suitability of data for point sampling approach (/forgotten all my DS theory)

[Ollie Bartlett]

I'll be upfront straight away and admit its been a long time since i have done any distance sampling. I was recently approached by an NGO in Costa Rica to create a predictive spatial model for howler monkey electrocution incidents. unfortunately the data they have generated are somewhat unsuitable for the task however the more i look at the observations the more im wondering if another question can be answered.

The observations (five years of howler monkey fatalities post contact with electrical lines) are uploaded by members of the public to an online survey, however the spatial distribution of observations around the activity centers of a collection of NGO's follows a near perfect distance decay model from human geography studies - as distance increases interactions between individuals decrease. this leads me to believe the observations are spatially biased towards only occurring in areas where the NGOs operate and have the ability to spread knowledge of the platform.

given that both total suitable habitat and total length of electrical lines across the wider area area can be known (plus a number f other geospatial variables), it seems to me that the total number of electric shock incidents per year could be estimated from the data as long as one works on the assumptions that observations can only occur if an observer knows about the online platform, and that any given individual is more likely to be aware of the platform the closer they reside to the activity center of one of the ngos that promote it.

i would be interested to hear other peoples opinions about whether this is a valid case for the use of a point sampling approach to estimate total number of annual incidents. there are 150 observations over a number of years, spatially distributed around the four NGO's activity centers. those activity centers are biased towards coastal areas however i don't believe that would make it any more or less likely that incidents occur there than inland.

any thoughts as to whether this might be a valid use of the data?

many thanks to anyone who can help

[Eric Rexstad]

Ollie

I perhaps don't have anything relevant to contribute. It seems to me this isn't as much a distance sampling issue as it is a sampling/extrapolation issue. From what I gather, you have four NGO activity centres you wish to treat as point transects; from which you intend to extrapolate fatality density to the remainder of your study area.

You mention a predictive spatial model, but that is not very many activity centres upon which to base your extrapolation. I would also guess that electrical line density (hence potential for monkey contact with them) is higher near the NGO centres, further complicating the inference you might derive from your proposal.

---

From: distance...@googlegroups.com <distance...@googlegroups.com> on behalf of Ollie Bartlett <ollie...@gmail.com>
Sent: 17 August 2023 19:01
To: distance-sampling <distance...@googlegroups.com>
Subject: {Suspected Spam} [distance-sampling] suitability of data for point sampling approach (/forgotten all my DS theory)

 

--
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/cb8ca462-9bb0-4d37-8302-d4f5af0fe120n%40googlegroups.com.

[Tiago Marques]

Hi Ollie,

Sounds like a creative way of seeing your data and framing it into a Distance Sampling (DS) context, but it is not void of dangers.

I see Eric's point about the distribution of electric lines not being random over space as a major issue. If you could quantify the number of power lines over space, then you could presumably have DS without the uniform assumption, but that would require a more sophisticated analysis. And it depends a lot on something else Eric noted, how much of the area for which you are making inferences lies around these NGO's you have sampled vs is completely out there and massive extrapolation is therefore required for areas where there are no NGO's whatsoever.

Further, it is unclear to me what might be the g0 effectively, in other words, a conventional DS analysis will assume that all incidents that occur in the near vicinity of the NGO's are reported. I have no idea whether this is a fair assumption or not, in particular if it is possible that some animals which might die but are never seen again after the initial incident.

Cheers,

Tiago

To view this discussion on the web visit https://groups.google.com/d/msgid/distance-sampling/DBAPR06MB6694795B1CE42E4FF74B9764EA1BA%40DBAPR06MB6694.eurprd06.prod.outlook.com.