Biomechanical factors influencing successful self-righting in the pleurodire turtle Emydura subglobosa

Self-righting, or the ability of an animal to recover from an inverted position, is a critical function for many animals. In terrestrial settings, inverted animals may risk stranding, exposure or increased predation if they cannot right themselves in a timely fashion. In this context, performance during self-righting has been considered as a factor related to fitness in several invertebrate and vertebrate taxa (e.g. Penn and Brockmann, 1995; Delmas et al., 2007; Jusufi et al., 2008; Jusufi et al., 2011; Porter et al., 2011; Kaspari et al., 2016; Mitchell et al., 2016). For example, better self-righting performance has been linked to greater rates of survival in horseshoe crabs (Penn and Brockmann, 1995) and higher fitness in turtles (Steyermark and Spotila, 2001; Delmas et al., 2007).

Self-righting performance depends on both the shape and flexibility of the body. Animals with flexible bodies can twist and bend their anteroposterior body axis to self-right (Jusufi et al., 2011; Evangelista et al., 2014; Singleton and Garland, 2018); however, animals with rigid bodies (e.g. beetles, crabs, turtles) cannot execute such movements. Self-righting can be particularly important for many rigid-bodied animals because they are often forced into procumbent positions through competition, predation attempts or falls during the navigation of complex environments (Penn and Brockmann, 1995; Mann et al., 2006; Golubović et al., 2013).

With their limited axial mobility, rigid animals employ a variety of alternative strategies to flip over. For example, beetles exhibit approximately 20 different, stereotyped self-righting behaviors, depending on the species (Frantsevich, 2004), and locusts rely on their large hindlegs to self-right (Faisal and Matheson, 2001). Turtles use a variety of strategies that are thought to be primarily dependent on shell morphology, which itself is strongly correlated with the habitat in which a species lives. In terrestrial taxa such as tortoises, the carapace (top of the shell) is typically domed, and turtles use a strategy primarily involving movements of the limbs to shift the center of mass and induce rolling of the body (Ashe, 1970; Chiari et al., 2017). However, in aquatic species with a flatter, more streamlined carapace, structures such as the limbs and head (though typically not the tail, which is reduced in most turtles) may be able to reach the ground and act as levers to flip the body (Ashe, 1970).

Previous studies have evaluated how various morphological factors influence self-righting in turtles, often through comparisons of how long it takes until an attempt is made (latency) or how long it takes until an attempt is successful (duration) (Mann et al., 2006; Delmas et al., 2007; Domokos and Varkonyi, 2008; Golubović et al., 2013; Mitchell et al., 2016; Chiari et al., 2017). However, this approach only provides insight into the correlation between morphological variation and the end-product of the self-righting behavior. An understanding of the actual mechanics that drive success versus failure during a righting attempt is still lacking. Such an understanding could establish a predictive framework of which factors, among many possible movements and exerted forces, are most likely to contribute to righting success and its critical consequences. To evaluate such factors, we used force-platform recordings synchronized with high-speed video to compare the magnitude and orientation of forces produced between successful and unsuccessful righting attempts by the pink-bellied side-neck turtle Emydura subglobosa, a pleurodire that primarily uses its head to flip. We predicted that during successful attempts, the head would limit anteroposteriorly directed forces and, instead, exert consistently greater moments to produce roll about the long axis of the body. Investigating these factors represents a promising avenue for bettering our understanding of self-righting in rigid-bodied animals and provides a framework for investigating the processes governing this behavior.

Four pink-bellied side-neck turtles, Emydura subglobosa (Krefft 1876) (carapace length 181.5±5.17 mm) were purchased from a commercial vendor (Turtles and Tortoises, Inc., Brooksville, FL, USA). Turtles were housed in stock tanks in a temperature-controlled greenhouse and fed pellets ab libitum. All experiments were conducted under Clemson University IACUC guidelines (protocol 2017-034).

To facilitate measurement of flipping kinematics from videos, turtles were marked with high contrast points on the ventral midline at the anterior and posterior margin of the plastron, and at three points along the anterior plastron margin (Fig. 1). To elicit self-righting attempts, turtles were inverted and placed on their carapace, such that the dorsal surface of the head contacted a custom-built force platform (K&N Scientific, Guilford, VT, USA). Specifications of the force platform and signal processing are reported by Butcher and Blob (2008) and Kawano et al. (2016). Three-dimensional forces were recorded at 5000 Hz using a custom-written LabVIEW (v.6.1, National Instruments, Austin, TX, USA) routine, while being filmed with digitally synchronized high-speed video in dorsal and frontal views at 100 Hz (Phantom v 5.1, Vision Research Inc., Wayne, NJ, USA). Video and force data were synchronized by a trigger that sent a light pulse to the video and a square-wave pulse to the force recordings. Forces were only recorded from the head because it is the only mobile structure that contacts the ground during righting attempts by E. subglobosa. To avoid exhaustion, trials were conducted no more than 10 times per day over five non-consecutive days for each turtle. We recorded approximately eight successful flips (Movie 1) and eight failed attempts (Movie 2) from each individual, for a total of ∼64 videos.

Force data were processed using the R package ‘Kraken’ (https://github.com/MorphoFun/kraken). Video data were tracked using DLTdataviewer software (Hedrick, 2008). Processed kinematic and force data were combined in custom-written Matlab routines to calculate net ground reaction force (GRF), and its anteroposterior (AP) and mediolateral (ML) inclination angles in the frame of reference of the turtle. We also calculated the angle of the head to the ground, total yaw (lateral rotation of the body) and roll velocity. We calculated the flipping moment as the vector product of GRF and the moment arm between the roll axis of the shell and the point where the head of the turtle contacted the ground (Fig. 1).

Statistical analyses were conducted using mixed-effects models, with individual as a random effect {full model: Success/Failure∼Mean anteroposterior–GRF angle+Mean mediolateral–GRF angle+Mean head angle+Mean roll velocity+Mean flip moment [the GRF standardized by turtle mass×moment arm of the head (body-weight×meters, BWm)] +Total yaw+1|Individual}. We used Akaike's information criterion (AIC) to assess the importance of variables in determining success or failure, and model averaging to find the variables that best predicted success (Burnham et al., 2011). All statistical analyses were conducted in R v. 3.3.2 (www.r-project.org).

We found that the best predictors of a successful flip were mean roll velocity, mean flipping moment and mean AP GRF angle (Table 1). Head angle, mean ML GRF angle and total yaw of the body were not substantial predictors of flipping success (Table 1). Successful flips were characterized by a much higher roll velocity (Success=175.39±22.81 deg s⁻¹; Failure=26.31±10.70 deg s⁻¹), double the flipping moment (Success=0.13±0.03 BWm; Failure=0.07±0.01 BWm), and a more vertically directed AP GRF angle (Success=28.03±6.79 deg; Failure=41.23±4.82 deg) (Fig. 2, Table S1).

For E. subglobosa, righting success was determined not only by the magnitude of the flipping moment exerted but also by the speed of the attempt (Fig. 2A,B). The role of speed in successful flipping suggests that success is probably determined very early in a self-righting attempt, with slow, continued straining likely proving to be fruitless. Among factors that might impact the effective production of a flipping moment, excess yaw (lateral rotation) might be expected to impede roll about the long axis of the shell. However, even though failed flips showed twice as much yaw as successful flips, yaw was limited in all attempts (averaging <20 deg) and had little influence on righting success (Table S1). The ML angle of the GRF and the angle of the head relative to the body also played negligible roles in determining self-righting performance (Table 1). The minimal effect of head angle suggests that the primary driver of differences in flipping moment between successful and failed self-righting attempts is the magnitude of the force being applied, rather than the orientation and length of the moment arm between the head and the roll axis of the shell.

By combining kinematic analyses of self-righting with data on force production, our analysis provides novel insight into the mechanisms through which successful self-righting is achieved. While most previous research has used patterns of flipping performance to measure exhaustion and other fitness-related traits (Penn and Brockmann, 1995; Delmas et al., 2007; Kaspari et al., 2016; Mitchell et al., 2016), there has been little focus on the actual mechanisms governing flipping performance and success. Our data provide a foundation for further evaluations of how such mechanics might influence performance across morphologically diverse systems. For example, neck posture and shell morphology differ dramatically in turtles across species, between sexes and throughout ontogeny (Ashe, 1970). Certain shell shapes are thought to facilitate self-righting and reduce the energy required to successfully flip (Ashe, 1970), but the ability to self-right is important for all species of turtle. By integrating biomechanical data with morphological comparisons, it may be possible to identify traits that enable turtles to self-right despite morphological constraints (Chiari et al., 2017).

In addition to anatomical factors, numerous environmental conditions (e.g. temperature, substrate) are known to influence terrestrial locomotion (Lailvaux, 2007; Kaspari et al., 2016) and could potentially influence self-righting performance for a range of taxa. Future studies could examine whether and how differences in the environment influence self-righting mechanics and performance. Furthermore, these data help to establish a framework for evaluating self-righting in other rigid animals (e.g. beetles, crustaceans) as well as in additional taxa, providing new perspectives for studies that use self-righting performance to estimate fitness. Such broader comparisons within and between species for a variety of conditions could also inform biomimetic applications in which rigid bodies with alternative constructions must self-right under variable conditions.

We thank A. Arellanez for her assistance with data collection, and members of the Blob lab for their assistance with animal care.