Our study demonstrates that while Eastern long-necked have a remarkable ability to navigate many structures, barriers exist from common landscape features (Coffin 2007; Forman & Alexander 1998; Jakes et al. 2018; Jones et al. 2022; McInturff et al. 2020; Smith et al. 2020; Trombulak & Frissell 2000).
The obstacle materials we tested here are commonly used in road, stream, and wetland restoration or common fence types used by commercial and private landholders and therefore are likely to impact Eastern-long necked turtles throughout much of their range. Chicken wire or similar small-diameter mesh fences presented unpassable obstacles for the sampled eastern long-necked turtles, and this reflects the barrier created for many other wildlife groups as well (Baines & Andrew 2003; Bock & Bock 1994; Moseby & Read 2006; Smith et al. 2020). Although other obstacle types were successfully crossed by turtles, the time required to successfully navigate through obstacles was often substantial and this likely affects an energy cost.
Crossing success
Rocks
Turtles crossed rock obstacles with seemingly little trouble, regardless of the size of rock encountered. We expected the sharp, irregular angularity of the quarried boulders to pose a challenge to turtles, particularly smaller individuals with carapace lengths below 160 mm (33.3% of turtles in our study) as boulders were generally taller than this. Yet, 73% of boulder crossing attempts were successful, and multiple crossings of the obstacle in each trial were common. Small turtles in the sample group also used their head and necks as a point of contact to both climb and cross the large boulder obstacles.
We expected the smallest obstacles such as gravel and gabion to be crossed far more often, however, gabion had only 12 more crossings (185 of 215 attempts) than boulders (173 of 236 attempts), a more difficult obstacle. Multiple attempts were unsuccessful for both gravel and gabion, where turtles either approached the obstacle and chose not to cross or aborted crossing part-way through. This may be attributed to several factors that were not measured in our experiment. Warm surface temperatures of rocks may explain aborted crossings, as burns and eventual death from hot rocks have been recorded in turtles attempting to gabion railway track beds (Abinesh et al. 2022; Kornilev et al. 2006). However, the effect of dangerously high temperatures is unlikely to have had a major impact on turtles in our experiment due to the use of shade cloth which shaded the entire mesocosm.
Fences
Turtle's success at crossing fences was highly dependent on the fence design. The chicken wire style mesh, which was completely impassable for all turtles, is prolific in agricultural regions with sheep. The wire mesh usually extends to ground level to prevent the separation of lambs from ewes or enforce weaning (Galeana et al. 2007; McGregor 1990) and exclude predators (Kondinin Group 2016; Long & Robley 2004). Despite the complete failure to cross, chicken wire had the third highest number of overall attempts and the longest amount of time attempting to cross. The average time per crossing attempt was far higher than other obstacles at 35.6 minutes per attempt, likely imparting a high energy cost. Increased energy expenditure may place turtles at elevated risk of mortality due to exhaustion, desiccation and overheating (Ferronato et al. 2014). While turtles with roads in their home range have only small increases in energy expenditure (Patterson et al. 2019), the presence of chicken wire-type fences without suitable crossing structures would likely result in turtle exhaustion and death, as has been witnessed in other ’turtle-proof fences’ (Ferronato et al. 2014).
Exclusion fencing (prefabricated hinge joint) may still pose a risk to C. longicollis which was not identified by our study. Chelodina longicollis of the New England region, particularly Armidale NSW, exhibit irregularly small adult body sizes, (average body sizes (SCL) of females = 239 mm; male = 199 mm; Parmenter 1976), which is far smaller than in other regions where SCL may exceed 260 mm (Chessman 1984). As such, turtles used in our experiment passed through the ~ 150 mm wide spaces of the exclusion fencing with relative ease. In other regions, the carapace width of C. longicollis may exceed this width, making passage through unmodified exclusion fencing difficult or impossible (Kennett et al. 2009). However, simple cost-effective solutions, such as turtle gates (Waltham et al. 2022), that permit turtle movement while also maintaining fence integrity have been developed and trialled for other, larger-bodied, freshwater turtles.
Although fences may prevent the movement of non-target species, fences are rarely completely intact, with breaks in fences acting as potential crossing points (Jones et al. 2018). Gates often act as weak points in fences, with substantial gaps beneath gates existing either through the gate design or wheel ruts from repeated vehicle traffic (Long & Robley 2004). Even in high-cost reinforced fences, holes may be created by burrowing species such as rabbits and foxes, compromising fence integrity (Moseby & Read 2006). While intermittent gaps may be beneficial for more motile species, freshwater turtles show a significant reduction in their ability to locate gaps within the space of 3m (Waltham et al. 2022), making the likelihood of encountering gaps in turtle proof fences unlikely in all but the most dilapidated of cases.
The ability of C. longicollis to climb the silt fencing used to divide the arenas demonstrates a remarkable capacity to tolerate certain linear barriers. The loose weave of the silt fencing allowed the turtles to use their claws to climb in the vertex where the edges met at right angles. Turtles were not seen climbing parallel edges, likely due to their sprawling posture and limited mobility from hips sitting inside their shells (Vitt & Caldwell 2013). Australian turtles are not typically known for their ability to scale vertical surfaces; however, a considerable ability to climb fences (Aresco 2005a; Griffin 2005), logs (Lindeman 1999), and rocks (Davenport et al. 1984) have been recorded in several North American turtle species.
Fatigue
The number of attempts per trial decreased consistently throughout the experiment, resulting in turtles making just over half (46%) of the attempts on the final day as compared to day one. Mean daily movement lengths for C. longicollis are generally below 10 m-1 day − 1 (Roe & Georges 2008), yet our subjects exceeded this distance in 60% of trials, likely expending significant energy that led to reduced attempts as trials progressed. Additionally, the tendency for turtles to climb their enclosures reduced with the number of treatments, with no climbing of the mesocosm walls observed after trial five. Therefore, we believe that fatigue rather than habituation towards food supplied in pools was the primary cause of reduced crossing attempts as trials progressed.
Boldness
As turtles retract into their shells when exposed to potentially threatening stimuli (Greene 1988), higher degrees of boldness may reduce the risk of mortality to turtles attempting to cross busy roads because bold individuals cross more quickly. Bold turtles crossed obstacles faster than turtles with lower boldness scores, which suggests that less bold individuals may be more vulnerable to threats such as vehicle strikes or predation as they linger exposed on obstacles.
Implications for management
Our study highlights the need for selecting wildlife-friendly fence designs when planning and installing fences around wetlands and waterways. Exclusion fencing is manufactured in a range of sizes and designs and populations with larger-bodied turtles may not be as successful at crossing exclusion fencing as in our study. However, simple, cost-effective mitigation methods exist for reducing the threat posed to turtles by exclusion fencing in these scenarios. For example, clipping small sections of vertical exclusion fence wires at regular intervals to permit the movement of larger turtles through exclusion fences incurs minimal cost to landholders while maintaining the functionality of the exclusion fence (Waltham et al. 2022). The agricultural sector, particularly the sheep industry, encourages the use of wire mesh fencing, similar to that tested here, for the exclusion of canines and foxes to limit livestock depredation (Kondinin Group 2016). Many of these designs encourage wire mesh to be anchored at ground level or installed below the surface to prevent canids from digging under (Green & Gipson 1994; Kondinin Group 2016). Alternatively, the implementation of species-specific one-way gates, such as those currently used for small mammals to transit through conservation areas (Coates 2013), may warrant further investigation for the possibility of turtle-specific modification.