Multiple surveys by different observers of a single 1km transect containing
150 wooden stakes placed based on expected uniform distribution throughout a
40 m strip (20m on either side).
Format
A data frame with 150 observations on the following 13 variables.
StakeNo
unique number for each stake 1-150
PD
perpendicular distance at which the stake was placed
from the line
Obs1
0/1 whether missed/seen by observer 1
Obs2
0/1 whether missed/seen by observer 2
Obs3
0/1 whether missed/seen by observer 3
Obs4
0/1 whether missed/seen by observer 4
Obs5
0/1 whether missed/seen by observer 5
Obs6
0/1 whether missed/seen by observer 6
Obs7
0/1 whether missed/seen by observer 7
Obs8
0/1 whether missed/seen by observer 8
Obs9
0/1 whether missed/seen by observer 9
Obs10
0/1 whether missed/seen by observer 10
Obs11
0/1 whether missed/seen by observer 11
Details
The 1997 survey was based on a single realization of a uniform distribution.
Because it was a single transect and there was no randomization of the
distances for each survey, we repeated the experiment and used distances
that provided a uniform distribution but randomly sorted the positions along
the line so there was no pattern obvious to the observer.
Source
Laake, J. 1978. Line transect estimators robust to animal movement.
M.S. Thesis. Utah State University, Logan, Utah. 55p.
References
Burnham, K. P., D. R. Anderson, and J. L. Laake. 1980.
Estimation of Density from Line Transect Sampling of Biological
Populations. Wildlife Monographs:7-202.
Examples
data(stake78)
data(stake77)
# compare distribution of distances for all stakes
hist(stake77$PD)
hist(stake78$PD)
# Extract stake data and put in the mrds format for model fitting.
extract.stake <- function(stake,obs){
extract.obs <- function(obs){
example <- subset(stake,eval(parse(text=paste("Obs",obs,"==1",sep=""))),
select="PD")
example$distance <- example$PD
example$object <- 1:nrow(example)
example$PD <- NULL
return(example)
}
if(obs!="all"){
return(extract.obs(obs=obs))
}else{
example <- NULL
for(i in 1:(ncol(stake)-2)){
df <- extract.obs(obs=i)
df$person <- i
example <- rbind(example,df)
}
example$person <- factor(example$person)
example$object <- 1:nrow(example)
return(example)
}
}
extract.stake.pairs <- function(stake,obs1,obs2,removal=FALSE){
obs1 <- paste("Obs",obs1,sep="")
obs2 <- paste("Obs",obs2,sep="")
example <- subset(stake,eval(parse(text=paste(obs1,"==1 |",obs2,"==1 ",
sep=""))), select=c("PD",obs1,obs2))
names(example) <- c("distance","obs1","obs2")
detected <- c(example$obs1,example$obs2)
example <- data.frame(object=rep(1:nrow(example),2),
distance=rep(example$distance,2),
detected = detected,
observer=c(rep(1,nrow(example)),
rep(2,nrow(example))))
if(removal) example$detected[example$observer==2] <- 1
return(example)
}
# extract data for observer 10 and fit a single observer model
stakes <- extract.stake(stake78,10)
ds.model <- ddf(dsmodel = ~mcds(key = "hn", formula = ~1), data = stakes,
method = "ds", meta.data = list(width = 20))
plot(ds.model,breaks=seq(0,20,2),showpoints=TRUE)
ddf.gof(ds.model)
# extract data from observers 5 and 7 and fit an io model
stkpairs <- extract.stake.pairs(stake78,5,7,removal=FALSE)
io.model <- ddf(dsmodel = ~mcds(key = "hn", formula=~1),
mrmodel=~glm(formula=~distance),
data = stkpairs, method = "io")
summary(io.model)
par(mfrow=c(3,2))
plot(io.model,breaks=seq(0,20,2),showpoints=TRUE,new=FALSE)
ddf.gof(io.model)