Herbivore management for biodiversity conservation: A case study of kangaroos in the Australian Capital Territory (ACT)

Populations of macropods are higher than estimated preEuropean densities in many parts of Australia. To achieve appropriate densities of macropods in the Australian Capital Territory’s nature reserves, multi-tenure kangaroo management units are used to tailor management of kangaroos and total grazing pressure to achieve conservation objectives. An adaptive management framework is recommended that monitors the state of the groundlayer vegetation and alters the cull accordingly. This case study may provide insights for kangaroo management in other temperate areas of Australia.


Introduction
L arge mammalian herbivores have a long co-evolutionary relationship with the plants that share their habitat Adaptations in animal anatomy, physiology and behaviour are related to the plants upon which they feed (e.g. Gordon & Prins, 2019), whilst plants have evolved structural and chemical defences against herbivory and may also respond to herbivore-driven nutrient cycling (e.g. deposition of faeces and urine; Gordon & Prins, 2019). Over ecological time, large herbivores are a significant factor influencing plant population dynamics, vegetation community composition and structure, the faunal diversity associated with these vegetation communities, as well as the intensity and frequency of fires (Gordon & Prins, 2019). As such, the individual and combined influences of herbivore species' feeding behaviour within an ecosystem, as well as their relative population densities, can affect ecosystem processes and services, and the biodiversity that relies on vegetation for food and cover (Owen-Smith 2002;Augustine et al. 2003;Fynn et al. 2016).
Over recent millennia, humans have extirpated or significantly reduced the impacts of carnivores in many ecosystems (Ritchie et al. 2012;Ripple et al. 2014). The resulting release of prey species' populations from these top-down effects has had profound consequences for the regulation of herbivore population densities (Prowse et al. 2014). Reductions in perceived predation risk have also increased the extent of foraging areas readily utilised by herbivores (Ripple & Beschta, 2004). Combined, these direct and indirect changes have skewed ecological plant-herbivore processes worldwide from their original evolutionarily 'coadapted' state (Augustine & McNaughton, 1998;Ripple & Beschta, 2004).
In Australia, the native Thylacine (Thylacinus cynocephalus) was eliminated from the mainland about 5,000 years ago. Livestock grazing was introduced over extensive areas of the continent following settlement by Europeans~250 years ago, after which the Dingo (Canis dingo) was eliminated from much of the mainland, and Indigenous hunting was suppressed (Prowse et al. 2014;Johnson 2015). This combination of actions is believed to have reduced predation/ hunting pressure on the larger marsupial herbivores, and in combination with land clearing and the promotion of grass growth, has led to significant increases in the distribution and density of many large macropod species, notably the Eastern Grey Kangaroo (Macropus giganteus), Western Grey Kangaroo (M. fuliginosus), Red Kangaroo (Osphranter rufus) and a number of the larger wallaby species (Wilson & Edwards, 2019).
These changes in macropod distribution and abundance have led to significant impacts on agricultural pasture, and native vegetation communities and their associated fauna ( Fig. 1; Howland et al. 2014;Prowse et al. 2014). Densities of a number of species of macropods remain above pre-European densities in many areas, despite most Australian states (except Tasmania) having a commercial harvest of kangaroos, all states and territories allowing licensed shooters to cull kangaroos on their properties to reduce total grazing pressure, and some jurisdictions allowing conservation culls (Wilson & Edwards, 2019).
The Australian Capital Territory (ACT) Government undertakes an annual 'conservation cull' of kangaroos within the urban reserves comprising Canberra Nature Park. The objective of this programme is to achieve densities of kangaroos that provide a grazing regime favourable for the conservation of the plants and animals that occur in the ground-layer vegetation. Here, we will review the history of the kangaroo management programme in conservation reserves of the ACT, describe the current management policy enacted by the ACT Government, and provide recommendations for future management of kangaroos and total grazing pressure to achieve conservation objectives in the ACT. This follows an Expert Workshop (attended by the authors of this paper, and other ACT Government staff) organised by the ACT Government's Conservation Research unit and held in Canberra, ACT, in August 2018, to discuss and review recent research relating to the management of kangaroo grazing in the ACTs nature reserves.
Kangaroo Management in the Australian Capital Territory (ACT) The context: Canberra Nature Park Canberra Nature Park is a network of 37 nature reserves dispersed throughout the city of Canberra, ACT. Individual nature reserves range in size from 47 to 2017 ha, with a total combined area of around 11,000 ha. Individual reserves within this network, whilst connected at the landscape scale, are fragmented at the finer scale by lakes and rivers, as well as anthropomorphic barriers such as suburbs, fences and major roads. Green corridors, and other natural or semi-natural spaces such as golf courses, urban parks and farmland, also contribute to habitat connectivity for some species throughout the ACT's urban matrix. The management goals for Canberra Nature Park are to conserve the natural environment, and to provide for public use for recreation, education and research (ACT Government 2019a).
The grasslands, woodlands and forests of Canberra Nature Park are of conservation significance, both regionally and nationally, due to the size and connectivity of these patches throughout the landscape. Over one third of the reserve system supports critically endangered Yellow Box -Blakeley's Red Gum Grassy Woodland, and a further 10 per cent supports critically endangered Natural Temperate Grassland or habitat for eighteen threatened grassland animals or plants (ACT Government 2019a).  Fig 3). The other three extant macropod species (Black Wallaby, Wallabia bicolor; Rednecked Wallaby, Notamacropus rufogriseus; and Common Wallaroo, Osphranter robustus) have increased substantially in distribution and abundance in the last four decades (D. Fletcher pers obs. since 1975),but none has been subject to licensed culling. Kangaroos are responsible for the majority of herbivory in Canberra Nature Park (ACT Government 2019b), and as such they occupy a central place in the ecology of these ecosystems (ACT Government 2010). They are also treasured by Canberrans as one of the best-known local native animal species (Micromex Research 2008). Kangaroos had been almost extirpated from the area where Canberra was to be built but began to increase from the early 1950s (ACT Government 2010). In 1975, only two reserves or future reserves of Canberra Nature Park appeared to contain kangaroos, but in 2020 kangaroos are obvious in all 37, and are also present in some of the larger suburban parks. Areas of kangaroo habitat within the urban matrix of the ACT are linked by green corridors; consequently, additional urban parks are continuing to be colonised by the species (Fletcher pers obs. since 1975). At the same time, kangaroos are being displaced from areas of former habitat by urban expansion. It is apparent that kangaroos are now more abundant in the Australian Capital Territory than at any time since 1900 but pre-European abundance is unknown because explorer accounts are ambiguous or equivocal.
Grazing pressure in the ACT reser ves Heavy grazing pressure, associated with high-density kangaroo populations, can cause loss of herbaceous biomass (McIntyre et al. 2010), and local loss of threatened plant and animal species (Coulson 2001(Coulson , 2006Howland et al. 2014Howland et al. , 2016a. Specifically, selective grazing by kangaroos in the ACT has been demonstrated to modify the habitats of grassland plants, birds, reptiles and invertebrates (Barton et al. 2011;Howland et al. 2014Howland et al. , 2016bMcIntyre et al. 2018). For example, by maintaining a short grass structure, kangaroos reduce the density of, or exclude, certain grassland birds (Neave & Tanton, 1989), and grazing-sensitive plants (McIntyre et al. 2018). Grazing lawns can also have high floristic diversity and may, concurrently, benefit floristic diversity at some scales (Vivian & Godfree, 2014;Snape et al. 2018). Efforts in Canberra Nature Park to reverse the effects of historic heavy livestock grazing, and ongoing kangaroo grazing, show only modest recovery of grazing-sensitive species, and document a perennial flora that is resistant to change, even after four years of favourable rainfall and reduced grazing pressure (McIntyre et al. 2010(McIntyre et al. , 2017. It should be noted that localised species losses would likely be permanent within the fragmented grassy ecosystems of Canberra Nature Park; therefore, a consistent landscape scale approach to the management of kangaroo populations is essential for the conservation of endangered ecological communities in this context.
The policy: Program histor y and development of a kangaroo management plan The history of the programme is summarised in Fig. 2. The conservation cull of kangaroos within Canberra Nature Park aims to maintain densities of kangaroos which allow for conservation of the grassy ecological community and habitat for all grassland plant and animal species. This objective, along with significant reference to the scientific literature which underpinned kangaroo management policies, was first described Box 1. The Herbivore: Eastern Grey Kangaroo The Eastern Grey Kangaroo (Macropus giganteus) exhibits a highly seasonal breeding pattern in the ACT, with 84 per cent of young born within three months of the third week of November ( Fig. 3; Fletcher 2006). Final pouch emergence generally occurs in spring around 9.6 months later. Fecundity is as high as 83 per cent even in dry years, and typically 71 per cent of adults are females although the birth ratio is even (Fletcher 2006). The principal limit to kangaroo abundance for unmanaged populations is food availability, which acts mainly through juvenile mortality due to starvation or related processes (Fletcher 2006(Fletcher , 2007Portas & Snape, 2018). Kangaroos are also preyed upon, and scavenged by, Dingoes (Canis dingo), Wedge-tailed Eagle (Aquila audax), European Red Fox (Vulpes vulpes), and a variety of carrionassociated invertebrate species (Robertshaw & Harden, 1989;Barton et al. 2013). They are also frequently killed through collisions with motor vehicles, especially during late winter and early spring, with an estimated 13,985 kangaroo-vehicle collisions occurring on ACT roads in 2015, based on the results of Micromex Research (2015), and the number of licensed drivers in the ACT. High collision rates also apply in a large adjoining region of NSW (Ramp & Roger 2008). The effects of fences on predation and vehicle-related mortalities appears to influence population growth rates of kangaroos (Table 1) in the ACT Kangaroo Management Plan (ACT Government 2010). The scale of the ACT conservation culling programme was kept deliberately small in its first year (494 kangaroos in five reserves) in recognition of the rigorous standards required for shooting in urban areas, and that this was a novel activity for those responsible (Fletcher pers obs., ACT Government 2017). A kangaroo shooting season from March to July (the austral winter) was also enforced to reduce the rate at which shooters encounter female kangaroos with a young-atfoot or large pouch young that would be deprived of maternal milk and face starvation if its mother was culled (Fig. 3;Fletcher 2007). However, substantial opposition to culling remained in some sectors of the Australian community (Ballard 2008;Micromex 2008Micromex , 2012Micromex , 2015ACAT 2009ACAT , 2014Ben-Ami & Mjadwesch, 2017), although kangaroo culling was supported by many if it reduces kangaroo deaths due to starvation or to meet conservation objectives (Sharp 2012).
The culling programme encountered significant resistance initially, within and outside ACT Government, and from 'The Canberra Times', the primary local newspaper. An important factor in overcoming this challenge was the formation of the 'Limestone Group' of Canberra-based scientists and conservation group representatives who provided supportive commentary independent of government ( Fig. 2).
Court injunctions delayed the programme in three of its first six years until legal appeals were heard and lost. The appeals ostensibly targeted the ecological evidence underpinning the culling programme (e.g. validity of counting methods, or evidence that the abundance of one native species might impact that of another (ACAT 2009(ACAT , 2014), although the three individuals nominally responsible for the appeals have stated their underlying motivation to be opposition to the killing of sentient individual animals. When the cull re-commenced, protestors cut fences, glued up padlocks on scores of gates, and on one occasion slashed tyres and smashed windows at two ACT Government depots, actions publicly defended by two of the most prominent protestors ( Fig. 4; Canberra Times 2014a,b). Community support for the protest appeared to decline suddenly after this vandalism, as it had done previously in response to vandalism of a fox-proof fence protecting eastern bettong (Bettongia gaimardi) recently re-introduced to the mainland (The RiotACT 2012), and the detention of two university researchers (Francis 2012) who were mistaken for kangaroo shooting personnel. Meanwhile, government efforts to explain the conservation cull, mainly on web pages and in the news media, were increased to include a wide range of media. This included some actions regarded as unusually bold for government media units and ministerial offices; for example, a novelist was given access to every part of the shooting operation, resulting in a more informed and objective treatment in the resulting Canberra-based novel (Viggers 2015). Also, a documentary film crew was allowed to accompany government scientists for one year of their daily activities, whilst also recording the thoughts and activities of kangaroo activists (360 Degree Films 2011). Protest declined and after 2015, protests at cull sites became insignificant. By 2017, 14 of the 37 conservation reserves in Canberra (and one ACT Government managed reserve in NSW) had been involved in the conservation culling programme at least once, and the operational capacity, and geographical extent, of the programme had increased significantly (Fig. 2).

Resulting policy
In recognition of the ongoing requirement for kangaroo management in the ACT, the Eastern Grey Kangaroo was declared a 'Controlled Native Species' under the ACT Nature Conservation Act 2014 in 2017, negating the need for the ACT Government to obtain a licence to undertake annual kangaroo management programmes. In the same year, an Eastern Grey Kangaroo: Controlled Native Species Management Plan (ACT Government 2017) was published as a statutory plan under the ACT Nature Conservation Act 2014, reaffirming the policy objectives identified in the ACT Kangaroo Management Plan, namely to: Maintain populations of kangaroos as a significant part of the fauna of the 'bush capital' and a component of the grassy ecosystems of the Territory; Manage and minimise the environmental, economic and social impacts of those kangaroo populations on other biota, grassy ecosystems and primary production.
Since the initiation of the conservation culling programme, when 59 per cent of the public were 'supportive' or 'very supportive' of conservation culling, public acceptance increased (70 per cent in 2011) then remained high (76 per cent in 2015) according to repeated community attitude surveys (Micromex Research 2012.
The approach: kangaroo management units and an annual culling program The ACT Kangaroo Management Plan gave significant consideration to the spatial and temporal scale of the conservation culling programme (ACT Government 2010). A major outcome was the identification of individual Kangaroo Management Units (or KMUs), which were defined as areas occupied by a discrete kangaroo population bounded by barriers to kangaroo movement (determined through GPS tracking studies) such as highspeed roads (≥ 80 km/h limit), significant water bodies or suburbs (Tolfts 2019; ACT Government unpublished data; Fig. 5). As many nature reserves are only 2-3 km 2 in area and are bounded, on at least one side, by other open spaces, such as rural properties, golf courses or horse agistment paddocks, some kangaroo home ranges straddle the boundary between the reserve and adjoining land.
Consequently, the multi-tenure KMU approach was adopted such that temporary movements of individual kangaroos into and out of a nature reserve, especially in relation to disturbance such as counting or culling activities (Pulsford & Snape 2019), would not impact on the capacity of land managers to undertake effective monitoring and management of the broader isolated kangaroo population. Overall, culling targets across all land tenures within the KMU were subdivided between landholders (e.g. ACT Parks and Conservation Service, rural leases and horse agistment operators). For legal reasons, licences to cull kangaroos were issued to landholders individually but there was mutual value in cooperative management given the shared nature of the kangaroo population within each KMU.
In many cases, new KMUs that were incorporated into the conservation culling programme required large initial reductions in kangaroo density to reach conservation targets, sometimes taking multiple years to achieve the target (Fig. 6). After kangaroo populations were reduced, they could usually be maintained at the reduced size thereafter by small annual culls. Modelling of this approach, based on local population demographic data, indicated the strategy of small, frequent culls to be the most effective means of maintaining conservation target densities of kangaroos, both operationally, and to limit the number of animals needing to be culled (ACT Government 2010). As this 'maintenance' mode of kangaroo management was achieved across the reserves already in the programme, further sites could be added without proportional changes in the total resources required for the culling programme each year. This gradual approach ensured management objectives were readily achieved at each site (Fig. 6), and enabled progressive expansion of the conservation culling programme to other priority sites over time.
The science: calculation of the number to cull Target densities of kangaroos, to achieve conservation objectives, were determined based on an ecological model developed specifically for the Eastern Grey Kangaroo in the temperate environment (Fletcher 2006). The model adopted the same structure as Caughley's (1987) interactive model based on red kangaroos in arid rangelands, which has been used for decades to guide the national commercial harvest of kangaroos and  Common Wallaroo (Pople et al. 2010(Pople et al. , 2018. As such, it incorporated local estimates of the pasture response (pasture growth rate as a function of weather and herbage mass), kangaroo numerical response (population growth in relation to pasture availability), and kangaroo functional response (food consumption rate as a function of pasture availability). Unlike Caughley's original model, the eastern grey kangaroo version considered the availability of green (living) herbage mass (rather that total herbage mass) and recognised the seasonality of temperate grasslands by incorporating monthly temperature and rainfall values from local weather stations (Fletcher 2006). Pasture and kangaroo numerical response rates were estimated using weather stations, exclosure cages and kangaroo counts at three sites in and near the ACT, and the functional response was estimated using three purposebuilt 'graze-down' pens in each of two pasture types (Fletcher 2006).
Model output (Table 2) simulated five management options: no reduction in kangaroo density; a commercial culling type approach (~20 per cent); winter culling to a mean density of approximately 1.0 kangaroo/ ha; winter culling to approximately 1.6 kangaroo/ha; and culling to a density considered consistent with achieving economic gains on rural land. Fig. 7 shows the effect of the five management options on kangaroo and vegetation density. The culling regimes tested in the model showed that reductions greater than those set by a commercial quota (i.e. approximately 20 per cent of population, Pople et al. 2018) were required to achieve vegetation responses that enabled the development of tussocky grass structures thought to be associated with conservation of threatened vertebrate species (e.g. Striped Legless Lizard, Delma impar, Howland et al. 2016b). Culling to a mean density of approximately 1.0 kangaroo/ ha resulted in increased herbage mass, yet still retained large numbers of kangaroos in the grassland ecosystem (Table 2, Fig. 7). It is interesting to note that contrary to intuition, high levels of kangaroo culling may not be associated with increased herbage mass where removals are insufficient to exceed ecological thresholds. For example, the removal of 800 kangaroos (0.5 kangaroos/ha) from Googong Nature Reserve was ineffective in meeting the stated objectives of increased herbage mass, and reduced soil erosion and associated water contamination in 2004, due to the compensatory influences of increased juvenile survival (ACT Government 2010, p. 37).
In order to apply the general 'grassland' target density of 1.0 kangaroo/ ha to KMUs comprising a mix of vegetation communities, this target density was adjusted for the negative effect of tree cover on herbage mass by making the target density for other vegetation structures inversely proportional to tree cover, that is 0.9/ha in Open Woodland, 0.5/ha in Woodland and 0.1/ha in Open Forest and Forest. To enable this adjustment to be applied, a vegetation map was prepared, by satellite image analysis, to provide consistent vegetation data for the full area of all KMUs (Wimpenny et al. 2015). As the proportions of each class of woody vegetation differ between KMUs, no two KMUs have the same overall target kangaroo density. Additional KMU-specific adjustments, based on expert judgement, were applied to this base formula in relation to factors such as strategic livestock grazing (e.g. for fire fuel management) or priority conservation values (e.g. the habitat preference of local threatened species; Fig. 8).
A key element of calculating numbers to cull is establishing the current kangaroo population density within a KMU. Each year from 2009, surveys were conducted to estimate kangaroo density across each KMU being considered for culling. Three counting methods, in order of preference and increasing cost, were used: Total Count, Walked Line Transect Count or Faecal Pellet Count (for details of methods see ACT Government 2010). A trial, on five sites in 2014, showed that the results of the different count methods were not significantly different, and no method produced consistently higher or lower results than the others (ACT Government unpublished data). The target density was deemed to be the average across the year (i.e. the population will start below, and end higher, than the nominated target density). Late autumn and early winter (i.e. April to June) were the preferred seasons to count kangaroos in the ACT, because this is a time of brief stability in the annual cycle of kangaroo populations (Snape et al. 2021 this issue). However, the overlap in timing with the delivery of the operational components of the culling programme presented both logistical hurdles and meant that population estimates calculated around 12 months prior to management decisions being made.
As well as conservation considerations, operational factors, including terrain, vehicular access, and public safety, influenced culling priorities. Experimental densities of kangaroos were applied in two areas of Canberra Nature Park to further inform grazing impacts on biodiversity and woodland restoration processes as part of the Mulligans Flat-Goorooyarroo Wood-land Experiment (www.mfgowoodla ndexperiment.org.au).
The future: an integrated approach within an adaptive management framework Shift from kangaroo density to density relative to grassland condition Biodiversity in grassy ecosystems is linked directly to the ground-layer vegetation, rather than to kangaroo density per se (Howland et al. 2014;Snape et al. 2018). Therefore, a major recommendation of the 2018 Kangaroo Management Research Workshop was to shift the focus of the conservation culling programme from one of maintaining 'grassland conservation densities' of kangaroos towards one more focused on the direct and integrated management of the grassy vegetation layer (Gordon & Snape, 2019). Whilst the objective of kangaroo management to date has been to avoid biodiversity loss resulting from excessive pasture depletion (ACT Government 2010), the detrimental impacts of excessive herbage mass on biodiversity have also been recognised within key conservation areas within Canberra Nature Park. Here, exclusion fencing or areas of exotic pasture species have resulted in a build-up of tall rank herbage that is avoided by kangaroos. In these cases, relying on the utilisation of rank pastures by resident herbivores was found to be an ineffective approach to reducing excess herbage which instead resulted in significant impacts of heavy grazing on adjacent areas of shorter (more preferred) grass. In this instance, the integrated use of livestock grazing and patch burning were used successfully to remove excess herbage and restore conditions favourable to grazing by kangaroos, such that the ongoing maintenance of appropriate grass structures in these areas might be achieved once again by native herbivores. The observation of decreased herbivory under higher herbage mass conditions presumably because there are structural (e.g. stems) and compositional (dead vs live leaf) components in high-biomass swards (c.f., Benvenutti et al. 2008 for cattle), which, once accumulated, are retained within grassy ecosystems, even at high kangaroo densities (> 4 kangaroos/ha). Previous studies of the functional responses of red and western grey kangaroos in chenopod shrublands (Short, 1987), and of eastern grey kangaroos in temperate grasslands (Fletcher 2006), have also demonstrated a benefit of using realtime data on grassy layer biomass to improve estimations of per capita off-take.
To address these issues, the preferred approach identified in the Kangaroo Management Research Workshop towards managing the grassy layer was to determine a 'safe operating environment' (adapted from Rockstr€ om et al., 2009) for key characteristics known to influence biodiversity responses, such as grass height, percentage of bare ground and thatch depth (Howland et al. 2014(Howland et al. , 2016aSmith et al. 2018;Snape et al. 2018). The Workshop also identified that detailed mapping of key grass communities within reserves, coupled with annual monitoring of grassy layer structure, and modelling of anticipated pasture growth rates at the management polygon scale, would provide an improved basis for managing the impacts of kangaroo grazing as part of an integrated approach to protecting conservation values.
Local research suggests that the greatest biodiversity benefits occur at an average grass height of between approximately 5 and 12 cm (Howland et al. 2014(Howland et al. , 2016aSmith et al. 2018;Snape et al. 2018) although specific 'safe operating environments' are likely to vary amongst grassy communities and/or reserves depending on individual priority biodiversity values. A heterogeneous grass layer (e.g. high variation in grass heights) is known to provide habitat for legless lizards   Brown et al. 2011;Howland et al. 2016b) and increase native plant diversity (Smith et al. 2018). The importance of promoting heterogeneity of grass structure is consistently emphasised as being key to grassland management globally (Vickery et al. 2001;Fuhlendorf et al. 2006;P€ oyry et al., 2006). At the landscape level, maintaining some areas where herbage mass exceeds this 'safe operating environment' may also contribute positively to a heterogenous grassy layer, especially where they persist to provide refugia during periods of drought (Howland et al. 2014). At the local patch level, preventing more than 30 per cent of the patch becoming short is recommended for native pasture in south-eastern Australia (McIntyre & Tongway, 2005).

Monitoring grassy layer condition
In implementing the Workshop's recommendations, management polygons were mapped across Canberra Nature Park in the 2018-2019 summer to reflect variations in grassy layer composition, and hence inherent structural attributes and appropriate use of management tools. Annual monitoring of these polygons will inform the calculation of target kangaroo densities, based on grassy layer condition within individual KMUs, providing a surrogate for effective threatened species habitat, and an evidence base for the strategic and integrated use of other herbage mass management tools such as fire (e.g. Driscoll et al. 2010 Manning et al. 2013;McDougall et al. 2016). Monitoring will include on ground and quadrat-based assessments of grass height and cover, as well as other information relevant to informing management decisions including dominant grass type, percentage of bare ground, and depth and cover of grass thatch. This approach will enable land managers to meet conservation aims at multiple scales, including within the broader KMU at which kangaroo management occurs (Gordon & Snape, 2019). Interventions are now in place, based upon existing scientific knowledge (where available), the system is monitored for its responses to those interventions, and the resulting information is used to update ecological models and future management recommendations (Morgan et al., 2018; see also Gordon 2009). With weather being the primary driver of grassy ecosystem processes, long-term weather forecasts will be important to predict future pasture growth. Improved weather forecasting information may be available from the Bureau of Meteorology (Bureau of Meteorology 2019); however, it is important to include uncertainty in the herbage growth predictions. Models already exist that could be adapted to provide seasonal and yearly predictions of vegetation production in the Nature Reserves (e.g. Moore et al. 1997;Hill et al. 2004). However, the predictions from these models would have to be tested with local data.

Current methods for estimating populations
Other recent adjustments to the kangaroo programme include more frequent evaluations of the geographic extent of KMUs, based on evidence of kangaroo population stability and predictability between annual counts, a shift in the timing of kangaroo counts from autumn to summer and the development of a simple model for estimating annual population growth rates from population density within each KMU (ACT Government 2018). It is predicted that the benefits of having more up to date density estimates when determining annual culling numbers will outweigh any disadvantage of estimating abundance when populations are unstable due to a high number of young emerging from the pouch (Fletcher 2006). The drivers of population dynamics likely deserve further attention in the ACT as research indicates that food availability, rather than population density per se, might be a stronger predictor of juvenile recruitment, adult survival and overall rates of annual population change (Portas & Snape, 2018;Festa-Bianchet in Gordon & Snape, 2019). Understanding the anticipated responses of culled populations to management, either in terms of increased immigration, emigration, juvenile recruitment, or mortality (e.g. roadkill), will have a profound influence over the overall effectiveness of reducing abundance below an ecological threshold, and so should continue to be addressed as a priority in this and future wildlife management programmes.

Recommendations
Taken together, the review of current practices and recent research presented at the Kangaroo Management Research Workshop (Gordon & Snape, 2019) recommended steps to setting culling targets that are shown in Fig. 9. In addition to this recommended approach, the Workshop identified the need to consider a mechanism for identifying priority actions at the landscape scale within the context of environmental risk. For example, management decisions may prioritise mitigating risk at the lower grass height threshold of the safe operating environment (e.g. to avoid loss of species and soil through erosion) rather than focus on areas where excessive herbage mass may also threaten conservation values. However, these decisions will have to consider sitespecific conservation values in addition to operational and budgetary constraints. Finally, the monitoring protocols used to inform this adaptive management framework, and the validity of using grassy layer structure as a surrogate for biodiversity, need to be reviewed at appropriate (e.g. five yearly) intervals.

Concluding Remarks
In this paper, we have reviewed the history of the kangaroo management programme in conservation reserves of the ACT, described the current conservation management policy enacted by the ACT Government, and provided recommendations for the future management of kangaroos, and total grazing pressure, in the ACT's nature reserves. Left unchecked, kangaroo grazing can have adverse impacts on the conservation values of grassy ecosystems. Recent research has demonstrated the importance of vegetation structure and composition for a range of native temperate grassy ecosystem taxa within the Australian Capital Territory (ACT), as well as specific grassy habitat requirements for species of conservation concern. Extensive research and modelling have identified approximately 1 kangaroo/ha to be appropriate to achieve biodiversity conservation objectives in these grassy ecosystems. Within the ACT, kangaroo densities are assessed and managed to meet these conservation objectives across multi-tenure landholdings within a kangaroo management unit (KMU).
The approach taken to conservation culls of the Eastern Grey Kangaroo in the ACT is likely of broad relevance to management of overabundant native wildlife elsewhere in peri-urban Australia, in terms of policy development, community responses, and the nature of the scientific evidence upon which such programmes would likely depend. The broader implications of the management approach for kangaroos in the ACT are the need for government agencies to work in collaboration with stakeholders across civil society to ensure a social licence to operate. Key to achieving conservation aims is the transition from a cull focused on density reduction to one that sets key performance indicators based upon the conservation goals. As with many other circumstances, the evidence base for these management actions in the ACT has grown over time, and so an adaptive management approach of plan, act, monitor and review will allow for further evidence to be brought to bear in future decision-making. Kangaroo management in the ACT demonstrates this cycle in action, and is a good model for macropod management across the Australian states and territories and wildlife management more generally across the globe. ing the workshop a success. Note that MS, BH, CW and GB declare a conflict of interest as employees of the ACT Government. Iain Gordon is additionally affiliated with the James Hutton Institute (Aberdeen, AB15 8QH, UK), Central Queensland University (Townsville, Queensland, Australia) and Land & Water (CSIRO, Townsville, Queensland, Australia), and Sue McIntyre is additionally affiliated with Land and Water (CSIRO, Canberra, ACT, Australia). This paper is part of the special issue 'Optimum management of overabundant macropods' published in Ecological Management & Restoration.