A field test of surveyors' influence on estimates in line transect sampling

Line transect sampling is a well-known method for estimating densities of w ildlife populations and can be a useful alternative to the traditional meth ods for timber cruising when assessing objects and species of interest in r elation to biodiversity, This paper presents results from a field test of l ine transect sampling for inanimate objects in boreal forest, focusing on t he surveyors' influence on estimates under different surveying conditions. The method was tested by II surveyors in northern Sweden in two forest type s and for two different object types. An underestimation of between -0.3 an d -22.2% was found, depending on the surveying conditions and the model use d for the detection function. The underestimation is partly due to violatio n of the assumption that all objects on or very close to the survey line ar e detected. It was found that systematic differences among surveyors were g enerally moderate. At most, the coefficient of variation for surveyor effec t was 6.8%. Still, this size of surveyor effect can be problematic for long -term monitoring, where systematic differences among surveyors' estimates m ay indicate illusory changes or disguise real trends. When comparing line t ransect sampling with other sampling methods, the increased variation in es timates caused by surveyors' systematic and random errors as well as the va riation caused by model selection must be considered. (C) 2000 Elsevier Sci ence B.V. All rights reserved.