BATS bottle records show a 10.8 umol/kg oxygen loss on the 26.5 density surface between 1989-1998 and 2016-2025

We tested whether vertical motion of water layers can explain long-term oxygen change at the Bermuda Atlantic Time-series Study site. We used BCO-DMO bottle data version 10, quality flag 2, and samples within 50 km of the site. We compared 1989-1998 with 2016-2025 with equal calendar-month weights and a three-year block bootstrap. On potential density anomaly 26.5 kg/m3, oxygen fell from 197.88 to 187.12 umol/kg. The change was -10.76 umol/kg, with an approximate 95% interval of -12.42 to -9.20. Equilibrium oxygen increased by 0.62 umol/kg. Apparent oxygen utilization increased by 11.38 umol/kg. The oxygen loss remained with 1994-2003 as the early period (-8.75) and after a paired comparison with 2000 m oxygen (-9.34 for the main periods). Oxygen near 300 m fell by 3.73 umol/kg against 1989-1998, but its change against 1994-2003 was uncertain. A stricter density interpolation rule left incomplete seasonal coverage and gave no valid contrast. Nitrate plus nitrite increased in the main density selection. Under the main sampling rules, an unchanged oxygen-density profile with vertical motion alone cannot explain the fixed-density loss. One site, wide interpolation brackets, missing samples, possible measurement errors and short period windows limit this result. It does not identify ventilation or respiration as the cause. Code, source hashes, aggregate results, simulation draws and figures are included. Source records remain external.

BATS · dissolved oxygen · density surfaces · deoxygenation · apparent oxygen utilization · Sargasso Sea · block bootstrap · sampling coverage