Factors that drive koala roadkill: an analysis across multiple scales in New South Wales, Australia

Keywords: coffs harbour, highways, koala, mitigation, monitoring, Phascolartos cinereus, road ecology, roadkill, threatened species, vehicle collisions.

Both in Australia and internationally, research into wildlife roadkill since the latter part of the 20th century has been perceptive and encouraging (Harris et al. 2008; Hobday and Minstrell 2008; Ramp and Roger 2008; Fahrig and Rytwinski 2009; Beckman et al. 2010; McCall et al. 2010; Taylor and Goldingay 2010; Jones et al. 2011, 2014; Roger et al. 2011, 2012; Jones 2012; Bond and Jones 2013; Lunney 2013; Soanes et al. 2017; Grueber et al. 2018; Santori et al. 2018; Ascensão et al. 2019; Fox et al. 2019), with Forman (2010) declaring that road ecology now reaches across much of the globe.

The koala (Phascolarctos cinereus) is listed as a threatened species under both New South Wales (NSW) and Commonwealth legislation, with the Commonwealth actively considering uplisting the koala to endangered (DAWE 2021a). Roadkill is recognised as a threat to koalas (DECC 2008; Commonwealth of Australia 2009, 2011; Shumway et al. 2015; OEH 2018; EES 2019; DAWE 2021b). A parliamentary inquiry (Parliament of NSW 2020, 4.1, p. 63) reported that, ‘Throughout the inquiry, many stakeholders told the committee that a key threat to local koala populations were roads, traffic and vehicle strikes.’ DPIE (2020a, 2020b, 2020c) has produced fact sheets on koala vehicle strike and how to avoid it. The Commonwealth of Australia (2011) has also recognised the importance of roadkill of koalas but the report was limited to federally-funded roads and did not include arterial or local roads (Shumway et al. 2015).

McAlpine et al. (2015) reviewed the contrasting regional trends, outlooks and policy challenges for conserving koalas, and concluded that there needs to be a concerted effort to take into account the unrelenting threat of road-related mortality of koalas. Gonzalez-Astudillo et al. (2017) examined the causes of the decline of koalas in south-east Queensland, and 26% of the 20 250 koala hospital admissions in their study comprised otherwise healthy koalas that had been hit by vehicles. Schlagloth et al. (2021) report a similar finding for Victoria, Charalambous and Narayan (2020) for NSW, and Lunney et al. (D. Lunney, I. Sonawane, H. Cope, E. Stalenberg and R. Haering, unpubl. data) have identified that it has been an ongoing issue for koalas in north-east NSW, as recorded by the local wildlife rehabilitation group Friends of the Koala. In fact, of their 596 koalas admitted to rehabilitation over 31 years from road injury, only 32% were released, the remainder died or were euthanised. Thus, the roadkill tally must include most of the koalas that were rescued.

There has been a number of insightful studies on predicting and mitigating koala roadkill. Dique et al. (2003), in a koala speed zone trial in SE Queensland, reported that 83% of the 1407 koalas hit by cars did not survive, and most were young, healthy males, but there was a higher survival rate of koalas on roads with lower speed limits. Ellis et al. (2016) found that daylight saving time can decrease the frequency of koala-vehicle collisions by shifting the timing of commuter traffic relative to darkness for this basically nocturnal mammal. Visintin et al. (2017) found a consistent pattern of vehicle collision risk in Victoria for six native mammal species, including the koala, with species occurrence, traffic volumes and traffic speed as predictors. Dexter et al. (2018) found that koalas were disproportionately more likely to cross a road if they had initially been found within a distance of 100 m of that road. Male koalas were more likely to cross roads than female koalas, as were koalas aged less than 5 years.

Wildlife roadkill is a problem on an international scale, and Australian policy writers and road planners can continue to benefit from knowledge from across the globe (Rajvanshi et al. 2001; Sherwood et al. 2002; Forman et al. 2003; Malo et al. 2004; McCall et al. 2010; Laurance et al. 2014; van der Ree et al. 2015; Dopp 2017; Srbek-Araujo et al. 2018; Collinson et al. 2019; Kreling et al. 2019). However, there remains the pressing need to manage specifically for koalas, meaning that purely generic approaches directed at any type of roadkill may not be adequate.

A challenge to understanding the impact of roads on koalas is the absence of a clear picture of the scales at which roadkill affects koala populations, or at which scale mitigation measures might be applied. Generically, the problem of koala roadkill has been identified at the state and national level (DECC 2008; Commonwealth of Australia 2011), with the most recent iteration being in the NSW Koala Strategy (OEH 2018). At a local scale, the Coffs Harbour Koala Plan of Management (Lunney et al. 1999) identified koala roadkill hotspots, or blackspots, and our 6-year study of an upgrade of a 1 km stretch of highway within Coffs Harbour local government area (LGA), at Lindsay’s cutting, examined the value of localised mitigation measures for koalas (Lunney et al. 2022). Thus conceptually, the scale ranges from an overview of the entire state to a local scale assessment of koala roadkill, with both policy and management needing to work at all scales when allocating resources to reduce the problem.

In this study, we looked at koala roadkill at two spatial scales – the entire state of NSW, and a local government area (Coffs Harbour) – for three road types (primary, arterial and local). These two scales complement our local-scale study of koala roadkill on a 1 km section of primary road in Coffs Harbour (Lunney et al. 2022). We also subdivided the state data into three major regions – coast, tableland and inland – to look at regional differences. Collectively, the aim was to elucidate the factors that drive koala roadkill across multiple scales in NSW in order to examine the conditions that predispose koalas to die on roads. By taking a simultaneous view of all scales, the relative importance of koala roadkill on different road types can be appreciated, such as the impact of major highways compared with arterial and local roads. This has particular relevance because all three levels of government have different responsibilities for roads and their impacts, for funding, mitigation strategies, management of different land tenures, and for fauna management.

We investigated the state-wide and regional-scale picture of koala roadkill in NSW using data on koalas from the Wildlife Atlas, within Bionet, managed by the NSW Department of Planning and Environment. This database contains spatially explicit records of species of animals within NSW, with date of record as well as observation type, including the option of dead on road. As with any large wildlife database comprising a range of sources, there exists the potential for biases, such as more records in the eastern third of NSW compared with western NSW. This has been graphically shown by Lunney et al. (2017), who explored the strengths and limitations of the Atlas. The database includes the records of koalas collected as part of a major 2006 state-wide community wildlife survey (Lunney et al. 2009, 2010), which had a focus on koalas. Participants in the survey were asked to mark on a map where they had seen both live and dead koalas. Our judgment was that, for the purposes this study, the Atlas data were robust. Records with dates between 1970 and 2014 were extracted from the Atlas.

To determine whether the extent of koala road deaths is related to the type of road, every koala record (including dead on road and others) within 100 m of the centreline of three road types was determined using ArcGIS (version 10.1). Road type, provided by Land and Property Information, NSW Department of Finance and Services, in five categories, was compressed into three categories: primary, i.e. highways and motorways; arterial, i.e. major roads within a particular locality, including sub-arterial and distributor roads; and local roads (Fig. 1). For each of the three road types we calculated the number of koalas recorded as dead on road and the total number of all koalas recorded (regardless of observation type) within 100 m from the centreline on either side of the road.

Fig. 1.  The distribution of the three road types within New South Wales. Examples are given of each. Photographs by Daniel Lunney.

In order to look at the extent of koala roadkill at a regional scale, records were assigned to one of three geographic regions derived from the seven koala management areas (KMAs) (DECC 2008) (Fig. 2). The coastal region consisted of KMA1 (North Coast), KMA2 (Central Coast) and KMA3 (South Coast). The tablelands region consisted of KMA4 (Northern Tablelands) and KMA5 (Central and Southern Tablelands). The western (inland) region consisted of KMA6 (Western Slopes and Plains) and KMA7 (Far West and South West).

Fig. 2.  The three regions within New South Wales used in the analysis of road type.

We used a logistic regression to test for patterns of how road type and location may influence the likelihood of a koala dying on the road. The model included koalas observed as being either alive or dead on road as the dependent binary variable. Two categorical explanatory variables were included: road type and geographic location. A full model was run including interaction between the two explanatory variables using the LOGISTIC procedure in SAS software (SAS Version 9.4, SAS Inc, Cary USA).

In addition to the Atlas of NSW Wildlife data used across the state, the analysis of koala roadkill at the scale of local government area (Coffs Harbour LGA) included data from a major community survey completed in 1990 that formed part of the preparation of a Comprehensive Koala Plan of Management for Coffs Harbour City Council (Lunney et al. 1999, 2000). The survey asked participants where they had seen koalas and if the koala was dead on the road. Using the same three road types as for the state-wide comparison, the number of koala records within 100 m of the road centreline within the Coffs Harbour LGA was determined.

Coffs Harbour LGA also provided the opportunity to look at the relationship of koala roadkill to koala habitat, a variable not available at the state scale. As part of the Comprehensive Koala Plan of Management, a map of koala habitat was prepared ranking koala habitat quality as primary, secondary and tertiary based on a combination of vegetation type, geology and koala vegetation use. Using ArcGIS (version 10.1), each koala record within 100 m of a road centreline was assigned to a koala habitat category (‘habitat type’) based on either the habitat polygon that it sat within or else the nearest koala habitat polygon based on a straight-line distance to the polygon edge. We were able to categorise each koala record, including the roadkill, based on the road type and the koala habitat type.

We used a logistic regression to test for patterns of how road type and koala habitat type may influence the likelihood of a koala dying on the road. The model included koalas observed as either alive or dead as the dependent binary variable and two categorical explanatory variables: road type and koala habitat type. A full model was run including interaction between the two explanatory variables using the LOGISTIC procedure in SAS software (SAS Version 9.4, SAS Inc, Cary USA).

At a local scale, the characteristics of traffic flow can influence, and possibly explain, koala road deaths. One characteristic of traffic that will influence the rate of roadkill is the number of vehicles on the road and, in particular, the number of vehicles passing a particular location (Rhodes et al. 2014). To determine the characteristics of traffic that may affect koalas within Coffs Harbour LGA, we looked at the average time between vehicles passing a set point on the Pacific Highway and the monthly Annual Average Daily Traffic counts (AADT). Data were based on a 2007 AADT count of 21 537 vehicles on the Pacific Highway at a station south of Sawtell Road (data provided by Roads and Maritime Services (RMS) Northern Region) and the hourly breakdown of vehicles from the Woolgoolga to Ballina Pacific Highway Upgrade Environmental Impact Statement (RMS 2012). We further looked at the weekly average daily traffic volume as a percentage of AADT based on traffic counts presented in RMS (2012).

Across NSW, 12 970 koala records (1970–2014) occur within 100 m of the centreline of a road. Of these records, 542 (4%) were recorded as being dead on a road (Table 1). While somewhat similar numbers of koalas were recorded as dead on road across the three road type categories, there is a major difference in the percentage of koalas recorded as dead compared with koalas seen within 100 m of a road across the three road type categories. Over a quarter of all koalas observed on or adjacent to primary roads were dead on the road, compared to only 6% on arterial roads and 2% on local roads (Table 1). The overall logistic regression model was significant (likelihood ratio chi-squared = 501.5, d.f. 8, P < 0.0001). There was a significant difference in the likelihood of a koala being dead related to road type (χ² = 430.16, d.f. = 2, P < 0.0001), no significant difference related to geographic region (χ² = 2.31, d.f. = 2, P > 0.05) and a significant interaction between road type and geographic region (χ² = 15.00, d.f. = 4, P = 0.005). Comparing the maximum likelihood estimates among the three road types showed that the levels of koala death were all significantly different from the overall average and contributed to the overall model: local (MLE = −2.82, χ² = 423.53, P < 0.0001); arterial (MLE = −1.60, χ² = 143.46, P < 0.0001); primary (MLE = 4.42, χ² = 318.27, P < 0.0001). Although an individual koala is more likely to die if adjacent to a primary road than the other two road types, the vastly different total length of roads across the state in the three categories (local roads 122 755 km; arterial roads 79 706 km; primary roads 12 972 km) accounts for the similarity in the actual number of koalas seen dead in the three categories, despite the differences in the percentage dead on road. Also, a pronounced result from the regional analyses is that koala records are predominantly coastal, and so are the koala roadkill records.

Table 1.  The number of koalas recorded (total/roadkill (percentage dead)) within 100 m of three road types, across three geographic regions and for NSW as a whole.

The regional percentage of koalas dead on road shows little variation from a pattern of the highest mortality being adjacent to primary roads (Table 1). Analysis of the Maximum Likelihood Estimates for the nine interaction combinations showed that only the interaction of local roads in the west was significant (MLE = 1.01, χ² = 8.20, P < 0.005). The remaining eight interaction terms did not contribute to the overall model.

There was a total of 2885 koala records from within the Coffs Harbour LGA within 100 m of a road (Table 2). Of these records, 477 (17%) were recorded as being dead on the road. Although the road mortality here is higher than the average for the state, the pattern shown at the LGA scale is similar to that seen across the state of NSW: a koala has a higher likelihood of being killed on a primary road, followed by an arterial road, and the lowest likelihood of being killed is on a local road (Table 2). However, the high number of koalas observed next to local roads means that the actual number recorded dead is higher than on primary or arterial roads due to the much greater length of local roads (Table 2). Habitat type adjacent to a road appears to have little influence on the likelihood of a koala being killed on the road (Table 2).

Table 2.  The number of koalas recorded (total/roadkill (percentage dead)) within 100 m of three road types, for Coffs Harbour LGA, split across three koala habitat types.

Overall the logistic regression model was significant (Likelihood ratio χ² = 274.01, d.f. = 8, P < 0.0001). Road type was a significant contributor to the model (joint test χ² = 189.58, d.f. = 2, P < 0.0001), whereas koala habitat type (joint test χ² = 3.36, d.f. = 2, P > 0.05) and the interaction of road type and koala habitat type (joint test χ² = 8.81, d.f. = 4, P > 0.05) were both not significant. Analysis of the maximum likelihood estimates showed that all three road types made significant contributions to the model: local (MLE = −1.90, χ² = 161.08, P < 0.0001); arterial (MLE = −0.46, χ² = 7.95, P < 0.005); primary (MLE = 4.42, χ² = 72.42, P < 0.0001).

The traffic count data illuminate the problem for a koala attempting to cross a primary road. Gaps between vehicles on the Pacific Highway, when wildlife may be able to cross without being hit, average between 2 and 22 seconds (Fig. 3). However, road traffic tends to clump into ‘platoons’, rather than flow evenly, so that the period from about 10 pm to 5 am likely contains some extended gaps, enabling a safer crossing for any animals able to utilise them. Throughout the year traffic densities fluctuate with school holidays and season (Fig. 4), with a major peak in the middle of the period (September/October) when koalas are most mobile.

Fig. 3.  Average time between vehicles throughout a 24 h period. Figures are based on AADT of 21 537 vehicles on the Pacific Highway at station 4144 south of Sawtell Road, Coffs Harbour LGA, and the hourly breakdown of vehicles from Woolgoolga to Ballina EIS (RMS 2012).

Fig. 4.  Weekly average daily traffic expressed as a percentage of AADT. Data are modified from the Woolgoolga to Ballina EIS (RMS 2012) and represent the average of four monitoring stations on the Pacific Highway north of Coffs Harbour. Weeks 34–45 (September/October) represent a recognised time for dispersal or increased movement for koalas.

We examined koala roadkill at two scales: the state-wide picture (also three NSW divisions of coast, tablelands and inland), and an LGA where we also included habitat type at three levels (primary, secondary and tertiary). At each scale koala roadkill was examined by road type, with three levels (primary, arterial and local). Scale has implications for koala threat management because knowing whether problems are widespread or localised can determine not only how the problems may be addressed, but also where responsibility lies for policy, funding and threat mitigation. This study has provided the first distinct picture of where koala roadkill occurs across NSW.

From our assessment, a number of points become clear. Wherever koala populations and roads occur together there is roadkill. Where more koalas occur in the landscape, there is more roadkill. This is seen spatially with higher numbers of koalas dead on road in the north coast of NSW, a geographic region recognised as having high numbers of koalas, and at the smaller spatial scale of Coffs Harbour LGA, where higher numbers of dead on road are seen in primary koala habitat, which supports higher numbers of koalas, although we note that we found no significant difference in the proportion of the koala population adjacent to primary, secondary or tertiary habitat being killed on the road.

Analysis of the probability of a koala being killed shows that road type is important, with primary roads having the highest likelihood of koalas dying. This is seen at the two spatial scales investigated, and in each of the three regions. This is likely due to the greater width of the primary roads and increased volume and speed of the traffic, and, as shown in Coffs Harbour LGA (Fig. 3), the time available for fauna to effect a safe crossing of a busy primary road is measured in seconds.

Primary roads provide a good opportunity for mitigation. They are managed by the one State government department in NSW (RMS) and mitigation measures, such as gap-free fauna fencing, are generally more accepted along such roads because of the level of disturbance already made by the road. It would be far harder to fence arterial or local roads and the value of this would be questionable given the lower rate of koala deaths per kilometre, the need to balance benefits against cost, and the overall barrier effect that fences produce. As an example, Goldingay and Dobner (2014) address the limitations for mitigation at the local scale in an urban area. Similarly, for arterial roads, where local government is required to pay the costs, more gaps for residences decrease the value of fences because koalas walk through the gaps (Lunney et al. 2022).

There is a need to correctly assess the relative importance of koala roadkill compared with other factors. Roadkill is visually confronting, but predation, particularly from dogs, is rarely witnessed. In our radio-tracking study of a large population of koalas after a wildfire at Port Stephens, a coastal LGA between Sydney and Coffs Harbour, no koalas were killed by cars, despite being adjacent to arterial roads, with the surprise finding that dogs were the primary killer (Lunney et al. 2007). The radio-tracking revealed that koalas that were killed by dogs were hidden under bushes behind locked gates on mining and water catchment land. The resulting action was to manage predatory dogs in the post-fire landscape, as well as manage traffic on the surrounding roads, where dead koalas were conspicuous and a big road notice showed the incremental koala road toll. Similarly, a detailed population study in south-east Queensland found that, apart from natural mortality, disease was the biggest killer, with roads being the second biggest (Rhodes et al. 2011). A key point here is that, without a detailed population study, an easy default assumption is that roadkill takes a greater proportion, compared with other threats, of the koala population than is actually the case if the other threats are hidden. This points to the need for population studies that consider all threats, not only studies of roadkill. Somewhat pessimistically, Preece (2007), who modelled threats to koalas in SE Queensland, concluded that urban koala populations will not be able to withstand the high rates of anthropogenic mortality, such as roadkill, in addition to natural mortality with the result being localised extinctions.

The matter of long-term impacts has received considered attention from a number of angles. Eberhardt et al. (2013) pointed out that road mortality hotspots are typically located where wildlife habitats are near to, or straddle, the road. They identify such locations as the most appropriate sites for mitigation where the populations have not already been reduced by road mortality. Polak et al. (2014) examined the vulnerable koala population living in habitat patches separated by roads on the Koala Coast in south-east Queensland, and applied decision science to find the best solutions for a range of possible costs of mitigation. They mathematically formulated the problem of prioritising different road mitigation actions, and in all of the model runs (no action at all, and fences with or without crossings, on different road segments) the ‘Do Nothing’ approach of no mitigation on any road segment was the worst for population numbers. To the authors’ surprise, full mitigation (Fence + Wildlife crossing, i.e. the most costly solution) on all road segments did not maximise mean population numbers. Their analysis also showed that achieving very good outcomes for koalas will be expensive, and that reducing investment will result in a much smaller koala population.

For koalas, this is a pertinent issue because the koala has been recognised as a threatened species under NSW legislation since 1992, with roadkill being recognised as a factor in its decline. The Senate committee (Commonwealth of Australia 2011) also recommended that where the Australian Government provides funding for roads or other infrastructure in or adjacent to koala habitat, it be contingent on the provision of adequate koala protections. Another recommendation was that the Australian Government work with the states to develop new national guidelines to ensure that all new roads and upgrades in or adjacent to koala habitat are koala-friendly. We urge that research both precede, and follow up, implementing such expensive measures to combat koala roadkill. Goldingay et al. (2018) found that no koalas were detected in the escape ramps on the Oxley Highway, which had been installed for koalas, and that only nine were detected in the underpasses.

If ever we are to minimise the impacts of roads, particularly highways, on koalas, then we need to respond to the practical requirements of transport agencies, which in NSW is the Roads and Maritime Services (RMS). The consultation by RMS with one of us (CM) over the Bonville deviation near Coffs Harbour was very responsive to environmental factors which arose during initial planning. Ecologists are now being engaged from the early stages of route selection by RMS. We need to avoid the danger that ecologists do not understand road development processes as much as the other way around. Jones et al. (2015, p. 395) expand this point diplomatically when they state that experience from committees world-wide on road design and the public has identified some common themes, with the most important being the extent to which one side is ignorant of the key concepts of the other. We also have formed the view that roadkill needs to be considered much earlier in the planning process, with Rhodes et al. (2014) providing a numerically worked example, also from coastal NSW. Finally, we need to define the issues more sharply, which is part of the point of this paper, and we need to overcome what is a blind spot in public perception of the issue (Lunney 2013).

The recognition of the value and importance of managing wildlife, especially threatened species and natural habitats, has mostly occurred since the mid-20th century, and New South Wales is typical (Lunney et al. 2017). By the time any environmental legislation was in place, NSW was criss-crossed with roads and filled with high speed motor vehicles, but there were very few provisions for understanding or managing wildlife roadkill. As is now all too apparent, the speed of economic growth and population increase has far exceeded our capacity to contain the environmental damage that follows in their wake. The focus on mitigating wildlife roadkill presents a clear-cut case that the conservation of the native fauna is an afterthought. The question then is ‘have we left it too late to prevent further roadkill?’.

Some clear recommendations can be made from this study. Specifically, our results indicate that different road types should be treated differently as they affect koala mortality in different ways. Highways are likely to be a good option for targeted mitigation measures as they are likely to provide the biggest benefit per dollar spent. Local and arterial roads, on the other hand, are a more problematic, ubiquitous issue where targeted retrofitting may be less effective overall due to the diffuse nature of the threat. Here education and regulation might be more effective.

We have two primary tools to approach mitigating the adverse impact of roads on native fauna and natural habitats – education, and research in an adaptive management framework, not only of physical structures, such as fences, but policy adjustments such as valuing daylight saving so that peak periods of traffic occur before dark (Ellis et al. 2016).

The case for education is compelling, and it will include relying on a broad-based understanding of, and insights into, issues that go well beyond the normal scope of the work of ecologists and engineers (Stratford et al. 2000; Lunney 2013). Recent studies have shown the potential of crowdsourced geographic citizen science to contribute to biodiversity research on koalas (Predavec et al. 2016; Brown et al. 2018, 2019a, 2019b), to the extent that local, State and Commonwealth governments can gain a better view of the conservation issues and management options for koalas that cross roads. This includes a sense of the willingness of citizens to exercise restraint in activities that endanger koalas and to support government initiatives to manage the issues.

In asking, what is policy-oriented environmental research worth, Pannell et al. (2018) recognise that environmental and conservation scientists are increasingly being asked to justify their work in terms of benefits. Koalas are no exception. The authors conclude that improved measurement of benefits from environmental research would assist environmental scientists. We add that while the cost of research is evident in any policy or management business case, the cost of not conducting research on roadkill and the broader issue of the long-term survival of the local koala populations remains uncosted. We include monitoring as being within the framework of research, because without a rigorous design, and one that draws on the rapidly changing international network of researchers in this subject, we shall never know whether a reduction in roadkill numbers is a measure of the success of a mitigation action or whether, more simply, there are fewer and fewer koalas. To that end, NSW has now adopted a koala monitoring framework (DPIE 2021).

The conclusion can be drawn that roadkill is an ever-present threat to koalas across their entire range in NSW. We can expect that the use of vehicles will steadily rise with the increasing human population size, and that koala habitat will continue to be lost, or fragmented, with more and more developments and thus more roads and road upgrades. The consequence is that koala roadkill will remove an ever-increasing proportion of what is recognised as a decreasing koala population in NSW (Adams-Hosking et al. 2016). Roads are fixtures on the landscape, and there is as yet little public perception of how to understand and tackle this vexatious matter of unwanted animal death (Lunney 2013). Our work provides a framework for planners, government decision-makers and other managers to take action on the issue of koala road death in their locality at a scale appropriate to the scale of the problem, and with an understanding of why koala black spots are where they are. Our answer to the terrible question, ‘is it too difficult to prevent further roadkill?’ is yes, it is too difficult to prevent roadkill, but it is not too difficult to mitigate roadkill, and we now have a better grasp of where to target those efforts.

The data that support this study are available in NSW BioNet | NSW Environment, Energy and Science, except for 1990 data from Coffs Harbour, collected before Bionet was created, which is displayed in the Coffs Harbour Comprehensive Koala Plan of Management (Lunney et al. 1999).

The authors declare no conflicts of interest.

This research did not receive any specific funding.