The HCS is characterized by its large extension along the Peruvian and Chilean coast (5–45°S), with several coastal upwelling centers driven by along shore winds, and seasonality that's partially associated with the oscillation of the SPA (Strub et al., 1998). Equatorial Subsurface Water (ESSW), transported by the poleward subsurface Peru-Chile Undercurrent (PCUC), is characterized by a pronounced DO deficit and high nutrient levels, being the main source of water for the wind-driven coastal upwelling (Wooster and Gilmartin, 1961; Fuenzalida et al., 2009). South of 28°S, the surface of the equatorward Humboldt Current is fed mainly by Subantarctic waters (SAAW), which, by contrast, has relatively low salinity and nutrients and high oxygenation (Silva et al., 2009; Llanillo et al., 2012). In addition, the HCS is among the most productive marine ecosystems worldwide, with annual mean net primary production rates of 1.1 kg C m⁻² yr⁻¹ (Daneri et al., 2000; Testa et al., 2018) which sustains the productive pelagic fisheries in Chile (Cubillos et al., 1998). Furthermore, high availability of OM stimulates its mineralization in bottom waters and sediment leading to DO consumption (Farías et al., 2015).

A time, series station at a fixed location (36°308′ S, 73°07.75′ W) off central Chile with 92 m depth (Figure 1), discrete samples at different depths (2, 5, 10, 15, 20, 30, 50, 65, 80 m) were taken and continuous profiles were recorded from 1997 to present. This time series station is called station 18 (hereafter TST18) due to its distance from the coast being 18 nautical miles. The TST18 has been visited monthly since 2002 and quarterly from 1997 to 2002. A CTD device was used (conductivity, temperature and depth), model SBE-25 and SBE 19plus with various sensors to measure DO, photosynthetically active radiation (PAR) and fluorescence. Calibrations for CTD sensors were annually performed by the manufacturer (Sea-Bird) and CTD data were processed using SeaBird data processing software. In addition, post-processing conductivity and DO validation was done using in-situ salinity (salinometer AUTOSAL model 8400B) and DO measurements. Discrete seawater sampling was conducted using a rosette sampler, mounted with 10-L Niskin bottles. Samples of DO (in 125 ml iodimetric bottles, immediately fixed with Winkler reagents) were analyzed by the Winkler method using an automatic AULOX measurement system.

The continuous high-frequency oceanographic and meteorological data used in this research was recorded by a POSAR buoy, an automated monitoring meteo-oceanographic buoy located close to Coliumo Bay (Chile) about 10 km from the mouth of the Itata River 144 (36.4°S–72.9°W), at a depth of about 50 m (Figure 1) and close to the TSS18. This POSAR buoy was operative during two consecutive upwelling seasons (austral spring-summer). POSAR data is freely available in real-time at the CR2 (www.cr2.cl/posar) and CDOM web portals (www.cdom.cl). High frequency DO, temperature and salinity measurements came from a Microcat CTD-DO (SBE 37-SMP-ODO) sensor located at 10 m depth, transmitting high frequency and real time oceanographic data. While wind measurements came from an ultrasonic anemometer (MM86106). Data analysis and interpretation for the DO sensor and others at 1 m depth and technical details of the sensors/instruments forming the POSAR buoy are available in Aguirre et al. (2021a) and in www.cr2.cl/posar.

In this study, the analysis of DO variability is based on profiles obtained by the CTDO; discrete Winckler DO measurements were used to validate CTDO, and there was a significant positive correlation between both types of measurements. Wind speed and direction based on an hourly register were obtained from a permanent meteorological station located at Carriel Sur (http://www.meteochile.gob.cl/) close to the continental shelf off central Chile (Figure 1). This station meets international standards and it is established as a coastal station. Meteorological data (daily average) were compared with daily satellite wind fields, averaged every 6 h and interpolated linearly from 0.25° to 0.05°. Two products related to satellite observations were sourced from Copernicus marine service (https://resources.marine.copernicus.eu/). The wind speeds taken from satellite data significantly correlated with those from the weather station, however the wind speed records from satellite data were 30% higher than those from the weather station (Faundez, unpublished data); thus, for conservative criteria, the data from Carrier sur weather station was used.

Alongshore or equatorward wind component (v = m s⁻¹) was calculated from the wind magnitude (|V| = m s⁻¹) and direction (dd) converted to radians, as follows; v = − V ∗ cos ⁡   d d ∗ π 180

The upwelling index (UI; m⁻³ s⁻¹) from 2002 to 2021 was obtained through Ekman zonal transport per 1,000 m of coastline (Bakun, 1973), as: U I = τ y ρ f ∗ 1000   m

where τ _y (Pa; kg m⁻¹ s⁻²) is the mean wind stress meridional component within box 73–74ºW, 36–39ºS, ρ represents the mean density of the water column at TST18 (1,026.2 kg m⁻³), and f indicates the Coriolis parameter corresponding to the latitude at ST18 (8.67 *10^–5 s⁻¹). Daily stress ( τ y ; N m⁻²) was calculated following Nelson (1977): τ y = C d · ρ a i r · V · v

where C d is the empirical static drag coefficient, corresponding to 0.0013 (Kraus et al., 1994), ρ a i r is the air density that corresponds to 0.00122 g cm⁻³ or 1.22 kg m⁻³, V is the wind magnitude in m s⁻¹, and v is the equatorward component of the wind in m s⁻¹, A 5-day running mean was used to evaluate the cumulative alongshore wind stress, according to Barth and Menge, (2007). Hypoxic volume (HV) was calculated as the hypoxic portion of an ideal cylinder of radius r=100 m and 80 m depth, around the TST18 as follows: H v o l = 10 0 2 π 80 − Z h y p o x i c   i s o l i n e

In this study a hypoxia event is defined as DO levels reaching below 89 μmol L⁻¹ for two or more days of duration. It is based on an analysis of a hypoxic event at the Gulf of Trieste (Stachowitsch, 1991, 1984), where 2 days of hypoxic conditions could cause a reduction in metazoan biomass to less than 25% of the initial conditions; other physiological effects are pointed out by Diaz and Rosenberg (1995) at those DO concentrations; this value serves as a rough reference as to how different taxa respond and adapt to hypoxic conditions.

Mean annual cycles (referred to as climatology) of the variables at TST18 were calculated using monthly averages across the time series. To quantify the deviation from the mean annual cycle, the former was subtracted from the observed data for each corresponding month of the cycle, thus obtaining the anomaly, defined as the difference in measuring units from the long-term average annual cycle.

Regarding interannual variability, ENSO and oceanic episodes of El Niño (EN) and La Niña (LN) were detected using two indexes, the Oceanic Niño Index (ONI) which is based on the temperature anomaly along the Central Equatorial Pacific Ocean (3.4 region 5°N–5°S, 120°–170°W) and the coastal El Niño index (ICEN) which is based on the temperature anomaly off the coast of Ecuador and Perú (1 + 2 region 90°−80°W, 10°S–0°). In both cases, low negative and positive values indicate an LN and EN episodes, respectively, (Takahashi et al., 2014). The data was obtained from https://origin.cpc.ncep.noaa.gov/products/analysis_monitoring/ensostuff/ONI_v5.php for ONI and from http://met.igp.gob.pe/datos/icen.txt. For ICEN.

Monthly time series for anomalies of temperature, salinity and DO were used to determine the decadal trends using a least squares linear model fit over 6 month averages of the original time series. Trends were grouped by pre-defined averages over depth layers including surface-layer (0–10 m), lower-surface (15–20 m), mid-layer (30–50 m), and deep-layer (65–80 m). Correlations between DO and wind stress anomalies were performed at different phase lags (0–4 days) to determine the best correlation between wind and the oceanographic variables.

High-frequency measurements of temperature, salinity and DO from the POSAR buoy were analyzed to quantify the variability occurring at the daily to weekly time scales (from now on referred to as high-frequency) and to quantify error and lost data in observed patterns when compared to a monthly sampling frequency at TST18. Pearson correlations between DO and wind stress were performed at different phase lags to estimate the optimum high frequency correlation of these two variables and to assess the effects of upwelling on surface DO levels.

Throughout the study period, the weather station recorded daily wind speeds of between 0.6 ± 0.7 m s⁻¹ (means±SD) and 13.3 ± 1.6 m s⁻¹ (means±SD), while the alongshore (northward or equatorward) component of wind speed fluctuated from −12.8 to 10.7 m s⁻¹. Figure 2A shows the meridional component of daily wind stress from 2001 to 2021, which ranges between −0.30 and 0.19 N m⁻². These winds are stronger and more frequent during austral spring-summer peaking around December-January, whereas a weakening or even inversion of the predominant winds occurs during late autumn-winter. In the latter period, wind stress is more variable than in the upwelling seasons (Figures 2A, B).

The annual cycle of wind reveals the predominance of upwelling favorable winds from August to April (70% of the year) with a maximum monthly mean of 0.022 N m⁻² in January; (Figure 2B). By contrast, between May and July, negative or close to zero stress values (−0.0014 N m⁻² July) are dominant. Cumulative upwelling-favorable wind stress shown in Figure 2C Indicates a rapid accumulation from spring to summer, which varies between different years, mainly in winter and springtime (Figure 2C). The climatology of the UI ranges from 25.15 m² s⁻¹ in April to 249.66 m² s⁻¹ in January.

On the continental shelf off central Chile, coastal upwelling events influence the vertical distribution of the physical and chemical properties in seawater, as shown in the vertical sections of temperature, salinity, DO (Figure 3), Upwelling of subsurface waters is demonstrated by the shoaling of 12°C isotherm in spring (Figure 3A), reaching its shallowest depth during mid-summer (<10 m depth in January). Salinity also shows seasonal patterns that correlate with the advective shoaling of isotherms. Low salinity is observed in a narrow surface layer (0–15 m) during winter and spring, which is associated with the Itata River discharge (plume) that reaches TST18, creating a notable haline stratification.

DO distribution in the water column (Figure 3C) presents higher concentrations (> 260 μmol L⁻¹) in the upper (0–15/20 m depth) layer, however, subsurface and bottom layers exhibit strong seasonal variation due to the influence of the poorly oxygenated ESSW from coastal upwelling. During the upwelling season, as much as 87% and 81% of the water column exhibits DO concentrations under 89 and 22 μmol L⁻¹, respectively. The average seasonal shoaling of hypoxic waters is strongly correlated with the UI (r = 0.81, p<0.05, n = 12).

DO levels rapidly decline with the spring onset (around September) when upwelling-favorable winds increase (Figure 2B) and poor oxygen (< 44 μmol L⁻¹) and nutrient rich ESSW is transported onto the continental shelf (Figures 3A, 4). During November-March, cumulative upwelling favourable wind stress has reached over half of its annual maximum (Figures 2A–C), the average DO in the water column reaches 63%–72% or 41%–48% volume, with hypoxia and severe hypoxia, respectively. On the contrary, DO levels increase in April/May as upwelling-favorable wind weakens prior to the autumn transition. During winter, the persistence of poleward winds lead to downwelling of oxygenated SAAW (Silva et al., 2009; Llanillo et al., 2012) and storms (Strub et al., 1998) provoke an oxygenation of the water column on the continental shelf (Figure 2A) through advective flushing and vertical mixing (Sobarzo et al., 2007). Maximum oxygenation is reached in July, where an average of 26% (2%) of the water column has DO concentrations under 89 (22) μmol L⁻¹ (Figures 3C, 4K).

Density, expressed as sigma-t, describes water mass structure and dynamics (Supplementary Figure S2). Density is negatively correlated with DO (r = −0.59; p<0.05, n = 2,209), which means that denser waters contain less DO. The water column at the sampling location consists of nutrient poor and oxygenated SAAW with a density (sigma-t) of approximately 25.6. The SAAW moves northwards through the HCS, whereas the ESSW, which forms at the equator is nutrient rich and DO poor, having a density (sigma-t) of approximately 26.2 and moves poleward by the PCUC (Silva and Neshyba, 1979; Silva et al., 2009). The DO content of the ESSW can be identified by projecting the 26.2 isopycnal over the DO time series (Supplementary Figure S1), showing a clear association of this water mass with hypoxic waters.

Figure 4 shows the climatological monthly hydrographic profiles from TST18. Temperature transitioned from almost an entirely mixed water column in late fall/winter to a strong thermocline between 10 and 20 m depth in spring/summer (Figure 4A). Salinity presents negligible depth differences in summer, except for a noticeable shallow (0–15 m depth) gradient during winter and spring (Figure 4C). DO climatological profiles capture the seasonal development of the vertical distribution of hypoxia and severe hypoxia, along with temperature and salinity (Figure 4C). The thermocline and oxycline shoal simultaneously in spring-summer. Vertical temperature gradients intensify between December to March, when a thin surface layer becomes warmer (with values up to 15°C), caused by increased radiative heating during this season. This is contrasted with the upwelling of cool subsurface water from the ESSW, creating a marked thermal stratification. By contrast, low salinity waters are detected in a thin surface layer between June to October, which coincides with high precipitation and river runoff (Figures 4G, H). Moreover, a low salinity area is associated with the Itata River plume which creates notable haline stratification in the winter (Saldía et al., 2012; Masotti et al., 2018) but does not affect surface DO concentration in wintertime nor the decadal trends presented below.

DO concentration consistently decreases with depth; with maximum values in the mixed layer (> 220 μmol L⁻¹), reaching values below hypoxia and severe hypoxia at the subsurface and bottom layers throughout the year. However, vertical gradients and differences between the surface and the bottom layers vary between unfavorable and favorable upwelling periods. High DO values throughout the entire water column (Figures 4I–L) correspond to periods of strong vertical mixing associated with winter storms, and dominance of the SAAW. On the contrary, during the spring and summer, the oxycline intensifies and becomes shallower as coastal upwelling progresses (September to March). This seasonal variability in DO levels agrees in part with Pizarro-Koch et al. (2019), describing model derived estimates with a minimum (maximum) hypoxic volume in winter (Summer), attributed to the advection of offshore waters depleted in DO associated with the ESSW and transported by the PCUC. This is supported by observations from TST18, as shown in Supplementary Figure S2, where the boundary of hypoxic waters is associated with the isopycnal (sigma-t = 26.2) in the ESSW. Also, low DO may be caused by DO consumption through the oxidation of OM, which is exported from the surface and accumulated close to and in the sediment throughout the upwelling period (Farías et al., 2004).

The development of hypoxia and severe hypoxia displays high-frequency, intra-seasonal and interannual variability (Figure 3C). High-frequency variability is superimposed on the seasonally varying winds, caused by the eastward passage of cyclonic storms (Rahn and Garreaud, 2013) and poleward propagating coastal lows (Garreaud et al., 2002; Garreaud and Rutllant, 2003; Aguirre et al., 2021b). Variability with 3–12 days fluctuations lead to active and less-active upwelling events (Aguirre et al., 2014; 2021a). Other intraseasonal processes affecting the water column include coastally trapped waves and other phenomena, such as eddies (Hormazabal et al., 2002; Pizarro et al., 2002). Annual variability of most oceanographic and biogeochemical variables is attributed to the ENSO cycle, which is the primary process that induces interannual variability in the HCS (Hormazabal et al., 2002, 2006).

Previous long-term studies on the PCUC along the continental slope (30°S) state that coastal trapped waves control intraseasonal variability in the OMZ, whereas Rossby waves provide a physical mechanism that lead to significant fraction of seasonal (annual cycle) DO variability off central Chile. Both physical mechanisms originate from equatorial wind forcing (Hormazabal et al., 2002; Pizarro et al., 2002). However, this study doesn't consider how different types of wave interactions are affected by local forcing, such as local winds, river discharges, topographic and geomorphological features and pelagic benthic coupling mechanisms, among others.

Understanding the DO regime and its long term response in different systems, helps to identify the key processes that it is controlled by, providing the knowledge base to predict potential changes under future climate change scenarios. Table 1 presents the DO decadal trends in different ocean depth layers. Significant negative DO trends are observed in the lower surface and mid layers, of −18.1 and −21.97 μmol L⁻¹ decade⁻¹ respectively, with a maximum rate of change in the lower surface layer at the location of the oxycline (just below the pycnocline). Also, the surface and bottom layers showed trends towards oxygen depletion, however not significant. The global average decrease in DO for the upper ocean (100–1,000 m depth) is estimated at −0.93 ± 0.23 μmol L⁻¹ y⁻¹ and DO depletion is higher in the mid-latitudes of both hemispheres (Helm et al., 2011). Approximately 15% of the global DO decrease is due to the reduced capacity for DO storage in a warmer mixed layer (Helm et al., 2011). However, this is not the case in central Chile where negative decadal trends in temperature (cooling) ranging from −0.09 to −.29°C decade⁻¹ are estimated and these rates of change increase towards the surface (Figure 8A). Also, the strongest reduction in DO is registered below the surface layer, and only the lower-surface and mid layer trends are statistically significant (Figure 8C). In addition, non-significant positive trends in salinity, between 0.01 and 0.04 psu decade⁻¹ occur throughout the water column (Figure 8B). These trends indicate that the over-accentuated decadal variation in the surface to mid layers with respect to the bottom layer (almost completely occupied by the ESSW) are probably not dominated by a change in the ESSW properties, but due to the shoaling of isotherms/isopycnals (i.e., increased vertical advection towards the (sub)surface layer) as equatorward winds strengthen and become more frequent shifting average conditions towards a more shallow ESSW. It is important to note that the decadal decrease in DO over several depths is not related to reduced DO solubility, but it is likely to be associated with changes in local wind forcing (see review Pitcher et al., 2021). This is further supported by the long term data of DO levels. Below 89 and 22 μmol L⁻¹, indicating a close relationship with UI, as shown in Figure 7.

Pitcher et al. (2021) provides a synthesis of drivers for DO reduction trends in coastal systems. In most cases, a decadal negative trend (deoxygenation) occurs in systems such as the California current system from Canada (Vancouver) to southern California (ranging from 8.3 to 21 μmol L⁻¹ decade ⁻¹) and coastal systems off Peru (ranging from 1.62 to 10 μmol L⁻¹ decade ⁻¹), whereas no trends are observed in the Canary upwelling system. In the last 60 years, shelf deoxygenation trends are significantly higher than the global ocean deoxygenation trend (i.e., 0.075 μmol kg^–1 yr⁻¹ or ∼4.5 μmol kg⁻¹) (Grégoire et al., 2021).

Climate model projections estimate that there will be considerable reductions in the open ocean DO inventory in response to anthropogenic climate forcing (Schmidtko et al., 2017; Oschlies, 2021), these projections also coincide with observed DO (Helm et al., 2011). However, the drivers of shelf hypoxia remain unresolved. The main driver of DO variability is wind driven upwelling, which is predicted to increase under climate change scenarios (e.g., Echevin et al., 2012; Sydeman et al., 2014; Bakun et al., 2015; Oyarzún and Brierley, 2019; Winckler-Grez et al., 2020). Observational evidence from this study supports these predictions, as significant negative trends in DO and temperature, and positive trends in salinity see (Table 1), indicates changes in the depth of waters depleted in DO. On average, the annual UI is four times higher, from 55.9 to 214.1 m³ s⁻¹ (Figure 7A) while the hypoxic volume became 1.2 times higher (Figure 7B) in the same period. Yearly averages since 1997 show a clearer trend of the hypoxic volume, changing by a factor of 2.3 from 1997. A similar trend is also estimated for the volume of severe hypoxia (Supplementary Figure S1) supporting the hypothesis that climate-dependent strengthening of upwelling wind forcing increases the intensity and spatial extent of hypoxia in coastal upwelling systems. The simultaneous increase in hypoxic volume and UI is probably due to the increase in upwelling intensity and/or duration of the upwelling favorable season, as described by Abrahams et al. (2021). Additionally, north of 30°S there is an intensification of winds due to increased land-sea temperature gradients, known as the “Bakun effect” (Bakun, 1990), reported by Weidberg et al. (2020).

In large EBUS, a general increase in upwelling favorable winds has led to a reported decline in SST (Gutiérrez et al., 2011; Sydeman et al., 2014; Abrahams et al., 2021), which is the case of the HCS (Aravena et al., 2014; Schneider et al., 2017; Aguirre et al., 2018). This is related to the well documented southward expansion of the SPA (Sydeman et al., 2014; Schneider et al., 2017; Aguirre et al., 2018; Winckler-Grez et al., 2020) and the more frequent occurrence of migratory anticyclones passing through the coast of central-southern Chile (Aguirre et al., 2021b). These trends agree with previous studies reporting positive trends in chlorophyll along upwelling regions (Gregg et al., 2005; Gutiérrez et al., 2011; Marrari et al., 2017), including in coastal upwelling off central Chile, as shown by Aguirre et al. (2018) and Winckler-Grez et al. (2020). Such trends may be explained by increased upwelling-favorable winds that pump nutrients toward the surface to promote the growth of phytoplankton and increase biomass, as reported in Aguirre et al. (2018) and Weidberg et al. (2020) but only north of 30°S. An increase in the phytoplankton biomass or OM could have an implication in the local consumption of DO, which requires further investigation.

DO dynamics, including hypoxic volume and intensity are highly responsive to a variety if parameters, including changes in source water (Emerson et al., 2004; Whitney et al., 2007, Espinoza-Morriberon et al., 2021), eutrophication (Howarth et al., 2011) and upwelling wind forcing (see previous references). Also, variability in ENSO, with increased frequency and intensity of extreme ENSO events during the last century (Gergis and Fowler, 2009; Cai et al., 2015; Wang et al., 2019), may interact with the regional changes in wind patterns and water mass distributions. It is unlikely that a change in DO consumption off central Chile is due to eutrophication in the coastal system as rivers have a moderate or low N and P load (Masotti et al., 2018). This contrasts to the Gulf of Mexico or the continental shelf off India (Breitburg, et al., 2018; Naqvi, 2021), but it remains unclear if changes in water masses at mid latitudes, mainly the ESSW, are responsible for the observed deoxygenation.

Finally, decadal variability associated with the Pacific Decadal Oscillation (PDO) is also a potential driver in the observed DO variability, which also impacts the seasonal migrations and intensification of the offshore SPA and its frictional interactions with the continent. A reconstruction of DO concentrations over the previous century (Srain et al., 2015) shows an interdecadal variability of >30 years periods associated with a negative (positive) phase of PDO, that favors (hinders) the development of wind driven upwelling, which in turn enhances (reduces) primary productivity and export.

Currently, climate change risk assessments are carried out for social, ecological and productive sectors (IPCC, 2022). Off central Chile, the observed reduction in DO over time could be considered as a climate-related hazard having an impact and risk on biological resources, ecosystems and coastal communities. In addition, wind-forced upwelling can propel hypoxic water onto the littoral zone (less than 20 m depth) and even into shallow water bays (e.g., Concepción Bay, Coliumo Bay, Arauco Gulf). This is due to the shoaling of oxygen isolines, including the 89 and 22 µmol L⁻¹, frequently observed in coastal upwelling zones (see Supplementary Figure S3) Thus, hypoxia events on the inner-shelf may become more frequent and severe, increasing exposure and sensitivity, therefore the risk of benthic organisms (that live on the littoral area less than 20 m depth) or even pelagic organisms to be more exposed, leading to more vulnerable coastal communities that depend on these resources.

Hernández and Tapia (2021) indicate that upwelling induced hypoxic events on nearshore subtidal systems are commonplace, but also significantly variable between locations; with half of the sites in their study showing mean DO values at the hypoxic threshold. In fact, several historical massive fish and crustacea mortality events were recorded in the study area (Hernandez-Miranda et al., 2010; Hernandez-Miranda et al., 2012). Recently, an analysis from reports and monitoring by the National Fisheries and Aquaculture Service (SERNAPESCA) over the last 15 years (2006–2021) has suggested that most massive fish mortally events occur during events of dominant south or southwest winds (i.e. with a significant equatorward component), These mass mortality events have intensified on the central Chilean coast, especially in small pelagic fish (anchoveta and sardine) and crustaceans (Sepúlveda, 2022).

The climatic risk is a probability indicator of the magnitude of damage that may occur due to changes in the environment. In this case, DO reduction is a phenomenon that has a climate forcing. Estimating risk requires knowledge of hazard, exposure and sensitivity of the system (Pörtner et al., 2022), and the spatial extension of a climatic hazard for the recent past (1980–2010) and the medium-term future (2035–2065) under different scenarios of greenhouse gas emissions (such as RCP8.5). Human coastal communities and ecosystems have an increasing risk of exposure to natural hazards (in this case hypoxia) over the EBUS due to the spatial expansion of this phenomena, and negative DO trends in upwelling systems (review in IPCC, 2019).

Equatorward wind stress is predicted to increase in the northern and southern HCS during the next decades. The Coupled Model Intercomparison Project (CMIP5) uses thirteen different models with both historical simulations and an extreme climate change scenario (RCP8.5) (Oyarzún and Brierley, 2019), indicates that the exposure of coastal marine ecosystems to hypoxia may increase direct mortality or severely affect different biological processes. For example, impact on feeding behavior, movement, growth, reproductive process, and reductions in habitat quality for plankton and benthic species is observed. In turn, these impacts scale up to population level, species composition, biodiversity, and commercial fisheries (e.g., Grantham et al., 2004; Vaquer-Sunyer and Duarte, 2008; Ekau et al., 2010; Low and Micheli, 2018; Thomas and Rahman, 2012). Moreover, severe impacts on marine organisms or even mas mortality events of fish and invertebrates are commonly associated with EBUS, where hypoxia frequently occurs (Grantham et al., 2004; Low et al., 2021). Thus, alterations in upwelling regimes as a consequence of climate change are likely to further increase the frequency and magnitude of upwelling-driven hypoxia and, in consequence, mass mortality events (Stauffer et al., 2012).

Vulnerability to hypoxia is variable across temporal and spatial scales (Low et al., 2021) and this is the case in central Chile. Since vulnerability is dependent on economic, sociocultural, geographic, demographic, and governance factors (Kelman, et al., 2016; Low et al., 2021), the Chilean coast could be considered as highly vulnerable to hypoxia. In Chile there are 100 coastal municipalities with around 4.5 million inhabitants. Fisheries and aquaculture are amongst the most important economic and socially productive activities. Climate change is expected to increase the intensity and frequency of deoxygenation over time; thus, hypoxia will increase the vulnerability of coastal communities and the fishing industry. This may lead to socioeconomic impacts such as reduced economic income for fishers and compromise food security. Furthermore, it is considered that there is a highly dependent socioeconomic relationship between artisanal fisheries and small-scale aquaculture, with the environmental services being used. This means they are more vulnerable and less resilient and require support to adapt to climate change. To mitigate this vulnerability requires, in part, continuous, high-resolution and real time DO observations in coastal upwelling ecosystems to inform decision makers and resource users in coastal zones. Unfortunately, insight into high frequency variability in the southern HCS system has been limited by the lack of in-situ and real time data of physical and biogeochemical properties in the coastal zone (Farías et al., 2019; Ramajo et al., 2020).

Thus, a monitoring network of academic, government, fishing cooperatives and social stakeholders could provide a valuable opportunity for local adaptation in the face of escalating exposure to environmental stressors (Low et al., 2021). Capacity building, community participation and adequate governance in coastal communities regions will help to reduce socio-economic impacts associated with these events.