analyzing track count data collected along transects

Forum for discussion of general questions related to study design and/or analysis of existing data - software neutral.

analyzing track count data collected along transects

Postby dvpopescu » Thu Nov 14, 2013 7:07 pm

hi phidot community,

I have a question about analysis options for a dataset of brown bear track count data collected across 3 seasons in Romania.

The goal of the study is to estimate density/abundance of brown bears using track count data from unmarked animals only. We selected 40 2-km long forest road segments, each located within 3x3 km cells of a grid overlaid to the study area (each cell contained a transect). Segments were located >2 km distance from each other (this distance was based on the maximum daily movements of 1.5 km estimated from an independent telemetry dataset).

Each segment was surveyed 3-4 times/season (at 1 week intervals, usually after fresh snow) across 3 seasons (2 x 1 month of Fall/Winter [pre-hibernation] and 1 month Spring [post-hibernation], when bear movements are of low magnitude; hence the 2 km between transects above). Fresh bear tracks (<24 hrs old) were counted and measured (track data was vetted based on 4 separate measurements as to remove double counting of individuals within a transect). Because the transects were segments of forest roads, track data was also collected along the entire length of the road, not only on the established transects.

My current analysis approach was to use the Royle N-mixture models for count data on the 2-km transect ONLY (the assumption is that transects were independent within a survey occasion (i.e., were far enough apart that the same bear was not counted on neighboring transects within a survey occasion), but this assumption might not hold between survey occasions). This approach does not make use of the data collected between the established transects either (which could be useful information), as well as any sort of spatial autocorrelation.

The next issue is inferring bear density across the study area from the "abundance per transect" estimated through the Royle N-mixture models. The 3x3 km grid was selected arbitrary, having to do more with spacing the transects, rather than reflecting home range sizes. We do have an independent dataset on home range sizes and movements during the survey months from 10 bears (not all within the study area), which could be useful. Thus far I simply used the home range size information to set the effective sampling area of each transect = mean home range size during the month of the survey (which is 14kmp, thus greater than the 3x3 km cells).

I was wondering if there are any suggestions of alternative ways to analyze the data, or perhaps pointing to to other issues that we didn't consider yet.

Thank you...

Cheers,
Viorel
dvpopescu
 
Posts: 2
Joined: Wed Nov 13, 2013 9:13 pm

Return to analysis & design questions

Who is online

Users browsing this forum: Google [Bot] and 18 guests