Multi-Method Occupancy

[bacollier]

All,
A student approached me with some questions on application of multi-method occupancy models and he has some data that looks amenable to the approach. Other than some 'it is possible' statements nested in reviews and such I cannot find anywhere it has actually been implemented in the literature (perhaps I am using the wrong search term?). I have a idea of how I think I could work it up but I wanted to check and see if anyone has published on the approach, tested simulated data, or whether it has been implemented in a yet to be available version of Presence before I attempted to answer the student's question?

Thanks,
Bret

[bacollier]

For anyone interested, a couple of colleagues emailed me this citation after my post.

Nichols et al. (2008). Multi-scale occupancy estimation and modellling using multiple detection methods. Journal of Applied Ecology 45: 1321-1329.

Bret

[bacollier]

Couple more:

O'Connell et al. (2006). Estimating site occupancy and detection probability parameters for meso- and large mammals in a coastal ecosystem. J. Wildlife Management 70: 1625-1633.

Manley et al. (2005). A field-based evaluation of a presence-absence protocol for monitoring ecoregional-scale biodiversity. J. Wildlife Management 69: 950-966.

b---

[darryl]

Bret,
It depends on what types of replication you have and what biological questions are of interest as to which approach might be best.

If you only have 2 levels (eg sampling units and different methods within sampling units checked only once each 'season') then you can use the standard approaches just fine: the different methods are your repeat surveys.

If you have 3 levels (eg sampling units, multiple nights, and multiple methods) then the Nichols et al. (2008) approach could be useful where you're interested in estimating device specific detection probabilities, or when rather than having full closure you have random changes within a season and you want to estimate that random availability component as well. If you're primarily interested in occupancy of sampling units though then you could pool across methods (so each night species is detected by at least one method or none of them) and use standard methods with no loss of precision.

Cheers
Darryl

[jhines]

Bret,

The multi-method model is available in PRESENCE (and GENPRES).

Jim

[bacollier]

Darryl/Jim (and others who emailed me offlist),

Thanks for the insight/manuscripts, seems pretty straightforward to implement.

Bret

[shar3840]

I recently conducted sign surveys to look at habitat associations of my species using occupancy models. During the surveys I documented whether the sign was scat or tracks. I originally combined the sign types for my habitat analysis but recently got interested in the detection probability of the sign types (which I found are quite different). Is there an advantage to now run my habitat analysis with the multi-method model or is it still okay to combine the sign types? Thanks.

-Mackenzie