Background
This measure relates to indicator 1.5.1 - Species composition and diversity.
Birds are a key indicator of environmental health. Traditionally, monitoring has involved field observers, however this presents challenges for birds that call infrequently and/or during the night. As an alternative, DOC is using automatic recording devices (ARDs), which have the advantage of generating a reviewable permanent record and being able to record for extended periods of time. DOC has been using ARDs to record presence of nocturnal bird species (and other birds that frequently call at night) since 2011, to improve our knowledge of their distributions across public conservation land (PCL).
morepork / ruru is most frequently detected and widespread nocturnal bird species, found at 84.6% of sites.
What did we measure?
DOC uses a National Biodiversity Monitoring Programme (NBMP) to assess the status and trends of biodiversity at nearly 1,400 sites that are spaced evenly across PCL. Approximately 280 randomly-selected sites are measured each field season (September–May) so that every site is measured on a 5-year rotation.
There are five bird count stations, spaced approximately 200 m apart, at each site. Observers conduct bird counts during the day, while ARDs are set to record day and night, and recordings are processed manually to identify which species are present. This factsheet presents a selection of the results from nocturnal ARD recordings for the first 10 years of DOC’s monitoring programme (2011/12 to 2020/21), focussing on species that are nocturnal (e.g. kiwi, morepork / ruru) or that can be active day or night but are frequently heard calling at night (e.g. kākā, kea, long-tailed cuckoo / koekoeā).
The data are used to update our knowledge of species distributions and improve our understanding of species behaviour/ecology via various research projects (e.g. Pryde et al. 2020), and can feed into other monitoring programmes, such as the Save Our Iconic Kiwi (SOIK) programme.
What did we find?
The most frequently detected and widespread species was morepork / ruru (84.6% of sites), followed by long-tailed cuckoo / koekoeā (54.3%), kākā (41.8%) and kiwi spp. (28.7%; Table 1).
Several rarely encountered species were also detected, including Australasian bittern / matuku hūrepo, marsh crake / kotoreke, New Zealand falcon / kārearea, blue duck / whio and takahē (Figure 1).
Naive (i.e. unmodelled) occupancy results indicate that kākā, kiwi spp., morepork / ruru and weka are much more commonly encountered in woody habitats than non-woody, whereas kea is more widespread in non-woody habitats (Figure 2).
Occupancy estimates (modelled) suggest that morepork / ruru is present at almost all the woody sites surveyed, whilst kiwi spp. are present at around one quarter of all the woody sites surveyed. The models indicate that there have been only minor changes in occupancy since the national monitoring programme began, ie no major increases or decreases in population distributions (Figure 3).
Figures
Figure 1: Distributions of nocturnal species based on automatic recording device (ARD) recordings, 2011/12 to 2020/21. Black dots = species detected, open circles = site surveyed but species not detected. All sites are on public conservation land (PCL).
Figure 2: Naive occupancy estimates for nocturnal species based on automatic recording device (ARD) recordings, 2011-12 to 2020-21.
Figure 3: Modelled occupancy estimates for nocturnal species based on automatic recording device (ARD) recordings, 2011-12 to 2020-21, with 95% confidence intervals.
Tables
| Species | Number of sites | Percentage of sites (%) |
|---|---|---|
| Morepork / ruru | 942 | 84.6% |
| Long-tailed cuckoo / koekoeā | 604 | 54.3% |
| Kākā | 465 | 41.8% |
| Kiwi spp. | 319 | 28.7% |
| Kea | 312 | 28.0% |
| Weka | 293 | 26.3% |
| Pūkeko | 104 | 9.3% |
| New Zealand falcon / kārearea | 42 | 3.8% |
| Marsh crake / koitareke | 17 | 1.5% |
| Australasian bittern / matuku hūrepo | 14 | 1.3% |
| Blue duck / whio | 6 | 0.5% |
| Little penguin / kororā | 6 | 0.5% |
| Cook’s petrel / tītī | 6 | 0.5% |
| Mottled petrel / kōrure | 5 | 0.4% |
| Spotless crake / pūweto | 4 | 0.4% |
| Little owl | 3 | 0.3% |
| Takahē | 3 | 0.3% |
| Banded rail / mioweka | 2 | 0.2% |
| Common diving petrel / kuaka | 1 | 0.1% |
| Grey-faced petrel / ōi | 1 | 0.1% |
Data quality
This measure is classified as a partial measure of high accuracy and complies with the data quality guidelines used in in New Zealand’s Environmental Reporting Series.
The data were first reduced to detected/non-detected for each species at each station in a season. The data were modelled separately for each species, with a separate zero-inflated binomial model was fitted for each of the two habitats. The probability of detection (given occupancy) was assumed to be constant across the stations within a site, but could vary between sites. Probability of occupancy was modelled as an inverse logit function of time, also with a site-level random effect. In addition to the woody and non-woody habitats, estimates for an “all sites” habitat were reported, but to avoid confounding between habitat and time, the two first seasons were omitted for that purpose.
Although DOC processes thousands of nocturnal ARD recordings each year, this does not represent all of the recordings collected. Resource constraints mean that only one 15-minute recording from every hour between sunset and sunrise is processed, leaving the other 45 minutes per hour unprocessed. Therefore, the presence of species may be missed at some sites, particularly those species that call infrequently and/or are rare. Call frequency will vary with species and may also be influenced by the time of year, weather and other factors. Automated species identification software (currently under development) may reduce costs and enable a much larger sample of recordings to be processed in the future.
Figures 2 and 3 show trends in occupancy (unmodelled and modelled respectively) for a selection of the most commonly-recorded species. Conclusions from these should be treated with caution, however, as the annual occupancy estimates are based upon a different random selection of sites each year, rather than a repeated measure of the same sites (it takes five years to survey a complete set of sites).
Glossary of terms
Automatic recording device (ARD) is any device that records audio/visual information and does not require the presence of an operator.
Processing is the conversion of recorded information (audio/visual) into data that can be analysed (a spreadsheet of identified bird species). Processing can be automated (performed by a computer algorithm) or manual (performed by a person).
Naive occupancy is the proportion of sites where a species has been recorded. For example, a naive occupancy of 0.8 means that the species was recorded at 80% of sites. This does not account for sites at which a species was present but remained undetected.
Modelled occupancy is the estimated proportion of sites where a species is present, using repeated surveys to calculate a probability of presence at sites where the species was not detected.
Additional resources
Lee, W.; McGlone, M.; Wright, E. 2005: Biodiversity inventory and monitoring: a review of national and international systems and a proposed framework for future biodiversity monitoring by the Department of Conservation. Landcare Research Contract Report LC0405/122 for the Department of Conservation, Wellington (unpublished). 213 p.
McGlone, M.; Dalley, J. 2015: A framework for Department of Conservation inventory and monitoring: intermediate outcomes 1–5. Landcare Research Contract Report LC2427 for the Department of Conservation, Wellington (unpublished). 75 p.
Pryde, M.A.; Mortimer, J.A.J.; Greene, T.C.; Thygesen, H.H. 2020: Optimising monitoring times for surveys of ruru (Ninox novaeseelandiae novaeseelandiae). New Zealand Journal of Ecology 44(1): 3401.