How to correct field compass bearing mistakes? A quick query

[Samundra Subba]

Hi,

Would appreciate if anyone could help me understand why my histogram look similar to this figure below. Survey was conducted by line transect for prey species- ungulates.

What does this mean? Is left truncation the only way? I believe surveyors weren't able to walk in a straight line using compass and the detections may have unequal perpendicular distance from the line. Is that the main reason for this particular survey? 

Thank you,

Best wishes,

Samundra

[Eric Rexstad]

Samundra

There are many reasons that could give rise to the observed distribution of perpendicular distances you show. Most common reasons for this distribution are a) evasive movement on the part of the animals or b) distribution of animals available to be detected is lower at small distances from the transect.  Cause b) is often associated with transects placed along features known to the animals; such as tracks.

If you have eliminated these possible causes, only then can you consider other explanations. Transects need not necessarily be straight, but perpendicular distances do need to be recorded to the transect; this is most easily accomplished with straight transects.

Even if it is correct that compass bearings are at fault, I do not know how you would correct the recorded data. Corrections you might make would also be based upon supposition; thereby introducing new uncertainty into the data.

If the cause of the problem is (b) above, then left truncation is not the remedy. See the short note published by Marques (2016) in response to a paper that employed left truncation in an ungulate survey:

• Marques, T. A. (2016). A comment on Horcajada-Sánchez and Barja (2015): A cautionary tale about left truncation and density gradients in distance sampling. Annales Zoologici Fennici, 53(1/2), 52–54.
  
  

---

From: distance...@googlegroups.com <distance...@googlegroups.com> on behalf of Samundra Subba <samundr...@wwfnepal.org>
Sent: 29 August 2024 12:46
To: distance...@googlegroups.com <distance...@googlegroups.com>
Subject: [distance-sampling] How to correct field compass bearing mistakes? A quick query

 

--
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/TY0PR01MB5412B27708E8D2609E60D002E4962%40TY0PR01MB5412.apcprd01.prod.exchangelabs.com.

[Duane Diefenbach]

I would second what Eric suggested. My experience with ungulate distance sampling surveys (white-tailed deer, Odocoileus virginianus). The histogram you shared would be typical for surveys conducted from roads (because deer avoid the areas near roads) or because the survey was conducted on foot and deer were able to detect observers first and move away from the transect line. In the first case (roads as transect), you need independent data to model the distribution of deer relative to the transect (Marques et al. 2013 https://doi.org/10.1007/s10260-012-0223-2). In the second case (deer avoid observers) I'm not sure what you do except modify your data collection methods.

Duane Diefenbach

To view this discussion on the web visit https://groups.google.com/d/msgid/distance-sampling/DBAPR06MB6694BB06EA06BF8F3F0E7BA2EA962%40DBAPR06MB6694.eurprd06.prod.outlook.com.

[Samundra Subba]

Hi Eric and Duane,

Thank you so much for your insights, these are very helpful. Yes my problem is the latter, as this was not conducted on trails/roads rather random line transects drawn on a map. We followed all principles of line transect- distance sampling. However, due to the habitat composition of the study area as some were difficult to traverse in a straight line, the transects might have oriented a little further. I corrected the perpendicular distances though, by calculating the distance from the point (we had GPS location) where the prey data was observed to the Euclidean transect (where it supposed to be) by near tool in arcgis and also by compass animal bearing (recorded during the field). So this helped to either add or subtract the distances. Now my rough or data exploration graph looks like the one below.

Am I doing it correctly?

Best wishes,

Samundra

---

From: Duane Diefenbach <duan...@gmail.com>
Sent: Thursday, August 29, 2024 6:40 PM
To: distance...@googlegroups.com <distance...@googlegroups.com>
Cc: Samundra Subba <samundr...@wwfnepal.org>
Subject: Re: [distance-sampling] Re: How to correct field compass bearing mistakes? A quick query

 

[Eric Rexstad]

Yes, Samundra. You want to record the perpendicular distance from the animal to the transect, not distance from animal to observer. Observers might deviate from the transect, but the distance of interest is between the animal and the transect.

Your histogram of detection distances looks reasonable; however, I would carry out truncation of the largest detection distances before fitting detection function models. I cannot read the x-axis scale because the test is small. But I think certainly the last two histogram bars, and perhaps also the third from greatest histogram bar distances might be truncated. The model being fitted is to the process that gives rise to the data (namely the detection process). Beyond some perpendicular distance, detections are influenced more by random chance than by the detection process. These most distant detections are removed such that noise in the data has less influence in the fitted model.

---

From: distance...@googlegroups.com <distance...@googlegroups.com> on behalf of Samundra Subba <samundr...@wwfnepal.org>

Sent: 02 September 2024 08:21
To: Duane Diefenbach <duan...@gmail.com>; distance...@googlegroups.com <distance...@googlegroups.com>

To view this discussion on the web visit https://groups.google.com/d/msgid/distance-sampling/TY0PR01MB54128D5FF60C03CF3E06822AE4922%40TY0PR01MB5412.apcprd01.prod.exchangelabs.com.

[Samundra Subba]

Hi Eric,

 

Thanks for verifying and also your insights regarding detections in far distances. Will use truncation for sure.

 

Thanks for all the help, appreciate it.

 

Best wishes,

Samundra

[Stephen Buckland]

I’m not sure there is much of an issue here. The histogram suggests some avoidance in the first 10 or 12m. If the goodness-of-fit test is not significant, that could be just chance variation and I would ignore it. If the test is significant, it may be that animals move slightly away from the line before detection. If these animals are still being detected, but in the second or third distance interval instead of the first, then the fitted detection function averages out the slight excess of detections in those 2 intervals with the slight shortage in the first interval, and estimation should not really be compromised. Only if many animals on or near the line move and avoid detection altogether might the bias become problematic.

Steve Buckland 

---

From: distance...@googlegroups.com <distance...@googlegroups.com> on behalf of Samundra Subba <samundr...@wwfnepal.org>

Sent: Tuesday, September 3, 2024 7:13 AM
To: Eric Rexstad <Eric.R...@st-andrews.ac.uk>; distance...@googlegroups.com <distance...@googlegroups.com>
Subject: RE: [distance-sampling] Re: How to correct field compass bearing mistakes? A quick query

 

To view this discussion on the web visit https://groups.google.com/d/msgid/distance-sampling/TY0PR01MB54120F64965E3B812ECAA4E8E4932%40TY0PR01MB5412.apcprd01.prod.exchangelabs.com.