[R] package unmarked: distsamp vs gdistsamp vs pcount vs gpcount

Rachel Field rdsfield at gmail.com
Sun Nov 16 19:56:37 CET 2014

Hello All,

I have been struggling with the following analysis in package 'unmarked':

I conducted repeated 10 min point counts over one season. I have 50 sites
(100 m radius) replicated over 3 surveys, and each individual observation
(n = 1108) is recorded at discrete distance intervals (i.e., the number of
observations for each site during each survey is not always equal). Habitat
variables were measured once for each site (n = 50), and detection
covariates were measured for each site on each survey (n = 150).

I wish to test the effect(s) of various habitat metrics on songbird
abundance/density, to include detection covariate(s) in my models, and to
account for repeated measures in my design. I think that 'distsamp' is most
appropriate for this, but am not sure (especially when it comes to how to
deal with repeated measures).

I have followed Chandler's 'Distance sampling analysis in unmarked (2011)'
and everything seems to work until I add detection covariates (using
distsamp; prior to adding abundance/density habitat predictors), when
running my models produce the warning: "*In lambda * A : longer object
length is not a multiple of shorter object length*".

(a) Am I using the appropriate fitting function (i.e., distsamp vs.
gdistsamp vs. pcount vs. ???)
(b) Why am I getting this warning message?

Here is my code:

dists <-read.csv("file/path.csv")

#sub-set of variables (to be used as detection covariates)

head(dists, 1108)

#'point' contains character+numerical site names (e.g., 'sweco03')
levels(dists$point) <- c(levels(dists$point), "sweco03")

#individual observations were recorded at 10 m distance intervals to 100 m
umf <-unmarkedFrameDS(y = as.matrix(yDat), survey = "point", dist.breaks =
  c(0,10,20,30,40,50,60,70,80,90,100), unitsIn = "m")

#to determine the best detection function
`hn_Null <- distsamp (~1 ~1, umf, keyfun = "halfnorm", output = "density",
unitsOut = "ha")
haz_Null <-distsamp (~1 ~1, umf, keyfun = "hazard")                #lowest
uni_Null <- distsamp (~1 ~1, umf, keyfun = "uniform")
exp_Null <- distsamp (~1 ~1, umf, keyfun = "exp") `

#to test the fit of detection covariates
model1 <-distsamp (~1 ~jdate, umf, keyfun = "hazard")
model2 <-distsamp (~1 ~daytime, umf, keyfun = "hazard")

#etc... When I attempt to run these models, I get the warning message.

Thank-you in advance,

	[[alternative HTML version deleted]]

More information about the R-help mailing list