On kiwi (Apteryx mantelli) vocal behaviour and activity : relations to population densities and applications to conservation : a dissertation presented in partial fulfilment of the requirements for the degree of Doctor of Philosophy in Zoology at Massey University, Manawatū, Aotearoa New Zealand

According to the International Union for the Conservation of Nature (IUCN), over 38,500 species of living organisms assessed (27.8%) are currently threatened with extinction. Reducing this startling percentage requires cost–effective monitoring of populations of many and varied species. Information regarding population trends is crucial to allow decision makers to judiciously allocate unavoidably limited resources. Acoustic monitoring has long been employed to document the presence and estimate populations of vocal species for conservation purposes. Determining populations trends without the need of sighting or capturing animals can drastically reduce costs and improve welfare. However, as with many other indirect monitoring practices, acoustic surveys impose a series of assumptions about the detectability of the observed animals and their vocal behaviour. Whereas the variability in detection distances and other observer–induced effects can be minimised using acoustic recorders, enabling the delivery of animal abundances using acoustic monitoring requires detailed knowledge of the target species’ behaviours to relate numbers of detected acoustic cues to those of animals in an area. The iconic North Island Brown Kiwi (Apteryx mantelli, Bartlett 1851) is a flightless nocturnal bird species endemic to Aotearoa New Zealand, fragmentedly distributed across its mainland range and some of its offshore islands. North Island Brown Kiwi are known for their characteristic vocalisations which differ between sexes, with males emitting series of whistle-like syllables, and females producing series of hoarser and lower frequency syllables. Indeed, acoustic surveys are routinely employed by conservation groups and the Department of Conservation Te Papa Atawhai to monitor North Island Brown Kiwi. These surveys, known as Kiwi Call Counts, require observers to annotate sex, direction of arrival, and distance of the detected Kiwi vocalisations over a set period of time. However, little is known of North Island Brown Kiwi vocal behaviour and how this may relate to animal abundance and the development of more accurate and objective monitoring practices is included among the objectives of the Kiwi Recovery Plan (Germano et al., 2018). This thesis aimed to investigate North Island Brown Kiwi vocal behaviour and activity to build more objective and accurate acoustic monitoring protocols. Firstly, results from an extensive literature review on the acoustic playback technique — which has been shown to have the potential to enhance acoustic surveys in other species — led to the development of a set of recommendations to enable reproducibility when using playback. Secondly, results from playback experiments showed how single microphone acoustic recording units (ARUs) can be used to localise sound sources with reasonable degrees of uncertainty. This enables the potential transition of Kiwi Call Counts from relying on human observers to ARUs, which would allow for objective interpretation of the data while creating a potentially perpetual record. One of the thesis aims was to ascertain the potential of using playback to standardise the response of Kiwi populations. The results of experiments testing the effect of playback and environmental factors on kiwi vocal response show that there is no real relationship between the vocal activity of the target Kiwi community and playback. However, they corroborate and add to existing knowledge of Kiwi vocal behaviour by identifying relationships between the latter and external factors, such as lunar illumination and weather conditions. This thesis finally concentrated on the issue of relating vocal activity to animal abundance by developing and trialling the use of animal-borne acoustic recorders in conjunction to fixed ARUs. Since using animal-borne acoustic recorders entails handling target animals, we first performed an experiment on post–handling vocal behaviour to ascertain whether the vocal activity of handled birds of our target community differed from that of birds that had never been handled. The results from this experiment showed that the vocal activity recorded from a gully inhabited by never handled Kiwi did not differ from that of a gully inhabited by birds that were handled during the survey — and have been regularly handled over the last 17 years — in any detectable way. This is encouraging both for animal welfare purposes, and for comparing acoustic surveys from both managed and more wild Kiwi populations. Finally, the results from employing the animal-borne acoustic recorders to inform density estimates showed how information about individual vocal activity informs more realistic and consistent population estimates than methods based only on community–level vocalisations. On all the occasions sampled, results of population estimates only accounting for environmentally recorded vocalisations delivered lower abundance expectations for both males and females. Repeated sampling results show how estimates that account for individual vocal activity are both more consistent and closer to real densities than traditional methods, as estimated by paired sampling with a specialised dog survey. Lastly, information from individual vocal activity in some populations informed more accurate estimates for other populations without individually tagged animals. Taking advantage of having multiple populations with tagged individuals, we estimated abundances of a target population with three different models: unmarked, tagged with animal-borne acoustic recorders, and with information from other populations’ tagged individuals. This last estimate was in between the unmarked and with animal-borne acoustic recorders and apparently more accurate than the unmarked model. This thesis provides methods and shows encouraging results to eventually employ passive acoustic monitoring to infer Kiwi abundance in a cost-effective and non–invasive fashion at large scale, and invites further employment of animal-borne acoustic recorders to confidently deliver abundance estimates, crucial information for conservation decision makers. Using animal-borne acoustic recorders and ARUs together as a way to estimate populations does involve some invasive trials, but has the potential to lead to fully non–invasive robust abundance estimates though passive acoustic monitoring