Measuring Orangutan nest structure using Unmanned Aerial Vehicle (UAV) and ImageJ

The nest is one of the crucial elements in orangutan daily activities. Previously, most of the nest structure studies were done manually by estimating measurement directly from visual observation. However, using the latest unmanned aerial vehicle (UAV) technology, we can reduce the workforce, time and energy while simultaneously ensuring the safety of the researcher conducting nest structure analysis. We recorded 49 pictures of orangutan nests at Sepilok Orangutan Rehabilitation Centre (SORC) using UAV (DJI Phantom 3 Quadcopter). The nest structure (length, depth, and width) was digitally measured by using ImageJ. Most of the nests were built at a strong, stable, and comfortable position at the top of the tree. Most orangutans chose Eusideroxylon zwageri to build nest compared to other tree species because of the strong and durable wood characteristic which would create a sturdy, strong and comfortable nest. We propose the use of drone with digital image analysis could provide a more accurate, less time consuming and safe method for studying orangutan nest structure.


42
Arboreal great apes especially orangutans need to master the nest building skill together with other 43 skills such as climbing, foraging and being able to identify their natural predators [1][2][3][4]. The nest 44 building is a skill inherited through observation on the mother's or other adults' nesting practices 45 [2,3,5,6]. Nests served as a "bed" for resting and sleeping, hiding from danger or predator, as 46 well as for a better thermoregulation [4,5,7]. 47 Previously, most of the orangutans' nest studies were based on ecological aspects such as density 48 estimation, distribution and population of orangutans in a habitat, nesting preferences and 49 mechanisms and material to build a nest and its decaying rates [7][8][9][10][11][12]. Nest measurement and 50 materials used are evaluating characteristics of the nest building skill of an individual and the 51 quality of the nest built [6,13]. However, researchers must conduct rigorous, time and energy 52 consuming techniques such as tree climbing to study the orangutan's nest structures. Even though 53 direct nests measurement by climbing techniques might provide a more accurate reading, a new 54 approach using Unmanned Aerial Vehicle (UAV) has been developed for nest survey and 55 observation during this decade [7,11,[13][14][15].

56
The application of UAV or drones technology was once limited to the military. However, 57 the usage of drones has spread widely and no longer exclusive to the military. Drones are used in 58 civilian use such as monitoring, transportation for goods delivery as well as site inspections [16, 59 17]. These drones have the capability to capture images through rapid data acquisition, which is 60 an advantage for scientific research especially for animals conservations [8,11,[16][17][18][19]. This study 61 aims to determine the variation of nest structure quantitatively via image analysis through 62 utilization of drones.  We tested a new experimental design by adapting to the nest description outlined by Samson and 76 Hunt (6). We utilized the use of Unmanned Aerial Vehicle (UAV) to capture the orangutan nests 77 images [11,20]. We used the DJI Phantom 3 Professional fitted with a 12-megapixel camera (f/2.8 78 lens 94 0 field view). Image capture process was done without the presence of the orangutans during 79 daytime for safety purposes (for the researchers and the orangutan). Our skilled licensed drone 80 pilots managed to fly the drone as close as within 1 to 3 meters of the nest. The side and top view 81 of the nest were recorded at a screen resolution of 72 dpi (dot per inch) which equals to 300 dpi in 82 print resolution. The digital scale measurement of the nests' structure was estimated by using leaf 83 samples of the tree where the nests were recorded. We used a slingshot to obtain the leaf samples 84 from the nearest branch of the nest. The leaf samples were measured by using a ruler to obtain the 85 length. Ten samples of the leaves were measured and we used the average value. In the ImageJ 86 software, we used the measured length as the digital scale for the overall image analysis. ImageJ      34.63% of the nests were closed nest (Fig 4).

127
There is a significant difference of nest depth based on the nest class (Kruskall-Wallis, χ² = 13.408, 128 df = 2, P = 0.001) (Fig 5). The depth of nest decreased with the nest class. Class 3 nest recorded 129 the thinnest depth (34.560 ± 3.613 cm) compared to nest from class 1 (52.200 ± 2.032 cm and class 130 2 (57.000 ± 2.430 cm). There was no significant difference for nest class according to the nest 131 width (F (3,77) = 2.187, P = 0.097. Tree height and nest height showed a strong positive and 132 significant correlation (r s =0.7867, p=0.0001) (Fig 6).  The aim of this study was to determine the variation of nest structure quantitatively using advance 144 drone and image analysis software. We aim to promote our method as a new approach in nest  Orangutans might have avoided small trees as they preferred strong and sturdy branches to support 158 their concave shape nests. This is similar to chimpanzee nest selection preference, in which they 159 preferred sturdy, strong and comfortable spot to build nest [1,[5][6][7].
Our results showed no significant differences between the nest sizes cumulatively. However, 161 several studies have shown that nests size increases with the age group. This is related to the body 162 size where adult flange male with large body size built larger nest compare to the juvenile with a 163 small body [7,13]. In our study site, since we did not detect any adult flanged male nesting and 164 most of the nests that we sampled were frequented by orangutans within a similar age group and 165 body sizes. There is a possibility that a larger sample size could provide more information in the 166 future.

167
The significant correlation between tree height and nest height supports the hypothesis which 168 orangutans were more likely to build the nest at the very top of the tree. The decaying process 169 might also have contributed to the significant difference between nest class and nest depth [10].

170
Class 3 is considered as the second last stage of nest decay. Orangutans usually repairs certain nest 171 spots that are their favorite, which is why nest class 1 and 2 is frequently used. However, the decay 172 rate or factors affecting the decay process and decay process related to nest class quality were not 173 recorded in this study. There is also the possibility that the depth of the nest was influenced by the 174 orangutan body weight. Orangutan would position themselves in the middle of the nest and their 175 body weight would depress the whole nest. Unfortunately, since we did not specifically track 176 individual orangutans to a specific nest, we could not confirm this effect. There was also no 177 assessment related to the animal physical measurement such as weight and body size as we were 178 not allowed to physically touch the orangutans. Orangutans shows intelligence in nest building by choosing the best materials from hard and 184 durable tree species such as E. zwageri tree (commonly known as a Belian tree). Compared to 185 other tree species, the Belian tree would provide the orangutans a sturdy and strong nest to support 186 their weight. This valuable timber is widely used by human to produce furniture, medicinal 187 purposes as well as in traditional rituals [23][24][25][26]. The frequent usage of this tree species also 188 indicates that orangutans chose their nesting site based on strong and sturdy tree more for their 189 comfort [27]. 190 In natural settings, orangutans are solitary and have the tendency to avoid crowd and predators 191 including human [4,28,29]. In this study, we noticed that most of the orangutans have become 192 habituated with the human presence since most of the nests recorded were located near the feeding 193 platforms and boardwalk; a place where the visitors could access and observe the orangutan's 194 activities. We also suspected that this phenomenon was also due to food availability since the food 195 will be given to them every 10 a.m. and 3 p.m. regardless to the presence of visitors.

196
However, we must highlight the fact that the orangutans in Sepilok were in a rehabilitation 197 program which indirectly means that they were already familiar with human presence since they 198 have been exposed to human care. To release them into the wild, the juvenile orangutans were 199 released in the rehabilitation area within the 4,294-hectare forest. This was the next step in 200 preparing the orangutans to be released in the wild. The orangutan need to forage, build their own 201 nest, and subsequently making them less dependent on human care. Nest building skills is a very 202 crucial and important skill for orangutan to survive in the wild. Therefore, research on the nest 203 structure of orangutan is a very important. We propose that our method would be used as a standard