Techniques for estimating wood duck density or abundance have not been
well-developed. This study was conducted to (1) evaluate the feasibil
ity of using line-transect sampl ing to estimate density of breeding w
ood ducks in forested wetlands, (2) determine sample size required for
obtaining estimates of breeding wood duck density at various levels o
f precision, and (3) evaluate whether wood duck behavior and field met
hods are compatible with the assumptions of line-transect theory. Samp
ling was conducted on Mingo National Wildlife Refuge in southeastern M
issouri. I established 8.23-10.31 km of transects in forested habitat
and sampled on 9 dates between 7 March and 2 April, 1989 and 1991. I a
nalyzed sighting distance and sighting angle data using program DISTAN
CE and a Fourier series model. Wood duck density estimates ((D) over c
ap) ranged from 0.3-1.0 birds/ha, with associated coefficient of varia
tion (CV[(D) over cap]) ranging from 24-52%. Based on precision of den
sity estimates I determined that 30 km of transects would yield a CV((
D) over cap) of 20%, whereas a CV((D) over cap) of 5% would require ab
out 480 km of transects. The level of precision required for density e
stimates depends on the intended application of results. Managers and
researchers should consider whether their study area will accommodate
the necessary level of sampling. Density estimates from line-transect
sampling will be useful in investigations of the breeding biology of w
ood ducks, especially in regard to nest box management.