Our study took place at the Chain Islets and Great Chain Island IBA, a cluster of small islets and rocky outcrops along the marine coastline south of the City of Victoria in the Strait of Juan de Fuca (48.42°N, 123.28°W), British Columbia (Fig. 1). This site also comprises the Oak Bay Islands Ecological Reserve. The IBA is 1.38 km² in size and consists of rocky flats and rocky shorelines with grass and sedge meadows, and sand flats throughout. The rocky islets within the IBA are sparsely vegetated, Great Chain Island is the largest and the most vegetated island with a herbaceous cover. Both Great Chain and the rocky islets are used by nesting and migrating seabirds. As this site lies within an introgression zone (Bell 1996), our references to Glaucous-winged Gulls include any Western Gull (L. occidentalis) × Glaucous-winged Gull hybrids.

Figure 1.

Great Chain Island & Chain Islets IBA, British Columbia, Canada. Environment and Climate Change Canada, 2022.

RPAS flight operations and protocol were selected to be consistent with a previous study in the region (Blight et al. 2019). Drone imagery for this study was collected using a multi-rotor RPAS (DJI Mavic 2 Pro) with a 20-megapixel Hasselblad L1D-20c aerial camera. Flights were performed on 21 June 2019 from 0859 to 1257 h, with operations run from the aft deck of a 42-foot former commercial fishing vessel. The survey date was chosen based on known local breeding phenology of Glaucous-winged Gulls and was meant to capture the maximum number of occupied nests, when breeding adults are likely either incubating eggs or brooding newly hatched chicks (Blight et al. 2014). Weather conditions on the day of the flight were clear skies with flat, calm water and minimal wind speed (~Beaufort 0–1).

Flights were pre-programmed through drone mapping software (DroneDeploy, www.dronedeploy.com) to run single straight-line paths across the colony, offset to achieve consistent sidelap and followed by a return to the stationary vessel. Due to strong tidal currents in the region, the vessel motored in place instead of dropping anchor, with all vessel operations occurring outside the Ecological Reserve boundary, approximately 650 metres from Great Chain Island; this distance was well beyond the 100 m launch distance recommended by Vas et al. (2015) for minimising effects on waterbird behaviour. The drone climbed vertically until survey altitudes were achieved and then flew horizontally over the colony. The altitude above ground level of the drone flights varied from 90–65 metres, with initial flights starting at 90 meters, a height at which we observed no disturbance in an earlier study in the region (Blight et al. 2019). Flight altitude was adjusted according to the natural topography of the islands and generally decreased for each subsequent flight after no Glaucous-winged Gull disturbance was observed.

Image capture was via a 28 mm lens on auto-exposure mode, certain flight specifics such as speed over ground and photo intervals were set by front-lap and side-lap requirements within the drone software. As flight altitudes and topography were variable throughout the six flights, ground sample distance (GSD) also varied. GSD ranged from 1.41 cm/pixel at 60 meters altitude to 2.11 cm/pixel at 90 meters. Drone flights were operated using one pilot and one visual observer, along with two additional observers monitoring the colony for any evidence of disturbance to nesting gulls during drone operations.

We used photogrammetry software to process drone flight images and create orthomosaics of the study area. Clear weather on the day of the flight generally created optimal light conditions for subsequent visual assessment of the orthomosaic imagery. However, solar glare on the less-vegetated small islets (approx. 33% of the colony) rendered a subset of the images overexposed; these islets were excluded from subsequent analysis. Orthomosaics of Great Chain were enlarged and visually scrutinised on-screen for occupied gull nests; these were readily apparent as the white and pale grey plumage of an incubating gull contrasts strongly with its background of rock and vegetation, and its posture is distinctive (Fig. 2). Thus, we assumed that each incubating bird represents a nesting pair; as two mates may attend one nest, two birds sitting side by side in an incubation pose were counted as a single pair. The Glaucous-winged Gulls we identified were primarily on nests, birds flying overhead and apparently roosting along the shoreline were also visible but were not included in the total nest count. As each individual nest was counted, a pin was placed over it to eliminate duplicate counts. The orthomosaic encompassed Great Chain Island as well as several of the small islets situated within the IBA (Chain Islets; Fig. 1).

Figure 2.

Gull nests detected within an enlarged portion of orthomosaic imagery.

Our count results were similar to ground-based survey data collected a decade earlier (Blight et al. 2015) as well as more recent estimates by Ecological Reserve wardens (J. Sirois and M. Lambert, pers. comm). Our drone-based count confirms that the breeding population now falls below the KBA threshold to designate this site, a finding consistent with regional declines in the species (Blight et al. 2015). The estimated total of 2952 individuals accounts for approximately 67% of the threshold needed to classify this site as a KBA. Currently, to reach a national threshold for Glaucous-winged Gull, an aggregation representing 1% or greater of the national population size over a season, and during one or more life history stages is required, resulting in a threshold of 4,400 individuals. Globally, the same requirements are necessary, resulting in a threshold of 4,700. While this specific site no longer qualifies for KBA status, there is potential to extend the boundaries of this site to create a larger KBA with the inclusion of a similar sized colony nearby, Mandarte Island. These two sites, combined with Mitlenach Island in the northern Strait of Georgia, account for a quarter of the breeding gulls in British Columbia (Booth 2001). In Pacific Canada, there are 10 other IBA sites with Glaucous-winged Gull as the trigger species that could be candidates for future RPAS analysis.

Canada is one of the first countries globally to transition IBAs to the KBA program. Two main challenges identified in the program crosswalk relating to survey data are data age and repeatability (Birds Canada, pers. comm.). “Lack of population data” was listed as a key methodological issue in six out of seven regions in an early assessment of KBA progress globally (Foster et al. 2012) and it remains a challenge in the development of the program. While drones and other remote technologies are not a panacea for this problem – particularly where lack of funding is a key issue – expanding the type of approaches available is important for some data-deficient sites, as demonstrated by this study. Recent literature explores the potential use of semi-automated counts on drone imagery of breeding waterbirds through free software and highlights the effectiveness and accessibility of this method as a low-cost alternative to manual counts for conservation use (Francis et al. 2020; Corregidor-Castro and Valle 2022). The cost of this survey was comparable to the 2-day field operation required to undertake this census as a ground-based survey. This semi-automated method can be especially cost-effective for areas such as IBAs that need accurate repeat surveys and for large areas that are difficult to access (Francis et al. 2020).

Though the use of RPAS technology has rapidly increased, gaps remain in establishing best practices for surveying sensitive species (Vas et al. 2015) and literature on methods and guidelines for this approach continues to evolve (Vas et al. 2015; Rush et al. 2018; Weimerskirch et al. 2018; Francis et al. 2020). The technique utilized in this study caused no evident disturbance to nesting birds, as was anticipated based on previous work in the region (Blight et al. 2019). However, our study’s IBA is located less than 2 km from the City of Victoria and birds may be habituated to air traffic due to their proximity to the Victoria aerodrome. Though Glaucous-winged Gulls are considered a non-sensitive species in this context, lack of disturbance is an important survey factor for colonial bird studies, especially during sensitive life cycle stages including breeding. Geldart et al. (2022) studied the heart-rate response of drone surveys on colonial-nesting Eider ducks (Somateria mollissima), in Nunavut, Canada; their results indicated no physiological stress response from drones flying 30 m above the nesting colony. However, the authors highlight that other species may perceive a greater threat to drones, specifically avian species that may experience adult predation from other avian predators (Geldart et al. 2022). This study, among others, allows researchers to better understand the behavioural and physiological response of birds to drones and highlights that, when done properly and in accordance with available guidelines to avoid disturbance to seabird and waterbird colonies, RPAS can allow for a less invasive surveying method that provides adequate coverage of a site in significantly less time compared to land-based surveys.

Drone surveying can not only assist in the establishment of KBAs but can provide critical population and habitat monitoring which can indicate habitat loss or degradation. It is evident that climate change will continue to pose stressors that may lead to fluctuations in biodiversity, potentially impacting elements for which a conservation site was initially designated (KBA Canada Coalition 2021). Bollard et al. (2022) explore drone technology for mapping protected areas in Antarctica to provide baseline vegetation inventories and ecological data for future monitoring. This method proved useful in obtaining vegetation maps that show spatial cover with higher accuracy and resolution, though small intricate taxa such as lichen and cyanobacteria could not be accurately monitored through this method (Bollard et al. 2022). The low impact of this survey method at a sensitive site, combined with its repeatability and inexpensiveness, made this a successful demonstration of collecting baseline data for a protected area.

While our study only used RPAS to obtain a viable count of breeding gulls, the imagery could be used to indicate the area of occupancy, the extent of suitable habitat, and the range of such habitat for this and other species; these are all metrics that are considered acceptable to assess qualifying KBA sites (KBA Canada Coalition 2021). Further, such metrics can be monitored over time to indicate future changes to habitat and ecosystems. The drone imagery was high-resolution enough and the visibility on our survey day clear enough for us to view the extent of the underwater kelp (Nereocystis luetkeana) forests surrounding the archipelago, as well as macroalgae in the intertidal zone (cf. Cavanaugh et al. 2021; Tait et al. 2021), indicating the utility of drone imagery for assessing multiple species and habitats at a target site. This methodology could prove useful for assessing a range of species across various landscapes to support the KBA assessment process.