NLIWs signatures were identified in synthetic aperture radar images (SAR images) acquired by the Sentinel-1A and Sentinel-1B satellites, operating in the C-band, using Interferometric Wide swath mode with spatial resolution of 20 m and swath width of 250 km. The satellite images were downloaded from the Alaska Satellite Facility. A total of 3491 SAR images were available for the period from January 2015 to December 2024. The spatial distribution of SAR image coverage across the study area is shown in Figure 1.

As shown in Figure 1a, the highest satellite data availability (over 700 SAR images over 10 years) is observed in the area of the Fourth Kuril Strait, the southeastern shelf of the Kamchatka Peninsula, and the northern part of the Kronotsky Bay. The continental shelf and upper continental slope are covered by 300 to 400 SAR scenes in ten years, while the open-ocean region beyond the shelf is covered by fewer than 200 scenes in ten years.

The SAR images were processed using the Sentinel Application Platform (SNAP) software developed by the European Space Agency, which is freely distributed under the GNU General Public License v3 (https://step.esa.int/main/download).

Surface manifestations of NLIWs are identified in SAR imagery as quasi-parallel, alternating bright (rough surface) and dark (smoothed surface) arc-shaped bands which are grouped into wave packets (Robinson, 2010). The visibility of NLIW surface signatures in radar imagery is due to the modulation of gravity-capillary waves by the divergent and convergent components of surface currents induced by internal waves (Alpers, 1985). In this study, NLIW manifestations were visually detected in SAR images, and further analysis of their properties followed the methodology described by Kozlov et al. (2022). A visual search for signatures resembling the shapes in Figure 1c was conducted on the SAR images. Using the built-in tools of SNAP software, a curve matching the shape of the leading crest of the manifestation was constructed as well as a cross-section of the radar signal intensity (Line A1–A2) along the entire extent of the NLIW manifestation (Figures 1C, D). For each identified feature the following parameters were determined: coordinates of the leading crest, number of waves in the packet (n), length of the leading crest (Cr), mean wavelength (λ), and propagation direction (φ). The number of waves per packet was visually counted as the number of visible arc-shaped bands or as the number of peaks in the radar signal intensity cross-section. The mean wavelength was calculated along A1–A2 by averaging the distances between adjacent bands within wave packet. The direction of propagation was determined along the normal to arc-shaped curve in middle of leading crest (Figure 1c). All derived NLIW characteristics were compiled into a unified database.

To identify “hot spots” in the field of NLIW manifestations — areas where NLIWs are most frequently observed, following Sabinin and Serebryany (2007) — an occurrence frequency was computed as the ratio of the number of NLIW events to the number of SAR images within grid cells of 0.2° latitude × 0.24° longitude. “Hot spots” were defined as areas where NLIW occurrence exceeded the background level (0.05) by at least a factor of two. This level was chosen as the long-term average occurrence frequency of NLIWs throughout the entire study region.

To estimate NLIW phase speeds, pairs of optical images taken within the same season (spring–summer) by OLI/TIRS sensors onboard Landsat-8 and MSI sensors onboard Sentinel-2 (levels L1 and L2, with spatial resolutions of 30 and 10 m, respectively) were analyzed, provided that the time difference between image acquisitions did not exceed 30 minutes. On these images visually similar NLIW manifestations—matching in shape and location—were identified, and the displacement of their leading crests was measured. The NLIW phase speed cph was then computed as the ratio of crest displacement to the time interval between image acquisitions. Propagation direction was defined as the direction of crest displacement. These satellite-based velocity estimates were compared to theoretical values derived from the dispersion relation for a two-layer stratification (Konyaev and Sabinin, 1992; Kozlov et al., 2022). In this case, the phase speed of the NLIW is defined as: cph=gΔρρh1(H−h1)H, where g is acceleration due to gravity, Δρ=ρ2−ρ1, ρ=0.5(ρ1+ρ2),  ρ1 and ρ2 are the density of the upper and lower layers, correspondently, h1 is the thickness of the upper layer, H  is full depth. Stratification parameters were found from in situ vertical profiles of sea temperature and salinity which were available and close to place and date of the detected NLIW packages.

To illustrate the role of tidal forcing in the generation of NLIWs, examples of SAR images containing two consecutive NLIW packets propagating in the same directions were selected. The distance between these NLIW packets was interpreted also as the wavelength of the internal tide which generates them. The time interval between the generations of the packets was calculated as ratio of this distance to internal wave speed cph and then was compared with period of the dominant diurnal tidal harmonic (K1) of 23.93 hours.

To assess the potential influence of NLIW propagation on the redistribution of chlorophyll-a, optical imagery from the MSI sensor onboard Sentinel-2 (L1, 10-m resolution) was used for the period from April to August in 2018 and 2024. For 2018, 26 cloud-free images (cloud cover<30%) were available; for 2024, 17 such images were analyzed. On these images structures resembling NLIW manifestations—displaying alternating bright and dark bands in the blue and green channels—were visually identified. Fragments containing such features were processed using the C2RCC algorithm [https://c2rcc.org/] to retrieve chlorophyll-a concentration. To jointly assess the interannual variability of chlorophyll-a and the number of wave packets in the identified NLIWs “hot spots”, MODIS chlorophyll-a data (4 km resolution) were averaged over 1° boxes centered on the “hot spot” locations using the GIOVANNI online tool [https://giovanni.gsfc.nasa.gov/giovanni/].

Total tidal currents were derived from a regional simulation of barotropic tidal dynamics off the southeastern coast of the Kamchatka Peninsula using the finite-volume model FESOM-C (Romanenkov et al., 2023). The regional model was implemented on an unstructured triangular mesh with spatial resolution adapted to local bathymetry. Barotropic tidal velocities were estimated for the entire study period and included the following tidal constituents: M2, S2, N2, K2, K1, O1, P1, and Q1. The use of a regional tidal model provides a more accurate representation of tidal currents compared to global tidal models. Notably, the tidal regime in the study area is classified as mixed with a dominance of diurnal constituents over semidiurnal ones.

Daily and monthly mean fields of seawater temperature and salinity from January 2015 to December 2024, obtained from the Copernicus Global Ocean Physics Reanalysis product (CMEMS GLORYS12v1 GLOBAL_MULTIYEAR_PHY_001_030) (https://doi.org/10.48670/moi-00021), were used to calculate buoyancy frequency and density gradient.

Additional parameters derived from this dataset included mixed layer thickness, which was used to analyze factors influencing the interannual and intraannual variability of NLIW manifestations. As an additional forcing factor, hourly wind speed data were used during period 2015–2024 from the WIND_GLO_PHY_L4_MY_012_004 reanalysis product (https://doi.org/10.48670/moi-00185) and then were averaged to monthly values.

To support the analysis of NLIW characteristics inferred from SAR imagery, in situ observations were conducted in August 2024 in the Avacha Bay (see marker in Figure 1). Measurements were performed from a drifting vessel over the shelf at depths ranging from 150 to 200 m using a CTD48M probe (Germany). The vessel’s drift speed throughout the survey did not exceed 0.05 m/s, as determined via a Garmin Ertex 21 GPS receiver. The CTD probe was deployed and recovered using a winch at an average vertical speed of 1 m/s, scanning the upper 50–55 m of the water column. The total duration of the intensive CTD profiling time series was approximately 6 hours. Vertical temperature profiles were linearly interpolated in depth while preserving the temporal resolution, allowing reconstruction of isotherm displacements. Based on the fluctuations of the isotherm associated with the pycnocline core, individual NLIWs were identified and their key properties—wave height and period—were determined. According to Kozlov et al. (2022), wave height was defined as the mean of the amplitudes of the leading and trailing slopes, and the wave period was computed as the time interval between successive isotherm deflection maxima. The resulting wave periods and phase speeds were used to estimate characteristic wavelengths via the dispersion relation for two-layer stratification. These estimates were then compared with parameters of NLIWs observed in satellite images.

In the discussion of how NLIWs may influence chlorophyll-a variability—and thus the food availability for juvenile fish—estimates of the abundance of 2-year-old Eastern Kamchatka walleye pollock cohorts (2017–2024 years classes) were used, as presented by Varkentin et al. (2024). These estimates were obtained using a cohort state-space model combined with an unscented Kalman smoother (Ilin, 2022). In addition to standard inputs such as the age–year catch matrix, mean body weight, and proportion of mature individuals, the model was calibrated using standardized indices of catch per unit effort, spawning stock biomass, and egg production in the main spawning area located in the deep-water canyons of the Avacha Bay.

During the analysis of SAR images, 3895 manifestations of NLIW packets were identified. The number of waves within a packet reached up to 45, while the length of the leading wave crest varied between 2 and 174 km. Most frequently, wave groups consisted of 8 to 14 waves with wavelengths within the packet ranging from 290 to 480 m and leading crest lengths from 22 to 51 km. The average packet comprised 10 waves with an approximate wavelength of 350 m and a leading crest width of about 29 km. The spatial distribution of the leading crests of NLIW packets is shown in Figure 2a, and the occurrence frequency is presented in Figure 2b. NLIW manifestations were observed almost ubiquitously over the shelf and continental slope. In areas with depths exceeding 2000 m, wave’s groups were predominantly recorded in the Avacha Bay and south of it. The maximum NLIW occurrence frequency (approximately on every seventh image) was found in the Kronotsky Bay region, near Cape Shipunsky, and south of the Fourth Kuril Strait. In these areas the occurrence frequency exceeds the background level by at least a factor of four, allowing these zones to be classified as “hot spots” for NLIW manifestations. These “hot spots” are detected near the northern Kuril Islands, on the southeastern shelf of the Kamchatka Peninsula, in the Kronotsky Bay, and near Cape Shipunsky.

Data on the interannual variability of detected NLIW manifestations are summarized in Table 1. It should be noted that the distribution of available SAR images across years is uneven: approximately 200 images per year were processed for the study area in 2015–2016, whereas more than 450 images per year were available in 2019–2021. The data show that when the number of available images is low, there is a correlation between data volume and the number of detected wave manifestations. However, once a certain data threshold is reached (over 300 images), this dependence disappears, indicating that such data volume is sufficient for a representative assessment of wave activity for a given year. On average, in years with adequate coverage of SAR images, about 420 NLIW packet manifestations are recorded in the study area annually. The highest ratio of waves per image was observed in 2024 and the lowest in 2020. Despite pronounced interannual fluctuations, there is an overall increasing trend in the number of recorded wave manifestations over the study period.

The average number of waves per packet and the length of the leading crests did not exhibit a consistent multi-year trend and showed no pronounced interannual variability (Table 1). Annually both individual wave manifestations (solitons) and large packets containing 35–45 waves (4–5 times greater than the average) were observed. The individual wavelengths varied widely, ranging from 70 to 1500 m. The length of the leading crests of NLIW packets ranged from 2 to 170 km. The mean values of this parameter during the first two years of the study period were 1.5–2 times higher than those in the subsequent eight years, which is likely due to the limited availability of satellite data in the initial years. In years with good SAR data coverage the average crest length was approximately 27 km.

The seasonal variability of internal wave manifestations can be inferred from Figure 3. The SAR data coverage for each season is substantial with more than 800 images available for the winter and spring seasons and over 900 images for summer and autumn.

Figure 3A shows that the winter period is characterized by low NLIW occurrence (less than 0.025) across the study region. Isolated, scattered manifestations are primarily recorded along the shelf break and near the continental slope. Overall, the number of waves detected during winter is several times lower than in other seasons. However, the wave packets tend to be more pronounced with the leading crest length averaging over 40 km. These packets also contain a larger number of waves with individual wavelengths exceeding 400 m compared to other seasons.

In the spring period (Figure 3B) the number of detected waves increases by a factor of 3.5 compared to winter, though their geometric dimensions are significantly smaller. The highest occurrence frequencies (0.03–0.04) are observed in the Kronotsky Bay, near the Cape Shipunsky, and in specific areas of the shelf and continental slope off the southeastern Kamchatka Peninsula. In the rest of the region occurrence frequencies remain below 0.025.

During summer (Figure 3C) the number of waves recorded increases by an order of magnitude compared to spring, while the geometric characteristics of the packets remain similar to those in the preceding season. The highest frequencies (exceeding 0.25) are found in the Kronotsky Bay, near the Cape Shipunsky, on the southeastern Kamchatka shelf, and near the Fourth Kuril Strait.

In autumn (Figure 3D) the number of detected waves decreases by half relative to summer. However, packets with larger crest lengths and higher numbers of waves are registered. The maximum occurrence frequency (above 0.12) is observed near the northern Kuril Islands, along the southeastern Kamchatka coast, and near the Cape Shipunsky, where the absolute peak in occurrence is recorded.

Thus, Figure 3 reveals significant seasonal variability in both the frequency and spatial distribution of areas with elevated NLIW occurrences within the study region.

Table 2 presents the intra-annual variability of NLIW occurrences by month. The distribution of SAR images throughout the year is fairly uniform; however, the number of detected waves varies substantially: the minimum values are observed in February, while the maximum (up to 70 times higher) occur in August. Overall, the lowest monthly wave counts (fewer than 21) are recorded from January to March, and the highest—in July and August (over 820). Meanwhile, other wave parameters show less pronounced fluctuations and generally follow the seasonal pattern described above.

In our previous studies (Svergun et al., 2023, 2024), we examined the relationship between high NLIW occurrence in certain years and months and factors that could explain the spatial distribution features of NLIWs. “Hot spots” were identified where more than 50% of all recorded NLIW manifestations were observed. Parameters such as tidal body force (TBF), topographic criticality, and internal Froude number were evaluated, demonstrating the importance of tidal forcing in explaining the causes and features of NLIWs in the region. The results obtained in the present study do not contradict this idea but rather complement it. The stability of these “hot spot” areas both seasonally and interannually supports the hypothesis of their tidal origin.

In the present work we expand our understanding of the variability features of NLIW manifestations for the area best covered by different observational data, which is close to the “hot spot” near Cape Shipunsky (marked by a black rectangle in Figure 2b). A joint analysis was conducted of NLIW variability together with the following factors: (1) upper layer stratification parameters, (2) surface wind speed, and (3) velocities of total tidal currents.

Figure 4 shows the curves of mean values of the above characteristics averaged over 10 years for each month in the “hot spot” near Cape Shipunsky. The shaded corridor indicates the minimum and maximum values of each characteristic per month. Although, as previously noted by Svergun et al. (2024), the masking influence of wind on the number of NLIW manifestations can be significant, the main factors controlling their variability are a combination of favorable hydrological conditions and the strength of tidal forcing. A high multi-year mean occurrence frequency of NLIW manifestations, corresponding to the upper quartile and equal to 1.86, is observed from June to September. This is associated with a multi-year mean density gradient ranging from 0.06 to 0.11 kg/m⁴, surface mixed layer depth of approximately 10 meters, wind speed between 4 and 8 m/s, and tidal current velocity from 0.22 to 0.26 m/s.

Thus, maximum probabilities of NLIW occurrences (defined as the ratio of detected NLIW events to the number of SAR images) occur during the warm months from June to September. During these months we also observe the sharpest pycnocline, minimum thickness of the upper mixed layer, and wind speeds generally below the threshold, above which NLIW detection by SAR dramatically decreases. Additionally, the highest tidal current speeds are observed, whose semiannual modulation in this region is primarily explained by the combination of K1 and P1 tidal harmonics (Svergun et al., 2023).

The pattern of interannual variability and the range of seasonal variability of NLIW manifestations closely coincide with the change in the maximum of density gradient. It is important that this maximum occurs in the pycnocline at the lower boundary of the upper mixed layer formed due to summer heating. Weak density gradient due to autumn-winter convection corresponds with the reduction in the number of surface NLIW manifestations even with high tidal current velocities. Thus, it can be concluded that only a combination of hydrological conditions such as strong subsurface pycnocline, intensified tidal currents and weak winds is linked to the maximum of NLIW occurrences.

As shown in the previous section, the area near Cape Shipunsky is characterized by high NLIW activity based on SAR data. To verify the correspondence between the characteristics of the detected NLIWs and fluctuations of the near-surface pycnocline (thermocline), results from intensive CTD profiling were used (Figure 5). The temperature records in the 5 to 50 m layer show short-period oscillations with maximum amplitudes observed in the 10 to 20 m layer, where CTD profiling revealed a pronounced seasonal pycnocline with a density gradient of 0.08 kg/m⁴ coinciding with the thermocline.

The amplitudes of NLIWs recorded by in situ methods ranged from 0.5 to 3.5 m with an average amplitude of 1.5 m. The NLIW period varied from 6 to 20 minutes. The phase speed cph  was about 0.4 m/s estimated from CTD data (see Section 2.1). Accordingly, the wavelengths as the product of the phase speed and periods ranged from 140 to 480 m and this is consistent with the wavelength range observed in satellite images.

To verify the validity of the two-layer model dispersion relation for estimating the phase speed of NLIWs, an analysis was performed on the propagation speed of surface manifestations of NLIWs using sequential satellite images. Figure 6 shows the positions of wave fronts superimposed on a bathymetric map along with their propagation directions and the vertical density distribution from in situ data. The phase speed.estimates obtained from tracking the displacement of successive NLIW manifestations closely agrees with one derived from the dispersion relation for two-layer stratification (0.17 m/s from the two-layer dispersion relation versus 0.18 m/s calculated as the ratio of the distance between NLIW manifestations on two consecutive images to the time interval between them). This confirms the adequacy of applying the two-layer approximation in this place. It is noteworthy that based on tidal model data and the observed propagation directions of NLIWs, their formation is likely caused by the interaction of tidal currents with the shelf break. To further test this hypothesis, another example is considered.

Figure 7 shows two successive packets of NLIWs, presumably propagating from a single generation source, which we believe, as in the previous case, is located at the shelf break. The distance between the leading crests is 27 km. The vertical distribution of seawater density (panel b), based on reanalysis data in the observed NLIW area, justifies using the dispersion relation for a two-layer stratification to estimate the internal wave propagation speed and thus we get cph = 0.29 m/s. The period between the two generation events was estimated as the ratio of the distance between the leading crests (25.6 km) to the internal wave speed of 0.29 m/s, yielding approximately 24 hours. This daily period between consecutive NLIW generation events serves as further evidence supporting the tidal generation mechanism.

Let us focus on the characteristics of tidal currents in the Avacha Bay area, specifically in the “hot spot” near Cape Shipunsky (see Figure 2b), where the individual cases of NLIW manifestations shown in Figures 6 and 7 were examined. Figure 8 presents ellipses of tidal currents for the semidiurnal M2 and diurnal K1 harmonics as well as background currents in the upper layer.

A significant difference between semidiurnal and diurnal tides in the study area is explained by the influence of trapped shelf waves (Romanenkov et al., 2023), which may arise due to intense scattering of the diurnal barotropic tide by coastal and bottom heterogeneities. As shown in Figure 8, the diurnal harmonic currents locally increase near Cape Shipunsky and directly in front of the deep submarine canyon, where they exceed the semidiurnal harmonic currents by an order of magnitude. It should be noted that the area of maximum NLIW occurrence (see Figure 2) corresponds to the area of maximal tidal currents. Considering the tide as a source of internal waves, it should be borne in mind that tides at subinertial frequencies can only generate internal waves that cannot freely propagate away from their generation sites but rather dissipate or break down into NLIW due to nonlinear effects. For the subinertial internal semidiurnal tide at high Arctic latitudes the process of disintegration into high-frequency waves has been confirmed by observations and modeling (Morozov and Paka, 2010; Morozov, 2018). In our case, the diurnal (subinertial) components of the total tidal current prevail over the semidiurnal components; therefore, the topographically trapped internal diurnal tide can be regarded as the source of numerous NLIWs.

In the case presented in Figure 7, the exact moment of NLIW packet generation cannot be determined with certainty. However, we suggest that the most probable timing corresponds to the interval of maximum tidal currents, when their direction aligns with the background flow and the internal Froude number approaches unity. The internal Froude number was estimated as the ratio of the total current speed to the NLIW phase speed: Fri=VmaxCph, where Vmax=(Vtidе+Vbackground), Vtidе  is the magnitude of the tidal current at the time of directional alignment with the background flow within the diurnal cycle, Vbackground is the magnitude of the background current according to reanalysis data, and Cph is the phase speed of the NLIW. According to Vlasenko et al. (2005), when Fri≥1 internal tidal waves undergo transformation and evolve into packets of NLIWs.

A total of 33 instances of banded brightness patterns in the blue and green channels, potentially associated with the redistribution of chlorophyll-a concentrations by propagating NLIWs (Vázquez et al., 2009; Muacho et al., 2013, 2014), were identified across 43 Sentinel-2 images collected between April and August of 2018 and 2024. For these cases, chlorophyll-a concentrations were retrieved using the C2RCC algorithm. An example of chlorophyll-a distribution in an area of frequent NLIW activity near Cape Shipunsky is shown in Figure 9.

Figure 9a illustrates the presence of suspended matter in the shelf region near Cape Shipunsky, ranging in color from blue-green to white. The scene exhibits distinct banded structures, bright stripes against a darker background, that closely resemble surface manifestation of NLIW packet. The chlorophyll-a retrieval using the C2RCC algorithm (Figure 9b) reveals that concentrations within the high-brightness bands range from 1 to 4 mg/m³. The same banded pattern, attributable to NLIWs and apparent in true-color composites, is also preserved in the chlorophyll-a concentration field. A transect across one such feature shows concentrations varying from 0.5 to 2 mg/m³, while surrounding waters display nearly zero values.

This spatial pattern is typical of nearly all observed cases of brightness banding in the blue and green channels during both spring and summer. The majority of observed NLIW manifestations propagate shoreward and are located over the shelf and continental slope. Thus, this example demonstrates that in “hot spots” NLIWs propagating onto the shelf may play a significant role in redistributing phytoplankton biomass.

Given that August is the month of peak NLIW activity (see Table 2), we analyzed the interannual variability of chlorophyll-a concentration and NLIW occurrence frequency in the “hot spot” region near Cape Shipunsky. The analysis covered Augusts from 2017 to 2024 (see Figure 10).

In general, the lowest frequencies of NLIW occurrences correspond to the lowest chlorophyll-a concentrations, while the highest frequencies align with peak concentrations. However, a notable area stands out where moderately high NLIW occurrence frequencies are associated with both low and high chlorophyll-a concentrations. Despite this scatter, the relationship between chlorophyll-a concentration and NLIW frequency can be roughly approximated by a linear trend, though with some caveats. This suggests that multiple factors likely influence chlorophyll-a variability. Nonetheless, the frequent appearance of NLIW-like surface patterns in regions of enhanced green-channel brightness in optical satellite imagery supports the potential existence of such a relationship.

It is well established that food availability during the transition of fish larvae to exogenous feeding plays a critical role in determining fish cohort recruitment success. This issue, as it pertains to walleye pollock in the Sea of Okhotsk, was addressed to some extent by Gorbatenko et al. (2004) and Maksymenkov (2007). The diet of larval and post larval stages is not particularly diverse with both the composition and size of consumed prey strongly dependent on larval size. Phytoplankton generally serves as a food source only at the very onset of exogenous feeding (Kamba, 1977; Maksymenkov, 1984; Nishiyama et al., 1986). However, in certain regions—such as the Korea Bay and the Pacific waters off Kamchatka Peninsula — phytoplankton plays a significant role in the diet from the earliest larval stages.

As larvae grow, smaller prey items are progressively replaced by larger ones. In many areas nauplii of copepods dominate the diet of larvae up to 14 mm in length. In some locations eggs of copepods and euphausiids are also commonly consumed by smaller larvae. With increasing size the larval diet becomes more diverse. By 6–8 mm length early copepodite stages, as well as small copepods such as Oithona similis, Pseudocalanus minutus, Acartia longiremis, and others, begin to appear in the diet. From 14 mm onwards Pseudocalanus minutus becomes the dominant prey item. The role of this species—across both early developmental stages and adults—is exceptional in the diet of pollock larvae. Pseudocalanus minutus typically constitutes a major portion of the plankton community, both as nauplii and copepodites, and in most cases determines the condition of the larval feeding base for pollock (Shuntov et al., 1993).

Abundance and availability of planktonic prey are not always the primary factors ensuring high survival of fish larvae (Drinkwater and Myers, 1987; Gaard, 1999; Sundby, 2000). Of particular importance for larval survival are the early-summer formation of subsurface “growth layers”, where large aggregations of walleye pollock larvae are often observed. If these layers contain sufficient concentrations of accessible prey, they may significantly enhance larval survival.

Given the potential role of NLIWs in delivering nutrients to shelf waters that stimulate phytoplankton growth, we examined the abundance of 2-year-old pollock and the frequency of NLIW manifestations over the period May–September. In Figure 11 lowest frequencies of NLIW occurrences correspond to the lowest abundance of 2-year-old walleye pollock with a two-year lag, while the highest frequencies align with peak abundance with a two-year lag. However, this trend is not observed in 2019/2017 and 2023/2021. A joint interpretation of Figures 9 and 10 suggests that NLIWs may play a role in regulating the abundance of ichthyoplankton and early fish life stages. Clearly, further testing of this hypothesis will require new targeted observational evidence.