed threshold to be used. Only gets used if thresholding technique FIXED
is selected.
MCE maximum omission error (%) allowed for the thresholding. Only gets used
if thresholding technique MCE is selected.
MEM a vetor with the species richness prediction of a MEM for each site. Only
needed if MEM is selected.
12 ecospat.CCV.communityEvaluation.bin
Details
The function uses the probability output of the ecospat.CCV.modeling function and creates
binary maps based on the selected thresholding methods. These binary maps are then
used to calculate the selected community evaluation metrics both for the calibration and
evaluation data of each modeling run.
Value
DataSplitTable a matrix with TRUE/FALSE for each model run (TRUE=Calibration point,
FALSE=Evaluation point)
CommunityEvaluationMetrics.CalibrationSites
a 4-dimensional array containing the community evaluation metrics for
the calibartion sites of each run (NA means that the site was used for
evaluation)
CommunityEvaluationMetrics.EvaluationSites
a 4-dimensional array containing the community evaluation metrics for
the evaluation sites of each run (NA means that the site was used for
calibaration)
PA.allSites a 4-dimensional array of the binary prediction for all sites and runs under
the different thresholding appraoches.
Author(s)
Daniel Scherrer ¡daniel.j.a.scherrer@gmail.com¿
References
Scherrer, D., D’Amen, M., Mateo, M.R.G., Fernandes, R.F. & Guisan , A. (2018) How to
best threshold and validate stacked species assemblages? Community optimisation might
hold the answer. Methods in Ecology and Evolution, in review
See Also
ecospat.CCV.createDataSplitTable;ecospat.CCV.communityEvaluation.prob
Examples
#Loading species occurence data and remove empty communities
testData <- ecospat.testData[,c(24,34,43,45,48,53,55:58,60:63,65:66,68:71)]
sp.data <- testData[which(rowSums(testData)>0), sort(colnames(testData))]
#Loading environmental data
env.data <- ecospat.testData[which(rowSums(testData)>0),4:8]
#Coordinates for all sites
xy <- ecospat.testData[which(rowSums(testData)>0),2:3]
#Running all the models for all species
myCCV.Models <- ecospat.CCV.modeling(sp.data = sp.data,
env.data = env.data,
xy = xy,
NbRunEval = 5,
minNbPredictors = 10,
VarImport = 3)
ecospat.CCV.communityEvaluation.prob 13
#Thresholding all the predictions and calculating the community evaluation metrics
myCCV.communityEvaluation.bin <- ecospat.CCV.communityEvaluation.bin(
ccv.modeling.data = myCCV.Models,
thresholds = c('MAX.KAPPA','MAX.ROC','PS_SDM'),
community.metrics= c('SR.deviation','Sorensen'),
parallel = FALSE,
cpus = 4)
ecospat.CCV.communityEvaluation.prob
Evaluates community predictions directly on the probabilities
(i.e., threshold independent)
Description
This function generates a number of community evaluation metrics directly based on the
probability returned by the individual models. Instead of thresholding the predictions
(ecospat.CCV.communityEvaluation.bin this function directly uses the probability and
compares its outcome to null models or average expectations.)
Usage
ecospat.CCV.communityEvaluation.prob(
ccv.modeling.data,
community.metrics=c('SR.deviation','community.AUC','probabilistic.Sorensen'),
se.th=0.01,
parallel = TRUE,
cpus = 4)
Arguments
ccv.modeling.data
a'ccv.modeling.data'object returned by ecospat.CCV.modeling
community.metrics
a selection of community metrics to calculate ('SR.deviation','community.AUC','probabilistic.Sorensen','probabilistic.Jaccard','probabilistic.Simpson')
se.th the desired precission for the community metrics (standard error of the
mean)
parallel should parallel computing be allowed (TRUE/FALSE)
cpus number of cpus to use in parallel computing
Value
DataSplitTable a matrix with TRUE/FALSE for each model run (TRUE=Calibration point,
FALSE=Evaluation point)
CommunityEvaluationMetrics.CalibrationSites
a 3-dimensional array containing the community evaluation metrics for
the calibartion sites of each run (NA means that the site was used for
evaluation)
CommunityEvaluationMetrics.EvaluationSites
a 3-dimensional array containing the community evaluation metrics for
the evaluation sites of each run (NA means that the site was used for
calibaration)
14 ecospat.CCV.communityEvaluation.prob
Note
If the community evaluation metric 'SR.deviation'is selected the returned tables will have
the following columns:
SR.obs = observed species richness,
SR.mean = the predicted species richness (based on the probabilities assuming poission
binomial distribution),
SR.dev = the deviation of observed and predicted species richness,
SR.sd = the standard deviation of the predicted species richness (based on the prob-
abilities assuming poission binomial distribution),
SR.prob = the probability that the observed species richness falls within the predicted
species richness (based on the probabilities assuming poission binomial distribution),
SR.imp.05 = improvement of species richness prediction over null-model 0.5,
SR.imp.average.SR = improvement of species richness prediction over null-model av-
erage.SR and
SR.imp.prevalence = improvement of species richness prediction over null-model
prevalence.
If the community evalation metric community.AUC is selected the returned tables will have
the following colums:
Community.AUC = The AUC of ROC of a given site (in this case the ROC plot is
community sensitiviy [percentage species predicted corretly present] vs 1 - community
specificity [percentage of species predicted correctly absent])
If any of the other community evaluation metrics ('probabilistic.Sorensen','probabilistic.Jaccard','probabilistic.Simpson')
is selected the returned tables will have the follwing colums:
METRIC.mean = The average Sorensen/Jaccard/Simpson based on a number of random
draws of the probabilities.
METRIC.sd = The standard deviation of Sorensen/Jaccard/Simpson based on a number
of random draws of the probabilities.
METRIC.CI = The 95% confidence intervall of the average Sorensen/Jaccard/Simpson
based on the standard deviation and number of draws. Should normally be ¡= se.th.
nb.it = number of draws used to estimate all the parameters. The draws stop as
soon as the desired precission (se.th) is reached or the limit of allowed iterations
(default=10’000).
composition.imp.05 = improvement of species compostion prediction over the null-
model 0.5.
composition.imp.average.SR = improvement of the species composition prediction over
the null-model average.SR.
composition.imp.prevalence = improvement of the species composition prediction over
the null-model prevalence.
For detailed descriptions of the null models see Scherrer et al. .....
Author(s)
Daniel Scherrer ¡daniel.j.a.scherrer@gmail.com¿
ecospat.CCV.createDataSplitTable 15
See Also
ecospat.CCV.createDataSplitTable;ecospat.CCV.communityEvaluation.bin;
Examples
#Loading species occurence data and remove empty communities
testData <- ecospat.testData[,c(24,34,43,45,48,53,55:58,60:63,65:66,68:71)]
sp.data <- testData[which(rowSums(testData)>0), sort(colnames(testData))]
#Loading environmental data
env.data <- ecospat.testData[which(rowSums(testData)>0),4:8]
#Coordinates for all sites
xy <- ecospat.testData[which(rowSums(testData)>0),2:3]
#Running all the models for all species
myCCV.Models <- ecospat.CCV.modeling(sp.data = sp.data,
env.data = env.data,
xy = xy,
NbRunEval = 5,
minNbPredictors = 10,
VarImport = 3)
#Calculating the probabilistic community metrics
myCCV.communityEvaluation.prob <- ecospat.CCV.communityEvaluation.prob(
ccv.modeling.data = myCCV.Models,
community.metrics = c('SR.deviation','community.AUC','probabilistic.Sorensen'),
se.th = 0.02,
parallel = FALSE,
cpus = 4)
ecospat.CCV.createDataSplitTable
Creates a DataSplitTable for usage in ecospat.ccv.modeling.
Description
Creates a DataSplitTable with calibration and evaluation data either for cross-validation
or repeated split sampling at the community level (i.e., across all species).
Usage
ecospat.CCV.createDataSplitTable(NbRunEval,
DataSplit,
validation.method,
NbSites,
sp.data=NULL,
minNbPresences=NULL,
minNbAbsences=NULL,
maxNbTry=1000)
16 ecospat.CCV.createDataSplitTable
Arguments
NbRunEval number of cross-validation or split sample runs
DataSplit proportion (%) of sites used for model calibration
validation.method
the type of DataSplitTable that should be created. Must be either
'cross-validation'or 'split-sample'
NbSites number of total sites available. Is ignored if sp.data is provided.
sp.data a data.frame where the rows are sites and the columns are species (values
1,0)
minNbPresences the desired minimum number of Presences required in each run
minNbAbsences the desired minimum number of Absences required in each run
maxNbTry number of random tries allowed to create a fitting DataSplitTable
Details
If a sp.data data.frame with species presences and absences is provided the function tries
to create a DataSplitTable which ensures that the maximum possible number of species
can be modelled (according to the specified minimum presences and absences.)
Value
DataSplitTable a matrix with TRUE/FALSE for each model run (TRUE=Calibration point,
FALSE=Evaluation point)
Author(s)
Daniel Scherrer ¡daniel.j.a.scherrer@gmail.com¿
See Also
ecospat.CCV.modeling
Examples
#Creating a DataSplitTable for 200 sites, 25 runs with an
#80/20 calibration/evaluation cross-validation
DataSplitTable <- ecospat.CCV.createDataSplitTable(NbSites = 200,
NbRunEval=25,
DataSplit=80,
validation.method='cross-validation')
#Loading species occurence data and remove empty communities
testData <- ecospat.testData[,c(24,34,43,45,48,53,55:58,60:63,65:66,68:71)]
sp.data <- testData[which(rowSums(testData)>0), sort(colnames(testData))]
#Creating a DataSplitTable based on species data directly
DataSplitTable <- ecospat.CCV.createDataSplitTable(NbRunEval = 20,
DataSplit = 70,
validation.method = "cross-validation",
NbSites = NULL,
sp.data = sp.data,
minNbPresence = 15,
ecospat.CCV.modeling 17
minNbAbsences = 15,
maxNbTry = 250)
ecospat.CCV.modeling Runs indivudual species distribuion models with SDMs or ESMs
Description
Creates probabilistic prediction for all species based on SDMs or ESMs and returns their
evaluation metrics and variable importances.
Usage
ecospat.CCV.modeling(sp.data,
env.data,
xy,
DataSplitTable=NULL,
DataSplit = 70,
NbRunEval = 25,
minNbPredictors =5,
validation.method = "cross-validation",
models.sdm = c("GLM","RF"),
models.esm = "CTA",
modeling.options.sdm = NULL,
modeling.options.esm = NULL,
ensemble.metric = "AUC",
ESM = "YES",
parallel = FALSE,
cpus = 4,
VarImport = 10,
modeling.id)
Arguments
sp.data a data.frame where the rows are sites and the columns are species (values
1,0)
env.data either a data.frame where rows are sites and colums are environmental
variables or a raster stack of the envrionmental variables
xy two column data.frame with X and Y coordinates of the sites (most be
same coordinate system as env.data)
DataSplitTable a table providing TRUE/FALSE to indicate what points are used for calibra-
tion and evaluation. As returned by ecospat.CCV.createDataSplitTable
DataSplit percentage of dataset observations retained for the model training (only
needed if no DataSplitTable provided)
NbRunEval number of cross-validatio/split sample runs (only needed if no DataSplitTable
provided)
minNbPredictors
minimum number of occurences [min(presences/Absences] per predicotors
needed to calibrate the models
18 ecospat.CCV.modeling
validation.method
either ”cross-validation” or ”split-sample” used to validate the communtiy
predictions (only needed if no DataSplitTable provided)
models.sdm modeling techniques used for the normal SDMs. Vector of models names
choosen among 'GLM','GBM','GAM','CTA','ANN','SRE','FDA','MARS','RF','MAXENT.Phillips'and 'MAXENT.Tsuruoka'
models.esm modeling techniques used for the ESMs. Vector of models names choosen
among 'GLM','GBM','GAM','CTA','ANN','SRE','FDA','MARS','RF','MAXENT.Phillips'and 'MAXENT.Tsuruoka'
modeling.options.sdm
modeling options for the normal SDMs. "BIOMOD.models.options"” ob-
ject returned by BIOMOD ModelingOptions
modeling.options.esm
modeling options for the ESMs. "BIOMOD.models.options" object re-
turned by BIOMOD ModelingOptions
ensemble.metric
evaluation score used to weight single models to build ensembles: 'AUC','Kappa'or 'TSS'
ESM either 'YES'(ESMs allowed), 'NO'(ESMs not allowed) or 'ALL'(ESMs
used in any case)
parallel should parallel computing be allowed (TRUE/FALSE)
cpus number of cpus to use in parallel computing
VarImport number of permutation runs to evaluate variable importance
modeling.id character, the ID (=name) of modeling procedure. A random number by
default
Details
The basic idea of the community cross-validation (CCV) is to use the same data (sites)
for the model calibration/evaluation of all species. This ensures that there is ”inde-
pendent” cross-validation/split-sample data available not only at the individual species
level but also at the community level. This is key to allow an unbiased estimation of
the ability to predict species assemblages (Scherrer et al. 2018). The output of the
ecospat.CCV.modeling function can then be used to evaluate the species assemblage predic-
tions with the ecospat.CCV.communityEvaluation.bin or ecospat.CCV.communityEvaluation.prob
functions.
Value
modelling.id character, the ID (=name) of modeling procedure
output.files vector with the names of the files written to the hard drive
speciesData.calibration
a 3-dimensional array of presence/absence data of all species for the cali-
bration plots used for each run
speciesData.evaluation
a 3-dimensional array of presence/absence data of all species for the eval-
uation plots used for each run
speciesData.full
a data.frame of presence/absence data of all species (same as sp.data
input)
DataSplitTable a matrix with TRUE/FALSE for each model run (TRUE=Calibration point,
FALSE=Evaluation point)
ecospat.climan 19
singleSpecies.ensembleEvaluationScore
a 3-dimensional array of single species evaluation metrics ('Max.KAPPA','Max.TSS','AUC of ROC')
singleSpecies.ensembleVariableImportance
a 3-dimensional array of single species variable importance for all predic-
tors
singleSpecies.calibrationSites.ensemblePredictions
a 3-dimensional array of the predictions for each species and run at the
calibration sites
singleSpecies.evaluationSites.ensemblePredictions
a 3-dimensional array of the predictions for each species and run at the
evaluation sites
Author(s)
Daniel Scherrer ¡daniel.j.a.scherrer@gmail.com¿
References
Scherrer, D., D’Amen, M., Mateo, M.R.G., Fernandes, R.F. & Guisan , A. (2018) How to
best threshold and validate stacked species assemblages? Community optimisation might
hold the answer. Methods in Ecology and Evolution, in review
See Also
ecospat.CCV.createDataSplitTable;ecospat.CCV.communityEvaluation.bin;ecospat.CCV.communityEvaluation.prob
Examples
#Loading species occurence data and remove empty communities
testData <- ecospat.testData[,c(24,34,43,45,48,53,55:58,60:63,65:66,68:71)]
sp.data <- testData[which(rowSums(testData)>0), sort(colnames(testData))]
#Loading environmental data
env.data <- ecospat.testData[which(rowSums(testData)>0),4:8]
#Coordinates for all sites
xy <- ecospat.testData[which(rowSums(testData)>0),2:3]
#Running all the models for all species
myCCV.Models <- ecospat.CCV.modeling(sp.data = sp.data,
env.data = env.data,
xy = xy,
NbRunEval = 5,
minNbPredictors = 10,
VarImport = 3)
ecospat.climan A climate analogy setection tool for the modeling of species dis-
tributions
Description
Assess climate analogy between a projection extent (p) and a reference extent (ref, used in
general as the background to calibrate SDMs)
20 ecospat.cohen.kappa
Usage
ecospat.climan (ref, p)
Arguments
ref A dataframe with the value of the variables (i.e columns) for each point
of the reference exent.
pA dataframe with the value of the variables (i.e columns) for each point
of the projection exent.
Value
Returns a vector. Values below 0 are novel conditions at the univariate level (similar to the
MESS), values between 0 and 1 are analog and values above 1 are novel covariate condtions.
For more information see Mesgeran et al. (2014)
Author(s)
Blaise Petitpierre <bpetitpierre@gmail.com>
References
Mesgaran, M.B., R.D. Cousens and B.L. Webber. 2014. Here be dragons: a tool for
quantifying novelty due to covariate range and correlation change when projecting species
distribution models. Diversity & Distributions,20, 1147-1159.
Examples
x <- ecospat.testData[c(4:8)]
p<- x[1:90,] #A projection dataset.
ref<- x[91:300,] #A reference dataset
ecospat.climan(ref,p)
ecospat.cohen.kappa Cohen’s Kappa
Description
Calculates Cohen’s kappa and variance estimates, within a 95 percent confidence interval.
Usage
ecospat.cohen.kappa(xtab)
Arguments
xtab A symmetric agreement table.
Details
The argument xtab is a contingency table. xtab ¡- table(Pred ¿= th, Sp.occ)
ecospat.CommunityEval 21
Value
A list with elements ’kap’, ’vark’, ’totn’ and ’ci’ is returned. ’kap’ is the cohen’s kappa,
’vark’ is the variance estimate within a 95 percent confidence interval, ’totn’ is the number
of plots and ’ci’ is the confidence interval.
Author(s)
Christophe Randin <christophe.randin@wsl.ch>with contributions of Niklaus. E. Zim-
mermann <niklaus.zimmermann@wsl.ch>and Valeria Di Cola <valeria.dicola@unil.ch>
References
Bishop, Y.M.M., S.E. Fienberg and P.W. Holland. 1975. Discrete multivariate analysis:
Theory and Practice. Cambridge, MA: MIT Press. pp. 395-397.
Pearce, J. and S. Ferrier. 2000. Evaluating the predictive performance of habitat models
developed using logistic regression. Ecol. Model.,133, 225-245.
See Also
ecospat.meva.table,ecospat.max.tss,ecospat.plot.tss,ecospat.plot.kappa,ecospat.max.kappa
Examples
Pred <- ecospat.testData$glm_Agrostis_capillaris
Sp.occ <- ecospat.testData$Agrostis_capillaris
th <- 0.39 # threshold
xtab <- table(Pred >= th, Sp.occ)
ecospat.cohen.kappa(xtab)
ecospat.CommunityEval Community Evaluation
Description
Calculate several indices of accuracy of community predictions.
Usage
ecospat.CommunityEval (eval, pred, proba, ntir)
Arguments
eval A matrix of observed presence-absence (ideally independent from the
dataset used to fit species distribution models) of the species with n rows
for the sites and s columns for the species.
pred A matrix of predictions for the s species in the n sites. Should have the
same dimension as eval.
proba Logical variable indicating whether the prediction matrix contains presences-
absences (FALSE) or probabilities (TRUE).
ntir Number of trials of presence-absence predictions if pred is a probability
matrix.
22 ecospat.CommunityEval
Details
This function calculates several indices of accuracy of community predictions based on
stacked predictions of species ditribution models. In case proba is set to FALSE the function
returns one value per index and per site. In case proba is set to TRUE the function generates
presences-absences based on the predicted probabilities and returns one value per index,
per site and per trial.
Value
A list of evaluation metrics calculated for each site (+ each trial if proba is set to TRUE):
deviance.rich.pred: the deviation of the predicted species richness to the observed
overprediction: the proportion of species predicted as present but not observed among the
species predicted as present
underprediction: the proportion of species predicted as absent but observed among the
species observed as present
prediction.success: the proportion of species correctly predicted as present or absent
sensitivity: the proportion of species correctly predicted as present among the species ob-
served as present
specificity : the proportion of species correctly predicted as absent among the species
observed as absent
kappa: the proportion of specific agreement
TSS: sensitivity+specificity-1
similarity: the similarity of community composition between the observation and the pre-
diction. The calculation is based on the Sorenses index.
Jaccard: this index is a widely used metric of community similarity.
Author(s)
Julien Pottier <julien.pottier@clermont.inra.fr>
with contribution of Daniel Scherrer <daniel.scherrer@unil.ch>, Anne Dubuis <anne.dubuis@gmail.com>
and Manuela D’Amen <manuela.damen@unil.ch>
References
Pottier, J., A. Dubuis, L. Pellissier, L. Maiorano, L. Rossier, C.F. Randin, P. Vittoz and
A. Guisan. 2013. The accuracy of plant assemblage prediction from species distribution
models varies along environmental gradients. Global Ecology and Biogeography,22, 52-63.
Examples
## Not run:
eval <- Data[c(53,62,58,70,61,66,65,71,69,43,63,56,68,57,55,60,54,67,59,64)]
pred <- Data[c(73:92)]
ecospat.CommunityEval (eval, pred, proba=TRUE, ntir=10)
## End(Not run)
ecospat.cons Cscore 23
ecospat.cons Cscore Constrained Co-Occurrence Analysis.
Description
Co-occurrence Analysis & Environmentally Constrained Null Models. The function tests
for non-random patterns of species co-occurrence in a presence-absence matrix. It calculates
the C-score index for the whole community and for each species pair. An environmental
constraint is applied during the generation of the null communities.
Usage
ecospat.cons_Cscore(presence,pred,nperm,outpath)
Arguments
presence A presence-absence dataframe for each species (columns) in each loca-
tion or grid cell (rows) Column names (species names) and row names
(sampling plots).
pred A dataframe object with SDM predictions. Column names (species names
SDM) and row names (sampling plots).
nperm The number of permutation in the null model.
outpath Path to specify where to save the results.
Details
An environmentally constrained approach to null models will provide a more robust eval-
uation of species associations by facilitating the distinction between mutually exclusive
processes that may shape species distributions and community assembly. The format re-
quired for input databases: a plots (rows) x species (columns) matrix. Input matrices
should have column names (species names) and row names (sampling plots). NOTE: a SES
that is greater than 2 or less than -2 is statistically significant with a tail probability of less
than 0.05 (Gotelli & McCabe 2002 - Ecology)
Value
Returns the C-score index for the observed community (ObsCscoreTot), the mean of C-
score for the simulated communities (SimCscoreTot), p.value (PValTot) and standardized
effect size (SES.Tot). It also saves a table in the specified path where the same metrics are
calculated for each species pair (only the table with species pairs with significant p.values
is saved in this version).
Author(s)
Anne Dubuis <anne.dubuis@gmail.com>and Manuela D‘Amen <manuela.damen@unil.ch>
References
Gotelli, N.J. and D.J. McCabe. 2002. Species co-occurrence: a meta-analysis of JM Dia-
mond‘s assembly rules model. Ecology,83, 2091-2096.
Peres-Neto, P.R., J.D. Olden and D.A. Jackson. 2001. Environmentally constrained null
models: site suitability as occupancy criterion. Oikos,93, 110-120.
24 ecospat.cor.plot
Examples
## Not run:
presence <- ecospat.testData[c(53,62,58,70,61,66,65,71,69,43,63,56,68,57,55,60,54,67,59,64)]
pred <- ecospat.testData[c(73:92)]
nperm <- 10000
outpath <- getwd()
ecospat.cons_Cscore(presence, pred, nperm, outpath)
## End(Not run)
ecospat.cor.plot Correlation Plot
Description
A scatter plot of matrices, with bivariate scatter plots below the diagonal, histograms on the
diagonal, and the Pearson correlation above the diagonal. Useful for descriptive statistics
of small data sets (better with less than 10 variables).
Usage
ecospat.cor.plot(data)
Arguments
data A dataframe object with environmental variables.
Details
Adapted from the pairs help page. Uses panel.cor, and panel.hist, all taken from the help
pages for pairs. It is a simplifies version of pairs.panels() function of the package psych.
Value
A scatter plot matrix is drawn in the graphic window. The lower off diagonal draws scatter
plots, the diagonal histograms, the upper off diagonal reports the Pearson correlation.
Author(s)
Adjusted by L. Mathys, 2006, modified by N.E. Zimmermann
Examples
data <- ecospat.testData[,4:8]
ecospat.cor.plot(data)
ecospat.co occurrences 25
ecospat.co occurrences
Species Co-Occurrences
Description
Calculate an index of species co-occurrences.
Usage
ecospat.co_occurrences (data)
Arguments
data A presence-absence matrix for each species (columns) in each location or
grid cell (rows) or a matrix with predicted suitability values.
Details
Computes an index of co-occurrences ranging from 0 (never co-occurring) to 1 (always
co-occuring).
Value
The species co-occurrence matrix and box-plot of the co-occurrence indices
Author(s)
Loic Pellissier <loic.pellissier@unifr.ch>
References
Pellissier, L., K.A. Brathen, J. Pottier, C.F. Randin, P. Vittoz, A. Dubuis, N.G. Yoccoz, T.
Alm, N.E. Zimmermann and A. Guisan. 2010. Species distribution models reveal apparent
competitive and facilitative effects of a dominant species on the distribution of tundra
plants. Ecography,33, 1004-1014.
Guisan, A. and N. Zimmermann. 2000. Predictive habitat distribution models in ecology.
Ecological Modelling,135:147-186
Examples
## Not run:
matrix <- ecospat.testData[c(9:16,54:57)]
ecospat.co_occurrences (data=matrix)
## End(Not run)
26 ecospat.Cscore
ecospat.Cscore Pairwise co-occurrence Analysis with calculation of the C-score
index.
Description
The function tests for nonrandom patterns of species co-occurrence in a presence-absence
matrix. It calculates the C-score index for the whole community and for each species pair.
Null communities have column sum fixed.
Usage
ecospat.Cscore (data, nperm, outpath)
Arguments
data A presence-absence dataframe for each species (columns) in each loca-
tion or grid cell (rows). Column names (species names) and row names
(sampling plots).
nperm The number of permutation in the null model.
outpath Path to specify where to save the results.
Details
This function allows to apply a pairwise null model analysis (Gotelli and Ulrich 2010) to a
presence-absence community matrix to determine which species associations are significant
across the study area. The strength of associations is quantified by the C-score index (Stone
and Roberts 1990) and a ’fixed-equiprobable’ null model algorithm is applied. The format
required for input databases: a plots (rows) x species (columns) matrix. Input matrices
should have column names (species names) and row names (sampling plots). NOTE: a SES
that is greater than 2 or less than -2 is statistically significant with a tail probability of less
than 0.05 (Gotelli & McCabe 2002).
Value
The function returns the C-score index for the observed community (ObsCscoreTot), p.value
(PValTot) and standardized effect size (SES.Tot). It saves also a table in the working
directory where the same metrics are calculated for each species pair (only the table with
species pairs with significant p-values is saved in this version)
Author(s)
Christophe Randin <christophe.randin@wsl.ch>and Manuela D’Amen ¡manuela.damen@msn.com¿
References
Gotelli, N.J. and D.J. McCabe. 2002. Species co-occurrence: a meta-analysis of JM Dia-
mond’s assembly rules model. Ecology,83, 2091-2096.
Gotelli, N.J. and W. Ulrich. 2010. The empirical Bayes approach as a tool to identify
non-random species associations. Oecologia,162, 463-477
Stone, L. and A. Roberts, A. 1990. The checkerboard score and species distributions.
Oecologia,85, 74-79
ecospat.cv.example 27
See Also
ecospat.co occurrences and ecospat.cons Cscore
Examples
## Not run:
data<- ecospat.testData[c(53,62,58,70,61,66,65,71,69,43,63,56,68,57,55,60,54,67,59,64)]
nperm <- 10000
outpath <- getwd()
ecospat.Cscore(data, nperm, outpath)
## End(Not run)
ecospat.cv.example Cross Validation Example Function
Description
Run the cross validation functions on an example data set.
Usage
ecospat.cv.example ()
Details
This function takes an example data set, calibrates it for various models, and then runs the
cross validation functions on the results. Mainly to show how to use the cross validation
functions.
Author(s)
Christophe Randin <christophe.randin@wsl.ch>and Antoine Guisan <antoine.guisan@unil.ch>
Examples
## Not run:
ecospat.cv.example ()
## End(Not run)
28 ecospat.cv.gbm
ecospat.cv.gbm GBM Cross Validation
Description
K-fold and leave-one-out cross validation for GBM.
Usage
ecospat.cv.gbm (gbm.obj, data.cv, K=10, cv.lim=10, jack.knife=FALSE)
Arguments
gbm.obj A calibrated GBM object with a binomial error distribution. Attention:
users have to tune model input parameters according to their study!
data.cv A dataframe object containing the calibration data set with the same
names for response and predictor variables.
KNumber of folds. 10 is recommended; 5 for small data sets.
cv.lim Minimum number of presences required to perform the K-fold cross-
validation.
jack.knife If TRUE, then the leave-one-out / jacknife cross-validation is performed
instead of the 10-fold cross-validation.
Details
This function takes a calibrated GBM object with a binomial error distribution and returns
predictions from a stratified 10-fold cross-validation or a leave-one-out / jack-knived cross-
validation. Stratified means that the original prevalence of the presences and absences in
the full dataset is conserved in each fold.
Value
Returns a dataframe with the observations (obs) and the corresponding predictions by
cross-validation or jacknife.
Author(s)
Christophe Randin <christophe.randin@unibas.ch>and Antoine Guisan <antoine.guisan@unil.ch>
References
Randin, C.F., T. Dirnbock, S. Dullinger, N.E. Zimmermann, M. Zappa and A. Guisan.
2006. Are niche-based species distribution models transferable in space? Journal of Bio-
geography,33, 1689-1703.
Pearman, P.B., C.F. Randin, O. Broennimann, P. Vittoz, W.O. van der Knaap, R. Engler,
G. Le Lay, N.E. Zimmermann and A. Guisan. 2008. Prediction of plant species distributions
across six millennia. Ecology Letters,11, 357-369.
ecospat.cv.glm 29
Examples
## Not run:
gbm <- ecospat.cv.gbm (gbm.obj= get ("gbm.Agrostis_capillaris", envir=ecospat.env),
ecospat.testData, K=10, cv.lim=10, jack.knife=FALSE)
## End(Not run)
ecospat.cv.glm GLM Cross Validation
Description
K-fold and leave-one-out cross validation for GLM.
Usage
ecospat.cv.glm (glm.obj, K=10, cv.lim=10, jack.knife=FALSE)
Arguments
glm.obj Any calibrated GLM object with a binomial error distribution.
KNumber of folds. 10 is recommended; 5 for small data sets.
cv.lim Minimum number of presences required to perform the K-fold cross-
validation.
jack.knife If TRUE, then the leave-one-out / jacknife cross-validation is performed
instead of the 10-fold cross-validation.
Details
This function takes a calibrated GLM object with a binomial error distribution and returns
predictions from a stratified 10-fold cross-validation or a leave-one-out / jack-knived cross-
validation. Stratified means that the original prevalence of the presences and absences in
the full dataset is conserved in each fold.
Value
Returns a dataframe with the observations (obs) and the corresponding predictions by
cross-validation or jacknife.
Author(s)
Christophe Randin <christophe.randin@unibas.ch>and Antoine Guisan <antoine.guisan@unil.ch>
References
Randin, C.F., T. Dirnbock, S. Dullinger, N.E. Zimmermann, M. Zappa and A. Guisan.
2006. Are niche-based species distribution models transferable in space? Journal of Bio-
geography,33, 1689-1703.
Pearman, P.B., C.F. Randin, O. Broennimann, P. Vittoz, W.O. van der Knaap, R. Engler,
G. Le Lay, N.E. Zimmermann and A. Guisan. 2008. Prediction of plant species distributions
across six millennia. Ecology Letters,11, 357-369.
30 ecospat.cv.me
Examples
## Not run:
glm <- ecospat.cv.glm (glm.obj = get ("glm.Agrostis_capillaris", envir=ecospat.env),
K=10, cv.lim=10, jack.knife=FALSE)
## End(Not run)
ecospat.cv.me Maxent Cross Validation
Description
K-fold and leave-one-out cross validation for Maxent.
Usage
ecospat.cv.me (data.cv.me, name.sp, names.pred, K=10, cv.lim=10, jack.knife=FALSE)
Arguments
data.cv.me A dataframe object containing the calibration data set of a Maxent object
to validate with the same names for response and predictor variables.
name.sp Name of the species / response variable.
names.pred Names of the predicting variables.
KNumber of folds. 10 is recommended; 5 for small data sets.
cv.lim Minimum number of presences required to perform the K-fold cross-
validation.
jack.knife If TRUE, then the leave-one-out / jacknife cross-validation is performed
instead of the 10-fold cross-validation.
Details
This function takes a calibrated Maxent object with a binomial error distribution and
returns predictions from a stratified 10-fold cross-validation or a leave-one-out / jack-knived
cross-validation. Stratified means that the original prevalence of the presences and absences
in the full dataset is conserved in each fold.
Value
Returns a dataframe with the observations (obs) and the corresponding predictions by
cross-validation or jacknife.
Author(s)
Christophe Randin <christophe.randin@unibas.ch>and Antoine Guisan <antoine.guisan@unil.ch>
ecospat.cv.rf 31
References
Randin, C.F., T. Dirnbock, S. Dullinger, N.E. Zimmermann, M. Zappa and A. Guisan.
2006. Are niche-based species distribution models transferable in space? Journal of Bio-
geography,33, 1689-1703.
Pearman, P.B., C.F. Randin, O. Broennimann, P. Vittoz, W.O. van der Knaap, R. Engler,
G. Le Lay, N.E. Zimmermann and A. Guisan. 2008. Prediction of plant species distributions
across six millennia. Ecology Letters,11, 357-369.
Examples
## Not run:
me <- ecospat.cv.me(ecospat.testData, names(ecospat.testData)[53],
names(ecospat.testData)[4:8], K = 10, cv.lim = 10, jack.knife = FALSE)
## End(Not run)
ecospat.cv.rf RandomForest Cross Validation
Description
K-fold and leave-one-out cross validation for randomForest.
Usage
ecospat.cv.rf (rf.obj, data.cv, K=10, cv.lim=10, jack.knife=FALSE)
Arguments
rf.obj Any calibrated randomForest object with a binomial error distribution.
data.cv A dataframe object containing the calibration data set with the same
names for response and predictor variables.
KNumber of folds. 10 is recommended; 5 for small data sets.
cv.lim Minimum number of presences required to perform the K-fold cross-
validation.
jack.knife If TRUE, then the leave-one-out / jacknife cross-validation is performed
instead of the 10-fold cross-validation.
Details
This function takes a calibrated randomForest object with a binomial error distribution
and returns predictions from a stratified 10-fold cross-validation or a leave-one-out / jack-
knived cross-validation. Stratified means that the original prevalence of the presences and
absences in the full dataset is conserved in each fold.
Value
Returns a dataframe with the observations (obs) and the corresponding predictions by
cross-validation or jacknife.
32 ecospat.Epred
Author(s)
Christophe Randin <christophe.randin@wsl.ch>and Antoine Guisan <antoine.guisan@unil.ch>
References
Randin, C.F., T. Dirnbock, S. Dullinger, N.E. Zimmermann, M. Zappa and A. Guisan.
2006. Are niche-based species distribution models transferable in space? Journal of Bio-
geography,33, 1689-1703.
Pearman, P.B., C.F. Randin, O. Broennimann, P. Vittoz, W.O. van der Knaap, R. Engler,
G. Le Lay, N.E. Zimmermann and A. Guisan. 2008. Prediction of plant species distributions
across six millennia. Ecology Letters,11, 357-369.
Examples
## Not run:
rf <- ecospat.cv.rf(get("rf.Agrostis_capillaris", envir = ecospat.env),
ecospat.testData[, c(53, 4:8)], K = 10, cv.lim = 10, jack.knife = FALSE)
## End(Not run)
ecospat.env Package Environment
Description
A package environment that is used to contain certain (local) variables and results, espe-
cially those in example functions and data sets.
Examples
ls(envir=ecospat.env)
ecospat.Epred Prediction Mean
Description
Calculate the mean (or weighted mean) of several predictions.
Usage
ecospat.Epred (x, w=rep(1,ncol(x)), th=0)
Arguments
xA dataframe object with SDM predictions.
wWeight of the model, e.g. AUC. The default is 1.
th Threshold used to binarize.
ecospat.ESM.EnsembleModeling 33
Details
The Weighted Average consensus method utilizes pre-evaluation of the predictive perfor-
mance of the single-models. In this approach, half (i.e. four) of the eight single-models with
highest accuracy are selected first, and then a WA is calculated based on the pre-evaluated
AUC of the single-models
Value
A weighted mean binary transformation of the models.
Author(s)
Blaise Petitpierre <bpetitpierre@gmail.com>
References
Boyce, M.S., P.R. Vernier, S.E. Nielsen and F.K.A. Schmiegelow. 2002. Evaluating resource
selection functions. Ecol. Model.,157, 281-300.
Marmion, M., M. Parviainen, M. Luoto, R.K. Heikkinen andW. Thuiller. 2009. Eval-
uation of consensus methods in predictive species distribution modelling. Diversity and
Distributions,15, 59-69.
Examples
x <- ecospat.testData[c(92,96)]
mean <- ecospat.Epred (x, w=rep(1,ncol(x)), th=0.5)
ecospat.ESM.EnsembleModeling
Ensamble of Small Models: Evaluates and Averages Simple Bi-
variate Models To ESMs
Description
This function evaluates and averages simple bivariate models by weighted means to Ensem-
ble Small Models as in Lomba et al. 2010 and Breiner et al. 2015.
Usage
ecospat.ESM.EnsembleModeling( ESM.modeling.output,
weighting.score,
threshold=NULL,
models)
Arguments
ESM.modeling.output
alist returned by ecospat.ESM.Modeling
34 ecospat.ESM.EnsembleModeling
weighting.score
an evaluation score used to weight single models to build ensembles:”AUC”,”TSS”,
”Boyce”,”Kappa”,”SomersD” #the evaluation methods used to evaluate
ensemble models
( see "BIOMOD Modeling"models.eval.meth section for more detailed
informations )
threshold threshold value of an evaluation score to select the bivariate model(s)
included for building the ESMs
models vector of models names choosen among ’GLM’, ’GBM’, ’GAM’, ’CTA’,
’ANN’, ’SRE’, ’FDA’, ’MARS’, ’RF’,’MAXENT.Phillips’, ”MAXENT.Tsuruoka”
(same as in biomod2)
#a character vector (either ’all’ or a sub-selection of model names) that
defines the models kept for building the ensemble models (might be useful
for removing some non-preferred models)
Details
The basic idea of ensemble of small models (ESMs) is to model a species distribution based
on small, simple models, for example all possible bivariate models (i.e. models that contain
only two predictors at a time out of a larger set of predictors), and then combine all possible
bivariate models into an ensemble (Lomba et al. 2010; Breiner et al. 2015).
The ESM set of functions could be used to build ESMs using simple bivariate models which
are averaged using weights based on model performances (e.g. AUC) according to Breiner
et al. (2015). They provide full functionality of the approach described in Breiner et al.
(2015).
Value
species: species name ESM.fit: data.frame of the predicted values for the data used to build
the models. ESM.evaluations: data.frame with evaluations scores for the ESMs weights:
weighting scores used to weight the bivariate models to build the single ESM weights.EF:
weighting scores used to weight the single ESM to build the ensemble of ESMs from differ-
ent modelling techniques (only available if ¿1 modelling techniques were selected). failed:
bivariate models which failed because they could not be calibrated.
A"BIOMOD.EnsembleModeling.out". This object will be later given to ecospat.ESM.EnsembleProjection
if you want to make some projections of this ensemble-models.
Author(s)
Frank Breiner <frank.breiner@wsl.ch>
with contributions of Olivier Broennimann <olivier.broennimann@unil.ch>
References
Lomba, A., L. Pellissier, C.F. Randin, J. Vicente, F. Moreira, J. Honrado and A. Guisan.
2010. Overcoming the rare species modelling paradox: A novel hierarchical framework
applied to an Iberian endemic plant. Biological Conservation,143,2647-2657.
Breiner F.T., A. Guisan, A. Bergamini and M.P. Nobis. 2015. Overcoming limitations
of modelling rare species by using ensembles of small models. Methods in Ecology and
Evolution,6,1210-1218.
ecospat.ESM.EnsembleModeling 35
Breiner F.T., Nobis M.P., Bergamini A., Guisan A. 2018. Optimizing ensembles of small
models for predicting the distribution of species with few occurrences. Methods in Ecology
and Evolution. doi: https://doi.org/10.1111/2041210X.12957
See Also
ecospat.ESM.Modeling,ecospat.ESM.Projection,ecospat.ESM.EnsembleProjection
BIOMOD Modeling,BIOMOD Projection
Examples
## Not run:
# Loading test data
inv <- ecospat.testNiche.inv
# species occurrences
xy <- inv[,1:2]
sp_occ <- inv[11]
# env
current <- inv[3:10]
### Formating the data with the BIOMOD_FormatingData() function from the package biomod2
sp <- 1
myBiomodData <- BIOMOD_FormatingData( resp.var = as.numeric(sp_occ[,sp]),
expl.var = current,
resp.xy = xy,
resp.name = colnames(sp_occ)[sp])
### Calibration of simple bivariate models
my.ESM <- ecospat.ESM.Modeling( data=myBiomodData,
models=c('GLM','RF'),
NbRunEval=2,
DataSplit=70,
weighting.score=c("AUC"),
parallel=FALSE)
### Evaluation and average of simple bivariate models to ESMs
my.ESM_EF <- ecospat.ESM.EnsembleModeling(my.ESM,weighting.score=c("SomersD"),threshold=0)
### Projection of simple bivariate models into new space
my.ESM_proj_current<-ecospat.ESM.Projection(ESM.modeling.output=my.ESM,
new.env=current)
### Projection of calibrated ESMs into new space
my.ESM_EFproj_current <- ecospat.ESM.EnsembleProjection(ESM.prediction.output=my.ESM_proj_current,
ESM.EnsembleModeling.output=my.ESM_EF)
## get the model performance of ESMs
my.ESM_EF$ESM.evaluations
## get the weights of the single bivariate models used to build the ESMs
my.ESM_EF$weights
36 ecospat.ESM.EnsembleProjection
## End(Not run)
ecospat.ESM.EnsembleProjection
Ensamble of Small Models: Projects Calibrated ESMs Into New
Space Or Time.
Description
This function projects calibrated ESMs into new space or time.
Usage
ecospat.ESM.EnsembleProjection( ESM.prediction.output,
ESM.EnsembleModeling.output,
chosen.models = 'all')
Arguments
ESM.prediction.output
a list object returned by ecospat.ESM.Projection
ESM.EnsembleModeling.output
a list object returned by ecospat.ESM.EnsembleModeling
chosen.models a character vector (either ’all’ or a sub-selection of model names) to re-
move models from the ensemble (same as in biomod2). Default is ’all’.
Details
The basic idea of ensemble of small models (ESMs) is to model a species distribution based
on small, simple models, for example all possible bivariate models (i.e. models that contain
only two predictors at a time out of a larger set of predictors), and then combine all possible
bivariate models into an ensemble (Lomba et al. 2010; Breiner et al. 2015).
The ESM set of functions could be used to build ESMs using simple bivariate models which
are averaged using weights based on model performances (e.g. AUC) according to Breiner
et al. (2015). They provide full functionality of the approach described in Breiner et al.
(2015).
For further details please refer to BIOMOD EnsembleForecasting.
Value
Returns the projections of ESMs for the selected single models and their ensemble (data
frame or raster stack). ESM.projections ‘projection files’ are saved on the hard drive pro-
jection folder. This files are either an array or a RasterStack depending the original
projections data type. Load these created files to plot and work with them.
Author(s)
Frank Breiner <frank.breiner@wsl.ch>
ecospat.ESM.EnsembleProjection 37
References
Lomba, A., L. Pellissier, C.F. Randin, J. Vicente, F. Moreira, J. Honrado and A. Guisan.
2010. Overcoming the rare species modelling paradox: A novel hierarchical framework
applied to an Iberian endemic plant. Biological Conservation,143,2647-2657.
Breiner F.T., A. Guisan, A. Bergamini and M.P. Nobis. 2015. Overcoming limitations
of modelling rare species by using ensembles of small models. Methods in Ecology and
Evolution,6,1210-1218.
Breiner F.T., Nobis M.P., Bergamini A., Guisan A. 2018. Optimizing ensembles of small
models for predicting the distribution of species with few occurrences. Methods in Ecology
and Evolution. doi: https://doi.org/10.1111/2041210X.12957
See Also
ecospat.ESM.Modeling,ecospat.ESM.Projection,ecospat.ESM.EnsembleModeling
BIOMOD Modeling,BIOMOD Projection,BIOMOD EnsembleForecasting,BIOMOD EnsembleModeling
Examples
## Not run:
# Loading test data for the niche dynamics analysis in the invaded range
inv <- ecospat.testNiche.inv
# species occurrences
xy <- inv[,1:2]
sp_occ <- inv[11]
# env
current <- inv[3:10]
### Formating the data with the BIOMOD_FormatingData() function form the package biomod2
setwd(path.wd)
t1 <- Sys.time()
sp <- 1
myBiomodData <- BIOMOD_FormatingData( resp.var = as.numeric(sp_occ[,sp]),
expl.var = current,
resp.xy = xy,
resp.name = colnames(sp_occ)[sp])
myBiomodOption <- Print_Default_ModelingOptions()
### Calibration of simple bivariate models
my.ESM <- ecospat.ESM.Modeling( data=myBiomodData,
models=c('GLM','RF'),
models.options=myBiomodOption,
NbRunEval=2,
DataSplit=70,
weighting.score=c("AUC"),
parallel=FALSE)
### Evaluation and average of simple bivariate models to ESMs
my.ESM_EF <- ecospat.ESM.EnsembleModeling(my.ESM,weighting.score=c("SomersD"),threshold=0)
38 ecospat.ESM.Modeling
### Projection of simple bivariate models into new space
my.ESM_proj_current<-ecospat.ESM.Projection(ESM.modeling.output=my.ESM,
new.env=current)
### Projection of calibrated ESMs into new space
my.ESM_EFproj_current <- ecospat.ESM.EnsembleProjection(ESM.prediction.output=my.ESM_proj_current,
ESM.EnsembleModeling.output=my.ESM_EF)
## get the model performance of ESMs
my.ESM_EF$ESM.evaluations
## get the weights of the single bivariate models used to build the ESMs
my.ESM_EF$weights
## End(Not run)
ecospat.ESM.Modeling Ensamble of Small Models: Calibration of Simple Bivariate Mod-
els
Description
This function calibrates simple bivariate models as in Lomba et al. 2010 and Breiner et al.
2015.
Usage
ecospat.ESM.Modeling( data,
NbRunEval,
DataSplit,
DataSplitTable,
Prevalence,
weighting.score,
models,
tune,
modeling.id,
models.options,
which.biva,
parallel,
cleanup)
Arguments
data BIOMOD.formated.data object returned by BIOMOD FormatingData
NbRunEval number of dataset splitting replicates for the model evaluation (same as
in biomod2)
DataSplit percentage of dataset observations retained for the model training (same
as in biomod2)
DataSplitTable a matrix, data.frame or a 3D array filled with TRUE/FALSE to specify
which part of data must be used for models calibration (TRUE) and for
models validation (FALSE). Each column corresponds to a ’RUN’. If filled,
arguments NbRunEval and DataSplit will be ignored.
ecospat.ESM.Modeling 39
Prevalence either NULL or a 0-1 numeric used to build ’weighted response weights’.
In contrast to Biomod the default is 0.5 (weighting presences equally to
the absences). If NULL each observation (presence or absence) has the
same weight (independent of the number of presences and absences).
weighting.score
evaluation score used to weight single models to build ensembles: ’AUC’,
’SomersD’ (2xAUC-1), ’Kappa’, ’TSS’ or ’Boyce’
models vector of models names choosen among ’GLM’, ’GBM’, ’GAM’, ’CTA’,
’ANN’, ’SRE’, ’FDA’, ’MARS’, ’RF’,’MAXENT.Phillips’, ’MAXENT.Tsuruoka’
(same as in biomod2)
tune logical. if true model tuning will be used to estimate optimal parameters
for the models (Default: False).
modeling.id character, the ID (=name) of modeling procedure. A random number by
default.
models.options BIOMOD.models.options object returned by BIOMOD ModelingOptions
(same as in biomod2)
which.biva integer. which bivariate combinations should be used for modeling? De-
fault: all
parallel logical. If TRUE, the parallel computing is enabled (highly recommended)
cleanup numeric. Calls removeTmpFiles() to delete all files from rasterOptions() tmpdir
which are older than the given time (in hours). This might be necessary
to prevent running over quota. No cleanup is used by default.
Details
The basic idea of ensemble of small models (ESMs) is to model a species distribution based
on small, simple models, for example all possible bivariate models (i.e. models that contain
only two predictors at a time out of a larger set of predictors), and then combine all possible
bivariate models into an ensemble (Lomba et al. 2010; Breiner et al. 2015).
The ESM set of functions could be used to build ESMs using simple bivariate models which
are averaged using weights based on model performances (e.g. AUC) according to Breiner
et al. (2015). They provide full functionality of the approach described in Breiner et al.
(2015).
The argument which.biva allows to split model runs, e.g. if which.biva is 1:3, only the
three first bivariate variable combinations will be modeled. This allows to run different
biva splits on different computers. However, it is better not to use this option if all models
are run on a single computer. Default: running all biva models. NOTE: Make sure to give
each of your biva runs a unique modeling.id.
Value
A BIOMOD.models.out object (same as in biomod2) See "BIOMOD.models.out"for details.
Author(s)
Frank Breiner <frank.breiner@wsl.ch>and Mirko Di Febbraro <mirkodifebbraro@gmail.com>
with contributions of Olivier Broennimann <olivier.broennimann@unil.ch>
40 ecospat.ESM.Modeling
References
Lomba, A., L. Pellissier, C.F. Randin, J. Vicente, F. Moreira, J. Honrado and A. Guisan.
2010. Overcoming the rare species modelling paradox: A novel hierarchical framework
applied to an Iberian endemic plant. Biological Conservation,143,2647-2657.
Breiner F.T., A. Guisan, A. Bergamini and M.P. Nobis. 2015. Overcoming limitations
of modelling rare species by using ensembles of small models. Methods in Ecology and
Evolution,6,1210-1218.
Breiner F.T., Nobis M.P., Bergamini A., Guisan A. 2018. Optimizing ensembles of small
models for predicting the distribution of species with few occurrences. Methods in Ecology
and Evolution. doi: https://doi.org/10.1111/2041210X.12957
See Also
ecospat.ESM.EnsembleModeling,ecospat.ESM.Projection,ecospat.ESM.EnsembleProjection
BIOMOD FormatingData,BIOMOD ModelingOptions,BIOMOD Modeling,BIOMOD Projection
Examples
## Not run:
# Loading test data
inv <- ecospat.testNiche.inv
# species occurrences
xy <- inv[,1:2]
sp_occ <- inv[11]
# env
current <- inv[3:10]
### Formating the data with the BIOMOD_FormatingData() function from the package biomod2
sp <- 1
myBiomodData <- BIOMOD_FormatingData( resp.var = as.numeric(sp_occ[,sp]),
expl.var = current,
resp.xy = xy,
resp.name = colnames(sp_occ)[sp])
### Calibration of simple bivariate models
my.ESM <- ecospat.ESM.Modeling( data=myBiomodData,
models=c('GLM','RF'),
NbRunEval=2,
DataSplit=70,
Prevalence=0.5
weighting.score=c("AUC"),
parallel=FALSE)
### Evaluation and average of simple bivariate models to ESMs
my.ESM_EF <- ecospat.ESM.EnsembleModeling(my.ESM,weighting.score=c("SomersD"),threshold=0)
### Projection of simple bivariate models into new space
my.ESM_proj_current<-ecospat.ESM.Projection(ESM.modeling.output=my.ESM,
new.env=current)
ecospat.ESM.Projection 41
### Projection of calibrated ESMs into new space
my.ESM_EFproj_current <- ecospat.ESM.EnsembleProjection(ESM.prediction.output=my.ESM_proj_current,
ESM.EnsembleModeling.output=my.ESM_EF)
## get the model performance of ESMs
my.ESM_EF$ESM.evaluations
## get the weights of the single bivariate models used to build the ESMs
my.ESM_EF$weights
## End(Not run)
ecospat.ESM.Projection
Ensamble of Small Models: Projects Simple Bivariate Models
Into New Space Or Time
Description
This function projects simple bivariate models on new.env
Usage
ecospat.ESM.Projection(ESM.modeling.output,
new.env,
parallel,
cleanup)
Arguments
ESM.modeling.output
list object returned by ecospat.ESM.Modeling
new.env A set of explanatory variables onto which models will be projected. It
could be a data.frame, a matrix, or a rasterStack object. Make sure
the column names (data.frame or matrix) or layer Names (rasterStack)
perfectly match with the names of variables used to build the models in
the previous steps.
parallel Logical. If TRUE, the parallel computing is enabled
cleanup Numeric. Calls removeTmpFiles() to delete all files from rasterOptions() tmpdir
which are older than the given time (in hours). This might be necessary
to prevent running over quota. No cleanup is used by default
Details
The basic idea of ensemble of small models (ESMs) is to model a species distribution based
on small, simple models, for example all possible bivariate models (i.e. models that contain
only two predictors at a time out of a larger set of predictors), and then combine all possible
bivariate models into an ensemble (Lomba et al. 2010; Breiner et al. 2015).
The ESM set of functions could be used to build ESMs using simple bivariate models which
are averaged using weights based on model performances (e.g. AUC) accoring to Breiner
et al (2015). They provide full functionality of the approach described in Breiner et al.
(2015).
The name of new.env must be a regular expression (see ?regex)
42 ecospat.ESM.Projection
Value
Returns the projections for all selected models (same as in biomod2) See "BIOMOD.projection.out"
for details.
Author(s)
Frank Breiner <frank.breiner@wsl.ch>
with contributions of Olivier Broennimann <olivier.broennimann@unil.ch>
References
Lomba, A., L. Pellissier, C.F. Randin, J. Vicente, F. Moreira, J. Honrado and A. Guisan.
2010. Overcoming the rare species modelling paradox: A novel hierarchical framework
applied to an Iberian endemic plant. Biological Conservation,143,2647-2657.
Breiner F.T., A. Guisan, A. Bergamini and M.P. Nobis. 2015. Overcoming limitations
of modelling rare species by using ensembles of small models. Methods in Ecology and
Evolution,6,1210-1218.
Breiner F.T., Nobis M.P., Bergamini A., Guisan A. 2018. Optimizing ensembles of small
models for predicting the distribution of species with few occurrences. Methods in Ecology
and Evolution. doi: https://doi.org/10.1111/2041210X.12957
See Also
ecospat.ESM.EnsembleModeling,ecospat.ESM.Modeling,ecospat.ESM.EnsembleProjection
BIOMOD FormatingData,BIOMOD ModelingOptions,BIOMOD Modeling,BIOMOD Projection
Examples
## Not run:
# Loading test data
inv <- ecospat.testNiche.inv
# species occurrences
xy <- inv[,1:2]
sp_occ <- inv[11]
# env
current <- inv[3:10]
### Formating the data with the BIOMOD_FormatingData() function from the package biomod2
sp <- 1
myBiomodData <- BIOMOD_FormatingData( resp.var = as.numeric(sp_occ[,sp]),
expl.var = current,
resp.xy = xy,
resp.name = colnames(sp_occ)[sp])
### Calibration of simple bivariate models
my.ESM <- ecospat.ESM.Modeling( data=myBiomodData,
models=c('GLM','RF'),
NbRunEval=2,
DataSplit=70,
ecospat.grid.clim.dyn 43
weighting.score=c("AUC"),
parallel=FALSE)
### Evaluation and average of simple bivariate models to ESMs
my.ESM_EF <- ecospat.ESM.EnsembleModeling(my.ESM,weighting.score=c("SomersD"),threshold=0)
### Projection of simple bivariate models into new space
my.ESM_proj_current<-ecospat.ESM.Projection(ESM.modeling.output=my.ESM,
new.env=current)
### Projection of calibrated ESMs into new space
my.ESM_EFproj_current <- ecospat.ESM.EnsembleProjection(ESM.prediction.output=my.ESM_proj_current,
ESM.EnsembleModeling.output=my.ESM_EF)
## get the model performance of ESMs
my.ESM_EF$ESM.evaluations
## get the weights of the single bivariate models used to build the ESMs
my.ESM_EF$weights
## End(Not run)
ecospat.grid.clim.dyn Dynamic Occurrence Densities Grid
Description
Create a grid with occurrence densities along one or two gridded environmental gradients.
Usage
ecospat.grid.clim.dyn (glob, glob1, sp, R, th.sp, th.env, geomask)
Arguments
glob A two-column dataframe (or a vector) of the environmental values (in
column) for background pixels of the whole study area (in row).
glob1 A two-column dataframe (or a vector) of the environmental values (in
column) for the background pixels of the species (in row).
sp A two-column dataframe (or a vector) of the environmental values (in
column) for the occurrences of the species (in row).
RThe resolution of the grid.
th.sp The quantile used to delimit a threshold to exclude low species density
values.
th.env The quantile used to delimit a threshold to exclude low environmental
density values of the study area.
geomask A geographical mask to delimit the background extent if the analysis takes
place in the geographical space.
It can be a SpatialPolygon or a raster object. Note that the CRS should
be the same as the one used for the points.
44 ecospat.grid.clim.dyn
Details
Using the scores of an ordination (or SDM prediction), create a grid z of RxR pixels (or a
vector of R pixels when using scores of dimension 1 or SDM predictions) with occurrence
densities. Only scores of one, or two dimensions can be used. th.sp is the quantile of the
distribution of species density at occurrence sites. For example, if th.sp is set to 0.05,
the the species niche is drawn by including 95 percent of the species occurrences, removing
the more marginal populations. Similarly, th.env is the quantile of the distribution of
the environmental density at all sites of the study area. If th.env is set to 0.05, the
delineation of the study area in the environmental space includes 95 percent of the study
area, removing the more marginal sites of the study area. By default, these thresholds are
set to 0 but can be modified, depending on the importance of some marginal sites in the
delineation of the species niche and/or the study area in the environmnental space. It is
recommended to check if the shape of the delineated niche and study area corresponds to
the shape of the plot of the PCA scores (or any other ordination techniques used to set
the environmental space). Visualisation of the gridded environmental space can be done
through the functions ecospat.plot.niche or ecospat.plot.niche.dyn If you encounter
a problem during your analyses, please first read the FAQ section of ”Niche overlap” in
http://www.unil.ch/ecospat/home/menuguid/ecospat-resources/tools.html The argument
geomask can be a SpatialPolygon or a raster object. Note that the CRS should be the same
as the one used for the points.
Value
A grid z of RxR pixels (or a vector of R pixels) with z.uncor being the density of occur-
rence of the species, and z.cor the occupancy of the environment by the species (density of
occurrences divided by the desinty of environment in the study area.
Author(s)
Olivier Broennimann <olivier.broennimann@unil.ch>and Blaise Petitpierre <bpetitpierre@gmail.com>
References
Broennimann, O., M.C. Fitzpatrick, P.B. Pearman, B. Petitpierre, L. Pellissier, N.G. Yoc-
coz, W. Thuiller, M.J. Fortin, C. Randin, N.E. Zimmermann, C.H. Graham and A. Guisan.
2012. Measuring ecological niche overlap from occurrence and spatial environmental data.
Global Ecology and Biogeography,21:481-497.
Petitpierre, B., C. Kueffer, O. Broennimann, C. Randin, C. Daehler and A. Guisan. 2012.
Climatic niche shifts are rare among terrestrial plant invaders. Science,335:1344-1348.
See Also
ecospat.plot.niche.dyn
Examples
## Not run:
spp <- ecospat.testNiche
clim <- ecospat.testData[2:8]
occ.sp_test <- na.exclude(ecospat.sample.envar(dfsp=spp,colspxy=2:3,colspkept=1:3,dfvar=clim,
colvarxy=1:2,colvar="all",resolution=25))
occ.sp<-cbind(occ.sp_test,spp[,4]) #add species names
ecospat.makeDataFrame 45
# list of species
sp.list<-levels(occ.sp[,1])
sp.nbocc<-c()
for (i in 1:length(sp.list)){sp.nbocc<-c(sp.nbocc,length(which(occ.sp[,1] == sp.list[i])))}
#calculate the nb of occurences per species
sp.list <- sp.list[sp.nbocc>4] # remove species with less than 5 occurences
nb.sp <- length(sp.list) #nb of species
ls()
# selection of variables to include in the analyses
# try with all and then try only worldclim Variables
Xvar <- c(3:7)
nvar <- length(Xvar)
#number of interation for the tests of equivalency and similarity
iterations <- 100
#resolution of the gridding of the climate space
R <- 100
#################################### PCA-ENVIRONMENT ##################################
data<-rbind(occ.sp[,Xvar+1],clim[,Xvar])
w <- c(rep(0,nrow(occ.sp)),rep(1,nrow(clim)))
pca.cal <- dudi.pca(data, row.w = w, center = TRUE, scale = TRUE, scannf = FALSE, nf = 2)
####### selection of species ######
sp.list
sp.combn <- combn(1:2,2)
for(i in 1:ncol(sp.combn)) {
row.sp1 <- which(occ.sp[,1] == sp.list[sp.combn[1,i]]) # rows in data corresponding to sp1
row.sp2 <- which(occ.sp[,1] == sp.list[sp.combn[2,i]]) # rows in data corresponding to sp2
name.sp1 <- sp.list[sp.combn[1,i]]
name.sp2 <- sp.list[sp.combn[2,i]]
# predict the scores on the axes
scores.clim <- pca.cal$li[(nrow(occ.sp)+1):nrow(data),] #scores for global climate
scores.sp1 <- pca.cal$li[row.sp1,] #scores for sp1
scores.sp2 <- pca.cal$li[row.sp2,] #scores for sp2
}
# calculation of occurence density and test of niche equivalency and similarity
z1 <- ecospat.grid.clim.dyn(scores.clim, scores.clim, scores.sp1,R=100)
z2 <- ecospat.grid.clim.dyn(scores.clim, scores.clim, scores.sp2,R=100)
## End(Not run)
ecospat.makeDataFrame Make Data Frame
Description
Create a biomod2-compatible dataframe. The function also enables to remove duplicate
presences within a pixel and to set a minimum distance between presence points to avoid
autocorrelation. Data from GBIF can be added.
46 ecospat.makeDataFrame
Usage
ecospat.makeDataFrame (spec.list, expl.var, use.gbif=FALSE, precision=NULL,
year=NULL, remdups=TRUE, mindist=NULL, n=1000, type='random', PApoint=NULL,
ext=expl.var, tryf=5)
Arguments
spec.list Data.frame or Character. The species occurrence information must be a
data.frame in the form: \’x-coordinates\’ , \’y-coordinates\’ and \’species
name\’ (in the same projection/coordinate system as expl.var!).
expl.var a RasterStack object of the environmental layers.
use.gbif Logical. If TRUE presence data from GBIF will be added. It is also
possible to use GBIF data only. Default: FALSE. See ?gbif dismo for
more information. Settings: geo=TRUE, removeZeros=TRUE, all sub-
taxa will be used. \’species name\’ in spec.list must be in the form:
\’genus species\’, \’genus species\’ or \’genus.species\’. If there is no
species information available on GBIF an error is returned. Try to change
species name (maybe there is a synonym) or switch use.gbif off.
precision Numeric. Use precision if use.gbif = TRUE to set a minimum precision
of the presences which should be added. For precision = 1000 e.g. only
presences with precision of at least 1000 meter will be added from GBIF.
When precision = NULL all presences from GBIF will be used, also pres-
ences where precision information is NA.
year Numeric. Latest year of the collected gbif occurrences. If year=1960 only
occurrences which were collected since 1960 are used.
remdups Logical. If TRUE (Default) duplicated presences within a raster pixel will
be removed. You will get only one presence per pixel.
mindist Numeric. You can set a minimum distance between presence points to
avoid autocorrelation. nndist spatstat is used to calculate the nearest
neighbour (nn) for each point. From the pair of the minimum nn, the
point is removed of which the second neighbour is closer. Unit is the
same as expl.var.
nnumber of Pseudo-Absences. Default 1000.
type sampling dessign for selecting Pseudo-Absences. If \’random\’ (default)
background points are selected with the function randomPoints dismo.
When selecting another sampling type (\’regular\’, \’stratified\’, \’nonaligned\’,
\’hexagonal\’, \’clustered\’ or \’Fibonacci\’) spsample sp will be used.
This can immensely increase computation time and RAM usage if ext is
a raster, especially for big raster layers because it must be converted into
a\’SpatialPolygonsDataFrame\’ first.
PApoint data.frame or SpatialPoints. You can use your own set of Pseudo-Absences
instead of generating new PAs. Two columns with \’x\’ and \’y\’ in the
same projection/coordinate system as expl.var!
ext a Spatial Object or Raster object. Extent from which PAs should be
selected from (Default is expl.var).
tryf numeric ¿ 1. Number of trials to create the requested Pseudo-Absences
after removing NA points (if type=’random’). See ?randomPoints dismo
ecospat.mantel.correlogram 47
Details
If you use a raster stack as explanatory variable and you want to model many species in
a loop with Biomod, formating data will last very long as presences and PA’s have to be
extracted over and over again from the raster stack. To save computation time, it is better
to convert the presences and PAs to a data.frame first.
Value
A data.frame object which can be used for modeling with the Biomod package.
Author(s)
Frank Breiner <frank.breiner@unil.ch>
Examples
## Not run:
files <- list.files(path=paste(system.file(package="dismo"),
'/ex', sep=''), pattern='grd', full.names=TRUE )
predictors <- raster::stack(files[c(9,1:8)]) #file 9 has more NA values than
# the other files, this is why we choose it as the first layer (see ?randomPoints)
file <- paste(system.file(package="dismo"), "/ex/bradypus.csv", sep="")
bradypus <- read.table(file, header=TRUE, sep=',')[,c(2,3,1)]
head(bradypus)
random.spec <- cbind(as.data.frame(randomPoints(predictors,50,extf=1)),species="randomSpec")
colnames(random.spec)[1:2] <- c("lon","lat")
spec.list <- rbind(bradypus, random.spec)
df <- ecospat.makeDataFrame(spec.list, expl.var=predictors, n=5000)
head(df)
plot(predictors[[1]])
points(df[df$Bradypus.variegatus==1, c('x','y')])
points(df[df$randomSpec==1, c('x','y')], col="red")
## End(Not run)
ecospat.mantel.correlogram
Mantel Correlogram
Description
Investigate spatial autocorrelation of environmental covariables within a set of occurrences
as a function of distance.
Usage
ecospat.mantel.correlogram (dfvar, colxy, n, colvar, max, nclass, nperm)
48 ecospat.max.kappa
Arguments
dfvar A dataframe object with the environmental variables.
colxy The range of columns for x and y in df.
nThe number of random occurrences used for the test.
colvar The range of columns for variables in df.
max The maximum distance to be computed in the correlogram.
nclass The number of classes of distances to be computed in the correlogram.
nperm The number of permutations in the randomization process.
Details
Requires ecodist library. Note that computation time increase tremendously when using
more than 500 occurrences (n¿500)
Value
Draws a plot with distance vs. the mantel r value. Black circles indicate that the values
are significative different from zero. White circles indicate non significant autocorrelation.
The selected distance is at the first white circle where values are non significative different
from cero.
Author(s)
Olivier Broennimann <olivier.broennimann@unil.ch>
References
Legendre, P. and M.J. Fortin. 1989. Spatial pattern and ecological analysis. Vegetatio,80,
107-138.
See Also
mgram
Examples
ecospat.mantel.correlogram(dfvar=ecospat.testData[c(2:16)],colxy=1:2, n=100, colvar=3:7,
max=1000, nclass=10, nperm=100)
ecospat.max.kappa Maximum Kappa
Description
Calculates values for Cohen’s Kappa along different thresholds, considering this time 0.01
increments (i.e. 99 thresholds).
Usage
ecospat.max.kappa(Pred, Sp.occ)
ecospat.max.tss 49
Arguments
Pred A vector of predicted probabilities
Sp.occ A vector of binary observations of the species occurrence
Value
Return values for Cohen’s Kappa for 99 thresholds at 0.01 increments.
Author(s)
Antoine Guisan <antoine.guisan@unil.ch>with contributions of Luigi Maiorano <luigi.maiorano@gmail.com>
and Valeria Di Cola <valeria.dicola@unil.ch>.
References
Liu, C., P.M. Berry, T.P. Dawson, and R.G. Pearson. 2005. Selecting thresholds of occur-
rence in the prediction of species distributions. Ecography,28, 385-393.
See Also
ecospat.meva.table,ecospat.max.tss,ecospat.plot.tss,ecospat.cohen.kappa,ecospat.plot.kappa
Examples
## Not run:
Pred <- ecospat.testData$glm_Agrostis_capillaris
Sp.occ <- ecospat.testData$Agrostis_capillaris
kappa100 <- ecospat.max.kappa(Pred, Sp.occ)
## End(Not run)
ecospat.max.tss Maximum TSS
Description
Calculates values for True skill statistic (TSS) along different thresholds, considering this
time 0.01 increments (i.e. 99 thresholds).
Usage
ecospat.max.tss(Pred, Sp.occ)
Arguments
Pred A vector of predicted probabilities
Sp.occ A vector of binary observations of the species occurrence
Value
Return values for TSS for 99 thresholds at 0.01 increments.
50 ecospat.maxentvarimport
Author(s)
Luigi Maiorano <luigi.maiorano@gmail.com>with contributions of Antoine Guisan <antoine.guisan@unil.ch>
References
Liu, C., P.M. Berry, T.P. Dawson, and R.G. Pearson. 2005. Selecting thresholds of occur-
rence in the prediction of species distributions. Ecography,28, 385-393.
See Also
ecospat.meva.table,ecospat.max.kappa,ecospat.plot.tss,ecospat.cohen.kappa,ecospat.plot.kappa
Examples
Pred <- ecospat.testData$glm_Agrostis_capillaris
Sp.occ <- ecospat.testData$Agrostis_capillaris
TSS100 <- ecospat.max.tss(Pred, Sp.occ)
ecospat.maxentvarimport
Maxent Variable Importance
Description
Calculate the importance of variables for Maxent in the same way Biomod does, by ran-
domly permuting each predictor variable independently, and computing the associated re-
duction in predictive performance.
Usage
ecospat.maxentvarimport (model, dfvar, nperm)
Arguments
model The name of the maxent model.
dfvar A dataframe object with the environmental variables.
nperm The number of permutations in the randomization process. The default
is 5.
Details
The calculation is made as biomod2 ”variables importance” function. It’s more or less base
on the same principle than randomForest variables importance algorithm. The principle is
to shuffle a single variable of the given data. Make model prediction with this ’shuffled’
data.set. Then we compute a simple correlation (Pearson’s by default) between references
predictions and the ’shuffled’ one. The return score is 1-cor(pred ref,pred shuffled). The
highest the value, the more influence the variable has on the model. A value of this 0
assumes no influence of that variable on the model. Note that this technique does not
account for interactions between the variables.
ecospat.mdr 51
Value
alist which contains a data.frame containing variables importance scores for each per-
mutation run.
Author(s)
Blaise Petitpierre <bpetitpierre@gmail.com>
Examples
## Not run:
model <- get ("me.Achillea_millefolium", envir=ecospat.env)
dfvar <- ecospat.testData[4:8]
nperm <- 5
ecospat.maxentvarimport (model, cal, nperm)
## End(Not run)
ecospat.mdr Minimum Dispersal Routes)
Description
ecospat.mdr is a function that implement a minimum cost arborescence approach to analyse
the invasion routes of a species from dates occurrence data.
Usage
ecospat.mdr (data, xcol, ycol, datecol, mode, rep, mean.date.error, fixed.sources.rows)
Arguments
data dataframe with occurence data. Each row correspond to an occurrence.
xcol The column in data containing x coordinates.
ycol The column in data containing y coordinates.
datecol The column in data containing dates.
mode ”observed”, ”min” or ”random”. ”observed” calculate routes using real
dates. ”min” reorder dates so the the total length of the routes are mini-
mal. ”random” reatribute dates randomly.
rep number of iteration of the analyse. if ¿ 1, boostrap support for each route
is provided.
mean.date.error
mean number of years to substract to observed dates. It is the mean of
the truncated negative exponential distribution from which the time to
be substracted is randomly sampled.
fixed.sources.rows
the rows in data (as a vector) corresponding to source occurrence(s)
that initiated the invasion(s). No incoming routes are not calculated for
sources.
52 ecospat.mdr
Details
The function draws an incoming route to each occurence from the closest occurrence already
occupied (with a previous date) and allows to substract a random number of time (years) to
the observed dates from a truncated negative exponential distribution. It is possible to run
several iterations and to get boostrap support for each route. itexp and rtexp functions
are small internal functions to set a truncated negative exponential distribution.
Value
A list is returned by the function with in positon [[1]], a datafame with each row corre-
sponding to a route (with new/old coordinates, new/old dates, length of the route, timespan,
dispersal rate), in position [[2]] the total route length, in position [[3]] the median dispersal
rate and in position [[4]] the number of outgoing nodes (index of clustering of the network)
Author(s)
Olivier Broennimann <olivier.broennimann@unil.ch>
References
Hordijk, W. and O. Broennimann. 2012. Dispersal routes reconstruction and the minimum
cost arborescence problem. Journal of theoretical biology,308, 115-122.
Broennimann, O., P. Mraz, B. Petitpierre, A. Guisan, and H. Muller-Scharer. 2014. Con-
trasting spatio-temporal climatic niche dynamics during the eastern and western invasions
of spotted knapweed in North America.Journal of biogeography,41, 1126-1136.
Examples
## Not run:
library(maps)
data<- ecospat.testMdr
fixed.sources.rows<-order(data$date)[1:2] #first introductions
#plot observed situation
plot(data[,2:1],pch=15,cex=0.5)
points(data[fixed.sources.rows,2:1],pch=19,col="red")
text(data[,2]+0.5,data[,1]+0.5,data[,3],cex=0.5)
map(add=T)
# mca
obs<-ecospat.mdr(data=data,
xcol=2,
ycol=1,
datecol=3,
mode="min",
rep=100,
mean.date.error=1,
fixed.sources.rows)
#plot results
lwd<-(obs[[1]]$bootstrap.value)
x11();plot(obs[[1]][,3:4],type="n",xlab="longitude",ylab="latitude")
arrows(obs[[1]][,1],obs[[1]][,2],obs[[1]][,3],obs[[1]][,4],length = 0.05,lwd=lwd*2)
map(add=T)
ecospat.mess 53
points(data[fixed.sources.rows,2:1],pch=19,col="red")
text(data[fixed.sources.rows,2]+0.5,data[fixed.sources.rows,1]+0.5,data[fixed.sources.rows,3],
cex=1,col="red")
title(paste("total routes length : ",
round(obs[[2]],2)," Deg","\n","median dispersal rate : ",
round(obs[[3]],2)," Deg/year","\n","number of outcoming nodes : ",
obs[[4]]))
## End(Not run)
ecospat.mess MESS
Description
Calculate the MESS (i.e. extrapolation) as in Maxent.
Usage
ecospat.mess (proj, cal, w="default")
Arguments
proj A dataframe object with x, y and environmental variables, used as pro-
jection dataset.
cal A dataframe object with x, y and environmental variables, used as cali-
bration dataset.
wThe weight for each predictor (e.g. variables importance in SDM).
Details
Shows the variable that drives the multivariate environmental similarity surface (MESS)
value in each grid cell.
Value
MESS The mess as calculated in Maxent, i.e. the minimal extrapolation values.
MESSw The sum of negative MESS values corrected by the total number of pre-
dictors. If there are no negative values, MESSw is the mean MESS.
MESSneg The number of predictors on which there is extrapolation.
Author(s)
Blaise Petitpierre <bpetitpierre@gmail.com>. Modified by Daniel Scherrer <daniel.j.a.scherrer@gmail.com>
References
Elith, J., M. Kearney and S. Phillips. 2010. The art of modelling range-shifting species.
Methods in ecology and evolution,1, 330-342.
See Also
ecospat.plot.mess
54 ecospat.meva.table
Examples
x <- ecospat.testData[c(2,3,4:8)]
proj <- x[1:90,] #A projection dataset.
cal <- x[91:300,] #A calibration dataset
#Create a MESS object
mess.object <- ecospat.mess (proj, cal, w="default")
#Plot MESS
ecospat.plot.mess (mess.object, cex=1, pch=15)
ecospat.meva.table Model Evaluation For A Given Threshold Value
Description
Calculates values of a series of different evaluations metrics for a model and for a given
threshold value
Usage
ecospat.meva.table(Pred, Sp.occ, th)
Arguments
Pred A vector of predicted probabilities
Sp.occ A vector of binary observations of the species occurrence
th Threshold used to cut the probability to binary values
Value
A contingency table of observations and predicted probabilities of presence values, and a
list of evaluation metrics for the selected threshold.
Author(s)
Antoine Guisan <antoine.guisan@unil.ch>with contributions of Luigi Maiorano <luigi.maiorano@gmail.com>
References
Pearce, J. and S. Ferrier. 2000. Evaluating the predictive performance of habitat models
developed using logistic regression. Ecol. Model.,133, 225-245.
See Also
ecospat.max.kappa,ecospat.max.tss,ecospat.plot.tss,ecospat.cohen.kappa,ecospat.plot.kappa
Examples
Pred <- ecospat.testData$glm_Agrostis_capillaris
Sp.occ <- ecospat.testData$Agrostis_capillaris
meva <- ecospat.meva.table (Pred, Sp.occ, 0.39)
ecospat.migclim 55
ecospat.migclim Implementing Dispersal Into Species Distribution Models
Description
Enables the implementation of species-specific dispersal constraints into projections of
species distribution models under environmental change and/or landscape fragmentation
scenarios.
Usage
ecospat.migclim ()
Details
The MigClim model is a cellular automaton originally designed to implement dispersal con-
straints into projections of species distributions under environmental change and landscape
fragmentation scenarios.
Author(s)
Robin Engler <robin.engler@gmail.com>, Wim Hordijk <wim@WorldWideWanderings.net>
and Loic Pellissier <loic.pellissier@unifr.ch>
References
Engler, R., W. Hordijk and A. Guisan. 2012. The MIGCLIM R package – seamless inte-
gration of dispersal constraints into projections of species distribution models. Ecography,
35, 872-878.
Engler, R. and A. Guisan. 2009. MIGCLIM: predicting plant distribution and dispersal in
a changing climate. Diversity and Distributions,15, 590-601.
Engler, R., C.F. Randin, P. Vittoz, T. Czaka, M. Beniston, N.E. Zimmermann and A.
Guisan. 2009. Predicting future distributions of mountain plants under climate change:
does dispersal capacity matter? Ecography,32, 34-45.
Examples
## Not run:
ecospat.migclim()
### Some example data files can be downloaded from the following web page:
### http://www.unil.ch/ecospat/page89413.html
###
### Run the example as follows (set the current working directory to the
### folder where the example data files are located):
###
data(MigClim.testData)
### Run MigClim with a data frame type input.
n <- MigClim.migrate (iniDist=MigClim.testData[,1:3],
hsMap=MigClim.testData[,4:8], rcThreshold=500,
envChgSteps=5, dispSteps=5, dispKernel=c(1.0,0.4,0.16,0.06,0.03),
barrier=MigClim.testData[,9], barrierType="strong",
iniMatAge=1, propaguleProd=c(0.01,0.08,0.5,0.92),
lddFreq=0.1, lddMinDist=6, lddMaxDist=15,
56 ecospat.mpa
simulName="MigClimTest", replicateNb=1, overWrite=TRUE,
testMode=FALSE, fullOutput=FALSE, keepTempFiles=FALSE)
## End(Not run)
ecospat.mpa Minimal Predicted Area
Description
Calculate the minimal predicted area.
Usage
ecospat.mpa (Pred, Sp.occ.xy, perc)
Arguments
Pred Numeric or RasterLayer predicted suitabilities from a SDM prediction.
Sp.occ.xy xy-coordinates of the species (if Pred is a RasterLayer).
perc Percentage of Sp.occ.xy that should be encompassed by the binary map.
Details
The minimal predicted area (MPA) is the minimal surface obtained by considering all pixels
with predictions above a defined probability threshold (e.g. 0.7) that still encompasses 90
percent of the species‘ occurrences (Engler et al. 2004).
Value
Returns the minimal predicted area.
Author(s)
Frank Breiner <frank.breiner@wsl.ch>
References
Engler, R., A. Guisan and L. Rechsteiner. 2004. An improved approach for predicting
the distribution of rare and endangered species from occurrence and pseudo-absence data.
Journal of Applied Ecology,41, 263-274.
Examples
obs <- (ecospat.testData$glm_Saxifraga_oppositifolia
[which(ecospat.testData$Saxifraga_oppositifolia==1)])
ecospat.mpa(obs)
ecospat.mpa(obs,perc=1) ## 100 percent of the presences encompassed
ecospat.niche.dyn.index 57
ecospat.niche.dyn.index
Niche Expansion, Stability, and Unfilling
Description
Calculate niche expansion, stability and unfilling.
Usage
ecospat.niche.dyn.index (z1, z2, intersection=NA)
Arguments
z1 A gridclim object for the native distribution.
z2 A gridclim object for the invaded range.
intersection The quantile of the environmental density used to remove marginal cli-
mates. If intersection=NA, the analysis is performed on the whole envi-
ronmental extent (native and invaded). If intersection=0, the analysis
is performed at the intersection between native and invaded range. If
intersection=0.05, the analysis is performed at the intersection of the
5th quantile of both native and invaded environmental densities.
Details
If you encounter a problem during your analyses, please first read the FAQ section of ”Niche
overlap” in http://www.unil.ch/ecospat/home/menuguid/ecospat-resources/tools.html
Value
A list of dynamic indices: dynamic.index.w [expansion.index.w, stability.index.w, restric-
tion.index.w]
Author(s)
Blaise Petitpierre <bpetitpierre@gmail.com>
See Also
ecospat.grid.clim.dyn
58 ecospat.niche.dynIndexProjGeo
ecospat.niche.dynIndexProjGeo
Projection of niche dynamic indices to the Geography
Description
Creates a raster in geography with each pixel containing a niche dynamic index (stability,
expansion, or unfilling) extracted from 2 niches generated with ecospat.grid.clim.dyn.
Usage
ecospat.niche.dynIndexProjGeo(z1,z2,env,index)
Arguments
z1 Species 1 occurrence density grid created by ecospat.grid.clim.dyn.
z2 Species 2 occurrence density grid created by ecospat.grid.clim.dyn.
env A RasterStack or RasterBrick of environmental variables corresponding
to the background (glob in ecospat.grid.clim.dyn).
index ”stability”, ”unfilling” or ”expansion”
Details
extracts the niche dynamic index of objects created by ecospat.niche.dyn.index for each
point of the background (glob) using extract (package raster). The values are binded to
the geographic coordinates of env and a raster is then recreated using rasterFromXYZ
Value
raster of class RasterLayer
Author(s)
Olivier Broennimann <olivier.broennimann@unil.ch>
References
Broennimann, O., M.C. Fitzpatrick, P.B. Pearman, B. Petitpierre, L. Pellissier, N.G. Yoc-
coz, W. Thuiller, M.J. Fortin, C. Randin, N.E. Zimmermann, C.H. Graham and A. Guisan.
2012. Measuring ecological niche overlap from occurrence and spatial environmental data.
Global Ecology and Biogeography,21:481-497.
Petitpierre, B., C. Kueffer, O. Broennimann, C. Randin, C. Daehler and A. Guisan. 2012.
Climatic niche shifts are rare among terrestrial plant invaders. Science,335:1344-1348.
See Also
ecospat.plot.niche.dyn,ecospat.niche.dyn.index,ecospat.niche.zProjGeo
ecospat.niche.equivalency.test 59
Examples
## Not run:
library(raster)
spp <- ecospat.testNiche
xy.sp1<-subset(spp,species=="sp1")[2:3] #Bromus_erectus
xy.sp2<-subset(spp,species=="sp3")[2:3] #Daucus_carota
?ecospat.testEnvRaster
load(system.file("extdata", "ecospat.testEnvRaster.Rdata", package="ecospat"))
env.sp1<-extract(env,xy.sp1)
env.sp2<-extract(env,xy.sp2)
env.bkg<-na.exclude(values(env))
#################################### PCA-ENVIRONMENT ##################################
pca.cal <- dudi.pca(env.bkg, center = TRUE, scale = TRUE, scannf = FALSE, nf = 2)
# predict the scores on the axes
scores.bkg <- pca.cal$li #scores for background climate
scores.sp1 <- suprow(pca.cal,env.sp1)$lisup #scores for sp1
scores.sp2 <- suprow(pca.cal,env.sp2)$lisup #scores for sp2
# calculation of occurence density (niche z)
z1 <- ecospat.grid.clim.dyn(scores.bkg, scores.bkg, scores.sp1,R=100)
z2 <- ecospat.grid.clim.dyn(scores.bkg, scores.bkg, scores.sp2,R=100)
plot(z1$z.uncor)
points(scores.sp1)
plot(z2$z.uncor)
points(scores.sp2)
ecospat.niche.overlap(z1,z2 ,cor=T)
#################################### stability S in space ##################################
geozS<-ecospat.niche.dynIndexProjGeo(z1,z2,env,index="stability")
plot(geozS,main="Stability")
points(xy.sp1,col="red")
points(xy.sp2,col="blue")
## End(Not run)
ecospat.niche.equivalency.test
Niche Equivalency Test
60 ecospat.niche.equivalency.test
Description
Run a niche equivalency test (see Warren et al 2008) based on two species occurrence density
grids.
Usage
ecospat.niche.equivalency.test (z1, z2, rep, alternative, ncores = 1)
Arguments
z1 Species 1 occurrence density grid created by ecospat.grid.clim.
z2 Species 2 occurrence density grid created by ecospat.grid.clim.
rep The number of replications to perform.
alternative To indicate the type of test to be performed. It could be greater or lower.
ncores The number of cores used for parallelisation.
Details
Compares the observed niche overlap between z1 and z2 to overlaps between random niches
z1.sim and z2.sim, which are built from random reallocations of occurences of z1 and z2.
alternative specifies if you want to test for niche conservatism (alternative = ”greater”,
i.e. the niche overlap is more equivalent/similar than random) or for niche divergence
(alternative = ”lower”, i.e. the niche overlap is less equivalent/similar than random).
If you encounter a problem during your analyses, please first read the FAQ section of ”Niche
overlap” in http://www.unil.ch/ecospat/home/menuguid/ecospat-resources/tools.html
The arguments ncores allows choosing the number of cores used to parallelize the computa-
tion. The default value is 1. On multicore computers, the optimal would be ncores = detectCores() - 1.
Value
a list with obs = observed overlaps, sim = simulated overlaps, p.D = p-value of the test
on D, p.I = p-value of the test on I.
Author(s)
Olivier Broennimann <olivier.broennimann@unil.ch>with contributions of Blaise Petit-
pierre <bpetitpierre@gmail.com>
References
Broennimann, O., M.C. Fitzpatrick, P.B. Pearman,B. Petitpierre, L. Pellissier, N.G. Yoccoz,
W. Thuiller, M.J. Fortin, C. Randin, N.E. Zimmermann, C.H. Graham and A. Guisan.
2012. Measuring ecological niche overlap from occurrence and spatial environmental data.
Global Ecology and Biogeography,21, 481-497.
Warren, D.L., R.E. Glor and M. Turelli. 2008. Environmental niche equivalency versus
conservatism: quantitative approaches to niche evolution. Evolution,62, 2868-2883.
See Also
ecospat.grid.clim.dyn,ecospat.niche.similarity.test
ecospat.niche.overlap 61
ecospat.niche.overlap Calculate Niche Overlap
Description
Calculate the overlap metrics D and I based on two species occurrence density grids z1 and
z2 created by ecospat.grid.clim.
Usage
ecospat.niche.overlap (z1, z2, cor)
Arguments
z1 Species 1 occurrence density grid created by ecospat.grid.clim.
z2 Species 2 occurrence density grid created by ecospat.grid.clim.
cor Correct the occurrence densities of each species by the prevalence of the
environments in their range (TRUE = yes, FALSE = no).
Details
if cor=FALSE, the z uncor objects created by ecospat.grid.clim are used to calculate the
overlap. if cor=TRUE, the z cor objects are used.
If you encounter a problem during your analyses, please first read the FAQ section of ”Niche
overlap” in http://www.unil.ch/ecospat/home/menuguid/ecospat-resources/tools.html
Value
Overlap values D and I. D is Schoener’s overlap metric (Schoener 1970). I is a modified
Hellinger metric(Warren et al. 2008)
Author(s)
Olivier Broennimann <olivier.broennimann@unil.ch>
References
Broennimann, O., M.C. Fitzpatrick, P.B. Pearman, B. Petitpierre, L. Pellissier, N.G. Yoc-
coz, W. Thuiller, M.J. Fortin, C. Randin, N.E. Zimmermann, C.H. Graham and A. Guisan.
2012. Measuring ecological niche overlap from occurrence and spatial environmental data.
Global Ecology and Biogeography,21, 481-497.
Schoener, T.W. 1968. Anolis lizards of Bimini: resource partitioning in a complex fauna.
Ecology,49, 704-726.
Warren, D.L., R.E. Glor and M. Turelli. 2008. Environmental niche equivalency versus
conservatism: quantitative approaches to niche evolution. Evolution,62, 2868-2883.
See Also
ecospat.grid.clim.dyn
62 ecospat.niche.similarity.test
ecospat.niche.similarity.test
Niche Similarity Test
Description
Run a niche similarity test (see Warren et al 2008) based on two species occurrence density
grids.
Usage
ecospat.niche.similarity.test (z1, z2, rep, alternative = "greater",
rand.type = 1, ncores= 1)
Arguments
z1 Species 1 occurrence density grid created by ecospat.grid.clim.
z2 Species 2 occurrence density grid created by ecospat.grid.clim.
rep The number of replications to perform.
alternative To indicate the type of test to be performed. It could be greater or lower.
rand.type Type of randomization on the density grids (1 or 2).
ncores The number of cores used for parallelisation.
Details
Compares the observed niche overlap between z1 and z2 to overlaps between z1 and random
niches (z2.sim) as available in the range of z2 (z2 Z). z2.sim has the same pattern as z2
but the center is randomly translatated in the availabe z2 Z space and weighted by z2 Z
densities. If rand.type = 1, both z1 and z2 are randomly shifted, if rand.type =2, only z2
is randomly shifted.
alternative specifies if you want to test for niche conservatism (alternative = ”greater”,
i.e. the niche overlap is more equivalent/similar than random) or for niche divergence
(alternative = ”lower”, i.e. the niche overlap is less equivalent/similar than random).
If you encounter a problem during your analyses, please first read the FAQ section of ”Niche
overlap” in http://www.unil.ch/ecospat/home/menuguid/ecospat-resources/tools.html
The arguments ncores allows choosing the number of cores used to parallelize the computa-
tion. The default value is 1. On multicore computers, the optimal would be ncores = detectCores() - 1.
Value
a list with obs = observed overlaps, sim = simulated overlaps, p.D = p-value of the test
on D, p.I = p-value of the test on I.
Author(s)
Olivier Broennimann <olivier.broennimann@unil.ch>with contributions of Blaise Petit-
pierre <bpetitpierre@gmail.com>
ecospat.niche.zProjGeo 63
References
Broennimann, O., M.C. Fitzpatrick, P.B. Pearman, B. Petitpierre, L. Pellissier, N.G. Yoc-
coz, W. Thuiller, M.J. Fortin, C. Randin, N.E. Zimmermann, C.H. Graham and A. Guisan.
2012. Measuring ecological niche overlap from occurrence and spatial environmental data.
Global Ecology and Biogeography,21, 481-497.
Warren, D.L., R.E. Glor and M. Turelli. 2008. Environmental niche equivalency versus
conservatism: quantitative approaches to niche evolution. Evolution,62, 2868-2883.
See Also
ecospat.grid.clim.dyn,ecospat.niche.equivalency.test
ecospat.niche.zProjGeo
Projection of Occurrence Densities to the Geography
Description
Creates a raster in geography with each pixel containing the occurrence densities extracted
from a z object generated with ecospat.grid.clim.dyn.
Usage
ecospat.niche.zProjGeo(z1,env,cor)
Arguments
z1 Species 1 occurrence density grid created by ecospat.grid.clim.dyn.
env A RasterStack or RasterBrick of environmental variables corresponding
to the background (glob in ecospat.grid.clim.dyn).
cor FALSE by default. If TRUE corrects the occurrence densities of each
species by the prevalence of the environments in their range
Details
extracts the occurrence density of z objects created by ecospat.grid.clim.dyn for each
point of the background (glob) using extract (package raster). The values are binded to
the geographic coordinates of env and a raster is then recreated using rasterFromXYZ
Value
raster of class RasterLayer
Author(s)
Olivier Broennimann <olivier.broennimann@unil.ch>
64 ecospat.niche.zProjGeo
References
Broennimann, O., M.C. Fitzpatrick, P.B. Pearman, B. Petitpierre, L. Pellissier, N.G. Yoc-
coz, W. Thuiller, M.J. Fortin, C. Randin, N.E. Zimmermann, C.H. Graham and A. Guisan.
2012. Measuring ecological niche overlap from occurrence and spatial environmental data.
Global Ecology and Biogeography,21:481-497.
Petitpierre, B., C. Kueffer, O. Broennimann, C. Randin, C. Daehler and A. Guisan. 2012.
Climatic niche shifts are rare among terrestrial plant invaders. Science,335:1344-1348.
See Also
ecospat.plot.niche.dyn,ecospat.niche.dynIndexProjGeo
Examples
## Not run:
library(raster)
spp <- ecospat.testNiche
xy.sp1<-subset(spp,species=="sp1")[2:3] #Bromus_erectus
load(system.file("extdata", "ecospat.testEnvRaster.Rdata", package="ecospat"))
#?ecospat.testEnvRaster
env.sp1<-extract(env,xy.sp1)
env.bkg<-na.exclude(values(env))
#################################### PCA-ENVIRONMENT ##################################
pca.cal <- dudi.pca(env.bkg, center = TRUE, scale = TRUE, scannf = FALSE, nf = 2)
# predict the scores on the axes
scores.bkg <- pca.cal$li #scores for background climate
scores.sp1 <- suprow(pca.cal,env.sp1)$lisup #scores for sp1
# calculation of occurence density (niche z)
z1 <- ecospat.grid.clim.dyn(scores.bkg, scores.bkg, scores.sp1,R=100)
plot(z1$z.uncor)
points(scores.sp1)
#################################### occurrence density in space ##################################
# sp1
geoz1<-ecospat.niche.zProjGeo(z1,env)
plot(geoz1,main="z1")
points(xy.sp1)
## End(Not run)
ecospat.npred 65
ecospat.npred Number Of Predictors
Description
Calculate the maximum number of predictors to include in the model with a desired corre-
lation between predictors.
Usage
ecospat.npred (x, th)
Arguments
xCorrelation matrix of the predictors.
th Desired threshold of correlation between predictors.
Value
Returns the number of predictors to use.
Author(s)
Blaise Petitpierre <bpetitpierre@gmail.com>
Examples
colvar <- ecospat.testData[c(4:8)]
x <- cor(colvar, method="pearson")
ecospat.npred (x, th=0.75)
ecospat.occ.desaggregation
Species Occurrences Desaggregation
Description
Remove species occurrences in a dataframe which are closer to each other than a specified
distance threshold.
Usage
ecospat.occ.desaggregation (xy, min.dist, by)
Arguments
xy A dataframe with xy-coordinates (x-column must be named ’x’ and y-
column ’y’)
min.dist The minimun distance between points in the sub-dataframe.
by Grouping element in the dataframe (e.g. species, NULL)
66 ecospat.occupied.patch
Details
This function will desaggregate the original number of occurrences, according to a specified
distance.
Value
A subset of the initial dataframe. The number of points is printed as ”initial”, ”kept” and
”out”.
Author(s)
Frank Breiner <frank.breiner@unil.ch>
with contributions of Olivier Broennimann <olivier.broennimann@unil.ch>
Examples
## Not run:
spp <- ecospat.testNiche
colnames(spp)[2:3] <- c('x','y')
sp1 <- spp[1:32,2:3]
occ.sp1 <- ecospat.occ.desaggregation(xy=sp1, min.dist=500, by=NULL)
occ.all.sp <- ecospat.occ.desaggregation(xy=spp, min.dist=500, by='Spp')
## End(Not run)
ecospat.occupied.patch
Extract occupied patches of a species in geographic space.)
Description
This function determines the occupied patch of a species using standard IUCN criteria
(AOO, EOO) or predictive binary maps from Species Distribution Models.
Usage
ecospat.occupied.patch (bin.map, Sp.occ.xy, buffer = 0)
Arguments
bin.map Binary map (single layer or raster stack) from a Species Distribution
Model.
Sp.occ.xy xy-coordinates of the species presence.
buffer numeric. Calculate occupied patch models from the binary map using a
buffer (predicted area occupied by the species or within a buffer around
the species, for details see ?extract).
ecospat.occupied.patch 67
Details
Predictive maps derived from SDMs inform about the potential distribution (or habitat
suitability) of a species. Often it is useful to get information about the area of the potential
distribution which is occupied by a species, e.g. for Red List assessments.
Value
a RasterLayer with value 1.
Author(s)
Frank Breiner <frank.breiner@wsl.ch>
References
IUCN Standards and Petitions Subcommittee. 2016. Guidelines for Using the IUCN Red
List Categories and Criteria. Version 12. Prepared by the Standards and Petitions Subcom-
mittee. Downloadable from http://www.iucnredlist.org/documents/RedListGuidelines.pdf
See Also
ecospat.rangesize,ecospat.mpa,ecospat.binary.model
Examples
## Not run:
library(dismo)
library(dismo)
# only run if the maxent.jar file is available, in the right folder
jar <- paste(system.file(package="dismo"), "/java/maxent.jar", sep='')
# checking if maxent can be run (normally not part of your script)
file.exists(jar)
require(rJava))
# get predictor variables
fnames <- list.files(path=paste(system.file(package="dismo"), '/ex', sep=''),
pattern='grd', full.names=TRUE )
predictors <- stack(fnames)
#plot(predictors)
# file with presence points
occurence <- paste(system.file(package="dismo"), '/ex/bradypus.csv', sep='')
occ <- read.table(occurence, header=TRUE, sep=',')[,-1]
colnames(occ) <- c("x","y")
occ <- ecospat.occ.desaggregation(occ,min.dist=1)
# fit model, biome is a categorical variable
me <- maxent(predictors, occ, factors='biome')
68 ecospat.permut.glm
# predict to entire dataset
pred <- predict(me, predictors)
plot(pred)
points(occ)
# use MPA to convert suitability to binary map
mpa.cutoff <- ecospat.mpa(pred,occ)
pred.bin.mpa <- ecospat.binary.model(pred,mpa.cutoff)
names(pred.bin.mpa) <- "me.mpa"
pred.bin.arbitrary <- ecospat.binary.model(pred,0.5)
names(pred.bin.arbitrary) <- "me.arbitrary"
mpa.ocp <- ecospat.occupied.patch(pred.bin.mpa,occ)
arbitrary.ocp <- ecospat.occupied.patch(pred.bin.arbitrary,occ)
par(mfrow=c(1,2))
plot(mpa.ocp) ## occupied patches: green area
points(occ,col="red",cex=0.5,pch=19)
plot(arbitrary.ocp)
points(occ,col="red",cex=0.5,pch=19)
## with buffer:
mpa.ocp <- ecospat.occupied.patch(pred.bin.mpa,occ, buffer=500000)
arbitrary.ocp <- ecospat.occupied.patch(pred.bin.arbitrary,occ, buffer=500000)
plot(mpa.ocp) ## occupied patches: green area
points(occ,col="red",cex=0.5,pch=19)
plot(arbitrary.ocp)
points(occ,col="red",cex=0.5,pch=19)
## End(Not run)
ecospat.permut.glm GLM Permutation Function
Description
A permutation function to get p-values on GLM coefficients and deviance.
Usage
ecospat.permut.glm (glm.obj, nperm)
Arguments
glm.obj Any calibrated GLM or GAM object with a binomial error distribution.
nperm The number of permutations in the randomization process.
ecospat.plot.contrib 69
Details
Rows of the response variable are permuted and a new GLM is calibrated as well as deviance,
adjusted deviance and coefficients are calculated. These random parameters are compared
to the true parameters in order to derive p-values.
Value
Return p-values that are how the true parameters of the original model deviate from the
disribution of the random parameters. A p-value of zero means that the true parameter is
completely outside the random distribution.
Author(s)
Christophe Randin <christophe.randin@unibas.ch>, Antoine Guisan <antoine.guisan@unil.ch>
and Trevor Hastie
References
Hastie, T., R. Tibshirani and J. Friedman. 2001. Elements of Statistical Learning; Data
Mining, Inference, and Prediction, Springer-Verlag, New York.
Legendre, P. and L. Legendre. 1998. Numerical Ecology, 2nd English edition. Elsevier
Science BV, Amsterdam.
Examples
## Not run:
ecospat.permut.glm (get ("glm.Achillea_atrata", envir=ecospat.env), 1000)
## End(Not run)
ecospat.plot.contrib Plot Variables Contribution
Description
Plot the contribution of the initial variables to the analysis (i.e. correlation circle). Typically
these are the eigen vectors and eigen values in ordinations.
Usage
ecospat.plot.contrib (contrib, eigen)
Arguments
contrib A dataframe of the contribution of each original variable on each axis of
the analysis, i.e. the eigen vectors.
eigen A vector of the importance of the axes in the ordination, i.e. a vector of
eigen values.
70 ecospat.plot.kappa
Details
Requires ade4 library. If using princomp , use loadings and sdev of the princomp object.
if using dudi.pca, use li and eig of the dudi.pca object.
Author(s)
Olivier Broennimann <olivier.broennimann@unil.ch>
References
Broennimann, O., M.C. Fitzpatrick, P.B. Pearman, B. Petitpierre, L. Pellissier, N.G. Yoc-
coz, W. Thuiller, M.J. Fortin, C. Randin, N.E. Zimmermann, C.H. Graham and A. Guisan.
2012. Measuring ecological niche overlap from occurrence and spatial environmental data.
Global Ecology and Biogeography,21, 481-497.
See Also
ecospat.plot.niche.dyn,ecospat.plot.overlap.test,ecospat.niche.similarity.test,princomp
ecospat.plot.kappa Plot Kappa
Description
Plots the values for Cohen’s Kappa along different thresholds.
Usage
ecospat.plot.kappa(Pred, Sp.occ)
Arguments
Pred A vector of predicted probabilities
Sp.occ A vector of binary observations of the species occurrence
Value
A plot of the Cohen’s Kappa values along different thresholds.
Author(s)
Luigi Maiorano <luigi.maiorano@gmail.com>with contributions of Valeria Di Cola <valeria.dicola@unil.ch>.
References
Liu, C., P.M. Berry, T.P. Dawson, and R.G. Pearson. 2005. Selecting thresholds of occur-
rence in the prediction of species distributions. Ecography,28, 385-393.
Landis, J.R. and G.G. Koch. 1977. The measurement of observer agreement for categorical
data. biometrics,33,159-174.
See Also
ecospat.meva.table,ecospat.max.tss,ecospat.plot.tss,ecospat.cohen.kappa,ecospat.max.kappa
ecospat.plot.mess 71
Examples
Pred <- ecospat.testData$glm_Agrostis_capillaris
Sp.occ <- ecospat.testData$Agrostis_capillaris
ecospat.plot.kappa(Pred, Sp.occ)
ecospat.plot.mess Plot MESS
Description
Plot the MESS extrapolation index onto the geographical space.
Usage
ecospat.plot.mess (mess.object, cex=1, pch=15)
Arguments
mess.object A dataframe as returned by the ecospat.mess function.
cex Specify the size of the symbol.
pch Specify the point symbols.
Value
Returns a plot of the the MESS extrapolation index onto the geographical space.
Author(s)
Blaise Petitpierre <bpetitpierre@gmail.com>
References
Elith, J., M. Kearney and S. Phillips. 2010. The art of modelling range-shifting species.
Methods in ecology and evolution,1, 330-342.
See Also
ecospat.mess
Examples
## Not run:
x <- ecospat.testData[c(2,3,4:8)]
proj <- x[1:90,] #A projection dataset.
cal <- x[91:300,] #A calibration dataset
#Create a MESS object
mess.object <- ecospat.mess (proj, cal, w="default")
#Plot MESS
ecospat.plot.mess (mess.object, cex=1, pch=15)
## End(Not run)
72 ecospat.plot.niche
ecospat.plot.niche Plot Niche
Description
Plot a niche z created by ecospat.grid.clim.dyn.
Usage
ecospat.plot.niche (z, title, name.axis1, name.axis2, cor=FALSE)
Arguments
zA gridclim object for the species distribution created by ecospat.grid.clim.dyn.
title A title for the plot.
name.axis1 A label for the first axis.
name.axis2 A label for the second axis.
cor Correct the occurrence densities of the species by the prevalence of the
environments in its range (TRUE = yes, FALSE = no).
Details
if z is bivariate, a bivariate plot of the niche of the species. if z is univariate, a histogram
of the niche of the species
Author(s)
Olivier Broennimann <olivier.broennimann@unil.ch>
References
Broennimann, O., M.C. Fitzpatrick, P.B. Pearman, B. Petitpierre, L. Pellissier, N.G. Yoc-
coz, W. Thuiller, M.J. Fortin, C. Randin, N.E. Zimmermann, C.H. Graham and A. Guisan.
2012. Measuring ecological niche overlap from occurrence and spatial environmental data.
Global Ecology and Biogeography,21, 481-497.
See Also
ecospat.grid.clim.dyn
ecospat.plot.niche.dyn 73
ecospat.plot.niche.dyn
Niche Categories and Species Density
Description
Plot niche categories and species density created by ecospat.grid.clim.dyn.
Usage
ecospat.plot.niche.dyn (z1, z2, quant, title,
name.axis1, name.axis2, interest, colz1, colz2,colinter, colZ1, colZ2)
Arguments
z1 A gridclim object for the native distribution.
z2 A gridclim object for the invaded range.
quant The quantile of the environmental density used to delimit marginal cli-
mates.
title The title of the plot.
name.axis1 A label for the first axis.
name.axis2 A label for the second axis
interest Choose which density to plot: if interest=1, plot native density, if interest=2,
plot invasive density.
colz1 The color used to depict unfilling area.
colz2 The color used to depict expansion area.
colinter The color used to depict overlap area.
colZ1 The color used to delimit the native extent.
colZ2 The color used to delimit the invaded extent.
Author(s)
Blaise Petitpierre <bpetitpierre@gmail.com>
ecospat.plot.overlap.test
Plot Overlap Test
Description
Plot a histogram of observed and randomly simulated overlaps, with p-values of equivalency
or similarity tests.
Usage
ecospat.plot.overlap.test (x, type, title)
74 ecospat.plot.tss
Arguments
xObject created by
ecospat.niche.similarity.test or ecospat.niche.equivalency.test.
type Must be either “D” or “I”.
title The title for the plot.
Author(s)
Olivier Broennimann <olivier.broennimann@unil.ch>
References
Broennimann, O., M.C. Fitzpatrick, P.B. Pearman, B. Petitpierre, L. Pellissier, N.G. Yoc-
coz, W. Thuiller, M.J. Fortin, C. Randin, N.E. Zimmermann, C.H. Graham and A. Guisan.
2012. Measuring ecological niche overlap from occurrence and spatial environmental data.
Global Ecology and Biogeography,21, 481-497.
See Also
ecospat.niche.similarity.test,ecospat.niche.equivalency.test
ecospat.plot.tss Plot True skill statistic (TSS)
Description
Plots the values for True skill statistic (TSS) along different thresholds.
Usage
ecospat.plot.tss(Pred, Sp.occ)
Arguments
Pred A vector of predicted probabilities
Sp.occ A vector of binary observations of the species occurrence
Value
A plot of the TSS values along different thresholds.
Author(s)
Luigi Maiorano <luigi.maiorano@gmail.com>
References
Liu, C., P.M. Berry, T.P. Dawson, and R.G. Pearson. 2005. Selecting thresholds of occur-
rence in the prediction of species distributions. Ecography,28, 385-393.
Liu, C., M. White and G. Newell. 2013. Selecting thresholds for the prediction of species
occurrence with presence-only data. Journal of Biogeography,40, 778-789.
ecospat.rand.pseudoabsences 75
See Also
ecospat.meva.table,ecospat.max.tss,ecospat.plot.kappa,ecospat.cohen.kappa,ecospat.max.kappa
Examples
Pred <- ecospat.testData$glm_Agrostis_capillaris
Sp.occ <- ecospat.testData$Agrostis_capillaris
ecospat.plot.tss(Pred, Sp.occ)
ecospat.rand.pseudoabsences
Sample Pseudo-Absences
Description
Randomly sample pseudoabsences from an environmental dataframe covering the study
area.
Usage
ecospat.rand.pseudoabsences (nbabsences, glob, colxyglob, colvar="all",
presence, colxypresence, mindist)
Arguments
nbabsences The number of pseudoabsences desired.
glob A two-column dataframe (or a vector) of the environmental values (in
column) for background pixels of the whole study area (in row).
colxyglob The range of columns for x and y in glob.
colvar The range of columns for the environmental variables in glob. colvar=”all”
keeps all the variables in glob in the final dataframe. colvar=NULL keeps
only x and y.
presence A presence-absence dataframe for each species (columns) in each location
or grid cell (rows).
colxypresence The range of columns for x and y in presence.
mindist The minimum distance from presences within wich pseudoabsences should
not be drawn (buffer distance around presences).
Value
A dataframe of random absences.
Author(s)
Olivier Broennimann <olivier.broennimann@unil.ch>
Examples
glob <- ecospat.testData[2:8]
presence <- ecospat.testData[c(2:3,9)]
presence <- presence[presence[,3]==1,1:2]
ecospat.rand.pseudoabsences (nbabsences=10, glob=glob, colxyglob=1:2, colvar = "all",
presence= presence, colxypresence=1:2, mindist=20)
76 ecospat.rangesize
ecospat.rangesize Quantification of the range size of a species using habitat suit-
ability maps and IUCN criteria)
Description
This function quantifies the range size of a species using standard IUCN criteria (Area of
Occupancy AOO, Extent of Occurence EOO) or binary maps derived from Species Distri-
bution Models.
Usage
ecospat.rangesize (bin.map, ocp, buffer, eoo.around.model, eoo.around.modelocp,
xy, EOO, Model.within.eoo, AOO, resol, AOO.circles, d, lonlat, return.obj,
save.obj, save.rangesize, directory)
ecospat.rangesize (bin.map = NULL,
ocp = TRUE,
buffer = 0,
eoo.around.model = TRUE,
eoo.around.modelocp = FALSE,
xy = NULL,
EOO = TRUE,
Model.within.eoo = TRUE,
AOO = TRUE,
resol = c(2000, 2000),
AOO.circles = FALSE,
d = sqrt((2000 * 2)/pi),
lonlat = FALSE,
return.obj = TRUE,
save.obj = FALSE,
save.rangesize = FALSE,
directory = getwd())
Arguments
bin.map Binary map (single layer or raster stack) from a Species Distribution
Model.
ocp logical. Calculate occupied patch models from the binary map (predicted
area occupied by the species)
buffer numeric. Calculate occupied patch models from the binary map using a
buffer (predicted area occupied by the species or within a buffer around
the species, for details see ?extract).
eoo.around.model
logical. The EOO around all positive predicted values from the binary
map.
ecospat.rangesize 77
eoo.around.modelocp
logical. EOO around all positive predicted values of occupied patches.
xy xy-coordinates of the species presence
EOO logical. Extent of Occurrence. Convex Polygon around occurrences.
Model.within.eoo
logical. Area predicted as suitable by the model within EOO.
AOO logical. Area of Occupancy ddervied by the occurrences.
resol Resolution of the grid frame at which AOO should be calculated.
AOO.circles logical. AOO calculated by circles around the occurrences instead of using
a grid frame.
dRadius of the AOO.circles around the occurrences.
lonlat Are these longitude/latidue coordinates? (Default = FALSE).
return.obj logical. should the objects created to estimate range size be returned
(rasterfiles and spatial polygons). Default: TRUE
save.obj logical. should objects be saved on hard drive?
save.rangesize logical. should range size estimations be saved on hard drive .
directory directory in which objects should be saved (Default = getwd())
Details
The range size of a species is important for many conservation purposes, e.g. to assess the
status of threat for IUCN Red Lists. This function quantifies the range size using different
IUCN measures, i.e. the Area Of Occupancy (AOO), the Extent Of Occurrence (EOO)
and from binary maps derived from Species Distribution Models (SDMs). Different ways
to extract range size from SDMs are available, e.g. using occupied patches, the suitable
habitat within EOO or a convex hull around the suitable habitat.
Value
A list with the values of range size quantification and the stored objects used for quantifi-
cation (of class RasterLayers, ahull, ConvexHull).
Author(s)
Frank Breiner <frank.breiner@wsl.ch>
References
IUCN. 2012. IUCN Red List Categories and Criteria: Version 3.1. Second edition. Gland,
Switzerland and Cambridge, UK: IUCN. iv + 32pp.
IUCN Standards and Petitions Subcommittee. 2016. Guidelines for Using the IUCN Red
List Categories and Criteria. Version 12. Prepared by the Standards and Petitions Subcom-
mittee. Downloadable from http://www.iucnredlist.org/documents/RedListGuidelines.pdf
Pateiro-Lopez, B., and A. Rodriguez-Casal. 2010. Generalizing the Convex Hull of a
Sample: The R Package alphahull. Journal of Statistical software,34, 1-28.
See Also
ecospat.occupied.patch,ecospat.mpa,ecospat.binary.model
78 ecospat.rangesize
Examples
library(dismo)
# get predictor variables
fnames <- list.files(path=paste(system.file(package="dismo"), '/ex', sep=''),
pattern='grd', full.names=TRUE )
predictors <- stack(fnames)
#plot(predictors)
# file with presence points
occurence <- paste(system.file(package="dismo"), '/ex/bradypus.csv', sep='')
occ <- read.table(occurence, header=TRUE, sep=',')[,-1]
colnames(occ) <- c("x","y")
occ <- ecospat.occ.desaggregation(occ,min.dist=1)
# fit a domain model, biome is a categorical variable
do <- domain(predictors, occ, factors='biome')
# predict to entire dataset
pred <- predict(do, predictors)
plot(pred)
points(occ)
# use MPA to convert suitability to binary map
mpa.cutoff <- ecospat.mpa(pred,occ)
# use Boyce index to convert suitability to binary map
boyce <- ecospat.boyce(pred, occ)
### use the boyce index to find a threshold
pred.bin.arbitrary <- ecospat.binary.model(pred,0.5)
pred.bin.mpa <- ecospat.binary.model(pred,mpa.cutoff)
names(pred.bin.mpa) <- "me.mpa"
pred.bin.arbitrary <- ecospat.binary.model(pred,0.5)
names(pred.bin.arbitrary) <- "me.arbitrary"
rangesize <- ecospat.rangesize(stack(pred.bin.mpa,pred.bin.arbitrary),
xy=occ,
resol=c(1,1),
eoo.around.modelocp =TRUE,
AOO.circles = TRUE,
d=200000,
lonlat =TRUE)
## Range size quantification
rangesize$RangeSize
names(rangesize$RangeObjects)
par(mfrow=c(1,3))
ecospat.rcls.grd 79
plot(ecospat.binary.model(pred,0),legend=FALSE, main="IUCN criteria")
## IUCN criteria & derivates
# plot AOO
plot(rangesize$RangeObjects$AOO,add=TRUE, col="red",legend=FALSE)
# plot EOO
plot(rangesize$RangeObjects$EOO@polygons,add=TRUE, border="red", lwd=2)
# plot circles around occurrences
plot(rangesize$RangeObjects$AOO.circle@polygons,add=TRUE,border="blue")
for(i in 1:2){
## plot the occupied patches of the model
plot(rangesize$RangeObjects$models.ocp[[i]],col=c("grey","blue","darkgreen"),
main=names(rangesize$RangeObjects$models.ocp[[i]]),legend=FALSE)
points(occ,col="red",cex=0.5,pch=19)
## plot EOO around model
plot(rangesize$RangeObjects$eoo.around.model[[i]]@polygons,add=TRUE,border="blue",lwd=2)
## plot EOO around occupied patches
plot(rangesize$RangeObjects$eoo.around.mo.ocp[[i]]@polygons,add=TRUE,border="darkgreen",
lwd=2)
## plot the modeled area within EOO
#plot(rangesize$RangeObjects$model.within.eoo[[i]],col=c("grey","blue","darkgreen"),legend=FALSE)
#points(occ,col="red",cex=0.5,pch=19)
#plot(rangesize$RangeObjects$EOO@polygons,add=TRUE, border="red", lwd=2)
}
### Alpha-hulls are not included in the function yet because of Licence limitations.
### However, alpha-hulls can easily be included manually (see also the help file of
### the alpha hull package):
# require(alphahull)
# alpha = 2 # alpha value of 2 recommended by IUCN
# del<-delvor(occ)
# dv<-del$mesh
# mn <- mean(sqrt(abs(del$mesh[,3]-del$mesh[,5])ˆ2+abs(del$mesh[,4]-del$mesh[,6])ˆ2))*alpha
# alpha.hull<-ahull(del,alpha=mn)
# #Size of alpha-hulls
# areaahull(h)
# plot alphahulls
# plot(rangesize$RangeObjects$models.ocp[[i]],col=c("grey","blue","darkgreen"),
# main=names(rangesize$RangeObjects$models.ocp[[i]]),legend=FALSE)
# plot(alpha.hull,add=TRUE,lwd=1)
ecospat.rcls.grd Reclassifying grids function
Description
Function for reclassifying grid files to get a combined statification from more than one grid
80 ecospat.recstrat prop
Usage
ecospat.rcls.grd(in_grid,no.classes)
Arguments
in grid The grid to be reclassified.
no.classes The number of desired new classes.
Details
This function reclassifies the input grid into a number of new classes that the user defines.
The boundaries of each class are decided automatically by splitting the range of values of
the input grid into the user defined number of classes.
Value
Returns a reclassified Raster object
Author(s)
Achilleas Psomas <achilleas.psomas@wsl.ch>and Niklaus E. Zimmermann <niklaus.zimmermann@wsl.ch>
Examples
## Not run:
bio3<- raster(system.file("external/bioclim/current/bio3.grd",package="biomod2"))
bio12<- raster(system.file("external/bioclim/current/bio12.grd",package="biomod2"))
B3.rcl<-ecospat.rcls.grd(bio3,9)
B12.rcl<-ecospat.rcls.grd(bio12,9)
B3B12.comb <- B12.rcl+B3.rcl*10
# Plotting a histogram of the classes
hist(B3B12.comb,breaks=100,col=heat.colors(88))
# Plotting the new RasterLayer (9x9 classes)
plot(B3B12.comb,col=rev(rainbow(88)),main="Stratified map")
## End(Not run)
ecospat.recstrat prop Random Ecologically Stratified Sampling of propotional numbers
Description
This function randomly collects a user-defined total number of samples from the stratifica-
tion layer.
Usage
ecospat.recstrat_prop(in_grid, sample_no)
ecospat.recstrat regl 81
Arguments
in grid The stratification grid to be sampled.
sample no The total number of pixels to be sampled.
Details
The number of samples per class are determined proportional to the abundance of each
class. The number of classes in the stratification layer are determined automatically from
the integer input map. If the proportion of samples for a certain class is below one then no
samples are collected for this class.
Value
Returns a dataframe with the selected sampling locations their coordinates and the strata
they belong in.
Author(s)
Achilleas Psomas <achilleas.psomas@wsl.ch>and Niklaus E. Zimmermann <niklaus.zimmermann@wsl.ch>
See Also
ecospat.recstrat regl ecospat.rcls.grd
Examples
## Not run:
bio3<- raster(system.file("external/bioclim/current/bio3.grd",package="biomod2"))
bio12<- raster(system.file("external/bioclim/current/bio12.grd",package="biomod2"))
B3.rcl<-ecospat.rcls.grd(bio3,9)
B12.rcl<-ecospat.rcls.grd(bio12,9)
B3B12.comb <- B12.rcl+B3.rcl*10
B3B12.prop_samples <- ecospat.recstrat_prop(B3B12.comb,100)
plot(B3B12.comb)
points(B3B12.prop_samples$x,B3B12.prop_samples$y,pch=16,cex=0.6,col=B3B12.prop_samples$class)
## End(Not run)
ecospat.recstrat regl Random Ecologically Stratified Sampling of equal numbers
Description
This function randomly takes an equal number of samples per class in the stratification
layer.
Usage
ecospat.recstrat_regl(in_grid, sample_no)
82 ecospat.sample.envar
Arguments
in grid The stratification grid to be sampled.
sample no The total number of pixels to be sampled.
Details
The number of classes in the stratification layer is determined automatically from the integer
input map. If the number of pixels in a class is higher than the number of samples, then
a random selection without re-substitution is performed, otherwise all pixels of that class
are selected.
Value
Returns a dataframe with the selected sampling locations their coordinates and the strata
they belong in.
Author(s)
Achilleas Psomas <achilleas.psomas@wsl.ch>and Niklaus E. Zimmermann <niklaus.zimmermann@wsl.ch>
See Also
ecospat.recstrat prop ecospat.rcls.grd
Examples
## Not run:
bio3<- raster(system.file("external/bioclim/current/bio3.grd",package="biomod2"))
bio12<- raster(system.file("external/bioclim/current/bio12.grd",package="biomod2"))
B3.rcl<-ecospat.rcls.grd(bio3,9)
B12.rcl<-ecospat.rcls.grd(bio12,9)
B3B12.comb <- B12.rcl+B3.rcl*10
B3B12.regl_samples <- ecospat.recstrat_prop(B3B12.comb,100)
plot(B3B12.comb)
points(B3B12.regl_samples$x,B3B12.regl_samples$y,pch=16,cex=0.6,col=B3B12.regl_samples$class)
## End(Not run)
ecospat.sample.envar Sample Environmental Variables
Description
Add environmental values to a species dataframe.
Usage
ecospat.sample.envar (dfsp, colspxy, colspkept = "xy", dfvar,
colvarxy, colvar = "all", resolution)
ecospat.SESAM.prr 83
Arguments
dfsp A species dataframe with x (long), y (lat) and optional other variables.
colspxy The range of columns for x (long) and y (lat) in dfsp.
colspkept The columns of dfsp that should be kept in the final dataframe (default:
xy).
dfvar A dataframe object with x, y and environmental variables.
colvarxy The range of columns for x and y in dfvar.
colvar The range of enviromental variable columns in dfvar (default: all except
xy).
resolution The distance between x,y of species and environmental datafreme beyond
which values shouldn’t be added.
Details
The xy (lat/long) coordinates of the species occurrences are compared to those of the
environment dataframe and the value of the closest pixel is added to the species dataframe.
When the closest environment pixel is more distant than the given resolution, NA is added
instead of the value. This function is similar to sample() in ArcGIS.
Value
A Dataframe with the same rows as dfsp, with environmental values from dfvar in column.
Author(s)
Olivier Broennimann <olivier.broennimann@unil.ch>
Examples
## Not run:
spp <- ecospat.testNiche
sp1 <- spp[1:32,1:3]
occ.sp1 <- ecospat.occ.desaggregation(dfvar=sp1,colxy=2:3,colvar=NULL, min.dist=500,plot=TRUE)
clim <- ecospat.testData[2:8]
occ_sp1 <- na.exclude(ecospat.sample.envar(dfsp=occ.sp1,colspxy=1:2,colspkept=1:2,dfvar=clim,
colvarxy=1:2,colvar="all",resolution=25))
## End(Not run)
ecospat.SESAM.prr SESAM Probability Ranking Rule
Description
Implement the SESAM framework to predict community composition using a ‘probability
ranking‘ rule.
Usage
ecospat.SESAM.prr(proba, sr)
84 ecospat.SESAM.prr
Arguments
proba A data frame object of SDMs probabilities (or other sources) for all
species. Column names (species names SDM) and row name (sampling
sites) (need to have defined row names).
sr A data frame object with species richness value in the first column. Sites
should be arranged in the same order as in the ‘prob‘ argument.
Details
The SESAM framework implemented in ecospat is based on 1) probabilities of individual
species presence for each site - these can be obtained for example by fitting SDMs. This
step represents the application of an environmental filter to the community assembly, 2)
richness predictions for each site - the richness prediction can be derived in different ways,
for instance by summing probabilities from individual species presence for each site or by
fitting direct richness models. This step represents the application of a macroecological
constraint to the number of species that can coexist in the considered unit, 3) a biotic rule
to decide which species potentially present in the site are retained in the final prediction
to match the richness value predicted. The biotic rule applied here is called ‘probability
ranking‘ rule: the community composition in each site is determined by ranking the species
in decreasing order of their predicted probability of presence from SDMs up to a richness
prediction.
Value
Returns a ‘.txt‘ file saved in the working directory that contains the community prediction
by the SESAM framework, i.e. binary predictions for all species (columns) for each site
(rows).
Author(s)
Manuela D‘Amen <manuela.damen@unil.ch>and Anne Dubuis <anne.dubuis@gmail.com>
References
D‘Amen, M., A. Dubuis, R.F. Fernandes, J. Pottier, L. Pellissier and A. Guisan. 2015.
Using species richness and functional traits predictions to constrain assemblage predictions
from stacked species distribution models. J. Biogeogr.,42, 1255-1266.
Guisan, A. and C. Rahbek. 2011. SESAM - a new framework integrating macroecological
and species distribution models for predicting spatio-temporal patterns of species assem-
blages. J. Biogeogr.,38, 1433-1444.
Examples
proba <- ecospat.testData[,73:92]
sr <- as.data.frame(rowSums(proba))
ecospat.SESAM.prr(proba, sr)
ecospat.shift.centroids 85
ecospat.shift.centroids
Draw Centroid Arrows
Description
Draw arrows linking the centroid of the native and exotic (non-native) distribution (con-
tinuous line) and between native and invaded extent (dashed line).
Usage
ecospat.shift.centroids(sp1, sp2, clim1, clim2,col)
Arguments
sp1 The scores of the species native distribution along the the two first axes
of the PCA.
sp2 The scores of the species invasive distribution along the the two first axes
of the PCA.
clim1 The scores of the entire native extent along the the two first axes of the
PCA.
clim2 The scores of the entire invaded extent along the the two first axes of the
PCA.
col Colour of the arrow.
Details
Allows to visualize the shift of the niche centroids of the species and the centroids of the
climatic conditions in the study area. To compare invasive species niche, the arrow links
the centroid of the native and inasive distribution (continuous line) and between native and
invaded extent (dashed line).
Value
Arrow on the overlap test plot
Author(s)
Blaise Petitpierre <bpetitpierre@gmail.com>
86 ecospat.testData
ecospat.testData Test Data For The Ecospat package
Description
Data frame that contains vegetation plots data: presence records of 50 species, a set of en-
vironmental variables (topo-climatic) and SDM predictions for some species in the Western
Swiss Alps (Canton de Vaud, Switzerland).
Usage
data("ecospat.testData")
Format
A data frame with 300 observations on the following 96 variables.
numplots Number of the vegetation plot.
long Longitude, in Swiss plane coordinate system of the vegetation plot.
lat Latitude, in Swiss plane coordinate system of the vegetation plot.
ddeg Growing degree days (with a 0 degrees Celsius threshold).
mind Moisture index over the growing season (average values for June to August in mm
day-1).
srad The annual sum of radiation (in kJ m-2 year-1).
slp Slope (in degrees) calculated from the DEM25.
topo Topographic position (an integrated and unitless measure of topographic exposure.
Achillea atrata
Achillea millefolium
Acinos alpinus
Adenostyles glabra
Aposeris foetida
Arnica montana
Aster bellidiastrum
Bartsia alpina
Bellis perennis
Campanula rotundifolia
Centaurea montana
Cerastium latifolium
Cruciata laevipes
Doronicum grandiflorum
Galium album
Galium anisophyllon
Galium megalospermum
Gentiana bavarica
ecospat.testData 87
Gentiana lutea
Gentiana purpurea
Gentiana verna
Globularia cordifolia
Globularia nudicaulis
Gypsophila repens
Hieracium lactucella
Homogyne alpina
Hypochaeris radicata
Leontodon autumnalis
Leontodon helveticus
Myosotis alpestris
Myosotis arvensis
Phyteuma orbiculare
Phyteuma spicatum
Plantago alpina
Plantago lanceolata
Polygonum bistorta
Polygonum viviparum
Prunella grandiflora
Rhinanthus alectorolophus
Rumex acetosa
Rumex crispus
Vaccinium gaultherioides
Veronica alpina
Veronica aphylla
Agrostis capillaris
Bromus erectus sstr
Campanula scheuchzeri
Carex sempervirens
Cynosurus cristatus
Dactylis glomerata
Daucus carota
Festuca pratensis sl
Geranium sylvaticum
Leontodon hispidus sl
Potentilla erecta
Pritzelago alpina sstr
Prunella vulgaris
Ranunculus acris sl
88 ecospat.testData
Saxifraga oppositifolia
Soldanella alpina
Taraxacum officinale aggr
Trifolium repens sstr
Veronica chamaedrys
Parnassia palustris
glm Agrostis capillaris GLM model for the species Agrostis capillaris.
glm Leontodon hispidus sl GLM model for the species Leontodon hispidus sl.
glm Dactylis glomerata GLM model for the species Dactylis glomerata.
glm Trifolium repens sstr GLM model for the species Trifolium repens sstr.
glm Geranium sylvaticum GLM model for the species Geranium sylvaticum.
glm Ranunculus acris sl GLM model for the species Ranunculus acris sl.
glm Prunella vulgaris GLM model for the species Prunella vulgaris.
glm Veronica chamaedrys GLM model for the species Veronica chamaedrys.
glm Taraxacum officinale aggr GLM model for the species Taraxacum officinale aggr.
glm Plantago lanceolata GLM model for the species Plantago lanceolata.
glm Potentilla erecta GLM model for the species Potentilla erecta.
glm Carex sempervirens GLM model for the species Carex sempervirens.
glm Soldanella alpina GLM model for the species Soldanella alpina.
glm Cynosurus cristatus GLM model for the species Cynosurus cristatus.
glm Campanula scheuchzeri GLM model for the species Campanula scheuchzeri.
glm Festuca pratensis sl GLM model for the species Festuca pratensis sl.
gbm Bromus erectus sstr GBM model for the species Bromus erectus sstr.
glm Saxifraga oppositifolia GLM model for the species Saxifraga oppositifolia.
glm Daucus carota GLM model for the species Daucus carota.
glm Pritzelago alpina sstr GLM model for the species Pritzelago alpina sstr.
glm Bromus erectus sstr GLM model for the species Bromus erectus sstr.
gbm Saxifraga oppositifolia GBM model for the species Saxifraga oppositifolia.
gbm Daucus carota GBM model for the species Daucus carota.
gbm Pritzelago alpina sstr GBM model for the species Pritzelago alpina sstr.
Details
The study area is the Western Swiss Alps of Canton de Vaud, Switzerland.
Five topo-climatic explanatory variables to calibrate the SDMs: growing degree days (with
a 0 degrees Celsius threshold); moisture index over the growing season (average values for
June to August in mm day-1); slope (in degrees); topographic position (an integrated and
unitless measure of topographic exposure; Zimmermann et al., 2007); and the annual sum
of radiation (in kJ m-2 year-1). The spatial resolution of the predictor is 25 m x 25 m so
that the models could capture most of the small-scale variations of the climatic factors in
the mountainous areas.
Two modelling techniques were used to produce the SDMs: generalized linear models (GLM;
McCullagh & Nelder, 1989; R library ’glm’) and generalized boosted models (GBM; Fried-
man, 2001; R library ’gbm’). The SDMs correpond to 20 species: Agrostis capillaris,
ecospat.testEnvRaster 89
Leontodon hispidus sl, Dactylis glomerata, Trifolium repens sstr, Geranium sylvaticum, Ra-
nunculus acris sl, Prunella vulgaris, Veronica chamaedrys, Taraxacum officinale aggr, Plan-
tago lanceolata, Potentilla erecta, Carex sempervirens, Soldanella alpina, Cynosurus cristatus,
Campanula scheuchzeri, Festuca pratensis sl, Daucus carota, Pritzelago alpina sstr, Bro-
mus erectus sstr and Saxifraga oppositifolia.
Author(s)
Antoine Guisan <antoine.guisan@unil.ch>, Anne Dubuis <anne.dubuis@gmail.com>
and Valeria Di Cola <valeria.dicola@unil.ch>
References
Guisan, A. 1997. Distribution de taxons vegetaux dans un environnement alpin: Appli-
cation de modelisations statistiques dans un systeme d’information geographique. PhD
Thesis, University of Geneva, Switzerland.
Guisan, A., J.P. Theurillat. and F. Kienast. 1998. Predicting the potential distribution of
plant species in an alpine environment. Journal of Vegetation Science,9, 65-74.
Guisan, A. and J.P. Theurillat. 2000. Assessing alpine plant vulnerability to climate change:
A modeling perspective. Integrated Assessment,1, 307-320.
Guisan, A. and J.P. Theurillat. 2000. Equilibrium modeling of alpine plant distribution
and climate change : How far can we go? Phytocoenologia,30(3-4), 353-384.
Dubuis A., J. Pottier, V. Rion, L. Pellissier, J.P. Theurillat and A. Guisan. 2011. Predicting
spatial patterns of plant species richness: A comparison of direct macroecological and
species stacking approaches. Diversity and Distributions,17, 1122-1131.
Zimmermann, N.E., T.C. Edwards, G.G Moisen, T.S. Frescino and J.A. Blackard. 2007.
Remote sensing-based predictors improve distribution models of rare, early successional and
broadleaf tree species in Utah. Journal of Applied Ecology 44, 1057-1067.
Examples
data(ecospat.testData)
str(ecospat.testData)
dim(ecospat.testData)
names(ecospat.testData)
ecospat.testEnvRaster Test Environmental Rasters for The Ecospat package
Description
A stack of 5 topoclimatic rasters at 250m resolution for the Western Swiss Alps. It includes
”ddeg0” (growing degree-days above 0C), ”mind68” (moisture index for month June to
August), ”srad68” (solar radiation for month June to August), ”slope25” (average of slopes
at 25m resolution) and ”topos25” (average of topographic positions at 25m resolution)
Format
ecospat.testEnvRaster is a RasterBrick encapsulated in a .Rdata that contains the following
rasters:
[1] ”ddeg0” [2] ”mind68” [3] ”srad68” [4] ”slope25” [5] ”topos25”
90 ecospat.testMdr
Author(s)
Olivier Broennimann <olivier.broennimann@unil.ch>
References
Zimmermann, N.E., F. Kienast. 2009. Predictive mapping of alpine grasslands in Switzer-
land: Species versus community approach. Journal of Vegetation Science,10, 469-482.
Examples
fpath <- system.file("extdata", "ecospat.testEnvRaster.RData", package="ecospat")
load(fpath)
plot(env)
ecospat.testMdr Test Data For The ecospat.mdr function
Description
Data frame that contains presence records the species Centaurea stoebe along years in
North America.
Usage
data("ecospat.testMdr")
Format
A data frame with 102 observations of Centaurea stoebe.
latitude Latitude, in WGS coordinate system.
longitude Longitude, in WGS coordinate system.
date Year of the presence record.
Details
Simplified dataset to exemplify the use of the ecospat.mdr function to calculate minimum
dispersal routes.
Author(s)
Olivier Broennimann <olivier.broennimann@unil.ch>
References
Broennimann, O., P. Mraz, B. Petitpierre, A. Guisan, and H. Muller-Scharer. 2014. Con-
trasting spatio-temporal climatic niche dynamics during the eastern and western invasions
of spotted knapweed in North America.Journal of biogeography,41, 1126-1136.
Hordijk, W. and O. Broennimann. 2012. Dispersal routes reconstruction and the minimum
cost arborescence problem. Journal of theoretical biology,308, 115-122.
ecospat.testNiche 91
Examples
data(ecospat.testMdr)
str(ecospat.testMdr)
dim(ecospat.testMdr)
ecospat.testNiche Test Data For The Niche Overlap Analysis
Description
Data frame that contains occurrence sites for each species, long, lat and the name of the
species at each site.
Usage
data(ecospat.testNiche)
Format
ecospat.testNiche is a data frame with the following columns:
species sp1, sp2, sp3 and sp4.
long Longitude, in Swiss plane coordinate system of the vegetation plot.
lat Latitude, in Swiss plane coordinate system of the vegetation plot.
Spp Scientific name of the species used in the exmaple: Bromus erectus sstr, Saxifraga oppositifolia,
Daucus carota and Pritzelago alpina sstr.
Details
List of occurence sites for the species.
Author(s)
Antoine Guisan <antoine.guisan@unil.ch>, Anne Dubuis <anne.dubuis@gmail.com>
and Valeria Di Cola <valeria.dicola@unil.ch>
See Also
ecospat.testData
Examples
data(ecospat.testNiche)
dim(ecospat.testNiche)
names(ecospat.testNiche)
92 ecospat.testNiche.inv
ecospat.testNiche.inv Test Data For The Niche Dynamics Analysis In The Invaded
Range Of A Hypothetical Species
Description
Data frame that contains geographical coordinates, environmental variables, occurrence
sites for the studied species and the prediction of its distribution in the invaded range.
These predictions are provided by SDM calibrated on the native range.
Usage
data(ecospat.testNiche.inv)
Format
ecospat.testNiche.inv is a data frame with the following columns:
xLongitude, in WGS84 coordinate system of the species occurrence.
yLatitude, in WGS84 coordinate system of the species occurrence.
aetpet Ratio of actual to potential evapotranspiration.
gdd Growing degree-days above 5 degrees C.
pAnnual amount of precipitations.
pet Potential evapotranspiration.
stdp Annual variation of precipitations.
tmax Maximum temperature of the warmest month.
tmin Minimum temperature of the coldest month.
tmp Annual mean temperature.
species occ Presence records of the species occurrence.
predictions Species Distribution Model predictions of the studied species.
Details
The study area is Australia, which is the invaded range of the hypothetical species.
Eight topo-climatic explanatory variables to quantify niche differences: ratio of the actual
potential evapotranspiration; growing degree days; precipitation; potential evapotranspi-
ration; annual variation of precipitations; maximum temperature of the warmest month;
minimum temperature of the coldest month; and annual mean temperature.
Author(s)
Blaise Petitpierre <bpetitpierre@gmail.com>and Valeria Di Cola <valeria.dicola@unil.ch>
References
Petitpierre, B., C. Kueffer, O. Broennimann, C. Randin, C. Daehler and A. Guisan. 2012.
Climatic niche shifts are rare among terrestrial plant invaders. Science,335, 1344-1348.
ecospat.testNiche.nat 93
See Also
ecospat.testNiche.nat
Examples
data(ecospat.testNiche.inv)
str(ecospat.testNiche.inv)
dim(ecospat.testNiche.inv)
names(ecospat.testNiche.inv)
ecospat.testNiche.nat Test Data For The Niche Dynamics Analysis In The Native
Range Of A Hypothetical Species
Description
Data frame that contains geographical coordinates, environmental variables, occurrence
sites for the studied species and the prediction of its distribution in the native range. These
predictions are provided by SDM calibrated on the native range.
Usage
data(ecospat.testNiche.nat)
Format
ecospat.testNiche.nat is a data frame with the following columns:
xLongitude, in WGS84 coordinate system of the species occurrence.
yLatitude, in WGS84 coordinate system of the species occurrence.
aetpet Ratio of actual to potential evapotranspiration.
gdd Growing degree-days above 5 degrees C.
pAnnual amount of precipitations.
pet Potential evapotranspiration.
stdp Annual variation of precipitations.
tmax Maximum temperature of the warmest month.
tmin Minimum temperature of the coldest month.
tmp Annual mean temperature.
species occ Presence records of the species occurrence.
predictions Species Distribution Model predictions of the studied species.
Details
The study area is North America, which is the native range of the hypothetical species.
Eight topo-climatic explanatory variables to quantify niche differences: ratio of the actual
potential evapotranspiration; growing degree days; precipitation; potential evapotranspi-
ration; annual variation of precipitations; maximum temperature of the warmest month;
minimum temperature of the coldest month; and annual mean temperature.
94 ecospat.testTree
Author(s)
Blaise Petitpierre <bpetitpierre@gmail.com>and Valeria Di Cola <valeria.dicola@unil.ch>
References
Petitpierre, B., C. Kueffer, O. Broennimann, C. Randin, C. Daehler and A. Guisan. 2012.
Climatic niche shifts are rare among terrestrial plant invaders. Science,335, 1344-1348.
See Also
ecospat.testNiche.inv
Examples
data(ecospat.testNiche.nat)
str(ecospat.testNiche.nat)
dim(ecospat.testNiche.nat)
names(ecospat.testNiche.nat)
ecospat.testTree Test Tree For The Ecospat package
Description
The tree object is a phylogenetic tree of class ’phylo’ (see read.tree) that contains data of
50 angiosperm species from the Western Swiss Alps.
Format
ecospat.testTree is a tree contains the following species:
[1] ”Rumex acetosa” [2] ”Polygonum bistorta” [3] ”Polygonum viviparum” [4] ”Rumex crispus”
[5] ”Cerastium latifolium” [6] ”Silene acaulis” [7] ”Gypsophila repens” [8] ”Vaccinium gaultherioides”
[9] ”Soldanella alpina” [10] ”Cruciata laevipes” [11] ”Galium album” [12] ”Galium anisophyllon”
[13] ”Galium megalospermum” [14] ”Gentiana verna” [15] ”Gentiana bavarica” [16] ”Gen-
tiana purpurea” [17] ”Gentiana lutea” [18] ”Bartsia alpina” [19] ”Rhinanthus alectorolophus”
[20] ”Prunella grandiflora” [21] ”Acinos alpinus” [22] ”Plantago alpina” [23] ”Plantago lanceolata”
[24] ”Veronica officinalis” [25] ”Veronica aphylla” [26] ”Veronica alpina” [27] ”Veronica chamaedrys”
[28] ”Veronica persica” [29] ”Globularia cordifolia” [30] ”Globularia nudicaulis” [31] ”Myoso-
tis alpestris” [32] ”Myosotis arvensis” [33] ”Aposeris foetida” [34] ”Centaurea montana”
[35] ”Hieracium lactucella” [36] ”Leontodon helveticus” [37] ”Leontodon autumnalis” [38]
”Hypochaeris radicata” [39] ”Achillea atrata” [40] ”Achillea millefolium” [41] ”Homogyne alpina”
[42] ”Senecio doronicum” [43] ”Adenostyles glabra” [44] ”Arnica montana” [45] ”Aster bellidiastrum”
[46] ”Bellis perennis” [47] ”Doronicum grandiflorum” [48] ”Phyteuma orbiculare” [49] ”Phy-
teuma spicatum” [50] ”Campanula rotundifolia”
Author(s)
Charlotte Ndiribe <charlotte.ndiribe@unil.ch>, Nicolas Salamin <nicolas.salamin@unil.ch>
and Antoine Guisan <antoine.guisan@unil.ch>
ecospat.varpart 95
References
Ndiribe, C., L. Pellissier, S. Antonelli, A. Dubuis, J. Pottier, P. Vittoz, A. Guisan and N.
Salamin. 2013. Phylogenetic plant community structure along elevation is lineage specific.
Ecology and Evolution,3, 4925-4939.
Examples
fpath <- system.file("extdata", "ecospat.testTree.tre", package="ecospat")
tree <- read.tree(fpath)
plot(tree)
ecospat.varpart Variation Partitioning For GLM Or GAM
Description
Perform variance partitioning for binomial GLM or GAM based on the deviance of two
groups or predicting variables.
Usage
ecospat.varpart (model.1, model.2, model.12)
Arguments
model.1 GLM / GAM calibrated on the first group of variables.
model.2 GLM / GAM calibrated on the second group of variables.
model.12 GLM / GAM calibrated on all variables from the two groups.
Details
The deviance is calculated with the adjusted geometric mean squared improvement rescaled
for a maximum of 1.
Value
Return the four fractions of deviance as in Randin et al. 2009: partial deviance of model 1
and 2, joined deviance and unexplained deviance.
Author(s)
Christophe Randin <christophe.randin@unibas.ch>, Helene Jaccard and Nigel Gilles
Yoccoz
References
Randin, C.F., H. Jaccard, P. Vittoz, N.G. Yoccoz and A. Guisan. 2009. Land use improves
spatial predictions of mountain plant abundance but not presence-absence. Journal of
Vegetation Science,20, 996-1008.
96 ecospat.varpart
Examples
## Not run:
ecospat.cv.example()
ecospat.varpart (model.1= get ("glm.Achillea_atrata", envir=ecospat.env),
model.2= get ("glm.Achillea_millefolium", envir=ecospat.env),
model.12= get ("glm.Achillea_millefolium", envir=ecospat.env))
## End(Not run)
Index
∗Topic \textasciitildekwd1
ecospat.CCV.communityEvaluation.bin,
11
ecospat.CCV.communityEvaluation.prob,
13
ecospat.CCV.createDataSplitTable,
15
ecospat.CCV.modeling,17
∗Topic \textasciitildekwd2
ecospat.CCV.communityEvaluation.bin,
11
ecospat.CCV.communityEvaluation.prob,
13
ecospat.CCV.createDataSplitTable,
15
ecospat.CCV.modeling,17
∗Topic file
ecospat.max.tss,49
ecospat.meva.table,54
ecospat.occupied.patch,66
ecospat.plot.kappa,70
ecospat.plot.tss,74
ecospat.rangesize,76
∗Topic package
ecospat-package,3
BIOMOD.EnsembleModeling.out,34
BIOMOD.models.out,39
BIOMOD.projection.out,42
BIOMOD EnsembleForecasting,36,37
BIOMOD EnsembleModeling,37
BIOMOD FormatingData,38,40,42
BIOMOD Modeling,34,35,37,40,42
BIOMOD ModelingOptions,18,40,42
BIOMOD Projection,35,37,40,42
dudi.pca,70
ecospat (ecospat-package),3
ecospat-package,3
ecospat.adj.D2.glm,5
ecospat.binary.model,6,67,77
ecospat.boyce,7
ecospat.calculate.pd,8
ecospat.caleval,10
ecospat.CCV.communityEvaluation.bin,
11,13,15,18,19
ecospat.CCV.communityEvaluation.prob,
12,13,18,19
ecospat.CCV.createDataSplitTable,12,
15,15,17,19
ecospat.CCV.modeling,11–13,16,17
ecospat.climan,19
ecospat.co occurrences,25,27
ecospat.cohen.kappa,20,49,50,54,70,
75
ecospat.CommunityEval,21
ecospat.cons Cscore,23,27
ecospat.cor.plot,24
ecospat.Cscore,26
ecospat.cv.example,27
ecospat.cv.gbm,28
ecospat.cv.glm,29
ecospat.cv.me,30
ecospat.cv.rf,31
ecospat.env,32
ecospat.Epred,32
ecospat.ESM.EnsembleModeling,33,36,
37,40,42
ecospat.ESM.EnsembleProjection,34,35,
36,40,42
ecospat.ESM.Modeling,33,35,37,38,41,
42
ecospat.ESM.Projection,35–37,40,41
ecospat.grid.clim.dyn,43,57,60,61,63,
72
ecospat.makeDataFrame,45
ecospat.mantel.correlogram,47
ecospat.max.kappa,21,48,50,54,70,75
ecospat.max.tss,21,49,49,54,70,75
ecospat.maxentvarimport,50
ecospat.mdr,51
ecospat.mess,53,71
ecospat.meva.table,21,49,50,54,70,75
ecospat.migclim,55
ecospat.mpa,7,56,67,77
ecospat.niche.dyn.index,57,58
97
98 INDEX
ecospat.niche.dynIndexProjGeo,58,64
ecospat.niche.equivalency.test,59,63,
74
ecospat.niche.overlap,61
ecospat.niche.similarity.test,60,62,
70,74
ecospat.niche.zProjGeo,58,63
ecospat.npred,65
ecospat.occ.desaggregation,65
ecospat.occupied.patch,66,77
ecospat.permut.glm,68
ecospat.plot.contrib,69
ecospat.plot.kappa,21,49,50,54,70,75
ecospat.plot.mess,53,71
ecospat.plot.niche,44,72
ecospat.plot.niche.dyn,44,58,64,70,
73
ecospat.plot.overlap.test,70,73
ecospat.plot.tss,21,49,50,54,70,74
ecospat.rand.pseudoabsences,75
ecospat.rangesize,67,76
ecospat.rcls.grd,79,81,82
ecospat.recstrat prop,80,82
ecospat.recstrat regl,81,81
ecospat.sample.envar,82
ecospat.SESAM.prr,83
ecospat.shift.centroids,85
ecospat.testData,86,91
ecospat.testEnvRaster,89
ecospat.testMdr,90
ecospat.testNiche,91
ecospat.testNiche.inv,92,94
ecospat.testNiche.nat,93,93
ecospat.testTree,94
ecospat.varpart,95
mgram,48
optimal.thresholds,7
princomp,70

Navigation menu