Use of new technologies for improved estimates of Tasmanian mammal abundance

(Honours project)

Screenshot_2018-07-23-15-02-22-1.png

Population density is a useful parameter in the study of animal ecology. Changes in population abundance reflects the processes and effects of ecological interactions which are fundamental to the conservation, research and management of wildlife populations. Abundance estimates of forester kangaroo (Macropus giganteus tasmaniensis), Tasmanian pademelon (Thylogale billardierii), Bennett’s wallaby (Macropus rufogriseus) and bare-nosed wombat (Vombatus ursinus) have been done in the past via walked-transect surveys, camera-trapping and/or spotlight surveys (Taylor, 1993; Southwell, 1994; Le Mar et al., 2001; Dique et al., 2003; Borchard and Wright, 2010).

Unmanned aircraft vehicles (UAV), or drones, are an emerging powerful tool in wildlife ecology that obviate many of these logistical and observational difficulties (Gross, 2014; Schiffman, 2014; Christie et al., 2016). Drones provide remote-sensing data at high-resolution temporal and spatial scales (Gross, 2014; Hodgson and Koh, 2016). A camera-equipped drone produces images of a study area which can later be post-processed at the desktop using object-recognition software to count the number of animals present in that landscape. Given these obvious benefits, what might the future hold for the use of drones in ecology?

This study aims to investigate new survey methods for estimating population abundance of forester kangaroo (Macropus giganteus tasmaniensis), Tasmanian pademelon (Thylogale billardierii), Bennett’s wallaby (Macropus rufogriseus) and bare-nosed wombat (Vombatus ursinus).

The key questions I want to address:

  1. Are drone surveys a feasible survey technique to be used to estimate macropods and wombat population in Tasmania?
  2. What are the behavioural responses of macropods and wombats to unmanned aerial vehicles?

The data will be collected across a range of land-use types in Tasmania, including: Narawntapu National Park, Bronte Park, Huon Valley and The University Reserve. Two drone models – Mavic Pro and Phantom Pro 4 will be used in the study. The drone surveys aim to determine the optimum altitude for macropods and wombat observation and monitoring, as well as to investigate the impact of drone noise level (dependent on flight height and drone model) on the study species.

  • Drone transect surveys (optimum altitude for macropods monitoring)

Drones will be flown along waypoints of 40m spacing along a total transect length of 200m at a speed of 2m/s, at altitudes of 40, 60, 80, 100, and 120m. Each flight will spend approximately 20-30 minutes. I will first use Mavic Pro, and then Phantom 4 Pro to conduct the surveys. The animal response will be recorded.

  • Drone transect surveys by horizontal flight (optimum altitude for macropods monitoring)

Horizontal flights will be operated along a 200 m transect at different altitudes of 40, 60, 80, 100, and 120m. From the take-off point, the drone will be ascended vertically to 120 m. Horizontal flights above the animals will then be carried out at a speed of 2m/s. After passing from a vertical perspective above the animals, the drone continued its flight horizontally towards to end of the transect, and then descended to a lower altitude. Each flight will spend approximately 20-30 minutes. Again, I will first use Mavic Pro, and then Phantom 4 Pro to conduct the surveys. The animal response will be recorded.

Phantom 4 Pro

Aerial shot taken from Phantom 4 Pro

 

The animal behaviour and response towards the drones will be measured using a scale:

 

Scale Definition
0 No discernible behavioural response
1 Freeze and look towards drone; increased movement rates
2 Moving away from the study area

 

When the animals are showing a response of scale 2, drone survey will be discontinued and the drone will be removed from the study area.

  • Noise level study

Drones will be flown along waypoints of 40m spacing along a total transect length of 200m at a speed of 2m/s, at altitudes of 40, 60, 80, 100, and 120m. Each flight will spend approximately 20-30 minutes. A standard microphone will be placed at the end of the transect. One of the researchers will record the sound of the drone engine from ground level when the drone is flown at different altitudes. Animal response will be recorded on data sheets. The noise level of drones will be determined by recording the sound of drone motors from the ground level using a standard microphone, calibrated to dB (sound will be recorded at each height level). The noise level study will be measured on clear, sunny days as well as on overcast or windy days, to compare the wind speed and the noise level under different weather conditions. Visual observations of habitat type and other environmental conditions will be recorded at the time of data collection.

  • Model study using drones (study image resolution)

30 tennis balls with different colours will be randomly placed on a 5 x 5m plot. Drones will be flown at altitudes of 40, 60, 80, 100, and 120m on the same axis at a speed of 2m/s. Images will be taken at each altitude. The position of the tennis balls will be changed after the first survey using both drones. Then, the drones will be flown again at different altitudes and images will be taken.

Aerial drone imagery will be interrogated visually (at known strip dimensions by height), and I will also trial the use of neural network classification for automated detection.

The major risks are associated with drone operations and commuting between locations. When drone surveys are conducted on windy days, it increases the risk of drones being swept away by strong winds and losing connection with the controller. To minimise the risk, wind speed and direction of the study location will be checked via Bureau of Meteorology (BoM) before a drone survey is undertaken. Drone operation will be cancelled in poor weather conditions.

  • Teo Yee Von

Check out what we are doing in our research group: DEEP lab

 

References

Borchard, P., and Wright, I.A. (2010). Using camera-trap data to model habitat use by bare-nosed wombats (Vombatus ursinus) and cattle (Bos taurus) in a south-eastern Australian agricultural riparian ecosystem. Australian Mammalogy 32, 16-22.

Dique, D. S., de Villiers, D. L., and Preece, H. J. (2003). Evaluation of line-transect sampling for estimating koala abundance in the Pine Rivers Shire, south-east Queensland. Wildlife Research 30, 127-133.

Le Mar, K., Southwell, C., and McArthur, C. (2001). Evaluation of line-transect sampling to estimate nocturnal densities of macropods in open and closed habitats. Wildlife Research 28, 9-16.

Southwell, C. (1994). Evaluation of walked line transect counts for estimating macropod density. The Journal of Wildlife Management, 348-356.

Taylor, R. J. (1993). Observations on the behaviour and ecology of the common wombat Vombatus ursinus in northeast Tasmania. Australian Mammalogy 16, 1-7.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s