occuMulti: Fit the Rota et al. (2016) Multi-species Occupancy Model

Description Usage Arguments Details Value Author(s) References See Also Examples

View source: R/occuMulti.R

Description

This function fits the multispecies occupancy model of Rota et al (2016).

Usage

1
2
occuMulti(detformulas, stateformulas, data, maxOrder, starts, method="BFGS",
    se=TRUE, engine=c("C","R"), silent=FALSE, ...)

Arguments

detformulas

Character vector of formulas for the detection models, one per species.

stateformulas

Character vector of formulas for the natural parameters. To fix a natural parameter at 0, specify the corresponding formula as "0" or "~0".

data

An unmarkedFrameOccuMulti object

maxOrder

Optional; specify maximum interaction order. Defaults to number of species (all possible interactions). Reducing this value may speed up optimization if you aren't interested in higher-order interactions.

starts

Vector of parameter starting values.

method

Optimization method used by optim.

se

Logical specifying whether or not to compute standard errors.

engine

Either "C" to use fast C++ code or "R" to use native R code during the optimization.

silent

Boolean; if TRUE, suppress warnings.

...

Additional arguments to optim, such as lower and upper bounds

Details

See unmarkedFrame and unmarkedFrameOccuMulti for a description of how to supply data to the data argument.

occuMulti fits the multispecies occupancy model from Rota et al. (2016), for two or more interacting species. The model generalizes the standard single-species occupancy model from MacKenzie et al. (2002). The latent occupancy state at site i for a set of s potentially interacting species is a vector Z_i of length s containing a sequence of the values 0 or 1. For example, when s = 2, the possible states are [11], [10], [01], or [00], corresponding to both species present, only species 1 or species 2 present, or both species absent, respectively. The latent state modeled as a multivariate Bernoulli random variable:

Z_i ~ MVB(psi_i)

where ψ_i is a vector of length 2^s containing the probability of each possible combination of 0s and 1s, such that sum(psi_i) = 1.

For s = 2, the corresponding natural parameters f are

f_1 = log(psi_10/psi_00)

f_2 = log(psi_01/psi_00)

f_12 = log((psi_11 * psi_00)/(psi_10 * psi_01))

The natural parameters can then be modeled as linear functions of covariates. Covariates for each f must be specified with the stateformulas argument, which takes a character vector of individual formulas of length equal to the number of natural parameters (which in turn depends on the number of species in the model).

The observation process is similar to the standard single-species occupancy model, except that the observations y_ij at site i on occasion j are vectors of length s and there are independent values of detection probability p for each species s:

y_ij | Z_i ~ Bernoulli(Z_i * p_sij)

Independent detection models (potentially containing different covariates) must be provided for each species with the detformulas argument, which takes a character vector of individual formulas with length equal to the number of species s.

Value

unmarkedFitOccuMulti object describing the model fit.

Author(s)

Ken Kellner contact@kenkellner.com

References

MacKenzie, D. I., J. D. Nichols, G. B. Lachman, S. Droege, J. Andrew Royle, and C. A. Langtimm. 2002. Estimating Site Occupancy Rates When Detection Probabilities Are Less Than One. Ecology 83: 2248-2255.

Rota, C.T., et al. 2016. A multi-species occupancy model for two or more interacting species. Methods in Ecology and Evolution 7: 1164-1173.

See Also

unmarked, unmarkedFrameOccuMulti

Examples

  1
  2
  3
  4
  5
  6
  7
  8
  9
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
## Not run: 
#Simulate 3 species data
N <- 1000
nspecies <- 3
J <- 5

occ_covs <- as.data.frame(matrix(rnorm(N * 10),ncol=10))
names(occ_covs) <- paste('occ_cov',1:10,sep='')

det_covs <- list()
for (i in 1:nspecies){
  det_covs[[i]] <- matrix(rnorm(N*J),nrow=N)
}
names(det_covs) <- paste('det_cov',1:nspecies,sep='')

#True vals
beta <- c(0.5,0.2,0.4,0.5,-0.1,-0.3,0.2,0.1,-1,0.1)
f1 <- beta[1] + beta[2]*occ_covs$occ_cov1
f2 <- beta[3] + beta[4]*occ_covs$occ_cov2
f3 <- beta[5] + beta[6]*occ_covs$occ_cov3
f4 <- beta[7]
f5 <- beta[8]
f6 <- beta[9]
f7 <- beta[10]
f <- cbind(f1,f2,f3,f4,f5,f6,f7)
z <- expand.grid(rep(list(1:0),nspecies))[,nspecies:1]
colnames(z) <- paste('sp',1:nspecies,sep='')
dm <- model.matrix(as.formula(paste0("~.^",nspecies,"-1")),z)

psi <- exp(f %*% t(dm))
psi <- psi/rowSums(psi)

#True state
ztruth <- matrix(NA,nrow=N,ncol=nspecies)
for (i in 1:N){
  ztruth[i,] <- as.matrix(z[sample(8,1,prob=psi[i,]),])
}

p_true <- c(0.6,0.7,0.5)

# fake y data
y <- list()

for (i in 1:nspecies){
  y[[i]] <- matrix(NA,N,J)
  for (j in 1:N){
    for (k in 1:J){
      y[[i]][j,k] <- rbinom(1,1,ztruth[j,i]*p_true[i])
    }
  }
}
names(y) <- c('coyote','tiger','bear')

#Create the unmarked data object
data = unmarkedFrameOccuMulti(y=y,siteCovs=occ_covs,obsCovs=det_covs)

#Summary of data object
summary(data)
plot(data)

# Look at f parameter design matrix
data@fDesign

# Formulas for state and detection processes

# Length should match number/order of columns in fDesign
occFormulas <- c('~occ_cov1','~occ_cov2','~occ_cov3','~1','~1','~1','~1')

#Length should match number/order of species in data@ylist
detFormulas <- c('~1','~1','~1')

fit <- occuMulti(detFormulas,occFormulas,data)

#Look at output
fit

plot(fit)

#Compare with known values
cbind(c(beta,log(p_true/(1-p_true))),fit@opt$par)

#predict method
lapply(predict(fit,'state'),head)
lapply(predict(fit,'det'),head)

#marginal occupancy
head(predict(fit,'state',species=2))
head(predict(fit,'state',species='bear'))
head(predict(fit,'det',species='coyote'))

#probability of co-occurrence of two or more species
head(predict(fit, 'state', species=c('coyote','tiger')))

#conditional occupancy
head(predict(fit,'state',species=2,cond=3)) #tiger | bear present
head(predict(fit,'state',species='tiger',cond='bear')) #tiger | bear present
head(predict(fit,'state',species='tiger',cond='-bear')) #bear absent
head(predict(fit,'state',species='tiger',cond=c('coyote','-bear')))

#residuals (by species)
lapply(residuals(fit),head)

#ranef (by species)
ranef(fit, species='coyote')

#parametric bootstrap
bt <- parboot(fit,nsim=30)

#update model
occFormulas <- c('~occ_cov1','~occ_cov2','~occ_cov2+occ_cov3','~1','~1','~1','~1')
fit2 <- update(fit,stateformulas=occFormulas)

#List of fitted models
fl <- fitList(fit,fit2)
coef(fl)

#Model selection
modSel(fl)

#Fit model while forcing some natural parameters to be 0
#For example: fit model with no species interactions
occFormulas <- c('~occ_cov1','~occ_cov2','~occ_cov2+occ_cov3','0','0','0','0')
fit3 <- occuMulti(detFormulas,occFormulas,data)

#Alternatively, you can force all interaction parameters above a certain
#order to be zero with maxOrder. This will be faster.
occFormulas <- c('~occ_cov1','~occ_cov2','~occ_cov2+occ_cov3')
fit4 <- occuMulti(detFormulas,occFormulas,data,maxOrder=1)

## End(Not run)

unmarked documentation built on May 27, 2021, 5:07 p.m.