Sexual selection and predation drive the repeated evolution of stridulation in Heteroptera and other arthropods

(1) Evolution of stridulatory organs in Heteroptera and prevalence in other arthropods

Our survey of Heteroptera found that stridulatory organs have evolved independently between 84 and 94 times, based on strict and relaxed estimates, respectively (Fig. 2A, B; Table 1). This encompasses a great diversity of different stridulatory types, including modifications of the tremulatory tergal plate (TTP) that add stridulatory properties (to form a stridulatory tergal plate, STP) to its pre-existing vibrational mechanism. Heteroptera therefore represent an unrivalled example of parallel and convergent evolution of vibroacoustic signal generation among insects. For comparison, the present analysis shows that the other major insect group that widely uses vibroacoustic signalling, Orthoptera, have evolved stridulatory mechanisms at least 25 times independently [Fig. 3; Table 2, relying primarily on the analysis of Song et al. (2020) and our own findings, described in Appendix S2]. Only spiders (Araneae), in which vibroacoustic signalling is similarly prevalent, approach the diversity of stridulatory mechanisms seen in Heteroptera, having evolved these at least 57 times independently (Fig. 3; Table 2). Non-insect pancrustaceans have evolved stridulatory mechanisms at least 40 times independently (Fig. 3; Table 2), further demonstrating the prevalence of such organs across arthropods.

Stridulatory organs in Heteroptera have not only evolved repeatedly, but are also widespread, being present in around 22% of approximately 40,000 extant species (Fig. 2A, B; Table S1). In most cases, stridulation is lineage specific, being present in only one or a few closely related species or genera (Table 1). Consequently, the number of times that stridulation evolves in a higher-level taxon may depend more on its species richness than its ecology. For example, the smallest heteropteran infraorders possess the least number of independent origins of stridulatory organs (if any), while the three most speciose ones have the most (Fig. 2A, B; Table S1), presumably due to the larger pool of lineages with the potential to evolve this trait. Based on their phylogenetic distribution, we conclude that stridulatory organs were invariably absent in the common ancestor of each heteropteran group (Fig. 2A, B). This is in contrast to chemical and non-stridulatory vibrational signalling (e.g. tapping, buzzing, tremulation), each of which is thought to be present at the root of the order (Fig. 2A, B) (Carayon, 1971; Davranoglou et al., 2017; Gogala, 1984; Schuh & Weirauch, 2020; Wheeler, Schuh & Bang, 1993). Thus, insects that already communicate by other means (i.e. chemicals, non-stridulatory vibrations) have repeatedly evolved stridulatory organs to produce more complex signals. Indeed, our results reveal a clear evolutionary tendency in Heteroptera and other arthropods towards the evolution of stridulatory organs, as multiple such mechanisms have evolved across distantly related groups over evolutionary time.

(2) Stridulatory organs are more frequent in particular body parts and habitats

We identified up to 20 distinct types of heteropteran stridulatory mechanisms (15 if functionally ambiguous types are excluded), each involving modified surfaces on different body parts striking against each other (Fig. 4; Tables 1 and S2). This is a remarkably diverse set of mechanisms, as only 14 distinct types are known from the even more speciose Coleoptera (Wessel, 2006), although the occurrence of stridulatory mechanisms in the latter group may be under-reported. However, in Heteroptera, only three stridulatory types collectively account for 82% of all cases of convergent evolution of stridulatory mechanisms according to our conservative estimates (Table S3, Fig. 4), namely the abdomen–leg mechanism (legs rub against the abdomen; Fig. 2C), the wing edge–leg mechanism (legs rub against wing margin) and the STP (up-and-down shaking of the abdomen by a vibrational organ known as the TTP, which is equipped with a plectrum that strikes against a stridulitrum on the hind wing; Fig. 1). The arthropod outgroups that we analysed have also repeatedly evolved stridulatory mechanisms that overwhelmingly involve the same body parts (Table 2): stridulatory structures on the chelicerae dominate in Arachnida; the pterygostomial and infraorbital regions dominate in brachyuran crustaceans; and wing–leg mechanisms prevail in Orthoptera. The above examples demonstrate that the evolution of the morphological framework for arthropod vibroacoustic signalling can be biased towards the use of specific body parts, even when there are multiple solutions for how to generate acoustic and substrate-borne signals using stridulation. It is evident that the motions performed by certain body parts are especially prone to evolving into stridulatory mechanisms; we present a hypothesis on the behavioural background that leads to this morphological bias in Section 7.2.

Besides being prone to evolving in particular body parts, the evolution of stridulation may also be affected by the signalling medium. Stridulation appears to evolve overwhelmingly often in taxa inhabiting soil and leaf-litter, plants, and water, but less often in those living on the water surface or bark (both on and under it) (Fig. 2A, B; Table S4). This may indicate that either the physical properties of particular substrates increase the probability that stridulation will evolve, or that certain important biotic factors (e.g. predation) are more prominent in particular habitats than others. The former may be the case for burrowing bugs (Cydnidae), where stridulation may provide a mechanical solution to signalling underground (Čokl et al., 2006), and also for certain water striders (Gerridae) that are reported to produce signals of a different frequency band to the environmental noise produced in their torrential waterfall habitat (Zettel & Thirumalai, 2000).

(3) Behavioural context of stridulation

The biological significance of stridulation is poorly documented; in 79% of stridulating heteropterans and in 52% of the arthropod outgroups, the behavioural context of stridulation is unknown (Tables S5 and S6). However, in arthropods where the behavioural significance of stridulation is known, we observe a strong association with sexual (courtship and mating) and defensive contexts (interactions with predators, disturbance/protest signals). In Heteroptera, roughly 30% of the described stridulatory behaviours are involved in defensive contexts only and 30% in sexual contexts only (Fig. 5A; Table S5). The remaining 40% of described stridulatory behaviours combine both defensive and sexual contexts, occur together with competition, and/or are observed with other contexts such as aggregation, territoriality, or maternal care. The role of defence is even stronger in the arthropod outgroups: 55% of stridulation takes place only in defensive contexts, and 18% in courtship only; the remaining 27% combine sexual and defensive behaviours together and/or with other signalling contexts (Fig. 5B; Table S6).

Based on the above, we conclude that stridulation is most commonly used in defensive and sexual behaviours, either in isolation or in combination. The use of stridulation in other contexts is rarer and occurs almost exclusively in combination with sexual and/or defensive functions. However, we caution that the apparent extent of the use of stridulation in defensive contexts could reflect observational bias, as many cases of stridulation have been documented only when specimens were collected and handled by researchers (Field, Evans & Macmillan, 1987; Field, 1993; Field & Bailey, 1997; Leston, 1954, 1957). Targeted studies on the ethology of stridulating species are likely to reveal significant overlap between sexual, defensive, and other contexts of stridulation.

(1) Sexual selection and predation explain the evolution of stridulation

Our comprehensive synthesis of the published literature reveals a clear link between stridulation and close-range defensive and sexual behaviours (Fig. 5; Tables S5 and S6). Most available ethological studies of Heteroptera show that stridulatory signals are used at close range in defensive, courtship, or mating behaviours (Gogala, 1984; Dodson & Marshall, 1984; Manrique & Lazzari, 1994; Schmidt, 1994; Zych et al., 2012). These same functions are also common for other arthropods (e.g. Alexander, 1957a; Blair & Bilton, 2020; Bouwma & Herrnkind, 2009; Desutter-Grandcolas, 1998; Esposito et al., 2018; Hrušková-Martišová et al., 2008; Field, 1993; Kronmüller & Lewis, 2015; Low et al., 2021; Lourenço & Cloudsley-Thompson, 1995; Field et al., 1987; Masters, 1980; Polidori et al., 2013; Pomini et al., 2010; Woodrow et al., 2021) and vertebrates (Cresswell, 1998; Bostwick & Prum, 2005; Gans & Maderson, 1973; Gould, 1965). More rarely, stridulatory signals are used in other close-range behaviours, such as rivalry, aggregation, and triggering synchronous egg-hatching (Gogala et al., 1974; Haskell, 1957; Jansson, 1972, 1989; Mukai et al., 2014). The mechanical characteristics of stridulatory signals presumably explain why their vibrational component is used primarily for close-range interactions, as their high frequency (2–20 kHz) (Čokl et al., 2006; Sueur, Mackie & Windmill, 2011; Yasunaga et al., 2019) is unsuitable for long-distance substrate-borne transmission in the main arthropod signalling substrates (soil, plants and water; Fig. 2A, B, Table S4) (Bennet-Clark, 1998; Casas, Magal & Sueur, 2007; Čokl & Virant-Doberlet, 2003; Gogala, 1984, 2006; Lang, 1980; Michelsen et al., 1982; Mortimer, 2017).

Stridulation is clearly an important signalling mechanism, as its repeated evolution across arthropods indicates strong selective pressures for its development. Stridulation is thought to be adaptive in defensive contexts by reducing predation risk in several ways, which are not mutually exclusive: (i) it can startle predators (Masters, 1979, 1980; Robinson, 1969); (ii) it can warn predators of undesirable qualities of the stridulating animal if an attack is realised, such as distasteful chemical secretions or painful stings (Ewing, 1984; Haskell, 1957; Masters, 1979; Kowalski et al., 2014); (iii) it can be part of a deimatic display (Kowalski et al., 2014; Low et al., 2021; Umbers, Lehtonen & Mappes, 2015); (4) it can imitate the sounds of a dangerous or toxic animal (Masters, 1979; Sandow & Bailey, 1978); or (5) it can produce vibrations that cause physical discomfort when stridulating prey is in the mouth of a predator, due to overstimulation of tactile receptors (Masters, 1979, 1980; Senter, 2008; Song et al., 2020); similar effects have been proposed for the presumably defensive buzzing of solitary bees (Larsen, Gleffe & Tengö, 1986). However, few of these hypotheses have been conclusively supported for the evolution of stridulation in arthropods (Alexander, 1957a; Masters, 1979, 1980; Rowe & Guilford, 1999; Sandow & Bailey, 1978; Stidham, 2020), due to the paucity of empirical studies. It is likely that stridulation is adaptive against different threats in different lineages of stridulating arthropods.

In a sexual context, the role of stridulation is less clear, and whether it conveys a particular quality of the signaller or serves simply as an advertisement remains to be determined. However, experimental and theoretical studies suggest that increased signalling complexity may be sexually adaptive (Ewing, 1984; Gerhardt, 1992). Multiple non-redundant signals may provide information on multiple qualities of the signaller (Chaine & Lyon, 2008; Stevens, 2013), whereas redundant multimodal signals are advantageous as they increase information transfer fidelity by acting as back-ups for each other (Ay, Flack & Krakauer, 2007; Stevens, 2013), are easier to detect by the receiver (Grafe & Wanger, 2007) and trigger faster behavioural responses in the receiver (Rowe & Guilford, 1999). In addition, interactions between different signals during their transmission may be informative in themselves (Partan & Marler, 1999). In Heteroptera, most species are limited in the number of distinct types of vibrational signals they can produce, as their primary vibrational organ is the TTP, which generates simple abdominal vibration (Benediktov, 2007; Gogala, 1984; Moraes et al., 2005; Kavčič et al., 2013). However, many heteropteran species complement their tergal plate vibrations with appendage and abdominal tapping, body shaking and wing buzzing signals (Kavčič et al., 2013; Takács et al., 2008; Žunič et al., 2008). We therefore suggest that stridulation is so successful as a vibroacoustic mechanism because it allows heteropterans and other arthropods to considerably expand their signalling repertoire with a minimal amount of morphological change.

Cuticle ultrastructure is highly malleable genetically (e.g. Finet et al., 2022; Khila, Abouheif & Rowe, 2012), so its modification to create stridulatory surfaces may be simple developmentally (Alexander, 1958; Masters, 1979), compared to the drastic morphological reorganisation required for the evolution of complex vibroacoustic organs such as tymbals. Based on morphological data alone, it is evident that cuticle ultrastructure exhibits remarkable developmental plasticity, as it may be modified in different life stages (e.g. Guidoti & Barcellos, 2013), sexes (e.g. Khila et al., 2012), species and higher-level groups (e.g. Galleti-Lima & Guadanucci, 2019; Hemala et al., 2021), and in adaptive shifts to accommodate different life styles (Giglio et al., 2003; Li et al., 2021). Based on the above, the morphological plasticity of the arthropod cuticle may be the primary developmental factor underlying the repeated evolution of stridulation in arthropods.

Although our analysis shows a clear, large-scale pattern in the functions of stridulation (Fig. 5), stridulation is frequently used in several contexts by the same species (Tables S5 and S6). This raises the question of which was the context in which stridulation first evolved. Stridulatory vibroacoustic signals that were adaptive in a sexual context may have been secondarily co-opted into defensive behaviours or vice versa, but this evolutionary sequence is challenging to determine empirically on the basis of the available data. In species whose stridulatory organs are sexually dimorphic, it is most probable that they arose in a sexual context, or else that one sex faces greater predation risk than the other (Kowalski et al., 2014). However, in most heteropterans, and in most other arthropods we have examined except Orthoptera, stridulatory organs are not sexually dimorphic, which may render a defensive rather than sexual function the more likely behavioural context for the origin of stridulation.

(2) Likely behavioural contexts involved in the evolution of stridulation

Our findings on the systematic distribution and ethological contexts of stridulation provide new insights regarding the behavioural basis of its evolution. In Heteroptera, the three most common stridulatory mechanisms (abdomen–leg, wing edge–leg, STP; Fig. 2A, B, Tables S2, S3) have frequently evolved in other insects as well (Aiken, 1985; Roth & Hartman, 1967; Wessel, 2006), meaning that certain areas of an animal's body are particularly prone to evolving into stridulatory organs. The cleaning/grooming motions performed by many insects, where the hind and mid legs scrape the abdomen, the wing edge, and themselves, may account for this morphological bias in the evolution of stridulation. These cleaning behaviours characterise the premating and copulatory rituals of many insects (Jansson, 1972; Sweet, 1964), and they also take place in certain species when they are attacked by predators (L.-R. Davranoglou, personal observations). We suggest that these characteristic motions may form the behavioural background that is modified during the repeated evolution of abdomen–leg and wing edge–leg stridulatory mechanisms, and consequently, vibroacoustic signalling. These observations are consistent with our identification of sexual selection and predation as major drivers in the evolution of stridulation (Fig. 5).

Ethological observations provide additional support for our hypothesis, as these grooming behaviours frequently produce sound that is audible even to human listeners, in the absence of any modified cuticular surfaces (Jansson, 1972; Roth & Hartman, 1967; Uvarov, 1966). We propose that if the signals produced during these grooming behaviours are sexually or defensively adaptive, then sexual or natural selection may lead to further modifications of the morphological structures involved that will render them better suited for producing bimodal vibroacoustic signals, thereby leading to the evolution of stridulatory organs. This pattern likely applies to other arthropod groups as well. For example, masticatory and cleaning motions of the chelicerae are thought to underlie the evolution of stridulatory mechanisms in wind scorpions (Arachnida: Solifugae) (Bird, Wharton & Prendini, 2016; Cloudsley-Thompson, 1961b; Stidham, 2020); pre-mating motions and/or defensive movements likely formed the behavioural background for the evolution of stridulation in cockroaches (Roth & Hartman, 1967); and defensive or offensive strikes may have led to the development of the dominant scorpion stridulatory mechanisms (Alexander, 1958).

It is easier to pinpoint the behavioural context for the evolution of the STP, a mechanism that is broadly used in Heteroptera (Table 1) and Coleoptera (Wessel, 2006; referring to abdominal stridulatory mechanisms striking against their counterparts on the elytra or the hind wings). The STP comprises a plectrum on the abdomen and a stridulitrum typically situated on a wing vein (Lis & Heyna, 2001) (Table 1). The frequent interaction between the abdomen and the hind wing during wing tucking, tremulation (i.e. body-shaking or trembling to produce vibrations), and when attempting escape from the mouth of a predator, together with their morphological proximity, are the likely factors that underlie its frequent evolution. The morphology of the STP, which in Heteroptera combines tergal plate-based low-frequency vibrations with stridulatory signals generated by rubbing the latter against the wings (Gogala et al., 1974), is congruent with hypotheses suggesting that fixed bimodal or multimodal displays tend to evolve using the same or adjacent structures (Caldwell, 2014). However, we show that any opposing body parts can be modified to generate mechanical-based multimodal signals, as Heteroptera possess at least 15–20 different types of stridulatory mechanisms (Fig. 4, Tables 1 and S2).

(3) Sensing of stridulation-based vibroacoustic signals

Whereas heteropterans primarily sense chemical signals using antennal olfactory receptors, they detect substrate-borne vibrations using Johnston's and subgenual organs on their antennae and legs respectively (Jeram & Čokl, 2006; Nishino, Mukai & Takanashi, 2016). Experimental evidence, although limited, suggests that most Heteroptera cannot perceive the acoustic component of their stridulatory signals, as their subgenual organs are sensitive only to substrate-borne vibrations (Čokl, 1983; Gogala et al., 1974). Many heteropteran groups possess highly sensitive hair-like mechanoreceptors known as trichobothria (Drašlar, 1973; Schuh & Weirauch, 2020), yet their removal does not appear to affect vibration perception, and the species studied did not respond to the airborne component of their stridulatory signals (Gogala et al., 1974). However, it cannot be excluded that trichobothria may allow at least some Heteroptera to detect nearfield airborne vibrations or air currents (Kavčič et al., 2013). In addition, the majority of heteropterans lack tympana, the typical acoustic sensors of other insects (Strauß & Lakes-Harlan, 2014), with the exception of the aquatic and semiaquatic infraorder Nepomorpha (Parsons, 1964).

Their lack of acoustic sense organs means that most Heteroptera must only exploit the vibrational component of their vibroacoustic signal, or that in some cases their signals are directed towards other species (e.g. predators) capable of perceiving acoustic sound (Fig. 5). True acoustic communication is confined to the aquatic bugs (Nepomorpha), all of which are able to sense the acoustic component of the stridulatory signals they produce through tympana (Papáček, Štys & Tonner, 1990; Parsons, 1964), in conjunction with vibrational and chemical signals (Jansson, 1972, 1989; Parsons, 1964). However, the presence of tympana in Nepomorpha is not uniformly associated with the possession of stridulatory mechanisms, as the latter are present in only about 28% of all nepomorphan species (Table S1). This suggests that tympanic organs in Nepomorpha may have evolved to detect cues or signals from other biotic or abiotic sources instead of stridulation, which may be particularly important for survival in their aquatic habitat. Overall, we can find no correlation between the use of acoustic signals and the development of tympanic organs in Heteroptera.

Caeliferan Orthoptera may have followed a similar trajectory, as it has been proposed that their tympana evolved either for evading predators or controlling flight manoeuvres, and only later became co-opted for communication (Song et al., 2020). Interestingly, ensiferan Orthoptera have been proposed to have used stridulation initially for defensive purposes, without being able to perceive the signals they produced. However, at a later stage in their evolution, the independent evolution of tympana led to the incorporation of stridulation into their mating ritual, and since then, the two structures have co-evolved (Song et al., 2020). Considering the above, the incorporation of acoustic stridulatory signals into nepomorphan and orthopteran communication may have evolved via sensory exploitation, a hypothesis which postulates that senders exploit pre-existing receiver biases by creating signals that match these biases (McClintock & Uetz, 1996). Evidently, more studies are needed to explain why Nepomorpha, among all other Heteroptera, have evolved (or ancestrally retained) tympanal ears.

The likely inability of most Heteroptera to detect the acoustic component of their stridulatory signals is important to our understanding of the origins of stridulation, as it demonstrates that the evolution of signals associated with acoustic energy is not always associated with, and does not necessarily lead to, the evolution of conspecific acoustic communication (i.e. acoustic signal generation and detection). Indeed, we show that use of a signalling modality can pre-date its perception by the species producing the signal, in the same way that some animals cannot perceive the acoustic component of their own defensive signals [e.g. hissing, stridulation and tail-rattling in snakes (Gans & Maderson, 1973), defensive ultrasounds in some moths (O'Reilly et al., 2019)]. Consequently, the sensory evolution of Heteroptera is highly unusual compared to that of most other animals, including the arthropod outgroups examined here, as in most other cases a sense organ evolves in parallel with the emitted signalling modality (Shaw, 1994; Strauß & Lakes-Harlan, 2014).

This unusual aspect of heteropteran sensory evolution has important implications for our understanding of the selective pressures that drive the development of stridulation. Stridulatory acoustic signals may evolve for interspecific receivers, especially in a defensive context, which does not require their sensing by their own species. This is indeed the case in many groups of animals (e.g. aposematic millipedes and brittle stars), which are blind or have limited optic receptors respectively (Grober, 1988; Marek & Bond, 2009; Marek et al., 2011). The above provides further support for the behavioural data we presented here (Fig. 5), and our hypothesis that predation is a major selective force for the evolution of stridulation in Heteroptera. Even so, the apparent ‘deafness’ of Heteroptera to the acoustic component of their signal is still unusual, as the examined outgroups, such as crustaceans and spiders (Araneae), which also employ stridulatory signals for their defence (Fig. 5B), possess mechanosensors that are highly sensitive to the signals they emit (Boon, Yeo & Todd, 2009; Breithaupt & Tautz, 1988; Gordon & Uetz, 2011; Horch & Salmon, 1969; Keil, 1997; Salmon & Atsaides, 1968; Sweger & Uetz, 2016).

Alternatively, when the acoustic signals are used for conspecific communication, the receiver must be detecting them through other mechanical senses. Therefore, vibroacoustic mechanical signals may evolve to be detected in conspecifics by sensors adapted for other mechanical modes, here by substrate-borne vibration sensors (Čokl, 1983; Gogala et al., 1974; Nishino et al., 2016). This also confirms our suggestion that sexual selection for increased courtship signal complexity is the other major driver in the evolution of stridulatory mechanisms.

(4) Is stridulation a persistent strategy through evolutionary time?

We have found that stridulation has evolved hundreds of times independently across arthropods, and we provide evidence that it serves as a valuable adaptive strategy, especially in defensive and sexual contexts (Fig. 5). However, a question that has never been addressed before is how long stridulatory organs persist in a lineage. In Heteroptera and the examined arthropod outgroups, stridulation is mainly confined to the tips of the phylogeny. Typically, stridulation is found to be present in a single species, genus, or cluster of closely related genera (Table 1) comprising only a small percentage of the total diversity of the group (Table S1). This may suggest that stridulation, although useful, provides only a short-term evolutionary solution, which is expected if it is part of an ongoing predator–prey or sexual arms race that requires rapid morphological and behavioural shifts. Indeed, even stridulatory structures necessary for mating can be lost within a few generations in the face of intense predation or parasitism (Zuk, Rotenberry & Tinghitella, 2006).

Within Heteroptera, there are a few notable exceptions where stridulatory organs are nearly ubiquitous and very ancient. These include assassin bugs (Reduviidae) [likely origin of mechanism 178 million years ago (Mya) (Hwang & Weirauch, 2012; Weirauch, 2008)], the Corixidae (about 170 Mya; Ye et al., 2019, 2023), the Micronectidae (about 150 Mya; Ye et al., 2019, 2023), and the cydnoid families (earliest fossil about 145 Mya, likely appeared much earlier; Yao, Cai & Ren, 2007). In these four groups, secondary losses of stridulatory mechanisms are extremely rare (Grazia, Schuh & Wheeler, 2008; Cai, Chou & Lu, 1994; Lis & Heyna, 2001), suggesting that stridulation plays a ubiquitous adaptive role in the lives of its bearers, and that there is therefore long-term selective pressure for its maintenance.

Likewise, among the arthropod outgroups that we have examined, the tegmino-tegminal mechanism of ensiferan Orthoptera dates back to the Carboniferous (about 300 Mya; Song et al., 2020), and in many families of spiders, the ubiquitous presence of certain stridulatory organs also suggests an ancient retention; examples include the Dipluridae (90 Mya) and Hexathelidae (200 Mya; Opatova et al., 2020). The detailed reasons why stridulatory mechanisms may persist for hundreds of millions of years in some taxa, but are short lived in other arthropods remain obscure. For example, the abovementioned groups do not share any particular behavioural, ecological, or morphological characteristics, with the possible exception of the assassin bugs and spiders, both of which are predators.