distance sampling of butterflies

[Amy Pocewicz]

I am beginning a study of butterfly density, using transect walks in
forested areas. I would like to use distance sampling rather than the belt
transects that are traditionally used. The only example in the Distance
Sampling book of moving objects is seabirds, and I'm wondering if anyone has
implemented the method in the field with butterflies and may have some
advice to share or practical suggestions for sampling.



I'm also interested in incorporating "looking behind" on the transect. I
believe this may be useful for butterflies as some may not "flush" until
after the observer has passed. During one of my first surveys this week, an
assistant watched a butterfly fly behind me and over my head. Had it not
then crossed in front of me, I would have missed it even though it was
directly on the transect line. However, I'm concerned that by looking
behind I would miss butterflies ahead or increase the chances of
double-counting an individual. I will not always have the luxury of two
observers in the field, so it would be difficult to rely upon a two-observer
protocol.



Thank you for any ideas or advice,



Amy



~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Amy Pocewicz, Ph.D. Candidate

Department of Forest Resources

University of Idaho, Moscow, ID 83844-1133

poce...@uidaho.edu

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

[Aaron Ellingson]

Hi Amy,

I have worked out seemingly good distance protocols for butterflies in
alpine environments. I would be happy to send you excerpts from my
dissertation if you want. But you seem to have identified the main
points.

1. Movement. The usual distance protocols are clear for butterflies that
were initially at rest. Flying butterflies are more difficult. I chose
to count butterflies only when the passed through a plane perpendicular to
the transect and at arms-lenght in front of the observer. Analagous to
"coming astern" in shipboard surveys. Only counting in this plane also
serves to reduce positive bias incurred if butterfly movement is fast
relative to the observer (this depends on how you measure it). Counting
butterflies at arms lenght helps one to observe their position before they
respond to the observer.

2. g0 is very important. I suggest using something like a net to sweep
the vegetation in front of you on the line, insuring that all butterflies
flush. Sweep some to the sides as well so that you enforce a decent
shoulder to g(x). Don't sample in weather that precludes flushing (cold
or wet). The problem with the "look behind" method is the butterfly has
already moved in response to the observer, a no no. Continue to evaluate
this assumption with a few double-observer surveys, record the results as
you will be asked about this...

3. Grouped distances are essential. Choose bins that are narrow near
observer and wide further away. I used 0, 0.5, 1, 1.5, 2, 3, 5 m as
cutpoints, pooling groups later if necessary.

4. FInd a method to reduce error in distance measurment. I used a 3m rod
marked with the intervals. Visual estimates are poor.

5. Plot your data early on and make sure you have a good shape to g(x). I
see a tendency for spiked detection functions unless some effort is made
in the field to enforce a shoulder. Search the line, but also next to the
line.

6. Plan for variation in detection probability by observer and habitat.
Butterfly species-specific detection can probably be adequately handled by
pooling speices according to size (small, meduin, large).

I also have some data shared by Leslie Reis (U Northern Arizona) using
point transect for butterflies. She marked the distance intervals with
flagging on the ground. It worked well, but lines are still more
effiicent.

A few papers in the area (sorry I don't have citations with me) include
Dave Boughton's (Am Nat?) looking at source-sink dynamics and Boyce and
Brown on the Karner Blue. Let me know if you can't track them down, I can
find them in my other office. Neither is an ideal model for distance
sampling of butterflies, but a start nonetheless.

Good luck!!

~aaron


Aaron R. Ellingson
Statistician
Brown Treesnake Project
USGS / Johnson Controls World Sevices
Fort Collins Science Center
2150 Centre Avenue, Bldg C
Fort Collins, CO 80526-8118
Ph: 970.226.9464 Fax: 970.226.9230





Amy Pocewicz <poce...@UIDAHO.EDU>
Sent by: "Wildlife abundance estimation using distance sampling
techniques." <DISTANCE...@JISCMAIL.AC.UK>
04/07/2004 08:39 AM
Please respond to Amy Pocewicz


To: DISTANCE...@JISCMAIL.AC.UK
cc:
Subject: [DIST-SAMP] distance sampling of butterflies