including offset variables for sampling area: density?

[Matt Giovanni]

Dear unmarked users-

We are using the pcount function in unmarked to model abundance from
bird count data sampled at transects varying in length and width (200m
max width and 800m max length). I'm specifying offset variables to
account for variation in sampling area/effort per transect, such as:

forb=pcount(~obs ~FORBDEN+offset(log(AREA)),umf)

Am I correct in interpreting model predictions as densities (rather
than relative abundance) given the inclusion of area-offset variables?

Secondly, I have been using log-transformed areas as offset factors
(e.g., Kery et al. 2005 EcoApps), but I'm wondering if and why this
log transformation may be preferred over transforming area data to
standard normal deviates as is typically done with other continuous
predictor variables for enhancing convergence.

Thanks for your input and expertise. Best regards,

Matt

[Jeffrey Royle]

hi Matt,
(1) yes, the interpretation is density
(2) log(area) is used because in such models it is natural that AREA
should be multiplicative on density. i.e., if lambda is the density
for a unit area then A*lambda should be the expected population size
for a sample unit of size A. Covariates are modeled on the log scale,
and thus log(A*lambda) = log(A) + log(lambda).
regards,
andy

[Matthew Giovanni]

Hi Andy.  Great.  Thanks for your help, as always.

Matt

____________________________________
Matt Giovanni
Postdoctoral Visiting Research Fellow
Canadian Wildlife Service
2365 Albert St., Room 300
Regina, SK S4P 4K1
306-780-6121 work
402-617-3764 mobile
Research website:
http://sites.google.com/site/matthewgiovanni/