{"id":{"repo_id":"exeter","oai_identifier":"oai:figshare.com:article/32055675"},"canonical_url":"https://search.dev.ndltd.org/etd/exeter/oai:figshare.com:article/32055675","repository":{"repo_id":"exeter","name":"University of Exeter","base_url":"https://api.figshare.com/v2/oai"},"display":{"title":"Towards multimodal monitoring of large African carnivores","abstract":"Technology has helped revolutionise the collection of ecological data to monitor African lions (Panthera leo) and leopards (Panthera pardus). The most used technologies are camera traps and GPS or VHF collars. Passive acoustic monitoring could act as a complementary monitoring tool because both species produce loud, recognisable vocalisations that propagate long distances. Yet, no such survey has occurred for leopards and passive acoustic monitoring for lions has only previously been conducted on a limited scale. In this thesis, from fieldwork I conducted in Nyerere National Park, Tanzania, I present the first large-scale passive acoustic monitoring survey for lions and leopards. In addition, this thesis describes one of the first large-scale, paired passive acoustic monitoring and camera trap surveys globally. I begin by demonstrating how using multimodal data facilitates the identification of individual leopards via their roars. In a similar sense, individual lions can be identified by their full-throated roars; however, this has previously relied upon expert judgement and was susceptible to human bias. I develop a machine learning method that automated the identification of lions’ full-throated roars, led to the classification of new type of lion roar – intermediary roar - and subsequently improved the ability to discriminate between individuals. Next, to process the vast passive acoustic monitoring dataset that I collected, I develop an automated acoustic species classifier for lions. However, lions inhabit vast home ranges and persist in low densities, so obtaining the necessary volume of wild training data to build an acoustic classifier is difficult. Therefore, I tested whether captive populations could provide the necessary vocalisation training data to detect lions in wild recordings. I showed that captive individuals could provide additional training data for acoustic species detection classifiers, but only when combined with wild data. Overall, I lay important foundations for the future investigation of lion and leopard vocalisations and outline how bioacoustics combined with camera trapping could result in improved monitoring for lions and leopards.<p></p>","abstract_html":"Technology has helped revolutionise the collection of ecological data to monitor African lions (Panthera leo) and leopards (Panthera pardus). The most used technologies are camera traps and GPS or VHF collars. Passive acoustic monitoring could act as a complementary monitoring tool because both species produce loud, recognisable vocalisations that propagate long distances. Yet, no such survey has occurred for leopards and passive acoustic monitoring for lions has only previously been conducted on a limited scale. In this thesis, from fieldwork I conducted in Nyerere National Park, Tanzania, I present the first large-scale passive acoustic monitoring survey for lions and leopards. In addition, this thesis describes one of the first large-scale, paired passive acoustic monitoring and camera trap surveys globally. I begin by demonstrating how using multimodal data facilitates the identification of individual leopards via their roars. In a similar sense, individual lions can be identified by their full-throated roars; however, this has previously relied upon expert judgement and was susceptible to human bias. I develop a machine learning method that automated the identification of lions’ full-throated roars, led to the classification of new type of lion roar – intermediary roar - and subsequently improved the ability to discriminate between individuals. Next, to process the vast passive acoustic monitoring dataset that I collected, I develop an automated acoustic species classifier for lions. However, lions inhabit vast home ranges and persist in low densities, so obtaining the necessary volume of wild training data to build an acoustic classifier is difficult. Therefore, I tested whether captive populations could provide the necessary vocalisation training data to detect lions in wild recordings. I showed that captive individuals could provide additional training data for acoustic species detection classifiers, but only when combined with wild data. Overall, I lay important foundations for the future investigation of lion and leopard vocalisations and outline how bioacoustics combined with camera trapping could result in improved monitoring for lions and leopards.&lt;p&gt;&lt;/p&gt;","abstract_has_math":false,"creators":["Jonathan Growcott (21049556)"],"institution":null,"degree_name":null,"degree_level":null,"degree_discipline":null,"degree_department":null,"school":null,"contributors":[],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2026,"date_issued":"2026-04-20T00:00:00Z","date_published":"2026-04-20T00:00:00Z","updated_at":"2026-07-27T19:33:24Z","subjects":["camera traps","passive acoustic monitoring","lions","leopards"],"languages":[],"rights":["CC BY","Open Access after 2027-04-20"],"rights_urls":[],"identifier_entries":[{"key":"dc:identifier","label":"Identifier","values":["10779/exe.32055675.v1"],"render_values":[{"text":"10779/exe.32055675.v1","href":null,"code":true}]}]},"links":{"outbound_url":null,"outbound_label":null,"outbound_source":null},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:creator","label":"Author","values":["Jonathan Growcott (21049556)"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date","label":"Dc Date","values":["2026-04-20T00:00:00Z"]},{"key":"dc:relation","label":"Dc Relation","values":["https://figshare.com/articles/thesis/Towards_multimodal_monitoring_of_large_African_carnivores/32055675"]},{"key":"dc:type","label":"Dc Type","values":["Text","Thesis"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["camera traps","passive acoustic monitoring","lions","leopards"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:rights","label":"Dc Rights","values":["CC BY","Open Access after 2027-04-20"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["10779/exe.32055675.v1"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description","label":"Description","values":["Technology has helped revolutionise the collection of ecological data to monitor African lions (Panthera leo) and leopards (Panthera pardus). The most used technologies are camera traps and GPS or VHF collars. Passive acoustic monitoring could act as a complementary monitoring tool because both species produce loud, recognisable vocalisations that propagate long distances. Yet, no such survey has occurred for leopards and passive acoustic monitoring for lions has only previously been conducted on a limited scale. In this thesis, from fieldwork I conducted in Nyerere National Park, Tanzania, I present the first large-scale passive acoustic monitoring survey for lions and leopards. In addition, this thesis describes one of the first large-scale, paired passive acoustic monitoring and camera trap surveys globally. I begin by demonstrating how using multimodal data facilitates the identification of individual leopards via their roars. In a similar sense, individual lions can be identified by their full-throated roars; however, this has previously relied upon expert judgement and was susceptible to human bias. I develop a machine learning method that automated the identification of lions’ full-throated roars, led to the classification of new type of lion roar – intermediary roar - and subsequently improved the ability to discriminate between individuals. Next, to process the vast passive acoustic monitoring dataset that I collected, I develop an automated acoustic species classifier for lions. However, lions inhabit vast home ranges and persist in low densities, so obtaining the necessary volume of wild training data to build an acoustic classifier is difficult. Therefore, I tested whether captive populations could provide the necessary vocalisation training data to detect lions in wild recordings. I showed that captive individuals could provide additional training data for acoustic species detection classifiers, but only when combined with wild data. Overall, I lay important foundations for the future investigation of lion and leopard vocalisations and outline how bioacoustics combined with camera trapping could result in improved monitoring for lions and leopards.<p></p>"]},{"key":"dc:title","label":"Title","values":["Towards multimodal monitoring of large African carnivores"]}]}],"canonical_facts":{"dc:creator":["Jonathan Growcott (21049556)"],"dc:date":["2026-04-20T00:00:00Z"],"dc:description":["Technology has helped revolutionise the collection of ecological data to monitor African lions (Panthera leo) and leopards (Panthera pardus). The most used technologies are camera traps and GPS or VHF collars. Passive acoustic monitoring could act as a complementary monitoring tool because both species produce loud, recognisable vocalisations that propagate long distances. Yet, no such survey has occurred for leopards and passive acoustic monitoring for lions has only previously been conducted on a limited scale. In this thesis, from fieldwork I conducted in Nyerere National Park, Tanzania, I present the first large-scale passive acoustic monitoring survey for lions and leopards. In addition, this thesis describes one of the first large-scale, paired passive acoustic monitoring and camera trap surveys globally. I begin by demonstrating how using multimodal data facilitates the identification of individual leopards via their roars. In a similar sense, individual lions can be identified by their full-throated roars; however, this has previously relied upon expert judgement and was susceptible to human bias. I develop a machine learning method that automated the identification of lions’ full-throated roars, led to the classification of new type of lion roar – intermediary roar - and subsequently improved the ability to discriminate between individuals. Next, to process the vast passive acoustic monitoring dataset that I collected, I develop an automated acoustic species classifier for lions. However, lions inhabit vast home ranges and persist in low densities, so obtaining the necessary volume of wild training data to build an acoustic classifier is difficult. Therefore, I tested whether captive populations could provide the necessary vocalisation training data to detect lions in wild recordings. I showed that captive individuals could provide additional training data for acoustic species detection classifiers, but only when combined with wild data. Overall, I lay important foundations for the future investigation of lion and leopard vocalisations and outline how bioacoustics combined with camera trapping could result in improved monitoring for lions and leopards.<p></p>"],"dc:identifier":["10779/exe.32055675.v1"],"dc:relation":["https://figshare.com/articles/thesis/Towards_multimodal_monitoring_of_large_African_carnivores/32055675"],"dc:rights":["CC BY","Open Access after 2027-04-20"],"dc:subject":["camera traps","passive acoustic monitoring","lions","leopards"],"dc:title":["Towards multimodal monitoring of large African carnivores"],"dc:type":["Text","Thesis"]},"updated_at":"2026-07-27T19:33:24Z"}