This post seems to conflate a number of issues, and I’m not clear which is the biggest problem. If the issue is over dispersion (e.g., if the vast majority of short segments contain zero sightings), it’d be worth trying a Tweedie distribution
in addition to trying longer segments. That helped in two of our studies that had a lot of zeroes:
http://www.nature.com/articles/srep04170
If there are many sightings, most segments contain sightings, and the poor fit is due to variability in environmental conditions along long segments, or lack of variability in environmental conditions among short segments, that’s a different problem
with a different solution. If it’s a tiny study area, maybe environmental conditions didn’t vary much throughout the survey region, and you need more informative covariates. Maybe that’s poor model fit is the answer — variability in X across the range of
values of X you observed, doesn’t explain much variability in encounter rate.
Choice of segment length was raised as a priority area for future research in Hedley’s original thesis:
http://www.creem.st-and.ac.uk/sharon/public_html/thesis.pdf
Hedley recommended checking how sensitive your model fit is to choice of segment length. If segment length is causing your answers to vary by a lot, is it worth choosing one of Hedley’s other methods, such as the waiting distances method, which is less reliant
than the count method on choice of segment length? Adrian raises a good point: density surface models relax a lot of assumptions about sampling design, but they still assume reasonable coverage across the range of all covariates in the model. If that assumption
was violated, I could imagine poor fit and wild extrapolations.
I think a bit more detail is needed to narrow down the laundry list of potential explanations, so people can offer more helpful advice.
All the best,
Rob