Great apes nest surveys: searching around nests

[haur...@gmail.com]

Dear all,

We are launching in Republic of Congo a wildife survey including distance sampling with nests of chimpanzees and gorillas.

Chimps and gorillas live in groups (most of the time) and individuals from a group build their nests in the vicinity of each other. After attending the Distance workshops in 2015, I moved from nest site analysis to nest analysis (considering the uncertainty of nest site size estimation being more important than the violation of observation independency, and that it allows to have more observations).

My basic protocol is : when a nest is detected, stay on the transect and walk for 50 m further to check if other nests are detected. When no more nests are detected from the transect, go back to the first detected nest to measure all perpendicular distances, keeping well in mind and discriminating in the collected data which nests were detected from the transect and which ones were not (in case other nests are detected when leaving the transect to measure perpendicular distances). Then, the Distance analysis is ran using only the nests detected from the transect.

We discussed this methodology with a project partner. Their protocol is quite different: when detecting a nest, observers leave the transect to search the surrounding on a radius of 50 m to detect all nests in the vicinity (including those not visible from the transect). They measure the perpendiculat distances from every nest observed (including those not detected from the transect), and use all these observations to run the Distance analysis.

From what I understand of Distance, this second method is not correct, as observers leave the transect and you consider indices that are not detected from it. Therefore, the detection function you build is not the real detection function. However, it seems that a lot of personns working on great apes census are using this second protocol.

Could you help me in explaining me (1) if this second protocol is correct or not? and (2) if it is correct, how the rationale of distance could be respected while considering observations that are not detectable from the transect?

Thank you very much!

Barbara

NB: I searched the forum and did not find any previous explanation on the subject, sorry if I missed it!

[Stephen Buckland]

Either approach should be fine.  In your approach, the detection function is exactly that – the probability of detection of a nest from the line, as a function of distance from the line.  In their approach, the detection function doesn’t have that interpretation.  And the only reason that their approach works is the ‘pooling robustness’ property of distance sampling, which is described and investigated in

 

Rexstad, E., Buckland, S., Marshall, L. and Borchers, D. 2023.  Pooling robustness in distance sampling:  avoiding bias when there is unmodelled heterogeneity.  Ecology and Evolution 13, e9684.

 

It would be interesting to do a comparative study, to see how the 2 approaches compare wrt precision achieved for the same level of resources.

 

Steve Buckland

--
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 view this discussion on the web visit https://groups.google.com/d/msgid/distance-sampling/9e8104ce-0f67-4e6c-a972-cbb94d2196f2n%40googlegroups.com.