Estimating Relative Densities of Breeding Birds by the Line Transect Method. II. Comparison between Two Methods

• Published in: Ornis Scandinavica 1976-05, Vol.7 (1), p.43-48
• Main Author: Jarvinen, Olli
• Format: Article
• Language: English
• Subjects: Approximation; Aviculture; Birds; Breeding; Density estimation; Estimate reliability; Estimation methods; Linear models; Standard deviation; Statistical discrepancies
• ISSN: 0030-5693
• DOI: 10.2307/3676173

In line transect censuses of breeding birds, each observation belongs either to the 'main belt' (extending to 25 m on both sides of the observer) or to the 'supplementary belt'. Estimates of density have earlier been derived from the main belt data (method A). A recent method (B) makes it possible to use all data accumulated during a census; thus the available body of data becomes roughly five-fold under Finnish conditions. Method B is shown to reduce the variances of density estimates considerably if data from other areas can be used in the calculation of the so-called main belt percentages. If not, the methods are essentially similar. Method B, however, diminishes interobserver variation in the estimation of the distance of birds. An important quantity required by method B, k, can be estimated from a simple, practically unbiased formula, unless the number of observations made on the species in all censuses is very small. A sample size of 20 is found to be sufficient to produce small variances of k; smaller samples yield quite heterogeneous estimates for similar species. Reliability of line transect censuses can be enhanced by method B if the main belt percentages of a given species do not vary considerably in larger regions, or if the possible variation can be corrected by general formulae.