Fermentation and Anaerobic Oxidation of Organic Carbon in the Oxygen Minimum Zone of the Upwelling Ecosystem Off Concepción, in Central Chile

We studied the dynamics of fermentation and anaerobic degradation of organic matter at a fixed station in the Oxygen Minimum Zone (OMZ) within the Humboldt Current System off Concepción, central Chile. Products of the main anaerobic microbial reactions [fermentation, denitrification, and reduction of Fe(OH)3 and SO42–] were analyzed during laboratory incubations of OMZ waters. Fermentation of glucose and amino acids resulted in the production of volatile fatty acids, mainly acetate; these compounds were detected year-round in in situ water samples and were associated with high primary production rates and presence of O2-deficient waters at the sampling site. In contrast, whilst ethanol was produced from glucose fermentation by OMZ water microorganisms under laboratory conditions, it was not detected in the water column during the annual cycle. Evidence of acetate oxidation (which is thermodynamically feasible), with Fe(OH)3 as an electron acceptor, suggests that microbial activity could reduce solid-phase Fe carried by rivers using fermented metabolites in oxygen-depleted water, thus releasing dissolved bioavailable Fe. Here we present evidence for productivity-driven seasonality of biogeochemical cycles in the Humboldt system, supported by fermentation and anaerobic consumption of fermentation products oxidized by a variety of electron acceptors including NO3–, Fe(OH)3, and SO42–. Our results suggest that products of fermentation in the OMZ may provide a source of labile organics for advection to oxygenated waters of subantarctic origin during austral winter. Fermentation, anaerobic oxidation and associated advection of fermentation products are likely to be enhanced during the twenty-first century due both to temperature increase and decrease in dissolved O2 in the water column.


INTRODUCTION
Hypoxic zones within the marine water column appear to be expanding due to warming and eutrophication (Diaz and Rosenberg, 2008;Rabalais et al., 2014;Schmidtko et al., 2017). These processes will enhance fermentative production of semi-reduced metabolites that act as substrates for anaerobic microbial oxidation of organic matter (Oremland and Polcin, 1982;Megonigal et al., 2004) through, for example, NO 3 − (denitrification) and SO 4 2− reduction, and methanogenesis (Sansone and Martens, 1981;Jørgensen, 1982). Presently, O 2depleted marine environments are typically found in sediments, enclosed or semi-enclosed water bodies, coastal ocean dead zones, and areas referred to as Oxygen Minimum Zones or OMZs (Helly and Levin, 2004;Diaz and Rosenberg, 2008). These OMZs can be hundreds of meters in depth, and are characterized by suboxic and anoxic levels of O 2 typically in the productive, weakly ventilated waters of the eastern Pacific and southeastern Atlantic Oceans, and the Arabian Sea (Wyrtki, 1962;Kamykowski and Zentara, 1990;Helly and Levin, 2004;Ulloa et al., 2012;Löscher et al., 2016;Pizarro-Koch et al., 2019). On a global scale, OMZs are significant sinks (via N 2 O and N 2 ) for oceanic N (Codispoti et al., 2001), and are zones hosting active S cycling (Canfield et al., 2010) through diverse communities of anaerobic microbes (Fossing et al., 1995;Ward et al., 2009;Ulloa et al., 2012;Wright et al., 2012;Srain et al., 2015).
In the present study, we examined chemical dynamics of the water column within the OMZ off Concepción, Chile (∼36 o S). This is a coastal area where intense seasonal upwelling (Sobarzo et al., 2001(Sobarzo et al., , 2007 of high-nutrient subsurface watersdepleted in dissolved O 2 -supports high rates of primary production of up to 20 g C m −2 d −1 in near surface waters (Montero et al., 2007;Testa et al., 2018). The area off Concepción constitutes the most southerly extent of the OMZ in the eastern South Pacific (Fuenzalida et al., 2009), the fourth largest (by volume) of the six permanent hypoxic regions in the world oceans (Schneider et al., 2006). At this latitude on the continental shelf, the OMZ is located beneath surface waters of subantarctic origin (eastern South Pacific Transition Water, ESPTW) and is fed by poorly oxygenated subsurface waters (Equatorial Subsurface Water, ESSW) which upwell toward the coast driven by favorable winds during austral spring and summer (i.e., from September-October to March-April). During austral winter, northerly converging winds generate coastal downwelling and offshore bottom Ekman transport, which removes ESSW from the shelf and replaces it with ESPTW throughout most of the water column (Sobarzo et al., 2007).
The goal of the present study was to reveal relevant microbial pathways within OMZ waters [fermentation, and reduction of NO 3 − , Fe(OH) 3 and SO 4 2− ]. A series of controlled incubations of OMZ microbial assemblages, supplemented with high concentrations of substrates, was interpreted in the context of anaerobic processes within the water column during an annual cycle. The study aimed to describe annual patterns within the water column of vertical distribution of volatile fatty acids (VFA), mainly acetate, and other indicators of anaerobic metabolism of OMZ. A further aim was to evaluate potential export of these indicators to oxygenated waters of subantarctic origin.

Sampling
Cruises were carried out within the framework of the project "Microbial Initiative in Low Oxygen off Concepción and Oregon" (MILOCO), and as a part of the Time Series at Oceanographic Station 18 (33 km from the coast, 90 m depth, 36 • 29.94 S, 73 • 07.8 W) conducted by the COPAS Center for Oceanographic Research in the eastern South Pacific (FONDAP CONICYT Chile) in the upwelling ecosystem off Concepción in central Chile (Figure 1). Water samples were collected monthly aboard L/C Kay-Kay II between November 2009 and January 2011, encompassing two austral springs, one austral fall, one austral winter, and two austral summers. Ancillary water column measurements of dissolved O 2 , fluorescence, temperature, salinity, and nutrients were provided by the database of the COPAS Center and the MILOCO Project. Data for hourly coastal winds were collected by the meteorological station at Carriel Sur Airport (36 • 47 S; 73 • 04 W), located less than 10 km from the coast of Concepcion.
Water samples for chemical measurements and incubations were collected with Niskin bottles (10 L) at six depths (0, 10, 30, 50, 65, 80 m) stored in acid-washed carboys in the dark at ca. 10 • C. For VFA, ethanol, and CO 2 measurements, 50 mL aliquots of seawater (in triplicate) were removed on board and dispensed into glass bottles in a N 2 saturated chamber using FIGURE 1 | The geographical location of the study site at Station 18, off Concepción in central Chile. Color scale represents the average chlorophyll-a concentration observed between November 2009 and January 2011. The black solid line shows the 200 m isobath and the magenta dashed square represents the area analyzed for Net Primary Production estimates. The map was generated using Ocean Data View (Schlitzer, 2010).
glove bags (Aldrich R AtmosBag) and immediately poisoned with HgCl 2 (0.001%). Gas-tight bottles were sealed with butyl rubber stoppers, crimped, and then stored in the dark at 10 • C. In the laboratory (within 12 h), 1 L seawater per depth was filtered through pre-combusted 0.7 µm glass fiber filters (Whatman GF/F); filtrates and filters were kept at −20 • C prior to analyses of dissolved amino acids, and elemental analysis of C and N and natural abundance of 13 C and 15 N in particulate material.
On November 29, 2010, inocula of seawater for laboratory incubations were taken from 65 m depth (10.2 • C, 34.6 PSU, 8.4 µM O 2 ), and transferred to 10 mL serum vacuum-tubes BD R . Tubes were stored at 10 • C in darkness until arrival at laboratory.
Dissolved inorganic N anomalies -defined as a linear combination of nitrate and phosphate -were used in order to determine the role and distribution of nitrogen fixation and denitrification in the water column. The parameter N * was estimated following Hansell et al. (2004, N * = [NO 3 − + NO 2 − + NH 4 + ] -16 [PO 4 +3 ] + 2.9). N * values lower than −3 µmol kg −1 are indicative of denitrification while values higher than 2 µmol kg −1 denotes nitrogen fixation (Gruber and Sarmiento, 1997).

Laboratory Incubations
Artificial seawater for incubations was prepared according to Lovley (2006). To remove O 2 , water was autoclaved, capped, and gently bubbled with N 2 for 15 min in a laminar flow hood LABCONCO Class II Type IIA. In addition to those measures, O 2 -sensitive methylene blue (Resazurin 0.0001%, Wolfe, 2011) was added to incubation vessels to detect unwanted traces of O 2 (higher than 700 nM).
Incubations were conducted in darkness in a Shell-Lab incubator (Sheldon Manufacturing, United States) at 10 • C for 7 days (glucose fermentation), 8 days (amino acids fermentation), and 35 days (acetate oxidation). Subsamples of 2 mL were removed and filtered through 0.22 µm filters (MILLEX GV TM filter unit) every ca. 24 h for analyses of amino acids, and pH (pH-indicator paper Neutralit, pH 5.5-9.0, Merck Millipore). Measurements of Fe 2+ aq were carried out by removing 500 µL of water at the beginning and the end of the incubations, and amending with Fe(OH) 3 as electron acceptor. For CO 2 and HS − measurements, 500 µL of gas was removed from the headspace at each subsampling time. Control treatments were prepared without inocula. Filtration of solutions (0.22 µm MILLEX GV TM ), bubbling with N 2 , and the addition of substrates and inocula were conducted under a N 2 saturated chamber.

Fermentation of Glucose and Amino Acids
Aliquots of 30 mL artificial seawater were transferred into gastight bottles (60 mL, in triplicate) containing 3 mL inocula of seawater, and then amended with glucose or the amino acids alanine, leucine, threonine, phenylalanine, glutamic acid, and ornithine to final concentrations of 40 mM glucose and 10 mM of each amino acid. Such high substrate concentrations are four orders of magnitude higher than natural DOC concentrations and were intended to provide a culture medium not limited by substrate in order to achieve quick-start of microbial growth and short incubation times. Consequently, no inferences were made regarding reaction rates from these incubations because of the transient nature of zero-order reactions; however, the experiments do reveal potential reactions within OMZ waters. Incubation vessels were treated with 200 µM molybdate to prevent consumption of fermentative products by SO 4 2− reduction (Oremland and Capone, 1988), and with 200 µM N-guanyl-1,7-diaminoheptane (GC 7 ) to stop acetotrophic methanogenesis (Jansson et al., 2000). Bottles were capped with pre-sterilized butyl-caps and aluminum seals, and then incubated as described. Anoxic incubations (35 days) and inoculations were conducted as described above in the glucose fermentation section. NO 3 − (NaNO 3 , 60 mM), 250 mM of poorly crystalline ferrihydrite-iron oxyhydroxide (Schwertmann and Cornell, 1991), Fe(OH) 3 , and SO 4 2− (Na 2 SO 4 , 50 mM) were added to incubations as electron acceptors, and acetate (NaCH 3 COO, 40 mM) supplemented as C source. Artificial seawater with 5% v/v of reducing solution (Na 2 O 2 S × 5H 2 O and cysteine) was used in the experiments for acetotrophic SO 4 2− reduction. Fe 2+ aq concentrations were measured with the 1,10 Phenanthroline method, using kit HACH (Hach Lange GmbH) and a spectrophotometer HACH DR-4000 (Method 8146, DOC316.53.01049). The detection limit of the assay was 3.5 ± 0.3 µM.
Analyses of VFA, Ethanol, HS − and CO 2 by Gas Chromatography Coupled to Mass Spectrometry (GC-MS) Acetate, propionate, isobutyrate, butyrate isovalerate, valerate, and ethanol were extracted from headspaces of both ambient samples and incubation vessels using solid-phase microextraction. Equilibrium partition between the aqueous phase and headspace was estimated by sonicating standard solutions of VFA (0.1, 1, 5 µM) in gas-tight bottles containing synthetic seawater at 30 • C for 15 min, followed by adsorption of analytes from the gaseous phase on 85 µm CarboxenTM/PDMS Stable Flex micro-extraction fibers (SUPELCO). An adsorption fiber was inserted through the septum in the headspace and maintained for 5 min under continuous stirring, followed by desorption for 5 min at 250 • C in the injection port of the gas chromatograph Agilent 6890N series coupled to a mass spectrometer Agilent 5973 Network. Compounds were separated with an HP-Plot/Q column 30 m (0.32 mm diameter, 0.20 µm film thickness), using He as the gas carrier.
Concentrations of the above compounds were determined using calibration curves (R 2 > 0.998), with a VFA mixture (Supelco 46975-U), and ethanol (HPLC grade, Fisher) added to artificial seawater in the range 5 nmol L −1 to 1 µmol L −1 of acetate, isobutyrate and ethanol (Fisher HPLC grade). Concentrations of VFA and ethanol in the gas phase were calculated as C HS = C ap /[K + (V HS /V S )] with the partition coefficient K = C AP /C HS (C HS is the concentration in the headspace, C AP is the concentration in the aqueous phase, and V HS and V S are headspace and sample volume, Slack et al., 2003). This resulted in partition coefficients K of 1.2 for acetate, 1 for isobutyrate, and 1.4 for ethanol.
HS − and CO 2 were measured in incubation bottles by removing 500 µL of gas from the headspace of either incubation or ambient sample with a Hamilton gas-tight syringe. The gas sample was injected into the GC-MS and detected in SIM mode (m/z 32-33 and 34 for HS − and m/z 28 and 44 for CO 2 ). CO 2 was quantified by comparing sample chromatographic areas with those of a CH 4 reference internal standard co-injected (1 ppm) as a surrogate (1750, NOAA). Henry's Law was used to calculate aqueous concentrations from partial pressure. HS − is reported as percent of maximum abundance since we lacked an appropriate standard.
Detection limits were calculated from slopes of calibration curves, and residual standard deviations were derived from linear regressions of calibration curves (three times residual error times slope, Shrivastava and Gupta, 2011). Detection limits were 40 nmol L −1 for acetate, 10 nmol L −1 for isobutyrate, 50 nmol L −1 for ethanol, and 10 µmol L −1 for CO 2 .

Analyses of Dissolved Free Amino Acids (DFAA) by High-Pressure Liquid Chromatography (HPLC) With Fluorescence Detection
Concentrations of DFAA from incubations, and from Station 18, were quantified as OPA-derivatized adducts (Lindroth and Mopper, 1979) with a Shimadzu LC-10AT HPLC coupled to a Shimadzu RF-10Axl fluorescence detector (set at excitation/emission of 340/450 nm), column oven CTO 10As and autosampler (Shimadzu SIL 10 ADvp). Aliquots of 600 µL sample, mixed with 400 µL methanol, were derivatized in the autosampler with 60 µL ortho-phthalaldehyde/2mercaptoethanol reagent (OPT, Lindroth and Mopper, 1979) and 100 µL sodium acetate buffer 0.1 N, pH 5, and injected (50 µL) into the HPLC. Fifteen amino acids (asp, glu, ser, his, gly, thr, arg, ala, tyr, val, met, phe, ile, leu, lys) were separated using an Alltima C18 (5 µm, 250 × 4.6 mm) column kept at 40 • C, with a mobile phase of 5% tetrahydrofuran in 25 mM sodium acetate and methanol, and at a flow rate of 1 mL min −1 . A gradient of 25-30% methanol in 35 min, 30-50% in 7 min, 50-60% in 18 min, 60-100% in 12 min was used. Initial conditions were restored in 7 min, and the column was equilibrated for 10 min between injections. Amino acids were identified and quantified by comparison with chromatograms of a standard amino acid mix (Pierce 20088) run under the same conditions every 10 injections. The coefficient of variation for quantification of duplicate samples was 9.4%.

Concentrations and Stable Isotope Composition of Particulate Organic C and N
Particulate samples were acid-fumed to remove carbonate (Nieuwenhuize et al., 1994), dried at 60 • C for 24 h, wrapped in tin capsules, and analyzed using continuous-flow isotope ratio mass spectrometry (IRMS, Finnigan Delta Plus) interfaced with an elemental analyzer Carlo Erba NC2500. Reproducibility of standard acetanilide was greater than 0.11% for 13 C, and 0.005 for 15 N. Isotope ratios were expressed as per mil ( ) deviations of isotopic values relative to PDB ( 13 C) or atmospheric N 2 ( 15 N).

Satellite-Derived Chlorophyll-a and Net Primary Production Estimates
We examined chlorophyll-a estimates (monthly averages at 4 km resolution) from the Aqua Moderate-Resolution Imaging Spectroradiometer (MODIS) mission between November 2009 and January 2011, extracting data from the CoastWatch project 1 . Net primary production (NPP) estimates were made using the standard Vertically Generalized Production Model (Behrenfeld and Falkowski, 1997, Ocean Productivity Home Page 2 ), previously validated for the upwelling ecosystem off central Chile (Testa et al., 2018). We analyzed satellite estimates within boundaries of 72 • 48 -73 • 30 W, 36 • 6 -36 • 45 S over the continental shelf surrounding Station 18 (Figure 1). Data were averaged every 8 days between 2009 and 2011 (at least 33% of pixels) at a spatial resolution of ca. 9 km and integrating to the depth of euphotic zone (11-51 m).

Export Flux of POC
We calculated export flux of carbon through the photic zone by deriving an empirical relation (Supplementary

Data Analysis
Both homogeneity of variances (Levene test) and normality of variables (Shapiro-Wilk test) were not fulfilled; therefore, we also tested for significant differences among environmental data using the non-parametric Kruskal-Wallis ANOVA test, and differences among experimental results using the paired sample Wilcoxon Signed Rank test. Correlations were examined using Spearman R coefficients.

Fermentation of Glucose and Amino Acids and Anaerobic Oxidation of Acetate During Laboratory Incubations
The microbial degradations of glucose, amino acids, and acetate were detected during laboratory incubations under excess concentration of substrates (millimolar range) in the absence of O 2 (Figure 2). Production of acetate (from 0 to 1 mM), ethanol (from 0 to 2 mM), CO 2 (from 0 to 4 mM), and H + (pH from 8 to 5) indicated fermentation of glucose during the 7 days incubation period (Figure 2A). Fermentation of the amino acid mix containing alanine, leucine, threonine, phenylalanine, glutamic acid, and ornithine (decrease from ca. 60 mM to ca. 6 mM) resulted in the production of VFA, acetate, propionate, isobutyrate, butyrate, isovalerate, and valerate (from 0 to ca.7.5 mM) during 8 days of incubation ( Figure 2B and  Supplementary Figure 3).
No decay of fermentation products (acetate, ethanol) was detected in 7 days of incubation likely due either to the absence of external electron acceptors for anaerobic oxidation in the culture media, the thermodynamically unfavorable conditions caused by buildup of fermentation products, or the slow growth rates of acetate consumers suggested by comparing decay of amino acids (90% in near 2 days, Figure 2B) with decay of acetate (90% in near 30 days, Figures 2C-E).
Microbial oxidation of acetate with excess NO 3 − (Figure 2C), Fe(OH) 3 (Figure 2D), and SO 4 2− as electron acceptors was detected in the inocula treatments of OMZ water incubated under anaerobic conditions ( Figure 2E). In control incubations, decrease in acetate concentration was not detected (Supplementary Figure 3d) and CO 2 was always below the detection limit. Estimated Gibbs Energy ( G) for fermentation of glucose and amino acids, and anaerobic oxidation of acetate with NO 3 − , Fe(OH) 3 , and SO 4 2− were in the range of −73 to −3777 kJ mol −1 suggesting that these reactions were exergonic (Table 1)

Variability in the Chemistry of the Water Column
Chlorophyll-a concentration increased from ∼2 to >10 mg m −3 during spring and throughout the productive summer season ( Figure 5A). Concentrations of POC and PN averaged 27 ± 46 mmol C m −3 , and 4 ± 8 mmol N m −3 , corresponding to an average C/N ratio of ∼ 7, with maximum values of C and N occurring during the productive season (Figures 5B,C). PO 4 3− concentrations were higher below 30m depth than in surface waters throughout the year, and a pronounced depletion of surface PO 4 3− occurred in surface water during austral summer ( Figure 5D). The annual cycle of NO 3 − generally followed expected patterns, with concentrations typically lower in near-surface waters, and higher in deeper waters. Concentrations were substantially higher throughout the water column (>10 µM) during austral winter and summer 2011, with a conspicuous depletion below 50-m depth, suggesting microbial consumption (Figure 5E). Satellite-based estimates of primary production varied from 30 mmol C m −2 d −1 (0.4 g C m −2 d −1 ) during winter to over 600 mmol C m −2 d −1 (7 g C m −2 d −1 ) during spring-summer ( Figure 5F).
Our sampling coverage for NH 4 + concentrations were rather limited but showed accumulation of up to 2.5 µM NH 4 + in both the deeper and subsurface layers during austral spring and summer ( Figure 6A). A surface maximum of ca. 1 µM NH 4 + was detected in the surface oxygenated ESPTW water mass, but otherwise, concentrations were below 0.5 µM ( Figure 6A). NO 2 − concentrations were generally less than 1.5 µM NO 2 − throughout the water column, except for measurements of up to 7 µM during the productive season in austral summer in suboxic ESSW below 50 m depth ( Figure 6B). The concentrations of VFA (ca. 66% acetate, ca. 33% isobutyrate, and <1% isovalerate) ranged from 0.05 to 4 µM throughout the year, with higher concentrations observed under low O 2 conditions below 50 m depth (Kruskal-Wallis Test, p < 0.05, N = 52, Figure 6C). VFA concentrations of over 2 µM were observed in the water column during suboxic conditions, peaking in austral fall 2009, austral spring 2010, and austral summer 2011. VFA correlated positively with NH 4 + concentrations (Spearman r = 0.2, p < 0.05, N = 78) and negatively with dissolved O 2 (Spearman r = −0.3, p < 0.05, N = 78) and temperature (Spearman r = −0.3, p < 0.05, N = 78). Concentrations of dissolved free amino acids (DFAA) reached up to ∼0.5 µM and generally decreased with depth during upwelling months (austral spring and summer). In contrast, concentrations of DFAA lower than 0.1 µM were observed throughout the water column in winter months ( Figure 6D). Concentrations of DFAA and VFA were higher in ESSW during the upwelling season than in ESPTW during winter (Figures 6C,D).

Fermentation of Glucose and Amino Acids and Anaerobic Oxidation of Acetate
In our experiments with excess concentrations of substrates (millimolar range), the fermentation of glucose produced acetate and ethanol, and the fermentation of DFAA, alanine, leucine, threonine, phenylalanine glutamic acid, and ornithine produced acetate, propionate, isobutyrate, butyrate, isovalerate and valerate. Along with sugars, DFAA are considered to be suitable substrates for microbial heterotrophic metabolism; this would explain why their concentration was maintained lower than 100 nM in the water column, since their removal rates are coupled to their production (Webb and Johannes, 1967;Crawford et al., 1974;Fuhrman, 1987). In the absence of O 2 , amino acids are known to undergo anaerobic degradation (Barker, 1981) as observed in our incubations of microbial assemblages from OMZ waters. Our observations are also consistent with the Stickland catabolic pathway (Stickland, 1934;Nisman, 1954;Barker, 1981), a process that couples anaerobic reduction and oxidation of amino acids and produces VFA, CO 2 , and NH 4 + . In Stickland fermentation, an electron-donating amino acid is oxidized to a VFA that is shorter by one carbon (Stickland, 1934;Nisman, 1954;Barker, 1981); thus, acetate can be produced by alanine oxidation, propionate by threonine, butyrate by glutamic acids, and valerate by leucine. An electronaccepting amino acid is reduced to a VFA of the same number of carbons (Stickland, 1934;Nisman, 1954;Barker, 1981) as in the reduction of threonine to butyrate, and of leucine to caproic acid. In the present study, incubation media remained mildly alkaline (Figure 2B), likely because of NH 4 + formation during fermentation of amino acids (Prüss et al., 1994;Wolfe, 2005).
In our experiments under induced conditions of denitrification and dissimilative Fe(OH) 3 and SO 4 2− reductions, anaerobic terminal oxidation of acetate was detected together with CO 2 production (Figures 2C-E and Table 1, reaction 4-6). This experimental evidence of thermodynamically feasible acetate oxidation -with Fe(OH) 3 as an electron acceptor -reveals the potential ability of microbes to reduce solid-phase iron which is carried by rivers in the region, using the fermented metabolites produced in oxygen-depleted water. This process would release dissolved bioavailable Fe for utilization by phytoplankton and chemolithotrophic microbes (Segovia-Zavala et al., 2013), and metalloenzymes required for reduction of nitrate and nitrite (Milligan and Harrison, 2000). In support of this working hypothesis, bacterial reduction of solid-phase metals associated with oxidation of acetate has in fact been detected in the Baltic Sea (Berg et al., 2013). We suggest that our data provides evidence that acetate, and likely other VFA such as butyric acid, represents a principal intermediate of anaerobic metabolism in OMZ waters on the Concepción shelf.
Twice as much ethanol than acetate was produced during incubations of inocula from OMZ waters (Figure 2A, Wilcoxon Signed Rank test p < 0.05, n = 8), but ethanol was not detected in any sample from the water column. The observed production of ethanol could be an experimental artifact resulting in selection for ethanologenic copiotroph microorganisms, or because of acidification in enclosed vessels (which would normally be buffered to some extent by alkalinity in natural seawater) promoting microbial excretion of neutral molecules such as ethanol (Michels et al., 1979;Wolfe, 2005). These processes may explain the observed onset of ethanol production when pH begins to drop in the experimental vessels, preventing further decrease (Figure 2A). Alternatively, the absence of detectable ethanol could be a result of microbial consumption and fast turnover times in the water column, which would preclude its accumulation. In fact, the activity of the enzyme ethanol dehydrogenase has been previously detected in OMZ waters on the Concepción shelf (González and Quiñones, 2009), thus providing evidence the potential for ethanologenic fermentation triggered by the high concentration of organic carbon and reduced pH in these waters.

Chemical and Physical Variability in the Water Column
General hydrographic patterns detected during this study were consistent with previous research in this area (Schneider et al., 2003(Schneider et al., , 2017Escribano and Schneider, 2007;Sobarzo et al., 2007;Letelier et al., 2009;Escribano and Morales, 2012). The surface layer was dominated by ESPTW which originates from Subantarctic Water (SAW) from the subantarctic front (Schneider et al., 2003), and which is characterized by cold and low salinity waters during austral winter and higher temperature and moderate salinity during austral summer. While ESPTW is transported northward along the rim of the Subtropical Gyre, ESSW is transported southward by the Peru-Chile Undercurrent off central Chile, constituting the water source for coastal upwelling (Sobarzo et al., 2007;Silva et al., 2009). The fundamental mechanisms proposed to explain seasonal variations in predominance of these two water masses on the continental shelf have been coastal upwelling, and downwelling induced by alongshore winds (Sobarzo et al., 2007;Schneider et al., 2017). Other processes, such as wind curl, coastally trapped waves, or topographic upwelling, have been less studied at these latitudes. These processes could explain a seemingly anomalous bottom cooling detected in July 2010 that could not be attributable to upwelling favorable wind.
As expected, upwelling favorable southerly winds resulted in a shallowing of the nutricline and upwelling of subsurface nutrient rich waters (up to ∼15 µM NO 3 − and ∼1.5 µM PO 4 2− ) closer to the surface, leading to the fertilization of the photic zone, as has been previously documented (Sobarzo et al., 2007;Farías et al., 2015). Satellite-based estimates of primary production during spring-summer 2009 and 2010 showed a three-fold increase in rates of primary production and significant increases in the concentrations of chlorophyll-a, POC and PN. The seasonal pattern of phytoplankton activity in surface waters, generally followed previously described annual patterns of primary production (e.g., Montero et al., 2007;Testa et al., 2018) during spring-summer 2009 and 2010. Sinking POC fluxes however did not follow the seasonal pattern of phytoplankton activity. Fluxes of POC sinking to 50-m depth averaged 23 ± 4 mmol C m −2 d −1 throughout the year, varying between 25 ± 3 mmol C m −2 d −1 during the upwelling seasons and 19 ± 1 mmol C m −2 d −1 in austral winter. Variation in sinking fluxes of POC at 50-m depth did not closely reflect the threefold increase in primary production observed in both water masses between winter and upwelling conditions. These sinking fluxes were therefore rather similar throughout the year (P sink upwelling ESSW/P sink winter ESPTW = 1.3), in good agreement with the low seasonal variability in export production previously detected (3-10% of surface waters PP production) for this area .
In a modeling study, Pizarro-Koch et al. (2019) concluded that the southern extension and seasonality of the OMZ at Station 18 is driven more by changes in the undercurrent transport and mesoscale eddy fluxes than by mixing and local input. Similar year-round sinking carbon fluxes reported here appear to support this conclusion.
At Station 18 off Concepción, elevated concentrations of VFA and NH 4 + were concurrently observed (Figures 6A,C), suggesting fermentation of amino acids and production of NH 4 + in the water column through the Stickland fermentation reaction. In the OMZ off Peru, the stoichiometric production of NH 4 + from fermentation of amino acids could explain previously unaccounted NH 4 + requirements of up to 17 mmol NH 4 + m −2 d −1 for anammox bacteria . Our monthly field determinations in the water column off Concepción show that both VFA and NH 4 + concurrently increased during November-March 2010 (austral spring-summer) whereas decay of VFA occurred during intrusion of oxygenated ESPTW in March-July (austral winter). In contrast, NH 4 + was depleted much later in July-November (austral autumn), a likely consequence of a slow build-up of anaerobic NH 4 + oxidizers (anammox) bacteria with relatively low growth rates (Kuenen, 2008;Kartal et al., 2013). Anammox reaction rates (ca. 0.4 nmol L −1 h −1 , Canfield et al., 2010) are about 20 times lower than suboxic rates of degradation of amino acids (ca. 10 nmol L −1 h, Pantoja et al., 2009) in the OMZ off the northern Chilean coast, implicating that Stickland fermentation could be a source of NH 4 + for anammox in OMZ waters.

Dissolved inorganic N anomalies (NO 3
− + NO 2 − + NH 4 + ; N * ) are probably associated with water column nitrogen cycle microaerophilic and anaerobic metabolism. N * showed a yearround NO 3 − deficit in the study area, with similar negative N * values that have been previously recorded in the area from 21 to 33 o S (De Pol-Holz et al., 2009;Galán et al., 2014;Galán et al., 2017), and have been attributed to denitrification and anammox. Under conditions of NO 3 − depletion, 15 N enrichment of remaining NO 3 − would be expected in these low O 2 waters. Upwelled water with isotopically heavy NO 3 − may result in the enrichment of δ 15 N of particulate organic nitrogen (PON). This mechanism is consistent with δ 15 N-PON values of up to 35 found during this study. The similarity of the isotopic range of PON found in both OMZ and subantarctic water masses during the winter, can be explained by enhanced mixing of both water masses at that time of the year. Similar conclusions were reported by De Pol-Holz et al. (2009) analyzing δ 15 N-NO 3 − for this study area, and by Cline and Kaplan (1975), Liu and Kaplan (1989), Castro et al. (2001), and Sigman et al. (2003) for the California Current System.
Since OMZ waters (ESSW) are pushed offshore by SAW during austral winter, waters enriched in δ 15 N-PON and acetate, and with negative N * , mix with the more oxygenated waters of subantarctic origin, thus imprinting a signal of anaerobic activity on ESPTW. Calculations using the mixing triangle shows that, even during austral winter, 50% of ESSW is located at depths as shallow as 50 m (the ESSW can be present at 25-m and above during austral summer). Concentrations of acetate (>1.5 µM) in ESSW were higher than in surface ESPTW, both during summer (0.01-0.8 µM) and winter (<0.2 µM). Mixing of water masses allows the byproducts of anaerobic metabolism to spread beyond the boundaries of OMZ source waters, thus providing labile and low molecular weight dissolved organics to surrounding oxygenated ESPTW.
Acetate, isobutyrate, and isovalerate were present at significant levels within the OMZ and surrounding waters as biomarkers for fermentative activity. Twice as much acetate (range 0.1-2.5 µM) was present in the water column during upwelling seasons than observed in winter (Kruskal-Wallis Test, p < 0.05, N = 52), and showed a similar trend to isobutyrate which ranged from 0.9-1.5 µM. Elevated concentrations of acetate (up to 10-60 µM) have been detected in the anoxic Black Sea (Mopper and Kieber, 1991;Albert et al., 1995), the Pettaquamscutt River Estuary (ca. 7 µM, Wu and Scranton, 1994), and at ca. 10 µM in anoxic waters of the Cariaco Basin (Ho et al., 2002). In more oxic waters, acetate concentrations have been detected in the range 0.01-2.4 µM at the Chemotaxis Dock in Woods Hole (Lee, 1992), surface waters of the Cariaco Basin (Ho et al., 2002), in Long Island Sound (Wu et al., 1997), and the Gulf of Mexico (Zhuang et al., 2019). Acetate concentrations in suboxic waters off Concepción were generally within the ranges in the oxic and anoxic waters described above, and were likely enhanced by higher rates of primary production ( Figure 5F) than observed in the Cariaco Basin, Long Island Sound and the Gulf of Mexico. Higher concentration of isobutyrate observed off Concepción (up to 1.5 µM) -associated with anaerobic degradation of iso-leucine (Mueller-Harvey and John Parkes, 1987) -cannot be readily compared with those of other coastal environments because of lack of measurements in the water column. As a reference, Sansone and Martens (1982) measured iso-butyrate in the eutrophic sediments of Cape Lookout Bight within the range < 0.5-6 µmole per liter sediment, and concentrations of up to 8 µM have been determined in pore waters at several coastal sites (Finke et al., 2006).
Enhanced water column concentrations of VFA up to 4 µM were detected during austral spring, summer, and fall throughout the water column (dominated by ESSW, Supplementary  Figures 4A,B and Figures 3, 6C). Intruding oxygenated ESPTW of subantarctic origin during winter contains up to 0.5 µM VFA that can be attributed to the fermentation of organic matter by ambient microbial assemblages within OMZ waters.

CONCLUSION
We identified several products of the main anaerobic microbial reactions (fermentation, denitrification, and reduction of FeOH 3 and SO 4 2− ) during laboratory incubations of OMZ waters, and we related these products to the annual cycle in water column chemistry within the upwelling ecosystem on the continental margin off Concepción, central Chile. Degradation of organic matter in anoxic systems relies on fermentation products and these can then be transported to neighboring oxygenated waters as labile substrates for microbial respiration. Within OMZ waters, acetate produced by fermentation is a major substrate for SO 4 2− reducers (Megonigal et al., 2004;Jørgensen and Kasten, 2006), and was shown to be active even in the presence of NO 2 − and NO 3 − (Canfield et al., 2010). The reduction of SO 4 2− , with acetate as electron donor, is thermodynamically feasible in OMZ waters (Table 1) when the inhibitor HS − remains in low concentrations (Arndt et al., 2013). A group of Gammaproteobacteria affiliated to HS − oxidizers have been identified in OMZ waters off the Chilean coast (Canfield et al., 2010), and a SO 4 2− -reducing microorganism has been isolated from the OMZ in Peruvian coastal waters (Finster and Kjeldsen, 2010); these observations provide support for the findings shown here.
Implications of HS − , SO 4 2− , amino acids, and acetate cycling are evident in cycling of N and C in OMZ waters. Fermentation of amino acids could conceivably fuel the NH 4 + requirements of anammox bacteria because degradation of organic matter by denitrification is insufficient to account for observed anammox rates in OMZ waters (Thamdrup et al., 2006;Lam et al., 2009). The potential production of NH 4 + through Stickland fermentation of amino acids in the OMZ could have implications for our understanding of the biogeochemistry of C and N, and could elucidate sources of NH 4 + for the microaerophilic NH 4 + oxidizers Thaumarchaeota (Venter et al., 2004;Könneke et al., 2005;Francis et al., 2005) and anammox bacteria (Kuypers et al., 2005;Dalsgaard et al., 2005;Thamdrup et al., 2006); these microbes have been unexplored to date.

DATA AVAILABILITY STATEMENT
The datasets generated for this study are available on request to the corresponding author.

AUTHOR CONTRIBUTIONS
BS and SP-G initiated and planned the study. BS conducted the field campaigns. BS, SP-G, GD, HG, LF, AS, and NP provided the chemical and isotopic analyses. GT and MS provided the satellite and physical analyses. All the authors contributed to the data analysis and writing of the manuscript based on an initial version written by BS and SP-G.

FUNDING
We thank the Hanse-Wissenschaftskolleg Delmenhorst, Germany, for sponsoring the "Marine Organic Biogeochemistry" workshop in April 2019. The workshop was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation), grant # 422798570.

ACKNOWLEDGMENTS
We acknowledge the support provided by the COPAS Oceanographic Time-Series Station 18 off Concepción maintained by the Center for Oceanographic Research in the eastern South Pacific (FONDAP COPAS Center). We would particularly thank Prof. Dr. Wolfgang Schneider for validating data used in this study. Prof. Dr. Donald Canfield is acknowledged for providing insightful comments to an earlier version of this manuscript, and Dr. David Crawford for valuable comments. We are also grateful to UdeC personnel Lilian Núñez, Víctor Acuña, and Eduardo Tejos of the Marine Organic Geochemistry Laboratory, as well as Captain Sergio Marileo and the crew of L/C Kay-Kay II for valuable help during sampling and laboratory work. BS acknowledges the support provided by a student fellowship from the Gordon and Betty Moore Foundation, and valuable insights on thermodynamics and microbial metabolism from Prof. Dr. Kurt Hanselmann and international graduate courses ECODIM (Concepción, Chile) and Geobiology (Santa Catalina Island, United States). GT acknowledges the supports provided by a student fellowship (CONICYT-PFCHA/Doctorado Nacional/2017-21170561). Figure 1 was produced using Ocean Data View software developed by Reiner Schlitzer (http://www.awi-bremerhaven.de/ GEO/ODV).