The development of invasive rodent monitoring methods has focussed on ground-based deployment, which may not sufficiently capture the activity of arboreal rodents (Nance et al. In review a). To tackle this issue, we present a novel comparison of three traditionally ground-based rodent monitoring tools deployed in arboreal settings. In doing so, we provide evidence that common ground-based tools can be used to monitor arboreal rodents. We demonstrate that camera traps, chew cards and tracking tunnels can all effectively detect arboreal invasive rodents in forested settings at heights of up to 12.3 meters, though detection performance can vary. In particular, despite rodent presence being detected at equal frequency across the three strata, chew cards were superior in detecting the presence of rodents above-ground relative to other methods. Through a direct comparison of methods across strata and forest type, we demonstrate that it is possible to generate a three-dimensional estimate of rodent presence using the same methods traditionally used to monitor activity at ground level.
Our findings suggest that device sensitivity can vary by strata: while the three device types performed equally on the ground, chew cards performed significantly better than other devices when deployed above-ground. To account for strata-specific sensitivity of different methods, our findings suggest that the optimal monitoring design would pair chew cards with camera traps, and omit tracking tunnels. Tunnels were the least effective standalone method (0.6% of station-level detections were derived from tunnels alone), thus could be omitted with negligible impact on the accuracy of a monitoring network. Such a design would also reduce the costs and complexity of implementing the monitoring program. Alternatively, deploying cameras exclusively at the ground level and chew cards across all strata would maintain 96% detection accuracy at significantly reduced costs. Lure-based methods (like chew cards) may however experience reduced sensitivity when local food sources are abundant (Wegmann et al. 2011; Holmes et al. 2015). Therefore, when detectability is paramount, using two or more functionally different methods (e.g. lure-based and non-lure-based) in periods of lower resource availability may result in improved detectability.
It is possible that chew cards were more effective than other methods in above-ground strata due to their functional simplicity and subsequent reduced likelihood of technical failure. We did not identify any chew cards that experienced technical failure, as their lightweight and simple design meant that ground-based chew cards functioned similarly to above-ground chew cards. Conversely, we experienced technical issues with camera traps and tracking tunnels. Like most camera traps, the model we used did not have a view finder, which meant that we could only estimate the accuracy with which we aimed camera traps at above-ground vegetation structures. As such, during image processing we identified a small number of above-ground cameras that did not have traversable vegetation within 1-2 m of the field of vision i.e. vegetation was visible but rodents could only be detected at further distances (3-5 m). We anticipate that this resulted in these cameras being less effective than others at detecting rodents during the observation period. Additionally, in arboreal settings the number of large branches decreases with height from the ground. The structure of mid-storey and canopy vegetation therefore imposes challenges for methods that require secure attachment to large branches, impacting camera traps and tracking tunnels but not chew cards. This challenge resulted in fewer options for, and therefore potentially sub-optimal camera placement, which can lead to reduced detection performance (Moore et al. 2021). Similarly, there were fewer opportunities for optimal placement of tracking tunnels above-ground. While only a single arboreal tunnel was excluded from the dataset due to stability issues, factors such as rodent neophobic responses, perceived or actual stability issues, or a combination of the two may reduce the likelihood of invasive rodents interacting with this device type above-ground (Cooper et al. 2018; Baldwin and Meinerz 2022). Further experimentation with above-ground deployment of camera traps and tracking tunnels is needed to improve detection performance. Until then, we recommend the use of simple methods such as chew cards when monitoring above-ground. Ultimately, chew cards appeared to be superior in both arboreal detection performance and ease of deployment, providing managers with a simple yet powerful tool for the detection of arboreal rodents.
The reduced efficacy of tracking tunnels more generally could also be explained by a learned aversion to entering artificial box-like objects (Allsop et al. 2017). The baiting program associated with the present study utilises standard bait station boxes as well as baited tunnels typically constructed from black PVC piping. Given the visual similarity between black bait tunnels and black tracking tunnels, it is plausible that rodents in the present study were ‘tunnel-shy’ and less likely to enter, and therefore less likely to be detected by tracking tunnels. Managers should therefore consider the use of different monitoring methods against potential mechanisms of behavioural avoidance specific to their system.
Our data suggest that the optimal period of deployment may vary among methods; chew cards and tracking tunnels can quickly reach saturation at high rodent densities, whereas camera traps are only limited by battery life and data storage. We observed chew mark and footprint saturation at multiple sites, which could be attributed to both high site-level rodent activity as well as the length of the deployment period (Sweetapple and Nugent 2011; Ruffell et al. 2015; Burge et al. 2017). In some tracking tunnels we also observed substantial card coverage of non-target (insect) footprints, which interfered with rodent species identification. As such, species identification could only be discerned by camera trap footage (for strata-specific species composition see Nance et al. In review a), though similar comparisons have been explored in other studies (e.g. Blackwell et al. 2002; Whisson et al. 2005; Sweetapple and Nugent 2011). Ideally, mixed-method monitoring designs would implement shorter deployment periods for lure-based methods, and longer deployment periods for camera traps or similar methods (Haysom et al. 2021). Such a design would be ecologically robust, while also reducing logistical complexity: the simplest device to deploy and retrieve requires shorter deployment periods, and the more technically challenging device requires longer periods.
Cost is an important consideration when prioritising conservation actions or choosing methods to implement them (Pienkowski et al. 2021). In this study, chew cards clearly emerged as the most cost-effective tool, driven by both extremely low equipment costs and increased detection performance. This is encouraging given the limited pool of funding available for conservation globally, though should be considered within a broader management context. Compared to camera traps and even tracking tunnels, the number of taxa that chew cards can reliably detect is low: chew cards were designed to detect small rodents, possums, mustelids and cats (Sweetapple and Nugent 2011), whereas tracking tunnel detection capabilities can extend to include some insects (Watts et al. 2011) and reptiles (Lettink, Young and Monks 2022). Detection capabilities of camera traps extend even further to also include birds and larger mammals (Delisle et al. 2021). Additionally, chew cards are single-use, whereas camera traps and the plastic tracking tunnels that house tracking cards are re-usable and durable. These two factors - taxonomic detection range and re-usability - mean that tracking tunnels and camera traps may be able to be used in a wider range of management activities, thus increasing their overall cost-effectiveness over time within an integrated management framework. However, when the detection of arboreal rodents is the primary management goal, our results demonstrate that chew cards are the superior tool as measured by both cost and effectiveness.
As with any management program, practitioners need to consider the trade-offs between accuracy and feasibility. In the present study, the method used to deploy devices above-ground was physically demanding and time consuming (tree-climbing using an arborist harness). We recognise that such methods are unlikely to be feasible for intensive rodent monitoring programs. Further research is required to develop more efficient ways of deploying monitoring tools above-ground. Numerous studies have tested the accuracy of deploying detection methods at heights within the reach of ground-based fieldworkers (e.g. Mills et al. 2016; Shiels and Ramírez de Arellano 2019; Baldwin and Meinerz 2022). While promising, it is unclear whether tools at these heights will adequately capture presence of canopy-occupying rodents across different forest types, particularly forest types with high canopies. There is therefore merit in exploring and developing innovative methods for the rapid deployment and collection of monitoring devices at the canopy level. Such novel methods could involve for example, attaching chew cards to long poles, using fixed rope and pulley systems, or developing a sling-shot or arrow-shooting method (for a good example see Innes et al. 2018).
Within the scientific literature, there is a growing emphasis on measuring arboreal activity of invasive rodents (e.g. Innes et al. 2018; Ringler et al. 2021; Shiels and Ramírez de Arellano 2019). This emphasis could be explained by differences in species-specific foraging and dwelling behaviour, which has led to the postulation that species such as the highly arboreal black rat may be less targeted by ground-based monitoring and control (e.g. Tobin et al. 1997; Nelson et al. 2002; Wegmann et al. 2014). This is supported by a recent study conducted in the present study system: black rats were most commonly detected in above-ground strata, whereas Pacific rats and house mice were most commonly detected on the ground (Nance et al. In review a). The increased detection performance of chew cards in the present study is therefore correlated with black rat activity: chew cards were more effective where black rats are more common. Together, these findings support the notion that arboreal monitoring in the presence of highly arboreal species like the black rat may provide managers with a more holistic view of the prevalence of these species in vulnerable systems. As such, while arboreal monitoring may be more logistically complex, it may be desired or even necessary in the presence of invasive rodents like the black rat. Indeed, researchers have already begun exploring novel ways to detect arboreal invasive rodent presence by - for example - surveying for gnawed palm seeds (Wilmshurst and Carpenter 2020). Our results, demonstrating the performance of three commonly used methods in arboreal settings, is a key step towards addressing this issue.