Results

Swimming behavior changes over the course of larval development

After preliminary experiments, we chose four ages for behavioral observations at which animals showed clearly distinct behavioral patterns. These timepoints roughly correspond to two phases of the early planula (48 and 72 h postfertilization (hpf)), the late planula stage (96 hpf), and the tentacle bud or metamorphosed stage (120 hpf), at our developmental conditions (21 °C) (Figure 1B). To assess detailed aspects of behavior at these ages, we developed an assay that allowed us to obtain comprehensive behavioral tracks and the associated metrics under standardized conditions by filming as the animals moved freely for 5 min (Figures 1C, S1A, and S1B, see STAR Methods section for details). As previously reported,8 larvae quickly develop into active swimmers between 48 and 72 hpf and become more explorative at 96 hpf before mostly ceasing activity around 120 hpf (Figure 1D). Our tracked data allowed us to extract detailed aspects of the swimming behavior as the animals developed. For example, we observed that not all metrics changed simultaneously or stereotypically between and within the different ages, indicating that behavioral control is more complex than previously appreciated and may depend on different aspects of larval development, specifically anatomical characteristics and information processing capacity.

As they age, more larvae swim faster and further, with the average farthest distance (59.5 mm) and highest mean speed (0.2 mm/s) occurring at 96 hpf; however, the individual with the highest total distance (224.5 mm) and mean speed (0.75 mm/s) was a 72 hpf-aged individual. The slowest age was 48 hpf (mean speed 0.12 mm/s), which also swam the shortest average distance (35.8 mm). At 120 hpf these values drop (mean speed 0.15 mm/s and total distance 45.5 mm) and reach similar values to those observed in 72 hpf larvae (Figures 1E–1H, S1C, and S1E; Tables S1–S4). The total distance traveled correlates with speed (_R_2: 48 hpf: 0.948, 72 hpf: 0.998, 96 hpf: 0.998, 120 hpf: 0.968), as these metrics depend on each other (Figures 1E and S1J, see also STAR Methods). However, total distance is relative (larvae could have a high total distance value by swimming spatially confined in circles), which would not contribute to an actual Euclidean distance, a value that is more relevant for biological phenomena such as dispersal behavior. To determine if larvae also travel further in space and do so more effectively, we measured maximum distance (Euclidean distance Figure 1G) and confinement as a measure of efficiency to reach a distant point in space (Figures 1H, S1F, and S1I and STAR Methods for calculations). Animals at 48 hpf were most confined on average (ratio 0.048) and displayed the lowest maximum distance (3.1 mm). In contrast, 96 hpf larvae showed lower confinement (ratio 0.12) and the highest average maximum distance (8.7 mm). Interestingly, when looking at the distribution of these two metrics, we found that 48 and 120 hpf old larvae show distributions with mostly low, but subpopulations of relatively high values. This pattern was not found at 72 and 96 hpf, which display a unimodal distribution of similarly high values, indicating that all larvae at these ages and only a small proportion of the youngest and oldest larvae are capable of dispersing efficiently. Intriguingly, these observations parallel the appearance of the apical tuft, a larval-specific sensory structure.29

As previously reported,8,11 we also observed that Nematostella larvae can switch between stationary and swimming modes (Figure 1I), indicating that active swimming behavior is an energetically costly behavior that the larvae will only display when conditions are advantageous, for example for dispersing away from the colony, or when conditions are so unfavorable that the larvae or the settled polyp risk physical damage (such as high UV intensity11,13). When quantifying the percentage of stationary versus moving individuals across age, we found that the ratio is shifting in favor of movement in 72 (80%) and 96 hpf (89%), while for the youngest and oldest larvae the ratio is approximately 50% between stationary and moving individuals (Figure 1J; Table S5).

The range between the minimal and maximal speed values that larvae can achieve is not drastically changing between different age groups (Figures S1G and S1H), suggesting that the physiological requirements for both inactive and fast swimming modes are given throughout the development of the animals. Consequently, the ability or the drive to swim must be actively regulated to explain our observations. Therefore, to understand why the animals display different modes of activity with distinct ages, we looked more closely at the distribution of the activity (total distance) over the time of our 5 min experiment and discovered that the largest shifts were observed during the first 60 s of our recordings (Figures 1K and 1L; Table S6). This time point corresponds to the arousal phase that is caused by the mechanical handling of the animals or the transfer to a novel environment, a common feature that has been reported before both in _Nematostella_8 and other organisms.30 This peak in arousal of larvae at 96 hpf coincides with the highest expression of putative sensory receptor proteins, such as TRP channels, some of which have been implicated in mechanosensitive behavior in _Nematostella vectensis_19,31 (Figures S1K and S1L), suggesting that the animals are more sensitive to external stimuli at this age and that this sensory information could be used to activate the swimming mode.

Cilia change in length but not basal beating frequency

Cnidarian larvae are equipped with motile cilia that propel them through the water. Since the animal’s activity is changing over time, we suspected this might be mediated through cilia-related changes. Along this line of thought, we next investigated which aspects of larval ciliation are changing over the course of development. In cnidarian planula larvae, cilia act both as sensory and motor units and can most likely modify swimming patterns and behavioral complexity through both functions.17,32,33 We therefore investigated the overall length of the epithelial ciliation in the aboral, side, and oral region, (Figures 2A and 2E) as well as their beating frequency (Figures 2B–2D) and the length of the sensory organ cilia, the tuft (Figure 2F), to understand more about how ciliary details might contribute to the enhanced sensitivity and better steering abilities in older larvae.

Figure 2: Cilia beating frequency and length during development(A) Representative image of a 96 hpf larva at 20× with regions of measurement denoted; scale bar,100 μm.(B) Image of region of interest (ROI) taken from a high-speed microscopy video taken for CBF analysis, scale bar, 10 μm.(C) Frequency map depicting peak frequency for each pixel.(D) Cilia beat at frequencies between 7 and 18 Hz with no significant differences (48–96 hpf n = 10–18, N = 6). Dashed line represents the means and letters denote statistical significance of ns p > 0.05 by the Kolmogorov-Smirnov test; asterisk (∗) denotes a significant deviation from unimodality by Hartigan’s Dip test.(E) Cilia at the aboral region of the body do not significantly change in length with age, whereas cilia along the side of the body decrease in length after 96 hpf and cilia at the oral end increase in length after 48 hpf (48–96 hpf n = 21–30, N = 5; 120 hpf n = 28–31, N = 1). Lines at mean ± SEM. Stars denote statistical significance of ns p > 0.05, ∗p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001 by Šídák’s multiple comparisons.(F) Tuft (cilia) length shows a trend of increasing length after 48 hpf (48–96 hpf n = 16–21, N = 5; 120 hpf n = 13, N = 1). Lines at mean ± SEM. Letters denote statistically significant differences in means of p < 0.05 by Dunn’s multiple comparisons (tuft cilia length). See also Tables S15–S17 for more information.

Our measurements in restrained larvae show that the ciliary beating frequency (CBF) ranges between 7 and 18 Hz for all ages and body regions (Figures 2D and S2A), with no significant differences between the means. This observation shows that there is no overall increase in beating frequency as the animals age, and that larvae at all ages might be able to actively modify the CBF between set boundaries. Larvae at 96 hpf show a bimodal distribution of CBF in the aboral region (Table S15). It is possible that this sharp distribution points to an enhanced control over the beating at this age. Our setup does, however, not permit statements about CBF changes during swimming behavior, which might be more relevant to determine its effect on the speed in freely moving animals.

Other aspects of ciliation, such as increased ciliary length, might correspond to a larger power stroke when beating, increasing the force generated to propel the larval body through the water. We measured the length of cilia across the larval body over development (48–120 hpf), and when comparing average values for the three regions within the same age we found that cilia in the side region are significantly longer (range of means 13.94–16.63 μm) than those in the aboral (10.45–11.58 μm) and oral (9.89–11.93 μm) regions for all ages (Table S16). When comparing ciliary length within the same region across different ages, we find three primary features (Figure 2E). Firstly, there are no significant differences in apical ciliary length across ages. Secondly, 120 hpf have significantly shorter side cilia compared to 48–96 hpf (mean 48 hpf: 15.55 μm, 72 hpf: 16.62 μm, 96 hpf: 16.31 μm, 120 hpf: 13.94 μm), hinting toward a possible role in locomotion of this subgroup of cilia. Finally, 72–120 hpf larvae have significantly longer oral cilia compared to freshly hatched 48 hpf larvae (mean 48 hpf: 9.89 μm, 72 hpf: 11.42 μm, 96 hpf: 11.49 μm, 120 hpf: 11.93 μm), possibly in preparation for a more prominent role of the oral end in feeding in the metamorphosed animals (Figure 2E).

In addition, we measured the length of the cilia comprising the tuft of the apical organ and found significantly longer tuft cilia after 48 hpf (Figure 2F, mean 48 hpf: 36.81 μm, 72 hpf: 67.91 μm, 96 hpf: 96.07 μm, 120 hpf: 78.86 μm). Longer cilia increase both the physical reach and can house a greater number of receptors due to a larger membrane surface, leading to enhanced sensitivity for sensory cues, providing better spatial information that could facilitate more agile or faster swimming behavior.34,35 This is especially important considering the putative sensory role of the specialized cilia of the apical tuft. As the tuft cilia lengthen, not only does the larva experience improved sensory capability, but also enhanced polarity. A sensory role of motile cilia along the side of the body would further increase environmental perception to the whole body of the animal, leading to better flow sensing and improved proprioception, thereby facilitating complex swimming behaviors.36,37 At 120 hpf, apical tuft cilia show a wide spread of overall lengths and high variability (SEM of 13.55 μm, more than double that of any other age) (Figure 2F; Table S17), indicating the onset of the loss of the apical organ as previously reported21 and indicative of a decreasing need for this sensory structure when the larvae become non-motile. The timing of the loss of this structure coincides with the loss of motility but occurs before settlement. This reinforces the notion that the apical tuft is required for processing sensory information related to dispersal and swimming behavior, rather than settlement, and possibly forms a structure that aids in steering.18,38

Developmental body shape changes correlate with locomotion

Given these observations, we wondered which other aspects of the larval body are changing over the course of development that would enable the animals to swim more linearly and therefore disperse faster and more efficiently when agitated. During our experiments, we frequently noticed that animals performed “shape shifts”. To investigate how the body shape is changing over time, we first measured it in immobilized animals (Figures 3A and 3B). Anatomical elongation can be quantified using the body axis ratio, a measurement where higher ratios between the length and the width indicate a more elongated, slender shape. Indeed, this body axis ratio was found to increase significantly with age (Figure 3B; mean ratio 48 hpf: 1.172, 72 hpf: 1.333, 96 hpf: 1.379, and 120 hpf: 2.082; Table S20).

Figure 3: Body shape changes during development and locomotion behavior(A) Representative images of larvae over the course of development at 48, 72, 96, and 120 hpf at 20×, scale bars, 100 μm.(B) Body axis ratio of larvae at rest significantly increases with age (48–96 hpf n = 31–37, N = 4; 120 hpf n = 31, N = 1). Lines at mean ± SEM; letters denote statistically significant differences in the means by Dunnett’s T3 multiple comparisons. See also Table S20.(C) View of freely swimming larvae through the 2× objective in microscopy assay (Videos S1 and S2), allowing for live body size measurements while observing the swimming behavior. Speed and body axis ratio correlate in example larvae 1, 2, and 3. A behavioral mode analysis can be applied to downsampled swimming traces, categorizing moments within a trace as either straight, swirl, turn, and pause. See STAR Methods for more information.(D) Swimming traces for 48 hpf (n = 70, N = 3), 72 hpf (n = 80, N = 3), 96 hpf (n = 84, N = 2) and 120 hpf (n = 80, N = 3) colored based on the momentary speed; scale bar, 1 mm.(E) Density plots displaying the distribution of swimming speeds per age in the resampled dataset.(F) The resampled swimming traces for 48 hpf (n = 70, N = 3), 72 hpf (n = 80, N = 3), 96 hpf (n = 84, N = 2), and 120 hpf (n = 80, N = 3) colored based on the momentary behavioral mode, scale bar, 1 mm.(G) Bar graphs show the relative percentage of behavioral modes per age in the resampled dataset. Different letters denote statistically different means of p < 0.05 by Dunn’s multiple comparisons test. For more information, see Table S21.(H) Linear regression lines with 95% confidence intervals, and Spearman’s Rank correlation coefficients between the momentary body axis ratio and momentary speed for each behavioral mode, in all ages in the resampled dataset (48 hpf: straight slope = −190.2, turn slope = 456.6, swirl slope = 868.6, pause slope = −8; 72 hpf: straight slope = 508.1, turn slope = 706.5, swirl slope = 1241, pause slope = −10.5; 96 hpf: straight slope = 885, turn slope = 459.4, swirl slope = 707.5, pause slope = 106.1; 120 hpf: straight slope = 156.8, turn slope = −166.6, swirl slope = 144.6, pause slope = 6.1. See also Table S23).(I) Density plots display the distribution of body axis ratios observed for each age in the resampled dataset. X axis is on the log10 scale.

Since it has been reported that cnidarian larvae routinely perform body shape shifts during active behavior,6,9 we wanted to investigate whether muscular contractions dynamically change the body axis ratio during locomotor behavior and if such contractions could be connected to the behavior of the animal. To answer these questions, we developed an assay with sufficient resolution to observe the animal’s swimming behavior concomitantly with body shape changes, using a microscopy-based observation arena (Figure 3C; see example larvae 1, 2, and 3, see also Videos S1 and S2). By integrating basic tracking metrics with information regarding local confinement and turn angles, we systematically classified four behavioral modes in larval swimming traces: “straight”, “swirl”, “turn”, and “pause”. In our assay, body shape, speed, and behavioral mode can then be assigned to the same coordinate, enabling the simultaneous analysis of three parameters for each individual (Figure 3C, right image, for details, see STAR Methods section).

Video S1. Example video from 2× microscopy in Figure 3Larvae at 96 h postfertilization swim freely under the 2× objective.

Video S2. Example video of body measurements on a binary version of Video S1, related to Figure 3A binary version of larvae aged 96 h postfertilization swimming under the 2× objective while taking body size measurements.

In agreement with data from the larger arena, swimming activity and speed are changing over development (Figures 3D and 3E). Additionally, the relative frequency of certain behavioral modes per age is changing when quantified by relative occurrence per track (Figures 3F and 3G). Planula larvae (48–96 hpf) are increasingly swimming “straight” and pause less before these values increase again after metamorphosis (120 hpf) (Figure 3G straight: 48 hpf 16%, 72 hpf 27%, 96 hpf 40%, 120 hpf 30%; Turn: 48 hpf 9%, 72 hpf 19%, 96 hpf 20%, 120 hpf 18%; Swirl: 48 hpf 13%, 72 hpf 23%, 96 hpf 22%, 120 hpf 10%; Pause: 48 hpf 63%, 72 hpf 30%, 96 hpf 19%, 120 hpf 43%. See also Table S21). Locally confined “swirling”, and reorientation behavior such as “turning”, might indicate heightened environmental sampling in search of sensory information. These modes occur more frequently in late planulae (72 hpf: 23% swirl, 19% turn; 96 hpf: 22% swirl, 20% turn), coinciding with increased TRP channel expression and other sensory behaviors as described in Figure 1L.

As both body axis ratio and locomotor patterns are changing over development (Figures 3E–3G and 3I), we were curious as to whether there is a correlative relationship between momentary elongation and momentary swimming speed during each of the behavioral modes (Figure 3H; Table S22). Indeed, longer animals swim, on average, faster during behavioral modes such as “turn”, “swirl” and “straight”, as indicated by the positive correlation in 72 hpf and 96 hpf (straight: 72 hpf _R_2 = 0.25, slope = 508.1; 96 hpf _R_2 = 0.59, slope = 885; turn: 72 hpf _R_2 = 0.49, slope = 706.5; 96 hpf _R_2 = 0.35, slope = 459.4; Swirl: 72 hpf _R_2 = 0.47, slope = 1241; 96 hpf _R_2 = 0.40, slope = 707.5). Interestingly, 48 hpf and metamorphosed (120 hpf) larvae did not swim at high speeds while swimming “straight”. In fact, there is a slight negative relationship between body elongation and speed in 48 hpf (_R_2 = −0.28, slope = -190.2) in this mode, and a less positive relationship in 120 hpf (_R_2 = 0.38, slope = 156.8) compared to 96 hpf. The correlation between body axis ratio and swimming speed during straight swimming is increasing until a maximum value is reached at 96 hpf. Together with the higher frequency of straight swimming behavior, we here find a major determinant of the larvae’s increase in dispersal efficiency at this age: animals coordinate body axis ratio and speed to optimize dispersal through enhanced straightness of the swimming trajectory.

These observations are not just due to larger larvae swimming with greater ease. The total size of the larvae (measured as area) is neither correlated with speed, nor with elongation (Figures S3A and S3B; Tables S23 and S24), showing that the specific body shape is contributing to increased swimming speeds in the larval stage. To our surprise, the body area (animal size) seems to be a determining factor for locomotion, as animals that reach a size threshold of around 60 mm2 (Figure S3B, dashed line) move quite slowly as compared to their smaller siblings. Whether size is a determining factor for metamorphosis or locomotion will have to be determined in future experiments.

How exactly shape changes of the larvae influence their swimming behavior and efficiency is unclear. One possibility is that the ciliary tuft of the apical organ is the reason for the elongation of the aboral area. We observed that larvae in straight and swirling behavior often show some constriction toward the apical tuft and that the tuft is bundled and pointing straight ahead (Figure S3C). This could enhance the polarity of the larvae either for better steering if the tuft serves a paddle function as suggested in38 or for further sensory reach.

Cilia and neurons play distinct roles in swimming behavior

Since the function of the larval cilia (particularly the apical tuft), the nervous system, and their interplay in locomotion remain poorly understood in Nematostella larvae, we were keen on describing these structures concomitantly. By using the same timepoints as the behavioral experiments, we hoped to deduce how anatomical differences correlate with the observed behavioral changes. To this end, we performed immunohistochemistry of larvae aged 48 hpf–120 hpf with anti-DsRed in the transgenic Elav::mOrange line that selectively labels neurons (Figure 4A)40 and with anti-acetylated tubulin to show the ciliary structures (Figure 4B). In addition to the histology, we re-analyzed single-cell transcriptomic data at relevant developmental stages39 to understand how cellular and molecular components endow the animal with distinct functional subsystems that enable a more controlled swimming pattern over time (Figures 4C–4E, S4B, and S4C).

Figure 4: Nervous system and cilia are distinct subsystems involved in swimming behavior(A) Elav::mOrange larvae stained with dsRed between 48 and 120 hpf. Arrow indicates nerve tracks; scale bars, 20 μm.(B) Ciliation of larvae between 48 and 120 hpf visualized using acetylated tubulin (yellow) and DAPI for nuclear stain (dark blue), (OE = oral end, AOE = aboral end, AT = apical tuft, scale bars, 20 μm).(C) Expression profiles of marker genes. Data show relative average expression values across developmental time.(D) Two-dimensional representation of single-cell transcriptional landscape at age td4 (∼96 hpf). Data points represent single cells labeled by cell groups reported in ref.39(E) Developmental average expression profiles (left) and 2D projection of single-cell expression values (right) of genes encoding Nav ion channels.(F) Representative image of a 96 hpf pipette-attached larva at 20×, scale bars, 100 μm.(G) Schematic of pipette-assay protocol with different perfused treatments as indicated by arrow and “media exchange”.(H) Comparison of CBFs at CBF 1 and CBF 2 video timepoints revealed no significant differences when exposed to the control ASW condition (n = 4, N = 2) or the MgCl2 treatment condition (n = 5, N = 2) for 5 min, where ns p > 0.05 by Wilcoxon matched pairs signed rank test. Lines at mean ± SEM.(I) Ratios of moving and stationary larvae for the control (n = 84, N = 3, 80% moving) and MgCl2-treated (n = 88, N = 3, 66% moving) larvae. Two-tailed two-proportion Z test ∗p = 0.039.(J) Distributions of maximum speeds between the control (n = 67, N = 3) and MgCl2 treated (n = 58, N = 3) larvae (moving animals only). Line at the mean, ∗∗p = 0.0016 by Kolmogorov-Smirnov test.(K) Distributions of axis ratios between the control (n = 75, N = 3) and MgCl2 treated (n = 84, N = 3) larvae. Line at the mean, ∗∗∗p < 0.0001 by the Kolmogorov-Smirnov test. See also Tables S25–S29.

We found that the larvae show an increasing number of neurons (Figure S4A), as previously reported41,42 with a small number of ectodermal sensory neurons in the early planula stage and larger numbers in both ecto- and endoderm in the late planula. Finally, clearly visible nerve tracks are found by the time animals are developing prominent tentacle buds (Figure 4A, arrow). Remarkably, we observed that the largest number of sensory neurons in the aboral area appears only in the 120 hpf stage (Figure 4A, right), suggesting that older larvae have an increased need for sampling substrates. Nematostella larvae indeed metamorphose before they settle and can reattach for a long period of the early polyp stage.8 In agreement with the literature and our measurements (Figure 2F), the 48 hpf larvae rarely possess an apical tuft. This structure appears later and remains prominent up until 5dpf, when it becomes less dense, and some larvae were observed to have already lost the tuft. Interestingly, when looking in the single-cell dataset from39 using markers for neurons (elavl3), the apical organ (Prd146) and the ectodermal epithelium (GP2) from,21 we found that, while the neuronal marker is increasing in agreement with our immunohistological observations, both the apical organ ciliated cells (Prd146) and some ectodermal cells labeled with GP2 seem to be largely constrained to the motile phase of the larvae (Figures 4C, 4D, and S4B). These observations are consistent with the expression of Nav2.4 as a proxy for the functional capacity of the ciliated epithelial cells. Nav2.4 is strongly expressed in younger larval ages and disappears almost completely after the animals seize swimming activity (Figure 4E). This sodium channel is exclusively found in the larval ectoderm and apical organ cells43 (Figure 4E right image), suggestive of a role in excitability in these ectodermal cells (e.g., ciliary beating, contractility). In contrast to Nav2.4, strong expression of Nav2.1 and 2.5, sodium channels that are characteristic of other excitable cells, such as neurons and cnidocytes, is observed only in later larval stages. These cellular and molecular data suggest that the larvae lose molecular and potentially also cellular components associated with ciliary locomotion. Altogether, these observations make it likely that individual systems play a role at different stages in the animal’s life and, in addition, serve distinct functions with respect to ontogenetically restricted behaviors.

To further test this hypothesis, we blocked synaptic transmission using magnesium chloride (MgCl2)44 and observed how the inhibition of the nervous system affects ciliary beating and swimming behavior. We used 96 hpf larvae, the motile stage with the most developed nervous system, most prominent apical tuft, and highest sensitivity to sensory stimuli. Baseline ciliary beating as measured in larvae fixed by a pipette (Figure 4F) did not change when MgCl2 was applied as compared with control animals, which remained in normal media (Figures 4G and 4H). Cilia in the larvae, therefore, beat without neuronal input and are, at least during basal conditions, not influenced by neuronal activity. Ciliated cells might create their own sensory-motor units, regulating beating frequency intrinsically in response to external information, such as mechanical or light stimuli, rather than being influenced by neuronal input. This finding is supported by previous observations of sensory receptors being expressed directly in epithelial cells in Nematostella planula larvae19,32 and by our expression analysis, which finds a large proportion of the highly expressed TRP channels at 96 hpf in epithelial cells (Figure S4C).

When MgCl2 was applied to freely moving animals, we observed that, despite the lack of an obvious change in CBF, fewer animals were among the actively swimming proportion (66% moving) compared to the control (80% moving) (Figure 4I, for the definition of moving vs. stationary, see STAR Methods). While the reduction of motile animals was significant, the majority of the planula larvae remained activated through arousal by handling. This result strengthens the theory of cilia as independent sensory-motor units and shows that a fully functioning nervous system is not basally required for swimming but rather constitutes an additional layer of locomotor control.

To further investigate what this role could be, we looked more carefully at the swimming and body shape differences between control and magnesium-treated animals. Magnesium-treated animals showed substantial changes in both parameters (Figures 4J and 4K). Actively swimming larvae (movers only) (Figures 4J and S4D) swam faster (median mean speed 0.88 mm/s) as compared to the control (median mean speed 0.63 mm/s). When plotting the distribution of the speed quartiles of all animals (Figure S4E), we found that magnesium-treated animals were essentially locked in two speed modes: either stationary (Q1) or at high speed (Q4), and only a few larvae were found in the intermediate speed modes.

Nematostella larvae do not show changes in the range between maximum and minimum speed, but rather appear to acquire the ability to utilize their body more efficiently over developmental time. Locomotion gait is mediated by matching internal and external sensory information to reach optimal swimming patterns. Therefore, larval swimming gait would include proprioceptive responses and appropriate adjustment of the body shape to a given external sensory environment. Indeed, when plotting the axis ratio and mean speed of the animals, it became evident that the inhibition of synaptic transmission leads to a disconnect of these parameters (Figure S4F). In particular, the relationship between speed, body axis ratio, and behavioral mode showed that neuronal activity is: 1. important for the positive correlation of these parameters in “straight” and “turn” behaviors in particular and 2. independent from the value of the axis ratio (swirl and pause slope are not affected but AR is) (Figure S4G).

Swimming gait is important for maximal locomotor efficiency and endows the animal with the capacity to fine-tune its behavior in agreement with external and internally collected sensory information, a process that seems to rely on functional synaptic transmission. Reorganizing the body shape through muscular contractions can influence ciliary orientation and consequently lead to speed modifications induced by flow field changes. These changes will in turn influence proprioceptive signaling, helping the animal to assess its behavior and orient itself with respect to the environment.37