Introduction

Density estimates are used to infer species ecology and guide biodiversity conservation and management. Estimating density, however, relies on successfully detecting animals, which is difficult for cryptic or rare species (Garrard et al. , 2015). Probability of detection can vary due to differences in survey method and effort, behaviour, habitat, season and weather. Only a fraction of a population may be detected during a survey, leading to a high risk of type two errors (i.e., when non-detection at an occupied site is mistaken for evidence of absence) (Guillera-Arroita, 2017, Kellner & Swihart, 2014) . Methods to account for imperfect detectability include those that do not require uniquely identifiable individuals (e.g., distance sampling and occupancy modelling) (Buckland, 2001, MacKenzie et al. , 2018) and those that require some fraction of the population to be uniquely identifiable (e.g., capture-recapture and spatially explicit capture-recapture) (Efford, 2004, Otis et al. , 1978, Pollock et al. , 1990). However, even when accounting for imperfect detection, cryptic and rare species generally have lower detection rates than common or conspicuous ones (Dettmers et al. , 1999). Survey designs for wildlife must therefore simultaneously account for imperfect detection, whilst striving to optimise the underlying detection rates. Incorporating such considerations can improve the accuracy and precision of density estimates, enabling better-informed conservation management decisions.
Spatially explicit capture-recapture (SECR) modelling using photographic identification is a popular technique for estimating densities of species with distinctive individual markings (Green, Chynoweth & Şekercioğlu, 2020). This method has been used with spotted or striped cats (Green et al. , 2020) and distinctive small Australian marsupials, including the sugar glider Petaurus notatus(Gracanin, Minchinton & Mikac, 2022) and numbat Myrmecobius fasciatus (Thorn et al. , 2022). Unlike traditional capture-recapture, SECR considers spatial variation in detection probability and records animal capture and recapture locations. This method enables estimation of population density across detector grids, and informs capture probability and home range estimates (Efford, Borchers & Byrom, 2009). To obtain a robust estimate of population density with SECR, repeated observations of numerous individuals at multiple detectors are required.
Arboreal nocturnal animals pose unusual challenges for survey design. They are cryptic, occupy challenging terrain and forest strata, and have variable detectability with traditional survey techniques (Catling, Burt & Kooyman, 1997, Davey, 1990). Forest dependent fauna are often of high conservation importance as forests globally are reduced and disturbed (Lindenmayer, 2023, Potapov et al. , 2022), resulting in the removal of mature habitat features and connectivity that such species need to persist. Consequently, there is a pressing need for reliable techniques that can be used to estimate population density for species which are of high ecological significance or threatened. Despite the biological and conservation importance of density information, few species receive targeted innovation in developing detection methods. For instance, most surveys of threatened arboreal marsupials that rely on live capture techniques use highly generalized lures (Commonwealth of Australia, 2011). In instances of detailed research (Austin et al. , 2017, Diete et al. , 2016, Morgan, 1990, Wayne et al. , 2005), substantial improvements can be made to tailor attractiveness of baits to target species, which can dramatically improve the probability of detection. Developing customized detection solutions for a target species can therefore be the difference between meaningful ecological inference and a failed study.
We consider a case study of an unusual ecological problem and aim to develop an effective approach for estimating density of an arboreal nocturnal marsupial. The sugar glider, Petaurus notatus (synonymP. breviceps, inland sugar glider, Krefft’s glider), is widespread in eastern Australia (Cremona et al. , 2021, Goldingayet al. , 2023), and was introduced to Tasmania in the 1830’s (Campbell et al. , 2018). Recent research has shown that sugar gliders are unexpected major predators of nesting birds in Tasmania, including the critically endangered swift parrot Lathamus discolor (Stojanovic et al. , 2018, Stojanovic et al. , 2014). There is an urgent need to understand the scale and severity of the predation threat to swift parrots but basic information about Tasmanian sugar gliders remains unknown.
Here, we aim to overcome this conservation problem regarding sugar gliders in three steps. First, we review the literature and evaluate the range and effectiveness of methods available for trapping sugar gliders. Next, we undertake a field study to evaluate approaches for optimizing detection. Detection probability can be influenced by survey method, behaviour, habitat, season and weather. Low detection probability can result in wrongly concluding the target species is absent and cause high uncertainty in resulting estimates of density. Cryptic species like the sugar glider generally have lower detection rates than common or conspicuous ones and survey designs must therefore aim to increase underlying detection rates while accounting for imperfect detection. Incorporating such considerations can improve the accuracy and precision of density estimates, enabling better-informed conservation management decisions. We do this by undertaking a camera trapping study and use SECR analysis to determine which types of bait influence detectability. Finally, we evaluate density of sugar gliders using SECR at a swift parrot important breeding area. Density estimates are crucial to infer species ecology and guide biodiversity conservation and management. We discuss our results in the context of conservation needs of swift parrots and research approaches for sugar gliders.