Gait Modulation in C. Elegans: It's Not a Choice, It's a Reflex!

In their perspectives article, Mesce and Pierce-Shimomura (2010) make a case for the existence of two locomotion gaits in C. elegans, and for the ability of this nematode worm to choose between them. Here, we offer a counter-perspective, namely, that the variety of observed behaviors are more appropriately described as a single gait. 
 
In what follows we focus on pure forward locomotion, which is generally believed to be controlled by a dedicated circuit, distinct from the circuits that control backward locomotion and head motor behavior. With regard to our question of gaits, we are not aware of any evidence that links head swings with the generation of propulsive (forward/backward) locomotion, so these will not be considered here.

In their perspectives article, Mesce and Pierce-Shimomura (2010) make a case for the existence of two locomotion gaits in C. elegans, and for the ability of this nematode worm to choose between them. Here, we offer a counter-perspective, namely, that the variety of observed behaviors are more appropriately described as a single gait.
In what follows we focus on pure forward locomotion, which is generally believed to be controlled by a dedicated circuit, distinct from the circuits that control backward locomotion and head motor behavior. With regard to our question of gaits, we are not aware of any evidence that links head swings with the generation of propulsive (forward/ backward) locomotion, so these will not be considered here.

Are swimming And crAwling chArActerized by quAlitAtively different pAtterns of body bending?
In C. elegans, forward locomotion, whether "swimming" (in liquids) or "crawling" (in or on gels) involves smooth propagation of undulations from head to tail. We argue that while such propagations can vary in kinematic parameters such as frequency, wavelength, and amplitude, these variations are continuous and there is no evidence for more than a single fundamental behavior 1 .
If, as Mesce and Pierce-Shimomura (2010) argue, "bends propagate at uniform velocity for crawling, but "whip" to the back of the body after characteristic pauses when the whole-body of the worm is bent into a C-shaped posture during swimming," then these characteristic pauses would be evident in plots of bending angle over time: the longer the pauses in extreme maximally bent positions, the squarer the bending wave would appear. In fact, in Pierce-Shimomura et al.'s own data, bending plots over time show proportionately little time spent in the extreme positions (Pierce-Shimomura et al., 2008;Vidal-Gadea et al., 2009). Our own data for worms in liquid buffer solution confirm this ( Figure 1C). More generally, Figure 1D clearly shows the equivalence of temporal patterns of dorso-ventral bending for "swimming," "crawling," and intermediate behaviors, upon normalization by amplitude and undulation period.
Furthermore, if, as claimed, swimming worms "whip" their bodies and propagate bends at non-uniform velocity, this would be evidenced by non-linear stripes in the contour plots for swimming worms ( Figure  1B of Pierce-Shimomura et al., 2008). Our own data ( Figures 1A,B, reproduced with permission from Berri et al., 2009) similarly show uniform bend propagation for worms exhibiting a range of behaviors (corresponding to a range of physical conditions). Whether looking at the spacing between the stripes (the period of undulation), the inclination of the stripes (the speed of bend propagation) or the amount of overlap (the wavelength along the body), we find only a gradual change between consecutive plots. Finally, we note that bending along the body directly maps muscle activation. Calcium imaging of muscle activity during locomotion of unrestrained worms in water and on agar (Pierce-Shimomura et al., 2008) indicates a clear one-to-one relationship between muscle activity and bending. This strong match holds in both environments and does not break down for fast undulations (during swimming). Therefore, it is reasonable to expect that the same one-toone relationship should hold in intermediate environments (up to a fixed phase lag which of course differs in different conditions). Another way of putting it is that body curvature is an equally appropriate (and easier to obtain) metric. In summary, the data strongly suggest that the entire transition from "swimming" to "crawling" corresponds to modulation of a single gait. Indeed, we find no evidence of a qualitative distinction between the undulation patterns.

Are swimming And crAwling chArActerized by different frequencies of undulAtion?
Mesce and Pierce-Shimomura (2010) state that C. elegans worms "crawl with a persistent S-shaped posture on a dry substrate with ∼1 Hz high-amplitude bends, and swim in liquid with ∼2 Hz lower amplitude bends." Indeed, the existence of two locomotion frequencies might be grounds to define two distinct gaits. However, it is important to point out the current consensus that the worm modulates its frequency of undulations as a function of the resistivity (viscosity or visco-elasticity) of the physi- Specifically, these three quantitative metrics capture the temporal and spatial properties of the body undulation to a good degree of accuracy (although additional metrics could always be added to capture the complex, non-sinusoidal features of the waveform).

Frontiers in Behavioral Neuroscience www.frontiersin.org
March 2011 | Volume 5 | Article 10 | 2 are these transitions really spontaneous and internally selected? Here we provide an alternative interpretation, motivated in part by our own, possibly related observations. One technical difficulty in working with many polymer solutions (e.g., agar, gelatin, methylcellulose) is obtaining homogeneous environments with different concentrations of the colloidal particles. Indeed, in characterizing a transition as a function of the environment, any observation of a change of behavior must be explicitly controlled against possible changes in the worm's local, microscopic environment before such differences can be attributed to an internal decision. We therefore needed an objective metric to quantify the local stiffness (or visco-elasticity) of our medium. Using image analysis, and treating the worm as a deformable body governed by Newton's equations of motion (applied to low Reynolds' numbers), we numerically estimated the ratio of drag coefficients for each short extract of recorded locomotion. Indeed, plotting the data against this physical metric, the smooth and continuous transition in Figures 1D-F of Berri et al. (2009) is immediately obtained. In the absence of such an objective characterization of the microscopic physical environment, it is impossible to have confidence in any statement about behavioral transitions being "spontaneous" and driven by "decision" rather than by some possible response of worms to local changes in the physical environment. In Vidal-Gadea et al. (2009), we might ask whether, e.g., the sudden (small) change in frequency observed for worms in 2-3% methylcellulose could be due to the worm moving into or out of contact with the more rigid agar surface underneath.

does the existence of A swimdefective mutAnt imply multiple gAits?
Mesce and Pierce-Shimomura argue that the identification of "swim-defective" mutants supports the existence of a distinct swimming gait, but we note that the three examples given by Pierce-Shimomura et al. (2008) exhibit some locomotion defect on agar as well. That aside, and even in the extreme case of a mutant with a clear phenotype in liquid and none on agar, it is important to which "crawling" and " swimming" (if they are indeed distinct behaviors) are characterized by quite similar frequencies.
Specifically the transitions are quantified by a difference of 0.2 Hz or less in the undulation frequency. This frequency difference is comparable in value to the variability in frequency (1.2-1.5 Hz) for locomotion in 0.5% methylcellulose (Vidal-Gadea et al., 2009), all of which is dubbed swimming. The small magnitude of the transitions may raise questions, not only about the validity of the classification, but even about the selective advantage worms might gain by such "decisions." That said, it is clear from the bending plots in Vidal-Gadea et al. (2009) that some small frequency transitions were observed by Pierce-Shimomura et al. But "crawling" appears to fall at about 0.8 Hz (Vidal-Gadea et al., 2009). It is therefore clear that C. elegans locomotion cannot be characterized by discrete sets of kinematic parameters, but rather by a broad and continuous range of such parameters. cAn worms switch spontAneously between the discrete gAits of swimming And crAwling?
Transitions between two undulation frequencies in a single environment, if spontaneous, could support the existence of two gaits. However, we note that any observed transitions (in the wild type) were only reported for certain high concentrations (2.25 and 2.75%) of methylcellulose (Vidal-Gadea et al., 2009), for Bending angle traces as in (C), taken from the 1st, 3rd, 5th, and 7th panels of (B; see color-coded arrow heads) and normalized by undulation period and amplitude. Note the similarity of the waveforms.