TY - JOUR
T1 - The potential for acoustic individual identification in mammals
AU - Linhart, Pavel
AU - Mahamoud-Issa, Mathieu
AU - Stowell, Dan
AU - Blumstein, Daniel T.
N1 - Funding Information:
Laela Sayigh, Frants Jensen, and Vincent Janik kindly discussed and shared yet unpublished individual identity content in bottlenose dolphin signature whistles. Justin Salamon, Tim Sainburg advised on collections of bioacoustic datasets and repositories. Alex Průchová, Ema Hrouzková, Siddharth Khopkar, Lucie Hornátová, Eliška Kovářová, and Jan Riegert commented on the manuscript within the Bioacoustic Journal Club at the University of South Bohemia. Work of PL and MMI was partially funded from the European Union's Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No. 665778 administered by the National Science Centre, Poland (UMO-2015/19/P/NZ8/02507). PL was also funded by the Czech Science Foundation (21-04023K). DTB is funded by the US National Science Foundation.
Funding Information:
The work of PL was partially funded by the Czech Science Foundation (GACR 21-04023 K). Furthermore, the work of PL and also MMI was partially funded from the European Union's Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No. 665778 administered by the National Science Centre, Poland (UMO-2015/19/P/NZ8/02507). DTB is funded by the US National Science Foundation.
Publisher Copyright:
© 2022, The Author(s) under exclusive licence to Deutsche Gesellschaft für Säugetierkunde.
PY - 2022/3/31
Y1 - 2022/3/31
N2 - Many studies have revealed that animal vocalizations, including those from mammals, are individually distinctive. Therefore, acoustic identification of individuals (AIID) has been repeatedly suggested as a non-invasive and labor efficient alternative to mark-recapture identification methods. We present a pipeline of steps for successful AIID in a given species. By conducting such work, we will also improve our understanding of identity signals in general. Strong and stable acoustic signatures are necessary for successful AIID. We reviewed studies of individual variation in mammalian vocalizations as well as pilot studies using acoustic identification to census mammals and birds. We found the greatest potential for AIID (characterized by strong and stable acoustic signatures) was in Cetacea and Primates (including humans). In species with weaker acoustic signatures, AIID could still be a valuable tool once its limitations are fully acknowledged. A major obstacle for widespread utilization of AIID is the absence of tools integrating all AIID subtasks within a single package. Automation of AIID could be achieved with the use of advanced machine learning techniques inspired by those used in human speaker recognition or tailored to specific challenges of animal AIID. Unfortunately, further progress in this area is currently hindered by the lack of appropriate publicly available datasets. However, we believe that after overcoming the issues outlined above, AIID can quickly become a widespread and valuable tool in field research and conservation of mammals and other animals.
AB - Many studies have revealed that animal vocalizations, including those from mammals, are individually distinctive. Therefore, acoustic identification of individuals (AIID) has been repeatedly suggested as a non-invasive and labor efficient alternative to mark-recapture identification methods. We present a pipeline of steps for successful AIID in a given species. By conducting such work, we will also improve our understanding of identity signals in general. Strong and stable acoustic signatures are necessary for successful AIID. We reviewed studies of individual variation in mammalian vocalizations as well as pilot studies using acoustic identification to census mammals and birds. We found the greatest potential for AIID (characterized by strong and stable acoustic signatures) was in Cetacea and Primates (including humans). In species with weaker acoustic signatures, AIID could still be a valuable tool once its limitations are fully acknowledged. A major obstacle for widespread utilization of AIID is the absence of tools integrating all AIID subtasks within a single package. Automation of AIID could be achieved with the use of advanced machine learning techniques inspired by those used in human speaker recognition or tailored to specific challenges of animal AIID. Unfortunately, further progress in this area is currently hindered by the lack of appropriate publicly available datasets. However, we believe that after overcoming the issues outlined above, AIID can quickly become a widespread and valuable tool in field research and conservation of mammals and other animals.
KW - Acoustic communication
KW - Acoustic individual identification (AIID)
KW - Acoustic signature
KW - Individual identity
KW - Individual variation
KW - Machine learning
KW - Mammal
KW - Vocalizations
UR - http://www.scopus.com/inward/record.url?scp=85127606911&partnerID=8YFLogxK
U2 - 10.1007/s42991-021-00222-2
DO - 10.1007/s42991-021-00222-2
M3 - Article
VL - 102
SP - 667
EP - 683
JO - Mammalian Biology
JF - Mammalian Biology
ER -