Inter-decadal changes in the intensity of the Oxygen Minimum Zone off Concepción, Chile (~ 36° S) over the last century

Introduction Conclusions References


Introduction
Oxygen Minimum Zones (OMZs) are epipelagic and mesopelagic subsurface layers of suboxic waters (e.g., ≤ 22 µM O 2 ) found along eastern boundary currents such as the Eastern Tropical North and South Pacific, and the Benguela current, as well as the Arabian Sea and the Equatorial Pacific, where upwelling of nutrient-rich waters promotes (ENSO) can also impact oxygenation of south Pacific waters (Blanco et al., 2002;Carr et al., 2002;Levin et al., 2002). In central-southern Chile, the upper edge of the OMZ deepens during El Niño thus allowing greater oxygenation of bottom waters (Gutiérrez et al., 2000;Neira et al., 2001;Escribano et al., 2004). Analyzing a sedimentary record from northern Chile, Vargas et al. (2007) related changes in coastal upwelling and biological production with variations in the Pacific Decadal Oscillation (PDO), characterized by an ENSO-like interdecadal variability in the Humboldt Current System. During the cold phase of the PDO, primary production intensifies in response to upwelling and fertilization of the upper ocean (Mantua et al., 1997(Mantua et al., , 2002Cloern et al., 2007), leading to enhanced oxygen consumption in the water column (Wyrtky, 1962;Sarmiento et al., 15 1998; Helly and Levin, 2004). Since patterns of biological production and oxygenation of the water column during PDO cycles resemble those of ENSO , we hypothesize that variations at the scale of the PDO promote chemical and biological changes in the OMZ off central-southern Chile.
Past redox variations can be analyzed using trace elements in sediments since re-20 dox sensitive metals are less soluble under reducing conditions resulting in authigenic enrichment in low oxygen and high organic matter environments (Algeo and Maynard, 2004;McManus et al., 2005). This chemical behavior makes uranium (U), molybdenum (Mo), and cadmium (Cd) useful paleoredox and paleoproductivity proxies (Algeo and Maynard, 2004;McManus et al., 2005;Riquier et al., 2005). 25 In the past decade, an abundant and diverse microbial community has been detected in OMZ waters off central and northern Chile (Stevens and Ulloa, 2008;Farías et al., 2009;Quiñones et al., 2009;Canfield et al., 2010;Molina et al., 2010;Levipan et al., 2012;Srain et al., 2015). Temporal and compositional variations in this microbial com- Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | munity can be studied by analyzing their cell membrane lipids (biomarkers) preserved in the sedimentary record, as has been done for other OMZ areas of the world ocean (Schouten et al., 2000a;Arning et al., 2008;Rush et al., 2012). We studied redox sensitive metals and organic biomarkers in a 110 year sedimentary record from the OMZ within the upwelling ecosystem off Concepción, central-southern 5 Chile (36 • S), to infer changes in biological production and oxygenation of the water column. Our goal was to assess whether the intensity of the OMZ has varied over the past century in response to ocean/atmosphere circulation patterns, and whether this has affected the prokaryote community.

Sampling
The study site (Station 18;36 • 30.8 S 73 • 7 W) is located in the coastal upwelling ecosystem off central-south Chile at ca. 18 nautical miles from the coast of Concepción ( Fig. 1). Sampling was carried out in the framework of the grant "Microbial Initiative in Low Oxygen off Concepción and Oregon" (http://mi_loco.coas.oregonstate.edu), 15 and the Oceanographic Time Series Program (Station 18) of the Center for Oceanographic Research in the eastern South Pacific (COPAS) at University of Concepción (www.copas.udec.cl/eng/research/serie). A 25 cm-long sediment core was collected at a depth of 88 m during austral summer (February 2009) using a GOMEX box corer onboard R/V Kay-Kay II. The top 5 cm 20 were sectioned on board every 0.5 cm, whereas the rest of the core was sampled at 1 cm resolution. Samples were stored in glass petri plates and kept frozen at −18 Fluorescence was transformed to chlorophyll a according to Parsons et al. (1984). All water column data were obtained from the database of the COPAS Center.

Sedimentary redox potential and organic carbon
Redox potential was measured in the top 15 cm of the sediment core using a redox potential sensor (Hanna) with an accuracy of ±0.1 mV. Sedimentary organic carbon 5 was determined by high temperature oxidation using a NA 1500 Carlo Erba elemental analyzer. Inorganic carbon was removed by putting the samples into an Ag cup and then fuming with concentrated HCl. After this, the samples were dried overnight at ca. 60 • C and then wrap the Ag cup into a tin cup for analysis.

10
210 Pb activities were determined in sediment core sections by Alpha spectrometry of its daughter 210 Po using 209 Po as a yield tracer (Flynn, 1968). The activities were quantified until 1σ error was achieved in a Canberra Quad Alpha Spectrometer. The ages (CE, common era) were established according the constant rate of supply model (CRS, Appleby and Oldfield, 1978), which considers the unsupported 210 Pb invento-15 ries ( 210 Pb xs ). Geochronology of the sediment core was established through the best fit of curves of the ages obtained from the CRS model and three 14 C control points from longer cores retrieved in 2006 at the same sampling site (Muñoz et al., 2012, core VG06-2). Radiocarbon ages were converted to calendar years before present using calibration curve MARINE09 (Reimer et al., 2009) and applying a DR = 137±164 years 20 and 2σ confidence interval (Table S1 in the Supplement).

Trace metal analysis
Trace metals molybdenum (Mo), uranium (U), and cadmium ( were prepared using sequential acid digestion (HNO 3 , HCl, HClO 4 , HF) until total dissolution. Analytical blanks were determined following the above procedure using 18.0 MΩ water and subtracted from sample measurements. Accuracy and precision of measurements were assessed by the analysis of the National Research Council of Canada reference material MESS-3. Excess metal (Me xs ) was calculated as  1-nonadecanol and squalene, and were transformed to contents by normalization by organic carbon content.

Analysis of glycerol dialkyl glycerol tetraethers (GDGTs) by High Performance Liquid Chromatography -Atmospheric Pressure Chemical Ionization -Mass Spectrometry (HPLC-APCI-MS)
Sedimentary material was sequentially extracted by ultrasonication (3x) with methanol, dichloromethane-methanol (1 : 1, vol/vol), and dichloromethane. Lipid extracts were concentrated using a rotary evaporator and dried over a small Pasteur pipette filled with combusted glass wool and anhydrous Na 2 SO 4 . Lipids were separated into nonpolar and polar fractions using a Pasteur pipette filled with activated Al 2 O 3 , after elu- 15 tion with hexane/dichloromethane (9 : 1, vol/vol) and dichloromethane/methanol (1 : 1 vol/vol), respectively. An aliquot of the polar fraction was dissolved in hexane/propanol 100 % eluent B for 5 min. Quantification of core GDGTs was achieved by co-injection of samples with a C 46 GDGT as internal standard (Huguet et al., 2006).

Statistical analysis
Homogeneity of variances was assessed using the Levene's test, whereas normality was determined using a Shapiro-Wilk test. Non-parametric Spearman correlations 5 were calculated between selected variables in order to determine statistical associations with significance < 0.05 (Software Statistica, version 12).

Oceanographic setting of the study site
During austral fall and winter (April to August), temperature ranged between 11 and

Sedimentary redox potential and organic carbon
Redox potential decreased from −176 mV at the water-sediment interface to −325 mV below 3 cm, indicating predominance of reducing conditions in near-surface sediments at the time of sampling during austral summer (Fig. 2e), consistent with the occurrence of 5 µM O 2 in bottom waters of (Fig. 2d). A surface fluffy layer with a Thioploca mat was 5 observed at the sediment-water interface. Organic carbon content varied between 2 to 4 wt.% (Fig. 2e).

Geochronology
210 Pb xs activity was detected down to 23 cm in the core where reached background values of 0.80 ± 0.02 dpm g −1 . The geochronology was estimated using CE ages from 10 210 Pb xs inventories (Table S1) transformed to Cal BP years and calibrated ages from radiocarbon measurements values fitted a polynomial curve (r 2 0.99) allowing to adjust ages at the bottom of the core, that generates errors with the CRS model (Binford, 1990). A recent sedimentation rate of 0.24 ± 0.02 cm yr −1 was established and an exponential decreased is observed due to sediment compaction (Fig. 2f). Thus, the core 15 represents ca. 110 years of sedimentation at Site 18.

Redox sensitive trace metals
Redox sensitive metals are most enriched in the interval ca. 1935-1970 CE ( Fig. 3ac; black bar). Excess molybdenum (Mo xs ) content ranged between 2.5 and 6.5 ppm ( Fig. 3a), showing a similar vertical distribution as uranium (U xs ) that ranged between 20 1.1 to 4.1 ppm (Fig. 3b), and Cd xs , which ranged between 1.9 and 0.8 ppm (Fig. 3c showed lower contents of redox-sensitive metals ( Fig. 3a-c; white bars), and presumably more oxygenated bottom waters and sediments.

Mono-O-alkyl glycerol ethers (MAGEs) indicators of fermentative and sulfate reducing bacteria
−1 (Fig. 4f). MAGEs content remained low during the period 1901-1928 CE with an average concentration of 50 µg (g C org ) −1 (Fig. 4f). From ca. 1935 CE, 5 MAGEs concentrations increased reaching the highest content in surface sediments (Fig. 4f). MAGEs correlated positively with Mo xs (R s : 0.4, p < 0.05) and Cd xs (R s : 0.6, p < 0.05). C 27 -trisnorhopene is favored in anoxic and euxinic environments, and during upwelling events (Grantham et al., 1980;Schouten et al., 2001), and is considered an indicator of anaerobic microbial degradation (Volkman et al., 1983;Duan et al., 1996;Duan, 2000;Peters et al., 2005). Fluctuations in bacterial hopanes and hopanols are related to variations in bacterial 5 groups (Ourisson and Albrecht, 1992; Innes et al., 1998;Rohmer et al., 1984;Talbot et al., 2007). MAGEs C 16 , C 17 , and C 18 mono-O-alkyl glycerol ethers are present in fermentative and sulfate reducing bacteria (Langworthy et al., 1983;Langworthy and Pond, 1986;Ollivier et al., 1991), although this biological source does not appear to be unique (Hernández-Sanches et al., 2014). That said, the statistical relationship found between MAGEs (R s < 0.05) and reducing conditions in the core collected from 18 St. is explained as changes in abundance and occurrence of bacteria involved in microareophilic and anaerobic metabolism, in response to variations in water column oxygenation over the continental shelf off Concepción. The downcore distribution of inorganic and organic proxies reveal a period of ca. 15 35 years between ca. 1935 and 1970 CE (Figs. 3 and 4; black bar) when redox sensitive metals (Fig. 3), sterols (Fig. 4a), GDGTs (Fig. 4b), C 27 trisnorhopene (Fig. 4c), C 31 hopanol (Fig. 4d), and MAGEs (Fig. 4f) were higher. Taken together, these patterns allow us to infer that water column O 2 was comparatively lower than during those periods immediately above and below, in association with enhanced primary produc-20 tion reflected in increase of sterols and GDGTs contents (Fig. 4a). Likewise, two periods with relatively more ventilated and oxygenated conditions are evident between ca. 1901 and 1919 CE, and between ca. 1979 and 2005 CE (Figs. 4 and 5). Both of these periods were characterized by low metal enrichments (Fig. 3), a lower content of bacterial biomarkers related to oxygen depleted conditions such as C 27 trisnorhopene, 25 C 31 hopanol, and MAGEs (Fig. 4c, d and f), and lower organic matter fluxes evidenced by low contents of sedimentary sterols (Fig. 4a) and GDGTs (Fig. 4b).

Patterns of redox depositional conditions, primary and exported
We suggest that from ca. 1935-1970 CE there was higher export production, and that this export is responsible for the increase in phytoplankton sterols (Fig. 4a) current with an increase in Cd (Fig. 3) and GDGTs (Fig. 4b). An enhanced sinking of organic matter leads to a subsequent increase in O 2 consumption by microbial degradation, potentially depleting O 2 in the water column (Helly and Levin, 2004;Canfield, 2006) and sediments. Such conditions lead to Mo, U and Cd enrichment sediments. Higher GDGTs content during this same time period (Fig. 4b) may reflect better preservation favored by severe O 2 depletion. The positive correlation between sterols, GDGTs, and U enrichments support this conclusion, since U enrichment occurs under low O 2 concentration and/or high organic matter deposition (Dezileau et al., 2002;Tribovillard et al., 2006;Muñoz et al., 2012). Schouten et al. (2004) and Zonneveld et al. (2010) reported that GDGTs preservation is lower in oxygenated than in suboxic-10 anoxic settings.

Changes in microbial communities in response to redox variation
Hopanols C 31 and C 32 are used to analyze changes in the bacterial community structure because they are the diagenetic products of bacteriohopanetetrols (BHPs), which in turn can have different bacterial sources (Talbot et al., 2003). The hopanol content 15 was dominated by C 32 hopanol whose predominance in recent sediments has been previously reported (Buchholz et al., 1993;Innes et al., 1997Innes et al., , 1998Talbot et al., 2003). An increase in C 31 hopanol content between ca. 1935 and 1970 CE (Fig. 4d) is indicative of low oxygen if analyzed in light of the positive correlation between Mo xs and Cd xs (R s : 0.6 and 0.4 respectively; p < 0.05). The content of C 32 hopanol, a diage-20 netic product of BHTs (Innes et al., 1998;Talbot et al., 2003), mostly produced by heterotrophic aerobic bacteria (Rohmer et al., 1984), exhibited a slight decrease (Fig. 4e) concurrent with the enrichment of C 31 hopanol (Fig. 4d) and redox sensitive metals (Fig. 3). Observed changes in abundance and distribution of C 31 and C 32 hopanols in concomitance with past variations of oxygen in the water column at the study site are 25 consistent with previous findings by Saenz et al. (2011) and Kharbush et al. (2013). These authors found that the abundance and structural diversity of BHPs, the biological sources of hopanoids, increase with decreasing oxygen in the water column of 6017 Introduction Trisnorhopanes are bacterial lipid markers associated with upwelling and anoxic depositional environments, although its biological source has not yet been identified (Schouten et al., 2001;Peters et al., 2005). The highest C 27 trisnorhopene (Fig. 4c) 5 contents occurred during the period of high primary production and O 2 depletion, suggesting a relationship between its abundance and upwelling-favorable conditions and anaerobic bacterial activity, as previously suggested (Grantham et al., 1980;Duan et al., 1996;Duan, 2000;Schouten et al., 2001).
The sedimentary content of MAGEs was also higher in the period 1935-1970 CE and10 in the topmost sediments (Fig. 4f). MAGEs have been detected in sediments from upwelling regions of Namibia, Peru, and central-southern Chile and are attributed to the occurrence of sedimentary sulfate reducing bacteria (Arning et al., 2008). The presence of sulfate reducing bacteria has been previously documented as well for coastal waters of Chile (Canfield et al., 2010) and Peru (Finster and Kjeldsen, 2010).

Forcing of variations in the intensity of OMZ in central-southern Chile
The combined records of redox-sensitive metals and biomarkers suggest the occurrence of enhanced reducing conditions, both in the water column and at the sedimentwater interface, from ca. 1935 until 1970 CE (Figs. 3 and4), that roughly coincide with a cool (negative) phase of the Pacific Decadal Oscillation (PDO) (Fig. 4g). This sug-20 gests a link between changes in continental shelf oxygenation off Concepción and the PDO cycle, with alternating phases of decreased (1901-1930and 1979CE) and enhanced upwelling (ca. 1935to 1970. The PDO is a recurring pattern of oceanatmosphere variability in which the Pacific central gyre cools down while the eastern margin warms up, with phases that last between two and three decades (Mantua et al., 25 1997, 2002). The PDO plays a major role in decadal-scale oceanographic variability in the Pacific Ocean (Mantua et al., 1997(Mantua et al., , 2002White and Cayan, 1998;Johnson and McPhaden, 1999 warmer while parts of the eastern Pacific become colder. The reverse pattern occurs during warm or positive phase. Periods of favorable upwelling conditions off central Chile may have been triggered by an enhanced thermal contrast between the sea-surface and land during negative phases of the PDO (Bakun, 1990;Vargas et al., 2007). Negative correlations be-5 tween PDO index values with algal sterols (Figs. 4a and g; R s : −0.3; p < 0.05) and GDGTs (Figs. 4b and 5g; R s : −0.2, p < 0.05) suggest an inverse relationship between the PDO and primary and export production at the study site, at least during the last 110 years. Interdecadal variations of enhanced coastal-upwelling conditions, recorded in sediments obtained from Mejillones Bay (23 • S) during the last century, have been previously reported by Vargas et al. (2007). These authors suggested that decreased anomalous sea surface temperatures during interdecadal ENSO-like conditions might have exacerbated the land-sea thermal contrast, which in turn intensified the wind stress responsible of upwelling events. Negative correlations between sedimentary C 27 -trisnorhopene, C 31 hopanol, 15 MAGEs, and PDO values (R s = −0.3, −0.4, 0.3, −0.2, respectively; p < 0.05) and a positive correlation between C 32 hopanol (R s = 0.3) and PDO suggest that this widebasin climatic anomaly has an impact on local oceanographic conditions off Concepción that in turn modulate the structure of the prokaryotic community. Bacterial C 31 hopanol and MAGEs derive from microorganisms associated with marked chemoclines 20 and redox gradients (Rohmer et al., 1984;Innes et al., 1997Innes et al., , 1998Talbot et al., 2003Talbot et al., , 2007 showed an inverse correlation with PDO index (R s : −0.4 with C 31 hopanol, −0.3 with MAGEs, p < 0.05). Thus, positive PDO phases (warm) were likely associated with a decrease in wind-driven upwelling, greater oxygenation, decreased primary productivity, and a concomitant decrease of microorganisms associated with low oxygen. Re- Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | tant fish populations, salmon catches and stocks, and recruitments of ground-fish stocks (Kawasaki, 1991;Adkinson et al., 1996;Yasuda at al., 1997;Hollowed et al., 1998;Peterman et al., 1998). Deutsch et al. (2011) reported that decadal climate oscillations could cause nutrient depletion for photosynthesis due to enhanced nitrogen removal in the marine water column during the positive phase of PDO (warmer 5 surface water). Here, we show the effects of the PDO on the structure of the planktonic prokaryotic community based on bacterial and archaeal biomarkers. Significant negative correlations were detected between the PDO and C 31 hopanol (R s = −0.4, p < 0.05), and GDGTs (R s = −0.2, p < 0.05). Microaerophilic methanotrophic bacteria that flourish under high methane and low oxygen environments (Hanson and Hanson, 10 1996) could be a biological precursor of the observed C 31 hopanol in the sedimentary record (Rohmer et al., 1984;Talbot et al., 2001). Higher content of sedimentary C 31 hopanol during the cold (negative) phase of PDO from ca. 1930-1970 and g) could be the result low oxygen concentrations in the water column that could have favored the presence of methanotrophic bacteria along the oxycline, where com- 15 paratively high methane and low oxygen concentrations prevail (Scranton and Brewer, 1977;Farías et al., 2009), and where micro-aerophilic methane oxidation occurs (Blumenberg et al., 2007;Farías et al., 2009;Berndmeyer et al., 2013). At Station 18, Farías et al. (2009) found that dark chemoautotrophy related to aerobic oxidation of methane occurs in the oxycline mainly during active upwelling. In addition, the rela-20 tively higher content of GDGTs during the negative phase of the PDO ( Fig. 5b and g) could be indicative of enhanced outgassing of N 2 O to the atmosphere, since marine ammonia oxidizing archaea have been suggested as responsible for a great proportion of oceanic N 2 O production (Santoro et al., 2011).