Dolphins, temp migration and robust design

[Susana García]

Hola,

I'm trying to run robust design model to estimate size of a population of dolphins with temporary e-inmigration. Is there any easy way to learn and test the statistical asummptions of the model???? (books, understandable papers, etc....)

Thanks so much from the sunny Spain...
Susana

[cschwarz@stat.sfu.ca]

The papers by Kendall et al and Schwarz et al outline the basic ideas of the model. Section 19.4.1 of the Williams et al book has a chapter on it.

Before going too far along this line, there are many types emigration possible and information on the study design would be helpful.

Carl Schwarz cschwarz@stat.sfu.ca

[Susana García]

Hola again,

Thanks for the quick answer,

We've been using Photo-identification to identify and resight the animals. I have 9 years of data with a different number of trapping-photographing sessions each year (ranging from 4 to 8 sessions). Some of the animals have been identified in year 1 and resighted in years 3,4, and 6 and never again... some others were identified in year 2 or 3 and dissapear and appear years after without any pattern. There are several reasons to explain this; (1)The study area doesn't cover the whole distribution range of the population, (2) the animals go outside the study area and come back later, (3) the animals stay in the study area but aren't photographed in certain occassion due to shy behaviour towards the boat or any other reason....But i can't distinguish if only 1, 2 or the 3 aspects are actually occurring in my case.

Some experienced people suggested the robust design model to account for this "movements", so I'm just learning how this model works and if I can use it.....

Susana

[cschwarz@stat.sfu.ca]

It may not be necessary to use the robust design as a preliminary step. Simply pool all the sightings from a year into one "super sighting" and use the usual Jolly-Seber model on the resulting estimates.

The pattern of animals appears for a year and then not being seen, may simply be random chance caused by the non-100% sightability in any particular year.

If the study doesn't cover the entire range, you will have to assume that the dolphins move in and out from the range other wise you won' get an estimate of the population of interest.

It really doesn't matter why dolphins aren't seen, as long as the "not being seen" is random and doesn't depend upon past behavior.