Although almost 990 species of bony fish (Osteichthyes) actively produce sounds, evidence for active sound production by elasmobranchs—sharks, rays, and skates—is scarce (Looby et al., 2022). To date, there have been only 27 examinations of sound production by elasmobranchs (Looby et al., 2022), and of the 13 recorded occurrences, the majority have been passive sounds associated with feeding (e.g., shell crushing; Ajemian et al., 2021). The only confirmed case of active sound production occurred when captive cownose rays Rhinoptera bonasus produced short, sharp clicks under duress, that is, forceful prodding (Fish & Mowbray, 1970). Two further examples of “active” sound production have been documented—“crunching” sounds with chewing and “mumbling” after ingestion in a captive common stingray Dasyatis pastinaca and “rumbles” when grabbing food by a captive picked dogfish Squalus acanthias (Shishkova, 1958)—but these were both associated with feeding and are less convincing.

There have been no confirmed examples of active sound production by elasmobranchs in the wild, despite attempts to record the behavior outside of captive settings. Although there are some anecdotal reports, they remain unproven or are given without sources. Bass and Rice (2010), for example, reported that “stingrays have been anecdotally documented to grind their teeth as an audible defence warning signal,” without providing a reference for this statement. By comparison, the hearing capabilities of elasmobranchs have received much more attention (Mickle et al., 2020; Myrberg, 2001). Elasmobranchs are most sensitive to low-frequency sounds between 40 and 1500 Hz, with peak sensitivities between 200 and 400 Hz, but audiograms have only been produced for 10 species (Chapuis & Collin, 2022). There is more evidence relating to behavioral responses to sounds. Many sharks are attracted to certain sounds, like those of struggling prey, and can change their behaviors in response to such sounds (Gardiner et al., 2012). Other sounds, such as the vocalizations of killer whales, Orcinus orca, reportedly repulse and cause a fleeing response in epipelagic sharks, which could fall prey to these odontocetes (Chapuis et al., 2019; Myrberg, 2001). Similarly, in some shark species an unexpected sound or the sudden increased intensity of a sound can result in rapid withdrawal from the sound source (Klimley & Myrberg, 1979; Myrberg, 2001; Myrberg et al., 1978). Sound may also elicit less obvious responses, for example, the southern stingray Hypanus americanus has been shown to alter its swimming behavior (i.e., resting less, increasing swimming activity, and breaching the surface more often) in response to certain sounds (Mickle et al., 2020).

Though it is clear that elasmobranchs can hear and many can also respond to sound in various ways, hearing capacity is not necessarily linked to the ability to produce acoustic sound (Mélotte et al., 2018), and until now there has been limited evidence to suggest that any elasmobranchs have the ability to actively produce sound themselves. Here we present the first records of voluntary active sound production in the wild by three individuals of two species of stingray: the mangrove whipray Urogymnus granulatus (Figure 1b) and the cowtail stingray Pastinachus ater (Figure 1c). The sounds recorded from all three individuals were characterized by a series of very short, broadband clicks (Figure 1d, Appendix S1: Table S1) and were associated with movement of the spiracles and cranial area. In all recorded observations, the ray commenced producing sounds in response to an observer approaching closely and ceased sound production when the distance between the ray and observer increased. We suggest hypotheses for the potential purposes and mechanisms of the sound production and highlight that further research into this ability is needed.

The three recorded observations occurred as follows. On 22 December 2018, Philip Christoff (PC) was undertaking a recreational closed-circuit rebreather dive at the Deep Turbo dive site northeast of Gili Trawangan, Gili Islands, Indonesia (Figure 1a, approximately −8.339491°, 116.048697°). At around 9:30 A.M., PC sighted an adult female mangrove whipray U. granulatus (disk width around 1 m) resting under the sand. Following a slow approach by PC, the ray appeared disturbed and slowly swam away parallel to the diver. It began making clicking sounds when PC came within ~2 m. Each click coincided with movement of the spiracle and partial retraction of the eye (Video S1). Eleven broadband clicks were recorded, ranging from 0.017 to 0.025 s in duration (mean ±SE = 0.021 ± 0.001) (Figure 1d-ii, Appendix S1: Table S1). Clicks 1–10 had a peak frequency of 1500 Hz, and the 11th click had a peak frequency of 1031 Hz (Appendix S1: Table S1). Mean bandwidth (±SE) across all clicks was 22.731 kHz ± 33.883 Hz (Appendix S1: Table S1). Secondary pulses were also noted in the waveforms of each click (Figure 1d-ii); however, based on their similarity to the primary pulses and lower relative amplitude, these were considered echoes within the camera housing (Nauticam housing on Sony RX100M5 digital camera).

In February 2018, J. Javier Delgado Esteban (JJDE) observed sound production by a juvenile mangrove whipray U. granulatus (disk width around 40 cm) while snorkeling in the shallow, inshore waters of Geoffrey Bay, Magnetic Island, Great Barrier Reef, Queensland, Australia (Figure 1a, −19.153243°, 146.867342°). The juvenile was part of a larger group, but it had been separated from the group at the time it was recorded. Seven distinct broadband clicks were observed, ranging from 0.01 to 0.017 s in duration (mean ±SE = 0.013 ± 0.001) (Figure 1d-i, Appendix S1: Table S1). The first six clicks had a peak frequency of 1687 Hz, but the seventh click had a peak frequency of 1875 Hz (Appendix S1: Table S1). Mean bandwidth (±SE) across all clicks was 22.314 kHz ± 902.754 Hz (Appendix S1: Table S1). The clicks were described as originating from the ventral area of the animal, with each visibly coinciding with contractions of the spiracles (Video S1). Immediately after the sounds were emitted, the rest of the group of stingrays approached both the individual ray and the stationary snorkeler. JJDE observed numerous instances of sound production in this group of juvenile mangrove whiprays over several days, but these were not captured on film.

The third observation was recorded in October 2017 by John Gaskell (JG) when snorkeling with a group of cowtail stingrays P. ater, which are known to aggregate in shallow waters off the southern beach of Heron Island (Figure 1a; −23.443510°, 151.913074°), Great Barrier Reef, Queensland, Australia. While filming in water approximately 70 cm deep, JG pursued one animal that was slowly swimming away from him. When JG came within a distance of less than a disk width of the ray, the animal started to produce loud clicking sounds that coincided with contractions of the cranial and spiracle area of the animal (Video S1). Five distinct broadband clicks were recorded, ranging from 0.021 to 0.091 s in duration (mean ±SE = 0.065 ± 0.012) (Figure 1d-iii, Appendix S1: Table S1). The first click had a peak frequency of 1406 Hz, while clicks 2–5 peaked at 1500 Hz (Appendix S1: Table S1). Mean bandwidth (±SE) across all clicks was 23.904 kHz ± 17.776 Hz (Appendix S1: Table S1). JG observed similar sound production two more times in the same species over 6 days, but these events were not captured on film.

In addition to the foregoing observations recorded on film, in the early 2010s commercial divers from Far North Queensland were hand-collecting sea cucumbers in the inshore waters of the Great Barrier Reef and Coral Sea up to 20 m in depth and reported that on multiple occasions cowtail stingrays P. ater, when approached in murky waters, produced loud clicking sounds while fleeing from divers (B. E. Wueringer, unpublished).

The observed sound production in both species of rays appeared to serve the purpose of agonistic displays. In sharks, agonistic displays are relatively common and mainly comprise visual components, such as lowering of the pectoral fins (silent) or tail slapping or popping, which does produce sounds (Martin, 2007), although these sounds are based on direct observations, and recording and analysis are still needed. In rays, agonistic displays observed to date generally have involved physical intra- and interspecific interactions, such as biting, chasing, and shoving (Newsome et al., 2004; Pini-Fitzsimmons et al., 2021). In contrast, the sudden loud sounds reported here appear to be more likely to represent a warning or serve to startle predators, such as sharks, which have been shown to rapidly flee from sudden unexpected sounds (Klimley & Myrberg, 1979; Myrberg, 2001). Further, since the rays are able to produce these sounds while fleeing from a fight-or-flight situation, they do not have to sacrifice their swimming efficiency in order to produce a warning signal (Martin, 2007).

Both juvenile mangrove whiprays U. granulatus and cowtail stingrays P. ater appear social and are often observed feeding and resting in groups, likely as a predator-avoidance strategy (Kanno et al., 2019; Martins et al., 2020a, 2020b). In the case of JJDE's observation, other juvenile mangrove whiprays were observed gathering around the individual filmed producing the clicks and appeared to be doing so in response to the produced sounds. Sound production may therefore alert conspecifics to the need to aggregate in response to danger, which also implies a role in intraspecific communication.

The bandwidth of the clicks produced by U. granulatus and P. ater examined here spanned the expected hearing range of elasmobranchs (40–1500 Hz; Chapuis & Collin, 2022), providing some evidence that their predators (Carcharhinus melanopterus and Negaprion acutidens; Kanno et al., 2019; Martins et al., 2020a, 2020b) and conspecifics can hear these sounds, although peak frequencies of the clicks occurred at the top or above this hearing range (1031–1875 Hz). However, audiograms have only been produced for a few elasmobranch species, and none has been produced for U. granulatus, P. ater, or their known predators (Chapuis & Collin, 2022). Further assessment of the hearing abilities of these species is therefore necessary to clarify the role of the produced sounds in agonistic displays or predator avoidance.

The exact mechanism of sound production remains unclear but appears to be similar in both species. In all video recordings, contractions of the spiracles and associated gill openings are visible simultaneously with the clicking sounds (Video S1), indicating that sounds may be produced through fast contractions of the cranial and gill area. Because both species lack myliobatiform grinding plates, which would be positioned on the palate, but instead possess teeth limited to their jaws, the anecdotally proposed mechanism of sound production using grinding plates (Bass & Rice, 2010) is likely incorrect in this instance. Whether the sound production is achieved through fast expulsion of water or another internal mechanism is plausible but awaits verification, and further research on the internal morphology of these rays is required.

The observations presented here highlight that further research on sound production in elasmobranchs is warranted, especially considering the limited number of examinations in this group to date (Looby et al., 2022). Our observations are of species that are encountered relatively often by snorkelers and yet were not previously known to produce sounds. Other similar species may also produce sounds, but anecdotal records may have not yet come to light; thus, our paper may serve to bring to light further examples from the public and researchers. All of the examples presented here were captured opportunistically with handheld digital cameras, and future targeted research should endeavor to use standardized hydrophones (Lindseth & Lobel, 2018; Rountree et al., 2006), where possible, to allow for better control of sound distortion and echoes.

Our observations and Fish and Mowbray's (1970) observations in captivity mean that three ray species (of approximately 245 Myliobatiformes [Stein et al., 2018]) have now been convincingly shown to actively produce sounds and to do so in the wild, voluntarily, and without artificial stimuli. Although elasmobranchs are generally not considered to be sound producing (Looby et al., 2022), our study illustrates that this is a misconception and more research into their ability to produce and hear such sounds is required.

尽管将近 990 种硬骨鱼 (Osteichthyes) 主动发出声音，但板鳃类动物（鲨鱼、鳐鱼和鳐鱼）主动发出声音的证据很少（Looby 等人，  2022 年）。迄今为止，只有 27 次检查软骨鱼类发出的声音（Looby 等人，  2022 年），而在记录的 13 次事件中，大多数是与进食相关的被动声音（例如，贝壳破碎；Ajemian 等人，  2021 年）。唯一已确认的主动发声案例发生在圈养的牛鼻鳐Rhinoptera bonasus在胁迫下发出短促、尖锐的咔哒声时，即强力刺激（Fish & Mowbray，  1970 年）). 记录了另外两个“主动”声音产生的例子——在被圈养的普通黄貂鱼Dasyatis pastinaca吞食后发出“嘎吱嘎吱”的咀嚼声和“咕哝”声，以及在被圈养的角鲨Squalus acanthias抓取食物时发出的“隆隆声” （Shishkova，  1958 年） )——但这些都与进食有关，而且不太令人信服。

尽管试图记录圈养环境之外的行为，但尚无证实的板鳃类动物在野外主动发声的例子。尽管有一些轶事报道，但它们仍未得到证实或没有来源。例如， Bass 和 Rice ( 2010 ) 报告称“据传闻，黄貂鱼磨牙是一种可听见的防御警告信号”，但并未为此声明提供参考。相比之下，软骨动物的听觉能力受到了更多关注（Mickle 等人，  2020 年；Myrberg，  2001 年）). Elasmobranchs 对 40 到 1500 Hz 之间的低频声音最敏感，峰值灵敏度在 200 到 400 Hz 之间，但只为 10 个物种制作了听力图 (Chapuis & Collin,  2022 )。有更多关于对声音的行为反应的证据。许多鲨鱼会被某些声音所吸引，例如挣扎中的猎物的声音，并且可以根据这些声音改变它们的行为 (Gardiner et al.,  2012 )。据报道，其他声音，例如虎鲸Orcinus orca的叫声，会排斥上层鲨鱼并引起逃跑反应，这些鲨鱼可能会成为这些齿鲸的猎物（Chapuis 等人，  2019 年；Myrberg，  2001 年）). 同样，在一些鲨鱼物种中，意外的声音或突然增加的声音强度会导致迅速从声源中退出（Klimley & Myrberg，  1979 年；Myrberg，  2001 年；Myrberg 等，  1978 年）。声音也可能引起不太明显的反应，例如，南方黄貂鱼Hypanus americanus已被证明会改变其游泳行为（即，减少休息，增加游泳活动，并更频繁地突破水面）以响应某些声音（Mickle 等人.,  2020 年）。

虽然很明显板鳃类动物可以听到并且许多也可以以各种方式对声音做出反应，但听力能力不一定与产生声音的能力相关（Mélotte 等人，  2018 年），并且直到现在，只有有限的证据表明表明任何板鳃类动物都有能力自己主动发出声音。在这里，我们展示了两种黄貂鱼的三个个体在野外主动主动发声的第一批记录：红树林鞭鳐Urogymnus granulatus（图 1b）和牛尾黄貂鱼Pastinachus ater（图 1c）。所有三个人记录的声音的特点是一系列非常短的宽带咔哒声（图 1d，附录 S1：表 S1），并且与气孔和颅骨区域的运动有关。在所有记录的观察中，射线开始发出声音以响应靠近的观察者，并在射线和观察者之间的距离增加时停止声音产生。我们提出了声音产生的潜在目的和机制的假设，并强调需要进一步研究这种能力。

三个记录的观察发生如下。2018 年 12 月 22 日，Philip Christoff (PC) 在印度尼西亚吉利群岛吉利特拉旺岸东北部的 Deep Turbo 潜水点进行休闲闭路循环呼吸器潜水（图 1a，大约 -8.339491°，116.048697°）。上午 9 点 30 分左右，PC 发现了一只成年雌性红树林 whipray U. granulatus（圆盘宽度约 1 m）在沙子下休息。在 PC 缓慢接近后，鳐鱼似乎受到了干扰，并慢慢地与潜水员平行游走。当 PC 进入约 2 m 以内时，它开始发出咔哒声。每次点击都与气门的移动和眼睛的部分收缩一致（视频 S1）。记录了 11 次宽带点击，持续时间从 0.017 到 0.025 秒不等（平均值±SE = 0.021±0.001）（图 1d-ii，附录 S1：表 S1）。点击 1–10 的峰值频率为 1500 Hz，第 11 次点击的峰值频率为 1031 Hz（附录 S1：表 S1）。所有点击的平均带宽 (±SE) 为 22.731 kHz ± 33.883 Hz（附录 S1：表 S1）。每次点击的波形中也记录了次级脉冲（图 1d-ii）；然而，基于它们与初级脉冲的相似性和较低的相对振幅，

2018 年 2 月，J. Javier Delgado Esteban (JJDE) 观察到幼年红树林 whipray U. granulatus的声音产生（圆盘宽度约 40 厘米）在澳大利亚昆士兰州大堡礁磁岛杰弗里湾近岸浅水区浮潜（图 1a，-19.153243°，146.867342°）。少年是一个更大群体的一部分，但在记录时它已与群体分开。观察到七种不同的宽带点击，持续时间从 0.01 到 0.017 秒不等（平均值±SE = 0.013±0.001）（图 1d-i，附录 S1：表 S1）。前六次点击的峰值频率为 1687 Hz，但第七次点击的峰值频率为 1875 Hz（附录 S1：表 S1）。所有点击的平均带宽 (±SE) 为 22.314 kHz ± 902.754 Hz（附录 S1：表 S1）。咔哒声被描述为源自动物的腹侧区域，每次明显与气孔收缩一致（视频 S1）。声音发出后，黄貂鱼群的其余部分立即靠近单独的鳐鱼和静止的浮潜者。JJDE 在几天内观察到这组幼年红树林鞭鳐中有许多声音产生的例子，但这些都没有在电影中捕捉到。

第三次观察是在 2017 年 10 月由 John Gaskell (JG) 与一群牛尾黄貂鱼P. ater一起浮潜时记录的，已知聚集在澳大利亚昆士兰州大堡礁苍鹭岛南部海滩附近的浅水区（图 1a；−23.443510°，151.913074°）。在大约 70 厘米深的水中拍摄时，JG 追赶一只慢慢游离他的动物。当 JG 进入小于射线圆盘宽度的距离内时，动物开始发出响亮的咔嗒声，这与动物的颅骨和气孔区域的收缩一致（视频 S1）。记录了五次不同的宽带点击，持续时间从 0.021 到 0.091 秒不等（平均值±SE = 0.065±0.012）（图 1d-iii，附录 S1：表 S1）。第一次点击的峰值频率为 1406 Hz，而点击 2-5 的峰值频率为 1500 Hz（附录 S1：表 S1）。所有点击的平均带宽 (±SE) 为 23.904 kHz ± 17.776 Hz（附录 S1：表 S1）。

除了上述记录在胶片上的观察结果外，2010 年代初期，昆士兰远北地区的商业潜水员还在大堡礁和珊瑚海的近岸水域中，深度达 20 米，手工采集海参，并多次报告说牛尾黄貂鱼P. ater在浑浊的水域中接近时会在逃离潜水员时发出响亮的咔嗒声（BE Wueringer，未发表）。

在两种射线中观察到的声音产生似乎都用于竞争性显示的目的。在鲨鱼中，竞争性表现相对常见，主要包括视觉成分，例如降低胸鳍（无声）或尾巴拍打或爆裂，这确实会产生声音（Martin，  2007），尽管这些声音是基于直接观察，并且仍然需要记录和分析。在射线中，迄今为止观察到的竞争性表现通常涉及种内和种间的物理相互作用，例如咬、追逐和推搡（Newsome 等人，  2004 年；Pini-Fitzsimmons 等人，  2021 年）). 相比之下，这里报道的突然响亮的声音似乎更有可能代表警告或惊吓捕食者，例如鲨鱼，这些捕食者已被证明会迅速逃离突然的意外声音（Klimley & Myrberg,  1979 ; Myrberg,  2001） . 此外，由于鳐鱼在逃离战斗或逃跑的情况时能够发出这些声音，因此它们不必为了发出警告信号而牺牲游泳效率 (Martin, 2007 )。

幼年红树林鞭鳐U. granulatus和牛尾黄貂鱼P. ater都表现出群居性，经常被观察到成群进食和休息，这可能是一种躲避捕食者的策略（Kanno 等人，  2019 年；Martins 等人，  2020a和2020b）。在 JJDE 的观察案例中，观察到其他幼年红树林鞭鳐聚集在拍摄的发出咔哒声的个体周围，并且这样做似乎是为了响应产生的声音。因此，声音的产生可能会提醒同种动物需要聚集以应对危险，这也意味着在种内交流中发挥作用。

此处检查的U. granulatus和P. ater产生的咔嗒声的带宽跨越了板鳃类动物的预期听觉范围（40–1500 Hz；Chapuis & Collin，  2022），提供了一些证据证明它们的捕食者（Carcharhinus melanopterus和Negaprion acutidens；Kanno et al.,  2019；Martins et al.,  2020a , 2020b ) 和同种动物可以听到这些声音，尽管咔哒声的峰值频率出现在该听觉范围 (1031–1875 Hz) 的顶部或之上。然而，只为少数软骨鱼类制作了听力图，没有为U. granulatus、P. ater制作过听力图, 或它们已知的捕食者 (Chapuis & Collin,  2022 )。因此，有必要进一步评估这些物种的听觉能力，以阐明所产生的声音在竞争性展示或捕食者回避中的作用。

声音产生的确切机制仍不清楚，但在这两个物种中似乎相似。在所有视频记录中，气孔和相关鳃孔的收缩与咔嗒声同时可见（视频 S1），表明声音可能是通过颅骨和鳃区域的快速收缩产生的。由于这两个物种都缺乏位于上颚的鳃足状磨盘，而是拥有仅限于下颌的牙齿，因此根据传闻提出的使用磨盘发声的机制（Bass & Rice，  2010 年）) 在这种情况下可能不正确。声音的产生是通过快速排出水还是其他内部机制实现的，这似乎是合理的，但有待验证，需要对这些射线的内部形态进行进一步研究。

此处呈现的观察结果强调，有必要进一步研究软骨鱼类的声音产生，特别是考虑到迄今为止该组的检查数量有限（Looby 等人，  2022 年）。我们的观察是浮潜者相对经常遇到的物种，但以前不知道它们会发出声音。其他类似物种也可能发出声音，但轶事记录可能尚未曝光；因此，我们的论文可能有助于揭示来自公众和研究人员的更多例子。此处介绍的所有示例都是用手持数码相机随机捕获的，未来的针对性研究应努力使用标准化水听器（Lindseth & Lobel，  2018 年；Rountree 等人，  2006 年）)，在可能的情况下，以便更好地控制声音失真和回声。

我们的观察以及 Fish 和 Mowbray（1970 年）在圈养条件下的观察表明，三种鳐鱼物种（大约 245 种蝠纲 [Stein 等人，  2018 年]）现在已被证明可以主动发出声音，并且在野外自愿发出声音，并且没有人工刺激。尽管板鳃类动物通常不被认为会发出声音（Looby 等人，  2022 年），但我们的研究表明这是一种误解，需要对它们发出和听到此类声音的能力进行更多研究。