Environmental conditions and resource availability are key determinants of the distribution of wildlife populations across a landscape (Orians and Wittenberger, 1991; López-López et al., 2014; Baruch-Mordo et al., 2014). Individuals are likely to be found in and move to locations where the probability of encountering resources is high (Stephens and Charnov, 1982; Jones et al., 2014). Resources can vary across space or through time, and the spatial aggregation, availability, and predictability of high-quality resources can all influence wildlife foraging movements (Pyke, 1984; Saracco et al., 2004; Jonzén et al., 2011; Riotte-Lambert and Matthiopoulos, 2020). Optimal foraging strategies suggest individuals will try to maximize nutritional benefits while limiting energetic costs (Stephens and Charnov, 1982; Pyke, 1984) and reducing the risk of predation (Lima and Dill, 1990). Anthropogenic land use change creates novel landscapes for animals to forage in, and ultimately alters the availability and distribution of foraging habitats for wildlife (Corlett et al., 1992; Foley et al., 2005; Bradshaw, 2012; Chase et al., 2020).

Across diverse taxa, animal populations are changing their foraging behavior in response to anthropogenically modified landscapes. Migratory seabirds forage in landfills (Corman et al., 2016) and fishery discards (Patrick et al., 2015). Black bears (Ursus americanus) forage in urban areas, regardless of native food resource availability (Merkle et al., 2013). Similarly, Australian flying foxes (Pteropus sp.) are urbanizing and changing their foraging behaviors, perhaps in response to profound environmental changes (Tait et al., 2014; Meade et al., 2021; Yabsley et al., 2021). This is particularly concerning, considering the beneficial role they play as pollinators (Eby, 2016) in the heavily cleared native eucalypt forests (Bradshaw, 2012) and the zoonotic disease risks they present as reservoir hosts for viruses, such as bat lyssaviruses and Hendra virus (Eby et al., 1999; Plowright et al., 2011; Mahalingam et al., 2012; Eby, 2016; Peel et al., 2022).

Australian flying foxes are nomadic, nocturnal nectivores and frugivores that roost in canopy vegetation during the day (Nelson, 1964; Marshall, 1985; Eby, 1991). They have typically foraged on small fruits such as figs and nectar from members of the Proteaceae and Myrtaceae families, notably Eucalyptus, Melaleuca, Banksia, and Corymbia species (Ratcliffe, 1931; Nelson, 1964; Marshall, 1985; Palmer, 1997; Markus and Hall, 2004; Eby and Law, 2008). Flying foxes historically made long-distance movements (>300 kilometers (km)) across their range to follow the seasonal flowering of their diet resources (Eby, 1991; Roberts et al., 2012), forming aggregations larger than 100,000 individuals when nectar was in great abundance (Ratcliffe, 1931). Recently, these large aggregations have become less common (Eby et al., 2023), and flying foxes have been observed roosting in smaller groups with roosts becoming more numerous across eastern Australia (Baranowski and Bharti, 2023), particularly after winters when typical diet resources were scarce (Eby et al., 2023).

Studies of flying fox diets reveal insights into their recent changes in foraging ecology. Meade et al., 2021 found grey-headed flying foxes (Pteropus poliocephalus) in urban areas primarily forage on a mixture of native and anthropogenic resources, including exotic species. Similarly, the availability of human-planted urban resources can attract flying foxes to areas beyond their historic range, such as the Melbourne and Adelaide Botanic Gardens (Van Der Ree et al., 2006; Yabsley et al., 2021). Studies that tracked black flying foxes (P. alecto) with radio telemetry in the urban center of Brisbane showed no individuals fed on any eucalypt species in winter (Markus and Hall, 2004). Rather, animals foraged on native and invasive fig species (Ficus sp.), and invasive Cocos palm (Syagrus romanzoffiana). Field et al. similarly found black flying foxes primarily fed in fragmented lands on weedy species, such as climbing asparagus vine (Asparagus africanus) and wild tobacco (Solanum mauritianum), as opposed to a typical blossom diet when foraging from a roost with little forest nearby (Field et al., 2016).

It is hypothesized this shift in resource consumption is driven by the increased spatiotemporal availability and reliability of these atypical diet resources (Welbergen et al., 2020; Meade et al., 2021; Yabsley et al., 2022), over typical Eucalyptus diet resources, which may flower only once over 1-5 years, depending on the species (Law et al., 2000). Longitudinal studies of flying foxes from 2012-2017 found their foraging movements were more quasi-random, and individuals could be drawn to urban or forested areas that are temporarily productive (Welbergen et al., 2020). These recent observations of flying fox winter foraging ecology are particularly concerning as they could indicate increased contact rates between human systems and black flying foxes, the main reservoir responsible for spillovers of Hendra virus.

Hendra virus is a zoonotic paramyxovirus that spills over from Pteropus bats to horses (Equus caballus) and subsequently to humans and other domestic animals (Murray et al., 1995; Plowright et al., 2011; Mahalingam et al., 2012). Nearly all known spillover events through 2023 (64/66) have occurred within the range of the black flying fox along the eastern coast of Queensland and northeastern New South Wales, Australia (Queensland Government Business Queensland, 2021). Two-thirds of all documented Hendra virus spillovers have occurred within the Austral winter months of June, July, and August, which coincides with a scarcity of typical native diet resources (Eby and Law, 2008). Studies on Hendra virus have linked winter food shortages (Becker et al., 2022), poor body condition (Edson et al., 2019), and flying foxes overwintering at new roosts containing less native winter habitat than historic overwintering roosts, (Becker et al., 2022) with greater detection of viral RNA in flying foxes (Edson et al., 2019) or their excreta (Becker et al., 2022). However, given the changes in flying fox movement and foraging ecology, it’s unclear how winter diet resources in roost foraging areas influence black flying fox winter distribution and roost occupancy.

We assess the relationships between typical and atypical winter foraging habitats, monthly weather conditions, and the winter roost occupancy of black flying foxes in Queensland from 2007 to 2020. We focus on these years due to the notable increase in the frequency of Hendra virus spillovers (Queensland Government Business Queensland, 2021). Given the findings from prior foraging studies on flying foxes, we hypothesized that black flying fox winter roost occupancy would be positively correlated with roosts containing abundant and annually reliable typical and atypical winter habitats. We also hypothesized that roosts with dispersed winter diet resources would be occupied more frequently than roosts with aggregated resources in their foraging areas, as black flying foxes have been noted to preferentially forage in fragmented forests (Field et al., 2016). We find that the location of roosts was associated with high reliability for typical winter diet species but the presence of black flying foxes at roosts in winter was not driven by the abundance, spatial clustering, or reliability of native or non-native winter habitats. Modeling the interactions of temporally lagged weather conditions on winter habitats identified no consistent environmental variables that explained black winter roost occupancy in Queensland.

We used flying fox roost surveys in Queensland from the National Flying Fox Monitoring Program (NFFMP) (Australian Government of Agriculture, Water and the E, 2021) to identify monitored roosts that were used by black flying foxes from 2007-2020 ( Supplementary Figure S1 ). These surveys, completed by government employees and volunteers, produce serial animal counts for nationally known or established roosts. They focus on roosts used by grey-headed and spectacled flying foxes (P. conspicillatus) because these species are listed as ‘Vulnerable’ and ‘Endangered,’ respectively (Westcott et al., 2015) but counts of all Pteropus species present are recorded. The species-specific roost selection may bias the data to overlook roosts occupied solely by black flying foxes. However, black flying foxes and grey-headed flying foxes often co-roost (Welbergen et al., 2020). Over 97% (266/274) of the roosts occupied by grey-headed flying foxes in Queensland were also occupied by black flying foxes in at least one roost survey from 2007-2020. We present results for the estimated average foraging area, a 20-kilometer radius buffer, which is based on foraging flights measured with radio and satellite telemetry (Roberts et al., 2012), equaling 1256.64 km², around each flying fox roost (n=455) surveyed by the NFFMP ( Figure 1B ). See Supplementary Figures S2 , S3 for analysis of additional buffer sizes.

Black flying foxes occupied roosts with a wide range of typical winter habitat extents, and, to a lesser degree, atypical winter habitat extent, in their proximal foraging areas. Black flying fox winter roost occupancy was not explained by the amount of winter habitat in the 20km foraging area ( Table 1 , Figure 2A ), the spatial arrangement of those winter habitats ( Table 2 ; Supplementary Figure S4 ), or the reliability of those winter resources ( Figure 3C ). However, the frequency at which annually reliable winter diet species occur in foraging areas may influence where roosts are established on the landscape ( Figure 3B ). We also found no evidence that interactions between the average past weather conditions and winter habitat attributes explained winter roost occupancy ( Supplementary Figure S5 ). Black flying fox winter roost presence was extremely dynamic and poorly correlated to static land metrics of vegetation presence, distribution, and reliability.

This study examined the relationships between winter habitat type, abundance, reliability, spatial distribution, and black flying fox occupancy at roosts in winter months, the season when Hendra virus spills over most frequently in eastern Queensland. We found that black flying fox winter roost presence was not impacted by the abundance of typical or atypical winter habitats in the 20 km foraging buffer ( Figure 2A ). This relationship was also consistent when we examined the grey-headed flying fox winter presence and the presence of both flying fox species in summer months ( Figure 2B ) as well as other foraging radii ( Supplementary Figure S2 ). We found all roosts were located near winter habitat that was dispersed, as noted by negative Moran’s I values. We likewise found that all roosts were near patches that contained annually reliable typical winter diet species in foraging areas, and reliability scores similarly did not differ between roosts that were occupied or unoccupied. The statistically significant interactions between the weather conditions in the previous spring and winter and the various winter habitat attributes identified in the mixed-effects model were not supported by a close examination of those variable relationships ( Supplementary Figure S5 ). Patterns of black flying fox winter occupancy were similar across all ranges of precipitation conditions. We found that the winter presence of black flying foxes was interannually dynamic, and the spatial land metrics quantified here could not explain those dynamics.

Our analyses were unable to explain much of black flying fox winter roost presence and absence, likely due to complex drivers influencing their winter distribution. As dietary generalists (Palmer, 1997), it’s possible black flying foxes do not need to find optimal foraging habitats, and instead roost in areas that are merely sufficient enough. A study of black flying foxes from 2012-2017 found weak support that foraging movements follow seasonal environmental cues (Welbergen et al., 2020); black flying fox movement may no longer be as strongly influenced by seasonal foraging resources as it once historically was. Our work did not support the suggestion that flying foxes may be occupying roosts with access to more atypical (i.e. non-native) diet resources due to their increased spatiotemporal availability or productivity ( Figure 3C ). Our results suggest the abundance of winter habitat is not as relevant for flying foxes as the frequency of winter diet species in foraging areas ( Figure 3B ). Roosts with high frequencies of diet species may increase the chance for individuals to encounter irregular and geographically focused bursts of productivity in winter habitats, which is likely the strongest driver of flying fox winter roosting presence (Giles et al., 2016). Unfortunately, estimating winter flowering for eucalypts across this scale is challenging, and accounting for all the potential resources in anthropogenically modified environments, such as fruit trees in residential yards, is not possible.

It’s likely that our data sources and methodology overestimated winter habitat extent and biased the relationship evaluated here. We defined winter habitat as the presence of diet species in a vegetation community, but we did not have information on the relative abundance or spatial distribution of the diet species within a patch. We likely defined some areas as ‘typical winter habitat’ that may not have been sufficiently abundant or productive to attract foraging black flying foxes in winter. Additionally, the abundance and distribution of atypical winter habitats were drastically constrained by the availability of geospatial data describing their locations. Atypical winter diet species located far from human settlements or in areas not commonly visited by humans are unlikely to be captured in the GBIF database. Varying levels of roost survey intensity in the NFFMP ( Supplementary Figure S1 ) may have contributed to the weak relationship between the environment and roost presence identified here. Finally, the temporal gaps around the biennial vegetation data used here may not reflect actual winter habitat abundance at roost locations during the time of the survey if conditions were changing rapidly. Collectively, these limitations could have contributed to our findings of weak environmental factors associated with winter roost occupancy. More examination of black flying fox foraging movement will help characterize the predominant vegetation types used for foraging resources in winter. This can inform management efforts on strategies to minimize winter food shortages for flying foxes, which may reduce the risk of Hendra virus spillover.

Ecological solutions for spillover prevention, such as conservation of existing habitats and replanting of winter diet resources, have also been proposed as a way to increase resource abundance and bat foraging activity in native forests instead of agricultural and anthropogenic lands (Giles et al., 2018; Eby et al., 2023). This solution would also aid forest health and benefit other wildlife species such as birds and small marsupials (Eby, 2016). However, our findings do not indicate that the extent or spatial distribution of winter foraging habitats strongly influences winter black flying fox presence. Further, some flying fox diet species take 5-10 years to become mature and produce flowers so these solutions would not resolve immediate problems (Law et al., 2000). More rapid solutions in addition to vegetation management should be considered to support flying foxes during winter food shortages.

Supplementing the availability of fruits or high-quality, native nectar for flying fox populations may be a more immediate way to ensure there is sufficient food in winter and mitigate the potential for a food shortage. While supplemental feeding programs for wildlife may lead to unintended consequences such as aggression or disease transmission (Forristal et al., 2012; Sorensen et al., 2014; Jones et al., 2014), Hendra virus infection in bats was not found to be related to roost size or density (Lunn et al., 2022). The aggregation of flying foxes at feeding sites would likely not necessarily increase viral transmission in the population. Feeding locations could also be strategically focused away from urban and peri-urban areas to help spatially separate bats from human systems and minimize the proximity to agricultural lands with horses. Supplemental feeding done carefully could mimic the conditions of large productive blooms of native eucalypt trees, which would not increase Hendra virus risk. Additional interventions that limit horse exposure to bat excreta and horse vaccinations would have direct impacts further minimizing the risk of Hendra spillover to horses.

The need for a thorough investigation of environmental conditions that are important for reservoir hosts broadly applies to many emerging viruses, including Nipah virus and Ebola virus. These bat-borne viruses have caused outbreaks in humans with case fatality rates as high as 90%. Both of these viruses have been linked to the conversion of forested bat habitats to anthropogenically managed lands (Hsu et al., 2004; Gurley et al., 2017; Rulli et al., 2017; Olivero et al., 2017). To reduce bat-borne viral spillovers, it will be imperative to identify the factors that drive bat movements. Whether human-driven changes negatively impact the reservoir hosts’ health, increase the frequency of contact between humans and wildlife, or both, disentangling the influences of these processes will help explain how and why zoonotic pathogens emerge.