Unreal density with DS

[alei...@gmail.com]

Hello,

We used DS on capercaillie (Tetrao urogallus) lek count data. It should be noted that the surveys were not specifically designed for distance sampling.

During lek counts we measured all detection distance and implemented it in Distance v7.3 as a point transect following the recommendation of Thomas et al. (2010).

On the other hand, we also estimated the population size of these leks by collecting, genotyping faeces and using CMR (N-mixture) models.

Both approaches were performed in the same surface area defined by the maximum detection distance MDD of a capercaillie. As we expected to have a comparable density with the two approaches. However, the densities obtained with DS are overestimated, in fact, they were twice as high as the densities estimated using CMR models.

We do not understand why, as with CMR models all males are sampled (singing and non-singing males by the faeces) and with DS only the density of singing males was estimated, so the density of DS should be lower than that of CMR models.

we have different hypotheses to explain this:

1- Violation of the random distribution of the counting points, as they were all placed in the centre of the lek, to ensure to detect individuals. In fact, all points detect capercaillies.

2- Possible impressions of the measurements of the detection distances, we have measured them very carefully but being in a forested area it is possible that there are errors. We have tried to correct this by applying classes of 10m by 10m but the result is still the same.

3- The hearing acuity of the observers. All observers are experienced in this type of sampling, but some have much better hearing acuity than others. Therefore, some detect individuals at a much greater distance and earlier than others.

 Can you help me to explain these differences or is distance sampling "simply" not adapted to calculate capercaillies density during lek counts?

Thank you very much for your help.

[Eric Rexstad]

The answer to your question is not obvious from the clues you provide.

Some common items that are overlooked include making sure the units of measure for radial distances and for areas are properly entered into the software.

You note placement of your point transects was non-random, but instead were placed at the centre of leks where capercaillie density is presumably very high.  By only sampling where animal density is high, it seems abundance estimates will be large.

Another problem that can lead to overestimation is data that has a "spike" (numerous detections) near 0.  Fitting models such as the negative exponential can result in unreasonable detection functions that fall away too quickly, leading to very low estimates of detection probability that, in turn, results in very large estimates of abundance.

If you want further insight, we could discuss in greater detail "off list", reporting any discoveries back to the list.

---

From: distance...@googlegroups.com <distance...@googlegroups.com> on behalf of alei...@gmail.com <alei...@gmail.com>
Sent: 10 May 2022 17:01
To: distance-sampling <distance...@googlegroups.com>
Subject: {Suspected Spam} [distance-sampling] Unreal density with DS

 

--
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/8864ef76-c2a0-43ce-af3d-be6e2ed522acn%40googlegroups.com.

[Eric Rexstad]

After some conversations with Gaël, we learned the point transect data set contained <40 detections. Detection function models has poor fit (chi square P-values<0.05).  The resulting abundance estimates had CV(abundance)>0.25. The resulting confidence interval around the abundance estimate (just) overlapped the confidence interval from the CMR estimate.

---

From: distance...@googlegroups.com <distance...@googlegroups.com> on behalf of alei...@gmail.com <alei...@gmail.com>
Sent: 10 May 2022 17:01
To: distance-sampling <distance...@googlegroups.com>
Subject: {Suspected Spam} [distance-sampling] Unreal density with DS

 

Hello,

--