Hydromedusae

Swimming and Associated Behaviors

Open in new tab

Figure 3. Schematic diagram of a hydrozoan medusa to show the structures that produce swim contractions. V—velum; VCM—velar circular muscle; ONR—outer nerve ring; INR—inner nerve ring; Tri—mesogleal triradius; NME—nonmuscular epithelium; SCM—subumbrellar circular muscle; M—mesoglea of the bell; RC—ring canal. Adapted from Satterlie and Spencer (1983).

Hydromedusae have representatives of both prolate and oblate forms. In both types, swimming involves coordinated contraction of sheets of circular, striated muscle that lines the inside of the bell (subumbrella). The efficiency of swimming is aided by a velum—a ring of contractile tissue that projects at a near-right angle to the subumbrellar walls and narrows the opening of the bell for more efficient propulsion (Figure 3). As noted in the biomechanical studies, the velum is very effective in aiding water jet formation in prolate forms, but relatively ineffective in doing so in oblate forms. The velum has circular, striated muscle on the subumbrellar side, which contracts with the rest of the swim musculature, and radial muscle on the exumbrellar side, which can deform the velum for directional jet formation and turning.

Two other behavioral reactions affect swimming, and while they are similar for all hydromedusae, they show species-specific differences in the details of their neuronal control. Feeding typically involves the transfer of prey from tentacles to the manubrium. Because all species vary in tentacular anatomy, the responses will be different, but all involve regional rather than global reactions of the tentacles and bell. On the other extreme, a protective reaction, called crumpling, involves a global contraction of tentacles and an inward curling of the margin of the bell (Mackie & Singla, 1975; Spencer, 1975, 1978; Singla, 1978b; Mackie, 2004b; Satterlie, 2002, 2008, 2011). Crumpling typically is triggered by injurious stimuli to any parts of the animal. The impact of feeding reactions on the swim system (inhibition or excitation) is species-specific, whereas crumpling always inhibits swimming.

Responses to different types of sensory input vary between species and depend on the unique sensory armament of each species. For example, a variety of statocystlike structures have been described from hydrozoan species (Singla, 1975), and similarly, variations in the presence or absence of eyelike structures have been found (Singla, 1974).

Common Elements of the Swim Systems

Open in new tab

Figure 4. A network of oversized neurons in the inner nerve ring of Polyorchis functions as both pacemakers and motor neurons for swim contractions. A. A single neuron of the network was injected with Lucifer Yellow through a recording microelectrode, and the dye spread throughout the entire network, showing the neurons are dye-coupled. B. Electrophysiological recordings from two neurons of the network show identical production of action potentials. Depolarizing (solid arrow) and hyperpolarizing (open arrow) currents injected into the neuron represented by the lower trace shows electrical coupling between neurons. Panel B adapted from Spencer and Satterlie (1980).

In 1977, intracellular recordings were made from oversized neurons in the inner nerve ring of Polyorchis penicillatus, a prolate hydromedusa (Anderson & Mackie, 1977). Subsequent recordings demonstrated that these neurons comprise an electrically coupled network in the inner nerve ring that activates the swim musculature (Anderson & Mackie, 1977; Singla, 1978b; Spencer & Satterlie, 1980; Spencer, 1981; Satterlie & Spencer, 1983; Satterlie, 1985a, 1985b; Meech & Mackie, 1995; Lin, Gallin, & Spencer, 2001; Figure 4). In addition, the network functions as a pacemaker network for swim contractions (recalling Romanes’s conclusions about the distribution of pacemakers in hydromedusae). Intracellular injections of Lucifer Yellow through a recording electrode revealed that the neurons were not only electrically coupled, but also dye-coupled (Spencer & Satterlie, 1980; Figure 4A). This network performs two functions: that of pacemaker cells and of motor neurons for swim contractions.

Open in new tab

Figure 5. The pacemaker/motor neuron network of the inner nerve ring of Aequorea. A. A dual recording from one of the network neurons and an overlying epithelial cell. A burst of action potentials in the neuron produces a long-duration action potential in the epithelial cell, and a contraction in the swim muscle. B. Neurons of the network are dye-coupled (Lucifer Yellow injection). The dotted circle outlines a statocyst. Adapted from Satterlie (1985b).

Later, recordings were made from a similar network in Aequorea victoria, an oblate hydromedusa; this network also functioned as the swim pacemaker and as motor neurons to the swim musculature (Satterlie, 1985b; Figure 5). A subsequent comparative study of 14 hydromedusan species from several different orders suggested that this network of oversized, electrically coupled, and dye-coupled neurons in the inner nerve ring is a common feature of hydromedusan swim systems, and that the presence of this network is independent of bell shape, present in both prolate and oblate species (Satterlie & Spencer, 1983).

Open in new tab

Figure 6. Swim musculature of hydrozoan medusae. A. The circular, striated muscle of Polyorchis is found at the base of the epitheliomuscular cells. M—mesoglea. B. These muscle cells are joined by desmosomes and gap junctions. The gap junction (arrow) is flanked by two desmosomes. C. A single epithelial cell overlying the inner nerve ring (INR) was injected with Lucifer Yellow through a recording microelectrode, and the dye spread into the epithelial cells and muscle cells of the velum (V) and subumbrella (S). The epithelial cells are postsynaptic to the pacemaker/motor neuron network of the inner nerve ring.

An additional common feature emerged from these comparative investigations. There is widespread electrical and dye coupling throughout circular muscle sheets of the subumbrella and the velum (Mackie & Singla, 1975; Keough & Summers, 1976; Singla, 1978a, 1978b; Spencer, 1979; Anderson, 1980; Satterlie, 1985a, 1985b, 2008; Satterlie & Spencer, 1983; Kerfoot, Mackie, Meech, Roberts, & Singla, 1985; Figure 6). This includes nonmuscular epithelial cells overlying the inner nerve ring, so there is electrical continuity from the velum to the subumbrella. Electrical recordings demonstrated that the epithelial cells were postsynaptic to the identified neuronal network in the inner nerve ring, such that synaptic potentials triggering long duration action potentials were recorded from the epithelial cells (Spencer, 1982; Satterlie, 1985a). Ultrastructural and electrophysiological work showed that these connections are via chemical synaptic transmission. Once initiated, the action potentials spread throughout the subumbrella and velum via the gap junctions, aided in some species by subumbrellar nerve nets. The long-duration action potentials in the epithelial and muscle cells are reflected in long-duration contractions in the swim musculature (Spencer, 1979; Satterlie, 1985a). The action potentials resemble those of another fluid pump, the vertebrate heart. This led to the conclusion that if a fluid pump is attached, it will move the fluid. If the fluid pump is not attached, it will move through the fluid.

One additional common property is related to the feeding and crumpling behaviors of hydromedusae (Mackie & Singla, 1975; Spencer, 1975, 1978; Singla, 1978b; King & Spencer, 1981; Mackie, 2004b; Satterlie, 2002, 2008, 2011). Both responses utilize the activation of subumbrellar radial muscle, although the organization of that radial muscle is species-specific, or at least specific for the two forms of bell structure. This will be addressed later in this article, when discussing individual species.

The common features of hydromedusae include an electrically coupled and dye-coupled network of oversized neurons in the inner nerve ring that function as the pacemakers and motor neurons for the swim system; widespread electrical coupling within the subumbrella tissue, including the circular muscle of the subumbrella and velum and the nonmuscular epithelial cells overlying the nerve ring; and use of the radial muscle to produce feeding and crumpling reactions of the bell. All these features are found in both prolate and oblate species.

Prolate Specializations: The Polyorchis Story

As with other hydromedusae, swim contractions are initiated by a network of electrically coupled neurons in the inner nerve ring (Anderson & Mackie, 1977; Anderson, 1979; Spencer & Satterlie, 1980; Satterlie & Spencer, 1983; Satterlie, 1985b; also see Figure 4). In Polyorchis, this motor network extends up each of the four radii and across the top of each muscle quadrant to surround the muscle sheets (Lin et al., 2001). Each contraction is preceded by a single action potential in this swim network. As noted previously, an action potential in the neurons of the inner nerve ring synaptically activates an action potential in the overlying epithelial cells, which then spreads throughout the subumbrellar muscle sheets and velum (Spencer, 1981, 1982). Presumably, similar synaptic activation occurs through the periphery of each muscle quadrant. The sheet of striated muscle cells in each quadrant is aneural, so spread of an action potential takes place via gap junctions (Spencer, 1978; also see Figure 6). This myoid conduction is relatively slow, so both radial and circular coordination of the four muscle quadrants depend on the faster conduction velocity of the neural network.

Circular coordination of the swim contraction has been well studied in Polyorchis, with a particular focus on conduction in the motor network of the inner nerve ring. Spencer (1981) noted that the action potential duration decreased as it was conducted around the bell from the site of initiation. This is translated as differences in synaptic transmission, and particularly synaptic delays in production of the muscle action potentials (Spencer, 1982). Because synaptic activation at any point depends on multiple inputs from the motor network, synchrony of activation of the neurons in the region influences the neuronal action potential. In the areas distant from the site of action potential initiation, the firing of neurons is more synchronous than in the initiation region, thus producing an action potential of shorter duration than in the former region (Spencer, 1981). This greater synchrony produces larger synaptic potentials and shorter latencies for both the synaptic potentials and for muscle action potential production. The shorter latencies that occur with larger distances from the initiation site promote great synchrony of contraction throughout the bell that partially compensates for conduction time through the motor network (Spencer, 1982).

Crumpling behavior in Polyorchis is accomplished by the contraction of four smooth muscle strips that run from the margin to the manubrium in each of the radii (Spencer, 1978). Their contraction pulls the bell margin upward and inward, and it is associated with inhibition of the motor network, as recorded in the inner nerve ring. This means that at least two separate, parallel conducting systems are localized to the radii of Polyorchis. This organization, compressed nerve nets to form nervelike pathways, represents a simple form of centralization in cnidarian nervous systems. Similarly, multiple, parallel conducting systems in these nervelike pathways add to the concept of cnidarian nervous centralization.

The swim network of Polylorchis is modulated by photic inputs that originate in two separate networks in the outer nerve ring (Spencer & Arkett, 1984; Arkett, 1985; Arkett & Spencer, 1986a, 1986b). The B system is comprised of electrically coupled neurons that are connected to ocelli found at the tentacle bases. During light conditions, the B system produces regular action potential activity. At light-off, a burst of action potentials produces contraction of the tentacles and of the subumbrellar swim muscle. The O system is directly photosensitive and nonspiking, exhibiting membrane potential oscillations when illuminated. At light-off, the system hyperpolarizes and the oscillations cease.

The inner and outer nerve rings contain multiple, parallel conducting systems, with the former including the motor networks for swimming and crumpling and the latter including sensory networks (Spencer & Arkett, 1984). Each hydromedusan species has unique sensory capabilities, highlighted by the variety of photoreceptors (Singla, 1974) and the variety of statocystlike structures (Singla, 1975) found in different medusae. The compression of multiple, parallel conducting system into the two nerve rings, as well as their interconnections, represent the height of neuronal concentration in hydromedusae and the centralized nervous system of the group.

Oblate Specializations: The Aequorea Story

The swim network of Aequorea is unique in that it involves firing a burst of action potentials for each individual swim contraction (Satterlie, 1985b; also see Figure 5). This burst produces a burst of summing synaptic potentials in the overlying epithelial cells (Satterlie, 1985a). As with Polyorchis, the epithelial cells are electrically coupled and dye-coupled with circular muscle cells of the subumbrella and the velum (Figure 6). Several differences are noted in the subumbrellar organization of Aequorea.

The subumbrella is divided into numerous wedge-shaped muscle sheets by radii that include a radial canal and gonad. In addition, each wedge contains more than circular muscle. The subumbrellar epithelium includes a sheet of radially oriented epitheliomuscular cells that lie immediately over the sheet of circular muscle cells (Satterlie, 1985a, 2008). The radial muscle of each wedge contracts during a crumple response. All circular muscle cells are electrically coupled, as in Polyorchis, while the radial muscle cells form a separate, electrically coupled muscle sheet (Figure 6).

Rather than being aneural, the subumbrella of Aequorea has two distinct nerve nets—one associated with the circular swim muscle and the other with the radial muscle sheet (Satterlie, 2008; also see Figure 7). A conducted contraction in the subumbrella thus requires synaptic input from the swim network in the inner nerve ring, input from the subumbrellar nerve net, and electrical coupling between muscle cells. Similarly, the radial muscle contraction requires initiation from the nerve rings, input from its nerve net, and electrical coupling within the muscle sheet.

Open in new tab

Figure 7. Subumbrellar nerve net of Aequorea. This nerve net has a directional orientation, from the margin in a radial direction toward the manubrium.

The oblate bell shape of Aequorea, the arrangement of muscle sheets in many segments, and the organization of radial muscle allow localized radial contractions for feeding purposes, as well as the widespread contractions that occur during a crumple response (Satterlie, 2008). Feeding responses are accompanied by an equally localized inhibition of burst activity in the motor network via hyperpolarizing input that decreases or delays the action potentials in a burst. A more significant hyperpolarization, as well as total inhibition, occur during crumpling.

Aequorea do not have ocelli, and a photic response has not been demonstrated. They do have multiple statocysts distributed around the bell margin, however, suggesting an effective righting response.

It is tempting to speculate on the prolate-oblate dichotomy in the case of the mechanism of activation of swim musculature, with aneural muscle sheets in prolate forms and oblate forms that have subumbrellar nerve nets. Three additional oblate species, Phialidium hemisphericum, Eutonina indicans, and Mitrocomella polydiademata, all have subumbrellar nerve nets (Satterlie & Spencer, 1983). However, all these species are in the order Leptomedusae, and further study of oblate species of the other orders (Anthomedusae, Limnomeduase, Narcomedusae, and Trachymedusae) is needed, as is an analysis of representative prolate species. Also, at least one prolate species has neuronal processes in the subumbrella (see Aglantha, discussed next), but this is a special arrangement, as the processes are neurons that are electrically coupled to giant neurons that run in the radii, so they do not represent a synaptic nerve net (Mackie, 2004a).

A Special Case: Escape Swimming and Giant Neurons in Aglantha

Aglantha digitale is a small, prolate medusa with a unique response to sudden, strong mechanical stimulation. They produce a ballistic swim contraction of the subumbrellar muscle that rapidly propels the medusa away from the stimulus (Donaldson, Mackie, & Roberts, 1980; Roberts & Mackie, 1980). Aglantha thus have two forms of swimming that use the same subumbrellar muscle.

Open in new tab

Figure 8. The radial giant neurons of Aglantha have basal rootlets in the inner nerve ring and a giant neurite that extend up each of the eight radii toward the manubrium. Iontophoretic injection of Lucifer Yellow. The injection point is shown by the narrowing of the neurite. The two photos are of the same cell. A—fluorescence only. B—fluorescence and dim white light illumination to show the nerve ring (nr).

Slow (normal) swimming contractions comply with the generalized structures of hydromedusan swim systems (Mackie, 2004a). An electrically coupled network of neurons in the inner nerve ring acts as the pacemaker and motor neuron for swim contractions. In addition, however, giant neurons with a radial orientation are found in each of the eight radii (called radial giants; Mackie, 2004b; Figure 8). These neurons also conduct motor impulses to the swim muscle during normal swimming in a unique way, described later in this article. In addition to the large-diameter radial process, numerous fine processes extend from neurons that are electrically coupled to the radial giants and run across the muscle sheets in a circular direction. Transmission from radial giants and the coupled neurons to muscle cells takes place via chemical synapses. This is critical because the electrical coupling between muscle cells of the subumbrellar sheets is not strong enough to allow myoid spread of the muscle activation. Also, there is no motor nerve net (MNN) in these subumbrellar regions. The radial giants and their electrically coupled neurons are thus essential for the spreading of the contraction during slow swimming (Singla, 1978a; Weber, Singla, & Kerfoot, 1982; Mackie, 2004b).

Appropriate stimulation of mechanoreceptive bristle pads (tactile combs) at the bases of the tentacles (Arkett & Mackie, 1988) activates another giant neuron that runs around the margin of the animal in the outer nerve ring (called the ring giant; Mackie, 2004b). This cell is electrically coupled to a giant neuron in each of the tentacles. The ring giant also synaptically activates the eight radial giants in the subumbrella, which then produce a forceful contraction of the subumbrellar muscle and a ballistic escape response (Meech & Mackie, 1995). The connection from the ring giant to the radial giants is polysynaptic, suggesting the intervention of an additional system, called the carrier system. This system also mediates the connection from the ring giant to the giant neurons of the tentacles (Mackie, 2004b).

An important question centers on how the radial giants can be involved in both normal and escape swim contractions. The answer is unique and surprising. The radial giants can produce two types of propagated action potentials: a low-threshold calcium spike and a high-threshold sodium spike (Mackie & Meech, 1985). During slow swimming, the calcium spikes are involved in activating the swim musculature and produce relatively weak subumbrellar contractions. The calcium spikes have slower rise times and lower peak voltages than the sodium spikes (Meech & Mackie, 1993a, 1993b). This system uses the swim network of the inner nerve ring, which also synaptically activates muscle sheets of the subumbrella and velum.

Each escape swim contraction can propel the animal five times as far as a single contraction of normal swimming, and the greater contractions are initiated by propagated sodium spikes in the radial giants. The sodium spikes provide greater excitation to the swim musculature, and with a shorter latency, to produce the stronger contractions.

Mechanical stimulation from within the bell does not activate the ring giant, but it still can produce an escape swim. In this case, coordination of the radial giants occurs via a system of rootlet interneurons, which are electrically coupled with the radial giants (Mackie, 2004a). Rootlet interneurons activate each other via chemical synapses, and their location around the bell margin allows activation of the entire system or radial giants to produce the escape swim. The normal pacemaker network for slow swimming produces subthreshold synaptic potentials in rootlet interneurons; however, action potentials in the rootlet interneurons produce 1:1 action potentials in the pacemaker network. In this way, the pacemaker network is also activated during escape swimming.

Overall, 13 conducting systems have been described in the nerve rings and tentacles of Aglantha. This complexity further argues for the concept of a centralized nervous system in hydromedusae and speaks for the integrative capabilities of this nervous organization.

Scyphomedusae: Common Features of Swim Control in Oblate Medusae

As noted by Romanes (1885), the location of pacemakers and the method of muscle activation are very different in scyphomedusae than in hydromedusae. Perhaps the most distinct feature of the scyphozoan swim system is the rhopalium (Figure 9A). These sensory/neural structures house the pacemakers for swim contractions, but they also represent integration centers for rhopalial and extrarhopalial sensory structures/systems (Passano, 1965; Satterlie, 1979; Satterlie & Spencer, 1979). In this way, they can be considered the cnidarian version of ganglia. As noted by Romanes, removal of the rhopalia stops swim contractions.

Open in new tab

Figure 9. The scyphozoan swim system. A. The rhopalium (R) of Aurelia. S—statolith; N—tissue of the rhopalial niche. Labeled with anti-FMRamide antibody. B. The MNN of Aurelia, labeled with an antitubulin antibody (part of a double-label experiment). C. The same area labeled with an anti-FMRFamide antibody, showing neurons of the DNN.

Each rhopalium includes one or more pigmented ocelli, a statolith, and ciliated sensory cells (Hyman, 1940; Horridge, 1956a; Yamasu & Yoshida, 1973; Hundgen & Biela, 1982; Nakanishi, Hartenstein, & Jacobs, 2009; Satterlie & Eichinger, 2014). Outgoing fibers join with a subumbrellar nerve net that activates the swim musculature (the MNN, formerly called the giant fiber nerve net), while incoming fibers from a second nerve net, the diffuse nerve net (DNN), are capable of modulating pacemaker activity and muscle contractility (Figures 9B, C).

The two primary properties of scyphozoan nerve nets help explain the conducting properties originally described by Romanes. The networks allow diffuse conduction that can survive a variety of interdigitating cuts. The morphological organization of neurons in the nerve net demonstrated the diffuse nature of this neuronal organization (Shafer, 1878; Horridge, 1954a, 1954b). In addition, conduction is nonpolarized—the initiation site for an excitation wave can be variable, and the resultant action potential will pass through the nerve net in any direction.

Horridge (1954b) first demonstrated a critical property of individual neurons with regard to the nonpolarized nature of the nerve nets. By dissecting a smaller and smaller bridge within the subumbrella of Aurelia, he was able to produce a bridge that contained a single neuron (visible with substage illumination). He found that the neuron could conduct a contraction wave from one side of the cut to the other, in either direction. That meant that communication between this neuron and other neurons in the intact tissue pieces had to be bidirectional. Subsequently, neurons of the MNN were found to produce individual action potentials that, in turn, activate other neurons, as well as contractions of the circular swim muscle, via chemical synapses (Patton & Passano, 1972; Passano, 1973; Schwab & Anderson, 1980, 1981; Anderson & Schwab, 1981, 1982, 1983, 1984; Anderson & Greenberg, 2001).

Neuroneuronal synapses in this network are symmetrical and are able to transmit synaptic excitation in either direction (Horridge & Mackay, 1962; Horridge, Chapman, & MacKay, 1962; Westfall, 1973, 1996; Anderson & Schwab, 1981; Anderson, 1985; Anderson & Grunert, 1988; Anderson & Trapido-Rosenthal, 2009; also see Figure 10). In this way, neurons communicating by symmetrical chemical synapses can achieve the same diffuse conduction as electrically coupled neurons or muscle cells. Individual neurons of the MNN are not electrically coupled or dye-coupled to one another (Anderson & Schwab, 1981). Similarly, electrical connections have not been found between muscle cells of scyphozoans. To date, gap junctions and gap junction proteins have not been found in any scyphozoans—a clear and basic distinction between scyphozoans and hydrozoans (Mackie, Anderson, & Singla, 1984).

Open in new tab

Figure 10. A symmetrical neuroneuronal synapse in the rhopalial neuropil of Carybdea. Adapted from Eichinger and Satterlie (2014).

The second subumbrellar nerve net, the DNN, is similarly diffuse and nonpolarized (Figure 9C). It conveys sensory information from the periphery to the rhopaliba, but it also can, in some species, modulate swim muscle contractions (Horridge, 1956a; Passano, 1965, 1982, 2004; Satterlie & Eichinger, 2014). This separation of MNN and DNN is supported by different immunohistochemical properties, as the MNN is labeled by antitubulin antibodies and the DNN is labeled with anti-RFamide antibodies (Satterlie & Eichinger, 2014; also see Figures 9B, C). Immunoreactive networks of both types innervate the tentacles and manubrium as well. In the tentacles, the tubulin-immunoreactive fibers are associated with the tentacular musculature, while the RFamide-immunoreactive network includes epithelial sensory cells.

The number of rhopalia in each species differs, with a minimum of eight. In species with larger numbers, the total number is typically in multiples of four (reflecting the body symmetry). A swim contraction can be initiated in any one of the rhopalia, and the activated wave of excitation is believed to reset the other pacemakers (Horridge, 1959; Lerner, Mellen, Waldron, & Factor, 1971). In this way, the rhopalium with the fastest rhythm drives swimming until its rate drops and another rhopalium takes over. Horridge and others examined this apparent pacemaker redundancy and found that each rhopalium had an inherent rhythm that was irregular. In both cutting and modeling experiments, additional pacemakers each contributed to the overall output of the swim systems in two important ways. They produced greater regularity in the overall swim system activity, and the overall rhythm was faster (Horridge, 1959; Lerner et al., 1971).

These two contributions are critical in terms of the properties of the neuromuscular system in scyphomedusae. Neuromuscular transmission occurs via chemical synapses (Westfall, 1973). Furthermore, the contractile activity of the muscles is based on frequency-dependent facilitation (Bullock, 1943; Pantin & Vianna Dias, 1952; Horridge, 1953, 1954b). In this system, the amplitude of muscle contractions depends on the interpulse intervals of the previous contractions. The efficiency of swim contractions thus depends on the number of rhopalia in the network. Reduction of the rhopalial number, either through damage or experimental excision, also reduces the overall rate and regularity of swimming, and therefore its facilitation-based efficiency. Because swimming and feeding are interrelated in scyphomedusae, changes in swimming efficiency can affect feeding efficiency as well.

Species-specific differences in scyphomedusae include the arrangement of the swim musculature and in the organization of marginal and oral tentacles. In Aurelia aurita, the subumbrellar muscle forms an uninterrupted circular sheet, while in Cynea capillata, circular muscle is interrupted by bands of radial muscle (both cocontract during swimming). In both cases, the MNN and DNN are similar in organization. The presence of marginal tentacles and the arrangement of oral tentacles are extremely variable and likely affect swim and feeding mechanics is species-specific ways.

Cubomedusae: Strong Swimmers Built on the Scyphomedusan Plan

Cubozoans were originally classified as an order within the Scyphozoa, but unique features of their development, their body plan, and their neuromuscular organization warranted the establishment of a new class—Cubozoa (Werner, Cutress, & Studebaker, 1971; Werner, Chapman, & Cutress, 1976). This development has been supported by subsequent phylogenetic analyses (Collins, 2002, 2009; Marques & Collins, 2004; Bentlage et al., 2009). Yet in terms of the Romanes dichotomy, they are clearly within the covered-eyed medusae category.

Cubomedusae are strong swimmers, so they can escape dominance by water movements like currents, tides, and waves (Hamner, Jones, & Hamner, 1995; Shorten et al., 2005). While they are members of the pelagic community, they are not planktonic. They can steer accurately, turn 90 degrees in one or two swim contractions, and avoid obstacles (Garm et al., 2007a; Garm, O’Connor, Parkefelt, & Nilsson, 2007b; Petie, Garm, & Nilsson, 2011). The structure and optics of their lensed eyes are unparalleled in the phylum (Berger, 1898; Martin, 2004; Garm et al., 2007a; O’Connor, Garm, & Nilsson, 2009).

The swim mechanics of Cubomedusae are more like those of prolate medusae than oblate medusae, but they are able to achieve a significant size through a hybrid mechanism of propulsion (Colin et al., 2013; see also Gladfelter, 1973; Shorten et al., 2005). Feeding responses include longitudinal contraction of the stimulated tentacle; inward bending of the pedalium, which attaches the tentacle to the bell; and a directional movement of the manubrium toward the prey-bearing tentacle (Larson, 1976). A protective crumple response involves similar contraction and inward bending of all four tentacles, as well as inhibition of swimming.

The cubomedusan nervous system is built on the same plan as scyphomedusae, with some modifications and some advances. Swim pacemakers are located in four rhopalia, each positioned near the margin in a perradial, exumbrellar position (Berger, 1898; Satterlie, 1979; also see Figure 11). Removal of the rhopalia stops swim contractions, as in scyphomedusae. The rhopalia are sensory-neural structures with a significantly greater complexity than their scyphozoan counterparts. Each rhopalium has two lensed, image-forming eyes on the oral midline, two lateral spot ocelli, and two lateral slit ocelli (Figure 11). A secreted statolith is located at the end of the rhopalium. Neuronal tracts have been described that interconnect the eyes and ocelli and the pacemaker region (Skogh, Garm, Nilsson, & Ekstrom, 2006; Garm & Ekstrom, 2010). These tracts are directional, suggesting the presence of discrete neuronal pathways that are unlike the traditional nerve nets.

Open in new tab

Figure 11. A schematic to show the subumbrellar organization of the cubozoan medusa, Carybdea. R—rhopalium; the nerve ring is in green; S—subumbrellar swim musculature; v—velarium; P—pedalium; F—frenulum (a triangular buttress of striated muscle that reinforces the velarium); Per—radial muscle overlying the perradial canal. Inset—A rhopalium showing the two lensed eyes (L) and the four ocelli (arrows).

The eyes and ocelli influence the swim pacemakers in different ways and contribute to a versatile, visually guided locomotory system (Garm & Bielecki, 2008; Garm & Mori, 2009; Garm, Hedal, Majken, & Gurska, 2013). The lensed eyes are of the camera type, opsin based, but with fairly poor resolution (Martin, 2004; Nilsson, Coates, Gislen, Skogh, & Garm, 2005; Coates, Garm, Theobald, Thompson, & Nilsson, 2006; Garm et al., 2007a; O’Connor et al., 2009; Gray, Martin, & Satterlie, 2009). Based on electroretinograms, the lensed eyes have relatively slow response times (Garm et al., 2007a; O’Connor, Nilsson, & Garm, 2010).

The cubomedusan rhopalium is constructed like a typical invertebrate ganglion, with an outer layer or nerve cell bodies and a central neuropil (Satterlie, 2002; Parkefelt, Skogh, Nilsson, & Ekstrom, 2005; Skogh et al., 2006). Neural pathways extend from the retinal layer of the eye and ocelli and form discrete neuronal tracts that show a bilateral symmetry (Martin, 2002, 2004; Satterlie, 2002; Parkefelt et al., 2005; Skogh et al., 2006; Parkefelt & Ekstrom, 2009). In addition, compressed nerve nets are present, as exemplified by a ringlike network of RFamide-immunoreactive neurons that are found in the pacemaker region of the rhopalium (Martin, 2002, 2004; Satterlie, 2002; Plickert & Schneider, 2004; Skogh et al., 2006; Parkefelt & Ekstrom, 2009; Eichinger & Satterlie, 2014). This organization of discrete, directed neuropillar tracts intermixed with compressed networks represents the height of cnidarian centralization.

The rhopalia are connected to the subumbrellar conducting system via a rhopalial stalk, and paired neuronal tracts exit the stalk and join a subumbrellar nerve ring (Satterlie, 1979; Garm et al., 2006; Garm, Poussart, Parkefelt, Edstrom, & Nilsson, 2007c; Eichinger & Satterlie, 2014; Satterlie, 2014). The nerve ring connects the four rhopalia and the four tentacle bases (located at the interradial margin; see Figure 11). The nerve ring represents an advance over scyphomedusae and allows rapid communication between rhopalia. At least four conducting systems are found in the nerve ring, including elements of the swim system, a system conducting photic responses between rhopalia, a crumpling system that coordinates all four tentacles, and a local system for feeding reactions (Satterlie, 2014).

Neurons diverge from the nerve ring and form subumbrellar nerve nets that innervate the circular swim muscle (striated cells) (Satterlie, 1979; Laska & Hundgen, 1984; Eichinger & Satterlie, 2014; also see Figure 12). As in scyphomedusae, the neurons of the subumbrellar nerve net interact via symmetrical chemical synapses and use polarized synapses to activate the muscle cells (Satterlie, 1979). Electrical and dye coupling have not been found in the nerve net or between muscle cells of cubomedusae (Satterlie, 2011, 2015b). The subumbrellar nerve net is the equivalent of the MNN of scyphomedusae. Action potentials in MNN neurons are similar to those of scyphozoan neurons, and they trigger long-duration muscle action potentials in a 1:1 relationship (Satterlie & Spencer, 1979; Satterlie, 1979). Frequency-dependent facilitation produces muscle plasticity through variations in pacemaker frequency and regularity (Satterlie, 1979).

Open in new tab

Figure 12. Subumbrellar nerve ring and nerve nets of Carybdea labeled with an antitubulin antibody (motor network for swim contractions). Neurons of the two nerve nets diverge from the nerve ring (NR). S—subumbrellar net; V—velarial net.

A velarium extends inward from the margin and functions like the velum of hydromedusae—it narrows the bell opening to enhance the forceful ejection of water from the subumbrella, but it also aids in turning via asymmetrical contraction and directional nozzle formation (Gladfelter, 1973; Shorten et al., 2005; Garm et al., 2007b; Petie et al., 2011). The velarium is considered analogous to, rather than homologous with, the hydromedusan velum. As with the velum, the velarium has circular, striated muscle that contracts with the subumbrellar muscle, but the latter lacks the exumbrellar radial muscle and must rely on a different mechanism for producing the asymmetrical contractions during turning (Satterlie, Thomas, & Gray, 2005). Neurons diverge from the nerve ring to innervate the velarium, as with the subumbrella (Eichinger & Satterlie, 2014; also see Figure 12).

Pacemaker redundancy was tested using rhopalial ablation, as well as network modeling (Satterlie & Nolen, 2001). As with scyphomedusae, each rhopalium in isolation had an irregular pacemaker output, and with the addition of pacemakers, the overall output of the swim system became more regular and the mean contraction frequency increased. The modeling experiments were unique, however. One tested model (the resetting model) included pacemakers that were linked with resetting connections, so that the active pacemaker reset all the others. Another model (the independent model) did not include any connections between the pacemakers. Evaluation of data from real animals and both models suggested that the animal data fell between the two (i.e., semi-independent) (Satterlie & Nolen, 2001). Since the rhopalial pacemakers receive several types of sensory inputs, some excitatory and some inhibitory, the semi-independent model suggests that while pacemaker interactions are resetting, some inputs can take a rhopalium out of the network temporarily or cause it to fire prior to receiving input from another active pacemaker.

A second subumbrellar conducting system, similar to the DNN of scyphomedusae, is missing in cubomedusae. Instead, a compressed network is found in the nerve ring, rhopalia, and tentacle bases that shows the same RFamide immunoreactivity as the scyphozoan DNN (Eichinger & Satterlie, 2014). In the rhopalia, immunoreactive neurons form a tight network in the pacemaker region that includes epithelial sensory cells that are similar to the touch pad of scyphomedusae (Satterlie, 2011; Eichinger & Satterlie, 2014). The cluster of immunoreactive neurons in the nerve ring area of the tentacle bases may be of similar origin as rhopalia, which develop from polyp tentacles.

Perradial strips of radial muscle (smooth type) run from just above the rhopalial-nerve ring junction to the manubrium (Satterlie et al., 2005; also see Figure 12). Two compressed networks of neurons are found in association with the radial muscle (Satterlie & Eichinger, 2014). These muscle strips contract to help pull the margin of the bell inward during either a single-tentacle feeding response or a coordinated crumple response. Cuts through the nerve ring destroy this tentacle coordination, suggesting that a crumple pathway runs in the nerve ring (Satterlie, 2014). Crumpling inhibits swimming as occurs in hydromedusae.

Cubomedusan advances (over scyphomedusae) include the elaborate rhopalia with lensed eyes and increased neural complexity. The rhopalia are the cnidarian version of ganglia. The DNN of scyphomedusae appears to be represented by a compressed network in the nerve ring that connects to well-developed networks of neurons in the rhopalia and tentacle bases.