MINIMIZING EFFORT IN LARGE-SCALE SURVEYS OF TERRESTRIAL BIRDS - AN EXAMPLE FROM THE ENGLISH UPLANDS

In 1990 a complete survey of the south Pennines in England (covering 7 25 km(2)) was used to estimate the population sizes of ten species of upland breeding birds. We investigate whether survey effort could have been reduced by using a sample based approach. Random sampling could have produced large reductions in survey effort for Curlew Numenius ar quata, Golden Plover Pluvialis apricaria and Red Grouse Lagopus lagopu s scoticus, which tended to be relatively evenly distributed across th e region. A random sample of less than 50% of the study area could est imate the populations of these species to within 10% of the value obta ined through the complete survey. Whinchat Saxicola rubetra and Twite Acanthis flavirostris is tended to be locally common and so over 80% o f the study area would need to be surveyed to provide equivalent popul ation estimates for these species. Regression sampling could not reduc e survey effort significantly in comparison with random sampling, beca use species-environment correlations tended to be relatively low. The largest reduction in survey effort occurred for Golden Plover in assoc iation with the distance to the moor edge in which a 41% regression sa mple had an equivalent accuracy to a 47% random sample. Stratified sam pling could produce the largest reductions in survey effort when the a llocation of survey effort to individual strata was related both to st rata size and to the variation of bird numbers within strata. Reductio ns in survey effort of over 25% in comparison with random sampling wer e obtained for Dunlin Calidris alpina stratified by percent moorland, and Twite stratified by northing. No single environmental variable cou ld provide the most efficient means of stratification for all species. We discuss the choice of sampling strategy for different bird species and given different survey objectives, and consider how the results o btained in the south Pennines may be used to determine appropriate sam pling strategies in other regions.