Siphonophores are ubiquitous and often highly abundant members of pelagic ecosystems throughout the open ocean. They are unique among animal taxa in that many species use multiple jets for propulsion. Little is known about the kinematics of the individual jets produced by nectophores (the swimming bells of siphonophores) or whether the jets are coordinated during normal swimming behavior. Using remotely operated vehicles and SCUBA, we video recorded the swimming behavior of several physonect species in their natural environment. The pulsed kinematics of the individual nectophores that comprise the siphonophore nectosome were quantified and, based on these kinematics, we examined the coordination of adjacent nectophores. We found that, for the five species considered, nectophores located along the same side of the nectosomal axis (i.e. axially aligned) were coordinated and their timing was offset such that they pulsed metachronally. However, this level of coordination did not extend across the nectosome and no coordination was evident between nectophores on opposite sides of the nectosomal axis. For most species, the metachronal contraction waves of nectophores were initiated by the apical nectophores and traveled dorsally. However, the metachronal wave of Apolemia rubriversa traveled in the opposite direction. Although nectophore groups on opposite sides of the nectosome were not coordinated, they pulsed with similar frequencies. This enabled siphonophores to maintain relatively linear trajectories during swimming. The timing and characteristics of the metachronal coordination of pulsed jets affects how the jet wakes interact and may provide important insight into how interacting jets may be optimized for efficient propulsion.

Propulsion patterns are closely linked to the ecology of pelagic organisms. Therefore, to fully understand the foraging behavior and trophic role of planktonic organisms, is it important to quantify their propulsion. Several different propulsive strategies exist among the plankton, including rowing and jet propulsion by medusae (Colin and Costello, 2002; Costello et al., 2008), flapping by pteropods (Murphy et al., 2016), paddling by krill, shrimp and ctenophores (Colin et al., 2020; Murphy et al., 2013) and jetting by copepods (Jiang and Kiørboe, 2011; Yen, 2000). Many of these planktonic animals use multiple propulsors, such as the swimming legs of krill and copepods or ctenes of ctenophores, and most often the timing of the kinematics of these propulsors is closely coordinated in the form of a metachronal wave (Colin et al., 2020).

Physonect siphonophores are colonial cnidarians that use multiple jetting bells, called nectophores (Fig. 1A,B), for propulsion. The nectophores are genetically identical clones and are arranged to form a coherent unit called the nectosome (Fig. 1A,B). Nectophores use muscular contractions to create jets of water that propel the whole colony through the water column (see Movies 1–3). During whole-colony swimming, the nectosome pulls the feeding and reproductive colony members that comprise the siphosome (Fig. 1A). The nectophores within a nectosome use a primitive nerve-net for neuro-muscular control (Mackie, 1964; Mackie et al., 1988). Despite having simple neural control, siphonophores exhibit different, sometimes complex, swimming behaviors. During escape swimming, siphonophores contract all of their nectophores simultaneously (termed synchronous swimming) to produce maximum thrust and evade a potential predator (Costello et al., 2015; Du Clos et al., 2022; Mackie, 1964). However, during routine swimming, used for migrations or repositioning between feeding bouts, siphonophores pulse their nectophores at different times (termed asynchronous swimming; see Movies 1–3). How well or even whether these nectophores coordinate their jets during routine swimming is not known. Further, the capability of the relatively simple neural system to control and coordinate the nectophores during swimming is little understood.

Physonectae is a suborder of siphonophores that are highly diverse morphologically, functionally and ecologically. Both the number and morphology of the nectophores vary greatly among physonects, and these differences impact the swimming proficiency and efficiency of the nectosome (Du Clos et al., 2022; Jiang et al., 2021) and, by extension, how the nectosomes function. However, no data exist on the relationships of morphology with swimming ability or behavioral strategies of siphonophores. This is primarily because of the limited number of species that can be effectively captured for analysis, but also because of the high diversity in morphology among the different species of siphonophores. Furthermore, the complexities of the multi-jet propulsion performed by these organisms pose a challenge in terms of understanding the function of these species. Understanding how these siphonophores use their nectophores to swim is an important step in developing a functional understanding of their diverse ecological roles and impacts.

The goal of this study was to investigate how morphologically distinct physonect siphonophores time their jetting nectophores for normal, straight, non-escape swimming and to determine whether the jets are coordinated in an organized manner. Normal swimming is defined as the behavior observed when siphonophores are swimming between feeding bouts or for migrations and has been characterized by the nectophores contracting at different times (Barham, 1966; Purcell, 1980). This behavior may involve both straight swimming and turns. Turning by physonects is primarily controlled by apical (or leading) nectophores (Costello et al., 2015). However, we only analyzed straight swimming bouts. In contrast to normal swimming, escape swimming is defined as the behavior siphonophores exhibit when disturbed and is characterized by all the jets contracting simultaneously (termed synchronous swimming; Mackie, 1964; Costello et al., 2015; Choy, 2017). We observed normal swimming using in situ video recordings of different species during routine swimming – non-escape swimming in the forward direction such that the siphosome is trailing behind the nectosome and moving in a straight line. Nectophore kinematics were quantified and compared for six physonect species: galaxy siphonophore (not yet described species; see Movie 3), Apolemia rubriversa (see Movie 1), Nanomia spp. (2 locations), Forskalia sp. and Agalma okeni (see Movie 2). These species vary morphologically, in both their nectophore and siphosome size, and their behaviors have yet to be quantified. Their morphologies and distributional data suggest that they also differ behaviorally (Mapstone, 2014; Munro et al., 2018). Unfortunately, we do not know much about the feeding and migratory behavior of the different taxa. Therefore, this study is a first step in understanding the functional morphology of siphonophore nectophores that will help inform our future understanding of their feeding and migratory behavior.

Kinematic data collection

Nectophore kinematics of the physonect species were quantified from videos during ‘normal’ swimming. Most of the data were collected from animals swimming in the field at two locations (Monterey Bay, CA, USA, and Friday Harbor, WA, USA); one was from the laboratory at a third location (Santa Catalina, Panama; Table 1). Field observations were collected using SCUBA or a remotely operated vehicle (ROV). ROV videos were captured using the ROV MiniROV in the Monterey Bay National Marine Sanctuary near Midwater Station 1 (latitude:36°41.8792N, longitude: 122°2.9929W), with bottom depths exceeding 400 m. Multiple dives with the ROV were made in the spring and summer of 2019 and 2020, and in the autumn of 2019. MiniROV is a small, 1500 m-rated electric ROV platform, designed to minimize acoustic and hydrodynamic disturbances so as to minimize organismal disruptions and behavior during observations. The vehicle is equipped with a main science camera (30 frames s−1, 1980×1080 pixels; Insite Pacific Incorporated Mini Zeus II), a stereo imaging system (Katija et al., 2021; Allied Vision G-319B monochrome cameras and Marine Imaging Technologies underwater housings with domed-glass optical ports), a pair of red lights (Deep Sea Power and Light MultiRay LED Sealite 2025 at 650–670 nm), and additional vehicle sensors. Red illumination was used whenever possible to minimize disruptions and changes in animal behavior (e.g. avoidance, attraction, escape).

Nectophore kinematics videos of Nanomia bijuga were collected in Friday Harbor by SCUBA divers using a FASTEC TS5 (250 frames s−1, 2560×2046 pixels) and brightfield illumination (Costello et al., 2015). Laboratory videos of Nanomia sp. from Panama were taken of individuals freshly hand collected by SCUBA and placed into a large gently circulating kreisel tank. Nanomia sp. from Panama are thought to be a different species from Nanomia bijuga but have not been fully described [S. Haddock, Monterey Bay Aquarium Research Institute (MBARI), personal communication]. Individuals in the tank appeared to have normal field behavior, switching between resting with tentacles deployed and swimming to reposition. Swimming to reposition was the behavior that we defined as routine swimming for subsequent kinematics analysis – non-escape swimming in the forward direction such that the siphosome is trailing behind the nectosome and moving in a straight line.

Kinematic data analysis

The videos were analyzed for multiple variables based on nectophore jetting (Table 1). Nectophores were numbered so they could be differentiated and labeled consistently throughout the study (Fig. 1B). For species where the nectophores are arranged bilaterally along a plane, we sequentially labeled the nectophores along each side of the nectosome from the apex to the posterior end. The nectophores were also labeled for whether they appeared in the video to be arranged along the left or right side of the nectosome. As the siphonophores were swimming straight and not spinning, the nectophores remained on the same side throughout the clip. The videos were observed frame by frame, and the contraction and relaxation of each individual nectophore was quantified through time. For each replicate individual, we analyzed a video sequence that consisted of at least 10 nectophore contraction–relaxation cycles (also termed swim cycles). Greater than 5 replicate individuals were analyzed per species (Table 1). The galaxy siphonophore, a very rarely observed and undescribed species, was the exception. This individual was pseudo-replicated (n=7) by analyzing different sequences (each consisting of 10 contraction cycles and lasting about 10 s in duration) that were separated in time by a minute throughout our single 10 min recording.

Four kinematic parameters were quantified: jet frequency, interpulse duration, pulse duration and offset time. Jetting frequency was quantified as the number of jets per second each nectophore performed. The interpulse duration (or time between pulses) was the amount of time in seconds that the nectophore was relaxed and not jetting, or the time between the sequential jets of one nectophore. The pulse duration was the time in seconds that each nectophore was contracted. In addition to individual nectophore kinematics, the timing of contractions of adjacent nectophores or offset time was analyzed. Offset was defined as the amount of time between jets of adjacent consecutive nectophores along a side of the siphonophore (termed axially aligned; e.g. nectophores 1L versus 2L or 2L versus 3L or 3L versus 4L in Fig. 1B) or between jets of nectophore on opposite sides of the nectosome (left versus right side; e.g. nectophores 1L versus 1R in Fig. 1B). This time was determined by subtracting the jet start time of each nectophore from the start of the nectophore contraction immediately behind it (for axial offset) or across from it (for left–right offset) on the nectosome. The values for each of these four characteristics were averaged together for all of the individuals of one species and for the left and right nectophores (for all five species, it was found that there was no significant differences between the right and left sides; t-test, P>0.05, P=0.0566–0.9923).

To evaluate how kinematic parameters scale with nectophore size, we measured the length of the longest axis of all the nectophores that comprised the nectosome and took the mean of these lengths to represent size.

The ambient temperature during filming of most of the siphonophores was 8°C except for Nanomia sp. from Panama, for which the ambient temperature was 29°C. Therefore, we calculated the Q10 of Nanomia spp. by comparing N. bijuga (8°C at Friday Harbor) with Nanomia sp. (29°C at Panama) using the equation:
(1)
where R is the pulse duration (which assumes that the two Nanomia populations have the same pulse duration – a reasonable assumption because the nectophores are the same size). We adjusted the rate processes for our comparison of contraction kinematics versus nectophore size by solving for R and adjusting the temperature T.

Statistical analysis

Jetting kinematic parameters were compared among species using single-factor ANOVA. All the data conformed to the assumption of normality; however, in some cases the data were square root transformed to achieve equal variance. The Tukey–Kramer method was used for post hoc comparisons. Kinematic comparisons between nectophores on either side of the nectosome used t-tests. If these kinematics were not significantly different, then all nectophores on a nectosome were pooled for comparisons among species.

Particle image velocimetry

To examine the hydrodynamics of adjacent, asynchronous jets, we used laser sheet particle image velocimetry (PIV) on N. bijuga from Friday Harbor, WA, USA. Individual siphonophores were hand-collected from the docks surrounding the Friday Harbor laboratories, and immediately transported to the laboratory for PIV videography (Sutherland et al., 2019a ). Individuals were placed into 8×8×15 cm glass filming vessels with filtered seawater that was seeded with 10 mm hollow glass beads. The vessel was large enough to minimize wall effects in the vicinity of the nectosome where the jets emerge (greater than 10× the width of the nectosome). The vessel was illuminated with <1 mm thick laser sheet (532 nm) and N. bijuga was recorded swimming in the laser sheet at 6400 frames s−1 (1024×1024 pixels) using a Photron AX200. It is rare for N. bijuga to swim asynchronously when it is in the PIV filming vessel; however, we did record one sequence where an individual swam asynchronously from rest. In this sequence, the plane of the laser passed through the middle of the nectophore and the jet wake. Image pairs of this sequence were analyzed using a cross-correlation PIV algorithm with multi-pass interrogation windows of decreasing size (64 to 32 pixels) and 50% overlap (LaVision DaVis 8.3). At 6400 frames s−1, there was no streaking in the jet wake.

During swimming, all the nectophores along the nectosome contract and relax with repeated cycles over time (Fig. 2A,B; Movie 1). Fig. 2 shows the raw cycle data for the galaxy siphonophore (Fig. 2A–C) and for Apolemia rubriversa (Fig. 2D,E). The raw data illustrate how each nectophore continues its own contraction–relaxation cycle over time. In addition, they show that nectophores contract and relax asynchronously and, qualitatively, nectophore contraction–relaxation cycles appear to be coordinated at some parts of the nectosome (seen as regular patterns in the raw data). Interestingly, for the galaxy siphonophore, the anterior four nectophores appear less coordinated than the posterior six nectophores (Fig. 2B versus C). Nectophore development occurs at the apex (i.e. apical nectophores are less mature than dorsal), and this developmental pattern may affect their ability to coordinate between adjacent nectophores. In addition, apical nectophores have been shown to be primarily used for maneuvering in N. bijuga and may not coordinate for steady-state straight swimming (Costello et al., 2015).

Kinematic parameters across all five species reveal that some parameters are much less variable than others. Pulse duration was the most consistent kinematic parameter among nectophores (Fig. S1), meaning that the jets produced by the nectophores within a species do not substantially change over time for normal swimming. However, kinematic parameters that quantify timing between adjacent nectophores (e.g. interpulse duration and offset time) and the time between successive pulses of any one nectophore (e.g. jetting frequency) were much less consistent (as indicated by larger error bars in Fig. S1).

None of the kinematic parameters differed significantly among the nectophores for each species (ANOVA comparison of individual nectophores within each nectosome, P>0.05 for all species). Therefore, we averaged the kinematic parameters among all the nectophores of each individual to compare the parameters among species. Pulse durations were highly consistent within each species; however, they differed significantly between species (single-factor ANOVA on log-transformed data, P<0.05; Tukey–Kramer post hoc test, P<0.05; Fig. 3A). The pulse duration for N. bijuga from Friday Harbor was more than double the duration of Nanomia sp. from Panama. However, the water temperature in Friday Harbor was 21°C colder than in Panama (8°C versus 29°C). Based on this temperature difference, we calculated a Q10=1.58±0.09 (assuming the two populations have the same pulse duration when at the same temperature). To examine how contraction durations related to the nectophore size of the different species, we adjusted the rates of Nanomia sp. from Panama based on the calculated Q10 to 8°C (the ambient temperature of the other species). Pulse durations were strongly related to nectophore length [regression P<0.05, R2=0.93; Fig. 3B (regression for non-temperature-adjusted data had a slightly higher R2=0.95)]. Interpulse duration also varied among species (Fig. 3C; single-factor ANOVA on log-transformed data, P<0.05); however, the duration between pulses was not as tightly dependent on nectophore length as pulse duration (Fig. 3D). The jet frequency also varied among species (Fig. 3E; single-factor ANOVA on log-transformed data, P<0.05). While the two species with the largest nectophores had significantly lower frequencies (Tukey–Kramer post hoc test, P<0.05), the regression with nectophore length among all the species was not significant (regression analysis, P>0.05).

To examine whether adjacent nectophores on opposite sides of the nectosome coordinated their pulses, we compared the offset times between left and right nectophores for the galaxy siphonophore, Nanomia sp. (Panama) and A. rubriversa (Fig. S2); these were the only animals where we could resolve both sides of the nectosome (Fig. 4). A negative versus positive offset meant that the left nectophore pulsed before versus after the right nectophore, respectively. The offset times between the left and right nectophores were highly inconsistent for all three species examined (Fig. 4). There was no consistent pattern for which nectophore pulsed first, with the left nectophores sometimes pulsing before the right, and also pulsing after the right just as often. In addition, the offset times were highly variable; occasionally, the offset was equal to or greater than the interpulse duration of individual nectophores. The offset for Nanomia sp. was shorter than the interpulse duration, indicating that the left and right nectophores of Nanomia sp. pulsed more simultaneously. However, there still was no pattern for which nectophore pulsed first (left versus right), making it difficult to conclude that they are coordinated. Consequently, the overall results indicate that nectophores on opposing sides of the nectosomes do not appear to be coordinated. However, a comparison of the pulse frequency of nectophores on each side showed that the nectophores along opposing sides of the nectosome pulsed with the same frequency (Fig. 5). Therefore, the same amount of thrust would be produced on each side of the nectosome, enabling the siphonophore to maintain a straight trajectory.

In contrast, nectophores aligned axially along the same side of the nectosome did appear to be consistently coordinated (Figs 2 and 4). To illustrate the coordination compared with the lack of coordination between the left and right nectophores, we similarly plotted the offset of adjacent nectophores along the same side of the nectosome, and these plots showed much less variable offset times. The leading nectophore always pulsed before the trailing nectophore (except for A. rubriversa, which had the opposite sequence), and the offset was much shorter than the interpulse duration. The consistent offset times indicated that the nectophores were coordinated metachronally. A positive offset time meant that the metachronal wave traveled from the apex to the dorsal end of the nectosome (Fig. 4). The negative offset times of A. rubriversa indicated that the wave traveled from the dorsal end to the apex. The offset times varied among the species, and appeared to be related to the pulse duration of each species (ANOVA, P<0.02, Tukey–Kramer post hoc comparison, P<0.05; Fig. 6A). The slope of the line of the offset time versus the pulse duration was 0.45, indicating that the adjacent nectophores initiated their pulses about halfway through the pulse of the previous nectophore (Fig. 6B).

PIV of adjacent jets, where the trailing nectophore initiated its contraction halfway through the pulse of the leading nectophore, shows that the wakes of the two jets eventually interact downstream from the nectophore orifice (Fig. 7). Very few PIV sequences of asynchronous jets exist because, in captivity, a siphonophore will exhibit mostly reactive escape swimming with synchronized jets. In their natural environment, as our data show, siphonophores use asynchronous jets during normal swimming. In Fig. 7 the siphonophore was starting from rest (velocity v=0) and its swimming speed was less than that observed in the field. This would result in the jet wake being further apart and interacting less than during steady-state swimming. Despite this, we can see that the leading jet is fully developed when the trailing jet is initiated. This pattern was repeated consistently although we note the limited extent of our PIV dataset. Most of the flow of the leading jet was a well-organized linear jet moving in the same direction as the jet of the trailing nectophore. The jets emerged at a mean±s.d. angle of 51.1±0.03 deg.

Multi-jet swimming is a relatively rare propulsive strategy for animals, and is observed for only a few known taxa, including salps and siphonophores. Both taxa appear to synchronize their jets to escape predators (Mackie, 1964; Sutherland et al., 2017, 2019b), which maximizes thrust production and acceleration. However, for routine swimming that is often used for feeding and repositioning, the jets of both taxa are timed asynchronously (Costello et al., 2015; Sutherland et al., 2017). Asynchronous jetting has been shown to produce less thrust and lower swimming speeds, but is more efficient than synchronous jetting. Efficiency is thought to be enhanced because asynchronous jetting produces steadier swimming speeds (i.e. fewer accelerations and decelerations), which serves to reduce drag (Du Clos et al., 2022; Sutherland et al., 2017). However, at the intermediate Reynolds numbers (Re) experienced by siphonophores and salps, other properties of the jets (e.g. timing and jet angle) can have important effects on drag and efficiency (Jiang et al., 2021). Previous studies have shown that the timing of most salp jets is uncoordinated (Sutherland et al., 2017), while siphonophore jets – mostly based on observations of N. bijuga – appear to be coordinated (Du Clos et al., 2022). We show that for several physonect siphonophores, nectophores on the same side of the nectosome are well coordinated in a metachronal wave pattern. However, nectophores on opposing sides of the nectosome do not appear to be coordinated with each other. Metachronal waves are typically initiated near the apex and travel dorsally along the nectosome. Apolemia rubriversa is an exception to this pattern because metachronal waves move from the nectosome base towards the apex. For all the species studied, the nectophore pulses were not always tightly coordinated and some nectophores either missed a wave or pulsed out of phase with their neighbors.

Metachronal coordination of propulsors is observed in many different taxa that use multiple propulsors for swimming, such as copepods, krill, polychaetes, shrimp and ctenophores. However, the metachronal waves used by most of these animals are antiplectic metachronal waves where the wave is initiated by the posterior-most limb/ctene and the wave travels toward the anterior or leading end of the swimmer (Colin et al., 2020). Therefore, the wave moves in the same direction that the animal is swimming. It has been suggested that use of antiplectic coordination is widespread, because such kinematics enable each propulsor to move without interfering with the adjacent propulsors (Sleigh and Barlow, 1980). In addition, in paddling animals, antiplectic kinematics may rely upon the negative pressure fields produced along their limbs that may assist in the movement and coordination of the limbs (Colin et al., 2020). However, the metachronal coordination of jets of all of the siphonophores in the current study, with the exception of A. rubriversa, was symplectic where the wave traveled in the opposite direction to the swimming nectosome from the apex to the posterior end. The different metachronal kinematics may arise from the differences in the hydrodynamic environments created by pulsed jets versus paddling limbs.

The advantages of metachronal coordination by nectophores remain unexplored. Wake interactions may influence efficiency or thrust production of the nectosome; however, this has been rarely examined. Among the few existing studies, most show that interactions between jets can reduce the thrust produced by a jet. For example, it has been shown that both thrust and efficiency of jets decreased as two synchronized pulsed jets moved closer together and interacted more (Athanassiadis and Hart, 2016). In addition, it is known that the flow environment surrounding pulsed jets further affects the thrust production where parallel co-flow decreases thrust, and counter-flow may increase thrust (Dabiri and Gharib, 2004; Krueger et al., 2003). Evidence from our current study demonstrates that adjacent nectophores generally initiate their pulse halfway through the jet of the previous nectophore, and, based on our observation of PIV of one jet starting from rest, the leading jet wake is linear in the co-flow direction (i.e. same direction) of the adjacent nectophore when it initiates jetting (Fig. 6; t=0.004 and 0.014 s). The wake of the second jet experiencing co-flow with an adjacent jet should therefore experience reduced thrust production (Dabiri and Gharib, 2004).

The orientation of jets, in addition to their timing, plays an important role in thrust and efficiency (Jiang et al., 2021) because as jet angles increase (jets directed more laterally relative to the nectosome axis), their wakes interact less. Jiang et al. (2021) found that two optima of jet angles exist depending upon the jetting parameters that are optimized. Larger jet angles (61–70 deg) optimized the quasipropulsive efficiency because these larger angles had fewer jet interactions and required less overall jet power to swim. Alternatively, smaller jet angles (34–45 deg) optimized Froude efficiency because jet interactions minimized wake kinetic energy, but required more jet power to overcome the drag associated with greater wake interactions. Interestingly, the wakes of many multi-jet salps appear to optimize quasipropulsive efficiency by directing their jets more laterally (Sutherland et al., 2017). This is intuitively appealing for salps that primarily swim to process fluid for feeding, and produce less powerful jets than siphonophores. In contrast, the sole role of siphonophore jets is to provide thrust for colony transportation. The jets of many siphonophores are angled more longitudinally with angles ≤50 deg (Costello et al., 2015). These angles produce more wake interactions but result in higher Froude efficiencies. Therefore, the metachronal timing of the jets may serve to further optimize the efficiency of jet interactions.

Contraction time of individual nectophores also affects the thrust produced by each propulsive unit. The jet thrust emerging from a nectophore is determined by the mass and acceleration of the ejected fluid, and the nectophore size, orifice diameter and contraction rate are the primary factors that affect the mass efflux from the nectophore (Colin and Costello, 2002; Daniel, 1983). We found that contraction times among the physonect species were directly related to nectophore size (measured as nectophore length) (Fig. 3B). Longer contraction times serve to decrease acceleration, but larger sizes should also eject a greater volume of fluid. We can estimate the net effect of nectophore size on thrust (T) as (ρ/Av)(dV/dt)2 where ρ is the density of seawater, Av is the area of the aperture opening, V is the volume inside the nectophore and t is the contraction duration (Colin and Costello, 2002; Daniel, 1983). Therefore, if we assume that Av scales with nectophore length and percentage dV is constant among the species, we can estimate thrust based on the morphology of N. bijuga. Based on this, we estimated that thrust increased with nectophore length (x in mm) by the relationship T (µN)=9.1×−19.9 (with an R2=0.97). Therefore, despite the longer contraction times, thrust appears to increase with nectophore length by a factor of almost 10 due primarily to the greater fluid volumes ejected by larger nectophores. This indicates that both the number and length of nectophores comprising the nectosome will directly increase thrust produced by different siphonophores species. Interestingly, the two species with the largest nectophores, A. rubriversa and the galaxy siphonophore, have nectophores with lengths greater than double that of the other species (Fig. 3) and they have siphosomes that are much longer and with much greater mass than those of the other species (for examples, see Movies 1 and 3). This is consistent with the need for producing more thrust to pull larger siphosomes through the water column and suggests that nectophore size may be directly related to siphosome biomass.

Although nectophores along the same side of the nectosomal axis are generally metachronally coordinated, the degree of coordination can be inconsistent because of variability in contraction timing by adjacent nectophores. The loose coordination may be the result of the simple neuromuscular system of cnidarian siphonophores. The nectophores are attached to a central nectosome stem through which the lower nerve tract travels; the lower nerve tract then connects the nectophores and stimulates forward swimming. However, the speed and fidelity of neural signal transfer are limited for this comparatively simple neural system (Mackie, 1964; Norekian and Meech, 2020). Despite their neural simplicity, cnidarians such as siphonophores and medusae are able to swim effectively and their level of neuromuscular control appears adequate for the behavior requirements of swimming. In other words, while the coordination of nectophores is not perfect, it may be ‘good enough’ to achieve the behavioral requirements of siphonophore swimming. The limited behavioral observations of siphonophores have shown that they alternate between feeding bouts of motionless drifting with repositioning bouts requiring movement to a new location (Madin, 1988). The asynchronous kinematics we observed occurred during repositioning bouts. Even though the nectophores on opposite sides of the nectosome were not synchronized, the colonies were able to maintain relatively linear trajectories because the pulse frequencies of nectophores on the two sides of the nectosome were equal. Coordination between sides along the nectosome does not appear necessary for achieving a linear heading. Likewise, the loose metachronal coordination of axial nectophores is adequate for effective repositioning of a siphonophore moving to a new feeding location.

Medusae demonstrate that, for individual cnidarians, propulsive mode is strongly related to foraging strategy (Costello et al., 2008; Dabiri et al., 2010). For individual medusae, ambushing species optimize thrust production over efficiency while feeding current-foraging medusae optimize efficiency over thrust. We also believe that similar relationships may exist for siphonophores. Some siphonophores are known to migrate over long distances (Pages and Gili, 2014), while others are known to swim much more continuously than others (J.H.C., J.D., K.K. and S.P.C., unpublished data). Nectophore design and operation are potential avenues for siphonophores to modulate tradeoffs between propulsive thrust and efficiency. Both thrust and efficiency (Jiang et al., 2021; Du Clos et al., 2022) are affected by nectophore jet interactions, and our results clearly demonstrate the ability of siphonophores to coordinate jets and therefore potentially the jet interactions. Although the complexities of coordinating multi-jet propulsion are substantial, comparison of multiple species allows us to establish a framework of how siphonophore colonies successfully accomplish this challenge.

We would like to thank Brad Gemmell and Kelly Sutherland for enlightened discussions about siphonophore behavior and propulsion, and the reviewers for their contribution to improving the manuscript.

Author contributions

Conceptualization: J.H.C., J.D., K.K., S.P.C.; Formal analysis: S.S., S.P.C.; Data curation: J.D., S.P.C.; Writing - original draft: S.P.C.; Writing - review & editing: S.S., J.H.C., J.D., K.K., S.P.C.; Visualization: S.S.; Funding acquisition: J.H.C., K.K., S.P.C.

Funding

This research was supported by the David and Lucile Packard Foundation and by the National Science Foundation [NSF-IOS 2114169 (S.P.C.), 2114171 (J.H.C.), 2114170 (K.K.) and NSF-CBET 2100156 (S.P.C.), 2100705 (J.H.C.)].

Data availability

All relevant data can be found within the article and its supplementary information.

Athanassiadis
,
A. G.
and
Hart
,
D. P.
(
2016
).
Effects of multi-jet coupling on propulsive performance in underwater pulsed jets
.
arXiv
1604.05382.
Barham
,
E. G.
(
1966
).
Deep scattering layer migration and composition: observations from a diving saucer
.
Science
151
,
1399
-
1403
.
Choy
,
C. A.
,
Haddock
,
S. H. D.
and
Robison
,
B. H.
(
2017
).
Deep pelagic food web structure as revealed by in situ feeding observations
.
Proc. R. Soc. B: Biol. Sci.
284
,
20172116
.
Colin
,
S. P.
and
Costello
,
J. H.
(
2002
).
Morphology, swimming performance and propulsive mode of six co-occurring hydromedusae
.
J. Exp. Biol.
205
,
427
-
437
.
Colin
,
S. P.
,
Costello
,
J. H.
,
Sutherland
,
K. R.
,
Gemmell
,
B. J.
,
Dabiri
,
J. O.
and
Du Clos
,
K. T.
(
2020
).
The role of suction thrust in the metachronal paddles of swimming invertebrates
.
Sci. Rep.
10
,
17790
.
Costello
,
J. H.
,
Colin
,
S. P. S. P.
and
Dabiri
,
J. O. J. O.
(
2008
).
Medusan morphospace: phylogenetic constraints, biomechanical solutions, and ecological consequences
.
Invertebr. Biol.
127
,
265
-
290
.
Costello
,
J. H.
,
Colin
,
S. P.
,
Gemmell
,
B. J.
,
Dabiri
,
J. O.
and
Sutherland
,
K. R.
(
2015
).
Multi-jet propulsion organized by clonal development in a colonial siphonophore
.
Nat. Commun.
6
,
8158
.
Dabiri
,
J. O.
and
Gharib
,
M.
(
2004
).
Fluid entrainment by isolated vortex rings
.
J. Fluid Mech.
511
,
311
-
331
.
Dabiri
,
J. O. O.
,
Colin
,
S. P. P.
,
Katija
,
K.
and
Costello
,
J. H. H.
(
2010
).
A wake-based correlate of swimming performance and foraging behavior in seven co-occurring jellyfish species
.
J. Exp. Biol.
213
,
1217
-
1225
.
Daniel
,
T. L.
(
1983
).
Mechanics and energetics of medusan jet propulsion
.
Can. J. Zool.
61
,
1406
-
1420
.
Du Clos
,
K. T.
,
Gemmell
,
B. J.
,
Colin
,
S. P.
,
Costello
,
J. H.
,
Dabiri
,
J. O.
and
Sutherland
,
K. R.
(
2022
).
Distributed propulsion enables fast and efficient swimming modes in physonect siphonophores
.
Proc. Natl. Acad. Sci. USA
119
,
e2202494119
.
Jiang
,
H.
and
Kiørboe
,
T.
(
2011
).
Propulsion efficiency and imposed flow fields of a copepod jump
.
J. Exp. Biol.
214
,
476
-
486
.
Jiang
,
H.
,
Costello
,
J. H.
and
Colin
,
S. P.
(
2021
).
Fluid dynamics and efficiency of colonial swimming via multijet propulsion at intermediate Reynolds numbers
.
Phys. Rev. Fluids
6
,
013103
.
Katija
,
K.
,
Roberts
,
P. L. D.
,
Daniels
,
J.
,
Lapides
,
A.
,
Barnard
,
K.
,
Risi
,
M.
,
Ranaan
,
B. Y.
,
Woodward
,
B.
and
Takahashi
,
J.
(
2021
).
Visual tracking of deepwater animals using machine learning-controlled robotic underwater vehicles
.
2021 IEEE Winter Conference on Applications of Computer Vision
.
Krueger
,
P. S.
,
Dabiri
,
J. O.
and
Gharib
,
M.
(
2003
).
Vortex ring pinchoff in the presence of simultaneously initiated uniform background co-flow
.
Phys. Fluids
15
,
L49
-
L52
.
Mackie
,
G. O.
(
1964
).
Analysis of locomotion in a siphonophore colony
.
Society
159
,
366
-
391
.
Mackie
,
G. O.
,
Purcell
,
J. E.
and
Pugh
,
P.
(
1988
).
Siphonophore biology
.
Adv. Mar. Biol.
24
,
97
-
262
.
Madin
,
L. P.
(
1988
).
Feeding behavior of tentaculate predators: in situ observations and a conceptual model l. p, Madin
.
Bull. Mar. Sci.
43
,
413
-
429
.
Mapstone
,
G. M.
(
2014
).
Global diversity and review of Siphonophorae (Cnidaria: Hydrozoa)
.
PLoS One
9
,
e87737
.
Munro
,
C.
,
Siebert
,
S.
,
Zapata
,
F.
,
Howison
,
M.
,
Damian-Serrano
,
A.
,
Church
,
S. H.
,
Goetz
,
F. E.
,
Pugh
,
P. R.
,
Haddock
,
S. H. D.
and
Dunn
,
C. W.
(
2018
).
Improved phylogenetic resolution within Siphonophora (Cnidaria) with implications for trait evolution
.
Mol. Phylogenet. Evol.
127
,
823
-
833
.
Murphy
,
D. W.
,
Webster
,
D. R.
and
Yen
,
J.
(
2013
).
The hydrodynamics of hovering in Antarctic krill
.
Limnol. Oceanogr. Fluids Environ.
3
,
240
-
255
.
Murphy
,
D. W.
,
Adhikari
,
D.
,
Webster
,
D. R.
and
Yen
,
J.
(
2016
).
Underwater flight by the planktonic sea butterfly
.
J. Exp. Biol.
219
,
535
-
543
.
Norekian
,
T. P.
and
Meech
,
R. W.
(
2020
).
Structure and function of the nervous system in nectophores of the siphonophore Nanomia bijuga
.
J. Exp. Biol.
223
,
jeb233494
.
Pages
,
F.
and
Gili
,
J.
(
2014
).
Vertical distribution of epipelagic siphonophores at the confluence between Vertical distribution of epipelagic siphonophores at the confluence between Benguela waters and the Angola Current over 48 hours
.
Hydrobiologia
216
,
355
-
362
.
Purcell
,
J. E.
(
1980
).
Influence of siphonophore behavior upon their natural diets: Evidence for aggressive mimicry
.
Science
209
,
1045
-
1046
.
Sleigh
,
M. A.
and
Barlow
,
D.
(
1980
).
Metachronism and control of locomotion in animals with many propulsive structures
. In
Aspects of Animal Movement
, pp.
49
-
67
.
Cambridge University Press
.
Sutherland
,
K. R.
,
Weihs
,
D.
and
Sutherland
,
K. R.
(
2017
).
Hydrodynamic advantages of swimming by salp chains
.
J. R Soc. Interface
14
,
20170298
.
Sutherland
,
K. R.
,
Gemmell
,
B. J.
,
Colin
,
S. P.
and
Costello
,
J. H.
(
2019a
).
Maneuvering performance in the colonial siphonophore, Nanomia bijuga
.
Biomimetics
4
,
62
.
Sutherland
,
K. R.
,
Gemmell
,
B. J.
,
Colin
,
S. P.
and
Costello
,
J. H.
(
2019b
).
Propulsive design principles in a multi-jet siphonophore
.
J. Exp. Biol.
222
,
jeb198242
.
Yen
,
J.
(
2000
).
Life in transition: balancing inertial and viscous forces by planktonic copepods
.
Biol. Bull.
198
,
213
-
224
.

Competing interests

The authors declare no competing or financial interests.

Supplementary information