simOcc: Simulate data for static occupancy models under wide range of...

In AHMbook: Functions and Data for the Book 'Applied Hierarchical Modeling in Ecology' Vols 1 and 2

Simulate data for static occupancy models under wide range of conditions

Description

Functions to simulate static occupancy measurements replicated at M sites during J occasions. Population closure is assumed for each site. Expected occurrence may be affected by elevation (elev), forest cover (forest) and their interaction. Expected detection probability may be affected by elevation, wind speed (wind) and their interaction.

simOccCat allows for a categorical habitat-type covariate (HAB) to affect occurrence and another, the observer (OBS), to affect detection.

Usage

simOcc(M = 267, J = 3, mean.occupancy = 0.6, beta1 = -2, beta2 = 2, beta3 = 1,
  mean.detection = 0.3, time.effects = c(-1, 1),
  alpha1 = -1, alpha2 = -3, alpha3 = 0, sd.lp = 0.5,
  b = 2, show.plots = TRUE)

simOccCat(M = 267, J = 3, mean.occupancy = 0.6, beta1 = 0, beta2 = 0, beta3 = 0,
  mean.detection = 0.3, time.effects = c(0, 0),
  alpha1 = 0, alpha2 = 0, alpha3 = 0, sd.lp = 0, b = 0,
  nHab = 5, range.HAB = 2, nObs = 10, range.OBS = 4,
  show.plots = TRUE)

Arguments

M	Number of spatial replicates (sites)
J	Number of temporal replicates (occasions)
mean.occupancy	Mean occurrence at value 0 of occurrence covariates
beta1	Main effect of elevation on occurrence
beta2	Main effect of forest cover on occurrence
beta3	Interaction effect on occurrence of elevation and forest cover
mean.detection	Mean detection prob. at value 0 of detection covariates
time.effects	bounds (on logit scale) for uniform distribution from which time effects gamma will be drawn
alpha1	Main effect of elevation on detection probability
alpha2	Main effect of wind speed on detection probability
alpha3	Interaction effect on detection of elevation and wind speed
sd.lp	standard deviation of random site effects (on logit scale)
b	constant value of 'behavioral response' leading to 'trap-happiness' (if b > 0) or 'trap shyness' (if b < 0)
nHab	number of categories to simulate for the site-specific categorical covariate, HAB.
range.HAB	a non-negative number that controls the size of the HAB effect.
nObs	number of observers, the categories to simulate for the survey-specific categorical covariate, OBS.
range.OBS	a non-negative number that controls the size of the OBS effect.
show.plots	if TRUE, plots of the data will be displayed; set to FALSE if you are running simulations

Value

A list with the values of the input arguments and the following additional elements:

gamma	The time effects, a vector of length J
eps	Individual random effects, a vector of length M
elev	Elevation, a vector of length M
forest	Forest cover, a vector of length M
wind	wind speed, a M x J matrix
psi	Probability of occurrence, a vector of length M
z	Realized occurrence (0/1), a vector of length M
p	probability of capture, possibly with a behavioral effect, a M x J matrix
p0	probability of capture when not captured on previous occasion, a M x J matrix
p1	probability of capture when captured on previous occasion, a M x J matrix
y	simulated capture history, a M x J matrix
sumZ	True number of occupied sites
sumZ.obs	Number of sites observed to be occupied
psi.fs.true	True proportion of occupied sites in sample (sumZ/N)
psi.fs.obs	Proportion of sites observed to be occupied (sumZ.obs/N)

In addition, simOccCat has

HAB	Categorical site covariate, a vector of length M
OBS	Categorical detection covariate, a M x J matrix
coefHAB	The effect of each category of HAB on occurrence
coefOBS	The effect of each category of OBS on detection

Author(s)

Marc Kéry, Andy Royle, Gesa von Hirschheydt, Mike Meredith,

References

Kéry, M. & Royle, J.A. (2016) Applied Hierarchical Modeling in Ecology AHM1 - 10.8.

Examples

# Generate data with the default arguments and look at the structure
tmp <- simOcc()
str(tmp)

# Simplest possible occupancy model, with constant occupancy and detection
str(simOcc(mean.occ=0.6, beta1=0, beta2=0, beta3=0, mean.det=0.3, time.effects=c(0, 0),
  alpha1=0, alpha2=0, alpha3=0, sd.lp=0, b=0))
# psi = 1 (i.e., species occurs at every site)
str(simOcc(mean.occ=1))

# p = 1 (i.e., species is always detected when it occurs)
str(simOcc(mean.det=1))

# Other potentially interesting settings include these:

str(simOcc(J = 2))                 # Only 2 surveys
str(simOcc(M = 1, J = 100))        # No spatial replicates, but 100 measurements
str(simOcc(beta3 = 1))             # Including interaction elev-wind on p
str(simOcc(mean.occ = 0.96))       # A really common species
str(simOcc(mean.occ = 0.05))       # A really rare species
str(simOcc(mean.det = 0.96))       # A really easy species
str(simOcc(mean.det = 0.05))       # A really hard species
str(simOcc(mean.det = 0))          # The dreaded invisible species
str(simOcc(alpha1=-2, beta1=2))    # Opposing effects of elev on psi and p
str(simOcc(J = 10, time.effects = c(-5, 5))) # Huge time effects on p
str(simOcc(sd.lp = 10))            # Huge (random) site effects on p
str(simOcc(J = 10, b = 0))         # No behavioral response in p
str(simOcc(J = 10, b = 2))         # Trap happiness
str(simOcc(J = 10, b = -2))        # Trap shyness


# Using categorical covariates only
str(simOccCat())
# Categorical and continuous covariates
str(tmp <- simOccCat(beta1 = -2, beta2 = 2, beta3 = 1,
    mean.detection = 0.3, time.effects = c(-1, 1),
    alpha1 = -1, alpha2 = -3, alpha3 = 0,
    sd.lp = 0.5, b = 2))
# Check how balanced the levels are for HAB
barplot(sort(table(tmp$HAB), decreasing=TRUE), xlab="Habitat category",
    ylab="Frequency", main="Frequency distribution of habitat categories")

AHMbook documentation built on Aug. 24, 2023, 1:07 a.m.