Journal of Comparative Physiology A

, Volume 200, Issue 7, pp 641–656

Sensory-evoked turning locomotion in red-eared turtles: kinematic analysis and electromyography

Authors

  • Dan B. Welch
    • Department of Growth Development and StructureSouthern Illinois University, School of Dental Medicine
    • Department of Cell Biology and NeuroscienceUniversity of California
Original Paper

DOI: 10.1007/s00359-014-0908-0

Cite this article as:
Welch, D.B. & Currie, S.N. J Comp Physiol A (2014) 200: 641. doi:10.1007/s00359-014-0908-0

Abstract

We examined the limb kinematics and motor patterns that underlie sensory-evoked turning locomotion in red-eared turtles. Intact animals were held by a band-clamp in a water-filled tank. Turn-swimming was evoked by slowly rotating turtles to the right or left via a motor connected to the shaft of the band-clamp. Animals executed sustained forward turn-swimming against the direction of the imposed rotation. We recorded video of turn-swimming and computer-analyzed the limb and head movements. In a subset of turtles, we also recorded electromyograms from identified limb muscles. Turning exhibited a stereotyped pattern of (1) coordinated forward swimming in the hindlimb and forelimb on the outer side of the turn, (2) back-paddling in the hindlimb on the inner side, (3) a nearly stationary, “braking” forelimb on the inner side, and (4) neck bending toward the direction of the turn. Reversing the rotation caused animals to switch the direction of their turns and the asymmetric pattern of right and left limb activities. Preliminary evidence suggested that vestibular inputs were sufficient to drive the behavior. Sensory-evoked turning may provide a useful experimental platform to examine the brainstem commands and spinal neural networks that underlie the activation and switching of different locomotor forms.

Keywords

TurtleLocomotionTurningKinematicsMotor patterns

Abbreviations

BP

Back-paddle

EE

Elbow extensor

EMG

Electromyogram

FS

Forward swim

HP

Hip protractor

HR

Hip retractor

KE

Knee extensor

SP

Shoulder protractor

SR

Shoulder retractor

Introduction

Animal locomotion in water, air, or terrestrial environments exhibits a variety of continual fine-steering adjustments along with more abrupt turns and reversals. During swimming or flight, in which animals are free to move in three dimensions, directional adjustments can involve changes in yaw, pitch, and roll orientations. In the present paper, we focus on lateral turning (changes in yaw, or horizontal orientation) during swimming locomotion in freshwater turtles. The biomechanical and neuromuscular mechanisms that mediate lateral turns during animal locomotion are much less thoroughly examined than those that underlie forward rectilinear locomotion. However, the inherent asymmetry of lateral turning behavior provides unique opportunities for examining the differential selection and modulation of locomotor patterns on the right and left sides. Among some arthropods, terrestrial turning locomotion, especially during tight turns or in-place rotation, can involve different locomotor forms in legs on the outer and inner sides of the turn with different intralimb timing (e.g., forward stepping on the outer side, backward stepping on the inner side) that rotates the animal around a vertical axis (jumping spiders: Land 1972; honeybee: Zolotov et al. 1975; cockroach: Bell and Schal 1980; fruitfly: Strauß and Heisenberg 1990; Mu and Ritzmann 2005). In other cases, particularly during more gradual curve walking, arthropods may shorten stride length on the inner side of the turn (ants: Zollikofer 1994; crayfish: Cruse and Silva Saavedra 1996; walking sticks: Gruhn et al. 2009). Similar asymmetries occur during aquatic turns. For example, whirligig beetles swim in a tightly curved trajectory via the asymmetric paddling of outer legs or sculling of the outer wing, combined with the extension of a stationary forelimb or elytra (wing cover) on the inner side that acts as a brake or pivot (Fish and Nicastro 2003). Vertebrates that swim via axial undulations (teleost fish: Gray 1933; lamprey: McClellan 1984; McClellan and Hagevik 1997; Saitoh et al. 2007) execute lateral turns via an abnormally large wave of muscular contraction passing down the trunk and tail on the inner side of the turn. In sea turtles, which swim via a lift-based mechanism involving synchronous flapping movements of both wing-like forelimb flippers, turns are executed either by reducing the duration of the forelimb cycle on the inner side of the turn (Renous and Bels 1993; Wyneken 1996) or by a combination of increasing the stroke amplitude in the outer-side forelimb and using the hindlimb flipper on the inner side of the turn as a stationary rudder (Walker 1971b; Lohmann et al. 1995; Avens et al. 2003). Some previous work has also been done on straight swimming vs. turn-swimming in fresh-water turtles. Rivera et al. (2006) described the pattern of limb movements during voluntary forward and backward lateral turns in free-swimming painted turtles (Chrysemys picta) in a study that investigated the dynamics of turning performance rather than the specifics of limb motor patterns and inter- or intralimb coordination. The bilateral movements and motor patterns of the turtle hindlimbs (but not forelimbs) were also described for spontaneous turning episodes in carapace-restrained red-eared turtles (Trachemys scripta elegans) held by a band-clamp (Field and Stein 1997a, b; Earhart and Stein 2000).

Surgically reduced vertebrate preparations have been developed to enable the study of actual and fictive behavior elicited by natural sensory stimulation, permitting the close examination of neural mechanisms. Lamprey tail fin–spinal cord preparations exhibited episodes of tail withdrawal followed by fictive escape swimming in response to cutaneous stimulation of the tail fin (McClellan and Grillner 1983), and head–brainstem–spinal cord preparations from lampreys could be induced to generate fictive turning (aversive withdrawal) and escape swimming bouts by mechanical or electrical stimulation of the trigeminal-innervated skin of the snout (McClellan 1984). There have been no comparable means of eliciting locomotion in restrained or semi-intact turtle preparations via sensory stimulation, and the search for such a method was part of the motivation for our present experiments. Freely behaving freshwater turtles are capable of making both gradual and abrupt voluntary lateral turns while swimming, especially during pursuit of fish prey (Rivera et al. 2006); however, these turns are typically executed by single, bilaterally asymmetric swim cycles.

In the present study, we describe sustained sensory-evoked turn-swimming in red-eared turtles, elicited by horizontal rotation of the animals around their mid-body vertical axes. We assessed the kinematics and electromyogram (EMG) motor patterns of bilateral forelimb and hindlimb movements during prolonged episodes of turn-swimming and during switches between right and left turn-swimming in response to switched rotational direction. Our results enabled us to identify a highly stereotyped set of bilaterally asymmetric movements and EMG motor patterns that underlie sensory-evoked turning behavior and provide a potential basis for the development of reduced preparations that express fictive turning motor patterns in response to rotational stimulation. Most of the data presented in this paper were submitted by the first author as part of his dissertation in partial fulfillment of the requirements of the doctoral degree (Welch 2011).

Methods

Adult red-eared turtles (n = 17; Charles D. Sullivan, Nashville, TN, USA), Trachemys scripta elegans, weighing 492–707 g with carapace lengths of 13.0–15.5 cm, were submerged in crushed ice for ≥2 h to induce hypothermic anesthesia (Lennard and Stein 1977; Marcus 1981; Samara and Currie 2007). Turtles remained partially submerged in crushed ice throughout all surgical procedures, including the suturing of marker beads to the skin over limb joints and the insertion of EMG electrodes in selected limb muscles (see below). After removing turtles from the ice, the ventral plastron was dried off so that reference markers could be glued or painted on. The turtles were then warmed to room temperature in a small tub before being placed in the motorized band-clamp apparatus (Fig. 1a) for testing. Following data collection, turtles without EMG implants were re-iced so that marker beads could be removed. Once those animals had warmed to room temperature, they were returned to the vivarium. Turtles with EMG implants were euthanized after experiments by an intrapleural injection of pentobarbital sodium (Beuthanasia-D™; Schering-Plough Animal Health, Union NJ; 390 mg/kg b.wt.).
https://static-content.springer.com/image/art%3A10.1007%2Fs00359-014-0908-0/MediaObjects/359_2014_908_Fig1_HTML.gif
Fig. 1

Apparatus used to elicit and record turn-swimming and rectilinear swimming in carapace-restrained turtles. a Animals were held by a band-clamp around the shell in a clear, water-filled tank, close enough to the water surface to breathe. Digital videos were recorded from below in a 45° mirror. Slow rotation of the animal to the right or left, via a reversible variable-speed motor attached to the band-clamp, reliably evoked robust turning behavior in the opposite direction. Forward swimming could be elicited by many stimuli, including movement within the animal’s visual field, touching the caudal body surface, etc. b Markers on the plastron (ventral shell), limb joints, chin and tail (am) permitted us to monitor forelimb, hip, knee, head and tail angles as a function of time, using video motion capture and analysis software (Datapac 5, Run Technologies) and synchronize those measurements with EMG recordings from selected hindlimb or forelimb muscles. Circular arrows in a and b indicate the axis of horizontal rotation induced by the motorized band-clamp encircling the middle of the shell

Rotation of carapace-restrained turtles

To rotate turtles in the horizontal plane, we attached a reversible geared DC motor (Reliapro model 38-010; 250:1 gear ratio; Jameco Electronics; Belmont, CA, USA) to the top of the band-clamp shaft that encircled each animal’s shell (Fig. 1a). The motor was driven by a variable DC power supply (Tenma, model 72-6628) and had sufficiently high torque (7.5 kg-force cm) to overcome easily the counter-acting forces produced by a medium-sized turtle’s vigorous turning movements in the opposite direction. Rotational speed was varied from 30 to 120°/s by changing the driving voltage, but for most measurements (Figs. 2, 3, 5, 6, 7, 8, 9, 10, 11) was kept at 90°/s during stimulation. Rotational direction was reversed via a manual polarity switch between the power supply and the motor. We timed the rotational stimuli via a Master-8 programmable pulse generator (A.M.P.I; Jerusalem, Israel), which gated power to the DC motor via a power MOSFET (IRF510) switching circuit. To produce triangular rotational stimuli, we switched the polarity of electrical power to the motor during the brief (0.5 s) delays between four consecutive gating pulses (each lasting 10 s) to the MOSFET switching circuit. We constructed a rotational position sensor for electrically monitoring turtle rotation (Fig. 1a). A plastic gear on the band-clamp post was placed in contact with an identical gear on the post of a 10-turn potentiometer, the terminals of which were connected to the resistance-measurement (PGR) inputs of a Grass 7P122 low-level DC amplifier (Astro-Med Inc., West Warwick, RI, USA). This converted rotations of the band-clamp into positive or negative voltage deflections corresponding to counter-clockwise and clockwise rotations, respectively. The voltage outputs of the sensor were calibrated in degrees of band-clamp rotation, and then recorded along with video and electromyograms on a PC computer and/or DVC tape.
https://static-content.springer.com/image/art%3A10.1007%2Fs00359-014-0908-0/MediaObjects/359_2014_908_Fig2_HTML.gif
Fig. 2

Limb trajectories compared for right and left turn-swimming and rectilinear swimming in the same animal (turtle D24). Right limbs are shown on the right and left limbs on the left. Stick figures show the trajectories of the left and right wrists, knees, ankles, head (indicating a dot painted on the chin; j in Fig. 1b) and tail (indicating a pigmented spot near the tip of the tail; m in Fig. 1b) for 2-s episodes of turn-swimming (elicited by 90°/s counter-clockwise and clockwise rotations, respectively) and voluntary rectilinear swimming. During turn-swimming (a and c), the “outer” forelimb and hindlimb exhibited out-of-phase forward swimming (FS), the “inner” hindlimb exhibited back-paddling (BP), the “inner” forelimb (star) was held motionless in a protracted position, and both the neck and tail were strongly bent toward the direction of the turn. During rectilinear swimming (b), right and left forelimbs and hindlimbs all exhibited bilaterally symmetrical forward swim movements, and both the head and tail were kept nearly straight (close to the center midline), although the tail oscillated slightly to the right and left. To visualize the overlapping limb trajectories of the rotating animal in a and c, we digitally stabilized the ventral mid-line of stick figures (line b, c in Fig. 1b) via the kinematic analysis software (see “Kinematics”)

https://static-content.springer.com/image/art%3A10.1007%2Fs00359-014-0908-0/MediaObjects/359_2014_908_Fig3_HTML.gif
Fig. 3

Average minimum and maximum forelimb, knee and hip angles on the right and left sides during left turn-swimming elicited by clockwise rotations (a) and right turn-swimming elicited by counter-clockwise rotations (b) (n = 5 turtles; 108–148 cycles per average). Left and right edges of bars represent average minimum and maximum angles, respectively. Errorbars = standard deviations. Statistical comparisons were made between corresponding minimum angles (e.g., right forelimb) and between corresponding maximum angles during left and right turns, using the Mann–Whitney U test (Siegel 1956). *p < 0.0005 and NSp > 0.05, respectively

Kinematic recordings

We recorded turtle locomotor behaviors from below with a digital video camera (Canon Optura 20 mini-DV), aimed at a 45° mirror beneath the clear Plexiglas bottom of the swim tank (Fig. 1a). Brightly colored plastic beads (3 mm diameter) were attached to the ventral-medial side of the wrists (h and i in Fig. 1b) and the ventral-anterior side of the knees (f and g) and ankles (k and l) to track movement of the limbs relative to fixed reference markers painted on the ventral plastron with white correction fluid (ae). Reference points a and c marked the rostral and caudal ends of the ventral midline. Reference point b marked the rostro-caudal level of pectoral-girdle pivot points along the ventral midline, where the scapulae meet the underside of the dorsal carapace. Our examination of red-eared turtle skeletons showed that this point could be closely estimated from external anatomy, as it occurs at nearly the same rostro-caudal level as the anterior border of the lateral shell bridge (the “axillary notch”; see Carr 1952, pp. 37–38). Reference points d and e marked the positions of the right and left hip joints (note that images are inverted in the mirror), while j and m marked spots painted on the chin and near the tip of the tail, respectively. Video recordings were obtained with a 30-Hz frame rate and 1/250-s shutter speed and recorded on a DVC tape. Selected video sequences were de-interlaced and digitized later on a lab PC at a resolution of 60 fields per second. We tracked marker locations on the limbs and plastron field-by-field, using Datapac 5 software (Run Technologies, Laguna Hills, CA, USA). A detailed description of limb angle measurements was given previously (Samara and Currie 2007). Briefly, forelimb angle was defined as ∠abh on the right and ∠abi on the left, and was measured relative to a stationary reference point on the plastron midline (b, see above) (Walker 1971a; Samara and Currie 2007). “Forelimb angle” was used instead of “shoulder angle” because the shoulder joint is not stationary during locomotion, but moves back and forth through an arc as the pectoral girdle rotates (Walker 1971a; Samara and Currie 2007). Hip angle was defined as the angle between the thigh line (df on the right or eg on the left in Fig. 1b) and a line parallel to the ventral midline (ac) with its origin at the stationary hip joint. Knee angle was defined as ∠dfk on the right and ∠egl on the left. Head angle was defined as ∠abj, where an angle of 0° meant that the head was pointed straight forward, and movements of the head to the right and left were indicated as positive and negative angles, respectively. Tail angle was not measured, but the position of the tail marker was tracked in Fig. 2.

To visualize the trajectories of digitized limb, head and tail markers during turning behavior in rotating turtles (Fig. 2a, c), we used a customized feature in the two-dimensional motion analysis module of our Datapac 5 software, developed by Youan Chang at Run Technologies. This allowed us to stabilize the animal’s ventral mid-line (line segment b–c in Fig. 1b) in digitized stick-figures and observe the overlapping movement trajectories of each limb marker relative to a stationary body axis.

Electromyography

EMGs were recorded with bipolar electrodes consisting of two H-ML-insulated 0.004-in. silver wires, approximately 24 in. in length and glued together with Permabond 910 adhesive (National Starch and Chemical, Bridgewater, NJ, USA). Electrodes were implanted into selected hindlimb muscles (n = 7), forelimb muscles (n = 3), or both (n = 1) on the right side, using previously described surgical techniques (Stein 1978; Earhart and Stein 2000; Samara and Currie 2007). The muscle-end of the paired EMG wires were separated over the last 1–2 cm and about 0.5 mm of insulation was scraped off at the tips; these were then hooked with fine forceps and inserted 4–8 mm apart in the thickest part of each muscle with a hypodermic needle. Recorded hindlimb muscles included (1) a hip retractor–knee flexor (HR: flexor cruris pars flexor tibialis internus, n = 5), (2) a hip protractor (HP: puboischiofemoralis internus pars anteroventralis, n = 4), and (3) a knee extensor (KE: triceps femoris pars femorotibialis, n = 4). Note that the hip retractor and protractor muscles (HR and HP) referred to here were called hip extensor and hip flexor muscles (HE and HF) in previous papers (Samara and Currie 2007, 2008a, b). Recorded forelimb muscles included (1) a shoulder retractor–adductor (SR: pectoralis, n = 3) that also rotates the pectoral girdle backward, (2) a shoulder protractor–abductor (SP: deltoideus, n = 1) that also rotates the pectoral girdle forward, and (3) an elbow extensor (EE: triceps brachii, n = 2) (Walker 1973; Stein 1978). Electrodes from different muscles were loosely bundled together to minimize tangling and movement artifacts, and were sufficiently long to wind around the band-clamp post for several complete rotations without breaking or becoming taut. We conducted a post mortem analysis to verify correct electrode placements. Recordings from electrodes that were not properly placed in the correct muscle were excluded from our counts. EMGs were amplified and filtered (30–1,000 Hz) with Grass P511 AC amplifiers (Astro-Med., West Warwick, RI, USA) and digitized on a lab PC, using Datapac 5 software (Run Technologies, Laguna Hills, CA, USA). A remote video synchronization unit (Peak Performance Technologies, Los Angeles, CA, USA) was used to synchronize analog recordings (EMGs and the output of the rotational position sensor) with the video data. Episodes chosen for figures were formatted with Datapac 5 and CorelDraw.

Data analysis

We used digitized kinematic data to calculate average minimum and maximum joint angles, cycle frequency, and intra- and interlimb phase values for turning responses, using the event selection and advanced spreadsheet modules of Datapac 5. The cycle frequency of turn-swimming was calculated for leftside forward swimming hip movements on the outer side of turns, using Datapac 5 software. The time between consecutive hip protraction onsets was measured in kinematic recordings and converted to cycle frequency. The data were imported into SYSTAT 11 (Systat Software, Inc. Chicago, IL, USA) for analysis. Repeated-measures ANOVA (Zar 1999) was used to assess the effects of rotation speed and duration on the cycle frequency of turn-swimming responses. We used t tests to isolate the differences between groups in our ANOVA analyses. Post hoc pair-wise comparisons (with Bonferoni corrections) were made to assess the differences between mean cycle frequency values for different rotation velocities within the same time-bin and for different time-bins at the same rotation velocity.

Dual-referent phase analysis was used to assess intralimb and interlimb coordination (Berkowitz and Stein 1994; Field and Stein 1997a, b); phase values were calculated with Datapac 5. One joint or limb angle was selected as “referent” and the other as “target”. The onsets of referent-limb protraction were defined by phase values of 0.0 and 1.0. The onsets of referent retraction were defined by a phase value of 0.5 (see Field and Stein 1997a, b). Phase data were imported into Oriana 3.0 (Kovach Computing Services, Anglesey, Wales, UK) to obtain circular statistics, which are appropriate for cyclical events (Batschelet 1981; Zar 1999; Mardia and Jupp 2000). The angle and length of the mean vector were calculated using standard trigonometric functions. The angle of the mean vector (μ) represents the average phase value on a circular scale ranging between 0.0 and 1.0. The length of the mean vector (r) indicates the directional concentration of data points around the mean vector angle. We used the Rayleigh test to discriminate between uniform and unimodal-clustered phase distributions and the Watson U2 test (Batschelet 1981; Zar 1999) to determine if phase values were significantly different.

Results

Kinematics

In response to slow horizontal rotation around their central vertical axes (Fig. 1), turtles that were held by a band-clamp and immersed in water displayed stereotyped forward turn-swimming against the direction of imposed rotations. In the absence of rotation, the same animals could be induced to execute episodes of rectilinear forward swimming by visual stimulation (e.g., movement of the experimenter) and/or tactile stimulation of the animal’s body surface. We recorded digital video of rotation-evoked turn-swimming and rectilinear forward swimming from below the tank and computer-analyzed forelimb and hindlimb movements as well as, in some cases, head and tail movements. In Fig. 2, we compared foot trajectories and head and tail positions in the same turtle during 2-s sequences of right and left turn-swimming evoked by counter-clockwise and clockwise rotation (Fig. 2a, c) and rectilinear forward swimming in the absence of rotation (Fig. 2b). During rectilinear swimming (Fig. 2b), the turtle displayed the typical bilaterally symmetric pattern of forward swim cycles in all four limbs, including 1:1 out-of-phase movement of contralateral limbs (right–left forelimbs and right–left hindlimbs) and ipsilateral limbs (right forelimb–hindlimb and left forelimb–hindlimb), and 1:1 nearly in-phase movement of diagonal limbs (right forelimb–left hindlimb and left forelimb–right hindlimb). During rotation-evoked turning (Fig. 2a, c), the same animal exhibited a highly stereotyped pattern of (1) tightly coordinated forward swimming motor patterns in the hindlimb and forelimb on the outer side of the turn, (2) back-paddling motor patterns in the hindlimb on the inner side of the turn, (3) a stationary and protracted forelimb on the inner side of the turn that appeared to act as a brake, and (4) neck and tail bending toward the direction of the animal’s turn. The same movement pattern was observed in all the turtles we examined (n = 17).

We present the average minimum and maximum limb angles for all four limbs during left and right turn-swimming in Fig. 3 (n = 5 turtles). Statistical comparisons were made between corresponding minimum angles and between corresponding maximum angles for a given limb during left and right turns, using the Mann–Whitney U test (Siegel 1956). The most obvious difference in angular range (average minimum to maximum angle) was between forelimb forward swimming, with very large-amplitude movements, and forelimb braking, with little movement at all. The next most striking difference was in the angular ranges of forward swimming vs. back-paddling movements at the hip. Hip movements during back-paddling occurred over a relatively narrow range and were protraction biased (biased toward smaller minimum and maximum hip angles) compared to forward swimming. These differences in minimum and maximum hip angles during back-paddling and forward swimming may only occur during rotation-evoked turning behavior, since no such difference was noted in previous work for spontaneous turning in non-rotated turtles (see Fig. 2 in Earhart and Stein 2000).

The cycle frequency of turn-swimming limb movements increased in proportion to the imposed rotational velocity (p < 0.01), and decreased over time during 24-s rotations (p < 0.042; repeated measures ANOVA) (Fig. 4). We used t tests to isolate the differences between groups in our ANOVA analyses. At a rotational velocity of 45°/s, the turn-swimming cycle frequency started at 1.16 ± 0.15 cycles per sec (cps), averaged over the first 8 s of rotation, and decreased to 0.82 ± 0.13 cps during the final 8 s (mean ± SE, n = 4 turtles). At a velocity of 90°/s, the cycle frequency went from 1.58 ± 0.10 cps in the first 8 s to 0.99 ± 0.09 cps in the final 8 s; and at the maximum velocity of 120°/s, the cycle frequency went from 1.89 ± 0.13 cps in the first 8 s to 1.19 ± 0.14 cps in the final 8 s. Altering the rotational velocity had the most pronounced effect on limb cycle frequency during the first 8 s of rotation. The most significant run-down of cycle frequency over time occurred between the first and second 8-s intervals during rotations at 120 and 90°/s. In all the other experiments in this study, we used the moderate rotation rate of 90°/s, since this speed reliably evoked robust turning responses without eliciting as much run-down in cycle frequency as faster rotation rates.
https://static-content.springer.com/image/art%3A10.1007%2Fs00359-014-0908-0/MediaObjects/359_2014_908_Fig4_HTML.gif
Fig. 4

The cycle frequency of turn-swimming increased with the velocity of rotational stimulation, but decreased as a function of time during prolonged rotations. Data were collected for three counter-clockwise rotational velocities: 45, 90 and 120 degrees per second, with each episode lasting 24 s (n = 4 turtles). Cycle frequency was measured for forward swim (FS) hip protraction onsets on the outer side of turns, pooled and averaged over three consecutive 8-s time bins for each rotational velocity. The errorbars are standard errors. We applied t tests to assess differences between specific groups (see “Data analysis”). Significantly different pairs are indicated by asterisks (p < 0.05)

We assessed intralimb (knee–hip) phase relationships for the forward swimming (FS) hindlimb on the outer side of the turn and the back-paddling (BP) hindlimb on the inner side of the turn during right turn-swimming episodes evoked by counter-clockwise rotation (Fig. 5, n = 5 turtles). Knee–hip phase values for FS and BP hindlimbs were found to be statistically different at p < 0.01 (Watson U2 test). The circular histograms in Fig. 5 indicate phase values for the onset of knee extension (target) movement within the ipsilateral hip protraction cycle (referent). Values close to 0.0 or 1.0 indicate an in-phase intralimb relationship, while values close to 0.5 indicate an alternating, out-of-phase relationship. Vector angle indicates the mean phase, while vector length indicates the strength of intralimb coupling (r) on a scale of 0.0–1.0. The mean knee–hip phase for the outer FS hindlimb (127 cycles) was 0.28 ± 0.09 (mean ± circular SD), r = 0.97, p < 0.0001 (Rayleigh test). The mean knee–hip phase for the inner BP hindlimb (110 cycles, pooled from 5 turtles) was 0.82 ± 0.04, r = 0.84, p < 0.0001. These knee–hip phase values are very close to those previously published by Field and Stein (1997a) from kinematic analysis of hindlimb forward swimming (0.25 ± 0.04) and back-paddling (0.77 ± 0.06) in non-rotating turtles held by band-clamp in water during spontaneous turning behavior.
https://static-content.springer.com/image/art%3A10.1007%2Fs00359-014-0908-0/MediaObjects/359_2014_908_Fig5_HTML.gif
Fig. 5

Intralimb (knee–hip) phase relationships for the forward swimming (FS) hindlimb on the outer side of the turn and the back-paddling (BP) hindlimb on the inner side of the turn during right turn-swimming episodes evoked by counter-clockwise rotation. Circular histograms (left side of figure) indicate phase values for the onset of knee extension movements within the hip protraction cycle (onset–onset); data pooled from five turtles. Each concentric circle represents 12 cycles for each range of phase values. The direction of the vectors (arrows) indicates the mean phase, and vector length indicates the strength of intralimb coupling (r) on a scale of 0.0 (innermost circle) to 1.0 (outermost circle). Representative kinematic sequences (right side of figure) were selected from turtle D19. Knee (target) angle is shown as dashed lines; hip (referent) angle is shown as solid lines. a Phase relationship between the “outer” (left side) forward swimming (FS) knee and hip (127 cycles pooled in histogram). Mean intralimb phase = 0.28, r = 0.97, p < 0.0001 (Rayleigh test). b Phase relationship between the “inner” (right side) back-paddling (BP) knee and hip (110 cycles pooled in histogram). Mean intralimb phase = 0.82, r = 0.84, p < 0.0001. A statistical comparison of knee extension onsets within the hip cycle between FS and BP was made with the Watson U2 test. The phase values for FS and BP were found to be statistically different, p < 0.001

In Fig. 6, we present interlimb phase relationships for rotation-evoked turning (n = 5 turtles). Figure 6a displays the phase distribution (110 cycles) for BP hip protraction-onsets within the contralateral FS hip protraction cycle. The mean hip–hip interlimb phase was 0.29 ± 0.14 (mean ± circular SD), r = 0.67, p < 0.0001 (Rayleigh test). Thus, BP and FS hips were out of phase with each other, but not clearly alternating as they were in a previous study of spontaneous turning in band-clamped turtles without rotation, where the mean phase relationship between contralateral BP and FS hips was 0.48 ± 0.1 (Field and Stein 1997b). Figure 6b shows the strong, alternating phase coupling between the ipsilateral FS forelimb and FS hip on the outside of the turn, where the mean phase between FS forelimb and FS hip was 0.53 ± 0.06, r = 0.94, p < 0.0001 (123 cycles). We also assessed the phase relationship between the diagonal forelimb–hindlimb pair. Figure 6c displays the out-of-phase coordination between the outer FS forelimb and the inner BP hip, where the mean phase was 0.27 ± 0.16, r = 0.60, p < 0.0001 (109 cycles).
https://static-content.springer.com/image/art%3A10.1007%2Fs00359-014-0908-0/MediaObjects/359_2014_908_Fig6_HTML.gif
Fig. 6

Interlimb phase relationships during right turn-swimming episodes evoked by counter-clockwise rotation. Circular histograms (left side of figure) indicate phase values pooled from 5 turtles. Representative kinematic sequences (right side of figure) were selected from turtle D19. a Phase relationship of “inner” (right side) back-paddling (BP) hip protraction onsets within the “outer” (left side) forward swim (FS) hip protraction cycle (onset–onset; 110 cycles pooled in histogram). Each concentric circle represents six cycles for each range of phase values. The direction of the vectors (arrows) indicates the mean phase, and vector length indicates the strength of interlimb coupling (r) on a scale of 0.0 (innermost circle) to 1.0 (outermost circle). Mean interlimb phase = 0.29, r = 0.67, p < 0.0001 (Rayleigh test). b Phase relationship of “outer” (left side) FS forelimb protraction onsets within the ipsilateral (left side) FS hip protraction cycle (123 cycles). Each concentric circle represents 16 cycles for each range of phase values. Mean interlimb phase = 0.53, r = 0.94, p < 0.0001. c Phase relationship of “outer” (left side) FS forelimb protraction onsets within the contralateral (right side) BP hip protraction cycle (109 cycles). Each concentric circle represents four cycles for each range of phase values. Mean interlimb phase = 0.27, r = 0.60, p < 0.0001

Turtles quickly switched the direction of their turns, including the asymmetric pattern of right and left limb activities and head position, when we abruptly reversed the direction of rotation. Figure 7 shows a representative sequence from one animal that was switched several times between counter-clockwise and clockwise rotations, with each period of constant-velocity rotation in one direction (90°/s) lasting just over 10 s. Counter-clockwise rotation (rotating the head to the left) always elicited robust right turning behavior against the direction of rotation, while clockwise rotation (rotating the head to the right) always elicited left turning behavior. The upper part of the figure shows kinematic records of head position, left and right forelimb angles, and right knee and hip angles. The EMGs in the bottom part of the figure will be discussed in the “Motor patterns” section, below. During right turns, this turtle displayed (1) head movement to the right, (2) forward swimming in the left (outer) forelimb, but virtually no movement in the right (inner) forelimb, (3) back-paddling in the right (inner) hindlimb, and (4, not shown) forward swimming in the left (outer) hindlimb. During left turns, the turtle exhibited (1) head movement to the left, (2) forward swimming in the right (outer) but not the left (inner) forelimb, (3) forward swimming in the right (outer) hindlimb, and (4, not shown) back-paddling in the left (inner) hindlimb. It should be noted from the figure that the angular range of right hip movements was relatively small and protraction biased during back-paddling, but considerably larger and more retraction biased during forward swimming. The same pattern is shown statistically for pooled data in Fig. 3.
https://static-content.springer.com/image/art%3A10.1007%2Fs00359-014-0908-0/MediaObjects/359_2014_908_Fig7_HTML.gif
Fig. 7

Synchronized kinematic and EMG recordings of right and left turn-swimming in response to abrupt switches in the direction of rotation (90°/s) from turtle D21. Top trace: output of rotational position sensor (see “Rotation of carapace-restrained turtles”). Second from top trace: head position, with the dashed line (0°) indicating that the head is pointed straight forward. The arrow indicates a pause in the head movement toward the left during a left turn. Middle 4 traces: kinematic recordings of left and right forelimb angles and right knee and hip angles. Bottom 3 traces: EMG recordings from right KE (knee extensor), HP (hip protractor) and HR (hip retractor–knee flexor) muscles (see “Electromyography” for muscle identities). Counter-clockwise rotation of the turtle via the motorized band-clamp evoked right turn-swimming: head turned to the right, forward swimming in left forelimb (shown) and hindlimb (not shown), motionless right forelimb, back-paddling in right hindlimb. Clockwise rotation evoked left turn-swimming: head turned to the left, motionless left forelimb, back-paddling left hindlimb (not shown), forward swimming in right forelimb and hindlimb. Triangles below bottom trace indicate the onset of 2-s back-paddle and forward swim recordings shown with expanded time-bases in Fig. 8

The lateral head movements (neck bending) that accompanied changes in rotational direction could be explained as passive responses to rotational inertia and water-associated drag forces acting on the head; however, our evidence suggests that the head movements were largely driven by active muscular control of the neck. First, head movements in water were often not smooth, as would be expected if they were passively driven, but jerky and discontinuous (see downward arrow at left side of head position trace in Fig. 7), suggesting muscular control. Second, identical head movements were observed when turtles were slowly rotated in the air, where water-induced drag on the head could not contribute to passive neck bending (see “Sensory activation”). Third, we observed similar head movement toward the direction of a turn even during spontaneous right and left turning behavior in non-rotating animals that were held stationary by a band-clamp (Welch and Currie, unpublished observations).

Motor patterns

In eleven turtles, we recorded EMGs from selected hindlimb muscles (n = 7), forelimb muscles (n = 3), or both (n = 1) on the right side during rotation-evoked turn-swimming (see “Electromyography” for detailed muscle identities and recording procedures). The bottom of Fig. 7 shows the hindlimb EMG motor patterns that accompanied the switching back and forth between right and left turn-swimming. With this slow time-base, the differences in right-side EMG amplitudes during back-paddling (right turns) and forward swimming (left turns) are very apparent. As was shown previously for spontaneous hindlimb activity in non-rotating turtles (Earhart and Stein 2000), the back-paddling power-stroke occurs during protraction of the hip and is associated with large-amplitude HP (hip protractor; puboischiofemoralis) and KE (knee extensor; femorotibialis) EMG bursts, while the return-stroke occurs during retraction of the hip and is associated with small-amplitude HR (hip retractor; flexor tibialis) bursts. In contrast, the forward swimming power-stroke occurs during retraction of the hip and is associated with large-amplitude HR EMG bursts, while the return-stroke occurs during protraction of the hip and is associated with small-amplitude HP and KE bursts. Figure 8 shows 2-s-long sequences of back-paddling and forward swimming from the recording in Fig. 7 (onsets of back-paddle and forward swim sequences in Fig. 8 indicated by unfilled arrowheads at the bottom of Fig. 7), displaying only the right hindlimb EMGs and hip angle traces with expanded time-bases. Note that during the back-paddle, the large HP EMG bursts began near the peak of hip retraction (maximum hip angle) in each cycle and continued through most of the hip protraction phase. Also, the large KE EMG discharge began before and continued throughout each HP burst, as was shown earlier by Earhart and Stein (2000). But during forward swimming, cycles were dominated by large HR EMG bursts that began near the peak of hip protraction (minimum hip angle) and continued through most of the hip retraction phase. The weak KE EMG discharge began during the latter part of each small HP burst (visible in the last two cycles) and continued just after HP offset and before each HR burst, similar to previous observations (Stein 1978; Earhart and Stein 2000).
https://static-content.springer.com/image/art%3A10.1007%2Fs00359-014-0908-0/MediaObjects/359_2014_908_Fig8_HTML.gif
Fig. 8

Hip angle (upper trace) and EMG recordings (lower three traces) from the right hindlimb of turtle D21 during 2-s back-paddle and forward swim sequences, taken from the turning episode shown in Fig. 7 and shown with expanded time-bases. The onsets of these sequences are indicated by triangles in Fig. 7. Note that the back-paddle motor pattern was dominated by large-amplitude HP and KE EMG bursts (with KE onsets preceding HP onsets), while the forward swim motor pattern was dominated by large-amplitude HR bursts (with KE onsets occurring during the latter part of each HP burst). These features are characteristic of BP and FS motor patterns, as previously described for spontaneous turning episodes (Earhart and Stein 2000). Corresponding EMGs during back-paddle and forward swim are shown with the same amplification

Figure 9 displays a representative example of forelimb EMG activity during switches between right and left turn-swimming, showing differences in right-side EMGs during braking (right turns) and forward swimming (left turns) with a slow time-base. Large-amplitude, rhythmic EMG bursts were only observed in SP (shoulder protractor; deltoid), SR (shoulder retractor; pectoralis) and EE (elbow extensor; triceps) muscles during forward swimming, when the right forelimb was on the outer side of left turns. During braking, when the right forelimb was on the inner side of right turns, we observed only tonic activity in the SR muscle, low-amplitude rhythmic discharge in SP, and no detectable activity at all in EE. This low-level tonic and weak bursting activity reflected the absence of significant movement in the braking forelimb. Figure 10 shows 2-s-long sequences of braking and forward swimming from the recording in Fig. 9 (onsets of Fig. 10 sequences indicated by unfilled arrowheads at the bottom of Fig. 9), displaying only the right forelimb EMGs and forelimb angle traces with expanded time-bases. Forelimb braking EMGs have not been recorded previously, however, the timing of EE, SP and SR muscle activity that we observed during forelimb forward swimming (Fig. 10) was similar to those reported elsewhere (see Fig. 8 in Stein 1978; Fig. 4 in Rivera and Blob 2010). The motor patterns were characterized by alternation between SP EMG activity during the swim return-stroke (forelimb protraction) and SR discharge during the power-stroke (forelimb retraction); brief, but large-amplitude EE bursts occurred at the peak of forelimb protraction (minimum forelimb angle), slightly overlapping the end of SP and the beginning of SR EMG discharge.
https://static-content.springer.com/image/art%3A10.1007%2Fs00359-014-0908-0/MediaObjects/359_2014_908_Fig9_HTML.gif
Fig. 9

Synchronized kinematic and EMG recordings of right and left turn-swimming in response to switches in the direction of rotation (90°/s) from turtle D20, focusing on forelimb activity. Top trace: output of rotational position sensor. Second and third traces from top: kinematic recordings of left and right forelimb angles. Bottom 3 traces: EMG recordings from right EE (elbow extensor), SP (shoulder protractor) and SR (shoulder retractor) muscles (see “Electromyography” for muscle identities). Counter-clockwise rotation of the turtle evoked right turn-swimming: forward swimming left forelimb, braking right forelimb (hindlimb movements not shown). Clockwise rotation evoked left turn-swimming: braking left forelimb, forward swimming right forelimb. Triangles below bottom trace indicate the onset of 2-s braking and forward swim recordings shown with expanded time-bases in Fig. 10

https://static-content.springer.com/image/art%3A10.1007%2Fs00359-014-0908-0/MediaObjects/359_2014_908_Fig10_HTML.gif
Fig. 10

Forelimb angle (upper trace) and EMG recordings (lower three traces) from the right forelimb of turtle D20 during 2-s braking and forward swim sequences, taken from the turning episode shown in Fig. 9 and shown with expanded time-bases. The onsets of these sequences are indicated by triangles in Fig. 9. Note that during braking, the largely stationary forelimb exhibited only weak rhythmic discharge in the SP muscle, while EE and SR were silent. In contrast, forward swimming was characterized by large-amplitude bursting EMG discharge in EE, SP and SR muscles that was correlated with cyclic forelimb movements. The timing of EE burst-onsets just before each SR burst is similar to that described previously for forelimb forward swimming movements evoked by spinal cord stimulation (Stein 1978). Corresponding EMGs during braking and forward swim are shown with the same amplification

Sensory activation

Horizontal rotation of turtles in water should activate vestibular, visual and proprioceptive sensory systems. To help assess the relative importance of these modalities in activating rotation-evoked turning behavior, we examined turning responses in three animals that had been blind-folded with opaque goggles (to prevent movement-related visual stimuli), suspended in the air (to abolish proprioceptive sensations caused by water currents against the limbs, head and neck) and rotated to the right and left at 90°/s. All three turtles exhibited fairly robust turning responses under these conditions, suggesting that sensory-evoked turning behavior may be driven to a significant extent by vestibular inputs. Figure 11 shows kinematic recordings of head position, right and left forelimb movements, and right hip and knee movements in turtle D21. Note that this is the same animal used for control turning responses (rotated in water without blindfolds) in Fig. 7. The main features of rotation-evoked turn-swimming still occurred, including strong, coordinated forward swimming in the outer forelimb and hindlimb, and braking and back-paddling of the inner forelimb and hindlimb, respectively. We did not quantify the limb movement parameters of turning responses obtained from animals in the air; however, the braking and back-paddling of the inner limbs appeared to be less consistent and sustained in all three animals than they were during control rotations in water without blind-folds (compare Figs. 7, 11). In Fig. 11, note that inner forelimb braking was interrupted by intermittent forward swim cycles, and hindlimb back-paddling appeared to fatigue rapidly, so that the second episode consisted of only two slow cycles.
https://static-content.springer.com/image/art%3A10.1007%2Fs00359-014-0908-0/MediaObjects/359_2014_908_Fig11_HTML.gif
Fig. 11

Rotation (90°/s) still elicited turn-swimming responses to the right and left when animals (n = 3) were suspended in the air (to abolish proprioceptive sensations caused by water currents) and blind-folded with opaque goggles (to prevent movement-related visual stimuli), suggesting that rotational vestibular stimulation was sufficient to activate turn-swimming motor patterns. The data shown are from turtle D21. The main features of rotation-evoked turn-swimming still occurred; although the braking and back-paddling of inner limbs were less consistent than in control responses (see same turtle in Fig. 7). However, note that inner forelimb braking was interrupted by two brief forward swim sequences (asterisks above left forelimb trace), and the second episode of hindlimb back-paddling was reduced to only two slow cycles (dots above right hip and knee traces)

Discussion

We evoked robust, sustained turning behavior (forward turn-swimming) in carapace-restrained turtles by slowly rotating animals to the right or left via a motorized band-clamp encircling middle of the shell. Turtles reliably exhibited forward turning in the opposite direction of the imposed rotation, in an apparent effort to counter the rotation and stabilize their yaw position. They also switched rapidly back-and-forth between right and left turns when the direction of rotation was reversed. We examined the hindlimb and forelimb contributions to turning behavior via video kinematics and synchronized EMG recordings of limb motor patterns. Our data confirmed and extended previous studies (Field and Stein 1997a, b; Rivera et al. 2006) that characterized some elements of turning behavior in freshwater turtles. During a turn, the “outer” forelimb and hindlimb produced tightly coordinated, out-of-phase forward swimming, while the “inner” hindlimb exhibited back-paddling, and the inner forelimb was held in a tonic protracted position, apparently acting as a brake, or “pivot” as earlier described by Rivera et al. (2006). Rotation-evoked turning was always accompanied by neck and tail bending toward the direction of the turn. In addition, the characteristic asymmetric pattern of limb and head movements during turn-swimming still occurred when animals were rotated in the air instead of in water (to prevent drag-associated movement of the limbs and head and the associated proprioceptive stimulation) and blindfolded (to abolish visual inputs), suggesting that turn-swimming was primarily triggered by rotation-evoked vestibular activity.

Sensory-evoked turning behavior was previously described in hatchling sea turtles (Lohmann et al. 1995; Avens et al. 2003). Lohmann et al. (1995) constructed a wave simulator that reproduced the circular motion of hatchling turtles swimming beneath ocean waves. Animals suspended in the air and subjected to this wave action consistently attempted to turn-swim into the simulated wave (against the wave direction), just as our adult freshwater turtles swam against the direction of horizontal rotation. In both cases, the animals were attempting to maintain a consistent heading in response to externally imposed accelerations. Hatchling sea turtles (green and loggerhead: Lohmann et al. 1995; Avens et al. 2003) swim in the same manner as their adult forms (green, loggerhead, Ridley, hawksbill, leatherback: Walker 1971b; Renous and Bels 1993; Wyneken 1996), via the synchronized strokes of powerful front flippers. Turns are initiated in sea turtle hatchlings by extending a rear flipper as a stationary rudder on the inner side of the turn and continued during sustained turns by increasing the amplitude of the stroke cycle in the outer-side front flipper (Lohmann et al. 1995; Avens et al. 2003). Similar movements, along with a shorter stroke cycles on the inner side of the turn compared to the outer side, have been observed in adult sea turtles during lateral turns (Walker 1971b; Renous and Bels 1993). However, in freshwater turtles like those in our present study, forward propulsion is achieved via the alternating forward swim strokes of large, webbed hind-feet, while the smaller, non-webbed forelimbs may serve mainly to counter the horizontal rotational torque produced by the alternating hindlimb power-strokes and regulate orientation, keeping the animal on a straight heading (Pace et al. 2001; Rivera and Blob 2010). During rotation-evoked turns, our red-eared turtles continued producing strongly coordinated, out-of-phase forward swim cycles in the outer forelimb and hindlimb, while protracting the inner forelimb as a brake to increase drag and switching the inner hindlimb from forward swimming to back-paddling. Turn-swimming motor patterns sped up in response to increased rotational velocity and were sustained for as long as the rotation continued (up to 24 s), albeit with gradually diminishing cycle frequencies (Fig. 3). There were both similarities and differences between the kinematics of our sustained, sensory-evoked turning episodes in red-eared turtles and those of the brief, voluntary forward turns in freshwater painted turtles, as described by Rivera et al. (2006). In both cases, turtles held the inner (or, “inboard”) forelimb in a stationary, protracted position during forward turns, which would be expected to increase drag on the inner side and contribute to rotational torque. However, Rivera et al. (2006) described the outer (“outboard”) forelimb and both hindlimbs as continuing with normal “rectilinear” forward swimming” during turns. Thus, they suggested that painted turtles produced forward turns primarily by increasing drag on the inner side of the turn via the braking forelimb, while the remaining limbs continued to execute rectilinear forward locomotion. In contrast, we found that while the outer forelimb and hindlimb did both exhibit robust forward swimming during a rotation-evoked turn, the inner hindlimb switched to a back-paddling motor pattern (see: Field and Stein 1997a; Earhart and Stein 2000) in which the power-strokes occurred during hip protraction. Also, we saw head and tail movements toward the direction of the turn in every animal, but these were not mentioned in the Rivera et al. (2006) article, and no neck or tail bending was apparent from their photographs of an animal during a right forward turn (see their Fig. 2b). It may be that the neck and tail bending that we always observed in our experiments (Figs. 2, 7, 11) occurs only in response to rotational stimulation, as vestibular reflexes, and that these do not normally accompany brief voluntary turns.

Forward swimming, back-paddling and braking motor patterns contribute to turn-swimming

Red-eared turtles exhibit two distinctly different hindlimb motor patterns during aquatic locomotion: forward swimming and back-paddling (Fig. 5; see also Field and Stein 1997a; Earhart and Stein 2000). During hindlimb forward swimming, the onsets of knee extension occur during the middle of the hip protraction phase of each swim cycle, and the power-stroke occurs during hip retraction. Knee extensor (KE) EMG activity begins near the middle of each hip protractor (HP) EMG burst and continues until shortly after HP offset (Fig. 8b; see also Earhart and Stein 2000). During hindlimb back-paddling, the onsets of knee extension occur before the onset of the hip protraction phase of each swim cycle, and the power-stroke occurs during hip protraction. Knee extensor (KE) EMG activity begins before the onset of each hip protractor (HP) EMG burst and continues until HP offset (Fig. 8a; see also Earhart and Stein 2000). In the present experiments, we always observed forward swimming in the outer hindlimb and back-paddling in the inner hindlimb during rotation-evoked turn-swimming (n = 17 turtles). This was previously noted in the spontaneous, voluntary turn-swimming of turtles held by a stationary band-clamp (Field and Stein 1997a, b; Earhart and Stein 2000); however, further kinematic measurements and EMG recording will be required to determine if a similar pattern of outer and inner hindlimb motor patterns, perhaps lasting only a single cycle, underlies some voluntary turns in free-swimming turtles.

We also recorded EMGs from three different forelimb muscles during turns, including SP (deltoid, shoulder protractor), SR (pectoralis, shoulder retractor) and EE (triceps brachii, elbow extensor). We observed only forelimb forward swimming and braking patterns, but never saw forelimb back-paddling like that of the inner hindlimb. Surprisingly, forelimb braking was not accompanied by significant EMG activity in any of the three recorded muscles, except for some very low amplitude, rhythmic discharge in the SP muscle (Fig. 10a). We expected to see higher levels of tonic or modulated discharge in both SP and EE, given the elbow-extended appearance of the braking forelimb. Forelimb forward swimming motor patterns, however, were robust, with intralimb EMG timing consistent with what Rivera and Blob (2010; see their Fig. 4; Table 2) described in great detail for intact, free-swimming animals, and what Stein (1978) observed during forward swimming elicited by spinal cord stimulation in high-spinal turtles.

We have not yet recorded EMGs from neck or tail muscles during turn-swimming, so we are unsure which muscles control the lateral neck and tail bending in the direction of forward turns. However, it appears likely that lateral neck bending is produced, at least in part, by the actions of cervico-capitis and sternomastoid muscles (Ashley 1955; Herrel et al. 2007).

Interlimb coordination during turn-swimming

During rectilinear swimming, freshwater turtles exhibit forward swim motor patterns in all four limbs, with 1:1 alternating coordination of contralateral limbs (right–left forelimbs and right–left hindlimbs, mean phase ≈ 0.5), 1:1 out-of-phase coordination of ipsilateral limbs (right forelimb–hindlimb and left forelimb–hindlimb, mean phase ≈0.4), and 1:1 nearly in-phase coordination of diagonal limbs (right forelimb–left hindlimb and left forelimb–right hindlimb, mean phase ≈ 0.9) (Stein 1978; Davenport et al. 1984; Field and Stein 1997b; Gillis and Blob 2001; Samara and Currie 2007, 2008b). Field and Stein (1997b) also showed that the hindlimbs continued to alternate 1:1 (mean phase = 0.48) during voluntary turning behavior in turtles held by a stationary band-clamp, in which one hindlimb executed forward swimming and the other executed back-paddling. During rotation-evoked turning in the current study, we found that contralateral back-paddling and forward swimming hips were still out of phase with each other (mean phase = 0.29), but not clearly alternating as they were in Field and Stein (1997b).

We demonstrated in previous work that the alternating right–left hindlimb movements that occur during voluntary rectilinear turtle swimming were not dependent on crossed commissural pathways in the spinal hindlimb enlargement (Samara and Currie 2007). By surgically bisecting the posterior spinal cord down the longitudinal midline, including the 5-segment hindlimb enlargement (segments D8–S2) and the pre-enlargement segment (D7), we showed that the phase-coupling of contralateral hindlimb forward swimming movements was unchanged after destruction of all crossed commissural axons in those segments, although hindlimb movement amplitudes were reduced. These results are consistent with a model involving alternating descending propriospinal drive from the left and right sides of the forelimb enlargement, sufficient to maintain the out-of-phase coordination of right and left hindlimbs in the bisected-cord preparation. Thus, descending propriospinal and crossed commissural spinal cord pathways appear to function together during rectilinear swimming as redundant mechanisms to sustain right–left hindlimb alternation during turtle locomotion. The fact that the braking inner forelimb is virtually motionless during rotation-evoked turn-swimming suggests that there is little or no propriospinal rhythmicity arising from inner forelimb circuitry to contribute to interlimb coordination, in contrast to the outer side, which is strongly rhythmic. In this regard, note the strong uncrossed interlimb coordination between the outer forelimb and hindlimb in Fig. 6b (r = 0.94), and the weaker crossed coordination between outer and inner limbs in Fig. 6a (r = 0.67) and c (r = 0.60). This is in contrast to bilaterally symmetrical, rectilinear forward swimming in which all four limbs are rhythmically active and strongly phase-coupled (Samara and Currie 2007, 2008b). We suggest that the weakened crossed coordination among the three rhythmically active limbs during rotation-evoked turning is due to the absence of significant descending propriospinal drive on the side of the braking (inner) forelimb that would normally augment the crossed coupling produced by commissural pathways.

Descending activation of turn-swimming by brainstem command systems

What is the nature of descending brainstem command systems that activate forward swimming in outer limbs versus braking and back-paddling in the inner limbs of turtles during sensory-evoked turning behavior? How do those commands switch between locomotor forms when the direction of imposed rotation is reversed? In vitro head–brain–spinal cord preparations from the lamprey made it possible to activate fictive forward swimming via mechanical or electrical stimulation of trigeminal afferents innervating the skin of the snout (McClellan 1984; Fagerstedt et al. 2001) and permitted intracellular characterization of identifiable reticulospinal neurons. Another approach in lamprey has been to record and sort extracellular spikes from the larger reticulospinal axons in the spinal cords of freely behaving animals and restrained animals subjected to rotational stimulation (Zelenin 2005, 2011); this enabled the activity of identified classes of cell to be correlated with specific rotational stimuli and different forms of locomotion (forward vs. backward swimming). In our own earlier work, we evoked fictive forward swim motor patterns in contralateral hindlimb nerves via electrical stimulation in the dorsolateral funiculus of the mid-body spinal cord in low-spinal immobilized animals (Juranek and Currie 2000) and by stimulation in the anterior-lateral brainstem of decerebrate-immobilized preparations (Currie 2003). Future studies should investigate the origin and identity of descending reticulospinal drive that selects and activates asymmetric locomotor forms on the right and left sides during sensory-evoked turtle turning behavior. We note that an in vitro turtle brain/ear-capsules preparation has been in use for years to examine the neuronal basis of rotation-evoked vestibulo-ocular reflexes (Fan et al. 1997; Jones and Ariel 2008). A similar preparation, with cervical-brachial spinal cord and selected forelimb nerves attached, might permit a cellular investigation of sensory-activated fictive turn-swimming and a better understanding of vertebrate locomotor pattern selection.

Acknowledgments

We thank Basilio Haro and Clifford Jung Hun Kye for assistance with video digitization. All procedures were performed according to protocols approved by the UC Riverside Institutional Animal Care and Use Committee in accordance with federal guidelines. This research was supported by U.C.R. Academic Senate grants to S.N.C.

Copyright information

© Springer-Verlag Berlin Heidelberg 2014