June 2025 | Oceanography
47
parameters. Predictive models were built using species-specific
call density as the response variable to identify persistent areas
of high trophic transfer or biodiversity in the ADEON study
site (Figure 6b). Ongoing analysis is examining how changes
in abundance and distribution of the forage assemblage var
ies relative to warm/poor and cool/good productivity years off
the US East Coast using taxon-specific community assemblage
metrics from the lander multi-frequency echosounder systems
(Figure 6c). These data can be used to examine regional and
seasonal differences in marine mammal species-environment
relationships. Subsequently, estimates of the acoustic commu
nity structure, for example, time series of different size classes
FIGURE 4. Net tows collected samples of the fish and zooplankton at each site. (a) Top row: flatfish larva, adult myctophid, siphonophore, salps. Bottom
row: copepod, krill, amphipod, pteropod. Aggregations of these animals in the water column are visible as backscattering layers in echograms of acous
tic transects. (b) Fine-scale acoustic surveys (FSASs) measured biological backscatter data in a grid of parallel transects covering an area up to 100 km2
centered on the lander location. (c) Spatial heterogeneity was assessed using the nautical area scattering coefficient (NASC, an acoustic measure pro
portionate to biomass; NASC = 4 pi × 18,522 × area backscattering coefficient in m2/nmi2), integrated as cells 100 m across and 5 m deep (b,c: Blair et al.,
2021). The example transect (b) and FSAS 5 m depth layers (c) were collected at the CHB site the night of December 4, 2017 (UTC). (d) Stationary back
scatter collected at VAC, HAT, and JAX landers (example echogram is two days at hourly resolution from HAT) were compared to FSASs for the two-
week period centered in time on that FSAS date. (e) The two-week time series (black line) was decomposed to the underlying trend component (red
line) for which the partial autocorrelation function (PACF) was calculated (inset). The autoregressive process order defining the temporal autocorrelation
of the time series (inset, green line) was divided by the mean current velocity collected during FSASs surrounding the stationary echosounder location
to calculate a distance estimate that could be compared with FSAS spatial autocorrelation estimates.