Problems estimating global abundance in R

90 views
Skip to first unread message

Paula GM

unread,
Jan 19, 2016, 7:35:02 AM1/19/16
to distance-sampling

Dear all,

I am using the package “dsm” in R to make density surface modeling of data collected from systematic line-transect surveys of 9 different years, for the same study area. For the DSM the detection function and cluster size have been fitted over all years. The question is:

How can I estimate global abundance for the whole data set, taking into account the annual variation in the environmental variables?

The package I am using now, estimates the MEAN abundance over these 9 years, calculating the mean abundance for each cell of the prediction grid with the model predictions of the 9 years separately.

I think that this is not a good way to predict global abundance over all years, because it does not take into account the spacial and temporal component of my data. So, does anyone knows how can I estimate the global abundance, avoiding use the mean of the annual predictions? Statistically, what is the best and right way to estimate it?

Thank you,
Paula.

Eric Rexstad

unread,
Jan 19, 2016, 8:01:03 AM1/19/16
to Paula GM, distance-sampling
Complicated question, with many possible answers.

Do you believe the detection function changes each year, do you believe the cluster size dynamics changes each year; do you believe the relationship of animal location and environmental covariates change each year?  If yes, then the flippant answer is to fit separate detection functions and spatial models to each years' data.  That approach is not commonly taken.

At the other extreme, if you believe the detection function, cluster size dynamics and covariate response does not change over the study, then fit a single detection and dsm model.  Build separate prediction grids using the annual values for the environmental covariates, produce annual predictions from the prediction grids, acknowledge the variation in the annual estimates. 

There are many more variants of those approaches that could be taken.  Others will doubtless offer you other suggestions.
--
You received this message because you are subscribed to the Google Groups "distance-sampling" group.
To unsubscribe from this group and stop receiving emails from it, send an email to distance-sampl...@googlegroups.com.
To post to this group, send email to distance...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/distance-sampling/4f5eae53-c8eb-4591-860a-15c6fd22bd23%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

-- 
Eric Rexstad
Research Unit for Wildlife Population Assessment
Centre for Research into Ecological and Environmental Modelling
University of St. Andrews
St. Andrews Scotland KY16 9LZ
+44 (0)1334 461833
The University of St Andrews is a charity registered in Scotland : No SC013532

Paula GM

unread,
Jan 25, 2016, 7:06:27 AM1/25/16
to distance-sampling
Thank you very much Eric,

I will consider your second advice as it gives more information and the results can be really interesting, although they aren't exactly what I was looking for.
Reply all
Reply to author
Forward
0 new messages