Estimating abundance for different regions with different areas

[Kirstie Hazelwood]

Dear All,

I am working with data from mammal line transects for two regions, and I want to know what the difference in abundance between the two sites is for various different groups of species. I have divided my species into 8 functional groups, and, using 'Distance2', I have come up with detection function models for each of the functional groups. Not all functional groups fit the same detection function, since some have very relevant covariates and others don't (e.g. for some species group size is very important, but others are only seen alone so group size is irrelevant). However I expect detection functions to be the same for each site, only actual abundance to be different, so detection function models are across both sites without site as a covariate. I now want abundance estimates for each region but I am hitting a couple of problems:

1. Some of the sample sizes are small for an individual site (e.g. large primates were seen in big groups every day at one site, and only one was ever seen at the other site. This is of course very important information but not very good for an abundance estimate). If I use 

>abundance_estimate(ds.model.sp1, ~Region.Label) # Region label = Site

will I get abundance estimates for each site without having to model each site and functional group separately?

2. I haven't been able to test the above because one of the sites has varying Areas since some of the transects were shorter than others. This seems to be throwing the function off. Should I manually divide group size by Area before setting out to resolve this? Or is there a clever way that Distance2 can deal with this?

Thanks in advance,

Kirstie

[David Lawrence Miller]

Hi Kirstie, hi listfolk,

Thanks for trying out Distance2! It's very much in an experimental stage
at the moment, so I'd recommend sticking to Distance/mrds for now, at
least for any publication/report-level analysis.

An example analysis of stratified data with Distance can be found at:

http://distancesampling.org/R/vignettes/minke.html

(with other examples at http://distancesampling.org/R)

Hope this helps!

cheers,
--dave

On 19/04/2017 10:43, 'Kirstie Hazelwood' via distance-sampling wrote:
> Dear All,
>
> I am working with data from mammal line transects for two regions, and I
> want to know what the difference in abundance between the two sites is
> for various different groups of species. I have divided my species into
> 8 functional groups, and, using 'Distance2', I have come up with
> detection function models for each of the functional groups. Not all
> functional groups fit the same detection function, since some have very
> relevant covariates and others don't (e.g. for some species group size
> is very important, but others are only seen alone so group size is
> irrelevant). However I expect detection functions to be the same for
> each site, only actual abundance to be different, so detection function

> models are across both sites /without/ site as a covariate. I now want

> abundance estimates for each region but I am hitting a couple of problems:
>
> 1. Some of the sample sizes are small for an individual site (e.g. large
> primates were seen in big groups every day at one site, and only one was
> ever seen at the other site. This is of course very important
> information but not very good for an abundance estimate). If I use
>
>>abundance_estimate(ds.model.sp1, ~Region.Label) # Region label = Site
>
> will I get abundance estimates for each site without having to model

> each site /and /functional group separately?

>
>
> 2. I haven't been able to test the above because one of the sites has
> varying Areas since some of the transects were shorter than others. This
> seems to be throwing the function off. Should I manually divide group
> size by Area before setting out to resolve this? Or is there a clever
> way that Distance2 can deal with this?
>
>
> Thanks in advance,
> Kirstie
>

> --
> 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
> <mailto:distance-sampl...@googlegroups.com>.
> To post to this group, send email to distance...@googlegroups.com
> <mailto:distance...@googlegroups.com>.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/distance-sampling/43f031c2-389a-4948-a4c5-7490c6b685d7%40googlegroups.com
> <https://groups.google.com/d/msgid/distance-sampling/43f031c2-389a-4948-a4c5-7490c6b685d7%40googlegroups.com?utm_medium=email&utm_source=footer>.
> For more options, visit https://groups.google.com/d/optout.

[Kirstie Hazelwood]

Hi Dave,

Thanks for your reply, it's a shame not to use Distance2 since I liked the package but it's not a problem to use Distance. Unfortunately I get the same problem with Distance:

Error in checkdata(data, region.table, sample.table, obs.table, formula) : 

  Region areas are not consistent.

It's not lying of course, my region areas are not consistent since some of my transects were cut short. I don't want to model these shorter transects separately since this will be a very small sample size. At the moment I have divided my 'group size' by my 'Area' and make every sighting 'monkeys per sq km', removing the area column (or making it all 1km), this is the only solution I have come up with. I have not found a way that Distance can deal with these types of anomalies.

Thanks for your thoughts,

Kirstie

[David Lawrence Miller]

Hi Kristie, hi listfolk,

Kristie: please feel free to send me your data off-list and I can take a
look.

One quick thing that might be the issue is what you're calling region
area. The region areas should be the stratum areas (the total areas you
want to estimate abundance for). So I'm not 100% sure of what you mean
about the transects being cut short.

As I say I can take a look off-list if that's useful and we can post our
conclusions back here.

cheers,
--dave

> /without/ site as a covariate. I now want abundance estimates for

> each region but I am hitting a couple of problems:
>
> 1. Some of the sample sizes are small for an individual site (e.g.
> large primates were seen in big groups every day at one site, and
> only one was ever seen at the other site. This is of course very
> important information but not very good for an abundance estimate).
> If I use
>
> >abundance_estimate(ds.model.sp1, ~Region.Label) # Region label = Site
>
> will I get abundance estimates for each site without having to model

> each site /and /functional group separately?

>
>
> 2. I haven't been able to test the above because one of the sites
> has varying Areas since some of the transects were shorter than
> others. This seems to be throwing the function off. Should I
> manually divide group size by Area before setting out to resolve
> this? Or is there a clever way that Distance2 can deal with this?
>
>
> Thanks in advance,
> Kirstie
>

> --
> 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
> <mailto:distance-sampl...@googlegroups.com>.
> To post to this group, send email to distance...@googlegroups.com
> <mailto:distance...@googlegroups.com>.
> To view this discussion on the web visit

> https://groups.google.com/d/msgid/distance-sampling/5ba78aa6-5283-4854-bece-be241f91dacc%40googlegroups.com
> <https://groups.google.com/d/msgid/distance-sampling/5ba78aa6-5283-4854-bece-be241f91dacc%40googlegroups.com?utm_medium=email&utm_source=footer>.