There is a global investment in offshore wind (OSW) infrastructure as many countries increasingly prioritize renewable energies over fossil fuels (Gielen et al., 2019). The increased human presence in the ocean poses challenges to marine life since the pile driving noise emitted during OSW construction has been shown to cause physical damage (Halvorsen et al., 2012), sensory harm (Kastelein et al., 2016), and behavioral changes (Jones et al., 2020) to a myriad of marine taxa. Consequently, anthropogenic noise is recognized as a global pollutant of paramount concern (Halfwerk et al., 2011; Kunc et al., 2014; Duarte et al., 2021). Noise-induced behavioral changes can have direct fitness consequences, and the spatial extent is likely greater than that of noise-induced physical and physiological harm (Popper et al., 2022). However, movement responses are rarely quantified. Fine behavioral changes are difficult to measure in marine environments where animals are largely in accessible, leading to key knowledge gaps on the effects of noise on behaviors that can influence individual fitness.

Much of the existing research on noise-induced behavioral changes has focused upon large marine mammals, and to some extent fishes (Miller et al., 2000; Southall et al., 2007; Miller et al., 2012; Popper and Hawkins, 2019). There is scant data on marine invertebrates such as cephalopods. This is a surprising fact considering their central position in many ocean food webs (Clarke, 1996) and their high commercial value exceeding $1 billion USD per year worldwide (Hunsicker et al., 2010). Cephalopods have been shown to detect sounds within the same frequency range (<500 Hz) as pile driving noise, indicating a likely susceptibility to adverse effects of noise (Mooney et al., 2010; Mooney et al., 2020). Indeed, recent laboratory studies showed that solitary longfin squid (Doryteuthis pealeii), an important U.S. fishery taxon, exhibit alarm responses to pile driving playbacks (Jones et al., 2020; Jones et al., 2021). However these studies used solitary squid in tanks, which makes behavioral inferencing challenging since D. pealeii is an aggregating species and the acoustic field differed from field conditions (Birkett and Newton-Fisher, 2011; Jones et al., 2019). One field study examined caged squid (Sepioteuthis australis) behavioral responses to seismic air-gun surveys (Fewtrell and McCauley, 2012). The authors found that both the proportion of alarm responses (e.g., escape jetting) and swimming speed were positively correlated with received noise levels. Nonetheless, this preliminary study only assessed movement qualitatively, leading to important questions regarding the ecological consequences, energetics, and duration of the observed behavioral changes.

Most bioacoustic studies have not measured the duration of noise-induced behavioral changes (but see Miller et al., 2012) despite being a key consideration for policy makers (Finneran et al., 2017; Southall et al., 2021). Measuring the duration of noised-induced behavioral impacts is critical because it is inherently linked to impact severity and persistence of effect. For example, the energetic cost incurred from a transient increase in acceleration is less severe than a prolonged heightened acceleration state if an individual does not habituate or desensitize to a noise stimulus (Southall et al., 2007). The few studies measuring disturbance durations in aquatic animals have been restricted to large vertebrates capable of carrying motion sensor tags (Miller et al., 2012). For many marine species, quantifying individual movement is difficult, particularly over time scales comparable to pile driving operations; yet such data are needed to quantify behavioral changes and energetic costs. As a result, most studies on smaller and more abundant animals are conducted in tanks, providing key data but limiting the knowledge that can be applicable to actual noise exposures in field settings. New tools and methods are thus needed to accurately describe and quantify noise-induced behavioral changes, especially in more real-world conditions (Popper et al., 2022).

To date, there has been no field study quantifying the movement behavior of cephalopods, or any invertebrate, during real-time pile driving construction. Given that construction is imminent and considering the spatial overlap of cephalopod fisheries and planned OSW development ( Figure 1 ), there is an urgent need to experimentally examine whether commercially-important cephalopods alter movement behaviors during pile-driving noise exposure, and if so, quantify how long those changes persist. In this context, our present aim was to quantitively examine the fine-scale swimming movements and kinematics of D. pealeii during field-based pile driving activities to assess potential ecological and energetic consequences of noise exposure. We utilized high-resolution movement sensors to measure individual-level swimming kinematics at sub-second to hourly temporal resolutions and at multiple spatial scales during the two main types of piling installation: continuous vibratory and impulsive impact hammering. Both installation methods are known to produce intense sounds, but the characteristics are vastly different (Amaral et al., 2020; Jézéquel et al., 2022). We then assessed the probability of squid changing their movement behavior associated with specific received noise levels, characterized the observed behavioral changes, and measured the durations of those alarm behaviors. These anthropogenically-induced alarm responses were then compared to natural swimming movements and gait disruptions observed throughout the course of quiet, control days to evaluate the potential biological and energetic implications of the noise-induced stress. To address these questions, we developed a new approach to quantify the movement of cephalopods that can be used to address similar questions for other species more broadly.

We present the first study quantifying the fine-scale movement behaviors of a marine invertebrate in response to an actual field-based anthropogenic noise source. We used high-resolution movement sensors to quantitively measure changes in swimming kinematics and measure how long gait disruptions persisted. Our results demonstrate that while field-conducted pile driving noise elicited clear alarm responses at high received levels, these were short-term evasions that persisted for only 4 s on average. Further, these escape behaviors were found only at a site of relatively high received sound levels, although the measured noise levels corresponded to roughly 1 km from actual windfarm construction pile driving (Sigray et al., 2022). Interestingly, alarm behaviors were shorter in duration than similar high acceleration movements during natural, intraspecific agonistic encounters observed during quiet control periods indicating that the animals quickly returned to sensory vigilance. Additionally, when considering overall jetting and finning gait behaviors throughout an exposure or control day, there was no detectable impact of pile driving noise on swimming behavior. Although, the experimental cage may have constrained certain swimming behaviors, particularly horizontal dispersion from the sound source.

This study used novel accelerometer-based particle acceleration measurements at multiple distances to create an acoustic propagation model and identify probabilities of movement behavior changes at specific received noise levels. Nine of 13 D. pealeii at the near site elicited at least one or more alarm movements in response to the IH between 122.96 and 112.32 PAL_zpk dB re 1 µm s^-2, which are noise levels greater than 880 m from a one OSW construction site (Sigray et al., 2022). We know of no other sites in which we there are comparable, published, particle acceleration data. This suggests that behavioral disruption will likely occur at the kilometer scale and at a relatively substantial range, especially if we consider wind turbine pile spaces to be roughly 1 km apart and noise levels to stay consistent. More intense or persistent responses may occur within that 880 m range especially if larger pilings are used or if multiple platforms are constructed concurrently. Hence, the alarm responses described here may impact a significant majority of animals within the entire OSW development area, leading to potential regional impacts on squid populations. However, more information on noise-induced disruptions to group-level behaviors is needed to better assess impacts on populations.

Although there were clear alarm behaviors in response to pile driving noise, we found no significant difference in the number of kinematic disturbances measured from the ITAG between control and noise exposure periods ( Figure 6 ). To be more representative of wild conditions, we used squid groups of mixed sexes in our experiments. D. pealeii are still reproductively active into September when our experiments took place (Stevenson, 1934), and squid are known to swim dynamically in breeding aggregations, and these movements were likely classified as kinematic disturbances in the present study (Shashar and Hanlon, 2013). This result provides more evidence that pile driving did not change long term swimming behaviors and it demonstrates the importance of considering the biology and group-level behaviors when quantifying noise-induced behavioral impacts. Future studies should avoid studying aggregating species in isolation because it may constrain individual behavior and limit interpretations.

Most alarm behaviors were associated with one or multiple rapid jet propulsion events; these jets resulted in elevated ODBA and a change in finning rate ( Figure 4 ). An increase in ODBA and a transition to primarily jet propulsion indicates a higher energetic cost (Webber and O’Dor, 1986, Halsey et al., 2009). Squid are thought to operate at or near their metabolic limit (O’Dor and Webber, 1991), which suggests that an anthropogenically-induced high energy alarm behaviors may be detrimental to squid energy budgets. However, because the disruptions were transient and only elicited a maximum of three times per individual over 3-4 hours of pile driving, we suspect the impact to be non-substantial, especially considering squid frequently elicited similar dynamic kinematics during non-noise exposure periods. Additionally, free-ranging muscular squid naturally display high acceleration jet propulsion at rates, > 9 jets min^-1 (Cones et al., 2022). Thus, the additional 0-3 jetting propulsion alarm responses over multiple hours of noise exposure are not likely detrimental to energetic expenditure.

No squid at the far site (with lower received levels) elicited alarm behaviors in response to either IH or VH pile driving noise despite noise levels occurring within D. pealeii sound detection abilities (Mooney et al., 2010). This result suggests there was either a dose-dependent response or there exists a minimum threshold that induces alarm behaviors, where animals detecting amplitudes 112-123 and 96 dB re 1 µm s^-2 have a 69% and <1% probability of eliciting at least one alarm response, respectively. In fact, dose dependence behavioral responses were found in S. australis exposed to air gun noise (Fewtrell and McCauley, 2012). Squid elicited a higher proportion of alarm behaviors with increasing noise levels, implying the severity of noise impact on squid is related to the distance from the noise source.

Interestingly, 16 of the 17 alarm behaviors were observed during IH (7 alarm behaviors at the first hammer strike) pile driving, with only one instance of reaction to the onset of VH pile driving. This finding suggests that high amplitude and transient signals are more detrimental to squid swimming kinematics compared to low amplitude and continuous signals. Previous noise studies have largely focused upon IH noise impacts on marine life (Herbert-Read et al., 2017; Jones et al., 2020; van der Knaap et al., 2022), while fewer have directly compared noise impact with temporal variation (Neo et al., 2014; Shafiei Sabet et al., 2015). These studies also demonstrated that intermittent noises, rather than continuous, induced more severe behavioral changes including more alarm behaviors. Further research should seek impact severity comparisons between IH and VH techniques for a broader range of species. Considering some OSW farms have been successfully installed with only the VH, it could serve as an important mitigation technique in areas with suitable substrate type (OSPAR, 2014).

The duration of a behavioral disturbance is a key metric to address impacts to individual fitness, and it can inform models and evaluations of impacts by managers as they develop policy recommendations (Southall et al., 2007; Tyack et al., 2011; Ranaweerage et al., 2015; Finneran et al., 2017). Observed D. pealeii alarm responses were transient and had similar movements as anti-predator behaviors observed in other squid species (Mather, 2010). By resuming baseline swimming within only a few seconds, squid may be selecting to maximize other sensory systems or detection needs, particularly audition, to enable vigilance for predators. In late summer, coastal Massachusetts waters and the habitat of this squid are turbid. Such conditions likely renders auditory cues more useful than vision for long-term sensory perception. Low acceleration swimming could serve to decrease chaotic flow around sensory hair cells, which aid in predator detection (Mooney et al., 2010; York and Bartol, 2014; Higham et al., 2015). Another explanation for the short-term alarm responses was that D. pealeii experienced temporary or permanent shifts in hearing thresholds as seen in other species (Smith et al., 2004; Mooney et al., 2009). If so, squid may lack perception of the noise stimulus, explaining the rapid decline in alarm behaviors throughout exposure. Future studies should aim to measure hearing thresholds before and after noise exposure to determine whether D. pealeii desensitized to pile driving noise or experienced physiological impairments.

There was no significant difference in finning rates over noise treatment periods, which is more evidence suggesting pile driving noise does not alter longer term natural swimming patterns. To our knowledge, these are the first data on squid finning rates in semi-wild conditions. Most research on squid locomotion, especially in the field, has focused upon jet propulsion despite finning being integral to squid energetics and ecology (Anderson and DeMont, 2005; Bartol et al., 2016; Cones et al., 2022). Fin-dominated movements increase propulsive swimming efficiency at certain speeds and allow for versatile maneuvers which are thought to aid in squids’ ability to compete with fishes (Hoar et al., 1994; Bartol et al., 2016). Although we did not measure specific energetic costs throughout noise exposure, the finning detection method described here could be used in combination with other metrics (i.e., speed) in the future to estimate free-ranging squid energetics in response to real OSW constructions and more broadly (Anderson and DeMont, 2005; Bartol et al., 2008).

This work revealed that pile-driving noise induced clear but transient disruptions to squid swimming behavior. However, the scale of our experimental pile driving was much smaller than planned future pile driving associated with OSW development within the D. pealeii range in the U.S. eastern coast. The diameter of our steel pile was 0.3 m, while OSW turbines are using piles exceeding 8 m in diameter, perhaps approaching or exceeding 10 m diameter (Steelwind Nordenham, FHL Corporation). As a result, noise propagating from OSW constructions will likely be higher in amplitude and farther reaching, which would expand the volume of ocean where behavioral impacts may be elicited. It also indicates the alarm behaviors seen in our present study may be wide-spread or even more severe.

Consequently, this study represents a significant step toward understanding how an abundant and commercially important species will be impacted by current and planned offshore constructions. Our novel high-resolution movement and particle acceleration data allowed us to be the first study to document both the probability of behavioral change and its duration in multiple spatial scales and noise exposure contexts. Future studies should aim to assess if pile-driving causes horizontal displacement, which is of particular concern the management of commercial fisheries.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

This study was carried out in accordance with the principles of the Basel Declaration and recommendations and approval of the Woods Hole Oceanographic Institution’s (WHOI’s) Institutional Animal Care and Use Committee scientific protocol to TM.

SC, YJ, and TAM designed research; SC, YJ, SF, and NA, performed research; SC and YJ analyzed the data; SC, YJ, SF, NA, and TAM wrote the paper. TAM acquired funding. All authors read and approved the last version of the manuscript.