Who’s been lost from the landscape? Identifying missing terrestrial fauna to inform urban rewilding

Introduction

Since the Late Pleistocene, there has been a period of unprecedented, rapid species loss, proclaimed the sixth mass extinction (Wake and Vredenburg 2008). Both mammals and reptiles are impacted by this extinction crises, with the former particularly hard hit (Davis et al. 2018). Since 1900, at least 53 squamates are estimated to have been extirpated globally, and 350 mammal species are estimated to have been extirpated since the late Pleistocene (Alroy 2015; Davis et al. 2018; Andermann et al. 2020). For mammals, this equates to the loss of 2.5 billion years of evolutionary history (Davis et al. 2018). Biodiversity loss has left many ecosystems profoundly altered (Soulé and Noss 1998) because species diversity regulates many ecosystem-level processes. For example, reductions in biodiversity have been linked to lowered ecosystem productivity (e.g. through trophic cascades), altered habitats (e.g. through the loss of key ecosystem engineers, such as beavers, resulting in changes to habitat formation), and lowered efficiency by which communities capture resources (e.g. nutrients, water, light, prey) (Cardinale et al. 2012). Ongoing biodiversity declines are predicted with 25% of mammals and 21% of reptiles categorised as threatened (Cox et al. 2022). Practitioners are responding to this emergency with a multitude of approaches spanning conservation biology and reintroduction biology, to restoration ecology. As biodiversity continues a downward spiral, exploring new conservation approaches and frontiers remains critical (Murphy and van Leeuwen 2021).

Rewilding is a potential conservation response to the biodiversity crisis (Jepson 2019). Initially characterised by the three Cs (i.e. cores, corridors, and carnivores), rewilding was positioned as a response to the emergence of the principles of island biogeography and the observation that the current extinction crisis had left ecosystems significantly altered and lacking in large keystone species (Soulé and Noss 1998). Since then, the term rewilding has quickly evolved. It has branched into use across a diversity of globally implemented conservation projects and experienced rapid growth in use, as evidenced by rising numbers of publications citing ‘rewilding’ (Pettorelli et al. 2019). Broadly, rewilding aims to restore ecological processes and functions, either passively or actively via human-initiated fauna translocations (Carver et al. 2021). Rewilding projects have predominantly occurred in large landscapes, in settings that remain relatively ‘wild’, or in human-altered areas that have been abandoned, left to undergo what has been termed ‘passive rewilding’ (Pettorelli et al. 2019; Sweeney et al. 2019). More recently, however, there has been growing interest in pursuing rewilding projects in people-dominated landscapes including cities (Van Heezik and Seddon 2018; Maller et al. 2019).

Although urbanisation has significantly disrupted the natural environment (McKinney 2002), rewilding urban areas via faunal reintroduction potentially presents unique opportunities for both biodiversity and humans. Arguably, cities could be good places for species protection as they have been found to support a diversity of both endemic and nonnative species (Lepczyk et al. 2017). Moreover, rewilding species into formerly occupied remnant patches of urban native vegetation can infill both local- and landscape-scale distribution gaps that may otherwise remain empty. This is particularly relevant for dispersal limited species due to challenges presented by a hostile urban landscape that may prevent them from naturally recolonising (Watson and Watson 2015; Leo et al. 2022). Restoring species to urban patches can also restore declining ecological processes such as pollination, pest control, and fire suppression (Seddon et al. 2014; Van Heezik and Seddon 2018; O’Rourke et al. 2020; Ryan et al. 2020). The more active rewilding approach of reintroducing historically present species likely presents a safer starting point to restore ecological function than other rewilding practices, such as reintroduction of surrogates, with lower risk of undesirable outcomes and mismatches with the local environment (IUCN/SSC 2013). Furthermore, exposure to nature has been linked to improvements in human well-being, both physical and psychological, and can also shift people towards pro-environmental behaviours (Berto 2014; Mackay and Schmitt 2019). Rewilding projects in urban areas thus offer opportunities to engage and educate local communities, and the possibility of nature exposure through hands-on conservation experiences (Parker 2008). Yet, only 38 animal translocations have been documented in urban areas worldwide (Brown et al. 2024), and within the IUCN Reintroduction Perspectives series (2010–2016), only 6 of 181 animal reintroductions occurred in urban areas (Van Heezik and Seddon 2018), highlighting a gap in rewilding efforts globally.

Rewilding projects in Australia echo these global trends. This continent has not been immune to the extinction crisis. At least 61 endemic species have gone extinct since European colonisation of Australia (10 invertebrates, 1 fish, 4 frogs, 3 reptiles, 9 birds, and 34 mammals) (Woinarski et al. 2019). Several traits have been identified and associated with these declines and extinctions, including body size (e.g. critical weight range (35–5500 g) mammals), ground-dwelling habits, and diet (e.g. carnivores) (Lunney et al. 1997). Mammals have been hardest hit, with their losses attributed to habitat loss and predation by introduced feral cats (Felis catus) and red foxes (Vulpes vulpes) (Woinarski et al. 2015). As such, rewilding initiatives in Australia have prioritised the conservation translocation of rare or threatened mammals into predator-free safe havens, including offshore islands and remote fenced areas (Sweeney et al. 2019). Only a handful of rewilding projects have been initiated in, or near to, urban settlements (Ryan et al. 2020; Dickman and La Marca 2021; Leo et al. 2022). For example, in south-eastern Australia, a fenced feral predator-free area has been established in urban western Sydney to reintroduce locally extirpated species, including eastern quoll (Dasyurus viverrinus) and eastern bettong (Bettongia gaimardi gaimardi) (Dickman and La Marca 2021). Three small mammal species, (bush rat Rattus fuscipes, eastern pygmy possum Cercartetus nanus and brown antechinus Antichinus stuartii) have also been reintroduced to a national park on the edge of Sydney, the most populous city in Australia (Leo et al. 2022). With 76% of people inhabiting major cities in Australia (Hill et al. 2021), urban rewilding presents a unique opportunity to restore biodiversity and repair human-nature connections (Maller et al. 2019).

To initiate further rewilding efforts in Australia, including outside of predator-free areas and within peopled landscapes, a key first step is to determine which historically present species are now locally missing (Morris et al. 2017). Species occurrence data is growing, particularly in urban areas. When species records are pooled in databases, they can provide a sizable dataset that can be used to evaluate species’ historical likelihood and persistence (Van Der Ree 2004; Major and Parsons 2010). Indeed, within in a patchy urban context, species or functional groups that may be missing across the entire landscape may present an initial suite of rewilding candidates with the broadest geographical scope for rewilding opportunities. Furthermore, this approach can identify candidates that are not limited by threat status. In Oceania, most conservation efforts have focused on rare and threatened species, yet common native species are fundamental to the structure of ecosystems and have also undergone substantial declines (Gaston and Fuller 2008). In a heavily altered urban landscape, working to ‘keep common species common’, may represent the best mantra for restoring ‘wildness’ (Watson and Watson 2015).

With a paucity of urban rewilding studies, our goal was to identify species that were historically likely to occur but are currently missing across an urban landscape in northern Sydney – generating an initial list of potential urban rewilding candidates – and consider how ecological traits influence their response to urbanisation. This is a critical step in the development of impactful and successful rewilding strategies in an urban setting (Hahs et al. 2023). To address this goal, we sought to answer the following questions: (1) which non-volant mammal and reptile species were likely to be present historically (i.e. pre-2000 and/or pre-European colonisation) in the landscape-scale study area of northern Sydney?; (2) which of these historically present species might be considered missing or declined from the study area today?; and (3) do missing and declined species share size, diet, habit (i.e. dominant foraging strata) and/or habitat traits? Our target landscape has remnant native vegetation within the urban matrix that has undergone vegetation restoration and feral predator (red fox) management over the past two decades, increasing the potential suitability for reintroductions of missing common native fauna species. To our knowledge, there has been no recent singular study spanning this urban landscape that systematically evaluates which historically present native fauna species are persisting and which may now be considered locally missing. We use the results to explore options for restoring species assemblages and ecological processes that may enhance or support local native biodiversity in northern Sydney and urban areas more broadly.

Materials and methods

Study area

The study area, situated north of the central business district (CBD) of Sydney, Australia, comprises seven Local Government Areas (LGAs) (Fig. 1). Covering ~110,000 ha, it stretches from Sydney Harbour in the south to the Hawkesbury River in the north, west to Lane Cove River and an arbitrary boundary along the Great Northern Road. The area is characterised by a strong urbanisation gradient moving from the south, with highest population densities around coastal areas, to peri-urban outskirts fringed by National Parks in the north (see Supplementary material Fig. S1). The study area contains 11 National Parks protecting ~44,000 ha of native vegetation, with a further ~500 council managed vegetation patches, ranging in size from <1 ha to ~300 ha, dispersed throughout the urban matrix. The National Parks in the study area were retained for this study as they are considered ‘urban impacted’ due to the relative ease of human access and high visitation compared with other protected areas (four of the parks were in the ‘top eight most visited parks in NSW’ in 2022 (DPE 2023)). The area is predominantly situated on Hawkesbury Sandstone, with some areas overlayed by shales, such as Narrabeen shales (Conroy et al. 2022). This geology contributes to the diversity of vegetation communities dominated by Sydney Coastal Dry Sclerophyll Forests, Sydney Coastal Heaths and North Coast Wet Sclerophyll Forests (DCCEEW 2016).

Fig. 1.

Study area (green) with boundaries and names of local government areas. (Satellite image base layer: Google Maps. 20 November 2024).

Locally missing species decision framework

A decision-making framework was developed to categorise native species’ records obtained from ALA into four categories: (1) extant; (2) extant, data deficient; (3) extant, declined; and (4) locally missing (Fig. 2). These categories were then combined into two status groups: (1) extant (extant and extant, data deficient); and (2) rewilding candidates (extant, declined and locally missing). Species in the latter status group were considered potential rewilding candidate species as they represent historically present species of relatively higher conservation value where they are currently either likely to be absent or declined. Prior to evaluating a species’ category, each species’ ALA records were organised by time period. ALA records dated prior to 2000 (i.e. 1788–1999) were assigned to be ‘historically present’ and those dated post 2000 (i.e. 2000–2023) were categorised as ‘currently present’ (Van Der Ree 2004).

Fig. 2.

Decision framework applied to occurrence records to assign species into one of four status categories: (1) extant; (2) extant, data deficient; (3) extant, declined; or (4) locally missing.

Status categories

Extant; and extant, data deficient

In endeavouring to determine which species are extant, several factors influence the level of confidence we might have in making such an assertion. Using data associated with ALA records, extant species (i.e. ‘historically present’ and ‘currently present’) were reviewed giving consideration to: (1) total number of records; (2) year of most recent record; and (3) geographic spread (Fig. 3; Supplementary material Table S1). With large variability in the number of occurrence records within and between our taxonomic groups, we used each taxonomic group’s lower inter-quartile range (IQR1) value of the number of records to represent how many records was considered ‘low’, as IQR better accounts for the outliers in record values within each taxonomic group (Wauchope-Drumm et al. 2020). Species were considered to have a low numbers of records since 2000 if they had fewer than the lower inter-quartile range value number of all records: mammals: IQR1 = 9; reptiles: IQR1 = 44. To assess geographic spread, the occurrence records were mapped in QGIS and visually examined. Species with records spread out across the study area were generally considered ‘well dispersed’. If a species’ records were clumped or missing from multiple LGAs, National Parks and larger Council managed reserves (i.e. >100 ha such as, Manly Dam, St. Ives Wildflower Garden) these were considered to have geographic gaps. Species that had some, or all, of these characteristics remained classified as ‘extant’ but noted as ‘extant, data deficient’ (see Supplementary material Table S1). For instance, we would have less confidence in the extant status of a species that has only a few, old records, that are spatially close together (i.e. clumped) (Shea 2010). In these cases, additional survey work/investigations may be worthwhile to validate their status.

Fig. 3.

The combinations of relationships between the factors (i.e. number of records, age of record, geographic distribution) can influence the confidence we have in assigning a species to be ‘extant’. For example, if a species had a low number of records, that were also old and geographically clumped we would have low confidence in their being extant, and thus we categorised them as ‘extant, data deficient’.

Addition

Species additions were those species that had some records in the period since 2000, but nil records historically, and were assessed as not being historically present using published accounts per above. However, if the species’ current known or predicted range per field guides or literature did not coincide with the study area, it was categorised as ‘never present, probable error’ and removed from further analysis (Menkhorst and Knight 2011; Wilson and Swan 2017).

Rewilding candidates

Extant, declined

Species categorised as extant (per above) that had fewer occurrence records in the current time period since 2000 than in the historical time period were classified as ‘extant, declined’.

Locally missing

Species classified as ‘historically present’ but had no records since 2000 and distributions that overlapped the study area were categorised as ‘Locally missing’. Species with low numbers of records since 2000 (i.e. possibility of being extant), were further scrutinised in the context of two factors: (1) individual records examined; and (2) published accounts evaluating the status of fauna in the study area and/or greater Sydney region (see literature listed above within ‘Historically present’ methods). Where records were dubious and/or literature identified a species as locally (or regionally) extinct, it was reclassified from ‘extant’ to ‘locally missing’. In these instances, the rationale and citation underpinning such a decision is in Supplementary material Tables S2 and S3.

Results

Insights from ALA occurrence records

There were 160,680 occurrence records of non-volant mammals (hereafter referred to as ‘mammals’) and 21,359 occurrence records of terrestrial reptiles (hereafter referred to as ‘reptiles’) across time (1788–2023) within the study area. The density of occurrence records was higher in areas with higher urban density (Supplementary material Fig S1).

Within the period 1788–2023, the ALA dataset identified 41 native mammal species in the study area. Six of these species were deemed unlikely to ever have been present in the study area: (1) long-tailed planigale (Planigale ingrami); (2) mountain pygmy possum (Burramys parvus); (3) northern brown bandicoot (Isoodon macrourus); (4) woylie (Bettongia pencillata); (5) red kangaroo (Osphranter rufus); and (6) mountain brushtail possum (Trichosurus cunninghami). These species either do not have suitable habitat (e.g. mountain pygmy possum occurs in alpine regions) and/or alternative sources indicated that their historical distributions (at the time of European settlement) were situated outside the study area (Recher et al. 1993; Burbidge et al. 2008; Yeatman and Groom 2012). Most remaining mammal species were ground-dwelling (70%) and critical weight range (i.e. 35–5500 g) (61%). The number of records for each species was highly variable, ranging from 71,493 records (swamp wallaby, Wallabia bicolor) to a single record (e.g. parma wallaby, Notamacropus parma), with an average of 4589 (IQR: 9–551) occurrence records (Supplementary material Table S2).

Within the same period, the ALA dataset included 92 native reptile species in the study area; however, 45 species were deemed unlikely to ever have been present in the study area. Alternative sources for these 45 species indicated that their historical distributions were situated outside the study area (Cogger 2010; Shea 2010) and current range maps did not overlap the study area (Wilson and Swan 2017). The remaining 47 species in this taxon had lower numbers of records per species, ranging from 3798 (eastern water dragon, Itellagama lesueruii) to a single record (mainland she-oak skink, Cyclodomorphus michaeli), with an average of 392 records (IQR: 44–273) (SM 4). Most were diurnal (n = 32) and in the Scincidae family (n = 18).

A temporal distribution of the records shows an increasing number of records for both taxa, highly biased towards the decades since 2000 (Fig. 4). There have been sizable increases in the volume of occurrence data, with 98% and 88% of mammal and reptile records respectively generated since 2000. In an urban landscape, this large growth in records does not necessarily imply changes to species’ abundances but is likely linked to increased sightings due to rising human population, environmental education, legislative obligations linked to development e.g. environmental impact statements and technological advances (i.e. smartphones, citizen science data collection applications).

Fig. 4.

Temporal distribution of ALA occurrence records. Graphs were split across time to better highlight temporal changes as the number of records was orders of magnitude higher post-2000 than the preceding period (1788–1999).

Species status

Most mammals and reptiles historically thought to be present within the study area remain extant (62% and 93% respectively). Nevertheless, both groups appear to have experienced declines in richness. Of those likely to be persisting, nearly 40% were either data deficient or potentially declining. Mammals were found have a higher number of species that could be confirmed as locally missing, likely due to additional lines of evidence confirming their local extirpation. In contrast, reptiles had higher numbers of species ‘extant, declined’. For both groups, when these categories were combined, they represented 51% and 32% of mammal and reptile species respectively (Table 2). There were no native species additions identified.

Table 2.Percentage of historically present species by status.

Status	Mammals	Reptiles
Extant	35.9%	57.5%
Extant, data deficient	12.8%	10.6%
Extant, declined	12.8%	25.5%
Locally missing	38.4%	6.4%

Within the study area, mammals have potentially lost representation of two entire families: (1) Canidae; and (2) Potoridae, with species missing or in decline within an additional six families. The Macropodidae family was found to have the highest numbers of locally missing species (Fig. 5), with the study area losing 75% of historically present species. For reptiles, all families continue to have representation. Nevertheless, seven families have species missing or in likely decline (Fig. 5).

Fig. 5.

Species status categories by taxonomic family: (a) non-volant mammals and (b) reptiles. Species silhouettes highlight taxonomic groups with potential landscape rewilding candidates.

Ecological traits

Of the 86 species in this study, 51 were grouped as extant and 35 as rewilding candidates (locally missing/extant, declined). Associations between individual ecological traits and status groups (i.e. extant and rewilding candidates) were significant for two traits: habit (Fisher’s Exact Test: P-value = 0.047) and diet breadth (Fisher’s Exact Test: P-value = 0.024) (Fig. 6). Most rewilding candidate species had ground dwelling habits (29 of 35 species), with only 8 species in other habit categories as compared to 30 extant species (arboreal, 16; aquatic, 7; saxicolous, 4; fossorial, 3). Rewilding candidates were dominated by species with 1 diet category (22 of 35 species; predominant categories were invertebrates (n = 10) or other plants (n = 8)), with none having more than 3 diet categories.

Fig. 6.

Ecological traits for two species’ status categories: extant (light green) vs rewilding candidates (i.e. locally missing/extant, declined) (dark green). Shaded charts are those where status category is influenced by ecological attribute (Fisher’s Exact Test: P-value <0.05).

There was an overall difference in ecological traits between the two status groups (Global PERMANOVA; R² = 0.02961, P = 0.012). Visual inspection of the nMDS plot indicated mild separation of extant and rewilding candidate species, particularly evident in a group of nMDS plot points of macropods and arboreal mammal rewilding candidate species situated close together, yet further apart from extant species points (Fig. 7) (stress value = 0.23).

Fig. 7.

Ordination produced by nMDS analysis of ecological traits of extant and rewilding candidates. PERMANOVA (R² = 0.02961, P = 0.012).

Discussion

Using occurrence records combined with a decision framework, 15 terrestrial mammals and 3 reptile species were identified as historically present but now locally missing in northern Sydney, representing a likely loss of 38% of mammals and 6% of reptiles. Seventeen additional species were found to be potentially in decline. Together, these 35 species are a first cut at potential urban rewilding candidates. Missing species included multiple families; notably, two entire mammal families (Potoridae and Canidae). Other impacted taxa included Macropodidae, elapids, skinks, and geckoes. These findings align with broader patterns of loss and decline in these taxonomic groups across Greater Sydney (DECC 2007; Shea 2010). Landscape-scale drivers of species’ declines are varied and often multi-faceted (Cox et al. 2022). At a species scale, however, a potential driver of decline is ecological traits. Here, we found broad ecological traits influence a species’ status (locally missing/declined (i.e. rewilding candidates) versus extant species) (Fig. 7). Specifically, status was found to be significantly influenced by diet breadth and habit. Most locally missing or declined species are ground-dwelling species and dietary specialists, reliant upon one diet category (Fig. 6). These findings are consistent with studies evaluating drivers of species’ extinction risk (Lunney et al. 1997; McKenzie et al. 2007; Böhm et al. 2016). These notable traits highlight vulnerabilities that may be faced by species rewilded into urban contexts and if exhibited by a potential rewilding candidate, may be used to inform development of targeted conservation initiatives, which we discuss further below along with other factors that may influence rewilding success.

As well as the importance for a suitable urban rewilding candidate to have been historically present, rewilding literature emphasises a requirement to restore degraded ecosystem functions and processes (Carver et al. 2021), thus a rewilding candidate should, by preference, restore a key ecological function. Our results found that the study area has potentially lost species from taxonomic groups providing important functions. For example, potoroids, such as the long-nosed potoroo and rufous bettong, are bioturbators – crucial ecological engineers (Davies et al. 2019) whose diggings trap organic matter, supporting higher soil fertility and moisture content (Davies et al. 2019). Further, they are mycophgagous mammals, contributing to fungi spore dispersal linked to ecosystem health (Standaloft and Kirkpatrick 2022). Elapids, such as tiger snakes (Notechis scutatus), are also key predators that may help with pest rodent suppression (Shine et al. 2024). Strategically rewilding such species could therefore restore lost or depleted ecosystem processes in urban landscapes.

Persistence of restored ecosystem function requires that species rewilded are capable of becoming self-sustaining, viable populations (IUCN/SSC 2013; Carver et al. 2021). In urbanised places, fragmented remnant vegetation means landscape factors such as patch size and connectivity significantly influence species’ population viability. Thus, species movement traits (e.g. home range (HR) and dispersal needs) will form key considerations for urban rewilding candidate selection. Indeed, within the study area’s urbanised core (<25 km from CBD) patch sizes range from <5–350 ha, with some connected to large national parks on the urban fringe. Thus, the ability to occupy smaller areas of suitable habitat may improve rewilding feasibility for certain species. For instance, the dingo, a locally missing apex predator, requires large areas with a HR of 1700 ha (urban) up to 41,400 ha (arid monsoonal), moving over 6 km/day in urban settings (McNeill et al. 2016). In contrast, the brush-tailed phascogale, a smaller-sized, locally missing species, is a marsupial carnivore typically associated with dry sclerophyll forests. Formerly widespread in eastern and south-western Australia it is now present in less than 50% of its former range, including on the urban fringes of Perth, Western Australia (DBCA 2024) and Melbourne, Victoria (Nillumbik Shire Council 2025). This species has recorded HRs of between 2.3–8 ha and 20–70 ha, foraging over a range of 4–5 ha (Strahan 1991; Van Der Ree et al. 2001) potentially making them suited to smaller urban patch sizes.

Other locally missing species identified with small ‘HR to patch size’ ratios and/or lower dispersal requirements include the common dunnart (movement 440–1100 m, mean distance 154 m between activity locations (Paull 2013)), red-necked pademelon (HR: 5–30 ha; (Strahan 1991)), parma wallaby (HR: 5 ha (Lentle et al. 2004)), Burton’s snake-lizard (Lialis burtonis (average movement 5 m per day; (Wall and Shine 2013)), and the copper-tailed skink (HR: unknown, however, though other Ctenotus sp. move 40–60 m (Read 1998)). Several successful rewilding projects in small urban patches targeted species with relatively small HRs (Vieira et al. 2015; Miskelly 2018; Saumure et al. 2021). For example, Eurasian red squirrels (Sciurus vulgaris) rewilded into small urban parks (13 ha, 35 ha) (Vieira et al. 2015) (HR: 2.7–7.5 ha; Wauters et al. 2001); the relict leopard frogs (Rana fisheri) into an urban wetland (73 ha); and north island Kaka (Nestor meridionalis) into a fenced sanctuary (225 ha) (HR: 2–21 ha (Recio et al. 2016)). Globally, rewilding projects have typically been large scale, yet these examples highlight that cities can be suitable for rewilding, particularly where they comprise multiple types of green spaces, such as native vegetation, gardens and road verges that may aid connectivity by reducing the hostility of the urban matrix (Lepczyk et al. 2017). Hence, landscape factors need not constrain rewilding projects. Nevertheless, these factors are likely to have relatively greater influence on rewilding suitability of a species in an urban context than in other landscape settings. As such, they form a fundamental consideration for decision-makers progressing species that will survive when rewilded into an urban context.

Other traits that were examined individually found status (extant/declined/locally missing) to be significantly influenced by diet breadth and habit (Fig. 6). Most locally missing or declined species are ground-dwelling species (n = 29) and dietary specialists (n = 22), reliant upon one diet category. In Australia, ground-dwelling habits have been associated with declines and extinctions of birds and mammals (Lunney et al. 1997). Ecological specialisation is another factor often correlated with vulnerability to decline and extinction (McKenzie et al. 2007). Ecological specialists are thought to be more vulnerable to changes in their environment and availability of resources compared to generalists (Harcourt et al. 2002). Urban-adapted species are often associated with having dietary flexibility (e.g. bush rats, long-nosed bandicoots, eastern blue tongues (Fardell and Dickman 2023)), and thus species with broader diets may have greater urban rewilding success. Rewilding candidates with the highest dietary breadth (n = 3) included eastern quoll, short-eared possum, and brush-tailed phascogale.

Body size is another trait associated with species persistence (Lunney et al. 1997). In Australia, the majority of small mammals (11–5000 g) are threatened with extinction, and there are more threatened reptiles in larger size classes (1–5 kg) than smaller classes (1–10 g) (Lunney et al. 1997). In our urban context, species have been lost or declined across all size categories, albeit fewer than those persisting in all but the largest size category (>5 kg), where more species (n = 6) have been lost/declined than remain extant. Several small mammals (n = 13, Fig. 6) were identified as potential urban rewilding candidates, including the eastern chestnut mouse (Pseudomys gracilicaudatus) and long-nosed potoroo. The eastern chestnut mouse is associated with four broad habitat types. However, these mainly occur in dense, wet heathland, experiencing peak densities in recently burnt heath (Pereoglou et al. 2011). The long-nosed potoroo is associated with three broad habitat types, however, vegetation communities dominated by species known to host hypogeous ectomycorrhizal associations (e.g. eucalypts, acacias, allocasuarinas, casuarinas, Leptospermum) are potentially favoured due to the species’ dietary intake of hypogeal fungi (Claridge et al. 1993).

The key threatening processes of predation by red foxes and cats and habitat loss/fragmentation threaten the persistence of small vertebrate species (McKenzie et al. 2007; Doherty et al. 2016). These feral predators, omnipresent in urban areas (Campbell et al. 2020), may pose a threat to establishment and persistence of both small and ground-dwelling (n = 29) rewilding candidates. In Australia, these feral predators are managed via exclusion fencing, trapping, poison baiting and shooting, not all of which are viable in urban settings. Thus, urban rewilding offers an opportunity to explore alternative predation risk mitigation techniques. Vegetation structure has been found to mediate predation risk in a variety of ecosystems (Braun et al. 2024), and there is interest in the use of artificial structures to provide refuge from predation (Cowan et al. 2021; Carthey 2025). Moreover, olfactory misinformation has also been posited as a potential tool to aid species reintroduction efforts in the face of invasive (or native) predators (Price and Banks 2012; Finnerty et al. 2022). Reptile rewilding candidates, including broad-headed snake, red-throated skink, wood gecko, and Lesueur’s velvet gecko, may be declining due to habitat fragmentation and loss, such as sandstone outcrops, crevices, or historical removal of bush rock for gardens (Shea 2010). However, sandstone habitats persist in the study area, and targeted habitat suitability surveys could assess the viability of this habitat for rewilding, or confirm localised distributions. Habitat restoration e.g. rock piles, is also relevant here to support these species’ restoration (Herbert et al. 2023). Ongoing threats, particularly where mitigation is unavailable, challenge rewilding feasibility, especially for at-risk species (IUCN/SSC 2013). Thus, decision makers might prioritise urban rewilding candidates that are more common and whose threats can be mitigated.

Vulnerability to extinction based upon threat status is often used as a criterion to prioritise species for conservation (Mace et al. 2007). Yet, urban rewilding candidates here spanned from common (unlisted) to highly threatened species. Of the 35 potential rewilding candidates, the majority (60%) were common species not listed as threatened under Australian legislation. Common species have not been immune to human impacts of urbanisation, with many populations in decline (Gaston and Fuller 2008; Finn et al. 2023). However, the importance of common species in ecosystems is often under-represented, as many provide crucial ecosystem structure and function roles, such as food web stabilisation, seed dispersal and pollination (Gaston and Fuller 2008; Lindenmayer et al. 2011). Notable common species identified include the common dunnart, yellow-footed antechinus and three gecko species (thick-tailed gecko, Lesueur’s velvet gecko, and wood gecko; noted above). These species represent taxonomic groups broadly lost from our study area and with them a potential erosion of ecosystem functions. For example, gecko species are viewed by some communities as playing an important ecosystem role because their diet includes mosquitoes and invertebrates (Ceríaco et al. 2011). Common dunnarts and yellow-footed antechinus are carnivorous marsupials with a wide range of habitat associations that include dry sclerophyll forests that still dominate the study area. These marsupials have small HRs (e.g. yellow-footed antechinus: 0.7–1.2 ha; Kelly 2006), which may improve the species’ viability in relatively small urban patches. They were also recently found to display dietary diversity for fungi, important for maintaining natural variation in fungal community composition (Nest et al. 2023). Furthermore, these small, relatively cryptic species pose no known serious threat to humans, making them potentially more appealing to communities with whom they will be sharing the broader ‘urban backyard’.

When rewilding in an urban context, social licence is another essential consideration to ensure a rewilding project’s success. Undertaking rewilding in human-dominated landscapes presents opportunities to engage communities and address the ‘nature deficit disorder’ associated with people’s increasing disconnection from nature (Maller et al. 2019). Yet social benefits may be eroded if social acceptability and engagement factors were ignored. For example, rewilding candidates that present human–wildlife conflict, perceived or otherwise, may spark community opposition and lost community engagement opportunities. This study identified several predators and rodents as locally missing from the landscape (e.g. dingoes, eastern, and spotted-tailed quolls). Concerns generated by these taxa include zoonotic disease transmission, predation on domestic pets/poultry, loss of amenity or fouling of recreational areas, and risks to human safety and psychological wellbeing, including elevated anxiety/fear of attacks (McNeill et al. 2016). As such, these species may not represent ideal candidates to prioritise in an urban rewilding context. Nevertheless, if such species were considered essential to meeting an urban rewilding project’s goal, then carefully constructed community engagement and educational programs that bring the community along the urban rewilding journey, an important aspect of any such project, will be even more critical in this scenario.

Our desktop approach efficiently used an existing database to identify potential rewilding candidates. Yet databases of occurrence records present opportunities and challenges for evaluating species presence and temporal distribution changes. With over 180,000 records for the study area, the ALA provided a sizeable dataset from which to begin evaluations of a species’ ongoing presence. The dataset combines diverse sources of occurrence sightings providing a breadth of data unattainable from a single survey, yet may include inherent biases, such as observer identification error. Examination of records and incorporation of additional evidence can mitigate biases. For instance, reviewing species distribution maps revealed species unlikely to have been present, particularly reptiles (n = 45). We also pursued a landscape approach for our analysis, yet the scale where a similar analysis is undertaken will be project dependent. For example, if only a single patch was available, then an analysis of species conducted at the patch level may be more suitable and would likely generate a different number and suite of missing species. This scale dependency was evident when analysing missing species from our largest urban reserve (Manly Dam), where an additional seven mammals and four reptiles had no ALA records in this reserve, presenting potential rewilding candidates worthy of consideration for this patch (Supplementary material Table S5).

Rewilding urban landscapes is an emerging area for conservation projects that diverges from current rewilding practices taking place in rural or remote areas, predominantly because of a higher degree of landscape fragmentation and activities of people (Maller et al. 2019). With the majority of people globally living in urban contexts, it presents an approach that may respond to the dual rising concerns of declining biodiversity and health and wellbeing of people living in cities (Maller et al. 2019). A critical first step of rewilding is identifying historically present, locally extinct species. Our approach leveraged occurrence records to pinpoint a suite of rewilding candidate species that were historically present, but currently missing or declined in a landscape, and assessed their ecological traits. This study provided greater awareness of trait generated vulnerabilities, current taxonomic gaps, and associated ecosystem functions that may be missing from our urban study area, irrespective of a species’ threat status. In an urban context, where the list of species lost might be high, there are several additional factors that require consideration to aid rewilding candidate prioritisation in a resource constrained environment, along with highlighting needs for follow-up investigations on the ground. Species that best suit these factors, such as the examples highlighted above, are likely to have the greatest chance of urban rewilding success. Urban ecosystems, often viewed as degraded and beyond repair, continue to support high concentrations of biodiversity, even despite species losses (Ives et al. 2016). Several rewilding paradigms seek re-integration of nature back into human places (Pettorelli et al. 2019), positioning it as an ideal response to biodiversity losses in urban landscapes, viewing them instead as conservation openings. Through rebuilding populations of locally missing species, urban rewilding can fortify resilience of native species’ and begin shifting our urban ecosystems along the degradation gradient towards a more biodiverse and functional state.

Supplementary material

Supplementary material is available online.