Post-stratifying by Survey Stand and Time Period in R

96 views
Skip to first unread message

Leah Breidinger

unread,
Jan 11, 2018, 5:24:04 PM1/11/18
to distance-sampling

Hello –

 

My project is examining how old-growth maintenance logging cuts affect the densities of birds.   I am using a Paired BACI design with old-growth timber stands that were logged paired to control stands that have similar characteristics, but were not logged (3 logged stands, 3 control stands).  Point count locations were randomly assigned in each stand with limitations on minimum distances between adjacent points and between points and stand edges.  I collected bird data 2x year for 6 years with data collected before and after logging occurred.  Similar to Buckland et al (2009 - https://doi.org/10.1198/jabes.2009.08030) I would like to fit one detection function for all observations of each species of interest and then use that detection function to estimate abundances for each study stand pre- and post-harvest.  Note, logging was completed in different years in each stand, but there is at least 2 year of pre-harvest and 2 year of post-harvest data for each stand.  I have been using ‘Practical distance sampling in R’ and have found it very helpful, but I’m stuck on a few things:

 

  • Is it possible to truncate 10% of observations in the R version of Distance?

  • Is it possible to fit a detection function in R and then post-stratify the data by both the Stand ID and Time Period (Pre/Post Harvest) so that I can get abundance estimates for each study stand both pre- and post-harvest?  Should this be done by creating a ‘Region.Label’ field that includes both the Stand Id information and the Time Period (Pre/Post Harvest) information?  If I restructure the data in this way would I also need to adjust the effort and visit fields to reflect how many surveys were conducted in each stand pre- and post-harvest?  Initially I tried taking a subset of the pre-harvest data and then applying the model I had selected based on all of the data, but this doesn’t seem like the right solution.

 

I also attached data from one of my survey stands so you can see how the data is currently structured.  Any advice that you can provide would be great, thank you!

 

Leah

SRSF_distance_data_C1.csv

David Lawrence Miller

unread,
Jan 16, 2018, 4:28:53 AM1/16/18
to Leah Breidinger, distance-sampling
Hi Leah, hi listfolk,

To your questions...

- Is it possible to truncate 10% of observations in the R version of
Distance?

Yes! You can specify truncation="10%" (see ?ds documentation)

- Using Stand ID and Time Period to stratify abundance estimates.

Using Region.Label seems like the right thing to do here. Since you're
only going to change the labels (Region.Label) and the Area column
(which is the area that you are estimating abundance for), you don't
need to do anything to the effort or visit fields.

Something like

data$Region.Label <- paste0(data$Region.Label, "-", data$Harvest)

should work? (Assuming your current Region.Label is the Stand ID and
Harvest is the time period.)

Feel free to contact me off-list if you have further issues and we can
post back to the list any further clarifications.


cheers,
--dave



On 11/01/2018 22:24, Leah Breidinger wrote:
> Hello –
>
> My project is examining how old-growth maintenance logging cuts affect
> the densities of birds.I am using a Paired BACI design with old-growth
> timber stands that were logged paired to control stands that have
> similar characteristics, but were not logged (3 logged stands, 3 control
> stands).Point count locations were randomly assigned in each stand with
> limitations on minimum distances between adjacent points and between
> points and stand edges.I collected bird data 2x year for 6 years with
> data collected before and after logging occurred.Similar to /Buckland et
> al/(2009 - _https://doi.org/10.1198/jabes.2009.08030_
> <https://www.google.com/url?q=https%3A%2F%2Fdoi.org%2F10.1198%2Fjabes.2009.08030&sa=D&sntz=1&usg=AFQjCNGgAV4LvMmmTAAMkD2vWTrpAEtNLA>)
> I would like to fit one detection function for all observations of each
> species of interest and then use that detection function to estimate
> abundances for each study stand pre- and post-harvest.Note, logging was
> completed in different years in each stand, but there is at least 2 year
> of pre-harvest and 2 year of post-harvest data for each stand.I have
> been using ‘Practical distance sampling in R’ and have found it very
> helpful, but I’m stuck on a few things:
>
> *
>
> Is it possible to truncate 10% of observations in the R version of
> Distance?
>
> *
>
> Is it possible to fit a detection function in R and then
> post-stratify the data by both the Stand ID and Time Period
> (Pre/Post Harvest) so that I can get abundance estimates for each
> study stand both pre- and post-harvest?Should this be done by
> creating a ‘Region.Label’ field that includes both the Stand Id
> information and the Time Period (Pre/Post Harvest) information?If I
> restructure the data in this way would I also need to adjust the
> effort and visit fields to reflect how many surveys were conducted
> in each stand pre- and post-harvest? Initially I tried taking a
> subset of the pre-harvest data and then applying the model I had
> selected based on all of the data, but this doesn’t seem like the
> right solution.
>
> I also attached data from one of my survey stands so you can see how the
> data is currently structured.Any advice that you can provide would be
> great, thank you!
>
> Leah
>
> --
> 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/d8367464-ac76-412f-a5e8-87cd8d87a5da%40googlegroups.com
> <https://groups.google.com/d/msgid/distance-sampling/d8367464-ac76-412f-a5e8-87cd8d87a5da%40googlegroups.com?utm_medium=email&utm_source=footer>.
> For more options, visit https://groups.google.com/d/optout.

Leah Breidinger

unread,
Jan 17, 2018, 5:27:34 PM1/17/18
to distance-sampling
Hi Dave -

Thank you very much for your advice.  I will give it a try and see how it goes.

Leah

Leah Breidinger

unread,
Jan 23, 2018, 4:09:07 PM1/23/18
to distance-sampling
Hello -

Just a quick update for anyone who is interested in post-stratification using more than one variable.  I ended up creating a new Region.Label field that included both my stand information and time period information as Dave suggested.  For example,  one of my study timber stands was called SRSFC1 and I was looking at two different time periods (pre-harvest and post-harvest), so I assigned the study stand SRSFC1 two region labels - SRSFC1_Pre and SRSFC1_Post.  I also had to update the Sample.Label field to reflect the new Region.Labels.  Area did not have to be adjusted but I did have to make sure that I divided up the Effort information to reflect the number of times that I visited SRSFC1 before logging and after logging occurred. 

Once the data was restructured I was able to use the abundance estimate function in R to get estimates for each study stand pre- and post-harvest:

abundance_estimate(amro_hn, ~Region.Label)

Hope this is helpful.

Leah
Reply all
Reply to author
Forward
0 new messages