Range-Wide Camera Trapping for the Australian Cassowary Reveals Habitat Associations With Rainfall and Forest Quality

Abstract

The Australian Wet Tropics rainforests are a biodiversity hotspot covering just 0.2% of the continent's land area. However, historic forest loss, modern fragmentation, and climate change continue to threaten these ecosystems. Southern cassowaries (Casuarius casuarius) are large flightless birds restricted to closed-canopy tropical forests in Australia. Cassowaries are obligate frugivores whose dispersal of large-seeded plants is considered a keystone species interaction supporting forest regeneration. We conducted camera trapping across cassowaries' Australian range and quantified habitat associations using hierarchical models that account for imperfect detection. Cassowary detections were significantly higher in rainforests compared to adjacent wet sclerophyll closed-canopy forests, confirming their status as habitat specialists. Cassowaries' relative abundance (λ in Royle-Nichols modelling) declined with forest degradation and rainfall but was not strongly affected by human footprint or elevation. This aligns with observations of them occasionally foraging on anthropogenic food sources at the edges of large intact forests (e.g., where there are human-planted fruit trees). These findings provide the ecological reasons underpinning known cassowary hotspots in large rainforests that are relatively dry. It would be valuable to deepen our understanding of their persistence in degraded rainforests near humans via diet and survival studies, and we caution that their association with rainfall means that they may be impacted by climate change.

Keywords: abiotic factors; anthropogenic disturbances; bird; conservation; detectability; hierarchical modelling; interactive effects; occupancy; species distribution modelling.

FIGURE 1

Camera trap photo of a cassowary in Daintree National Park from a bush camera (as opposed to a road camera).

FIGURE 2

(A) Map showing the study area in Northeast Queensland with bounding boxes indicating sampled landscapes. (B) Inset of the Atherton tablelands where much sampling took place; each point represents a site where two cameras were deployed at paired trail and bush sites. (C) Inset of Wooroonooran National Park, showing spatially resampled camera deployments into 3 km² hexagonal sampling units that are colour‐coded by camera effort in trap nights.

FIGURE 3

The number of cameras with cassowary detections versus non‐detections across three forest types: Rainforest, wet sclerophyll, and dry sclerophyll. Bars represent the total number of cameras deployed in each forest type, categorised by whether cassowaries were detected or not detected. A detailed map of forest type is provided in Figure S1.

FIGURE 4

Predicted cassowary relative abundance (λ in RN hierarchical models) for the lowest‐scoring AIC model. Panels (A) and (B) depict the trend for each covariate while holding the other covariate constant at its mean value. Black tick marks along the x‐axis represent the distribution of the sampled covariate. Panel (C) illustrates the effect of rainfall on cassowary abundance while forest integrity is held at 10% (index value = 7) and 90% (index value = 9) of its maximum value. Shaded areas represent the 95% confidence intervals.

FIGURE 5

(A) Predicted cassowary relative abundance across Northeast Queensland, derived from the Royle–Nichols model, including an interaction between forest integrity and rainfall. (B) Cassowary density and distribution based on sign transects (redrawn from Westcott et al. 2014).