Detectability of Western Derby elands in West-African savanna by aerial imagery

B.Sc. Carlos Castillo Gómez, BSc

Master's thesis

Abstract:

Western Derby (or Giant) eland (Taurotragus derbianus ssp. derbianus) is one of the two subspecies of Derby eland that exist on the planet and the most threatened of them. According to IUCN its status is Critically Endangered (CR) and has a single population of about 120-150 mature individuals in Niokolo Koba National Park, Senegal. Due to its critical conservation status, in 2000, one male and eight females were captured (of which only five females survived) and transferred to Bandia Reserve (to create a semi-captive population with conservation purposes), where this research has been conducted. Due to its small population and the vast extension of the only place where they live in the wild, the use of Unmmaned Aerial Systems (UASs) for census and monitoring has been valued, as it has been successfully developed with other ungulates. The objective of this pioneering study with this antelope is, first, to determine the parameters, such as flight height, sensor angle, pros and cons of photos and videos, flight speed and ethical considerations, that affect the detectability of individuals in aerial images (thanks to flight tests under different conditions to determine their impact on probabilities of detection) and, second, to evaluate two systems of detection of specimens, Manual (using untrained volunteer observers) and Automated Supervised (thanks to Ecognition machine learning software), to know the scope of each one. After this research it was determined that the most efficient flight height for the detection of specimens is 182.9 meters, the most appropriate angle of the camera: 34-35º, the use of photographs or videos will depend on the final use of them and our objectives, the most appropriate speed is 3-5 m/s and the height at which the UAS begins to cause discomfort, is 25 meters. On the other hand, the detection of specimens by untrained observers was affected by the speed of the video (the higher the speed, the greater the difficulty in detecting the individuals), by the possibility, or not, of stopping it (greater possibility of detection when it is possible to stop the video), by its level of studies and age (individuals 18-38 years old and with university studies were more skilled in detection) but by their sex. On the other hand, the Automated Supervised Classification showed partial results that should continue to be investigated due to its great potential, which has been demonstrated in numerous studies with other species. This study is a first step in the methodology of work with Western Derby eland and its detection through aerial images, opening a door to future studies that may shed light on the situation of this threatened and magnificent antelope.

Keywords

Western Derby eland, WDE, UAV, Drone, Senegal, Detection, Woody savannah, Manual classification, Automated classification

Language used: English

Date on which the thesis was submitted / produced: 26. 4. 2019

Thesis defence

• Supervisor: prof. RNDr. Pavla Hejcmanová, Ph.D.

Czech University of Life Sciences Prague

Faculty of Tropical AgriSciences

Master programme / field:
Tropical Agriculture / Wildlife Management in the Tropics and Subtropics