An assessment of three survey methods on common bird species in low Arctic tundra.Comparing line transect distance sampling, repeated point counts and double observer point counts

Abstract

Long-term studies of bird communities in Arctic regions are rare in comparison to studies performed in temperate regions, particularly studies that takes detectability into consideration. Finding a suitable and robust method for long-term monitoring of bird populations in the low Arctic which includes a correction of imperfect detection, can contribute to increased ecological understanding and easier identification of changes over time. During two field seasons (June-July) in 2015 and 2016, the bird population on Erkuta tundra monitoring site, Yamal, Russia, was surveyed using the three different methods: distance sampling on line transects, repeated point counts and double observer point counts. The comparison between the methods was done by estimating abundances and densities of the three most common species in the area, i.e. Lapland bunting (Calcarius lapponicus), Red-throated pipit (Anthus cervinus) and Wood sandpiper (Tringa glareola). I found that the most plausible density estimates of the common target species were provided by the line transect distance sampling. Although all assumptions were difficult to completely fulfil in the field, the method could be improved by more exact distance measurements in the field, since the degree of overestimation is limited to the extent of which the assumptions are violated. In addition, the line transect distance sampling method have the least impact of “floaters” (non-breeders), which also makes this method more robust. Repeated point counts heavily overestimated the densities, and the models fitted poorly for Red-throated pipit. The overestimation was likely due to violation of the assumption of closed population, and could be reduced by using a model allowing for temporary emigration. Also, too short distance between points, and thus a risk of double counting, could have contributed to the overestimation. However, the double observer point counts offer an alternative approach for long-term monitoring, where the slight overestimation also can be decreased with increased distance between points. In addition, the double observer method requires less effort than the other two methods.