Inventory using a drone
A DJI Phantom drone (version 4 Pro V2.0) was used for the fieldwork. This is a remotely controlled quadcopter device with a total weight of 1,388 grams, equipped with a camera capable of taking both still photographs (max quality 5,472x3,648 pixels) and videos (max quality: 4,096 x 2,160 pixels), with the possibility of continuous tracking of unrecorded images transmitted to the display coupled with the remote control. Each photo records the geographic location and altitude in the metadata.
A drone flight was performed over sites where birds were known to be regularly present (Ławicki et al. 2010; Marchowski et al. 2018). The distance from the observer to these sites was usually several hundred metres, but never less than 100 m. This precluded any influence on the part of the observer on the birds’ behaviour. But if birds were scared away as a result of the observer’s presence, no attempt was made to fly the drone. The flight height was set at about 100 m – the same as recommended for bird counts from aircraft (Meissner 2011). From this height, the planned census area was scanned for birds. If a flock or individual birds were spotted on the remote controller display, a photo was taken, and the birds were approached. Up to a height of 30 m, birds were approached diagonally in such a way that the height and distance from the birds were about the same. After reaching 30 m (when the birds were still not responding to the drone), the drone was moved over the birds, the distance was shortened while the height was maintained, after which the drone was lowered to about 10 m. The birds’ reactions (if the terrain permitted them to be seen) were recorded by a second observer using a spotting scope mounted on a tripod. The birds’ reactions were also monitored in real time via the display on the drone’s remote controller. The behaviour of the birds was observed on a sample plot of approximately 40,000 m² (200 x200 m), but if the site was smaller (e.g. a small lake, pool or islet), the entire site was treated as a sample plot.
My main research question was whether it would be possible/safe to conduct bird research using a drone. How do birds react to a drone: do they ignore it, attack it, or does the drone flush them? The birds’ reactions to the drone were first divided into two basic categories: reaction – no reaction. In the reaction category, the following subcategories were created: the birds 1 – moved slowly away, 2 – dived, 3 – were flushed over a short distance but remained in the sample plot, or quickly returned, 4 – were scared away, exhibited a panic reaction, left the sample plot and did not return, 5 – attacked the drone.
Reactions 4 and 5 were placed in an “unfavourable” category, i.e. where the study of birds using a drone would not be recommended. No reaction and non-invasive reactions (1-3) were placed in a separate category. Statistical significance was tested using the chi-square goodness of fit. The statistical analyses were carried out using the software program R (R Core Team. 2021).
The second aim was to test the ability to count birds in non-breeding aggregations or nests in breeding colonies using a drone. After the initial recognition from a long distance of whether the birds were responding to the drone’s presence, attempts were made to approach the colony, with photos being taken continuously. If it was noticed that the drone had disturbed the birds in any way, it was recalled beyond the disturbing distance. The dates of drone flights over the breeding colonies were adjusted so as to take place when the colony was at the egg incubation stage. As a result, most of the nests present in the colony, represented by an incubating bird, were visible.