The environmental and ecological determinants of elevated Ross River Virus exposure in koalas residing in urban coastal landscapes

Koala populations in many areas of Australia have declined sharply in response to habitat loss, disease and the effects of climate change. Koalas may face further morbidity from endemic mosquito-borne viruses, but the impact of such viruses is currently unknown. Few seroprevalence studies in the wild exist and little is known of the determinants of exposure. Here, we exploited a large, spatially and temporally explicit koala survey to define the intensity of Ross River Virus (RRV) exposure in koalas residing in urban coastal environments in southeast Queensland, Australia. We demonstrate that RRV exposure in koalas is much higher (> 80%) than reported in other sero-surveys and that exposure is uniform across the urban coastal landscape. Uniformity in exposure is related to the presence of the major RRV mosquito vector, Culex annulirostris, and similarities in animal movement, tree use, and age-dependent increases in exposure risk. Elevated exposure ultimately appears to result from the confinement of remaining coastal koala habitat to the edges of permanent wetlands unsuitable for urban development and which produce large numbers of competent mosquito vectors. The results further illustrate that koalas and other RRV-susceptible vertebrates may serve as useful sentinels of human urban exposure in endemic areas.

The koala (Phascolarctos cinereus) is an important emblem of Australia's biodiversity and a globally-recognised Australian icon, yet many populations continue to decline 1 . This decline is largely attributable to wide-scale habitat loss, much of it unregulated. During the period from 2000 to 2017, over 7.7 million ha of Australia's wildlife habitat was cleared with less than 10% being subject to scrutiny under the Environment Protection and Biodiversity Conservation Act 2 . Of the land lost, ca. 1 million ha was known or potential koala habitat. Such significant habitat losses and associated population declines have resulted in the koala being listed as vulnerable or extinct in much of its historic range 3 . There are pockets where the koala is thriving, but these areas represent a fraction of its native range 4,5 . In the state of Queensland, the koala is currently listed as vulnerable in response to ongoing habitat loss and the threats of climate change 1,6,7 . The historical clearing of primary eucalypt forest habitat and accelerated coastal development in the southeast of the state has been associated with particularly sharp population declines [8][9][10] .
Habitat loss and climate change are not the only threats to the survival of the koala. Infectious disease has also contributed to its decline. The bacterium Chlamydia pecorum causes debilitating ocular and urogenital tract disease while the koala retrovirus (KoRV) has been implicated in host immunosuppression and exacerbation of chlamydial pathogenesis [11][12][13][14] . Koalas are also known to be infected by endemic mosquito-borne viruses such Barmah Forest virus (BFV) and Ross River virus (RRV) 15,16 , yet the threat of these viruses to koala health remains poorly studied. RRV is Australia's most medically important mosquito-borne virus that causes debilitating polyarthritis in humans, among other symptoms 17 . The pathology of RRV in marsupial hosts is currently unknown although myositis and arthritis are reported in mice 18 and domestic horses 19 . Primary vertebrate hosts Results RRV seroprevalence. More than 80% of koalas tested in the lacustrine and estuarine habitats surveyed had anti-RRV antibodies. Out of 218 koalas tested, 172 koalas were seropositive, 35 were seronegative, and 11 seroconverted from negative to positive over the course of the study (2015 to 2017). The low number of seroconversions during the study reflects the high prevalence of RRV exposure in each population prior to the beginning of the survey. Seroprevalence rates in female and male koalas were similar (84%, n = 95 vs. 82%, n = 61; χ 2 = 0.07, p = 0.80) and not influenced by environment (Table 2). This latter analysis contained data from 187 koalas after accounting for relocated individuals and animals with too few position locations (n < 5) to confirm their assignment to a particular environment. Overall seroprevalence in the estuarine and lacustrine populations surveyed was 82% (n = 73) and 85% (n = 83), respectively. Multivariate logistic regression analyses further revealed that seroprevalence increased with increasing koala age (Table 3), and that these increases were independent of koala sex (male vs female, OR = 1.1, 95% CI 0.5-2.7, p = 0.83) and the environment in which they resided (lacustrine vs estuarine, OR = 1.0, 95% CI 0.43-2.4, p = 0.98). Koalas aged between 2-4 years, 4-6 years, and > 6 years of age were 6.9 (95% CI 2.5-22.2, p < 0.001), 8.2 (95% 2.6-37.2, p = 0.001), and 23.7 (4.5-439.9, p = 0.003) times more likely to test RRV positive, respectively, relative to the baseline age group (< 2 years of age) (Fig. 1).
Koala tracking and home range comparisons. Home ranges were calculated for a total of 155 koalas across the lacustrine (54 females and 37 males) and estuarine (36 females and 28 males) environments after accounting for translocated koalas, those with inadequate sampling duration and those whose home ranges  www.nature.com/scientificreports/ failed to satisfy the Minimum Convex Polygon (MCP) 60% cutoff during visual assessment of home range asymptotes. Home ranges did not differ by environment but did differ by sex ( Fig. 2a; F 1, 151 = 24.79, p < 0.001) with males (mean = 6.1 ha, 95% CI 4.9-7.3 ha) having greater home ranges than females (mean = 3.0 ha, 95% CI 2.6-3.6 ha). However, no significant differences in home ranges was observed within a sex when considering serostatus ( Fig. 2b; F 1,151 = 1.57, p = 0.21). When home ranges were compared between age groups, males and females exhibited a similar trend of increasing home range size with increasing age (Fig. 2c). Significant differences (F 1,147 = 6.28, p = 0.01) between males and females were only observed within the oldest age group, wherein male home ranges (mean = 10.8 ha, 95% CI 9.3-12.4 ha) were 26% larger than those of similarly aged females (mean = 8.0 ha, 95% CI 7.0-9.0 ha). Differences between the sexes within the older age group may be attributable to the greater movement of older, dominant males during the mating season 27 . The home ranges of individuals that seroconverted during the study were similar (mean = 6.9 ha, 95% CI 3.4-10.4 ha) to those that were seropositive prior to the study or those that remained seronegative, but too few seroconverted to confidently compare the three groups. Table 3. Summary of seroprevalence results by koala age group and logistic regression outputs for the analysis of the association of koala age with RRV seroprevalence in koalas residing in coastal lacustrine and estuarine environments. Percentages are calculated for the number of individuals in each row. The data shown is only for animals for which their location could be confirmed (n = 187) after accounting for relocated individuals and animals with too few position locations (n < 5) to confirm their assignment to a particular environment.   (Table 4). Koalas favoured Eucalyptus tereticornis (blue gum) and E. siderophloia (grey ironbark) in the estuarine environment and E. racemosa (scribbly gum) and E. tereticornis in the lacustrine environment. Other species commonly used by koalas included Corymbia intermedia (pink bloodwood) and Melaleuca quinquenervia (broad-leaved paperbark) in the estuarine environment and Cinnamomum camphora (camphor laurel) and C. intermedia in the lacustrine environment. Tree selection between seropositive and seronegative koalas was similar with nine of the top ten most used tree species being shared between them (

Mosquito community composition.
As expected, the estuarine environment contained more of the saltmarsh associated species Aedes vigilax, Culex sitiens and Aedes alternans relative to the lacustrine environment that harbored a greater diversity and abundance of freshwater associated species such as Culex annulirostris and Aedes procax (Table 6). Cx. annulirostris dominated both habitats, accounting for 86% and 41% of all collections from the lacustrine and estuarine environments, respectively. While the abundance of mosquitoes in both environments was substantial, the estuarine environment had the greatest number of mosquitoes per surveillance event (11,199 ± 2,688 vs. 5,576 ± 2,696).

Discussion
The risk of arbovirus transmission to both humans and wildlife is influenced by the environment and is therefore spatially heterogeneous. Distinct patterns of increased human RRV risk in southeast Queensland (SEQ) appear to be driven by environmental changes in vector-vertebrate communities 21,28 , but little is known of the factors driving environmental and spatial risks of RRV transmission to endemic wildlife. In the present study, we demonstrate that RRV seropositivity in koalas can be far greater than that reported for other marsupial species 17,29 and other circulating arboviruses such as BFV 15,30 . Uniformity in exposure is related to the presence of the major RRV mosquito vector, Culex annulirostris, and similarities in koala movement, tree use, and age-dependent increases in exposure risk. The findings suggest that elevated RRV exposure in coastal koala populations is related to the confinement of remaining koala habitat to the edges of permanent wetlands that are unsuitable for further development and which produce large numbers of mosquito vectors. The discovery of high RRV prevalence in www.nature.com/scientificreports/ Table 4. Dominant tree species and characteristics of trees in which koalas were located during the study. Count represents the total number of observations (animal relocations) during which a koala was observed on an individual tree species. The data presented for tree diameter at breast height (DBH), tree height, and koala height in trees represents the mean data for all koalas surveyed in each environment. The tree species shown represents > 95% of animal relocations in each environment. www.nature.com/scientificreports/ these populations supports the need for future investigations into the population health consequences of RRV and other endemic arboviruses in koalas.
Habitat loss and fragmentation disrupts many important ecological processes including population dynamics and resource use [31][32][33] . It is further linked to increases in human disease risk for a variety of zoonotic diseases within urban environments through processes related to population isolation, reductions in host species richness, and increases in the abundance of urban adapted, highly competent host and vector species [34][35][36][37] . The current study reveals, for the first time, high RRV seroprevalence in koala populations residing in degraded urban coastal habitats. Resource related reductions in immunological fitness 22,23 , combined with often elevated animal densities in remnant coastal habitat patches 7 , may contribute to elevated RRV and other infectious disease prevalence in the surveyed populations (35% C. pecorum, 100% KoRV-A and 24% KoRV-B) 38 . The influence of habitat disturbance on RRV risk to koalas is further supported by lower seroprevalence rates (13%; n = 93) in koalas residing in the coastal Gippsland Lakes region of Victoria, Australia 15,16 , a region containing ca. 20,000 ha of protected natural habitat. However, we cannot attribute these differences to quality of habitat alone, as climate variability between regions will have differing impacts on RRV transmission intensity. Of note, although tree use by koalas in this study was similar to that reported elsewhere [39][40][41] , recent infrastructure developments in the study area have resulted in the loss of ca. 53 ha of koala habitat and the removal of at least 17,000 maturing trees 42,43 . This has clearly degraded the quality of this habitat overall and may influence future patterns of koala tree use and resource-related associations with arbovirus exposure in the areas surveyed.
The risk of disease exposure in wildlife often increases with animal age. Age-related increases in seropositivity has been demonstrated in koalas for KoRV 44 , as older animals are more likely to become infected due to having more contact with other koalas. The likelihood of vector-host contact also increases with age and reports on humans 17 and other marsupial species 24,29 demonstrate that antibody prevalence often increases with increasing age. Our data reflects the same pattern and is supported by the majority of seronegative koalas being < 2 years of age and the relatively young age of individuals that positively seroconverted during the study. It is difficult to give much weight to the latter finding, however, considering the low number of seroconversions observed. Our results further suggest that exposure is related to age-dependent patterns of movement, as both exposure and movement increased proportionally with increasing koala age. The home ranges observed in this study (males = 6.1 ha, females = 3.3 ha) are similar to those of koalas residing in another urban coastal habitat in Victoria, Australia (males = 9.1 ha, females = 4.3 ha) 45 , and general patterns of movement observed may reflect a combination of age and resource-limited 27,45,46 effects. Exposure ultimately appears to be dependent upon multiple demographic and habitat related factors, and the interactions between them. Thus, additional studies performed across varied habitats are needed to better understand how such factors influence exposure risk in koalas to RRV and other arboviruses. www.nature.com/scientificreports/ Although several mosquito species captured in each environment are known to transmit RRV, including Cx. annulirostris, Ae. vigilax and Cx. sitiens, historical evidence suggests Cx. annulirostris is the most likely vector for koalas. This mosquito has a tendency to seek hosts above ground level 47,48 , and commonly blood-feeds on marsupials, including tree-dwelling species like the brushtail possum 49,50 . The presence of Cx. annulirostris may expose koalas to other circulating arboviruses, as it is also a principal vector of BFV, Murray Valley encephalitis and West Nile Kunjin viruses 49,51 . Of these, BFV is the second most medically important endemic mosquitoborne disease in SEQ and serological investigations implicate marsupials as the largest and most diverse group of reservoirs 52 . Although previous reports suggest exposure to BFV in koalas is low (9% seroprevalence) 15 , the limited number of studies justifies additional investigations.
Lastly, infection in vertebrate hosts at the wildlife-human interface may be used to forecast the risk of spillover to the human population 53,54 . Although the spatial distribution of human RRV cases is extensive, there are elevated risks associated with suburban areas harbouring a greater proportion of wetlands and bushland and an increased presence of Cx. annulirostris and other freshwater mosquitoes 25,28,55 . These observations are corroborated by reports of RRV outbreaks being preceded by above-average rainfall events 56,57 that provide temporary freshwater larval habitats for Cx. annulirostris populations for extended periods of time. Increased monitoring of disease exposure in non-human vertebrate populations, such as koalas, may enhance our understanding of the environmental and ecological determinants of human exposure.

Conclusion
This survey represents one of the largest single-species marsupial seroprevalence surveys performed to date. The scale and scope of the survey provided key insights into the environmental and ecological determinants of RRV exposure in koalas residing in the urban coastal landscape. Substantial risks of exposure are associated with confinement of remnant coastal koala habitat to the edges of permanent wetland features that produce large numbers of competent mosquito vectors. Particular importance is attributed to the presence of freshwater habitat suitable to Cx. annulirostris mosquitoes. These results demonstrate that the careful study and sampling of wildlife populations can yield insights relevant to animal conservation and public health in rapidly changing urban landscapes.

Methods
Study area. Koalas were surveyed in coastal estuarine and lacustrine environments (Fig. 1a, Table 1) located in the Moreton Bay Region (MBR) of SEQ (27.2337° S, 153.0683° E and 27.2685° S, 152.9896° E, respectively), Australia. Koala habitat loss and fragmentation due to urban development in the two study areas has been severe over the last decade and remaining koala habitat is largely confined to the edges of swampy and/or flood-prone areas unsuitable for further development (Fig. 1b) 43 . Much of the remaining habitat in the more developed lacustrine study area is heavily fragmented and classified as medium quality, or suitability, for koalas 7 , whereas the less developed estuarine environment contains a greater proportion of high quality habitat. Both surveyed environments harbor mosquito faunas typical of the region's saltwater (Ae. vigilax) and freshwater (Cx. annulirostris) habitats because of their proximity to permanent water features. These species are significant vectors of RRV and readily blood-feed on humans and a variety of non-human vertebrates including marsupials 50,58,59 .
Endemic human circulation of RRV occurs in the region, with an average of 259 human cases/year notified between January 2012 and December 2016 (Data provided by the Queensland Department of Health, Communicable Diseases Branch, QIMRB Human Research Ethics Committee Approval no. P2238; Fig. S1). Human disease cases are reported in the region (and in broader SEQ) year-round, commonly peaking between February and May each year. This coincides with the late austral summer and autumn seasons, when seasonal increases in mosquito abundance occur.
Koala sampling, collaring and aging. We surveyed 223 koalas as part of a regional koala management program undertaken in response to a large, multi-year transport infrastructure development (Moreton Bay Rail Project; see Hanger et al. 43 ) that spanned both study areas. Koalas were captured using traps or the flag and pole technique and transported to the Endeavour Veterinary Ecology facilities in Toorbul, SEQ. Here, experienced veterinarians conducted a comprehensive clinical examination under anaesthesia to assess koala health which included the collection of a blood sample from the cephalic vein. During this examination, an estimated year of birth was determined for each koala based on the wear of the upper premolar and molar teeth 60 . For joeys, a date of birth was determined based on the developmental characteristics and size of the joey relative to a reference chart of known-age joeys. Initial age estimation enabled us to determine the approximate age of each koala during subsequent serum collection events. Following examination, healthy koalas were fitted with identification and telemetry devices to facilitate monitoring in the field. After recovery from anaesthesia, they were transported back to the study area for release at their point of capture (or as close as practicable based on safety or welfare concerns). Collaring and tracking of koalas began in March 2013 and ceased in January 2017.
Koala tracking. Koala tracking and monitoring followed the methods of Robbins et al. 61 Briefly, koalas weighing more than 3 kg were monitored using near-real-time biotelemetry devices (K-Tracker telemetry system, LX Group, Sydney, New South Wales) that reported global positioning system (GPS) locations and activity levels from each tagged koala via 12-hourly data uploads to an internet-based server. Koala position and activity levels were monitored remotely via the internet every 24 h but koalas were also field-tracked using very high frequency (VHF) radiotelemetry at least once per fortnight. Koalas weighing between 1 and 3 kg were not large enough to be fitted with the K-Tracker collars, so were field-tracked using VHF radiotelemetry several times a week. Koalas were monitored more frequently if there were health or welfare concerns, or if activity data www.nature.com/scientificreports/ reported by K-Tracker collars indicated low or zero activity. At each field monitoring event, koalas were examined with binoculars and various data recorded, including GPS location, tree DBH, tree species, tree height, the koala's height in the tree, external signs of health and the presence or absence of joeys. DBH was measured using a DBH tape and the maximum DBH measurement was recorded for each tree. Tree height and koala height was estimated by experienced field technicians. Koalas included in this study were monitored for a period ranging from 1 to 46 months (mean = 7.98; 95% CI = 7.54 to 8.54 months), which enabled determination of movement patterns and resource preferences.
Home range determination. The geo-referenced position data were used to calculate the home range sizes for each individual koala using the adehabitatHR package in R 62 . Prior to analysis, each position was mapped with QGIS software 63 to provide a visual indication of home ranges and ensure quality of data. Home range sizes were determined by minimum convex polygon (MCP 95%) estimation for each year the animal was monitored. Mean home range sizes were determined by averaging home range estimates across all surveyed years for each individual. In select cases, annual home ranges were estimated with lower confidence (cutoff MCP ≥ 60%) based on visual inspections of home range asymptote plots.
Serum collection and determination of seroprevalence. Koala sera were collected opportunistically between 2015 and 2017. A total of 529 blood samples were taken from 218 koalas. Some koalas were bled more than once (range = 1-6 serum samples/koala; mean = 2.3), creating the opportunity to investigate seroconversion in koalas. The presence of neutralising antibodies against RRV in each serum sample was tested in duplicate by preparing a monolayer of mammalian cell lines (Vero) in a 12-well tissue culture plate. Dilutions (1:10) of koala sera in RPMI-1640 (Sigma-Aldrich, USA) and the virus isolate RRV T48 64 were introduced to each well, sufficient to produce 50 to 60 plaques in the absence of antibody. Plates were incubated for two hours after which the virus-serum mixture was supplemented with 0.75% carboxymethylcellulose (CMC, Sigma-Aldrich, USA) overlay medium in ds-RPMI. Plates were incubated at 37 °C and 5% CO 2 for a further 2 days, fixed with crystal violet and examined for cytopathic effects (CPE), which can be determined by counting the number of plaques.
Neutralising antibodies bind to the virus, preventing CPE and the formation of plaques. Koala sera that neutralised ≥ 50% of plaques in these assays were considered anti-RRV seropositive 65 . This methodology has been deployed in a number of studies 66,67 , and has been tested at dilution rates of > 160 fold confirming strong and specific binding by the antibody 67 . We considered an animal sampled multiple times as seropositive if the animal returned a positive test result during any individual sampling event.
Mosquito community composition. The mosquito community in each location was characterized from February to March 2020; months associated with high mosquito and RRV activity in the region 68 . Mosquitoes were collected using CDC light traps (Pacific Biologics, Brisbane, Australia) baited with CO 2 from dry ice on 4 dates. Two traps were set within each environment either weekly or fortnightly depending on weather. Traps were operated from 18:00 to 06:00 during each surveillance event. Captured mosquitoes were transported back to the laboratory where they were stored at − 20 °C until identified to species using standard taxonomic keys.

Statistical analysis.
Fisher's exact test was used to compare differences in the number of positive sera results obtained in each surveyed environment. Two-way ANOVA was used to compare differences in home range sizes and the mean height of koalas in trees between seropositive and seronegative individuals. We used a generalized linear logistic model with the logit link function to determine the association between sero status and koala age. For this analysis, we grouped koalas into four age groups. Age groups included koalas < 2 years of age, those between 2 and 4 years of age, those between 4 and 6 years of age, and those > 6 years of age. Koala sex and location were included as co-variables in the model. www.nature.com/scientificreports/ Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/.