Fig 1.
(a) Satellite surface absolute dynamic height (ADT, m) on April 28 within the cyclonic (blue) and anticyclonic (red) eddies. The African and Madagascar Island landmasses are shown in grey. (b) The ship’s track is shown by black lines, and (c) the moving vessel profiler locations are shown by purple lines and are superimposed on the mesoscale eddy field (d) Mean current velocity (m s-1) for the u and v velocity components (red data points) and all data points (black) during Leg 1 of the RESILIENCE cruise in the transition zone (TZ), anticyclone (AC) and cyclone (C). A mix zone was determined and discarded from further analyses.
Table 1.
Classification of Leg 1 acoustic transect based on the location of the data points within the cyclonic, anticyclonic eddies, transition and mix zones.
Table 2.
Echosounder parameter settings used during the acoustic data acquisition and processing.
Fig 2.
Flowchart summarizing the Escore methodology introduced in this study.
Table 3.
Glossary of terms used in the acoustic backscatter classification approach.
Fig 3.
Vertical profiles of mean eddy kinetic energy (m2 s-2), fluorescence (mg m-3), salinity and temperature (°C) within the transition zone (TZ), cyclone (C), and anticyclone (AC).
Fig 4.
Day and night mean vertical Sv (dB) profiles ± standard deviations (SD) within the anticyclone (AC), cyclone (C) and transition zone (TZ) at the 18, 38, and 70 kHz frequencies.
Table 4.
Mean ± standard deviation daytime (D) and nighttime (N) 18–200 kHz acoustic metrics in the anticyclone (AC), cyclone (C), and transition zone (TZ).
The maximum echo-integrated depths were 1000 m, 800 m, 450 m, 250 m, and 120 m for the 18, 38, 70, 120 and 200 kHz, respectively.
Fig 5.
Stacked bar charts of NASC, sA, during day (D) and night (N) at the 18, 38, 70, 120, and 200 kHz within the surface (top 200 m), intermediate (200–400 m), and deeper layers (> 400 m) in the anticyclone (AC), cyclone (C), and transition zone (TZ).
Table 5.
Mean ± standard deviations daytime (D) and nighttime (N) nautical area scattering coefficient, sA (m2 nmi-2) from 15 to 250 m, of echo-classes 1 to 4 in the anticyclone (AC), cyclone (C), and transition zone (TZ).
Fig 6.
Frequency response curves at 18, 38, 70, and 120 kHz frequencies of echo-classes 1 to 4 during nighttime.
Mean Sv is shown by the solid lines and confidence intervals with the dashed lines.
Fig 7.
Left-hand panels: Echograms of Sv values from the surface to 250 m for the echo-classes 1 to 4 in the (a) anticyclone (AC), (b) cyclone (C) and (c) transition zone (TZ). Nighttime is denoted by the black rectangles below the echograms, sunrise in orange, and daytime in white. Right-hand panels: Vertical section plots of temperature (°C), salinity, and fluorescence (mg m-3) in the (a) AC, (b) C and (c) TZ, corresponding spatially and temporally to the RGB echograms of the left-hand panels.
Fig 8.
Smooth functions for the GAMMs showing the influence of significant covariates on the NASC values of echo-classes 1 to 4.
The y-axes show the smooth function of each covariate, with the estimated degrees of freedom in brackets. The predicted models are shown by solid lines and the 95% confidence intervals by filled contours. Data observations are shown on the x-axis.
Fig 9.
Daytime and nighttime canonical analysis of principal components (CAP) for (a) echo-class 1, (b) echo-class 2, (c) echo-class 3, and (d) echo-class 4. Each point represents one sA value coloured by oceanographic structure (AC: anticyclone, C: cyclone, and TZ: transition zone). The direction and length of the arrows mark the direction and rate of steepest increase of the given significant environmental variable. Percentage variability along each axis is given in between parentheses.
Table 6.
Environmental predictor loadings for the daytime and nighttime CAP analyses of echo-classes 1 to 4.
Predictors ≥ 0.6 on either of the first two CAP axes are shown in bold. Percentage variability is given in between parentheses.
Table 7.
Theoretical scattering models classified into the four echo-classes 1 to 4.
The Escore value gives the relative position of each scattering model to the centroid of the echo-class. The smaller the Escore value, the closer the model is to the centroid of that echo-class.