Over 11,000 analyzed vocalizations from Weddell seals were recorded during the 1991 and 1992 breeding seasons near the Vestfold and Larsemann Hills in Antarctica. The dataset, managed by the Australian Antarctic Data Centre, classifies calls into 12 major types and contains 39 variables on acoustic properties and recording context. This collection supports comparative studies of seal vocal dialects.
Use Cases
- Classify CTYPE (call type) using acoustic features like SFREQ (start frequency), HFREQ (highest frequency), and CALLD (call duration).
- Analyze temporal patterns in vocalizations using DATE, INCD (interval between calls), and O_LAP (overlap indicator).
- Compare geographic variation in call repertoire by grouping data by LOCATION and YEAR.
- Model call structure complexity by relating NOELM (number of elements) to waveform (WVFRM) and shape (CLSHP) codes.
- Study individual or group calling behavior by sequencing calls using CALNO (call number) and CASSETTE identifiers.
Strengths
- Contains over 11,000 analyzed vocalization instances.
- Provides 39 detailed variables per call, including spectral, temporal, and contextual features.
- Covers 8 distinct recording locations across two Antarctic regions.
Limitations
- Sample size is limited to two breeding seasons (1991, 1992), offering no long-term temporal trend analysis.
- The DATE field has incomplete reference information, as noted for day 1.
- Some variables, like S2STM, are described as 'unknown measure', indicating potential data ambiguity.
Provenance
- Source
- Australian Antarctic Data Centre (AU_AADC), based on research published in the Australian Journal of Zoology.
- Collection Method
- Underwater recordings made using hydrophones and recorded to Digital Audio Tape and analogue cassette formats near seal groups on ice.
- Time Range
- 1991 to 1992 breeding seasons.
- Freshness
- null
- Geography
- Vestfold Hills (7 locations) and Larsemann Hills (1 location), Antarctica.