Data from: Exotic megaherbivores as ecosystem engineers in Australian savannas: Do they facilitate predator movement?
Data files
Jul 21, 2025 version files 9.21 KB
-
Neave_2024_gametrail_cam_detections.csv
4.42 KB
-
README.md
4.78 KB
Abstract
An understanding of how terrestrial mammalian predators use their environment is critical for the development of effective management and monitoring. Mammalian predators often use anthropogenic linear features—such as roads, fencelines, and infrastructure corridors—to increase movement efficiency and prey encounter rates. However, there has been little investigation into how predators use more subtle linear features such as game trails (i.e., well‐trodden paths created by megaherbivores). This is despite native and exotic megaherbivores being abundant across many of Earth's most intact landscapes and conservation areas. We investigated how the two largest terrestrial mammalian predators in northern Australian savannas—the dingo (Canis familiaris, introduced ca. 4000 years ago) and cat (Felis catus, introduced ca. 200 years ago)—use game trails created by exotic megaherbivores (Asian water buffalo Bubalus bubalis and horse Equus caballus). We deployed two camera traps at 52 sites, with one camera positioned on a game trail and another in undisturbed vegetation < 60 m away. We compared the activity of predators on game trails to adjacent undisturbed vegetation and explored how trail use varied with vegetation structure and prey activity. Dingoes and cats were 34 times and 6 times more likely to be detected on game trails than in adjacent vegetation, respectively, suggesting these predators preferentially use game trails. We speculate that the extensive network of game trails created by exotic megaherbivores across northern Australia's vast savannas has potentially facilitated terrestrial mammalian predator movement at very large scales. Controlling exotic megaherbivores may, therefore, provide a means of disrupting the activity of dingoes and cats, thereby benefiting predation‐susceptible native species. However, further research is needed to understand the ecological implications of game trails in Australian savannas and other habitat types.
https://doi.org/10.5061/dryad.0zpc86776
Description of the data and file structure
The experimental method involved the deployment of 104 camera traps at 52 sites on Melville Island to compare faunal (both native and invasive spp.) activity along game trails created by exotic megaherbivores versus in undisturbed vegetation in savanna habitats. At each site, paired cameras were positioned: one facing along a game trail and another in undisturbed vegetation within 60 meters. The cameras were unbaited, set at a height of 50 cm, and programmed to capture five consecutive images per trigger.
Detections were considered independent when >30 minutes had elapsed since the previous detection of that species on the same camera. We investigated each species’ use of game trails using logistic generalised linear models. The modelled response variable was the proportion of independent on-trail detections relative to all independent detections recorded at each site. Sites with no detections of species being modelled on either camera were removed from the analysis. Models were fit with a quasibinomial error structure using the VGAM package (Yee, 2015) in R (version 4.3.1, R Core Team, 2023) to account for overdispersion. To examine how biotic variables influenced a species' use of game trails, we analysed all combinations of the explanatory variables outlined in Table S1, with no model interactions as none were deemed ecologically relevant.
Files and variables
File: Neave_2024_gametrail_cam_detections.csv
Description:
Variables
- site: site identity
- buff_off: number of independent buffalo detections off-trail camera
- buff_on: number of independent buffalo detections on-trail camera
- dingo_off: number of independent dingo detections off-trail camera
- dingo_on: number of independent dingo detections on-trail camera
- horse_off: number of independent horse detections off-trail camera
- horse_on: number of independent horse detections on the trail camera
- cat_off: number of independent cat detections off-trail camera
- cat_on: number of independent cat detections on the trail camera
- bandi_off: number of independent northern brown bandicoot detections off-trail camera
- bandi_on: number of independent northern brown bandicoot detections on-trail camera
- wallaby_off: number of independent agile wallaby detections off-trail camera
- wallaby_on: number of independent agile wallaby detections on-trail camera
- treerat_off: number of independent black-footed tree-rat detections off-trail camera
- treerat_on: number of independent black-footed tree-rat detections on-trail camera
- deli_off: number of independent delicate mouse detections off-trail camera
- deli_on: number of independent delicate mouse detections on the trail camera
- tunneyi_off: number of independent pale field-rat detections off-trail camera
- tunneyi_on: number of independent pale field-rat detections on-trail camera
- dunnart_off: number of independent dunnart detections off-trail camera
- dunnart_on: number of independent dunnart detections on-trail camera
- poss_off: number of independent northern brushtail possum detections off-trail camera
- poss_on: number of independent northern brushtail possum detections on-trail camera
- feral_herbivore_off: number of independent horse and buffalo detections off-trail camera
- feral_herbivore_on: number of independent horse and buffalo detections on-trail camera
- all_prey_spp_off: sum of independent detections of all native mammal species <2.5kg off-trail camera
- all_prey_spp_on: sum of independent detections of all native mammal species <2.5kg on-trail camera
- sm_mam_off: sum of dunnart, delicate mouse, and pale field-rat independent detections off-trail camera
- sm_mam_on: sum of dunnart, delicate mous,e and pale field-rat independent detections off-trail camera
- veg_visibility: average visibility height at each site (cm). Vegetation visibility was a proxy for vegetation density and was quantified using a custom vegetation index based on visual obstruction measurements along a transect.
- cam_nights_off: number of nights off-trail the camera was active
- cam_nights_on: number of nights the on-trail camera was active
Code/software
Neave_et_al_glm_qbinomial_code.R : This code fits and compares multiple binomial and quasibinomial logistic regression models to analyze species’ trail-use preferences, using QAIC for model selection while checking for overdispersion.
R version 4.3.1 (2023-06-16 ucrt) -- "Beagle Scouts"
Packages required:
VGAM
car
dplyr
ggplot2
At each site there were two cameras. One camera was placed on a game trail termed 'on' and the other camera was placed approximately 60m away in undisturbed vegetation termed 'off'.
Data csv includes the the number of independent detections of a species on the 'on-trail' camera and 'off-trail' camera. Detections were considered independent when >30 minutes had elapsed since the previous detections of that species on that same camera.
'feral_herbivore' category includes detections of two species: Bubalus bubalis and Equus caballus
'sm_mam' category includes detections of Sminthopsis spp., Pseudomys spp. and Rattus spp.
'all_prey_spp' category is the sum of independent detections of all native mammal species less than 2.5kg i.e. all native mammal species excluding agile wallaby
