Bellbird song recording as a field technique: a feasibility study
Bellbird songs are instantly recognisable to most New Zealanders, although many people have never seen the bird in the wild. Because the song is so distinctive, researchers carrying out field work often locate the bellbird first by sound, and only then by sight.
This is why we felt that recording of sound in the field – along with semi-automated sound analysis back in the lab – might have potential as a useful field method. With such a scheme, recording equipment could be placed in an area (or a number of areas, simultaneously) and left there over a period of days to capture information about bird presence.
BIRD SONG AS A FIELDWORK METHOD
As far as bellbird fieldwork is concerned, we would ideally like to identify the sex and age class (adult/juvenile) of the bird, along with its “home” territory. We know this information for the birds we have colour-banded (colour-banding research), because the bird can be individually identified by the colour combination. However, if it were possible to identify one or more characteristics of the bellbird from its song, we might be able to reduce our reliance on colour-banding and associated labour-intensive resighting studies.
SONG VARIATIONS
But is it possible? There is anecdotal evidence that male and female bellbirds sing different songs, at least during some parts of the year. The song of juveniles is not fully developed straight away so an expert can distinguish their song from adult song. And we know that bellbirds in different parts of New Zealand sing with regional “dialects”.
In our study, we focused mainly on the geographical differences – we recorded bellbirds during the breeding season in three of the Port Hills bush areas. These were Victoria Park, Kennedy’s Bush and Omahu Bush. We were surprised how different they were! Click on the buttons below to hear the different calls.
Victoria Park bellbird song (109kb .wav file)
Kennedy’s Bush bellbird song (113kb .wav file)
Omahu Bush bellbird song (74kb .wav file)
In each bush area, all the birds we heard had the same song, but in separate
bush areas the songs were different. This probably supports the commonly-held
view that bellbird pairs go back to the same breeding territory year after
year, so that any group of birds tends to be quite localised. There is
also a tendency amongst many bird species (including tui) for young birds
to find a breeding territory near their parents. At present we don’t
know whether young birds in the Port Hills ever move to other bush patches
to find territories, and potentially learn the “local dialect”
when they do so.
If bellbirds retain that same local song when they move down out of the Port Hills during the winter to forage, maybe we could use it to identify where they came from! There is some evidence to suggest that they do, so a bellbird observer listening in their garden in Spreydon might be able to identify that the bellbird feeding in their kowhai tree originates from Omahu Bush. However, to test this, we would need to record the songs of colour-banded bellbirds while they are foraging in the city and vicinity, and match up origin (known from band colour combination) with song. We have not yet done this.
RECORDING AND ANALYSIS
There were two other very important parts to our project:
- Assembling suitable equipment that was both light and robust for use in the field, and of sufficient quality to faithfully record bellbird song, and
- Writing software for partially automated analysis of the songs.
If sound recording is ever to be used as a practical field method, then efficient methods for analysing large quantities of data must be available. We began with two aspects of this. First, we wrote software to convert the call into a spectrogram.
Next, we wrote software to detect each of the notes represented in the spectrogram. Each note was analysed to determine its frequency (i.e., how low or high pitched it is), start time and duration. The “sweeps” are analysed to find their starting frequency and rate of frequency change.
These few numbers define each note and sweep in a simple way. They are easy to store, and offer a formal way of comparing two songs for similarity.
CONCLUSIONS
We have assembled recording equipment and written computer software to analyse bellbird song. This is a good foundation for any further work on bird song analysis. We have shown that bellbird songs are very different from one Port Hills valley to the next – maybe it will offer us a way of identifying the origin of a bellbird while it’s foraging in Christchurch during the winter!