Physical effects of sound exposure from underwater explosions on Pacific mackerel (Scomber japonicus): Effects on non-auditory tissues^a)

High-intensity underwater explosions can cause injury and death in fishes (Yelverton et al., 1975; Keevin and Hempen, 1997; Carlson et al., 2019; Dahl et al., 2020). There have been few well-designed studies that have related the physical impacts on fishes to the measured received level, or “acoustic dose,” from an underwater explosion [Popper et al. (2014); as reviewed in Dahl et al. (2020)]. Moreover, many of the early studies on effects of explosives were hindered by lack of controls, limited reporting of acoustic metrics, small or heterogenous samples, and/or insufficient necropsy procedures [e.g., Fitch and Young (1948), Hubbs and Rechnitzer (1952), and Yelverton et al. (1975)]. Notwithstanding these limitations, this historical research, in combination with additional studies [e.g., Gaspin (1975), Gaspin et al. (1976), Wiley et al. (1981), Settle et al. (2002), Govoni et al. (2003), and Govoni et al. (2008)], provides useful insight into the susceptibility of fishes to injury from underwater explosions.

These earlier studies demonstrate that barotrauma [physical damage to tissues caused by a pressure differential between an internal gas space (e.g., swim bladder in fishes) and the surrounding medium—see Carlson (2012)] can occur when the swim bladder compresses and then rapidly over-expands, risking rupture and potentially damaging nearby organs and tissues. Furthermore, over-expansion of the swim bladder can be subsequently exacerbated by a rapid pressure drop due to arrival of a negative pressure signal from the explosive signal that is reflected from the sea surface (Wiley et al., 1981).

Barotrauma has also been observed during rapid pressure changes and decompression as fishes pass through hydro turbines (Stephenson et al., 2010; Carlson, 2012) and from exposure to other impulsive sources such as impact pile driving (Halvorsen et al., 2011; Halvorsen et al., 2012a; Halvorsen et al., 2012b; Casper et al., 2017). Other potential impacts from explosive exposure include organ and tissue damage such as hemorrhaging of various internal organs (e.g., kidney, liver, or spleen), embolism (obstruction of blood vessels by clotting or the formation of air bubbles), tears or breaks in organs or membrane lining within the body cavity, and stunning (Hubbs and Rechnitzer, 1952; Settle et al., 2002; Govoni et al., 2003).

Our group reported injuries for the Pacific sardine (Sardinops sagax) resulting from exposure to underwater explosions (Dahl et al., 2020). In that study, examination of external and internal anatomy following explosive exposure found that the swim bladder and kidney were most susceptible to barotrauma as compared with other tissues. Results from Dahl et al. (2020) further demonstrated that injuries generally decreased with increasing range from the explosive source, corresponding to decreasing acoustic dose. However, there was also increased injury rate from approximately 125 to 150 m from the explosion. A plausible explanation was proposed, connecting increased injury with negative pressure as influenced by arrivals from the bottom bounce path with grazing angle less than the critical angle [for more information on the hypothesis for this effect, see Dahl et al. (2020)]. The pressure drops may have increased the extent of swim bladder over-expansion per a mechanism proposed by Goertner (1978) and Wiley et al. (1981).

Since fish species vary in anatomy which could affect susceptibility to underwater explosions, this study extends the approach of Dahl et al. (2020) to an unrelated species, the Pacific mackerel (Scomber japonicus). A companion study (Smith, 2022) examines the potential impact on the inner ear of Pacific mackerel exposed to the same underwater explosions.

The experimental approach and methods used in this effort were similar to those described in Dahl et al. (2020). A brief outline of methods is provided here except in the few cases when there have been substantive changes. The general paradigm involved taking fish to an offshore site, lowering them to depth in cages at specific distances from the explosive source, exposing them to the explosion, retrieving the fish to the surface for euthanasia, and then performing necropsy on each animal to look for physical effects of the explosive exposure.

Pacific mackerel are a member of the tuna and mackerel family (Scombridae). They are found in large schools that vertically migrate over the diel (day-night) cycle from the surface to 300 m (Love and Passarelli, 2020). Pacific mackerel have a swim bladder that is approximately one-third the length and approximately one-half the width of the fish. The swim bladder is loosely connected to the roof of the abdominal cavity. The kidney sits directly above, and in close contact with, the swim bladder.

Adult Pacific mackerel were acquired using hook and line in the nearshore waters off San Diego, CA over a period of a few days in August 2019. The fish were held for approximately three weeks prior to the beginning of the study in an open water holding facility near the entrance of San Diego Bay, CA. Mortality rates in the collection were approximately 10% over a 5-week period (three-week acclimatization period and two-week experiment duration), with most mortality occurring within a few days of collection.

Fish were collected under California Department of Fish and Wildlife scientific collection permit SC-13924. All procedures were approved by the Naval Information Warfare Center Pacific Institutional Animal Care and Use Committee (IACUC, protocol No. 131–2018), and the Navy Bureau of Medicine and Surgery. The study followed all relevant U.S. Department of Defense guidelines for the care and use of laboratory animals.

The experimental site is characterized by a flat sandy bottom with a depth of 19.5 m (±1 m) and is located approximately 5 km offshore from San Diego within the U.S. Navy's Silver Strand Training Complex. Five trials (one per day), each with a single explosion, were conducted between 12 and 18 September 2019. Wind and wave conditions varied little across trial days, with a root-mean-square (rms) wave height of 0.18 m (data from NOAA NDBC station 46235 located nearby at Imperial Beach, CA).

On the morning of each trial, fish were taken from the holding facility and placed into transport containers aboard the research vessel. Once at the experimental site, 12 fish were randomly placed in each of the five exposure cages (round enclosures measuring approximately 0.6 m high by 1 m in diameter) which were deployed at ranges of 21–807 m, starting with the cage closest to the detonation site and then working outward. Each cage was lowered from the surface to about 10.5 m depth over an approximately 2-min period. Twelve randomly selected control fish deployed in a sixth cage were handled identically to treatment fish except that they were removed from the water prior to the explosion.

The explosive for each trial consisted of 4.5 kg of C-4 [equivalent to 6.2 kg of trinitrotoluene (TNT)]. Detonations were also set at a depth of 10.5 m (±0.5 m) (approximately half the water depth at the site) to avoid immediate interaction of the detonation with the bottom. The detonations were conducted by Navy explosive ordnance disposal personnel following the U.S. Navy's Hawaii-SOCAL Environmental Impact Statement (Department of the Navy, 2022) and associated permits.

A remotely operated vehicle (ROV) was used to observe and record video of the fish in each cage before the exposure. During each underwater explosion, the ROV also recorded video of one of the fish cages.

Post-exposure, fish were raised to the surface in the order in which they were deployed at a rate of approximately 2 min per cage and carefully observed in the net. If no movement was apparent the fish were declared dead and placed on ice packs. Otherwise, fish were euthanized with buffered MS-222 before being placed on ice packs. Two fish from each treatment and control cage were randomly selected and prepared for examination of inner ears (Smith, 2022). The remaining ten fish per cage were taken for necropsy.

Twenty-one (of 25) exposure cages were instrumented with non-directional, tourmaline blast sensors (PCB W138A01, PCB Piezotronics, Depew, NY) with nominal sensitivity of 0.73 mV/kPa, equivalent to a received sensitivity of −243 dB re V/μPa. These sensors were time-synchronized and recorded at a 2 MHz sampling rate using the Naval Submarine Medical Research Laboratory's (NSMRL) custom designed Aquarius recording system. The four remaining cages located at 400 m or more from the detonation were instrumented with an autonomous SoundTrap recorder (Ocean Instruments, Aukland, New Zealand) using an HTI-96-min hydrophone (High Tech Inc., Long Beach, MS) with received sensitivity of −221.9 dB re V/μPa, and sampling rate 288 kHz.

Measurements of sound speed versus depth, using a conductivity, temperature, and depth instrument (Sontek, Inc.) were taken within a few minutes and within a few hundred meters of the explosion. Over the 12–18 September period of the experiment, sound speed versus depth was characterized by an approximately linear thermocline over which sound speed decreased from about 1522 m/s starting at depth 4 m, reaching 1505 m/s at depth 16 m; the upper and lower 4 m in the water column having approximately constant speed of 1522 and 1505 m/s, respectively.

Three acoustic metrics were obtained from the explosive measurements. Defining p(t) as the measured time series of pressure at any given range, then peak sound pressure level (SPL) is the maximum absolute value expressed in dB re 1 μPa. Sound exposure level (SEL) is the time integral of p²(t) expressed in dB re 1 μPa² s. A third dosage measurement, the pressure impulse, is based on the cumulative integral of p(t) over time with the maximum value taken as the specific measure of pressure impulse and expressed in Pa s.

Necropsy of each fish included collecting data on length and weight, examining the fish externally, and then dissecting the fish to examine the organs and tissues in the abdominal cavity. Investigators performing necropsies were blind to the fish's treatment and assessed external and internal injuries based on predesignated categories, although other observations were also recorded.

Injuries to internal organs and tissues, as well as observations of blood pooling within the body cavity and swim bladder, are summarized in Table I. Following Halvorsen et al. (2012b), injuries were placed into trauma categories based on their physiological significance. Internal hematoma or bruising (i.e., discoloration of tissue) was considered a moderate injury, which are those injuries that could potentially affect fish health but are recoverable if fish are not subjected to additional stressors (Casper et al., 2013; Casper et al., 2017). Internal hemorrhaging (loss of blood from damaged blood vessel), blood pooling indicative of internal hemorrhaging, or rupture of internal organs was presumed to be a mortal injury and placed in the “severe” trauma category. Observations of blood pooling are not injuries per se but will be referred to as “injuries” for simplicity in this manuscript.

External structures including the fins, eyes, and vent were examined. The few external injuries observed were similar in frequency of occurrence between controls and treatments and therefore not carried forward in the analysis.

Analysis was conducted in r (R Core Team, 2021). A Fisher's exact test was used to test for homogeneity of each potential injury type across the five control cages. The number of exposure cages that had significantly higher rates of each injury as compared to controls was determined by plotting the response for each injury type within each exposure cage (n = 25), along with 95% confidence intervals (calculated using Wilson binomial method), versus the pooled control response. Further inspection was conducted using Fisher's exact test to determine statistical significance at the α = 0.05 level. There were no corrections made for multiple comparisons, therefore, at least 4 out of 25 exposure fish cages needed to indicate a significant difference (test-wise α = 0.05) for the injury type to be considered indicative of exposure to an underwater explosion resulting in a family-wise error rate of α = 0.03. Internal injury types were then collapsed into “severe” or “moderate” according to the trauma category indicated in Table I.

Severe and moderate injuries were fitted to the acoustic metrics peak SPL, SEL, and pressure impulse using logistic regression models in r (GLM function) with logit link function and binomial error structure. Covariates included the fish's standard length, weight, and the interaction of standard length and weight. This was done to determine which acoustic metric was most predictive of these injuries and examine the influence of fish size on injury susceptibility.

The five trials used 303 Pacific mackerel for necropsy, including 50 control fish and 253 exposed fish. Fish had a mean standard length of 211.6 mm (SD 16.8 mm) and mean weight of 106.8 g (SD 27.9 g). The total time-in-water for the control fish was 99 min (SD 9 min), compared with 162 min (SD 21 min) for exposed fish.

Control fish showed no statistical differences (α = 0.05) for any injury type between the five trials, demonstrating that pooling of their responses across all days was appropriate (Table II). Injury types that had significantly higher rates in four or more exposure cages as compared to the pooled controls and that had pooled control responses ≤0.10 were considered for further investigation. Analysis (Table II) shows that five injury types met these criteria: kidney rupture, swim bladder rupture, blood pooling in swim bladder, swim bladder bruising, and blood pooling in abdominal cavity.

The five types of injury indicative of exposure to the underwater explosions were collapsed into trauma categories of either severe or moderate based on Table I. The only moderate injury carried forward was swim bladder bruising. The remaining four injury types were collapsed into severe injury.

The photos in Fig. 1 show healthy swim bladder and kidneys in control fish, alongside several of the most common injuries to these organs. The correlation matrix represented in Fig. 2 shows a strong relationship between swim bladder rupture, kidney rupture, and blood in abdominal cavity. Blood pooling in the swim bladder and swim bladder bruising also show strong correlation. Conversely, swim bladder rupture, kidney rupture, and blood in the abdominal cavity show a weak, non-significant relationship to blood pooling in the swim bladder and swim bladder bruising.

Peak pressure, pressure impulse, and sound exposure level responded with an expected monotonic decay with increasing range from the explosive source (Table III). Correspondingly, kidney rupture, swim bladder rupture, and blood pooling in the abdominal cavity decreased monotonically with increasing range (Fig. 3) and decreasing acoustic dosage (Table III). Swim bladder bruising and blood pooling in the swim bladder increased with decreasing range to about 50 m from the explosion, then decreased to the closest ranges tested (i.e., maximum dosages).

Post-exposure times, defined as the time between the detonation and retrieval of the fish cage onto the deck of the research vessel, were from 23 to 77 min (mean = 49 min, SD = 16 min). There was no correlation between the elapsed time to retrieval and the proportion of dead fish in each cage indicating that significantly more fish did not die between the beginning (23 min) and end (77 min) of this timeframe. Twenty-four out of 253 exposed fish were dead upon retrieval (Table III) starting at the closest cage to the explosion and extending out to 157 m. Logistic regression models identified all three acoustic dose metrics (pressure impulse, SEL, and peak SPL) as significant predictors of fish being dead upon retrieval (p < 0.001).

The 233 fish in Table IV includes all exposed fish that were individually tracked from condition upon retrieval through necropsy. Fish that were dead upon retrieval had significantly higher rates of kidney rupture, swim bladder rupture, and blood pooling in the abdominal cavity as compared to fish that lived (Fisher's exact test p < 0.001). Moreover, all dead fish had one or more of the four severe injuries, whereas only 39% of living fish showed any severe injury.

Logistic regression models were fit to groupings of “severe injury” and to “any injury” (i.e., moderate or severe) versus the three measures of acoustic dosage (peak pressure, SEL, and acoustic impulse). All models demonstrated a very significant relationship (p < 0.001) between the proportion of fish that suffered injury versus the acoustic dosage. Other factors considered as covariates included fish standard length (p = 0.08–0.34), fish weight (p = 0.45–0.76), and the interaction of standard length and weight (p = 0.22–0.45); but as indicated by the p-values, none of these covariates were significant. Likewise, the Akaike's information criterion (AIC) values for the simple models, using only injury versus acoustic metric, were two to four points lower than the full model with covariates indicating the simple model was more parsimonious (Table V). For both “severe” and “any” (i.e., moderate and severe) injuries, the simple model produced nearly identical AIC values for fits using peak pressure and SEL, which were lower by 6 to 12 points than models using pressure impulse.

Plots of severe injury and any injury versus range are shown in Fig. 4 along with plots of the acoustic dosage versus range. Responses in most individual exposure cages were above control levels from the closest exposure cage (21 m) to the cage located at 333 m from the explosion for both severe and any injury groupings.

Figure 5 provides a more detailed look at acoustic dosage in terms of the received time series of pressure, grouped into three range intervals. For each interval, two examples (measured on different days) demonstrate both the degree of variation and relative consistency. For example, for each case, the first and second bubble pulse arrives consistently at 275 and 520 ms after the direct arrival (time 0), respectively. For the longest range, an interesting low-amplitude arrival can be seen prior to the direct arrival, suggestive of a head wave arrival that propagates at a higher, sediment sound speed (Dahl and Choi, 2006). The three measures of acoustic dosage peak pressure in dB re 1 μPa, SEL in dB re 1 μPa²·s, and impulse in Pa-s, are identified for each interval based on an average derived from the two examples.

ROV examination of each cage showed that all fish were alive and exhibiting schooling behavior before the explosion. Fish did not exhibit any apparent buoyancy issues such as swimming with vertical component or otherwise struggling to maintain depth.

Additionally, during four of the five trials, the ROV recorded fish in a single cage just prior to, during, and for approximately 10–20 s after the detonation at cages located at 35, 51, 103, and 257 m. Since these videos only captured a single exposure cage per trial, they must be considered as general observations and interpreted with considerable caution. Among the four cages observed, 42 of 47 fish were stunned and sank to the bottom of the cage immediately following the explosion. Only the fish at range 257 m partially recovered and began swimming rapidly in an uncoordinated manner (i.e., not schooling) prior to the end of the video capture. Upon retrieval, all fish were alive in cages located at 257 and 103 m, indicating that they recovered from the stunning. However, 25% and 58% of fish located at 51 and 35 m, respectively, were dead.

The current study showed that, as expected, injuries and near-term mortality (i.e., within 23–77 min) increased with increasing acoustic dosage associated with underwater explosions. Exposure to underwater explosions at the ranges tested did not appear to affect external tissues such as fins, scales, or eyes.

Consistent with past findings (Gaspin, 1975; Yelverton et al., 1975; Gaspin et al., 1976; Goertner et al., 1994; Dahl et al., 2020), swim bladder injuries were the most prevalent and included three of the five injuries that were clearly related to exposure to underwater explosions. Moreover, organs directly adjacent to the swim bladder, in this case the kidney, were also affected. The kidney sits immediately dorsal to the swim bladder in the Pacific mackerel as it does in many teleost species. Thus, the proximity to the rapidly expanding and contracting swim bladder is the likely cause of kidney rupture. Similar observations and conclusions were reached both for Pacific sardines (Dahl et al., 2020) and in animals exposed to impulsive pile driving (Halvorsen et al., 2011, 2012b; Halvorsen et al., 2012a).

The correlation matrix (Fig. 1) shows that five injuries can be placed into two groups. The first group consists of swim bladder bruising and blood pooling in the swim bladder, while the second group consists of kidney rupture, swim bladder rupture, and blood pooling in the abdominal cavity. Severe levels of swim bladder rupture likely obscured observations of swim bladder bruising and blood pooling in the swim bladder and released any blood contained in the swim bladder into the abdominal cavity. Based on this and the individual injury plots in Fig. 2, progression of observed injuries with increasing acoustic dose likely begins with swim bladder bruising followed by hemorrhaging of tissues that release blood into the swim bladder. The second group of injuries follow, with higher acoustic doses leading to greater swim bladder excursions, which damages the adjacent kidney and ruptures the swim bladder, releasing blood into the abdominal cavity.

Severe and any (i.e., moderate and severe) injury groupings both began exceeding control levels in cages located at ≤333 m to the explosion. This confluence is likely due to swim bladder bruising being the only moderate injury observed, so that the any injury grouping was dominated by the four severe injuries. Control response for severe and any injuries groupings were 0.14 and 0.18, respectively, which is primarily due to blood pooling in the abdominal cavity which is responsible for 0.10 of these results. Blood pooling in the abdominal cavity was not observed in reference fish taken directly from collection, euthanized, and dissected; therefore, it is likely that handling of fish during the trials was responsible for blood pooling in the body cavity in controls and in at least some of the exposed fish. This injury was not noted in our study of Pacific sardines (Dahl et al., 2020), although handling procedures were nearly identical.

Pressure impulse, sound exposure level, and peak pressure were all good predictors of the five injury types, with injuries generally increasing with increasing acoustic dosage (Table V; Figs. 3 and 4). Logistic regression models indicated that peak SPL and SEL were better predictors of injury than acoustic impulse and that the size of the fish was not a factor. Although, the overall size range of the fish in the study may not have varied enough to be able to detect a difference in susceptibility with differing weight or length.

Dahl et al. (2020) showed an increase in injury rate (for kidney and swim bladder rupture) corresponding to ranges between about 125 and 150 m for Pacific sardines, which was postulated to be associated with waveguide propagation effects. A similar increase within this range interval is not apparent for Pacific mackerel. Sound speed profiles show better mixing of near-surface water and subsequent disruption of the thermocline in October 2018 during the Pacific sardine study as opposed to the current study. This has the effect of changing the multipath arrival pattern so that in 2019 the surface reflecting path traverses a higher sound speed than a bottom reflecting path, with the effect of lessening the rapid drop to negative pressure observed at closer ranges in 2018 and thereby reducing the potential injurious effect of this phenomena.

The ROV video footage captured during the explosion of fish in four exposure cages (35, 51, 103, and 257 m), although limited in sample size, provides a general sense of what happens to fish immediately following exposure to an underwater explosion. Past studies could only speculate on the type and severity of such immediate responses [e.g., Fitch and Young (1948) and Govoni et al. (2003)]. The videos show that most fish out to a range of 257 m in the four observed cages were immediately stunned and sank following exposure. Although all fish in the ROV-observed cages at 103 and 257 m were retrieved alive after 38 and 45 min, respectively, it is likely these fish would have experienced increased risk of predation within the time frame that stunning occurred.

The number of fish that were dead upon post-exposure retrieval was low overall but did decrease with decreasing acoustic dose as would be expected. Even at the closest range corresponding to the highest acoustic dose, only 33% of fish were dead at 26 min post exposure. Fish that were dead upon retrieval showed significantly more severe injuries than fish that were alive. This suggests that these severe injuries, especially kidney rupture and swim bladder rupture, caused mortality over a fairly short period. Based on the number of severe injuries overall, it is likely that more individuals, especially those with kidney and swim bladder ruptures, would have succumbed to their injuries with increased time post-exposure due to blood loss and the inability to perform vital physiological functions such as osmoregulation (kidney) or buoyancy regulation (swim bladder).

This study, along with its companion investigation of effects on the ear (Smith, 2022) and our earlier work on Pacific sardines (Dahl et al., 2020), provides significant insights into the potential effects to fish from exposure to explosions at different distances from the source. However, the two species studied only give a broad indication of potential effects from exposure to explosions, and the results should only be extrapolated to other species with caution. This is particularly the case since there is so much variation among the more than 34 000 extant fish species in terms of morphology and physiology. Thus, additional studies are needed to better understand the susceptibility of a wider array of marine fishes to underwater explosions.

As seen in this and past research, acoustic dosage metrics such as peak SPL and SEL are good predictors of physiological effects in fish exposed to impulsive signals (Gaspin, 1975; Yelverton et al., 1975; Gaspin et al., 1976; Wiley et al., 1981; Govoni et al., 2003). In our studies, exposures were conducted in a single area across a similar suite of environmental conditions with both the fish and explosive charges set mid-water. This design purposely limited these factors to make results easier to interpret; however, there was little variability across trials in the pattern and timing of multipath arrivals at any given distance (i.e., fish cage) from the explosion.

In our 2018 study with Pacific sardines (Dahl et al., 2020) and in past studies (Yelverton et al., 1975; Wiley et al., 1981), potential waveguide effects were observed. Since the pattern of sound arrivals will change with explosive charge depth, fish depth, water depth, sound velocity profile, and many other environmental factors, it is important that future studies consider these variables to gain a better understanding of effects of underwater explosions on fishes.

Finally, a critical issue is whether there are long-term fitness consequences for fishes that manifest themselves over time due to very small effects that we were unable to observe. Indeed, fitness post-exposure may be the most important question, and one that can be most directly answered with longer-term survival studies. Similarly, longer-term survival studies would help to determine mortality directly. One of the challenges with the current research effort is discerning which injuries may be mortal and which are recoverable over periods lasting for more than approximately an hour. Correlating injuries seen at certain acoustic dosages with survival at similar dosages may further reveal the lethality of injuries observed during necropsies.

This project could not have been completed without the aid of Alyssa Accomando, Dana Schrimpf, Jennie Shield, Kevin Carlin, Maria Zapetis, Michelle Tishler, and Cameron Martin, research vessel captain Brad Davidson, David Dall'Osto (APL-UW), NMSRL acousticians and field techs Matthew Babina and Matthew Daley, and the overwhelming support of several other NIWC Pacific teams: facilities (Depot), veterinary technicians, and the ROV team. The authors thank Explosive Ordnance Disposal Mobile Unit Three Detachment Southwest, the U.S. Third Fleet, and the U.S. Pacific Fleet for enabling a safe and successful field effort and our sponsors, the Living Marine Resources (LMR) program of the U.S. Navy, for making this study possible. The investigators also appreciate valuable discussions and guidance from Dr. Rich Townsend and Dr. Rebecca Buchanan of the University of Washington relating to statistical analysis. The views expressed in this publication reflect the results of research conducted by the author(s) and do not necessarily reflect the official policy or position of the Department of the Navy, Department of Defense, or the U.S. Government. Some of the authors are employees of the U.S. Government. This work was prepared as part of their official duties. Title 17 U.S.C. §105 provides that “Copyright protection under this title is not available for any work of the United States Government.” Title 17 U.S.C. §101 defines a U.S. Government work as a work prepared by an employee of the U.S. Government as part of that person's official duties.