Pile-driving and other impulsive sound sources have the potential to injure or kill fishes. One mechanism that produces injuries is the rapid motion of the walls of the swim bladder as it repeatedly contacts nearby tissues. To further understand the involvement of the swim bladder in tissue damage, a specially designed wave tube was used to expose three species to pile-driving sounds. Species included lake sturgeon (Acipenser fulvescens)—with an open (physostomous) swim bladder, Nile tilapia (Oreochromis niloticus)—with a closed (physoclistous) swim bladder and the hogchoker (Trinectes maculatus)—a flatfish without a swim bladder. There were no visible injuries in any of the exposed hogchokers, whereas a variety of injuries were observed in the lake sturgeon and Nile tilapia. At the loudest cumulative and single-strike sound exposure levels (SELcum and SELss respectively), the Nile tilapia had the highest total injuries and the most severe injuries per fish. As exposure levels decreased, the number and severity of injuries were more similar between the two species. These results suggest that the presence and type of swim bladder correlated with injury at higher sound levels, while the extent of injury at lower sound levels was similar for both kinds of swim bladders.
Anthropogenic noise has been established as a potential concern related to the health and survival of worldwide fish stocks . Among the types of potentially dangerous noise sources that could result in injury are shipping, sonar, seismic surveying and construction sounds. However, there are very few quantitative data on the physiological effects of these sounds on fishes [1–3].
Recent studies [4–7] have evaluated the effects of impulsive pile driving on fishes using methods that enable investigators to bring high-intensity sound sources into the laboratory for evaluation of effects. Halvorsen et al. [6,7] provided a scientific recommendation for the onset of injury threshold in juvenile Chinook salmon (Oncorhynchus tshawytscha, Salmonidae) at a cumulative sound exposure level (SELcum) of 210 dB re 1 μPa2·s derived from 960 pile strikes, each having a single-strike SEL (SELss) of 180 dB re 1 µPa2·s. These results were further supported in an injury recovery study on the same species . Moreover, Bolle et al.  found no difference in mortality between control and exposed common sole (Solea solea, Solidae) larvae (including stage 3–4a larvae which had inflated swim bladders) at an SELcum of 206 dB re 1 µPa2·s over 100 pile strikes and a SELss of 186 dB re 1 µPa2·s. While these initial studies provided data for understanding effects of pile driving on these two species of fishes, there is still too little information to be able to make generalized predictions and/or recommendations regarding what injuries to expect from other sizes of these species as well as anatomically and physiologically different species of fishes exposed to pile driving.
One key variable that could potentially alter the types and frequency of occurrence of injuries is the presence or absence of a swim bladder as well as the type of swim bladder . There are two general types of swim bladders in fishes: physoclistous and physostomous. The physostomous swim bladder (found in salmonids, sturgeons and many of the more evolutionarily ancestral fishes) is connected to the gut via a pneumatic duct, thus allowing the fish to gulp air from the water surface or expel air to quickly adjust the volume of air within the swim bladder. The physoclistous swim bladder (found in many of the more recently evolved fish species, including bass, perch and rockfish) has a gas gland that provides gas exchange by diffusion between the swim bladder and blood. There are also many species of fishes such as flatfishes, gobies and elasmobranchs that do not possess a swim bladder.
When a fish with a swim bladder is exposed to low-frequency impulsive energies, including pile driving, the swim bladder acts like an air bubble that vibrates with sufficient magnitude to cause damage to tissues and organs within close proximity as well as to the swim bladder itself [5–7]. Therefore, it would seem likely that fishes with physostomous swim bladders could potentially expel air, thereby diminishing the tension on the swim bladder and decreasing damaging effects during exposure to impulsive sounds. By contrast, fishes with a physoclistous swim bladder are incapable of decreasing the volume of gases fast enough to avoid sustaining more severe injuries. Perhaps the least likely to yield damage would be fishes without a swim bladder.
This study presents a comparative analysis of responses to pile-driving stimuli in the lake sturgeon (Acipenser fulvescens, Acipenseridae), a species with a physostomous swim bladder; the Nile tilapia (Oreochromis niloticus, Cichlidae), a species with a physoclistous swim bladder; and the hogchoker (Trinectes maculates, Achiridae), a flatfish without a swim bladder. While none of these species are currently known to be receiving high levels of exposure by anthropogenic sound sources, other species of sturgeon are listed under the Endangered Species Act in the United States and therefore are of particular concern with regard to effects of sounds from pile-driving and other intense sounds, making the lake sturgeon a valuable proxy for understanding sound exposure.
Fishes in this study were exposed to several different SELss and SELcum levels using the same high-intensity-controlled impedance fluid-filled wave tube (HICI-FT) used by Halvorsen et al. [6,7] and Casper et al.  to assess the frequency and types of injuries that occurred. The results not only provide some insights into the potential impacts of how physostomous versus physoclistous swim bladders may impose damage on body tissues, but it also extends earlier work to provide a more generalized understanding of impulsive sound levels that may help determine the onset of tissue damage in fishes.
2. Material and methods
(a) Fish species information
Nile tilapia (84.2 ± 9.6 mm standard length (SL) and 18.8 ± 8.1 g; approx. 6 months old) were obtained in April 2011 from a breeding colony in the laboratory of Dr Thomas Kocher of the Department of Biology at the University of Maryland. These fish were acclimated for a minimum of 2 weeks following transportation between laboratories before being used in experiments. The fish had their caudal fins clipped for identification during experiments.
Hogchokers (61.5 ± 18.9 mm SL and 11.6 ± 11.9 g; ages ranged from juveniles to adults) were collected in July 2011 using a bottom trawl in the Patuxent River, Calvert County, MD, USA and held at the Solomon's Island research laboratory until transportation to the University of Maryland. Owing to difficulties in finding a suitable food source for these fish in the laboratory, acclimation time following arrival was only 3 days. Any fish that had external injuries as a result of being caught in the trawl were removed before experimentation. Individual hogchokers were easily identified through natural markings and so did not require a tail clip.
Lake sturgeon (65.7 ± 5.2 mm SL and 1.6 ± 0.4 g; 3 to 4 months old) were obtained from the Wisconsin Department of Natural Resources Wild Rose Inland Hatchery (Wild Rose, WI, USA) in August 2011. Following shipping from the hatchery, these fish were acclimated for a minimum of 2 weeks before being used in experiments. Owing to the small size of the sturgeon fins, fin clipping was not used for identification.
In all cases, fish were maintained on a 14 L : 10 D cycle in 235 gallon round tanks. Fish that were scheduled to be used in experiments were not fed during the experimental week so that the digestive system would be void of food during sound exposure, because initial work with Chinook salmon showed that increased numbers of injuries were found in fish with food in their guts versus those without (M. Halvorsen 2011, personal observation).
(b) Pile-driving exposure equipment and signal presentation
Pile-driving exposure was conducted using the HICI-FT. This device has a cylindrical holding chamber that is 45 cm long with a 25 cm internal diameter and 3.81 cm-thick stainless steel walls. Large shakers on either end of the chamber were used to create sounds that accurately reproduced the acoustic characteristics and sound levels of pile-driving sounds under far-field plane wave acoustic conditions. For a detailed description of the equipment and its development, see Halvorsen et al. [6,7].
Signal generation and data acquisition for the HICI-FT are described in detail in Halvorsen et al. [6,7]. The pile-driving sounds used in this study were field recordings at a depth of 5 m taken at a range of 10 m from a 76.2 cm steel shell pile (outer diameter) driven using a diesel hammer at the Eagle Harbor Maintenance Facility . Eight different recordings of pile-driving strikes were normalized to the same SEL. Twelve repetitions of each of the eight strikes generated a file of 96 strikes that were randomized each day using MATLAB (The MathWorks, Inc., Natick, MA, USA). That file was repeated 10 times for a 960-strike presentation.
(c) Fish exposure
Four fishes were allowed to swim freely in an acrylic chamber mounted around the opening of the HICI-FT exposure chamber for a 20 min acclimation period. The fishes were then allowed into the exposure chamber and the upper shaker/lid was sealed over the chamber opening. The acrylic chamber was drained, and the HICI-FT rotated from the vertical position to the horizontal position for each exposure or control treatment.
Buoyancy was documented in all fishes as done in previous studies [6,7]. However, the hogchokers always sat on the bottom, the Nile tilapia always displayed neutral buoyancy and the lake sturgeon always swam along the walls but were never observed to gulp air (or expel air) as was the case with the physostomous Chinook salmon [6,7].
In total for these experiments, 125 Nile tilapia were exposed plus 32 controls, 57 hogchokers were exposed plus 10 controls and 141 lake sturgeon were exposed plus 32 controls. Control fishes were subject to the identical process as exposed fishes but without the pile-driving sound. Exposure sound levels for each species began with a SELcum of 216 dB re 1 µPa2·s, derived from 960 pile strikes and 186 dB re 1 µPa2·s SELss (treatment 1). The SELcum for subsequent trials was decreased in 3 dB steps for each treatment, as summarized in table 1. From here forward, the study will refer to treatments 1–5 to simplify the reference to the exposure paradigms.
Following each treatment, fishes were euthanized in a buffered MS-222 solution, necropsied and examined for external and internal signs of barotrauma (e.g. damage to eyes, fins, gills), using methodology from previous Chinook salmon studies [6,7,10]. Each potential injury was noted as present or not (for a detailed list of all potential external and internal barotrauma injuries, see Halvorsen et al. ).
(d) Evaluation of barotrauma injuries
Each observed injury was assessed and then assigned to weighted trauma categories [6,7]: Mortal, Moderate or Mild. The Mortal trauma category, weighted 5, included injuries that were severe enough to lead to death. The Moderate trauma category, weighted 3, included injuries likely to have an adverse impact on fish health but might not lead directly to mortality. Finally, the Mild trauma category, weighted 1, referred to injuries of minimal to no adverse physiological effects to fishes. The weight assignments applied to each of the three trauma categories were based on the assessment of physiological significance that considered the influence of multiple injuries and inspection of data for the occurrence of injury combinations. The response-weighted index (RWI) is the sum of the presence of each injury multiplied by the trauma weight assigned to each injury type (see the electronic supplementary material). The formula is: 2.1
(e) Statistical analysis
A one-way ANOVA was used to evaluate injuries within fish species with two-way ANOVA used for evaluating injuries between fish species. A post hoc Tukey test on the multiple comparisons (SigmaPlot 11, SYSTAT Software, Inc.) was used to evaluate any differences between species and treatments in terms of both RWI values and number of injuries observed. All statistical information is displayed in tables 2–5.
None of the fishes died during the pile-driving exposure within the HICI-FT. Also, at the loudest exposure paradigm, treatment 1, there were no external or internal injuries observed in any of the hogchokers (figure 1); thus, this species was not tested at any of the lower treatment levels. In the Nile tilapia (figure 2), a variety of injuries were observed, ranging from Mortal injuries such as ruptured swim bladders and renal haemorrhages to Moderate haematomas. In the lake sturgeon (figure 3), injuries ranged between Moderate haematomas and Mild partially deflated swim bladders. There were no external injuries in any of the species tested.
Four different injuries were observed in the lake sturgeon (figure 4), including haematomas on the swim bladder, kidney and intestine (all Moderate injuries), as well as partially deflated swim bladders (Mild injury). Injuries were most frequently observed in treatments 1 and 2 (figure 5). The level of injury (2.4 in treatment 1 and 2.6 in treatment 2) did not differ significantly from each other (table 2). The frequency of injuries then dropped to a statistically significant amount between treatments 2 and 3 (mean of 1.1 injuries; table 2).
The RWI values in lake sturgeon also followed a similar pattern (figure 6). Even though there was a slight increase in the average RWI values at treatment 2 compared with treatment 1 (17.7 and 18.9, respectively), there was no significant difference between those RWI values (table 3). Following a statistically significant drop in RWI value between treatments 2 and 3 (average RWI value of 5.3; table 3), and between treatments 3 and 4 (average RWI value of 2.9; table 3) there was, however, no significant difference between treatments 4 and 5 (average RWI value of 0.4; table 3).
Eight different injuries were observed in the Nile tilapia, including ruptured swim bladders, haemorrhaging of the kidney and intestines, and haematomas on the swim bladder, liver, intestines, gonads and adipose tissue (figure 4). The highest average number of injuries per fish was observed in treatment 1, with a significantly decreasing trend as sound levels were lowered (table 2 and figure 5). The RWI values observed at each treatment level also showed a significant decrease from treatment 1 through to treatment 4 but not between treatments 4 and 5 (table 3 and figure 6).
Comparisons were made between RWI values as well as between the average number of injuries per fish at each of the treatment levels between the Nile tilapia, lake sturgeon and data previously collected with the Chinook salmon [6,7] (figure 6). It should be noted that a spleen haemorrhage was found in one lake sturgeon but was not included in this analysis as this injury was not included in the previous Chinook salmon study. The low-frequency of occurrence of this injury (0.007% of all exposed fishes) would have a negligible effect only on the RWI values in lake sturgeon for that treatment.
For treatment 1, both the number of injuries per fish (table 4) and RWI values (table 5) were significantly different between the lake sturgeon, Nile tilapia and Chinook salmon. In treatment 2, the number of injuries and the RWI values were significantly different between salmon and tilapia and between salmon and sturgeon in that there were more injuries and higher RWI values for tilapia and sturgeon, respectively (table 5). For treatment 3, both variables were significantly different among all of the species except Chinook salmon and lake sturgeon (tables 4 and 5). In treatments 4 and 5, there was no significant difference in values for either variable among any of the fishes (tables 4 and 5).
The results of this study show that each species yielded different injury responses to the pile-driving stimuli, especially at the highest sound levels (figure 6). However, as SELs decreased, the responses between fishes became similar. The variability in results at the higher levels could potentially be a function of the presence or absence of a swim bladder and the difference in physostomous versus physoclistous swim bladder.
Variability is most evident when comparing the other tested species with the hogchoker. As a representative species without a swim bladder, there was no visual evidence of any external or internal injuries in hogchokers at treatment 1 (figure 1). It is possible that there could have been a delay in the onset of injuries, which could have appeared in the days following exposure, but a recent study  did not find evidence to support onset of new injuries in the days post-exposure in juvenile Chinook salmon.
However, those post-exposure analyses were not conducted in hogchokers owing to the availability of only a limited number of animals. Many species of flatfishes are considered commercially important, and the physiology among flounder and sole are fairly similar, which suggests that these fishes would be able to survive pile-driving exposures up to 960 strikes at levels equivalent to treatment 1 without sustaining damage. Follow-up studies could focus on other fishes without swim bladders to investigate the consistency of injuries in other fish families. Specifically, testing elasmobranch fishes could be valuable because their physiology is unique among other fishes without swim bladders, and many of these species are under pressure from other anthropogenic causes .
The Nile tilapia yielded the highest average RWI value at treatment 1 of all of the tested species (figure 6) with several Moderate injuries, including 100 per cent occurrence of swim bladder and gonadal haematomas as well as high levels of hepatic and intestinal haematomas. There was also a 38 per cent occurrence of Mortal renal haemorrhaging and one ruptured swim bladder. This was not surprising as the physoclistous swim bladder of this fish is incapable of quickly modulating its mass of gas, thereby making it much more susceptible to damage from overexpansion. All of the injuries observed from treatment 1 were also present in treatment 2 (figure 4), though the frequency of occurrence decreased, resulting in a large decline in the average RWI value. Haematomas of the swim bladder and gonads remained through treatment 2 and 3 and even remained, albeit at a lower frequency, throughout treatment 5, suggesting that these organs were the most sensitive to barotrauma. Any damage, even haematomas, to the reproductive organs could be significant at a population level if it resulted in a long-term decrease in the reproductive output of these fishes, though this was not measured within this study.
When compared with the Chinook salmon, Nile tilapia had higher average RWI levels for treatments 1–3. For the Nile tilapia in treatment 3, the RWI values and number of injuries were statistically higher than Chinook salmon from the Halvorsen et al. [6,7] study, which suggests that Nile tilapia are more sensitive to pile driving than the previously tested juvenile Chinook salmon. This could be an important finding for physoclistous fishes; yet there are many physiologically diverse physoclistous species, thus further testing would be necessary to make any generalizations.
The lake sturgeon RWI values were higher than the Chinook salmon [6,7] and lower than the Nile tilapia when exposed to treatment 1 (figure 6). Like the Chinook salmon and other salmonids, sturgeon have a physostomous swim bladder that, whereas sensitive to damage from pile driving, does not appear to be as damaging to surrounding tissues as the physoclistous swim bladder in the Nile tilapia. The most common injuries observed were haematomas of the swim bladder, liver and intestine, all of which could result from overexpansion of the swim bladder.
For the lake sturgeon, it is important to note that after treatment 2, the RWI values were not significantly different from Chinook salmon, suggesting that the overall responses between the two fishes are biologically similar. Treatment 3 used equivalent sound level metrics as those in Halvorsen et al. [6,7], and those sound levels helped define the threshold of injury onset in Chinook salmon (the threshold being an RWI value of ≥2).
It is interesting to note the diversity of injuries between the Nile tilapia, lake sturgeon and Chinook salmon. While there was a small overlap between the three species (swim bladder and intestinal haematoma), there were many more types of injuries (14) observed in the Chinook salmon [6,7] and Nile tilapia (10) than in the lake sturgeon (5) at equivalent treatment levels and number of pile strikes. There are many potential reasons for this diversity, with the most likely being anatomical and physiological differences between species. Looking at the morphology of swim bladders between the three species, the Chinook salmon and Nile tilapia have swim bladders that extend the length of the internal cavity, thereby taking up a large portion of the abdomen of the fish while the lake sturgeon swim bladder is much smaller and located more anteriorly. The swim bladder location within the lake sturgeon is not in as close proximity to as many internal organs as it is in the other species, thus decreasing the potential to cause damage. Future studies should make certain to consider not only the type of swim bladder (physostomous/physoclistous) but also the size and location within the abdomen.
A possible explanation to the diversity of observed injuries is the size and general body shape of the different species. The lake sturgeon were significantly smaller in standard length as well as weight compared with the other species. Of the lake sturgeon, 92 per cent were less than 2 g (none of the Chinook salmon or Nile tilapia were less than 6 g), a weight that has previously been hypothesized as having the potential to be more susceptible to injury than fishes larger than 2 g , based on a previous study looking at underwater explosive damage in different sizes of fishes . There is no evidence to support or refute this claim within the confines of this study as so few lake sturgeon larger than 2 g were tested. The differences in body shape between these species could also be an important factor; however, this variable was not specifically addressed in this study but would be a valuable exercise in a future experiment that could incorporate more examples of these and other body shapes.
Another important aspect of this study is that it supplies data that can be considered for use to inform decision-makers considering regulatory criteria for exposure of fishes to pile-driving sounds. The current guidelines developed on the United States West Coast for maximum sound exposure levels to which fishes can be exposed during pile driving before the onset of injury are a SELcum of 187 dB re 1 µPa2·s for fishes above 2 g and a SELcum of 183 dB re 1 µPa2·s for fishes below 2 g [14,15]. As four fish species that represent the range of swim bladder physiology and without swim bladder have now been exposed in the HICI-FT, results support potentially increasing current criteria levels to207 dB re 1 µPa2·s SELcum derived from 177 dB re 1 µPa2·s SELss and 960 impulsive signals, as suggested by Halvorsen et al. [6,7]. In other words, the data from the HICI-FT studies support an argument that fishes appear to be less susceptible to energy from impulsive pile driving than is currently allowed before the onset of physiologically significant injuries.
This study provides a new understanding of the importance of the swim bladder in fishes exposed to impulsive pile driving. The results from this study can lead scientists and regulators to understand and potentially predict the kinds of injuries that might appear in fishes when there is a general understanding of the fish's specific anatomy/physiology. With further testing, these predictions will become more refined for setting regulations with pile-driving projects.
Experiments were conducted under supervision and approval of the Institutional Animal Care and Use Committee of the University of Maryland (protocol no. R-09-23).
The work reported here was supported by the Bureau of Ocean Energy Management (BOEM), with additional support from the California Transportation Authority (CALTRANS). Initial studies and development of the HICI-FT were partially supported by a contract with the National Cooperative Highway Research Programme (NCHRP). We thank Elizabeth Burkhard and her colleagues of BOEM for their intellectual support and valuable ideas as we developed this project and to Dave Buehler from ICF Jones and Stokes for his help in obtaining Caltrans support. We also thank Steve Fajfer and the Wisconsin Department of Natural Resources Wild Rose Inland Hatchery for providing the lake sturgeon, Thomas Kocher and Jennifer Ser for providing the Nile tilapia, and Thomas Miller, Edward Houde and David Lowensteiner for the hogchoker. The authors acknowledge that there are no conflicts of interest.
↵† Co-first authors for this publication.
- Received July 6, 2012.
- Accepted September 17, 2012.
- This journal is © 2012 The Royal Society