Urban expansion is a major threat to natural ecosystems but also creates novel opportunities that adaptable species can exploit. The grey-headed flying-fox (Pteropus poliocephalus) is a threatened, highly mobile species of bat that is increasingly found in human-dominated landscapes, leading to many management and conservation challenges. Flying-fox urbanisation is thought to be a result of diminishing natural foraging habitat or increasing urban food resources, or both. However, little is known about landscape utilisation of flying-foxes in human-modified areas, and how this may differ in natural areas. Here we examine positional data from 98 satellite-tracked P. poliocephalus for up to 5 years in urban and non-urban environments, in relation to vegetation data and published indices of foraging habitat quality. Our findings indicate that human-modified foraging landscapes sustain a large proportion of the P. poliocephalus population year-round. When individuals roosted in non-urban and minor-urban areas, they relied primarily on wet and dry sclerophyll forest, forested wetlands, and rainforest for foraging, and preferentially visited foraging habitat designated as high-quality. However, our results highlight the importance of human-modified foraging habitats throughout the species’ range, and particularly for individuals that roosted in major-urban environments. The exact plant species that exist in human-modified habitats are largely undocumented; however, where this information was available, foraging by P. poliocephalus was associated with different dominant plant species depending on whether individuals roosted in ‘urban’ or ‘non-urban’ areas. Overall, our results demonstrate clear differences in urban- and non-urban landscape utilisation by foraging P. poliocephalus. However, further research is needed to understand the exact foraging resources used, particularly in human-modified habitats, and hence what attracts flying-foxes to urban areas. Such information could be used to modify the urban foraging landscape, to assist long-term habitat management programs aimed at minimising human-wildlife conflict and maximising resource availability within and outside of urban environments.

This study analysed P. poliocephalus satellite tracking data collected between 9th May 2012 - 27th April 2017 in NSW. Data were subsetted such that of 100,463 data points, all 51,585 high quality ARGOS location data classes 2 and 3 were initially retained. Positional fixes of these classes are estimated to be accurate to within 250 m and 500 m of the true location, respectively [55]. To investigate foraging locations, following Meade et al. (2021), we selected all positional fixes collected during the 10 h ‘on’ periods for which both daytime and night-time location data were available. The daytime fix allowed the roosting colony to be identified, and the night-time fix furthest from the roost site was selected as the assumed foraging location. This resulted in 5,118 paired roosting and foraging locations. Next, we excluded all paired locations (n = 52) where their distance was greater than 50 km, as 99% of foraging takes place within 50 km from a roost [54, 56, 57] so that greater distances likely represent movements between roosts. Recent research suggests that P. poliocephalus individuals travel directly to a foraging site early in the night and then undertake smaller movements between foraging sites before returning to the roost [28]. Thus, while we cannot be certain that these locations are ‘foraging locations’ it is likely that the location furthest from the roost site in a night is in an area that an individual was foraging. Finally, we subsetted the data to those animals foraging in NSW to allow for comparison with available data layers (below), resulting in 4,198 paired roosting and foraging locations for 98 of the 99 tracked individuals in this study area.

Land-use categories were extracted for each of the foraging locations in NSW (n = 4,198). For this we used a shapefile of Urban Centre and Locality data obtained from the Australian Bureau of Statistics [58], to classify NSW into three land-use categories. Land was defined as ‘of urban character’ based on dwelling density and population density [59]. ‘Major-urban’ areas were defined as urban centers with a population of > 100,000 [59]. ‘Other-urban’ areas were urban centers with a population of between 1,000 and 99,999 [59]. All other areas were defined as ‘non-urban’. For ease of interpretation, we refer to the ‘other-urban’ land-use category as ‘minor-urban’ throughout.

Vegetation types were extracted for each of the foraging locations in NSW (n = 4,198). For this we used the Vegetation Formations and Classes of NSW (version 3.03 - 200 m Raster) to classify vegetation type in NSW to 16 core classes [60]. The raster was created and published in 2012 and is thus concurrent with our tracking data. In the Vegetation Formations and Classes of NSW, ‘cleared land’ is defined as land that is not structurally intact native vegetation [61]. Thus, cleared land comprises human-modified land including agriculture, parks, gardens, and tree-lined streets. Cleared land may also include small remnant patches of native vegetation up to 2 hectares. For clarity, we refer to ‘cleared land’ as ‘human-modified land’ henceforth.

Flying-fox foraging habitat quality ranks from Eby and Law (2008) were available for 3,757 of the n = 4,198 foraging locations in NSW. These habitat quality ranks are based on a complex algorithm incorporating the spatial availability of known P. poliocephalus blossom food plant species and indices of productivity and nectar flow, as well as species richness scores of fruit food plant species [62]. Here, habitat quality was ranked from 1 (high quality) to 4 (poor quality), and areas were ranked as 0 if neither the dominant nor subdominant species were known P. poliocephalus food plant species. We extracted likely food plant species from vegetation shapefiles from Eby and Law (2008). These vegetation shapefiles only contained food plant species in the blossom diet of P. poliocephalus, as insufficient data were available on the productivity and reliability of food plant species in the fruit diet [62]. Eby and Law’s diet plant list comprised 59 species in the blossom diet including species from the Myrtaceae, Proteaceae, Arecaceae, Fabaceae, and Pittosporacea families. Only dominant and sub-dominant species [63] were considered resulting in a list of 55 species (see Table 4.1 in [62]). Habitat quality rank data were split into bi-months to account for seasonal variations in flowering phenology of the food plant species (Eby & Law 2008); December-January, February-March, April-May, June-July, August-September, October-November. Where more than one dominant or sub-dominant food plant species was available in the bi-month that a foraging fix was recorded, the species that flowered most often and that was most abundant was selected as the most likely food plant species [62]. We used a shapefile of P. poliocephalus’ range [40] and of NSW [64] to clip all data layers.

Definition of habitat rank values:

Habitat ranks are a representation of the 'quality' of habitats for P. poliocephalus foraging. Habitat quality ranks were extracted from Eby and Law [62], and are based on a complex algorithm incorporating the spatial availability of known P. poliocephalus blossom food plant species and indices of productivity and nectar flow, as well as species richness scores of fruit food plant species [62]. Here, habitat quality was ranked from 1 (high quality) to 4 (poor quality), and areas were ranked as 0 if neither the dominant nor subdominant species were known P. poliocephalus food plant species. Flying-fox foraging habitat quality ranks from Eby and Law [62] were available for 3,757 of the n = 4,198 foraging locations in NSW.  

[62] Eby P, Law B. Ranking the feeding habitat of grey-headed flying foxes for conservation management. Department of Environment, Heritage, Water and the Arts, Canberra. 2008.