The deep pelagic ocean likely contains a greater abundance and higher biomass of organisms than any other major ecosystem. Yet, this ecosystem is largely under-represented in global databases of marine biological records, and deep pelagic communities remain the least-known faunal group on Earth (Webb et al., 2010). Mesopelagic micronekton, small fishes and invertebrates (∼2–20 cm in size) that primarily inhabit depths between 200 and 1,000 m, form an essential trophic link between zooplankton and top predators such as tuna and billfishes (Choy et al., 2013; Young et al., 2015; Drazen and Sutton, 2017) and thus support pelagic fisheries. Diel vertical migration (DVM) by micronekton and zooplankton is the largest migration on Earth, and this contributes significantly to carbon and nutrient cycling in the ocean as organisms feed near the surface and respire at depth. Accordingly, the role of these communities in upper ocean carbon and nitrogen transport is substantial (Longhurst et al., 1990; Davison et al., 2013; Ariza et al., 2015).

When dense aggregations of micronekton and zooplankton form at midwater depths, these are frequently detected by shipboard sonars as acoustic backscatter and are thus termed “deep scattering layers” (DSLs). While swimbladdered fishes and gas-bearing siphonophores can dominate these acoustic signals, DSLs vary globally across pelagic ecosystems in their faunal composition, as well as daytime depths and vertical movements (Klevjer et al., 2016; Sutton et al., 2017). The fish families Myctophidae and Gonostomatidae (particularly Cyclothone spp.) are typically the dominant marine vertebrates caught in midwater trawls around the globe (e.g., Sutton et al., 2010) and likely comprise a large portion of mesopelagic scattering layers. Vertical migrations of DSLs are readily observable in echograms and this enables a visualization of largescale vertical community dynamics in the pelagic ocean (Dietz, 1948; Proud et al., 2018). The number of DSLs and their acoustic backscattering strength (a proxy for animal density) can vary widely at local and global scales (Urmy et al., 2012; Proud et al., 2017). Not all mesopelagic fauna migrate, which results in behavioral variation across the globe. Strong regional differences have been observed acoustically in the proportion of DSLs migrating out of the mesopelagic at night, from 20% migrating in the Indian Ocean and up to 90% migrating in the Eastern Pacific Ocean (Klevjer et al., 2016). However, information about the proportions of mesopelagic fauna actually carrying out DVM on local and regional scales is minimal, and this is a key attribute affecting active carbon flux and the coupling strength between the epi- and mesopelagic.

The ultimate driver of DVM by mid-trophic organisms is visual predator avoidance (Lampert, 1989, 1993). The behavior is thus thought to be driven primarily by downwelling light intensity in the water column where visual predation is common, as changes in the light environment at sunset and sunrise trigger these largescale animal migrations to and from the surface (Dickson, 1972; Kampa, 1975). The depth of daytime DSLs varies regionally (Bogorov, 1946; Dietz, 1948; Isaacs et al., 1974), and gradients in environmental conditions such as oxygen, chlorophyll, sea surface height, and water-column temperature have more recently been observed to drive largescale variability in scattering layer dynamics at regional and global scales (e.g., Bianchi et al., 2013; Netburn and Koslow, 2015; Urmy and Horne, 2016; Proud et al., 2017; Escobar-Flores et al., 2018). Oxygen-depleted zones act as strong bounding conditions for scattering layer depths globally, as hypoxia limits the depths to which animals can migrate and the duration they can spend at those depths. Phytoplankton chlorophyll affects light penetration in the ocean as light is scattered and absorbed by enhanced particle concentrations in surface waters (Yentsch and Phinney, 1989), and it can also be an indicator of surface food availability (Legendre and Michaud, 1999). Sea surface height anomalies can indicate the presence of oceanographic features (e.g., eddies) that may influence micronekton distribution via upwelling or downwelling (e.g., Drazen et al., 2011; Béhagle et al., 2016). Disturbances that alter these environmental drivers, such as climate change and human activities in the ocean, may stress midwater communities and alter their migration behavior. Knowledge of natural ecosystem dynamics and community sensitivities to environmental cues will thus enhance our ability to predict how the deep pelagic will respond to these stressors.

Growing interest in deep-sea polymetallic nodule mining may present a new human-induced stressor for midwater communities. This emerging industry could begin operations in the next several years in the Clarion-Clipperton Zone (CCZ)—a 4 million-km² area in the eastern Central Pacific Ocean (Wedding et al., 2015). This expansive region currently has mining exploration licenses by 15 countries regulated by the International Seabed Authority (ISA). Around the licensed exploration areas, a network of nine Areas of Particular Environmental Interest (APEIs) have been set up as no-mining zones by the ISA. They were developed on the basis of biophysical gradients, biologically unique seamounts, and socioeconomic impacts (Wedding et al., 2013). While nodule mining is expected to have major consequences for abyssal seafloor fauna (Amon et al., 2016; Jones et al., 2017), the industrial-scale operations are likewise expected to impact midwater ecosystems (Drazen et al., 2020). Namely, noise will be produced from vibrations in riser systems used to pump nodules to the surface, and heavy metal-laden sediment plumes generated from surface dewatering of the collected minerals will be released back into the water column. The resulting environmental influences could potentially hinder feeding, reproduction, DVM, and growth among organisms which may ultimately reduce pelagic biomass and diversity or alter their distributions (Christiansen et al., 2019; Drazen et al., 2020).

Only a handful of trawl and acoustic studies have documented midwater communities in the CCZ (e.g., Loeb and Nichols, 1984; Wishner et al., 2018), and this substantial knowledge gap hinders our ability to predict the influences of mining activities on the mesopelagic ecosystem. A recent global biogeographic classification suggests that the CCZ includes two mesopelagic “ecoregions” defined by geographically distinct faunal assemblages (Sutton et al., 2017): the Northern Central Pacific and Eastern Tropical Pacific ecoregions. In this framework, the Northern Central Pacific ecoregion is characterized as an oligotrophic gyre system with a fauna distinct from the central Equatorial Pacific, and the Eastern Tropical Pacific ecoregion is characterized as eutrophic with endemic taxa adapted to low oxygen. While this classification is broad, it is based on existing knowledge of faunal distributions and oceanographic patterns, and it may provide a good foundation from which to build hypotheses about mesopelagic ecosystems across the remote CCZ.

Light intensity and dissolved oxygen levels are known to be strong drivers of DVM and depth structure in deep scattering layers (Netburn and Koslow, 2015; Aksnes et al., 2017) and these parameters vary strongly within the CCZ mining region. Across the eastern Central Pacific Ocean, midwater oxygen partial pressures vary greatly with very low oxygen (∼0.5–2 kPa) extending from east to west and higher oxygen (∼4–7 kPa) in the north and south (Figure 1A). Further, surface chlorophyll and thus light penetration varies north to south and east to west with higher chlorophyll occurring to the east and to the south near equatorial upwelling (Figure 1B). Sea surface height anomalies, which may also influence DVM communities, are highly dynamic across the region due to the transient nature of the mesoscale features they indicate. Mesoscale eddies, for instance, may have a radius up to 100 km, a lifespan of several months, and typically propagate westward across the Pacific Ocean (Cheng et al., 2014; Aleynik et al., 2017). Such oceanographic gradients likely drive variability in the depths and behaviors of deep scattering layers across the CCZ.

The goal of this research is to build a baseline of deep scattering layer depths and DVM behaviors within and around the CCZ using opportunistic active acoustic data, and to reveal the dominant oceanographic factors driving variability in these dynamics. With the Eastern Tropical Pacific oxygen minimum zone (OMZ) extending throughout most of the region, we hypothesize that oxygen will play the largest role in structuring these communities by shallowing daytime DSL depths and increasing the proportions of migrating fauna in more hypoxic waters. The applied goal of this work is to inform developing resource exploitation regulations, which the ISA proposes to have in place by 2021, and to provide some mesopelagic baseline information for subsequently monitoring changes that may occur once largescale mining operations begin.

Our analyses of scattering layers across the CCZ use publicly available shipboard acoustic Doppler current profiler (ADCP) measurements. At low frequencies (75 and 38 kHz), acoustic backscatter in pelagic ecosystems typically indicates the presence of large zooplankton and micronekton, and especially small mesopelagic fishes and siphonophores with gas-filled organs (Simmonds and MacLennan, 2008; Proud et al., 2019). Each frequency highlights the community of organisms whose size and morphology cause them to resonate most strongly at that frequency (Davison et al., 2015). We selected these two frequencies because the acoustic signals extend deep enough (roughly 650 m for 75 kHz and 1,000 m for 38 kHz) to encompass the daytime depths of dominant mesopelagic scattering layers, especially in the eastern Central Pacific where hypoxia limits migration depths (Bianchi et al., 2013; Klevjer et al., 2016; Aksnes et al., 2017; Proud et al., 2018). However, the shallowest measurements from ADCP systems are determined by the combination of transducer depth, blanking distance, and vertical resolution (cell size) which is a function of pulse length. Thus, the narrowband 75 kHz ADCPs in our dataset (with a cell size of 16 m, default blanking distance of 8 m, and transducer depth around 4 m) have the shallowest measurements at about 28 m, and the narrowband 38 kHz instruments (with a cell size of 24 m, default blanking distance of 16 m, and transducer depth around 7 m) have the shallowest measurements at about 47 m. Since these ADCP datasets include many instruments, all with uncalibrated backscatter, we analyzed relative structure within each dataset by evaluating the behaviors of dominant scattering layers, rather than assessing backscatter distributions as proxies for biomass (which is often done with calibrated echo sounder data).

Our analyses of acoustically measured micronekton and large zooplankton migratory behaviors highlight the significance of the Eastern Tropical Pacific OMZ in structuring mesopelagic communities across the CCZ. Amongst the environmental variables explored, midwater oxygen appears to be the strongest driver of daytime habitat depth and vertical migration strength in the CCZ. Midwater oxygen was present in all component models with the highest relative importance (sum of Akaike weights; Tables 2, 3), and it largely structured the predicted behavioral patterns of these communities. As expected, DSL depths shoal and vertical migration strength increases along decreasing midwater oxygen gradients, leading to patterns in these behaviors that primarily follow the “tongue” shape of the OMZ (Figures 1A, 5). Under low oxygen conditions, Klevjer et al. (2016) similarly observed midwater oxygen as a primary driver of global daytime scattering layer depths and migrating proportions. At more regional scales, other studies have also observed scattering layer organisms to closely follow oxyclines (Netburn and Koslow, 2015; Wishner et al., 2018). Although our 38 kHz dataset does not cover the southern and northern CCZ, we suspect that DVM_p in this community is lower across the southern and northwestern CCZ where oxygen is higher, and DVM_p is higher to the northeast following the trend with low oxygen.

By putting a physiological constraint on organisms, oxygen can impact the vertical distribution and duration spent at depth when oxygen levels are low. Above an OMZ, many organisms may only inhabit waters as deep as their hypoxia tolerance allows, and greater proportions of the mesopelagic community vertically migrate to higher oxygen surface waters at night in areas where midwaters are most hypoxic to offset oxygen limitation (Seibel, 2011; Klevjer et al., 2016). Metabolic suppression at depth in OMZs is a common strategy for vertical migrators (Seibel, 2011). Many mesozooplankton and micronekton can only adjust their capacity to acquire oxygen down to a critical oxygen partial pressure threshold of ∼0.8–1.0 kPa (though this varies regionally and between taxa), but some can migrate well into hypoxic waters of OMZs using a variety of adaptations to cope with low oxygen (Sameoto, 1986; Childress and Seibel, 1998; Hofmann et al., 2011; Seibel, 2011; Wishner et al., 2013; Seibel et al., 2016). Climatological oxygen partial pressures at DSL depths in this study had median values of 1.55 and 1.65 kPa for the 75 and 38 kHz communities, respectively, and were similar to critical oxygen thresholds reported for mesozooplankton and micronekton in OMZ regions (Seibel, 2011; Cade and Benoit-Bird, 2015; Seibel et al., 2016).

We identified mean oxygen partial pressure between 200 and 1,000 m as the most parsimonious oxygen metric to include in our models. This range likely encompasses the species-specific depths at which hypoxia tolerance thresholds are encountered, which follows the notion that there is no common oxygen preference across the global ocean (Aksnes et al., 2017). It is important to emphasize that the scattering layer is not singular; rather, multiple layers migrate at variable times and to different depths. Since responses to hypoxia are species-specific, community-wide metrics like single frequency acoustic backscatter might miss variability that is significant to food web dynamics and ecosystem services such as carbon cycling. Future work should thus employ additional techniques like broadband surveys and trawl sampling to assess the taxon-specific details underlying the community patterns we observed in this study.

DSL depths and migration strength were also influenced by other environmental factors according to our models. DSL depths shoaled and migration strength increased with increasing surface chlorophyll for both frequency-defined communities. Enhanced chlorophyll concentration in surface waters can absorb and scatter downwelling light (Yentsch and Phinney, 1989), leading to lower visual predation pressure that could allow zooplankton and micronekton to reside at shallower depths. Chlorophyll-a may also fuel enhanced secondary production that can be channeled up to higher trophic level organisms (Legendre and Michaud, 1999; Hirst and Bunker, 2003), and this food availability in surface waters could lead to enhanced migration strength. While our results evaluate DVM behaviors, recent work using global satellite-based light-detection-and-ranging (lidar) data calculated total DVM biomass (not DVM strength) to be higher in more productive regions with higher food availability, including the tropical Pacific (Behrenfeld et al., 2019).

Our models further suggest that DSL depths deepen and migration strength increases with increasing SSHA. SSHA can act as a surface proxy for meso- and macroscale oceanographic features such as eddies and fronts, both of which have been observed to influence micronekton vertical structure and aggregation in the pelagic ocean (e.g., Drazen et al., 2011; Béhagle et al., 2016). In the northern hemisphere, downwelling (+ SSHA) by anticyclonic eddies can bring micronekton and zooplankton to deeper waters, while upwelling (− SSHA) from cyclonic eddies may do the opposite by shoaling the thermocline and enhancing secondary production (food availability) in shallower waters (Bakun, 2007; Drazen et al., 2011; Béhagle et al., 2014). Enhanced surface chlorophyll associated with both cyclonic eddy interiors and anticyclonic eddy peripheries (e.g., Mizobata et al., 2002) could also decrease light penetration and further influence the shoaling of scattering layers. It is worth noting that SSHA was not important in every component model (Tables 2, 3), and it is likely a weaker indicator of scattering layer dynamics across the CCZ than pO₂ and chl-a. SSHA indirectly influences the region-wide predictions of DSL depths and DVM_p because its inclusion in the empirical models changes the parameter estimates for chl-a and pO₂ during model fitting. However, the direct effect of SSHA is excluded from the region-wide predictions as mesoscale eddies and fronts only have short-term influences on scattering layer behaviors (i.e., weeks to months) in a fixed area (D’Ovidio et al., 2013). Although we lack a complete mechanistic understanding of how each environmental factor influences DVM behavior in micronekton and zooplankton, such satellite-based and climatological variables provide useful proxies that account for a portion of the variability in DSL behaviors in remote regions like the CCZ.

While we observed similar behaviors in relation to environmental factors in both frequency-defined communities, there were significant disparities between DSL depths and migration strengths worth noting. The ranges of observed daytime DSL depths were deeper for the 38 kHz community (290–890 m) than for the 75 kHz community (270–510 m). While there are variable hypoxia thresholds among marine fauna (Hofmann et al., 2011), oxygen partial pressure at daytime DSL depths was similar for 75 kHz (median 1.55 kPa, mode = 0.77 kPa) and 38 kHz (median = 1.65 kPa; mode = 0.73 kPa) communities in our observations. Thus, it is unlikely that differences in oxygen tolerance are driving the disparities in daytime depths between these communities. Clearly, the increased depths observed by the 38 kHz frequency itself may skew DSL locations toward greater possible depths. However, there was no evidence of strong DSLs approaching the maximum depth range in our 75 kHz dataset (i.e., the peak acoustic intensity was well above 650 m). Additionally, there may be biological explanations for these differences. For instance, it is likely that larger organisms dominate lower frequency (longer wavelength) signals where near-resonance effects would amplify scattering by larger organisms with gaseous inclusions. It is possible that these organisms, such as swim-bladdered myctophids, are simply more mobile and thus can reach greater daytime depths. Following size-dependent predator-avoidance hypotheses (Lampert, 1993; De Robertis, 2002), larger micronekton are also more conspicuous to visual predators, and thus may need to reside at deeper daytime depths than smaller organisms to stay hidden, as has been observed with zooplankton (Bollens et al., 2010). Our observations of DVM_p likewise differed by frequency, with a higher median proportion for 75 kHz (0.86) than for 38 kHz (0.63), though total ranges were similar. One possible explanation for this difference is that non- or weakly migratory, lower mesopelagic and bathypelagic fishes (e.g., stomiids) contribute to 38 kHz nighttime backscatter at depths below 650 m (Andersen et al., 1992; Kenaley, 2008; Sutton et al., 2010), which is the limit of our 75 kHz observations. This would lower the day-to-night backscatter ratios and thus lead to lower values of DVM_p.

Although oxygen was an important factor controlling daytime DSL depths for both communities, it was not as influential for the 38 kHz community, and we were ultimately unable to account for as much variability in this group with the environmental variables in this study (R² = 0.31; Table 2). Although DSL depth predictions for this community still largely follow the shape of the Eastern Tropical Pacific OMZ similarly to the 75 kHz community, they deviate slightly from this pattern and appear to reflect a stronger influence of surface chlorophyll, with shallower DSL depths appearing across the southern CCZ (Figure 5B). DVM_p predictions for the 38 kHz community similarly reflect a stronger influence of chl-a, with higher migrating proportions extending south of the OMZ (Figure 5D). This model had higher explanatory power (R² = 0.56; Table 3) than that for DSL depth, but it underestimated DVM_p east of ∼130°W, highlighting the variance unaccounted for by this model (Supplementary Figure 7D). More 38 kHz observations across the southern and northern CCZ (Figures 3B, 4B) are required to fully evaluate predictions there. The relative influences of environmental factors on micronekton and zooplankton behavior may be different between these two mesopelagic communities in the CCZ, and it is possible that there are additional factors affecting the 38 kHz community that are not accounted for in this study.

The acoustic properties of organisms themselves can be a source of variability affecting the signals obtained. Evidence from recent depth-stratified broadband acoustic surveys in the Atlantic Ocean (Bassett et al., 2020) suggests that ship-based echo sounders at frequencies near 38 kHz under sample scattering layers at deeper depths. This is because the contribution to backscatter from marine organisms can be strongly frequency and depth dependent. For example, gaseous swimbladder-bearing fish that resonate near 38 kHz in the upper water column have a higher resonance frequency at greater depths. The combination of this shift in the resonance frequency of those fish, and increased attenuation of higher frequency signals (e.g., 75 kHz), may mean that neither the 38 kHz nor 75 kHz frequency is detecting many organisms in the lower mesopelagic that resonate near 38 kHz in shallower waters. Thus, the ship-based systems we employ here likely have a “blind-spot” at greater depths. In this regard, our observations of DSL depths across the CCZ assume that peak acoustic intensity represents the strongest or densest biomass layers, but we recognize that this may not necessarily be the case. Depths of peak acoustic intensity may also represent the depths of specific communities of organisms that happen to resonate most strongly at that particular frequency and depth (Davison et al., 2015). It is important to consider such complexities in acoustic data for addressing vertical community structure in the pelagic ocean when concurrent trawl or specimen data is lacking.

The community compositions of DSLs we observed from ADCP backscatter in the CCZ are unknown. However, the aforementioned mesopelagic biogeographical ecoregions defined for the CCZ (Sutton et al., 2017) are consistent with our observations of disparate scattering layer behaviors; shallower daytime DSLs and stronger vertical migration in the eastern and central CCZ (Eastern Tropical Pacific ecoregion), and deeper daytime DSLs and weaker vertical migration in the northwestern CCZ (the Northern Central Pacific ecoregion). Thus, it is possible that differences in scattering layer behaviors across the region are indicative of changes in mesopelagic faunal composition. The Northern Central Pacific ecoregion appears to hold high mesopelagic fish diversity and a dominant abundance of fish in the genus Cyclothone (Clarke, 1973, 1974, 1976; Maynard et al., 1975; Barnett, 1983). Despite high abundances of Cyclothone collected in trawls, several species within this genus have regressed or fat-invested swim bladders (Phleger, 1998; Davison, 2011; Dornan et al., 2019), rendering them weak acoustic scatterers. Thus, Cyclothone may not contribute significantly to the acoustic backscatter we observed and could be considered “acoustically cryptic” (Dornan et al., 2019). However, those species containing a gas-filled swim bladder (e.g., C. pseudopallida) are likely non-migrators (Davison, 2011) and may contribute to observed non-migratory scattering layers. Similarly to our observed DSL depths and DVM_P in the NW corner of the CCZ (Figures 3, 4), nearly half of daytime micronekton biomass below 400 m has been found to migrate into the upper 400 m at night around Hawai’i (Maynard et al., 1975). Also consistent with our acoustic observations, micronekton and macrozooplankton vertical distributions in parts of the Eastern Tropical Pacific Ecoregion suggest that the upper water column down to 550 m is strongly influenced by DVM in and out of the upper oxycline and OMZ core (Maas et al., 2014). Fishes and crustaceans (particularly euphausiids) appear to dominate this community. Secondary biomass peaks in zooplankton and micronekton have also been observed at OMZ lower oxycline depths from trawls and acoustics (Supplementary Figure 6; Bianchi et al., 2013; Wishner et al., 2013; Maas et al., 2014; Klevjer et al., 2016). This is consistent with our observations of additional, weaker scattering layers extending well below ∼500 m in the OMZ (Figure 2 and Supplementary Figure 1), following the hypothesis that these low-oxygen waters provide some organisms with a refuge from large visual predators with greater oxygen demands (Seibel, 2011).

Despite our minimal understanding of mesopelagic community composition across the CCZ, these results provide basic information about deep pelagic communities that is relevant and significant to deep-sea mining in the region. After nodules are pumped from the abyssal seafloor to the surface, sediment plumes will be released into the water column after dewatering of polymetallic ore. As these sediment plumes will likely impact midwater communities, Drazen et al. (2020) recently recommended that they be discharged below depths of 1,500–2,000 m (i.e., well below the lower bounds of the mesopelagic), or be released all the way back to the abyssal seafloor. However, the ISA has yet to set that discharge depth, and if released too shallow (i.e., into the mesopelagic), these plumes could interact with and influence DVM behaviors and their associated ecosystem functions. An understanding of how DVM behaviors vary among licensed exploration areas and APEIs is crucial. For example, our overall findings suggest that dominant daytime DSLs are shallowest and vertical migration strength is highest in the eastern-most license areas (Figure 5). Mesopelagic mining impacts in this region could therefore have a greater influence on vertical coupling and DVM-mediated carbon transport (Longhurst et al., 1990; Ducklow et al., 2001) depending on the depth zones affected by mining.

Importantly, the majority of licensed mining exploration areas in the CCZ lie within the OMZ, and consequently these areas contain the shallowest scattering layer depths and strongest vertical migrations. However, the system of APEIs surrounding these mining zones lie on the boundaries or just outside of the OMZ, exhibiting relatively deeper scattering layers and weaker migrations. Many pelagic OMZ fauna are highly adapted (behaviorally and biologically) or endemic to low-oxygen habitats (Antezana, 2009; Seibel et al., 2016; Sutton et al., 2017). Our observations of disparate behavioral dynamics, combined with this knowledge of OMZ-specialist fauna, indicates that these protected, no-mining reserves likely do not adequately represent or protect the OMZ organisms in oceanic midwaters that could be impacted by deep-sea mining (Drazen et al., 2020). The APEI network of nine 400-km² reserve areas was spatially designed to protect the diversity of benthic habitats and fauna within the CCZ mining region (Wedding et al., 2013), so it is not surprising that these areas may not protect midwater habitats as they are currently placed. However, in order to prevent excessive harm to midwater ecosystems, whose services are closely linked to humans, the deep pelagic must be included in precautionary management plans and future environmental research in the CCZ (Christiansen et al., 2019; Drazen et al., 2020). Future work combining broadband acoustics and depth-stratified trawl sampling would provide a more thorough understanding of the organisms comprising DSLs in the CCZ (Kloser et al., 2002; Escobar-Flores et al., 2019), and to what extent different taxa contribute to the patterns of habitat depth and vertical migration that we observed here in opportunistic shipboard ADCP data. Nonetheless, this study provides the first major examination of midwater community dynamics across the remote CCZ, highlights the complexities of mesopelagic micronekton and zooplankton behavior in the open ocean, and raises concerns about the effectiveness of current no-mining areas for pelagic habitats.

The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author/s.

JP and JD were responsible for the project design and funding was secured by JD. JP led ADCP and statistical analyses as well as manuscript writing. EF led the development of ACDP backscatter processing tools and interpretation. JD and EF aided in the refinement and selection of environmental variables. JG contributed to statistical analyses and statistical model development. BJ contributed to interpretation of the acoustic results. All authors contributed to manuscript writing and have approved the submitted document.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.