Sound detection and production mechanisms in aquatic decapod and stomatopod crustaceans Free

During World War II, naval warfare required underwater listening devices – hydrophones – to be developed for detecting submarines. For years, sonar operators used to consider anything that was not a signal from a structure as noise. However, when scientists took the time to listen to this ‘noise’, it became apparent that it was derived from a myriad of sound sources. These sources include geophony, i.e. sounds produced by natural physical phenomena, such as weather, ocean conditions and geophysical activity; biophony, i.e. natural sounds produced by the animals themselves, such as vocalisations that may convey biological-specific information; and anthrophony, i.e. human-generated sounds, such as shipping and marine construction. Together, these comprise the soundscape of an ecosystem (Pijanowski et al., 2011). Furthermore, these soundscapes can vary both temporally (over daily and lunar cycles, or seasonally; Cato, 1978; Radford et al., 2008) and spatially (Cato, 1978; Kennedy et al., 2010; Radford et al., 2014, 2010). Therefore, the underwater soundscape provides aquatic animals with a rich sensory environment that they can take advantage of in an environment in which sensory cues are often limited.

Underwater, the detection of sound is the one sense that is least affected by environmental variables; it can provide directional and habitat quality information rapidly over large distances (Larsen and Radford, 2018; Radford et al., 2011; Rogers et al., 1988; Urick, 1983). Therefore, it is not surprising that many different aquatic animals have evolved ways to detect and produce sound. There have been decades of pioneering work on hearing and sound production in fish (hearing reviewed by Popper and Fay, 2011; Popper and Hawkins, 2021; sound production reviewed by Amorim, 2006; Ladich, 2019) and cetaceans (reviewed by Richardson et al., 1995), but there has been comparatively little on crustaceans, even though it is now well known that they rely on sound for many aspects of their life-history strategies (Roberts and Laidre, 2019; Stanley et al., 2015, 2012). The last review on sound detection and production in crustaceans was by Popper et al. (2001) and focused on decapods only. Here, we review hearing and sound production in fully aquatic Decapoda (crabs, shrimp and lobsters/crayfish), as well as the Stomatapoda (mantis shrimp), herein referred to as crustaceans. We focus on fully aquatic species because of the stark difference between how animals make and hear sounds in the air versus the water (Larsen and Radford, 2018). In this Review, we begin with the basics of the physics of underwater sound, and then follow with an overview of crustacean sound detection, sound production and associated behaviours, finishing with a synthesis of views and important questions raised.

Understanding the physics of sound is critical when considering how animals detect and utilise sound signals. The physics of sound as a biological stimulus has been reviewed in-depth by Rogers and Cox (1988) and Larsen and Radford (2018); here, we present a condensed version of the basics for the purpose of understanding sound detection and production in crustaceans. Sound is a waveform that travels through a medium accompanied by a transfer of energy. There are two components to the sound field: (1) sound pressure, which consists of alternating pressure fluctuations causing localised regions of compression and rarefaction; and (2) particle motion, vibrations of the individual particles of the medium that do not travel with the wave.

In addition to sound transmitting through the water, it can also transmit through the substrate (e.g. sea floor) or along the interface between the water and the substrate, which is often termed ‘ground roll’ (Popper and Hawkins, 2018). Ground roll is characterised by large particle motion amplitudes, and it propagates at higher speeds than water-borne sounds and over large distances (Hazelwood and Macey, 2016; Popper and Hawkins, 2018). These waves are typically lower than 30 Hz. Furthermore, within the ground roll, the particle motion within the sediment and water follow a closed elliptical path (evanescent waves) in the vertical axis parallel to the propagation direction (Hazelwood and Macey, 2016). As such, ground roll may be a significant sound stimulus for crustaceans, as they are typically well coupled to the substrate or to the water just above the substrate (Fig. 1).

Aquatic crustaceans possess a range of mechanoreceptors, including: (1) the statocysts, historically suggested to be the primary mechanism for sound detection; (2) superficial receptors, setae or hair-like structures found on the body surface; and (3) chordotonal organs associated with the joints of the antennae and legs or other body appendages (Bush and Laverack, 1982). Little has changed since the last review by Popper et al. (2001) regarding the anatomical descriptions of the sound detection mechanisms; therefore, in this section, we only provide an overview of the general mechanisms of sound detection in aquatic crustaceans. We discuss species-specific mechanisms in more detail in the subsequent section.

Statocysts are internal sensory receptors that are similar in design among various crustacean species. However, they can vary in location, such as at the base of both antennae in crabs and lobsters (Popper et al., 2001; Sekiguchi and Terazawa, 1997) or the telson or uropods (see Glossary) in mysids and isopods (Sekiguchi and Terazawa, 1997). The statocyst is a fluid-filled chamber that contains a gelatinous mass of sand grains, the statolith (Fig. 2A), which sits upon a crescent-shaped bed of sensory hairs (Budelmann, 1992; Lovell et al., 2005). It is well established that the statocyst functions as an equilibrium organ by triggering righting movements (Budelmann, 1988; Cate and Roye, 1997; Cohen and Dijkgraaf, 1961; Fraser, 2001; Fraser et al., 1987). However, statocysts also respond to a variety of stimuli, such as hydrostatic pressure (Fraser and Macdonald, 1994; Macdonald and Fraser, 1999) and water-borne particle motion (Budelmann, 1992; Dinh and Radford, 2021; Hughes, 1996; Jézéquel et al., 2021; Lovell et al., 2005, 2006; Radford et al., 2022).

The statocyst is a mass-loaded sensory system analogous to the fish otolithic system (Montgomery et al., 2006; Popper and Fay, 2011; Popper et al., 2001; Putland et al., 2019), hence it functions as a differential accelerometer; that is, a mass that moves in relation to a receptor and responds to sound-induced motions of the animal's body. In short, the animal and surrounding water are approximately the same density; the animal therefore moves at the same frequency and phase as the surrounding water. The statolith moves at a different frequency and phase owing to the density differences between it and the water. This differential movement deflects the sensory hair through shear force, and stimulates a nervous response (Budelmann, 1992; De Vries, 1950; Montgomery et al., 2006; Popper and Hawkins, 2019; Popper et al., 2001).

Sensory hair fans, setae or sensilla are found in large numbers on the external surface of the body and appendages (Fig. 2B), with some being both chemosensitive and mechanosensitive, whereas others are only mechanosensitive (Cate and Derby, 2002; Derby, 1982). One or more hair fans can make up a sensory bundle; hair fans within a sensory bundle can range in size from 200 to 2000 µm (Breithaupt and Tautz, 1990; Cate and Derby, 2002; Tautz, 1979). It has been suggested that the size of the hair fans and the boundary layer (see Glossary) over the cuticle is directly related to their sensitivity (Tautz, 1979). Assuming laminar flow (see Glossary), the boundary layer thickness is inversely proportional to frequency of the sound, hence at lower frequencies the boundary layer is thicker. Tautz (1979) has estimated that the hair fans need to be 2–6 times longer than the thickness of the boundary layer to work effectively. However, for crustaceans that have rough surfaces, such as spiny lobsters, this relationship breaks down, as the flow across the cuticle is turbulent. To date, the sensitivity of hair fans has only been described for two Homarus species. For one, the sensitivity to pure particle motion stimuli was tested (Laverack, 1962) and, for the other species, more recently, sensitivity to underwater sound (which generates both particle motion and pressure stimuli) was determined (Breithaupt and Tautz, 1990; Jézéquel et al., 2021). The European lobster (Homarus vulgas) responded to pure particle motion frequencies up to 100 Hz (Laverack, 1962), and the American lobster (Homarus americanus) responded to frequencies up to 300 Hz emanating from an underwater speaker (Jézéquel et al., 2021).

Chordotonal organs are widespread among crustaceans and are generally associated with the joints of flexible body appendages, with the most common locations being the antennae or legs (Fig. 2C) (Whitear, 1960). These organs contain bipolar sensory neurons that are embedded in the muscle, apodeme (see Glossary) or connective tissue and connect to the central nervous system (Boon et al., 2009; Budelmann, 1992; Bush and Laverack, 1982; Popper et al., 2001; Taylor, 1967). For example, an extremely sensitive chordotonal organ with two sets of sensory cells has been described in the antennal flagellum of the big-claw purple hermit crab (Petrochirus californiensis) (Taylor, 1967). A similar system has also been described in the small and large antenna of spiny lobsters (Hartman and Austin, 1972; Laverack, 1964; Rossi-Durand and Vedel, 1982) and the narrow-clawed crayfish (Astacus leptodactylus) (Bender et al., 1984; Masters et al., 1982; Tautz et al., 2005), and in the meral-ischail joint of the leg in face-banded mangrove crabs (Perisesarma eumolpe and P. indiarum; Boon et al., 2009). Depending on organ structure, these sensory organs can respond to joint position, movement and tension (Boon et al., 2009; Budelmann, 1992). They have also been shown to respond to water-borne particle motion (Taylor, 1967).

A few crustacean species have long been known to produce sound (Fish, 1966; Hazlett and Winn, 1962; Horch and Salmon, 1969; Johnson et al., 1947; Moulton, 1957). However, today, more than 10 families of aquatic Crustacea have been documented to produce acoustic signals (for previous reviews, see Nakamachi et al., 2021; Popper et al., 2001; Schmitz, 2002). Crustaceans possess a hard exoskeleton that they can use to produce sound through stridulation (Covich and Thorp, 2001). There are many types of stridulation mechanism: palinurid lobsters use the ‘stick and slip’ mechanism; paddle crabs rub their first walking leg across ridges on the chela (see Glossary); and various species potentially rub the gastric teeth together. The various mechanisms are discussed in detail below. Although stridulation appears to be the most dominant form of sound production among crustaceans, there are other forms, such as vibration/muscle contraction, and snapping or cavitation.

There has been a long-standing debate as to what constitutes hearing. If we adopt a narrow definition of hearing, which includes only the mechanisms that allow ‘detection of far-field sound, or pressure waves’ (van Bergeijk, 1967), then crustaceans cannot hear. By contrast, a broader definition of hearing includes ‘the reception of vibratory stimuli of any kind and nature, provided that the sound source is not in contact with the animal's body’ (Pumphrey, 1950). Here, we use the broader and more inclusive definition of hearing because it provides a non-anthropocentric view and includes most aquatic animals, which are likely to hear using accelerometer-based particle motion sensors (Budelmann, 1992; Mooney et al., 2010; Popper and Hastings, 2009).

Hearing thresholds can be determined using physiological methods (e.g. auditory evoked potentials, AEPs; see Glossary) or through behavioural methods (e.g. conditioning experiments). However, it is difficult to compare physiologically derived thresholds with behaviourally derived thresholds (Hawkins, 1981). Physiological methods require the animal to be restrained, often by use of anaesthesia, thus providing information on the response properties of the auditory system at the level of the brainstem (AEP) or nerve (single unit); by contrast, behavioural methods provide details on auditory thresholds at the level of the whole animal (Sisneros et al., 2016). In addition, physiologically determined absolute hearing thresholds vary greatly between different studies, owing to differences in experimental setups and acoustic environments, making quantitative comparisons of threshold levels difficult (Ladich and Fay, 2013; Ladich and Wysocki, 2009). However, physiological and behavioural audiograms can be used to compare auditory frequency bandwidths (i.e. the frequencies detectable by the animal) of different species (Ladich and Fay, 2013; Vetter et al., 2018). Furthermore, it is important to understand the different stimuli used in these experiments. Underwater speakers provide both a pressure and a particle motion stimulus, whereas shaker tables, which are becoming increasingly popular, provide a pure acceleration stimulus. Owing to the statocyst being a mass-loaded sensory system, it can be directly stimulated using shaker tables in the absence of water and any potential confounding pressure stimuli.

The genus Ovalipes (swimming crabs) has long been presumed to produce acoustic signals through stridulation, by rubbing of hard body parts along ridges on the lower section of the chela, owing to the presence of specific anatomical structures (Stephenson, 1969). The New Zealand paddle crab (Ovalipes catharus) has recently been described to produce three distinct acoustic signals, the ‘rasp’, ‘zip’ and ‘bass’ (Flood et al., 2019). The low-mid frequency (peak frequency 662±0.1 Hz) zip is a stridulatory signal produced when the chela propodus ridges are rubbed against the flexed meropodite–carpopodite joint of the first walking leg while it is stationary (Flood et al., 2019). The zip, along with the bass, have been directly correlated with post-copulatory mate-guarding and courtship behaviour, and are only known to be produced by male adult crabs in the presence of other males competing for a receptive female. The rasp signal has also been reported in the rowing crab (O. trimaculatus; Buscaino et al., 2015), with a peak frequency of 3.5 kHz, whereas the peak frequency for O. catharus varies with crab size (carapace width <50 mm, ∼11 kHz; carapace width >111 mm, ∼2.5 kHz). Although Buscaino et al. (2015) reported the rasp signal to be involved in reproduction, Flood et al. (2019) did not, instead reporting it to be associated with feeding behaviours. Both the rasp and bass are hypothesised to be produced internally, as there are no detectable movements of any appendages and/or mouthparts during their production.

Most research investigating the sound detection abilities of brachyurans has used AEPs to measure evoked responses from the statocyst/supraoesphogeal ganglion region. Radford et al. (2022) investigated the sound detection ability of two reef-associated species, the purple rock crab (Leptograpsus variegatus) and the red rock crab (Plugisa chabrus), and two soft sediment-associated species, the tunnelling mud crab (Astrohelice crassa) and O. catharus, using both shaker table and underwater speaker stimuli. Overall, there were differences between shaker table-derived thresholds and speaker-derived thresholds, with individuals being more sensitive to low-frequency (<200 Hz) speaker sounds than they are to shaker-table stimuli. This suggests that placing the animal in water provides particle motion stimuli that can be detected by one of the other sensory receptors outside the statocyst, such as the chordotonal organ or superficial receptors.

The only evidence for hearing in crustacean larvae is from Stanley et al. (2011, 2012). Here, the metamorphic response of settlement-stage (megalopa) animals was utilised as a behavioural indication of hearing. Underwater sound from a temperate reef was replayed at four sound levels (90, 110, 126 and 135 dB re. 1 µPa) to three reef-associated species – common rock crab (Hemigrapsus sexdentatus), smooth shore crab (Cyclograpsus lavauxi) and L. variegatus – and a soft sediment-associated species, A. crassa. Overall, the reef-associated species showed an increase in metamorphosis with reef sound being replayed compared with the silent control. Leptograpsus variegatus was the most sensitive, with an increased metamorphic response to 90 dB re. 1 µPa, followed by C. lavauxi, responding at 100 dB, and H. sexdentatus, responding at 126 dB re. 1 µPa. Unsurprisingly, the soft sediment-associated A. crassa did not show an increase in the metamorphic response to replayed reef sound. However, this is not an indication that the megalopae did not hear the sound, because even if they heard the reef sound they would not metamorphose and settle in an unfamiliar habitat. Although this work does not provide any indication of the frequency bandwidth these animals can hear, it shows that they are able to hear sound at relatively low sound levels.

Here, lobsters are defined as marine crustaceans from one of three families, the spiny (Palinuridae), clawed (Nephropidae) and slipper (Scyllaridae) lobsters, whereas crayfish are defined as being freshwater species. Spiny lobsters can be further divided into two groups based on historical perceptions of whether they produce sound: the Stridentes (Palinurus sp.) produce a rasp-type sound through the ‘stick-and-slip’ mechanism (Fig. 3A) (Patek, 2001, 2002; Patek and Baio, 2007), and the Silentes (Jasus sp.) were previously thought to be silent (Fornshell and Tesei, 2017). The stick-and-slip mechanism involves rubbing the basal extension of each antenna (plectrum) over a file on the antennular plate below the eyes. The resulting rasp is composed of a series of pulses, and the frequency range is species specific: Californian spiny lobster (P. interruptus) peak frequency is 633 Hz in the field (Patek et al., 2009), and European spiny lobster (P. elephas) peak frequency is 770 Hz (Jézéquel et al., 2019). Regardless of the species, the sounds are produced during perceived predator threats and are typically accompanied by anti-predatory behaviours (e.g. tail flap, lunge and point). A second sound type has been described for P. elephas – a rasp comprising a broadband pulse (peak frequency 19.2 kHz). These rasps are also produced in an antipredator context (Buscaino et al., 2011). This sound type is similar in nature to the sounds produced by the New Zealand spiny lobster (Jasus edwardsii), which was once thought to be silent. However, they are now known to produce a rasp-type sound (peak frequency 1436 Hz >150 mm, 2976 Hz ≤150 mm) during feeding events (Smith, 2021). The mechanism for the rasps produced by P. elephas and J. edwardsii is unknown.

Two species of Nephropidae, the American (Homarus americanus) and European (H. gammarus) lobsters, are known to produce a low-frequency buzz (87–261 Hz and 50–250 Hz, respectively) when threatened or handled (Fish, 1966; Henninger and Watson, 2005; Jézéquel et al., 2019, 2018; Ward et al., 2011). This sound is produced by antagonistic contraction of the remotor and promotor muscles, which generate carapace vibrations. Furthermore, Jézéquel et al. (2018) described a rattle-type sound from H. gammarus that was produced when the lobsters were feeding; however, the mechanism for this sound remains unknown. It could be similar in nature to the internally produced rasp observed in the two spiny lobsters mentioned above. To the authors' knowledge, there is no evidence for sound production in slipper lobsters.

The only crayfish reported to produce sound is the red swamp crayfish (Procambrus clarkii), which produces a rasp signal (peak frequency 28 kHz) during intraspecific interactions (Buscaino et al., 2012). This signal has similar features to the rasp of P. elephas, and the rattle of J. edwardsii and H. gammerus. The mechanism used to generate all three of these signals remains undescribed but is likely to involve very similar structures due to similar signal characteristics. We hypothesise that this rasp signal is produced by the gastric teeth in the stomach of the animals, which are responsible for producing similar signals in the ghost crab (Ocypode quadrata). (Taylor et al., 2019).

There have been several studies showing that lobsters and crayfish use a range of receptors to detect underwater sound. An early behavioural study using an underwater speaker suggested that H. americanus has a narrow frequency bandwidth, 10–150 Hz. Ablating the statocysts by removing the antennae did not change the observed response. This result was recently reinforced through a physiological study using the AEP technique (Jézéquel et al., 2021). This study found H. americanus to have a slightly wider frequency bandwidth (80–250 Hz; Fig. 4), and also found that ablation of the statocysts did not change the response. However, ablation of the hair fans on the body, chela and legs did. Furthermore, isolated preparation studies recording from the hair fans of the chela from the Australian crayfish (Cherax destructor) have shown responses to particle motion from 20 to 500 Hz (Tautz and Sandeman, 1980). Using a dipole source (see Glossary) and similar preparation, but recording from the telson in P. clarkii, Wiese (1976) observed responses from 0.05 to 200 Hz. There was also a gradient in sensitivity: hair fans closest to the last tail segment on the telson were more sensitive than those near the posterior end of the telson. Together, these electrophysiological and behavioural studies highlight the fact that lobsters and crayfish have a narrow frequency response and that the hair fans are likely to play a much bigger role in sound detection than previously thought.

Snapping shrimp are one of the most well-studied animals in terms of underwater acoustics, owing to the fact that this group is one of the most ubiquitous and abundant sources of sound in coastal areas (Au and Banks, 1997; Everest et al., 1948; Lillis and Mooney, 2018). The most diversified group, Alpheidae shrimp (600 species in more than 36 genera), possess a pair of asymmetric chelae, with the dominant having a snapping mechanism (Fig. 3B). This involves a protruding plunger on the dactyl and a matching chamber in the propus of the dominant chela (Versluis et al., 2000). This enables the production of an impulsive, broadband (up to 200 kHz, typically peaking between 2 and 20 kHz), high-intensity signal or ‘snap’, which results from the collapse of a cavitation bubble produced upon the rapid closure of the dominant chela (Au and Banks, 1998; Everest et al., 1948; Johnson et al., 1947; Versluis et al., 2000). There is no published information regarding sound production or behaviour in prawns.

Snapping shrimp use snaps for a multitude of behaviours. Snaps can be used for conspecific interactions for mate selection (Heuring and Hughes, 2020; Hughes et al., 2014), to compete with same-sex individuals (Dinh and Radford, 2021; Herberholz and Schmitz, 1998; Nolan and Salmon, 1970) or in agonistic encounters to defend territories or burrows (Duffy et al., 2002; Schmitz and Herberholz, 1998). The sound pressure levels of snaps are positively correlated with body size (Au and Banks, 1998); therefore, it is possible that snaps could be used to assess fighting ability, individual size or mate quality (Dinh and Radford, 2021). Lastly, snaps are also produced spontaneously without external cues or when individuals are held in isolation (Lillis et al., 2017; Rossi et al., 2016). This suggests that snaps are used by shrimp for behaviours other than mate assessment or direct aggressive encounters. Two potential uses of isolated snaps could be to maintain spacing or territory (Lillis et al., 2017; Staaterman et al., 2011; Waser and Brown, 1986) or to attract mates or territory partners (Heuring and Hughes, 2020; Lillis et al., 2017). Future work should be directed towards assessing what the behavioural contexts are for these isolated snaps.

Several prawn species – giant tiger prawn (Penaeus monodon; Smith and Tabrett, 2013), Pacific white shrimp (Litopenaeus vannamei; Peixoto et al., 2020a,b) and the giant freshwater prawn (Macrobrachium rosenbergii; Hamilton et al., 2021) – have been found to produce incidental sound during feeding, through closure of the mandibles when they are shredding food. Furthermore, M. rosenbergii produces a rasp that is similar in nature to the broadband rasp sounds made by several lobster and brachyuran species. Although Hamilton et al. (2021) suggest the source of this sound is the mandibles grinding together, we propose here that this sound is produced by the gastric mill owing to its similarity with the rasp signals of other crustaceans.

To date, only two species, the common prawn (Palaemon serrata; Lovell et al., 2005, 2006) and the snapping shrimp Alpheaus richardsoni (Dinh and Radford, 2021), have had their sound detection ability characterised using AEPs. Using an underwater speaker and recording from the statocyst region, the common prawn was found to be sensitive to frequencies between 100 and 3000 Hz (Fig. 4B; Lovell et al., 2005). It was also found that there was no correlation between the number of hair cells within the statocyst and sound detection ability (Lovell et al., 2006). Again recording from the statocyst region, but using a shaker table, Dinh and Radford (2021) observed that the snapping shrimp could detect particle acceleration frequencies from 40 to 1200 Hz. However, particle acceleration detection is found to be significantly more sensitive when determined using an underwater speaker compared with using a shaker table. This suggests that the statocyst is not the only sensory system used for sound detection in this group of animals.

Anomura is another group of crustaceans that remains relatively understudied, and those studies that exist have been conducted on hermit crabs (Trizopagurus sp. and Pagurus bernhardus; Briffa and Elwood, 2001; Field et al., 1987; Herbst, 1782; Roberts et al., 2016). During hermit crab shell fights, sound production occurs in two ways: incidentally, through a shell ‘rap’ where the aggressor hits the defender's shell (Briffa and Elwood, 2001), and actively, through a mechanism used by the defender. Active sound production occurs through reciprocal flexion and extension of the carpo-propodite and mero-carpopodite joints of tightly apposed chelae, and is known as the ‘chirp’ (Field et al., 1987). During shell fights, the attacker starts agonistically ‘shell rapping’ repeatedly on both its own and the defender's shell; and in response, the defender subsequently produces the chirp. Eventually, either the defender is evicted from the shell or the attacker gives up (Briffa and Elwood, 2001; Briffa et al., 2003; Lane et al., 2022). The chirp sound has also been recorded in other behavioural contexts, such as during confinement, after shell exchange, after introduction of novel stimuli and when handled (Borradaile, 1903; Imafuku and Ikeda, 1990), but none of these contexts are reliable triggers. There are no detailed reports on the characteristics of the signal produced by the shell rap; however, the peak frequency of Trizopagurus sp. chirps is between 6 and 8 kHz (Field et al., 1987).

To date, there has only been one study investigating the hearing ability of an anomuran, in P. bernhardus (Roberts et al., 2016). Using a behavioural paradigm that assessed antennule flicking and burst movement in response to substrate-borne vibrations, P. bernhardus was found to have a hearing frequency range of 5–400 Hz for antennule flicking and a narrower frequency range of 20–400 Hz for burst movement. At 90 and 210 Hz, the threshold for antennule flicking was significantly lower than that for burst movement. The authors suggested that burst movement is more energetically demanding, and it therefore requires a more intense sound to elicit a response compared with antennule flicking. Also, it was found that animals that had spent less time in captivity were more sensitive than those that had been in captivity for longer. This suggests that animals that have been held under laboratory conditions for longer periods of time show higher levels of habituation to substrate-borne vibrations.

Sound production mechanisms and associated behaviours have been described in detail in stomatopods, but very little is known about their sound detection capabilities. Stomatopods can produce sound in two different ways: incidentally, when using the raptorial appendage (see Glossary) to strike objects (Hazlett and Winn, 1962), or actively, using vibrating muscles under the carapace (Fig. 3C). Stomatopods actively produce low-frequency tonal sounds (fundamental frequency 20–60 Hz; termed the ‘rumble’) by contracting the mandibular remotor muscle, which connects to a stiff lateral extension of the carapace (Patek and Caldwell, 2006; Staaterman et al., 2011). Laboratory investigations of California mantis shrimp (Hemisquilla californiensis) found that only males produce the rumble sound, and it is typically accompanied by a strike of the raptorial appendage; this suggests that it is part of an agonistic/defensive display (Patek and Caldwell, 2006). Field investigations have supported this notion: multiple males produce rumbles in rhythmic series or ‘rumble groups’ during crepuscular periods while guarding their burrows (Staaterman et al., 2011).

How stomatopods detect sounds remains a mystery because, unlike most crustaceans, they do not possess a statocyst organ. However, their bodies are covered in sensory hairs – the detection of water-borne particle motion presumably occurs through this sensory receptor pathway (Staaterman et al., 2011). Stomatopods are typically well coupled to the seafloor, as they are benthic animals that live in burrows; therefore, substrate-borne vibrations may be an important sensory pathway for this group. The rumbling sounds are likely to propagate well through the substrate (although there is currently no evidence for this). As well as the sensory hairs, they are also equipped with an array of chordotonal organs (Wales and Ferrero, 1987), which could act as receptor pathways for substrate-borne vibrations.

Our understanding of the auditory and sound-production systems of crustaceans is limited, and many questions remain unanswered. In this section, we provide a synthesis of the main themes identified between the different groups and consider some of the knowledge gaps (summarised in Table 1). We also summarise what we consider to be the critical questions; we hope that these areas will be the focus of future studies.

For sound production to be useful, the intended receiver needs to be able to detect the sound signals; however, there are multiple examples above that suggest there is a mismatch between sound detection abilities and the sounds produced by a given species. For example, the common rasp signal (>2 kHz) that is produced by several species of lobsters and brachyurans is outside their sound detection ability: the maximum detectable frequency for O. catharus is 2000 Hz and for H. americanus is 400 Hz. However, several studies (e.g. Flood et al., 2019) have shown that these species can hear this sound. Thus, another sensory channel, such as substrate-borne vibrations, may be important here. Compared with aquatic crustaceans, there has been extensive research into understanding sound production and transmission in terrestrial crustaceans (e.g. Coenobita sp.; Imafuku and Ikeda, 1990; Roberts, 2021). Roberts (2021) measured chirps of the land hermit crab (Coenobita compressus) in the air to be 800–8400 Hz. This is likely to be outside their sound detection range. However, the same signals measured through the substrate were 40–1120 Hz, which falls within the sound detection range of this species. Thus, there is a major difference in the frequency components of the chirp between the two sound-transmission channels. If the same frequency differences occur for the rasp signals transmitted through the water and substrate by lobsters and brachyurans, then it is likely that these species detect the substrate-borne signals. This highlights the importance of measuring the particle motion of sound signals produced by crustaceans in both sound-transmission channels – water and the substrate.

Substrate-borne vibrations consist of several types of wave propagation. The two that are most relevant to crustaceans are vibrations along the substrate–water interface (more commonly known as ground roll) and vibration within the seafloor substrate (reviewed by Hawkins et al., 2021). Aquatic crustaceans live in a range of habitats, from mud to rocky reefs, and the nature of the habitat affects propagation of the vibrations. Furthermore, substrate-borne vibrations, although not devoid of noise, may be advantageous in water soundscapes that are dominated by geophony and biophony, which may hinder sound transmission and detection. To date, there have been no measurements of biological sources of substrate-borne vibrations in aquatic environments. Therefore, direct measurements need to be made to understand: (1) whether crustaceans can transmit acoustic information through the substrate; and (2) the propagation distances of crustacean signals in both the particle motion (in water) and substrate-borne vibration domains.

This Review has highlighted that there has been little research in the last 21 years describing the anatomical structures of the three main crustacean sensory receptors that are thought to be sensitive to sound. Furthermore, and more critically, there have been few measurements, physiological or behavioural, made of the sensitivity of different crustaceans to particle motion and even fewer to substrate-borne vibrations. This is especially true of stomatopods. Given the importance of sound in many life-history strategies of crustaceans, it is critical that we achieve a better understanding of sound detection by: (1) using state-of-the-art medical imaging, such as microCT (Camilieri-Asch et al., 2020), to produce high-resolution in situ images of the different sensory structures; (2) measuring physiological responses from the different sensory receptors (e.g. statocysts, superficial receptors and chordotonal organs); (3) investigating the potential of using operant or classical conditioning paradigms to investigate sound detection and ‘hearing’; and (4) using serial lesion experiments (e.g. ablating individual sensory receptors) in combination with the experiments suggested in 2 and 3. Finally, all such studies must be conducted under appropriate acoustic conditions (tank versus field) and using appropriate acoustic stimuli (e.g. water-borne particle motion and substrate-borne vibrations).

Although many sound production studies in crustaceans have been conducted in tanks, it is important to be aware of the limitations of such experiments. Of course, these studies can be useful. They allow animals to be isolated, so that researchers can confirm sound production in particular animals, without the ambiguity of other potential sources in the vicinity of the recorder. Tanks also allow a controlled experimental set-up, which helps us to understand animal behaviours associated with sound production. However, sound characterisation in tanks presents challenges, particularly when investigating broadband sounds, such as many of those reported here (e.g. antennal rasps and snaps). A tank can have several effects on acoustic signals, depending on its characteristics: (1) reverberation is the persistence of sound after the signal stops, and it effectively extends the sound duration (Jézéquel et al., 2018; Meyer-Rochow and Penrose, 1976; Parvulescu, 1963); (2) resonant frequencies (Akamatsu et al., 2002) can distort the spectral shape of the signal (Jézéquel et al., 2022, 2019, 2018; Meyer-Rochow and Penrose, 1976; Parvulescu, 1963); and (3) low frequencies are highly attenuated and cannot be measured properly (Rogers et al., 2016). Although tank recordings have their place, we recommend where possible that these should be complemented with in situ field recordings to provide accurate sound characterisation of the sound signals produced by crustaceans, especially broadband sounds.

The rasp sound produced by several prawn, lobster and brachyuran species is similar in acoustic structure; however, the exact mechanism underlying the production of this sound remains unknown. This rasp sound is similar in nature to the rasp signal produced by O. quadrata, which was shown by Taylor et al. (2019) to be produced by the gastric mill in the animals' stomach. Given that the rasp is produced across different families of malacostracans, including semi-terrestrial crustaceans, we propose that this is an acoustic signal that most, if not all, malacostracan crustaceans can produce and utilise.

Although there are now many examples of sounds used by crustaceans in a social setting, including reproductive displays, agonistic encounters and territory defence, relatively little is known regarding how a receiver specifically responds to a signal. To begin to better understand these behavioural observations in both field and tank settings, we need to observe the behavioural context of the signals with respect to both conspecifics and heterospecifics. Furthermore, research has shown that many larval crustaceans use ambient sound to orient and settle in their adult habitat. However, it is not known what part of the sound field they are responding to. Are they responding to general reef sound, conspecific signals or specific parts of the spectrum? Field behavioural choice experiments (Buwalda et al., 1983; Kough et al., 2014; Radford et al., 2007) can be used to answer these questions.

In conclusion, it is clear that sound is a vital sensory cue for aquatic malacostracan crustaceans and stomatopods, as it is for aquatic vertebrates. The transmission of sound underwater is very different to that in air: it can travel vast distances with little attenuation. However, as is the case for air-borne signals, underwater sound can carry important biological information. Here, we have shown that stridulation mechanisms, both internal and external, are the dominant method used by crustaceans to produce sound. In terms of determining the sound detection abilities of these species, the large majority of studies have used physiological techniques, and these results – coupled with knowledge of crustacean sound production characteristics – suggest that there is a mismatch between sound detection abilities and the frequencies of the sounds that crustaceans produce. Thus, there may be another sound transmission channel used by these animals: substrate-borne vibrations. Given that the large majority of these animals live near or on the sea floor, this sound-transmission channel potentially could be their dominant means of communication. Future research should be directed at understanding the transmission of substrate-borne vibrations and determining whether crustaceans can detect and ‘hear’ these signals.

We thank Shelia Patek for the sound files used to make the spectrograms in Fig. 3, as well as three anonymous reviewers for their constructive comments, which made for a more informative review.