Environment Protection and Biodiversity Conservation (National Recovery Plan for the Sandhill dunnart (Sminthopsis psammophila)) Instrument 2026
2Species information
Start here
Get a plain-English read of 2
Turn the raw legal text into a practical explanation grounded in Environment Protection and Biodiversity Conservation (National Recovery Plan for the Sandhill dunnart (Sminthopsis psammophila)) Instrument 2026.
This is the National Recovery Plan for the Sandhill dunnart (Sminthopsis psammophila), made in accordance with Part 13 of the Commonwealth Environment Protection and Biodiversity Conservation Act 1999 (EPBC Act). This plan guides conservation of the species across its range and identifies the research and management actions necessary to halt decline and support its recovery, so that its chances of long-term survival in nature are maximised.
1.1 Conservation status
The Sandhill dunnart is listed as 'Endangered' under the EPBC Act.
The Action Plan for Australian Mammals (Woinarski et al. 2014) assessed the national status of the species as 'Vulnerable’, a conclusion which was supported by the most recent assessment for the International Union for Conservation of Nature (IUCN) Red List of Threatened Species (Woinarski and Burbidge 2016). Both assessments used the following IUCN criteria (2001 ver. 3.1):
(B2) area of occupancy < 2,000 km2;
(a) number of locations ≤ 10;
(b) continuing decline in:
(ii) area of occupancy,
(iii) area, extent and/or quality of habitat,
(iv) number of locations or subpopulations, and (Woinarski et al. 2014 only)
(v) number of mature individuals);
(D2) restricted area of occupancy, where number of locations ≤ 5.
In South Australia (SA), the Sandhill dunnart is listed as ‘Vulnerable’ under Schedule 7, National Parks and Wildlife Act 1972 (NPW Act). In Western Australia (WA), the species is listed as ‘Endangered’ under the Biodiversity Conservation Act 2016 (BC Act), based on IUCN criteria B2ab(ii,iii,iv,v) (2012 ver. 3.1 2nd ed.). In the Northern Territory (NT), the species is listed as ‘Data Deficient’ under the Territory Parks and Wildlife Conservation Act 2006 (TPWC Act).
Internationally, the Sandhill dunnart is listed on Appendix 1 of the Convention on International Trade in Endangered Species of Wild Fauna and Flora (CITES), which prohibits international trade in specimens of the species except when the purpose of the import is not commercial (such as scientific research).
1.2 Conservation action
1.2.1 Surveys and monitoring
Several surveys targeting Sandhill dunnarts have been conducted and are continuing in the Great Victoria Desert (GVD) and on Eyre Peninsula since 2000. These have resulted in two new spatially independent locations 150 km north of the known range in WA (near Laverton) and several new locations recorded within the species’ known range. In addition, information has been gained on the ecology, diet, habitat, distribution, genetics, and threats to Sandhill dunnarts (McLean 2015, Riley 2020).
Environmental Impact Assessments (EIA) associated with mining activities have increased survey efforts and added data to assist in understanding the species (Gaikhorst and Lambert 2014). Most of these surveys from north of Queen Victoria Spring to the north-west of Plumridge Lakes Nature Reserve have been associated with the EIA for the Tropicana Gold Mine (Tropicana Joint Venture 2010; EPA 2010). The Great Victoria Desert Biodiversity Trust (GVDBT) (Sandhill Dunnart Workshop 2014), setup as an offset for the Tropicana Gold Mine, are also seeking to develop standardised survey methods and long-term monitoring protocols for monitoring Sandhill dunnarts in the two western subregions (Shield and Central) of the GVD in WA (2015a; 2015b). These are intended to build upon the 'Survey guidelines for Australia’s threatened mammals’ (Australian Government 2011). Iluka Mine in Yellabinna have conducted surveys for the species over many years as well an including collecting baseline and monitoring data.
There have been irregular surveys by Ecological Horizons, other researchers and the SA Department for Environment and Water (DEW) in the Middleback ranges over recent decades. Landscape Board National Landcare Program (NLP) funded projects in the Yellabinna, Pinkawillinie and Gawler Ranges areas.
Surveys have been ongoing in the Yellabinna Mallee country via the Alinytjara Wilurara Landscape Board (AWLB). Surveys in the far-west including Yellabinna were undertaken over the past ten years via the Alinytjara Wilurara NRM (now Alinytjara Wilurara Landscape Board) from 2018 to 2023, which are planned to continue if funding permits. The Middleback Alliance undertook aerial fox baiting over the past 5 years and a survey to attempt to quantify the impacts, although this was made difficult by drought (R. Brandle pers comms).
Turpin and Lloyd (2014) and Turpin and Riley (2017) have conducted the Sandhill dunnart Monitoring Program for APA Group biannually since 2014 along the Eastern Goldfields Pipeline in the GVD in WA, representing the only long-term monitoring studies for the species in WA. Riley (2020) and Riley et al. (2021b) were also funded by the Goldfields Environmental Management Group (GEMG) to perform motion camera surveys for the species throughout WA.
Further surveys are required in areas which are likely to be suitable for Sandhill dunnarts in SA and WA, including in the GVD of WA and SA (Map 2 ). Potential survey areas within the GVD include:
the mallee/spinifex habitat north of Yellabinna Regional Reserve (YRR),
the spinifex and sand dune communities in the Maralinga Tjarutja Lands,
Mamungari Conservation Park, and
the southern Anangu Pitjantjatjara Yankunytjatjara Lands (APY Lands).
Sandhill dunnart records are included in the Atlas of Living Australia (Atlas of Living Australia 2025).
The Alinytjara Wilurara NRM Board has been working with communities across the region to implement the draft Sandhill dunnart recovery plan (Churchill 2001a). This has included undertaking Sandhill dunnart surveys in the Maralinga Tjarutja Lands and training in sand-plot tracking through the Oak Valley Land Management Engagement program (Alinytjara Wilurara Landscape Board 2015).
In addition, long-term monitoring has been undertaken with surveys occurring at the Immarna site since 2007 – a location within the YRR just to the east of the Ooldea siding. Surveys have focused on pitfall trapping and measuring Triodia densities and heights (Moseby, et al. 2016).
In 2018, Alinytjara Wilurara Landscape Board (AWLB) secured National Landcare Program 2 funds to continue these works, not only at the Immarna site, but also across much of the Yellabinna and Yumbarra reserves area, focusing on determining the key habitats for the species, as well as attempting to extend the current known areas of occupancy within this area. Surveys have been planned by examining fire-scar mapping to target areas with a post-fire age of 15 years, or more. This has resulted in the species being found at three more locations, suggesting that these dunnarts occur across Yellabinna Regional Reserve. This project was funded until 2023 and aimed to determine whether there are any “core” habitat areas or refuge areas (Pavey et al. 2017) for the Sandhill dunnart and, if so, how these might best be defined spatially and ecologically. It also includes threat abatement works focused on reducing predation pressures from foxes and feral cats, and surveillance against the establishment of the habitat transforming weed, buffel grass.
1.2.3 Captive breeding research
In 2005, five Sandhill dunnarts (3 males and 2 females) were collected from the GVD in WA and transferred to Perth Zoo for captive breeding research (D. Jose pers. comm.). A total of six litters were produced over three seasons, with 41 animals bred. Research was published on their oestrus cycle, gestation period, and the growth and development of their young (Lambert et al. 2011), and a Husbandry Manual was also produced (Ferguson 2008). At the conclusion of the project in 2007, the remaining animals were relocated to Cleland Wildlife Park, Alice Springs Desert Park, or left on display at Perth Zoo. None were released back into the wild. Moonlit Sanctuary kept a single male from 2010 to 2011. The last recording of a Sandhill dunnart in captivity is a single male from Alice Springs Desert Park in 2012.
1.2.4 The Great Victoria Desert Biodiversity Trust
The Great Victoria Desert Biodiversity Trust (GVDBT) was established in 2014 as an independent not-for-profit entity as part of the approval process for the Tropicana Gold Mine. The GVDBT aims to establish a bioregional plan (for two GVD sub-regions in WA), undertake priority research and management (including research on the Sandhill dunnart), and facilitate the involvement of Traditional Custodians.
Workshops on Sandhill dunnarts were held in 2014 and 2019 to determine and prioritise research and management activities to help conserve viable populations of Sandhill dunnart in the WA GVD. Knowledge gaps identified during the workshops included, distribution, threats, effectiveness of mitigation strategies, monitoring, habitat requirements, diet and life history, and appropriate management. Based on the workshops, the GVDBT developed a Research and Adaptive Management Plan (RAMP) for the Sandhill dunnart (GVDBT 2016), and the recovery actions in this plan (Summary of recovery objectives, recovery actions, performance criteria, potential partners and priority ratings) are designed to complement the research priorities and conservation action priorities identified in the RAMP. In 2016 DBCA produced for the GVDBT a report on defining a baseline survey design for the Sandhill dunnart in the WA GVD and this was updated to include more guidance on the use of camera traps. These guidelines were revised in 2020 (GHD 2020).
1.2.5 Habitat protection in declared conservation areas
The GVD covers an enormous area, approximately 420,000 km2, and is the largest dune desert in Australia. It extends from north of the Nullarbor Plain to the Gibson Desert and Central Ranges in WA and to the central Australian ranges. From west to east, it extends from Laverton in WA to Coober Pedy in SA. Approximately half of this land is managed by Traditional Custodian groups, including Mirning, Anangu Pitjantjatjara Yankunytjatjara, Maralinga, Kokatha, Wirangu and Barngarla groups, and Spinifex people (Pila Nguru Aboriginal Corporation). A further 70,000 km2 (about one fifth) of the GVD is in conservation parks and reserves. Almost all the remaining area is land owned by the State (known as ‘Crown land’ or ‘Unallocated Crown Land’), much of which is overlain by mining tenements (including Tropicana Gold Mine and Iluka Resources) and native title claims.
While large areas of potential habitat of the Sandhill dunnart are reserved in several conservation areas, Queen Victoria Spring Nature Reserve is the only one known to have Sandhill dunnarts in WA. All other known populations occur on Unallocated Crown Land. In the SA portion of the GVD, most Sandhill dunnarts have been found in YRR with some additional records in Yellabinna Wilderness Protection Area and Yumbarra Conservation Park, despite survey efforts further north (Moseby et al. 2016).
The majority of Eyre Peninsula (54%) has been cleared for agriculture (DEWNR 2014a) and much of the uncleared land is under grazing leases. One of the largest areas of remnant vegetation is the Central-Northwest Linkage, spanning from Hincks Conservation Park through to the Kulliparu Conservation Park, approximately 9,900 km2. This area contains 6,820 km2 (69%) of native vegetation, of which 3,170 km2 (46%) is protected either in national parks, conservation parks or Heritage agreements. The discovery of Sandhill dunnarts in Hincks Wilderness Protection Area in 2005 has prompted further survey work within this broader area.
There are several conservation areas on northern Eyre Peninsula that contain known or potentially suitable habitat for Sandhill dunnarts, including Pinkawillinie Conservation Park (1,271 km2), Ironstone Hill Conservation Park (197 km2), Hambidge Wilderness Protection Area (380 km2), Hincks Wilderness Protection Area (669 km2), Gawler Ranges National Park (1786 km2) and Pureba Conservation Park (2264 km2). While brief initial surveys of Pinkawillinie Conservation Park (Churchill 2001b) and Hambidge Wilderness Protection Area (NCSSA 1967) failed to find any Sandhill dunnarts, survey work in 2005 successfully captured two animals in Pinkawillinie Conservation Park. In 2010, another Sandhill dunnart was found in Pinkawillinie Conservation Park (July 2010), and two were found in Hambidge Wilderness Protection Area (July 2010). During 2020 one animal was captured in the north-east of the Pinkawillinie Conservation Park (Moseby and Lynch 2020) with camera detections during 2020 and 2021 at 4 nearby locations and two locations in the north-western area of the park (all within 5 km of previous known locations). During 2022 follow up work (Bell 2022) captured a sub-adult near the eastern boundary of Pureba Conservation Park. Two 14 camera arrays set during this period, one near this capture location, the other on the southwestern boundary of the Gawler Ranges National Park, detected sandhill dunnarts at one camera each (DEW datastore). Further survey work is needed to verify the presence of the species in contiguous habitat between these locations and within other reserves of the region. Gawler Ranges National Park and Yumbarra Conservation Park are co-managed by Traditional Owners.
On the northeastern Eyre Peninsula, several large conservation reserves have the potential to contain Sandhill dunnarts. Lake Gilles Conservation Park (451 km2) and Munyaroo Conservation Park (123 km2) are of particular interest given their close proximity to Sandhill dunnart capture locations in the Middleback Range area (particularly Ironstone Hill Conservation Park and Secret Rocks). While some surveys have been conducted at Lake Gilles Conservation Park (Churchill 2001b; EP Project Group 1973), Munyaroo Conservation Park has not been adequately surveyed and requires further investigation. Further survey is also warranted in Pureba Conservation Park, as sandhill dunnarts have been recently recorded near the boundary (Bell 2022).
Some of the Sandhill dunnarts found in the Middleback Range area also occur on grazing leases (such as Moola and Cooyerdoo) and are subject to disturbance by pastoralism. Individuals have been trapped in areas that are adjacent to iron-ore mines, as well as sites that are located between cleared lines for high-voltage electricity cables and a railway line that services the mines. The most southern capture site in the Middleback Range area is adjacent to a major highway between Whyalla and Port Lincoln and is on the boundary of another pastoral property that has been cleared previously for wheat production. The Middleback Alliance, a partnership between DEW, Arrium Mining, Ecological Horizons and other local landholders operates in this area (Middleback Alliance 2015). The Middleback Alliance aims to sustainably manage private and public land in the region including goat, rabbit and fox control, fire management, and threatened species management, including surveying and monitoring Sandhill dunnart sites.
1.3 Cultural significance
A large proportion of the wild populations of Sandhill dunnart are found on Indigenous lands managed by First Nations people. Indigenous rangers have and continue to make significant contributions to annual surveys for Sandhill dunnart. The plan aims to continue strengthening collaboration between Traditional Custodians and other land managers, with the goal of keeping Country healthy for both the Sandhill dunnart and the people. It is expected that further engagement of Indigenous groups involved in the conservation of the Sandhill dunnart will occur on a site-by-site basis as this recovery plan is implemented.
Traditional Custodians are significant partners involved in the recovery of the Sandhill dunnart and its habitat including through:
• developing and implementing land management activities to support Sandhill dunnart recovery on land managed by, or associated with, Traditional Custodians, including linkages to Healthy Country plans;
• developing and implementing research projects on land managed by, or associated with Traditional Custodians;
• opportunities to increase land management capacity through sharing of cultural and scientific knowledge;
• documenting new records of Sandhill dunnart; and
• two-way science and knowledge sharing.
The potential distribution of Sandhill dunnart overlaps with several Indigenous Protected Areas (IPAs):
Maralinga Tjarutja Lands IPA South Australia
Watarru IPA South Australia
Walalkara IPA South Australia
Yalata IPA South Australia
Apara - Makiri - Punti IPA South Australia
Anangu Tjutaku IPA Western Australia
Ngaanyatjarra IPA Western Australia
Map showing the distribution of Sandhill dunnart within SA and WA. It also includes the relevant IPA boundaries that intersect with potential Sandhill dunnart habitat.
Map 1: Indigenous Protected Areas (IPAs) overlapping with Sandhill dunnart distribution
Note: Source: Base map Geoscience Australia; species distribution data Species of National Environmental Significance database.
Note: Caveat: The information presented in this map has been provided by a range of groups and agencies. While every effort has been made to ensure accuracy and completeness, no guarantee is given, nor responsibility taken by the Commonwealth for errors or omissions, and the Commonwealth does not accept responsibility in respect of any information or advice given in relation to, or as a consequence of, anything contained herein.
Note: Species distribution mapping: The species distribution mapping categories are indicative only and aim to capture (a) the habitat or geographic feature that represents to recent observed locations of the species (known to occur) or habitat occurring in close proximity to these locations (likely to occur); and (b) the broad environmental envelope or geographic region that encompasses all areas that could provide habitat for the species (may occur). These presence categories are created using an extensive database of species observations records, national and regional-scale environmental data, environmental modelling techniques and documented scientific research.
1.4 Key partners
Implementation of this Recovery Plan will depend on a coordinated approach between a range of stakeholders, which include the following:
Australian Government
Department of Climate Change, Energy, the Environment and Water
Parks Australia (Uluru-Kata Tjuta National Park)
Department of Defence
State Government
Country Fire Service
Local councils, and Local Government Authorities
SA – Department for Environment and Water (DEW) including Alinytjara Wilurara, Eyre Peninsula and South Australian Arid Lands Landscape Boards
NT – Department of Land Resource Management
WA – Department of Biodiversity, Conservation and Attractions (DBCA), Rangelands Natural Resource Management, Department of Fire and Emergency Services, Department of Water and Environmental Regulation, Department of Mines, Petroleum and Exploration
Traditional Owners
SA – Mirning, Anangu Pitjantjatjara Yankunytjatjara, Maralinga, Kokatha, Wirangu and Barngarla
WA – the Spinifex People (Pila Nguru Aboriginal Corporation) Upurli Upurli Nguratja, Untiri Pulka, Pilki People, Nangaanya-ku, Yilka, Desert Support Services, Central Desert Native Title Services
NT – Central Land Council (Katiti-Petermann Indigenous Protected Area)
Mining/gas companies
Including, but not limited to – AngloGold Ashanti Australia Ltd (Tropicana Joint Venture Gold Mine); Iluka Resources Limited; Vimy Resources (Mulga Rock), Deep Yellow, Lost Sands Pty Ltd
Other
Friends groups (such as the Friends of the Great Victoria Desert)
Great Victoria Desert Biodiversity Trust (GVDBT)
Middleback Alliance, SA
Private landholders
Goldfields Environmental Management Group (GEMG)
2 Species information
2.1 Description
The Sandhill dunnart, Sminthopsis psammophila, is the second largest of 19 species of the marsupial genus Sminthopsis (Dasyuridae) that occur predominantly in arid and semi-arid regions of Australia. All species are nocturnal, predominantly insectivorous and are similar in appearance, characterised by their long-pointed snouts, large eyes and ears, and relatively long slender hind feet.
The Sandhill dunnart differs from other members of its genus by its large size at maturity – mass 25–55 g, body length 85–114 mm, and tail length 106–133 mm (Pearson and Churchill 2008). It has comparatively large eyes and ears, and a distinctive tail that is an excellent identifier as it is clearly contrasted, pale above and dark grey below, tapering towards the tip and has a crest of stiff black hairs along the ventral surface of the distal portion (Archer 1981; Figure 1). It has distinctive bicoloured fur patterning, drab grey to buff fur on the dorsal surface and white fur on the underside with darker hair nearer the skin. It has a pale grey head and black pencilling extending from the shoulders to the wedge between the eyes and a black eye-ring.
Two photographs of the Sandhill dunnart.
More information on the Sandhill dunnarts description is available in Section 2.1: Description
The Sandhill dunnart is known to occur in three core population areas as shown in Map 2 (near Queen Victoria Spring Nature Reserve in the southwestern corner of the GVD in WA; YRR in the southeastern GVD in SA; and Eyre Peninsula in SA). The estimated extent of occurrence (EOO) is 151,635 km² and area of occupancy (AOO) is 500–1900 km2 (Woinarski and Burbidge 2016; Map 2). Habitat clearance in the southern portion of Eyre Peninsula greatly reduces available habitat. The chenopod shrubland communities in the areas south and east of Whyalla do not support significant habitat favoured by Sandhill dunnart.
The EOO is now considered stable, but the AOO is declining due to the limited availability of suitable habitat as a result of the frequent loss of suitable habitat by wildfire and other threats (Churchill 2001, Woinarski et al. 2014, Riley 2020). While wildfire is an important ecological process, the limited availability of structurally suitable habitat means that burning of suitable habitat in the current matrix constitutes a threat to Sandhill dunnart (Section 2.5.1: Inappropriate fire regimes).
Targeted efforts in the southwestern GVD in 2018 identified four additional sites within the WA GVD population where Sandhill dunnarts occur, extending the WA GVD population range approximately 70 km south and confirming previous baseline study findings (Greening Australia 2019). The surveys confirmed the presence of the species from a previously un-surveyed area situated on the eastern fringe of a local broad area of undulating yellow sandplains. The Sandhill dunnart was recorded within patches of long-unburnt vegetation situated within landscapes affected by widespread recent fire (several high-intensity burns from 2000 to 2017; US Geological Survey 2018). Sandhill dunnarts were found in areas where the local topography provided some protection from fire (occurring within swales (between tall sand dunes), sand plains, and dune slopes (Riley et al. 2021a) or adjacent to mulga shrublands) (Greening Australia 2019).
Riley et al. (2021b) used predictive species distribution models with ground validation in April 2018 to confirm 18 new spatially independent Sandhill dunnart records in the WA GVD, including several in a remote region 150 km northwest of the known range. In SA, new records were obtained in Yellabinna National Park, near the southeastern Boundary of Pureba Conservation Park, and on the western boundary of Gawler Ranges National Park (R Brandle, pers. comm.) in areas that were predicted as suitable for the species using models (Riley et al. 2021b).
Modelled distribution of Sandhill dunnart across SA and WA.
More information is available in Section 2.2: Distribution and population structure
Map 2: Modelled distribution of Sandhill dunnart
Note: Source: Base map Geoscience Australia; species distribution data Species of National Environmental Significance database.
Note: Caveat: The information presented in this map has been provided by a range of groups and agencies. While every effort has been made to ensure accuracy and completeness, no guarantee is given, nor responsibility taken by the Commonwealth for errors or omissions, and the Commonwealth does not accept responsibility in respect of any information or advice given in relation to, or as a consequence of, anything contained herein.
Note: Species distribution mapping: The species distribution mapping categories are indicative only and aim to capture (a) the habitat or geographic feature that represents to recent observed locations of the species (known to occur) or habitat occurring in close proximity to these locations (likely to occur); and (b) the broad environmental envelope or geographic region that encompasses all areas that could provide habitat for the species (may occur). These presence categories are created using an extensive database of species observations records, national and regional-scale environmental data, environmental modelling techniques and documented scientific research.
Investigation into the contemporary genetic structure of the Sandhill dunnart show that the three core populations are genetically differentiated and, therefore, may represent functionally separate populations (McLean 2015). However, examination of the historic genetic structure found limited genetic differentiation between core populations (Gaikhorst et al. undated; McLean 2015), suggesting the three core populations were once connected. This absence of major geographical barriers between populations suggests the genetic pattern may be a result of fragmentation, with connecting populations going locally extinct over time (McLean 2015). Alternatively, core regions may still be connected by undetected populations, particularly in the southern GVD where there may be large areas of suitable habitat (McLean 2015; McLean, et al. 2019). Further survey effort is needed to clarify the presence or absence of Sandhill dunnart populations between the three core population areas.
The distribution of the Sandhill dunnart has declined since European settlement. Its pre-European distribution extended further north and west. The species was previously found in NT where it was first captured near Lake Amadeus in 1894 (Spencer 1896), and bones in caves were recently verified from near Yalgoo and Lake Barlee in Western Australia, 400–600 km west of the known WA GVD population (Riley et al 2021b). Subsequent surveys have failed to relocate the species in the NT, but further surveys are warranted.
Owl pellets containing remains of Sandhill dunnarts have come from caves at the base of Uluru, NT (Parker 1973; Archer 1981), and caves at Darke Peak on Eyre Peninsula, SA (Baynes 1987). Owl pellets containing sub-fossil Sandhill dunnart remains have been collected from Black Point Sinkhole at Venus Bay, SA. These sub-fossil deposits were radiocarbon dated at 3,030–2,160 years before present (Baynes 1987; Copley et al. 1999).
2.2.2 Important sub-populations
The three core populations (WA GVD, YRR and Eyre Peninsula) of Sandhill Dunnarts consist of five known sub-populations. All sub-populations are considered important for the long-term recovery and survival of the species. These sub-populations are located in the following areas:
Southwestern Great Victoria Desert, WA (WA GVD)
Great Victoria Desert, SA (SA GVD) including:
Yellabinna Regional Reserve,
Yellabinna Wilderness Protected Area,
Yumbarra Conservation Park, and
East of Pureba Conservation Park.
Northcentral Eyre Peninsula (EP), SA including:
Pinkawillinie Conservation Park (near Kyancutta), and
Gawler Ranges National Park.
Middleback dunefields (near Whyalla), northeastern Eyre Peninsula (EP), SA
Central Eyre Peninsula (EP), SA including:
Hincks Conservation Park, and
Hambridge Conservation Park.
2.3 Habitat
The presence of dense pre-senescent, usually long unburned (greater than 20 years without fire) spinifex (Triodia spp.) hummocks in association with sand dune systems appears to be the most significant factor explaining known Sandhill dunnart habitat in SA (Churchill 2001; Ward 2009), although current distribution over fire history suggests that there are also other factors that explain distribution. In Western Australia, Riley et al. (2021a) concluded through radio tracking that the survey and conservation of the Sandhill dunnart necessitate attention to habitats that have not burned for over 32 years. These habitats include dense lower stratum swales, sand plains, and dune slopes. This was supported by MaxEnt modelling (Riley 2020). The distribution is also constrained by semi-arid climatic conditions (Riley 2020). Modelling supports the view that Sandhill dunnart distribution is constrained by both the availability of Triodia spp. habitat (that is restricted to sandy soils) in dense southern semi-arid deserts and desert influenced by higher winter rainfall than Australia’s northern and central deserts (Riley 2020).
In WA, Sandhill dunnarts are primarily found in plains and dunes (longitudinal and ring dunes) with interdune corridors and plains; occasional salt pans at a mean ± standard deviation elevation of 402 ± 22 m a.s.l. (DAFWA 2014; Riley 2020). Vegetation communities are dominated by spinifex, predominantly T. desertorum and T. basedowii, hummock grasslands with scattered eucalypts including marble gums, Eucalyptus gongylocarpa, situated over wattle scrub, Acacia spp., and mallees, Eucalyptus spp. (Beard et al. 2013). The study site soils are ‘Yellow deep sand, Soil Group 446’ except for ‘Mulga’ habitat class soils where a variable depth sandy loamy clay crust is present (Schoknecht and Pathan 2013).
Sandhill dunnart models (Riley 2020) responded to surface geologies of ‘Qd’ (Dunes, sandplain with dunes and swales; this category may include numerous interdune claypans and may be locally Gypsiferous), ‘Czs’ (Sand or gravel plains that may include some residual alluvium, quartz sand sheets commonly with ferruginous pisoliths or pebbles, local clay, calcrete, laterite, silcrete, silt, and colluvium); and ‘Ln’ (Middleback subgroup: comprising Jaspilite, quartzite, gneiss, quartz-mica schist; and dolomitic marble). Understanding these geological formations is important for modelling the presence of Sandhill dunnart (Raymond et al. 2007).
In SA, species associations vary but are generally low, open mallee woodland over a diverse shrub layer and relatively dense spinifex; there is also a greater diversity of Triodia species which may be present, including T. irritans, T. scariosa and T. lanata.
In the WA GVD recent radio-tracking showed that Sandhill dunnarts are most often associated with dense, long-unburned sand plain, swale and dune slope habitats (Riley et al. 2021a, Riley et al. 2022), with a strong preference for dense lower habitat stratums and mature vegetation. Sandhill dunnart speciation occurred approximately 15 million years ago (Archer 1982), while dune development in the GVD was geologically recent (up to 781 000 years ago; Madigan 1936), so it is likely that sand dunes are probably not a prerequisite habitat feature for the species to persist (Riley 2020). Riley (2020) also found that Sandhill dunnart sheltering habitats could be far away from dune crests within swales or sand plains. Nonetheless, 82% of all shelters were within 500 m of a dune crest (Riley et al 2021a), suggesting that dunes and their influence on vegetation and fire dynamics enhance habitat suitability.
There appear to be additional areas of potentially suitable habitat in the southern GVD where the species has not yet been recorded – due to low, no or inappropriate survey effort, although they are patchily distributed (extract from Riley et al. 2022). Suitable habitat on Eyre Peninsula is limited due to historical land clearance.
2.3.1 Shelter preferences
Sandhill dunnarts typically use burrows as diurnal refuges to maintain thermodynamic equilibrium (Porter and Gates 1969; Riley 2020, Riley et al. 2021a), and individuals have been observed using sun basking (passive rewarming) at burrow entrances, particularly during cooler conditions (Churchill 2001a, 2001b, Riley et al. 2021a). Shelter selection appears to be non-random within warmer habitat classes, and around 50% of entrances in a WA GVD radio tracking study were observed to face north (Riley et al. 2021a). Sandhill dunnarts have been observed using other types of shelter, including spinifex hummocks (Riley 2020, Churchill 2001a; Figure 2 f-h). In the Eyre Peninsula 68% of shelters were spherical nests within ‘Stage 3’ spinifex hummocks, 9% of shelters were within logs and 23% were burrows. In YRR, 96 % of shelters located by Churchill (2001b) were burrows (Riley 2020). In addition, fluorescent tagging of Eyre Peninsula individuals by Philp (2011) demonstrated that burrows were preferred as refuges during nocturnal foraging. In the Eyre Peninsula, when sheltering in spinifex, Sandhill dunnarts appeared to prefer large hummocks which have either an intact structure or have just started to die off in the centre (Churchill 2001a; Figure 3, life stages 2–4). However, it is unclear whether there are climatic or ecological differences affecting shelter use among populations, or whether these nests are diurnal or temporary shelters. In the WA GVD, when spinifex was chosen, Sandhill dunnarts opted for softer leaved species that would likely provide lower protection from predation, but which appear to have superior thermal properties, suggesting that thermal factors could be significant for the species (Riley 2020).
Together, these findings suggest that the habitat and shelter preferences of the species are complex, and may be population specific, and that site-specific characteristics are important in targeting management and recovery actions.
Composite of 8 images, all examples of different Sandhill dunnart burrows
Figure 2: Examples of Sandhill dunnart burrows
Note: (a) Exposed burrow with emerging individual inset; (b) Concealed burrow under a ‘Stage 5’ Triodia desertorum; (c) ‘Stage 3’ T. desertorum hummock; (d) Lepidobolus deserti hummock with individual inset; (e) Schoenus hexandrus hummock with individual inset; (f) Log shelter; (g) Bark shelter; (h) Burned mallee stump hollow with individual inset. Extract from Riley et al. 2021a Supplementary Data SD5.
2.3.2 Spinifex hummocks – Fire and structure
The habitats preferred by the Sandhill dunnart, including dense sand plains, swales, and dune slopes, thrive in the absence of wildfire. In some areas, it has been over 40 years since a fire occurred. This long period without fire allows for the development of a range of suitable sites that provide adequate shelter and reduces the risk of predation (Brown et al., 1999; Laundré et al., 2001, 2010).
Post-fire age of spinifex may be an important habitat indicator due to its influence on the cover and structure of Triodia. Recently burnt hummocks are too small for Sandhill dunnarts and long unburnt sites generally provide limited cover due to senescence of Triodia. The post-fire age of spinifex that Sandhill dunnarts will inhabit varies between sites and regions. In WA, most captures have been in habitat that was 17–26 years post-fire, with an optimum age structure recommended of 10–30 years post-fire (Gaikhorst and Lambert 2014). Riley et al. (2021a, 2021b) confirmed with tracking and satellite imagery that the mean fire age of preferred Sandhill dunnart nesting and foraging habitat in the WA GVD is 32+ years (range: 0–40+ years). In SA, captures have been in habitat ranging from 5 to >60 years post-fire (Churchill 2001b; Moseby et al. 2016; Ward 2009; Philp 2011; McLean 2012). Sandhill dunnarts captured at some SA sites with older spinifex (for example, Ooldea in the GVD and occasionally on Eyre Peninsula), have been shown to occupy burrows rather than nesting within hummocks (Churchill 2001a).
Given the apparently non-constant post-fire growth of Triodia at SA Sandhill dunnart capture sites (being also affected by rainfall, substrate, soil nutrients, and species), structural characteristics may be more useful predictors of Sandhill dunnart presence than post-fire age (Moseby et al. 2016). McLean (2015) found no preference for post-fire age. The number of logs (≥5 cm diameter), average height of Triodia hummocks and the overall complexity of vegetation were found to be the three most important habitat variables for predicting Sandhill dunnart abundance. Using modelling, Moseby et al. (2016) also found Sandhill dunnarts were more likely to populate areas which were ≥10 years post-fire where at least 10% of the Triodia had a minimum height of 330 mm. Moseby et al. (2016) recommend that fire be used as a management tool for maintaining habitat where Triodia cover declines below 10%, rather than using ‘post-fire age’ as a sole indicator.
2.3.3 Habitat important to persistence and recovery
Although habitat availability does not appear to be a limiting factor to the Sandhill dunnart, at least in the southern GVD, due to the relatively high mobility of the species, large areas of suitable habitat will need to be available to protect remaining populations (McLean 2015).
Further research is required to identify and map the habitat critical to survival of the species based on known habitat characteristics rather than relying on trapping programs which fail to consistently detect the species (Moseby et al. 2016). Maximum entropy models combined with accurate fire mapping have been valuable in locating unknown Sandhill dunnart populations in WA (Riley et al. 2021b). Landsat imagery is recommended for fire scar mapping in the GVD (Ruscalleda-Alvarez et al. 2021).
The most important habitat is considered to comprise:
Unburned habitat at extant Sandhill dunnart sites (based on the current available knowledge; Map 2);
Marble gum, mallee and Triodia spp. between dunes in the GVD of WA;
All Triodia sand dunes with a mallee Eucalyptus overstorey on Eyre Peninsula;
All Triodia sand dunes with a mallee Eucalyptus overstorey which occur between the three core population areas (Map 2); and
Areas within 500 m of a Dune crest within occupied sites in the GVD.
A composite of photographs which show the life stages of some Triodia spp.
Figure 3: Life stages of some Triodia spp.
Note: Drawings by Vicki Reynolds, photos are from Gaikhorst and Lambert (2014). At Stage 1 (Fig. 3a) the Triodia is a small tusk of grass. It then grows into a dense hummock by Stage 2 (Fig. 3b). At Stage 3 (Fig. 3c), some species of Triodia start to die off in the middle and continues to grow outwards until it reaches a ring (Stage 4; Fig. 3d), which breaks down into fragments (Stage 5; Fig. 3e).
Source: Churchill 2001b
2.4 Ecology
2.4.1 Diet
Sandhill dunnarts typically consume invertebrates including ants, beetles, spiders, grasshoppers, termites, wasps and centipedes (Pearson and Churchill 2008). Small reptile bones (possibly a gecko) have been recorded from one scat (Churchill 2001b). At least two other Squamata species have been recorded, and Gastropoda have been detected in the faecal pellets of several individuals (Riley 2020, Riley et al. 2022). Sandhill dunnarts were previously thought to be largely opportunistic feeders, eating prey types in similar proportions to their availability (Churchill 2001b). Recent research indicates that (rarely for small arid zone mammal species) Sandhill dunnarts have an ant-rich diet, consuming a diverse range of ants in greater than expected proportions (Riley et al. 2022), and demonstrates metabolic controls that can allow for an ant-rich diet (which is difficult to digest), such as the ability to manipulate their metabolic rate with sun basking, shelter choice, and selective torpor (Riley 2020). It is difficult to say whether they have developed adaptations to facilitate consumption of ants, broadening their niche to deal with the scarcity of other prey at certain times of the year, or prefer ants (Shipley et al. 2009). Nonetheless, an ant-based diet appears to be an important survival adaptation for desert fauna in the WA GVD.
Sandhill dunnarts in the WA GVD generally foraged continuously from dusk to dawn, except during heavy rainfall and very cold (0–5°C) temperatures (Riley et al. 2021b).
2.4.2 Reproduction
Sandhill dunnart females are polyoestrous, which means females have multiple oestrus cycles with 22–23 days between cycles (Riley 2020). Both male and female Sandhill dunnarts reach sexual maturity in their first year. Wild individuals usually mate following winter rains in August and September. They have a 16–19 day gestation period (Lambert et al. 2011), and young are typically born in September or October. The young remain in the pouch for up to 45 days and weaned young have been recorded between October and March (Lambert et al. 2011). The breeding season may be delayed or reduced in response to a reduction in available food resources (McLean 2015).
Females possess eight teats (Pearson and Churchill 2008) and have been recorded with up to eight young (How and Bignall 2009), although five to six are more common in the wild (Lambert et al. 2011). They usually produce a single litter each year but could possibly produce a second litter during good seasons (Churchill 2001b).
Individuals typically survive for just one breeding year in the wild, as with other Sminthopsis species, but males probably do not experience post-mating “male die off” as recorded in the Antechinus, and Phascogale genera because of differing physiologies (Morton 1982, Dickman and Braithwaite 1992, Pearson and Churchill 2008). The maximum age recorded for males and females is 19 months and 25 months, respectively, but in favourable conditions both sexes have been observed breeding over two consecutive years (McLean 2015). In captivity they have been found to breed for up to 3 seasons (Lambert et al. 2011) and live for up to six years (Sandhill Dunnart Workshop 2014).
Wild juveniles typically disperse between December and January, and both sexes are highly mobile to track intermittent resource patches (Pearson and Robinson 1990, McLean et al. 2019). Resource availability mediates condition, breeding and survival. In drought, breeding is delayed, and the survival of second year females and juveniles is reduced due to increased resource competition (McLean 2015).
Populations exhibit classic boom-bust dynamics typical of arid-zone mammals across Australia and population growth and decline are linked to rainfall (Southgate and Masters 1996, Dickman et al. 2001, Letnic and Dickman 2010).
Movement ecology studies (extensive male ranging; see Section 2.4.4: Movement ecology) suggest a polygynous or polyandrous social organisation (Riley 2020).
2.4.3 Temperature and diurnal shelter microclimate
The climate over the distributional range of the Sandhill dunnart is characterised by high diurnal variation sometimes fluctuating by over 35°C. Seasonal variation is also high with summer maximum temperatures sometimes exceeding 45°C and minimum winter temperatures sometimes falling to –4.5°C (Riley et al. 2021a; Australian Bureau of Meteorology 2025). Ground surface temperatures can reach up to 61°C because of reflective heat from the sand (Riley et al. 2021a).
The presence of large spinifex hummocks is considered an important habitat feature, as hummocks may provide a moderated microclimate suitable for Sandhill dunnart nest sites and reduce predation risk (Churchill 2001b). Sandhill dunnarts commonly nest within hummocks or in burrows excavated beneath hummocks, although they are known to also use hollow logs and Mitchell’s hopping-mouse (Notomys mitchellii) burrows (Churchill 2001b). Other hummock species (Lepidobolus deserti and Schoenus hexandrus), a burned mallee tree stump and a piece of bark have also been used during diurnal sheltering in the WA GVD (Riley et al. 2021a). In SA, the interior of spinifex hummocks has been found to be significantly cooler at high ambient temperatures (12.5°C cooler on average when the ambient temperature was >40°C), and significantly warmer at low ambient temperatures (6.5°C warmer on average when the ambient temperature was <0°C) (Churchill 2001b). However, no thermal benefit of spinifex hummocks was found in the WA GVD (Riley et al. 2021a).
Sandhill dunnarts significantly prefer burrows as diurnal shelters in the WA GVD and Ooldea in the SA GVD. Burrows provide even greater moderation of the microclimate with temperatures in the burrow (30 cm below the surface) showing virtually no variation throughout the day and night (Churchill 2001b; Riley et al. 2021a). Burrows also maintained a near constant humidity level (Churchill 2001b).
2.4.4 Movement ecology
The mean home range area (100% minimum convex polygon) recorded for Sandhill dunnarts in the WA GVD is 70 ha (Riley 2020; Riley et al. 2022). A previous study in SA suggested that the species has a mean home range of 7.8 ha, although ranging data from Churchill’s (2001b) unpublished study also supports a larger home range; Sandhill dunnarts had a maximum home range area of 42 ha in SA in that study (Churchill 2001b). This uncertainty warrants further investigation of home range sizes in SA.
There was a strong effect of sex and reproductive status on the ranging behaviour of Sandhill dunnarts in the WA GVD (Riley et al. 2022). Reproductively active females had comparatively small mean home ranges of approximately 16 ha (Churchill 2001b; Riley 2022), presumably to secure or maintain breeding territories or good quality nesting burrows (Churchill 2001b). Limited data from SA indicates that Sandhill dunnart remain within a given home range for at least 8 months, but the boundaries of territories may drift over time (Churchill 2001b). Non-reproductively active individuals (of both sexes) in the WA GVD had mean home range areas between 21 and 29 ha (Riley et al. 2022). The maximum home range area of reproductively active males is significantly larger (274 ha recorded; Riley et al. 2022). The home ranges of males can overlap those of other males and females, although it is unclear how common this is (Churchill 2001b; Riley 2020). Potentially, if female territories do not overlap, then individual males may copulate with up to as many as 17 females within their range (Riley 2020). If female territories do overlap, this number may be higher.
Sandhill dunnarts have been recorded in SA in densities of up to 30–90 individuals/km2 in SA (Ward unpublished; Ward et al. 2008), however this represents an extreme upper limit. In high quality habitat in a typical good year may be around 0.1 individuals/km2, and in a bust year density may be as low as 0.01 individuals/km2. However, densities are likely to be dependent on habitat availability, habitat quality, predation, and food resources (Gaikhorst and Lambert 2014). Making accurate population density estimates for the species is challenging. In a key SA study, the population density of Sandhill dunnart was estimated at 25 individuals/km2 of suitable habitat (Churchill 2001a). This figure was a projection of a maximum of five individuals captured at one 20 ha SA trapping site and may be an overestimate. More often, the capture rate is much more variable (likely due to droughts and deficits in resource availability) and not usually this high (McLean 2015, Turpin and Riley 2021, Riley 2020).
In SA, individuals may move 200–300 m per foraging period (two foraging periods have been recorded on warm nights), but they are able to traverse long distances in short periods of time, up to 1,960 m in two hours (Churchill 2001b). In WA, individuals were extremely mobile and had a maximum speed of 3 km/h (Riley et al. 2022). In contrast to previous dasyurid studies, Sandhill dunnart foraged continuously throughout the night until just before sunrise and then quickly returned to shelters that were up to 1 km away (Riley et al. 2022). The high mobility of the species appears to be necessary in a system with spatially and temporally variable food resources (McLean 2015).
2.5 Threats
Since European settlement there has been a major episode of mammal declines and extinctions in Australia (>10% of the 273 endemic terrestrial species) (Woinarski et al. 2015). The major drivers of this decline have been predation by introduced species (feral cats and foxes) and altered fire regimes (Woinarski et al. 2015). In some parts of the Sandhill dunnart’s range, other land use practices such as pastoralism and clearing for agriculture, have also been a major cause of decline, particularly on the Eyre Peninsula.
Along with habitat loss and fragmentation (particularly on Eyre Peninsula), predation by cats and foxes and altered fire regimes are considered severe threats to Sandhill dunnart survival (Woinarski et al. 2014). Other threats include the direct threat of predation by native carnivores, and the indirect threat of declining habitat quality caused by introduced herbivores (especially sheep, goats, rabbits, camels, and cattle) and weed species.
Climate change impacts may also increase the risk of wildfire through lower rainfall and increased lightning, as well as result in larger, hotter and more frequent fires. Moreover, a potential emerging threat affecting spinifex that may be linked to climate is the recently observed die-off events.
Sandhill dunnarts are within the Critical Weight Range (CWR) of Australian mammals, which includes medium-sized mammals of 35–5000 g. Extinction risk has been, and is predicted to remain, high among these CWR species (Burbidge and McKenzie 1989). The CWR species at greatest risk are ground-dwelling species from low-rainfall areas, as the sparse vegetation provides limited cover from predation by cats and foxes (Johnson and Isaac 2009).
2.5.1 Inappropriate fire regimes
All populations, except eastern GVD
Change in fire patterns is considered a major threat to Sandhill dunnarts (Woinarski et al. 2014). It is typically believed that historic “firestick farming” conducted by First Nations people produced complex mosaics of different age classes of vegetation (Burrows et al. 2006); this benefited species through encouraging new vegetation growth, increasing structural diversity, providing cover in unburnt patches and typically increasing food availability in unburnt areas (Latz 1995). Latz (1995) considered that most vegetation communities throughout arid Australia have become dependent on a regime of fires lit by First Nations people; however in areas of low First Nations population density (for example, waterless regions like the eastern GVD) it is more likely that lightning strikes had a greater influence on the fire regime than First Nations people (Hateley and Hateley 2010); this is also believed to be the case for Eyre Peninsula (J. Tilley pers. comm.).
Fire is a major threat to biodiversity in the WA GVD, including the Sandhill dunnart (Turpin and Johnstone 2017, Turpin and Riley 2021). Destructive summer fires (from lightning strikes) remove the mature vegetation required for protection and shelter. Repeated fire events inhibit the development of habitat. The current fire regime, arising from the cessation of traditional burning practices, is ongoing and as such, large areas of the GVD are burnt each summer. Since the commencement of monitoring in the WA GVD, fires have rendered large areas of habitat unsuitable for the Sandhill dunnart and have resulted in the loss of several known populations, including at monitoring sites known to formerly support the species (Turpin and Riley 2021). Previously occupied sites affected by fire do not support the species post burn for many years (Turpin and Riley 2021).
In general, long unburned, spinifex-dominated natural refugial habitats are important to reduce predation risk of Sandhill dunnart. Conversely, Sandhill dunnart successfully forages during periods of high predation risk (during or after fire and other disturbance events), probably by using dense habitats, burrows, and evasive techniques. Wildfire management may be the most important aspect for the conservation of Sandhill dunnart and sympatric small arid zone mammals in the Australian arid zone.
Mitigating against the likelihood of large-scale wildfires will be critical for Sandhill dunnart survival, particularly as climate change causes an increase in the frequency and severity of wildfires (McLean 2015; Riley et al. 2021b).
Summer lightning strikes are the most frequent cause of wildfire in the arid-zone, and they are likely to increase in frequency as temperatures rise – there is an approximate 6% increase in global lightning frequencies for every 1°C rise in temperature (Price and Rind 1991, 1994, Williams et al. 2001). With no active fire management when wildfires occur, large areas are likely to burn once initiated.
In some areas, the frequency of weekly wildfires has increased by 40% over the past five years, and this abruptly altered pattern is most likely related to climate change (Clarke et al. 2013; IPCC 2014; Dutta et al. 2016).
In addition to climate action, prescribed burning is recommended to reduce this threat and maintain a mosaic of suitable habitat for Sandhill dunnarts (DENR 2012; DEWNR 2014b; DEWNR 2014c; Gaikhorst and Lambert 2014; Moseby et al. 2016), but not in the eastern GVD (Armstrong 2015).
The fire management plan for Central Eyre Peninsula 2014–2024 (DEWNR 2014b) recognises that Sandhill dunnart habitats are strongly influenced by fire history. Strategies in the plan include implementing fuel management to maintain suitable age classes of hummocks for known populations, fuel management to protect against large-scale fires, and trialling ecological prescribed burns to improve Sandhill dunnart habitat. In 2010 and 2011, prescribed burns were undertaken in Sandhill dunnart habitat in the Secret Rocks area on Eyre Peninsula (northeastern Eyre Peninsula), burning a total of approximately 1300 ha. In 2014 and 2015, prescribed burns were conducted in Ironstone Hill Conservation Park to limit the spread of bushfire through Sandhill dunnart habitat and to redistribute age class of spinifex hummocks (J. Tilley pers. comm.).
Alinytjara Wilurara Landscape Board and the former SA Department of Environment, Water and Natural Resources (DEWNR), now Department for Environment and Water (DEW), also developed an ecological fire management strategy for the Sandhill dunnart in SA (DEWNR 2012). The strategy focuses on fire risk management for the Sandhill dunnart. The strategy recommends undertaking prescribed burns which maintain 66% of a given area of suitable habitat, are not within the breeding and weaning season, avoid burning optimum habitat, and avoid burning areas adjacent to habitat burnt <8 years earlier. The strategy also recommends prescribed burns in habitat >8 years post-fire and <20 years post-fire. The variability in recommended times acknowledges that vegetation growth rates in habitats with highly variable rainfall are not linear. As such the physical structure of Triodia spp. favoured by Sandhill dunnarts cannot be directly correlated to time since fire but is more likely to be associated with cumulative rainfall or microhabitat variables.
Land managers of the WA rangelands have developed Guiding Principles for fire management in the region, including areas containing Sandhill dunnart habitat (RFLG 2015). The principles recognise that prescribed burns which create mosaics of post-fire vegetation growth stages, including long unburnt vegetation, will benefit biodiversity and buffer against large wildfires.
Fire regimes that cause declines in biodiversity are a key threatening process listed under the EPBC Act. Advice on this key threatening process includes more information about the threats of inappropriate fire regimes to species (DAWE 2022).
2.5.2 Habitat loss and fragmentation
All populations, particularly Eyre Peninsula
Sandhill dunnarts have been lost from parts of their former range on central and northern Eyre Peninsula through historic clearance for agriculture (Aitken 1971). Around 54% of land has been cleared and the remaining vegetation is heavily fragmented with only 42% still connected (DEWNR 2014a). This fragmentation reduces opportunities for Sandhill dunnarts to recolonise areas of potential habitat. Sandhill dunnarts occurring in isolated native vegetation are also vulnerable to local extirpations as a result of stochastic events such as drought and large wildfires. Due to the isolated nature of remnant vegetation patches on Eyre Peninsula, McLean, et al (2019) undertook genetic analyses to determine whether gene flow is occurring between isolated vegetation patches to inform management. This was to determine the need for translocations or creation of habitat corridors (McLean et al. 2019).
At a fine spatial scale, McLean, et al. (2019) found no significant genetic structure or sex-biased dispersal within a study site of 240 km2. Their findings suggest that both sexes are highly mobile, which allows individuals to locate localised resource patches when they become available. They detected evidence of a genetic bottleneck within the population, possibly caused by a recent drought highlighting the importance of maintaining connectivity across the landscape for semi-arid and arid-zone species to enable them to track resource pulses and maintain genetic diversity. They found that the level of genetic diversity within the Eyre Peninsula sites is at risk of decline given the reduced allelic richness and heterozygosity, as well as the isolated nature of remnant vegetation patches compared with the populations in the GVD. Relatively recent (Churchill 2001b; Kemper et al. 2011; Moseby et al. 2016) and historic (Aitken 1971; Baynes 1987) records suggest that Sandhill dunnart was widespread over the central and northern Eyre Peninsula before European settlement. Since European settlement, Eyre Peninsula has lost over half (57%) of its native vegetation (Churchill 2001a; Brandle 2010). Habitat clearance in the southern portion of Eyre Peninsula greatly reduces available habitat, and its effects are significant. The chenopod shrubland communities south and east of Whyalla do not provide significant habitat preferred by the Sandhill dunnart.
The extent to which Sandhill dunnart may be affected by habitat loss and fragmentation elsewhere is unknown. However, fragmentation will likely result in isolated populations where movement, needed to track resources, is hindered by unfavourable landscapes between remnant vegetation. Loss of vegetation appears to have resulted in the contraction of the once widespread Sandhill dunnart to remnant patches in central and northern Eyre Peninsula. Research is needed to determine the long-term viability of these populations and the level of connectivity (gene flow) between them. If remnant patches are isolated, populations may become locally extinct due to environmental, demographic, and other stochastic events, such as wildfires (Lande 1988; McLean et al. 2018).
There has been no land clearance for agriculture in the GVD, however, new threats are emerging that may cause irrevocable damage to Sandhill dunnart habitat. Mining and other industry operations have been proposed or undertaken within arid and semi-arid regions of SA and WA that support Sandhill dunnart habitat. These activities often involve large scale vegetation clearance and substrate disturbance in sandy spinifex and mallee environments. In YRR, exploration tracks have been created using a roller which is designed to crush the vegetation but leave the soil surface intact (Warland and Mills 2011). Subsequent observations suggest that rolled Triodia typically dies (Sandhill dunnart Workshop 2014), however, regeneration of Triodia is likely to be faster using this method compared to bulldozing or grading, given that the rootstock and seedbank are left intact after rolling. While rolling may allow some habitat recovery, expanding mining activities continue to threaten Sandhill dunnart populations.
Given the dispersed impacts of exploration activity, it is hard to assess what impacts it may have on Sandhill dunnarts. If rolled tracks are positioned in tight grids, then direct impacts on Sandhill dunnart nests may be a concern. Land clearance and the construction of tracks also have the indirect, but potentially significant, effect of facilitating the spread of weeds such as buffel grass (Cenchrus ciliaris).
2.5.3 Predation by cats and foxes
All populations
Cats and foxes are known predators of some Sminthopsis species in arid Australia (Catling 1988; Paltridge et al. 1997; Read and Bowen 2001; Paltridge 2002), but the extent of their impact on the Sandhill dunnart is unknown. Currently core populations of Sandhill dunnarts are surviving in the presence of both cats and foxes (McLean 2015; Turpin and Riley 2021).
Analysis of 128 predator scats taken from known areas with Sandhill dunnarts has found a small percentage of scats (<2%) contained Sandhill dunnart remains, however, the predators were unidentified, but may have been a dog, dingo or fox (Boulton and Foulkes 2000). The Sandhill dunnart is ranked by Woolley et al. (2019) within the top Australian species at most risk of predation by the feral cat.
Further research into predation impacts of cats and foxes (and predation impacts of other species such as dingoes) on Sandhill dunnart populations is recommended. Several projects on feral cat control in Australia were funded as part of the Threatened Species Strategy (Australian Government 2015) and the Threatened Species Action Plan 2022-2032 (Australian Government 2022). This included developing and testing bait dispensing technologies and refining the use of feral cat baits. None of the funded projects directly relate to Sandhill dunnarts, however they may benefit and inform management of cats within the Sandhill dunnart’s range. Control of feral cats and foxes is desirable but will need to consider predator-prey interactions. For example, cat predation on Dasyurids is known to increase when consumption of rabbits is decreased (Doherty et al. 2015).
Furthermore, in a study of feral cats in northern Australian savanna, McGregor et. al. (2015) found the hunting success of the cats (per attempt; n = 13) was only 17% in habitats with dense grass or complex rocks, compared to 70% in open areas (n = 101). This suggests that dense and complex microhabitats are likely to decrease cat predation rates.
Predation by feral cats is a key threatening process listed under the EPBC Act. The threat abatement plan for predation by feral cats provides useful information on managing feral cats in a variety of landscapes and habitats (DCCEEW 2024).
Predation by the European red fox (Vulpes vulpes) is a key threatening process listed under the EPBC Act. The threat abatement plan for predation by European red fox provides useful information on managing fox in a variety of landscapes and habitats (DEWHA 2008).
2.5.4 Introduced flora (including buffel grass)
All populations
There are comparatively few serious weeds that are invading spinifex grasslands, except for buffel grass (Cenchrus ciliaris). Buffel grass is considered one of Australia’s worst environmental weeds (Humphries et al. 1991). It is a perennial tussock grass native to Africa, India and Asia and was introduced into Australia for pasture and dust control. It has spread widely across northern and central Australia (Marshall et al. 2012) including known and potential habitat of the Sandhill dunnart in WA, SA and NT (Lawson et al. 2004).
Buffel grass can increase the intensity and frequency of fire events, threatening plant and animal communities that are not adapted to these modified fire regimes (Adair and Groves 1998). Buffel grass dries quickly and produces a more combustible and contiguous flammable fuel than native grasses, resulting in hotter and more intense fires (Humphries et al. 1993). It regenerates rapidly after fire suppressing the regeneration of native species (Paltridge et al. 2009) and forms dense monocultures, reducing diversity (Clarke et al. 2005) and productivity of some vegetation types (Humphries et al. 1993).
Buffel grass is considered a major threat to Sandhill dunnarts due to the way that its dominance across landscapes leads to changes in availability of bare ground and the intensity and frequency of fires, resulting in the loss of nesting spinifex hummocks and potential changes to food availability (Sandhill dunnart Workshop 2014). Mitigation of this threat includes preventing the spread of buffel grass and controlling buffel grass where it establishes near known subpopulations of Sandhill dunnarts, including along incursion pathways (such as, access tracks and roads, and along adjacent watercourses). A SA Buffel Grass Strategic Plan (Biosecurity SA 2012) provides options for controlling the spread of buffel grass during the early stages of invasion; however, there is currently no feasible control method available once it is well established over an extensive area.
Novel biota and their impact on biodiversity is a key threatening process listed under the EPBC Act. Advice on this key threatening process includes more information on the management of invasive flora, including buffel grass (TSSC 2013).
At the time this plan came into effect a specific threat abatement plan for buffel grass is in development by DCCEEW and will be published on the DCCEEW website when finalised.
2.5.5 Introduced herbivores
All populations
Introduced herbivores (cattle, goats, sheep, rabbits and camels) may have an impact on the survival of Sandhill dunnarts indirectly in a variety of ways, including:
altering the structure of spinifex hummocks;
changing the fire regime by removing biomass;
reducing floristic diversity of habitats over time;
causing soil compaction and disturbance;
sustaining higher numbers of introduced predators (for example, rabbits and house mice provide an increased food source for foxes and feral cats [Holden and Mutze 2002], which may then switch to preying on native species when rabbit or mouse numbers decline during droughts or when rabbit numbers crash due to Rabbit Haemorrhagic Disease or myxomatosis outbreaks); and
encouraging the spread of introduced predators such as foxes and feral cats into areas that usually have only very low densities (Read and Bowen 2001, Moseby and Hill 2011).
Introduced herbivores are unlikely to have more than localised impacts on the habitats occupied by Sandhill dunnarts. The southern GVD is not suitable for introduced stock due to a lack of water sources and the low nutritional value of spinifex. On Eyre Peninsula, some sites in the Middleback dune fields are grazed by sheep, goats and rabbits; however, most of these animals appear to avoid the spinifex-dominated habitats in favour of other habitat types with more nutritious plants.
2.5.6 Climate change
All populations
Predicted anthropogenic climate change is likely to influence the distribution, habitat suitability and recovery prospects of Sandhill dunnart. Over the past century, Australia’s annual mean temperature has increased more than 1°C, while the south-west of the continent has experienced a decline in annual rainfall of up to 20% (Riley et al. 2021b).
Semi-arid desert specialists such as the Sandhill dunnart are often dependent on winter rainfall and are susceptible to rainfall deficits, wildfires and extreme temperatures caused by anthropogenic climate change (Riley 2020). Future distributions of the Sandhill dunnart, considering the effects of climate change, current fire, and habitat fragmentation, have been modelled by Riley (2020) and Riley, et al. (2021b). Suitable habitat for Sandhill dunnart is predicted to contract significantly towards the more mesic southern parts of its range. As a sandy desert specialist, the distribution of the species is also geographically limited at its southern bounds due to the cessation of suitable habitats. Consequently, further extension of the range southwards is unlikely without assisted migration (Riley et al. 2021b).
Riley et al. (2021b) note that the WA GVD and YRR populations are at a higher climatic risk, with the Eyre Peninsula population also predicted to contract in range as a result of climate change.
The decline in Sandhill dunnart populations appears to be occurring from north to south, tracking the effects of climate change (Riley 2020, Riley et al. 2021b). Unlike other arid zone mammal species where decline and extinction has occurred from south to north following the displacement of First Nations peoples (who were managing the land to prevent fire), the Sandhill dunnart’s contraction aligns with increasing aridity and warming (Riley et al. 2021b). The species reliance on dense, semi-arid habitats makes it highly sensitive to climate driven habitat shifts, as increasing temperature and limited rainfall are likely to degrade the refugial environments on which it depends (Riley et al. 2021b).
Riley (2020) and J. Turpin (pers. comm.) have observed that extreme heat events have caused mass vegetation die-offs in the WA GVD (Figure 4). Rapidly increasing global temperatures, irregular rainfall patterns and more frequent wildfires and droughts are particularly dangerous in the Australian arid zone and threaten Sandhill dunnart habitat. Water is already scarce, vegetation is highly flammable, and the flat arid landscape offers little elevation for species to escape rising temperatures (Pittock 2009, Lindenmayer et al. 2010).
Northern and western habitats are predicted to become increasingly unsuitable for Sandhill dunnart and sympatric desert species over the next 50 years. Regions of the southern central WA (outside of the arid zone) were indicated as climatic refuges by the modelling. Translocations to constructed artificial fenced reserves with transplanted soils, seeds and vegetation from natural arid zone habitats may be required. However further research into their feasibility and value for sandhill dunnarts will be required before this may be considered an option to mitigate the projected loss of distribution due to climate change (Riley 2020).
Photographs showing habitat destruction and degradation from fire, droughts and hot winds in the Great Victoria Desert.
Figure 4: GVD habitat destruction
Note: Large wildfires (left), droughts and hot winds (right) destroyed sensitive Western Australian Great Victoria Desert (WA GVD) habitat in 2019. Extract from Riley 2020.
Further, recent observations of Triodia die-off in Sandhill dunnart habitat in the WA GVD, and the APY Lands of far northern SA (including Wamitjara and Kenmore Park mail road), may be linked to climate change, although this is unclear. If climate change is the cause, then Sandhill dunnart habitat (and other semi-arid species), may be nearing a tipping point. Active investigation of the causes underlying this die-off is warranted.
2.5.7 Native herbivores
All populations
Native herbivores such as kangaroos are also a potential issue in the Middleback dune fields. Euros (Osphranter robustus) graze regenerating Triodia, sometimes preventing recruitment after fire and Western Grey Kangaroos (Macropus fuliginosus) in high density lay on Triodia hummocks, squashing them and presumably limit their value for dunnarts (J. Read pers. comm.). This is not an issue in other areas such as Yellabinna where dingoes control the usually low kangaroo numbers.
Euros, Western Grey Kangaroo and Red Kangaroo (O. rufus) represent a significant threat to Triodia structure and abundance where there are permanent water sources nearby (R. Brandle pers. comm.). Native herbivores may also be partially responsible for the impacts summarised in Section: 2.5.5 Introduced herbivores in areas where Dingoes (Canis lupus dingo) are persecuted (inside the dog fence in SA). Waterpoint management and control of numbers in fenced areas are key to managing these impacts.
Table 2: Risk matrix for the Sandhill dunnart
Consequence (x) // Likelihood (y)
Not significant
Minor
Moderate
Major
Catastrophic
Almost certain
Very-high riskInappropriate fire regimesHabitat loss and fragmentation
Likely
Moderate riskNative herbivores
High riskIntroduced herbivoresClimate change
Very-high riskPredation by cats and foxesIntroduced flora
Possible
Unlikely
Unknown
Risk matrix key:
Low risk
Moderate risk
High risk
Very-high risk
Definition of likelihood and consequence categories
Consequence is a measure of the severity of the impact of the threat on the survival of the species.
Catastrophic – affecting survival, reproduction or essential movement of individuals in >80% of the population or across >80% of the distribution within the next 3 generations or 10 years, whichever is longer.
Major – affecting survival, reproduction or essential movement of individuals in 50-80 % of the population or across 50-80% of the distribution within the next 3 generations or 10 years, whichever is longer.
Moderate – affecting survival, reproduction or essential movement of individuals in 20-50% of the population or across 20-50% of the distribution within the next 3 generations or 10 years, whichever is longer.
Minor – affecting survival, reproduction or essential movement of individuals in 5-30% of the population or across 5-30% of the distribution within the next 3 generations or 10 years, whichever is longer.
Not significant – affecting survival, reproduction or essential movement of individuals in <5% of the population or across <5% of the distribution within the next 3 generations or 10 years, whichever is longer.
Likelihood is a measure of how regularly the threat impacts the species.
Almost certain – >90% chance that threat will have an impact on the species within the next 3 generations or 10 years, whichever is longer.
Likely – 66-90% chance that threat will have an impact on the species within the next 3 generations or 10 years, whichever is longer.
Possible – 33-66% chance that threat will have an impact on the species within the next 3 generations or 10 years, whichever is longer.
Unlikely – <33% chance that threat will have an impact on the species within the next 3 generations or 10 years, whichever is longer.
Unknown – probability that threat will have an impact on the species within the next 3 generations or 10 years, whichever is longer, could be 0-100%.
3 Recovery planning
3.1 Vision
Sandhill dunnart populations are healthy and stable and have expanded across their potential range. The quality and connectedness of habitat is being improved by regional communities and land managers who have adopted management practices to support the long-term sustainability of the species so that it becomes eligible for removal from the EPBC Act List of Threatened Species.
3.2 Objectives
The Objectives in place to enact the Vision for the Sandhill dunnart are as follows:
Threats to Sandhill dunnart are appropriately managed reducing their impact to Sandhill dunnart sub-populations and habitat;
Sandhill dunnart habitat is rehabilitated, connected and protected;
Collective understanding of Sandhill dunnart populations, habitat, ecology and threats has improved; and
Conservation of Sandhill dunnart and their habitat is led and championed by Traditional Custodians, community, experts and governments.
Part 2 - Environment Protection and Biodiversity Conservation (National Recovery Plan for the Sandhill dunnart (Sminthopsis psammophila)) Instrument 2026 - Species information | Zoe