distsamp: transect covariates and study design

184 views
Skip to first unread message

PaulS

unread,
Jun 26, 2012, 5:10:05 AM6/26/12
to unma...@googlegroups.com
Hello all,

   I am establishing line transects that we will drive with a vehicle and 2 observers to estimate ungulate abundance across a landscape with varying types of habitat (open grasslands to dense woodland).  A permanent river bisects the study area.  We want to use habitat and distance to river as transect-level covariates.  I have established 10 straight-line transects in GIS, varying in length from 4 - 25 km, but I hope to partition these into shorter sampling units based on habitat type and distance to river for analysis.  I have a few questions about how to set up the transects to analyze the data in distsamp.  These are kind of survey design questions with an objective of using distsamp.  Hope this is appropriate...

1) Can I subdivide each of the 10 transects by habitat type and use each of these as a 'transect'?  For example, say I have a 4 km east to west transect, which is composed of 1 km riparian, 2 km grassland, and 1 km bushland.  I would prefer to analyze these as 3 separate, but contiguous, transects, but was concerned about issues of spatial autocorrelation.  Any suggestions of how to account for this in distsamp?

2) In some areas, we may have to meander away from a perfect straight line transect, making more of a curved transect.  Though we will not be following an established road, we will be using a path of lesser resistance.  I am trying to decide what would be an appropriate method to calculate the perpendicular distance from the transect.  Hiby and Krishna (2001) calculated the distance to the nearest portion of the curve.  We were considering taking the vehicle's bearing, the bearing of the animals, and the sighting distance to calculate the perpendicular distance from our point on the curve.  I'd be interested in any suggestions any of you might have.

3) Any plans for multi-season distsamp models?  We plan to continue our animal counts every 8 weeks for 3 years.  I could do multiple, single-season analyses, but thought I'd ask out of curiosity.

Thanks again for developing these models and putting together great manuals, forums, and workshops.  I just received news that one of my multi-season occupancy modeling papers using the colext model was accepted for publication.  More details to come soon!  Thanks for your help.


Andy Royle

unread,
Jun 26, 2012, 9:55:22 AM6/26/12
to unma...@googlegroups.com, unma...@googlegroups.com
hi Paul,
 (1) yes this seems reasonable to me. If you have good covariates then these might do a satisfactory job of accounting for the autocorrelation. In any case, autocorrelation is only ever a minor problem, not a major problem.

 (2) I don't know about that.

 (3) Yes, there is a multi-season version in unmarked now called gdistsamp()  ("generalized hierarchical distance sampling").  We have examples of its use in the workshop materials which you should be able to find online also.

regards
andy
 




From: PaulS <pasch...@gmail.com>
To: unma...@googlegroups.com
Date: 06/26/2012 05:11 AM
Subject: [unmarked] distsamp: transect covariates and study design
Sent by: unma...@googlegroups.com


Richard Chandler

unread,
Jun 26, 2012, 10:04:00 AM6/26/12
to unma...@googlegroups.com
Hi Paul,

1) You could try analyzing the segments as independent transects, but you would not be able to formally account for spatial correlation. There is a paper by Niemi and Fernadez (JABES 2010) that might be useful.

2) I think the issue of non-linear transects is discussed in one of the Buckland et al. books. To be safe, make sure you have the coordinates of travel route and the coordinates of the detected animals.

3) Yes. Actually, I have most of the code written for a model like the Dail and Madsen model but for distance sampling data. I don't know when it will be released though. Oh, I see that Andy just responded. gdistsamp could be useful too, but I think a Dail-Madsen-like model might be more appropriate for multi-year data.

Richard

_____________________________________
Richard Chandler, post-doc
USGS Patuxent Wildlife Research Center
301-497-5696



From: PaulS <pasch...@gmail.com>
To: unma...@googlegroups.com
Date: 06/26/2012 05:10 AM
Subject: [unmarked] distsamp: transect covariates and study design
Sent by: unma...@googlegroups.com





PaulS

unread,
Jul 15, 2012, 9:14:27 AM7/15/12
to unma...@googlegroups.com
I am a bit delayed, but I wanted to say thanks for your input.... This was really helpful!

Paul
Reply all
Reply to author
Forward
0 new messages