|dc.description.abstract||Kōkako (Callaeas wilsoni) are endangered, duetting songbirds endemic to New Zealand and are confined to a small number of managed mainland reserves and offshore islands. Each fragmented population exhibits distinctive vocal traditions. Conservation of the species is centred on intensive site management of introduced mammalian predators – the current leading cause of kōkako population decline – followed by re-establishment of populations through translocation. Translocated populations are often sourced from multiple areas, leading to an artificially created scenario of secondary contact between behaviourally diverged populations.
I studied the consequences of kōkako song traditions, and the effect of population-specific behaviours on conservation of the species. During a transfer of 20 birds from two distinct "song neighbourhoods" (c. 25% between-neighbourhood phrase sharing), I explored the utility of neighbourhood specific acoustic playback as a conservation tool in preventing excessive post-release dispersal. I found that birds dispersed less far than predicted by a random walk model, yet were no more attracted to same- than different-neighbourhood song playback. These results suggested that while playback appeared to reduce dispersal, this effect was not driven by neighbourhood-specific song. Following release, kōkako also used the available habitat disproportionately, preferring to remain in the short term, and to establish breeding territories within a forest type similar to that where they were caught. The vocal differences I detected between the neighbourhoods at the source sites were also insufficient to promote assortative mating following release. The findings of this experiment confirmed that neighbourhood-scale song variation does not act as a barrier to gene flow, or inhibit post-translocation population establishment.
To assess whether more distinct vocal differences between populations (c. 5% between-population phrase sharing) represent more salient signals to kōkako, I performed reciprocal stereo playback experiments in two populations frequently used as translocation source sites. In one experiment I tested for discrimination between local and foreign dialect duets and found that pairs responded vocally to local duets with less delay, and produced more song phrases, with a lower diversity, compared to playback of foreign duets. This suggested that local song represented more of a threat to pairs than unfamiliar song. In a second experiment I presented pairs with two types of synthesised local-foreign mixed-dialect duets; each stimulus differed depending on the sex of the local duet stimulus component. I found that pairs responded with equal overall strength to both stimuli, and did so in a qualitatively similar way to pure local duets. As pairs did not discriminate between these stimuli, this indicates that if mixed-dialect pairs form, they should be able to effectively communicate with, and defend against territorial intruders. From an evolutionary perspective this further suggests that dialectal differences might not prevent gene flow in kōkako if pair formation between dialects can occur. Importantly, these findings also suggest that animals translocated for conservation purposes need not necessarily exhibit homogeneous cultures to acquire and defend resources.
Sexual selection leading to positive assortative mating based on song dialect may act as an isolating barrier to gene flow, thus promoting speciation. From a conservation perspective this process could be problematic by hindering population establishment following translocation of a small number of individuals from multiple sources. To assess this tendency in kōkako, I harnessed an existing 18-year data set from reports prepared by the New Zealand Department of Conservation documenting multi-source translocations to six sites from 11 source populations. Based on these reports, in each breeding season at each site I compiled a list of the possible same- and mixed-dialect pairs that could have formed, which I then compared to the actual pairs in each season. In doing so, I statistically confirmed that, following release, kōkako pairs form assortatively based on dialect more often than expected by chance. However, mixed-dialect pairs did form in low numbers at two sites, either when an existing "core" population was present at the time of translocation, or following the release of a captive-reared sub-adult bird with potentially plastic song. These results suggest that sufficiently large differences in a sexually selected behavioural trait, such as song, can inhibit gene flow between populations.
Taken together, my findings suggest that vocal variation among fragmented, allopatric populations may influence a species' cultural and genetic history, and could affect the success of conservation strategies. However, as kōkako pairs can form across dialects, and postnatal dispersal sometimes occurs between areas with different song traditions, complete genetic isolation of populations is unlikely to occur. Further research on the timing of song-learning in kōkako, dispersal patterns, and vocal plasticity in territorial adults will greatly advance our knowledge base and shed light on the function and conservation implications of song dialects in kōkako.||