North Atlantic temperature control on deoxygenation in the northern tropical Pacific

The study addresses how natural climate variability on interannual to multidecadal timescales controls oxygen levels in the Eastern Tropical North Pacific (ETNP) oxygen minimum zone (OMZ). Declines in marine oxygen inventory, already about 2% since 1960, have major biogeochemical and ecological consequences, yet instrumental oxygen records are short, hindering attribution to known climate modes (ENSO, PDO, NPGO, AMO). The Gulf of California (GoC), at the core of the North Pacific OMZ, preserves annually varved sediments that record denitrification intensity via sedimentary nitrogen isotopes (δ15N), enabling reconstruction of past deoxygenation. The study reconstructs interannual to multidecadal variability over select 200-year windows spanning the last 55 ka to identify dominant climate periodicities and mechanisms linking Atlantic and Pacific variability to Pacific deoxygenation.

Prior work shows global deoxygenation due to warming-driven stratification and reduced O2 solubility, with OMZ expansion affecting N2O production, fixed-nitrogen inventory, productivity, and the biological carbon pump. The North and Tropical Pacific contribute up to ~40% of existing and projected oxygen loss and host the largest OMZs. Multidecadal climate variability (PDO, AMO) modulates oxygen trends via temperature and ventilation effects. Since the late 1970s, increased denitrification in North Pacific OMZs has been linked to a shift to positive PDO phase. Similar multidecadal trends are observed between ETNP O2 anomalies and the AMO index over the last ~60 years, but short instrumental records limit robust attribution. PDO and related modes (ENSO, NPGO) affect SST, mixing, and nutrient supply; AMO variability (~70–80 years with ~35-year harmonic) is increasingly linked to global teleconnections (e.g., ITCZ shifts, wind stress curl changes) that can influence Pacific overturning and gyre dynamics, although the stability and drivers (internal vs external) of these oscillations remain debated.

Study site and proxies: Annually varved sediment sections from the Guaymas Basin, Gulf of California (core MD02-2515 piston and MD02-2517 CASQ; 27°29.01′N, 112°04.46′W; 881 m) were used. Ten intervals of continuously laminated sediments, each ~160–200 years long (~0.3–0.5 m), were selected to represent key climate periods from the last glacial to recent pre-industrial. All slabs were sub-sampled at annual resolution. Proxies measured included sedimentary nitrogen isotopes (δ15N) as a denitrification (deoxygenation) marker, and biogenic silica (BioSi) accumulation as an upwelling/productivity indicator. Analytical methods: Opal content was determined via molybdate-blue spectrophotometry after NaOH dissolution (absorbance at 812 nm) following Mortlock & Froelich. Bulk sediment δ15N (‰) was measured by CF-IRMS (EA-IRMS; VG Isotech Prism) with analytical precision ±0.2‰ (1σ). Age control followed published models for MD02-2515/17. Time-series analyses: Annual records were interpolated to even annual spacing (cubic spline), detrended, and pre-whitened. Spectral analyses used Blackman-Tukey and Maximum Entropy methods implemented in Analyseries 2.0.8. Statistical significance of peaks was evaluated via multitaper (MTM) F-tests; only frequencies with >90% confidence retained. Two analysis sets were performed: (1) filtered to remove periodicities >70 years to examine interannual to multidecadal variability; (2) unfiltered to detect longer AMO/PDO periodicities. Band-pass filtering at 4–6 dominant frequencies reconstructed variance contributions by short (≤10 years; ENSO-/NPGO-like) versus long (>10 years; PDO-/AMO-like) periodicities. Wavelet analyses assessed nonstationary periodicities. Modern reanalyses: Observed low-pass filtered annual-mean SSTs (HadISST v1, 0.5°) and subsurface ocean currents (SODA v2p2p4, 0.5°; 96 m depth, representative of EUC core depth) for 1950–2012/2008 were linearly detrended and regressed on positive phases of AMO and PDO indices. Equatorial Under Current (EUC) maximum zonal velocity (eastward) and zonal gradients (Δu between western and eastern equatorial Pacific) were derived from SODA for 1950–2012; extended EUC transport estimates for 1880–2012 were computed (Supplementary Note 4). Cross-correlation and coherence analyses were conducted between δ15N and BioSi annual records.

- Across all 10 filtered, ~200-year annual δ15N records, dominant periodicities are ~25 years, followed by ~10 years (±2 years), with minor 3–8-year variability. Average spectral power for ~25-year period is 2.6 (±0.6) times greater than 3–8-year (ENSO-like) and 1.8 (±0.8) times greater than 10–12-year (NPGO-like) peaks. Variance contributions indicate ~48% PDO-like (20–25 years), 34% NPGO-like (10–12 years), and 18% ENSO-like (3–8 years) influence on deoxygenation variability; PDO-like multidecadal oscillation is the main mode in all records (n=10). - Unfiltered spectra reveal AMO-like periodicity “doublets” centered at ~83 years and 31–35 years in warm periods (Holocene sections #III and #2; Bølling-Allerød sections #9 and #10). During Dansgaard–Oeschger events (IS3), AMO-like periods span ~60–100 years and 31–37 years. AMO-like peak powers are comparable to PDO-like peaks, indicating a substantial AMO contribution to oxygen variability. - During cold periods (Heinrich Event 1 #17; LGM #26), AMO-like doublets are less clear; different multidecadal modes appear (~41 years and <125 years, respectively). δ15N shows larger amplitude swings and lighter values, implying intermittent oxygenated conditions; δ15N then reflects nitrate utilization/availability rather than denitrification (supported by significant negative correlation between denitrification and productivity in section #26). - BioSi vs δ15N dynamics: Band-pass analyses show for δ15N, >10-year periods explain 67.4% (average) of variance versus 32.6% for ≤10 years; conversely, BioSi variability is dominated by ≤10-year periods (54.8% on average). Cross-correlations and coherence reveal no significant relationship (|r| ≤ 0.4) or systematic lag between BioSi and δ15N at annual–decadal scales, indicating distinct controls (local upwelling/productivity vs basin-scale oxygen supply). - Modern reanalyses (1950–2010/12) show AMO+ associates with cooler equatorial SST anomalies and strengthened trades/Walker circulation, but with negative (westward) velocity anomalies at EUC core depth in the east, i.e., a slowdown of the EUC; PDO+ shows weaker, spatially ambiguous subsurface signals. EUC maximum velocity increases during AMO− and decreases during AMO+, with the zonal velocity gradient Δu strengthening in AMO−; estimated EUC transport doubles during AMO− relative to AMO+. These changes would enhance oxygen advection to the eastern Pacific during AMO− and reduce it during AMO+. - Historical context links: Periods of negative AMO (1970s–1990s) coincide with reduced denitrification (improved oxygenation) in ETNP basins and peak EUC strength, while AMOC indices show weakening—supporting a teleconnected Atlantic control on Pacific subsurface oxygen via atmospheric-oceanic coupling. - Overall, PDO and AMO dominate ETNP deoxygenation variability at multidecadal scales; ENSO/NPGO contribute modestly, primarily via nutrient/upwelling pathways rather than oxygen supply.

The findings demonstrate that multidecadal climate modes, particularly PDO- and AMO-like variability, predominantly govern denitrification (and thus deoxygenation) in the ETNP OMZ across diverse climate states over the last 55 ka. ENSO and NPGO exert smaller, higher-frequency influences largely tied to nutrient supply and local upwelling, which do not translate into major subsurface oxygen variability in the GoC. Mechanistically, the AMO modulates atmospheric circulation (ITCZ position, trade winds, Walker circulation) and equatorial Pacific subsurface zonal circulation: during AMO+, intensified trades and equatorial upwelling raise oxidant demand while increased surface westward flow enhances frictional braking of the EUC, slowing eastward oxygen transport and reducing oxygen supply to the eastern Pacific; AMO− reverses these anomalies, increasing EUC transport and oxygen advection, improving oxygenation. The PDO’s influence on STC strength, stratification, and isopycnal heave can partially offset between oxygen supply and consumption pathways, yielding a comparatively modest net effect, but PDO can act constructively or destructively with AMO. These results extend modern observations linking North Atlantic SST variability to Pacific oxygen changes back into the pre-industrial Holocene and Pleistocene, underscoring a dominant Atlantic control on Pacific OMZ oxygenation via teleconnections and subsurface circulation changes.

Annually resolved δ15N records from GoC varved sediments reveal that ETNP deoxygenation variability is dominated by multidecadal modes matching PDO- and AMO-like periodicities, with minor contributions from ENSO/NPGO. Modern reanalyses show AMO phase strongly modulates EUC strength and thus eastern Pacific oxygen supply, providing a mechanistic link from North Atlantic temperature patterns to tropical Pacific deoxygenation. The combined action of AMO and PDO can amplify or mitigate deoxygenation depending on phase alignment. Given rising Northern Hemisphere and North Atlantic temperatures and strengthening trades under ongoing global warming, the study suggests intensified deoxygenation episodes in the Pacific, especially when AMO+ and PDO− co-occur. The results highlight the need to consider Atlantic–Pacific teleconnections and AMO/AMOC variability when projecting future oxygen trends and ecosystem impacts in OMZ regions.

- Instrumental records used for reanalysis (SST, currents, oxygen) span mainly 1950–2012 and are relatively short for capturing low-frequency variability and nonstationarity. - The nature and stability of climate modes (AMO/PDO) and their drivers (internal vs external forcing) remain debated, potentially affecting interpretation of periodicities. - During cold periods (e.g., LGM, Heinrich events), δ15N can reflect nitrate utilization/availability rather than denitrification due to intermittent oxygenation, complicating attribution to oxygen changes. - The 10 annually resolved sections each cover ~160–200 years and are not a single continuous annual record across 55 ka; periodicities > the section length are less constrained. - The GoC is semi-enclosed; although prior work indicates representativeness for the North Pacific OMZ, regional processes could imprint local signals. - Initial band-pass filtering (>70-year removal) precludes detection of the longest periodicities; addressed by subsequent unfiltered analyses but still limited by section length and dating uncertainties.