Use of distance sampling to measure long-term changes in bird densities in a fenced wildlife sanctuary

Long-term data are needed to assess the impact of management initiatives such as mammalian predator-exclusion fences, but long-term monitoring programmes can be difficult to maintain. We used annual line transect distance sampling data collected by undergraduate students to model trends in native bird densities at Bushy Park, New Zealand, from 2002 to 2018, including 14 years of data collection following the installation of a predator-exclusion fence in 2005. We corrected for known breaches to the distance sampling assumptions for North Island robins/toutouwai (Petroica longipes) by calibrating raw transect counts with mark-recapture data. Two of the three reintroduced species, North Island robins and North Island saddlebacks/ tieke (Philesturnus rufusater), showed marked increases in density, and were the numerically dominant species in Bushy Park by the end of the study. The distance data for hihi (Notiomystis cincta), which were reintroduced in 2013, were too sparse to show a trend. Comparison with independent data for these three species showed that uncorrected distance data greatly over-estimated densities of robins (6-fold) and hihi (9-fold) but were accurate for saddlebacks. The methodology used to calibrate North Island robin estimates could be applied to hihi if the current intensive monitoring for that species is discontinued. In contrast to the reintroduced species, densities of the original bird populations all remained relatively constant (kereru Hemiphaga novaeseelandiae, tomtit Petroica macrocephala) or declined (grey warbler/riroriro Gerygone igata, fantail/piwakawaka Rhipidura fuliginosa, silvereye/tauhou Zosterops lateralis) after the installation of the fence, or had too few observations to estimate densities (tui Prosthemadera novaeseelandiae, bellbird/korimako Anthornis melanura). This study demonstrates that simple low-intensity monitoring data collected by non-experts can provide useful information on long-term trends in bird densities. However, we stress the importance of including uncertainty in estimates when inferring population trends, and the potential need to calibrate distance data with independent density estimates.