Abstract
Growing concerns about significant biodiversity decline due to agricultural intensification are increasingly leading consumers to seek agricultural products that are produced sustainably. To raise awareness of sustainable land management and direct policy and research to mitigate adverse impacts, large-scale bird monitoring programmes are being used in Europe. New Zealand's first farmland bird monitoring scheme was established in 2004 to quantify bird abundance on 98 farms across three sectors (sheep & beef, dairy and kiwifruit). Distance methods were considered ideal because they minimised disruption by nuisance variables that affected detectability (most often observer and whether birds were seen or heard; less frequently, effects of wind, habitat and farming systems). However, distance detection functions could only be measured for half the species present on the study farms, and sampling uncertainty remained high for several of those species. Gradually more species with reduced sampling uncertainty can be added as sufficient detections are gathered to generate their global detection functions. This will likely increase the scheme's power to detect any ongoing decline, but simulations that combine sampling uncertainty with observed inter-annual variation in abundance are now needed to test whether population-decline thresholds can be reliably detected using the current and alternative survey designs.