Individual identification patterns in the Visayan spotted deer and implications on the study of their behavioural ecology

Natálie Šimáčková

Master's thesis

Abstract:

The Visayan spotted deer (Rusa alfredi) is considered one of the rarest deer in the world. The extensive hunting and deforestation of its primary habitat have brought the species to the brink of extinction. The remaining populations on the islands of Negros and Panay in the Philippines are estimated to be less than 1000 mature individuals. The threats and lack of current species knowledge are concerning; thus, conservation action is urgently needed to save this species. Within this context, camera traps offer an effective and non-invasive method for monitoring this elusive species. In our thesis, we focused on studying the individual identification patterns of Rusa alfredi in the Bayawan Nature Reserve, Negros, Philippines. Over a period of 2 years, we monitored a population of 33 individuals in semi-wild conditions using camera traps, observing four artificial feeding stations. We collected 641 images out of 4,510 videos featuring deer. In the initial analysis, the individual identification software IBEIS identified 45 of 101 images, with a matching score of 44.55 %. By the end of the study, IBEIS achieved an impressive 95.50 % matching success rate in identification. It should be emphasised, however, that this accomplishment was the result of extensive training of the software. When the same analysis was conducted without the training and extensive ID database, IBEIS only identified 50 out of 89 images, scoring 56.18 %. We highlighted the advantages of utilising camera trap videos as opposed to images for individual recognition studies and stressed the significance of a comprehensive identification database with real-life images for the software to operate effectively. The findings of this study will be used for behavioural studies at the individual level. Furthermore, this study serves as an important foundation for understanding the ecology of the species and guiding conservation efforts. Our recommendations can serve as a valuable reference point for others working with spotted cervid species, offering key recommendations for leveraging camera traps and identification software in similar conservation studies.

Keywords

artificial intelligence, camera-trapping, IBEIS, Philippines

Language used: English

Date on which the thesis was submitted / produced: 16. 8. 2024

Thesis defence

• Date of defence: 12. 9. 2024
• Supervisor: doc. Francisco Ceacero Herrador, Ph.D.
• Reader: Meyer De Kock, externi

Czech University of Life Sciences Prague

Faculty of Tropical AgriSciences

Master programme / specializace:
Wildlife and Livestock Production, Management and Conservation / Wildlife Management and Conservation