• Short answer:

"By the habitat… is meant the kind of situation in which the organism lives." (McDougall, 1927)

• A little bit more complex:

"Resources and conditions present in an area that produce occupancy—including survival and reproduction—by a given organism." (Hall et al., 1997)

→ Inherently multivariate!

Critical to characterize a species habitat and how it is used.

• Research (ecology, evolution)
• Management
• Conservation

Aim of habitat selection is to highlight relationships between the distribution of locations and environmental features.

• Environment: A set of environmental variables → AVAILABILITY
• Locations: Individuals can be identified or not → USE

→ Habitat selection is statistically defined as a disproportionate usage of available habitat.

"An n-dimensional hypervolume, every point in which corresponds to a state of the environment which would permit the species to exist indefinitely"

• p environmental variables
• N locations in the ecological space
• (can be generalized to the individual)

• Exploration: multivariate analyses (PCA, ENFA, Mahalanobis distances, etc.)
• Modeling: regression analyses (RSF)

• A difference of means: \(\mu - \mu\prime\)

• A ratio of variance: \(\frac{\sigma}{\sigma\prime}\)

The data come from the PhD of Daniel Maillard on wild boars (Sus scrofa) in Puéchabon (France). Six wild boards have been captured and monitored with radio-telemetry in July and August in the years 1993–1995.

During the day, wild boars rest in shelters, which have been located almost daily. We are trying to assess what are the habitat characteristics that wild boars look for to find shelter.

Maillard D. (1996) Occupation et utilisation de la garrigue et du vignoble méditerranéens par le sanglier (Sus scrofa L.). In: Biologie des populations et des écosystemes, 235 pp. Université de droit, d'économie et des sciences d'Aix-Marseille III.

Install necessary packages:

install.packages(c("adehabitatHS", "rgdal"))

Download and extract data.zip in your working directory, and download the R code 1.R:

download.file(
    "https://mablab.org/project/r/2018-04-11-niche-analyses/data.zip",
    "data.zip")
unzip("data.zip", exdir = "data")
file.remove("data.zip")
download.file(
    "https://mablab.org/project/r/2018-04-11-niche-analyses/1.R",
    "1.R")

library("adehabitatHS")
locs <- read.delim("data/locs.txt")
coordinates(locs) <- ~X+Y
head(locs)

##         coordinates   NOM
## 1 (699889, 3161560) Brock
## 2 (699291, 3161120) Brock
## 3 (700046, 3161540) Brock
## 4 (700050, 3161580) Brock
## 5 (699811, 3161390) Brock
## 6 (698840, 3161030) Brock
## Coordinate Reference System (CRS) arguments: NA

library("rgdal")
elev <- readGDAL("data/altitude.asc")
image(elev)
plot(locs, add = TRUE)

li <- lapply(list.files("data", pattern = ".asc",
    full.names = TRUE), readGDAL)
names(li) <- sub(".asc", "", list.files("data", pattern = ".asc"))
maps <- do.call(cbind, li)
fullgrid(maps) <- FALSE
names(maps) <- sub(".asc", "", list.files("data", pattern = ".asc"))
mimage(maps)

We check the presence of any structure with a Principal Component Analysis:

tab <- slot(maps, "data")
pc1 <- dudi.pca(tab)
s.corcircle(pc1$co)

maps$pc1 <- pc1$li[, 1]
maps$pc2 <- pc1$li[, 2]
mimage(maps)

• Population level: all individuals together (perfect for presence data)
• One axis of marginality
• Successive axes of specialization
• Pixels and variables are projected in the factorial plane just as for a PCA.

pr <- slot(count.points(SpatialPoints(locs), maps), "data")[, 1]
enfa1 <- enfa(pc1, pr)
scatter(enfa1, pts = TRUE)
hist(enfa1)
hist(enfa1, scores = FALSE, type = "l")

maps$mar <- enfa1$li$Mar
mimage(maps)

• Stable system (movement, population dynamic, resources)
• Available habitat correctly identified. Maybe the most complicated hypothesis… Must be defined by the biologist!
• Unbiased use data
• Free access to the entire habitat for all animals (selection is a choice)
• Resources affecting habitat selection correctly identified.

Several multivariate niche analyses:

• Population level: GNESFA (General niche-environment system factor analysis; Calenge & Basille, 2008), a generalization of the ENFA with several complementary methods (MADIFA, FANTER). Check the MADIFA to compute habitat suitability.
• Individual level: K-select (Calenge et al., 2005) and EISERA (Eigenanalysis of selection ratios; Calenge & Dufour, 2006) to investigate second- and third-order selection.

Habitat selection is not simple:

• False absences: that the species is not detected does not mean the habitat is avoided!
• Detection problems
• Historical reasons
• Low use of some habitat does not mean it is useless (example: waterhole)
• Habitat selection, yes, but for what kind of activity? Foraging? Resting? Looking for mates?