An underwater glider (Seaglider – SG675) was deployed between 22 March to 23 May 2023 (63 d) within the Cape Basin, SW of South Africa. The glider sampled between the surface and 1000 m depth, completing 297 profiles and covering a distance of approximately 1650 km (Fig. ). The Seaglider was equipped with an unpumped Seabird CT sensor that measured conductivity, temperature, and pressure, an Aanderaa optode 4831 that measured oxygen and Wet Labs ECO puck that measured Chl a and backscatter at 470 and 700 nm.

To ensure data quality, conservative temperature and salinity outliers were removed using the wrapper function in GliderTools . Within this function, any detected spikes were eliminated using a 5-point rolling median and subsequently smoothed with a Savitzky-Golay filter. Raw data was corrected for sensor thermal-lag using the IOP basestation code that applied a first-order thermistor-response lag . The remaining thermal lag in the final dataset was found negligible, as the absolute difference between the mean of all climbs and dives in conservative temperature and absolute salinity at the thermocline was 0.04 °C and 0.015 gkg-1, respectively. Part of the climb–dive differences may also reflect spatial variability, especially in strong frontal regions. The mixed layer depth (MLD) is defined as the depth z at which the difference in potential density referenced to the potential density at 10 m first exceeds a threshold value of 0.03 kgm-3 .

Oxygen is measured by the glider with an Aanderaa Optode 4831. The oxygen dataset is smoothed using a Savitzky-Golay filter with a 2nd order polynomial. The ship CTD is used to calibrate the glider's oxygen measurements during deployment. An offset of 22.28 µmkg-1 is added to the glider's oxygen concentration along the transect.

AOU is calculated: 1AOU=[O2 solubility]-[O2 observed] where [O2 solubility] represents the solubility of oxygen in seawater and [O2 observed] the observed oxygen concentration measured with the Oxygen Optode 4831. The [O2 solubility] represents the amount of oxygen that can dissolve in a given volume of seawater depending on temperature, pressure and salinity and is calculated with TEOS-10 (http://www.teos-10.org/, last access: 14 July 2025).

The glider data were mapped onto a regular grid with a vertical resolution of 0.5 m and onto a horizontal resolution of 1.5 km. To account for the spatial variability discussed in Sect.  and to minimize the effect of glider advection, distance was calculated relative to the surrounding current. From this point onward, this will be referred to as “distance”. After this correction, the cumulative distance was reduced to approximately half of the along-track distance.

This distance is derived using the horizontal speed of the glider obtained by the Glide Slope Model (GSM) and calculated by averaging the horizontal speed of two consecutive glider measurements and multiplying the result by the time interval, Δt. While Δt was nominally 5 s, occasional clock resets introduced variability. To maintain consistency, Δt was assumed to be 5 s for the calculations. The GSM-derived distance differed by 3.55 % from the distance obtained using the horizontal speed from the Hydrodynamical Model. Both time and glider positions were interpolated to the horizontal distance grid.

To identify surface-level events that may influence ventilation processes, we have analyzed the Sea Level Anomaly (SLA), from AVISO, with a 0.125° resolution and the Sea Surface Temperature obtained from the Global SST and Sea Ice Analysis, L4 OSTIA, 0.05° daily. Where the SLA is estimated by merging L3 along-track measurements from various altimeter missions.

Eddy kinetic energy (EKE) is calculated using the geostrophic velocities from the AVISO product. 7EKE=12(u′2+v′2), where u′ and v′ are the deviations from the mean (u‾,v‾) .

Ageostrophic motions at fronts can be indicated by an increased FSLE . The FSLE is defined as the inverse time of separation of two particles from their initial distance δ0 to a final distance δf . The particles are advected by altimetry-derived velocities and their trajectories are computed by forward-time λ+ and backward-time λ- integration of the altimetry velocities. Large timescales of separation of the particles λ- indicates an intense strain field . The separation's growth rate FSLE is defined as: 8λ+=1τln⁡δfδ0;9λ-=-1τln⁡δfδ0; where δ0 is the initial distance between a particle at (x,y,t) and its four closest neighbors. δ0 corresponds to the spatial resolution of the FSLE grid. δf is the final distance between particles. τ is the minimum time (among the 4 particle pairs) to reach the distance δf . FSLE is calculated following the method of , using τ = 7 d, δ0 = =0.05° (∼ 6 km) and δf = 0.5° (∼ 56 km). The initial separation δ0 is set close to the altimetry grid spacing (18°) and smaller than the regional first-baroclinic Rossby radius (∼ 25 km) to resolve meso- to submesoscale frontal features. The final separation δf is set so that δf = 10δ0 following the method of . This choice ensures that FSLE captures the growth of submesoscale frontal features into larger mesoscale structures, representing the overall strain field. The time integral τ is chosen to align with typical mesoscale mixing timescales observed in the Cape Basin . The chosen parameters represent the lower limits permitted by the resolution of altimetry, bringing the FSLE fields closer to glider-scale observations. Parameter sensitivity tests indicate that further reductions in τ or δ0 predominantly enhance noise rather than reveal additional coherent structures. Maximum values of the λ+ and λ- fields identify divergent and convergent flows, respectively . In particular, intersections of intense converging and diverging FSLE lines identify Lagrangian hyperbolic points, where particles and tracers are simultaneously being stretched along one direction and compressed along the other direction .

During the deployment, the glider navigated in the Cape Basin, north of the Agulhas Retroflection. The glider's trajectory was significantly influenced by strong currents in the vicinity of a cyclonic eddy located around 18° E and 38° S. In general, this region is also influenced by filaments and anticyclonic eddies (Agulhas Rings) shedding off from the Agulhas Current, introducing warm and saline Indian Ocean waters into the South Atlantic, resulting in elevated SST (Fig. a). Using water- mass classification, the glider's observations reveal that these Indian Ocean waters (absolute salinity Sa = 35.5 gkg-1; conservative temperature CT = 16 °C) are not confined to the surface layers and can be transported along isopyncals to depths of 250 m (Fig. ). In the subsurface (50–200 m), they interact with South Atlantic waters (Sa = 34.7 gkg-1, CT = 13 °C), highlighting the role of mesoscale eddies in stirring and redistributing heat and salinity. The strong differences in temperature (T) and salinity (S) between these water masses create strong horizontal T and S gradients, particularly between 550–620 and 700–750 km along the glider's track (Fig. a and b). These thermohaline gradients influence stratification and may further drive isopycnal stirring and shear-induced mixing in the subsurface layers.

At intermediate depths (500–700 m), the glider reveals waters of Subantarctic origin with characteristics of Antarctic Intermediate Water (AAIW, Sa < 34.6 gkg-1 ), identified by its salinity minimum. These waters originate north of the Subantarctic Front, where strong wintertime convection ventilates the ocean, allowing AAIW to equilibrate with the atmosphere and become enriched with oxygen before it subducts . Once subducted, this recently ventilated AAIW is advected into the Cape Basin, where it flows beneath the warmer, saltier Agulhas waters and interacts with Agulhas Rings and filaments, which trap and stir the AAIW water masses, and maintaining elevated oxygen levels at intermediate depths . The layering of different watermasses in the Cape Basin is evident in the glider observations (Fig. a and b), and when combined with mesoscale stirring, generates interleaving structures that can lead to enhanced lateral mixing . This lateral mixing, combined with mesoscale eddy activity, redistributes oxygen-rich water masses and facilitates the vertical exchange across density layers. This is evident in the glider AOU profiles (Fig. c), where the distribution of oxygen is closely associated with interleaving signals in the T and S properties. When ventilation is limited, AOU is expected to increase with depth, indicating prolonged isolation from the surface, during which oxygen is consumed by respiration and other biological processes . This pattern is not consistently observed across the entire glider section. For instance, at approximately 380, 600, and 780 km along the glider's track, low AOU values – that have recently been at the surface – are observed to extend from the surface down to 700 m depth, indicating a well-ventilated region (Fig. c, black dashed lines and arrows). As these locations align with low-salinity patches at intermediate depths (Fig. b), they likely correspond to recently ventilated AAIW. In contrast to this broader ventilation pattern, a more localized ventilation event is observed at approximately 80 km along the glider's track, where low AOU values (50 µmolkg-1) extend to 400 m depth over a horizontal distance of only 10 km. This suggests that the isopycnal transport of oxygen-rich waters at this location may result from a small-scale subsurface-intensified eddy or the glider is advected across a narrow filament of distinct water properties.

We contextualise the ventilation signatures in Fig.  by comparing the glider-measured lateral buoyancy gradients with altimetry-derived FSLE, which demonstrates the mesoscale circulation patterns in Fig. . The interaction between mesoscale eddies can drive horizontal strain, leading to the deformation of pre-existing buoyancy gradients and thereby intensifying fronts through frontogenesis . These fronts can become unstable, leading to stronger mixing and enhanced vertical motions. These processes are often associated with baroclinic instabilities, where density gradients between water masses drive vertical motions and intensify mixing responsible for the vertical transport of tracers . FSLE can be used as an indicator of the horizontal strain developing at fronts, with elevated values highlighting regions of intensified vertical motions and frontal instabilities.

Satellite altimetry observations reveal maximum diverging FSLE reaching over 2 d-1 (Fig. ) at 405, 465 and 620 km along the glider's track, indicative of strong frontal structures. At 405 km, this enhanced FSLE coincides with increased horizontal buoyancy gradients observed by the glider between 300–700 m (Fig. d), the depth range associated with ventilation and renewal of intermediate water . Localized FSLE peaks are observed to coincide with tracer subduction, as seen at 400 km along the glider's track, where pronounced salinity, temperature and AOU gradients indicate vertical transport of tracers reaching 700 m in depth (Fig. a–c). This aligns with the findings of , which demonstrate that enhanced frontal activity from localized strain fields can generate strong tracer gradients at depth.

The pronounced FSLE peaks do not always coincide with strong horizontal buoyancy gradients observed by the glider – for example at 465 km in distance (Fig. d). Although the FSLE parameters were chosen at the lower limits permitted by altimetry to better approach glider scales (Sect. ), the temporal and spatial resolution of the FSLE field remain coarser than the glider resolution. As a result, it may not fully resolve sharper, short-term frontal structures observed by the glider or capture immediate short-term surface dynamics. Additionally, surface dynamics may take time to propagate to depth, creating a temporal lag in the events visible at depth. Furthermore, the horizontal buoyancy gradients derived from the glider are likely underestimated, as the glider is largely advected with the background current and does not always cross ocean fronts perpendicularly. A full quantification of cross-frontal gradients requires sampling orthogonal to the front, and oblique sampling can lead to an underestimation of up to 50 %–70 % . To partially account for this effect, the distance covered by the glider is adjusted using the Glider Slope Model, as described in Sect. . Nevertheless the current calculation does not account for the full cross-frontal buoyancy gradient, potentially leading to an underestimation of its magnitude in regions where fronts are sampled at an oblique angle.

Assessing the regional significance of these localized processes can be done by considering the broader circulation of the Cape Basin. The highest strain fields are observed at the Agulhas Retroflection, indicating strong mesoscale activity in this region (Fig. ). The strong and highly variable strain field in the region of 38° S and 18° E coincides with elevated EKE, indicative of intensified mesoscale turbulence. This suggests that mesoscale instabilities play a key role in driving transport to depth across the Agulhas Retroflection, highlighting how localized strain fields, such as those observed along the glider's track, contribute to larger-scale ventilation and tracer subduction pathways across the Cape Basin.

Vertical velocities were derived from the Omega equation , using boundary conditions following . Methods and results are provided in the Supplementary information. Strong vertical velocities O(100 md-1) are generally found in regions with elevated horizontal buoyancy gradients and high Rossby numbers, suggesting the role of submesoscale eddies in tracer subduction (Fig. S1 in the Supplement). These enhanced vertical velocities often coincide with elevated FSLE and low AOU at depths, however this alignment is not consistent throughout the glider section. The glider observations were often orientated along-front rather than cross-front, limiting the accuracy of derivatives such as ∂ug/∂x and ∂b/∂x. Consequently, the calculated vertical velocities are subject to large uncertainties, and the available estimates should be considered indicative rather than quantitative.

To understand the role of meso- to submesoscale processes on the vertical and horizontal transport of tracers, we focus on a specific subset of the glider transect, spanning 600 to 720 km (yearday 126 to 136) along the glider's track. This period of the glider's trajectory was characterized by intense horizontal buoyancy gradients or fronts, where enhanced subsurface ventilation of AOU was observed. The vertical velocities in Fig. S2 in the Supplement, show an alternating pattern indicative of a secondary circulation driven by frontogenesis, with the strongest downwelling roughly aligned with a strong POC subduction event near 650 km in distance along the glider's track (Fig. ). Within this transect, the MLD exhibits more variability (Fig. ), deepening initially, then shoaling, and deepening again. The lower rates of stratification within the pycnocline beneath the MLD (< 50 m) between 600 and 630 km along the glider's track (Fig. a) provides favorable conditions for the exchange of tracers between the surface boundary layer to the deeper ocean interior. Accordingly, the glider observations in Fig.  reveal that AOU, backscatter slope (γbbp), and POC are primarily confined to the mixed layer, yet between 630 to 670 km, diapycnal exchange occurs to about 500 m depth, where they are then advected along sloping isopycnals between 630 and 710 km to reach deeper waters.

To explain the diapycnal exchange from the mixed layer to deeper layers, we examine a stratified (3 × 10⁻⁴ s-2) layer between 630 and 660 km along the glider's track at 80 m depth that is surrounded with elevated vertical shear (Fig. b), suggesting that this region could be prone to growing shear instabilities. This is reinforced by a peak in FSLE at 630 km (Fig. d) and yearday 128 (Fig. ), indicating the presence of an ocean front, that promotes enhanced mixing . Synonymously, Richardson numbers below 10 (red contour in Fig. c) suggest a transitional regime where stratification weakens, allowing shear instabilities to grow, that could be responsible for driving cross-isopycnal mixing .

The downward mixing across isopycnal is supported by the presence of elevated spiciness values observed between 630 and 660 km along the glider's track in Fig. c, indicative of warm, salty water, that is transported across the pycnocline spanning density layers 25.7 to 26.5 kgm-3 . AOU exhibits a similar pattern between 630 and 660 km, suggesting enhanced turbulence and vertical mixing between density layers 25.7 to 26.5 kgm-3. The observed elevated vertical shear in the presence of the ocean front at 630 and 660 km along the glider's track suggests conditions favorable for generating turbulent mixing of POC across the pycnocline, further enhancing the redistribution of tracers and particles in the upper 400 m (Fig. ). A reduction in POC within this generally high POC regime (26 < σ < 26.5 kgm-3) suggests that POC is transported across isopycnals. Once below this shear-influenced region, POC can sink more freely or be redistributed along density surfaces, as demonstrated along the 27.0 kgm-3 isopycnal (Fig. , blue thick contour). Particles that pass through the pycnocline are eventually trapped by a secondary stratified layer at 400 m depth, corresponding to a density of 27.0 kgm-3, which acts as a barrier to mixing, limiting further vertical transport, leading to localized accumulation of POC between 80 and 400 m depth (Fig. ).

The cross-isopycnal mixing enhances exchange between the surface boundary layer and the ocean interior between 630 and 660 km along the glider's path, providing a pathway for particles to move from the surface towards 400 m depth. Elevated AOU values observed below the pycnocline at 80 m depth indicate active remineralization of sinking POC, as microbial decomposition depletes oxygen in subducted waters . Upward doming isopycnals between 630–660 km indicate upwelling, which could potentially bring nutrients to the surface layer, stimulating photosynthesis and biological activity. However, elevated chlorophyll levels were not observed at the surface, nor was chlorophyll detected at depths below 80 m. At the same time, POC concentrations remain elevated at depth but are lower in the mixed layer compared to surrounding regions (Fig. b), suggesting an enhanced export of POC.

Beyond 660 km along the glider's path, the enhanced geostrophic velocity – which extends below 500 m depth (Fig. d) – facilitates isopycnal stirring, which contributes to mixing along the density surfaces. As a result, POC is efficiently advected along isopycnals, reaching 800 m in depth (Fig. c).

As discussed in Sect. , the horizontal buoyancy gradients derived from the glider may be underestimated due to its oblique sampling of ocean fronts. Consequently, the calculated buoyancy gradients, which is used to derive vertical shear and the Richardson number are underestimated. Therefore, the shear instabilities are likely larger than those represented and may have a greater impact on the cross-isopycnal transport than is apparent from the observations.

Satellite-derived daily average of SLA, SST and FSLE, between 6 and 16 May 2023 in Fig. , combined with glider observations from Fig. , reveal the important role of small-scale flow structures and fronts in driving vertical tracer transport. Around 37.5° S (and between 600–630 km), the glider follows a small anticyclonic trajectory, steered by the background depth-averaged current, as indicated by the red arrows in Fig. a and d. This small-scale variability in the flow structure is not captured in the satellite images of SLA, due to the low resolution of the AVISO product. However, high variability in temperature and salinity sections suggests that this is a highly dynamic region that can facilitate the interaction and mixing of different water masses. Additionally, elevated EKE and FSLE values were observed in this region (Fig. ), indicative of enhanced vertical and lateral transport of tracers, including AOU to greater depths . At intermediate depths (300–700 m), low mean AOU values of 80 µmolkg-1 were observed, supporting the dynamic interaction that drives vertical transport of surface waters to depth. The glider traverses a high diverging FSLE field at 37.5° S (Fig. c), making this area prone to frontogenesis . From 600–630 km along the glider's track, the glider is passing through a region with significant SST variability (Fig. e) and sharp SST gradients of approximately 0.4 °Ckm-1, as indicated by the dashed contour lines in (Fig. b and e). These surface gradients also extend to depth (Fig. a). These subsurface gradients are likely to enhance vertical transport of tracers by driving localized turbulence, facilitating mixing across density layers . This process is evident in our observations, where pronounced cross-isopycnal transport occurs near these strong fronts (Fig. ).

Continuing the trajectory from 630 to 660 km along the glider's path, the glider enters a transitional regime where we find elevated AOU values of 120 µmolkg-1 averaged between intermediate depths (300–700 m) in Fig. . The glider samples across sharp SST gradients of approximately 0.6,°Ckm-1, as indicated by the dashed contour lines in (Fig. b), crossing the edges of small-scale cyclones and anticyclones. The path begins at the edge of a strong cyclone and progresses northward, eventually reaching the edge of a smaller secondary cyclone that recently split from the larger cyclone south of 37.5° S. This cross-structure sampling captures vertical transport of tracers across density layers 25.7 to 26.5 kgm-3 (Fig. ), consistent with shear instabilities driving localized diapycnal transport (Fig. ). In this regime AOU is accumulated at depths near the stratified layer, where oxygen consumption continues, amplifying the processes of respiration and re-mineralization of organic matter.

In contrast, beyond 660 km, the glider primarily follows the edge of a mesoscale eddy and moves along the front. In this regime, vertical tracer transport occurs primarily along tilted isopycnals, rather than through across-isopycnal mixing. The glider is following a path along the edge of the secondary cyclone that is aligned with sharp SST gradients of approximately 0.6,°Ckm-1, as indicated by the dashed contour lines in (Fig. b). Strong geostrophic velocities and elevated depth-averaged current beyond 660 km suggest the glider experiences a transition from a dynamic, high-energy environment of submesoscale turbulence to a more structured, geostrophic regime dominated by advection along density surfaces.

The ventilation observed in this study is found to be driven most likely by a combination of shear instabilities, lateral advection and submesoscale fronts (Fig. ).

Shear instabilities emerge as a significant contributor to vertical mixing, driving diapycnal transport (Fig. ). In these regions, Richardson numbers approach zero (Fig. ), and show a reduction in upper ocean stratification. This dynamical regime suggests conditions favorable for enhanced cross-isopycnal mixing and vertical transport, consistent with previous findings on shear instabilities observed in the South China Sea . Additionally, our observations support the model results of in the Cape Basin, confirming that shear instabilities are crucial for vertical transport of tracers in this region.

Lateral advection plays a crucial role in redistributing tracers along isopycnals. Our observations demonstrate how mesoscale eddies facilitate the transport of AOU and POC along isopycnals. This aligns with previous findings, which highlight how mesoscale eddies and standing meanders enhance submesoscale variability, driving a cascade of energy and momentum that strengthens vertical fluxes and tracer exchange . The eddy driven subduction of POC and AOU is concentrated along the edges of these mesoscale eddies, consistent with observations in the North Atlantic , where dynamic eddy flow fields subduct surface waters rich in non-sinking POC from eddy peripheries into the ocean interior.

In addition, another key mechanism evident in Fig.  is the role of fronts in cross-isopycnal transport. These fronts, characterized by strong buoyancy, temperature, and salinity gradients, enhance the vertical flux of tracers. Our observations suggest that the combination of elevated EKE and high buoyancy gradients at these fronts lead to enhanced vertical motions and stronger mixing, enabling AOU to be transported to depth. This observation aligns with previous studies , which emphasize the role of submesoscale features in facilitating vertical exchange. The alignment of high FSLE values with strong mixing signatures underscores the importance of surface strain fields in enhancing subduction.

In some instances, the glider may cross into a different water mass, making it difficult to precisely locate the source of the ventilated waters. Localized features, such as the subsurface eddy described in Sect. , can trap and transport recently ventilated waters to depth over small horizontal scales. While this event reflects ventilation associated with the eddy, it is also possible that these waters were ventilated through surface processes in a neighboring region and subsequently advected into the observed area. Therefore, while the vertical transport of tracers appears to be influenced by local mixing and submesoscale dynamics, it is important to consider the possibility that these processes are linked to larger-scale advection and ventilation occurring outside the glider's immediate path.

The findings of this study extend beyond local oceanographic dynamics, with the redistribution of heat and carbon impacting both the regional thermohaline structure and potentially large-scale ocean circulation. We show that waters stirred by the oceanic circulation of the Agulhas Current can facilitate the interaction and mixing of water masses that originate from the Indian, Atlantic and Southern Oceans. As ventilated waters are advected downstream along isopycnals, the cumulative anomalies of high POC and low AOU are gradually reduced by remineralization and respiration. While this process is taking place, they can interact with surrounding water masses and contribute to the transport of heat, salt, and tracers toward other parts of the Atlantic, potentially influencing intermediate-depth circulation and branches of the AMOC . This exchange contributes to the transport of heat and salt from the Southern Hemisphere to other parts of the Atlantic, and potentially influencing the strength and variability of the AMOC . Meso- to submesoscale eddies and fronts are found to enhance lateral and vertical fluxes, driving the downward transport of warm, carbon-rich (high POC) surface waters into the ocean interior. Regions of consistently high FSLE, as shown in Fig. c, coincide with elevated EKE and mark persistent hotspots where strong mesoscale stirring and fronts are likely to subduct and ventilate waters, highlighting this particular area (the Cape Basin and particularly just west of the Agulhas Retroflection) as a region of potentially enhanced vertical transport. This has direct consequences for ocean-atmosphere interactions, given that the vertical supplies of heat and carbon can alter air–sea fluxes .

Similarly, the downward flux of carbon via subduction and subsequent remineralization at depth impacts oceanic carbon sequestration and oxygen consumption, regulating the biological pump and overall carbon cycling. Given that current estimates suggest the biological carbon pump is underestimated , a more comprehensive understanding of the fine-scale physical processes driving carbon sequestration, such as those resolved by the glider survey, is needed for closing carbon budgets. Integrating such observational data with high-resolution models will improve estimates of carbon fluxes and enhance predictions of future carbon sequestration in response to climate change. Although this study focuses on the Cape Basin, similar processes might occur in other regions with strong frontal dynamics, such as the Brazil–Malvinas Confluence. These areas remain the focus of active research. Integrating observational datasets with high-resolution models and emerging satellite missions such as SWOT might provide a more holistic view of ventilation and carbon fluxes across the South Atlantic.