World Ocean Database bottle casts 1899-2025: one Japanese source of 1-3 m Secchi values causes a false transparency decline; fixed-location Secchi and chlorophyll show no 1 %/yr decline

We parsed all World Ocean Database bottle (OSD) yearly files and kept 439,361 Secchi depths (1899-2018) and 225,899 bottle chlorophyll values at 0-10 m (1934-2025). We compared each measurement type only with itself at the same 1-degree cell and calendar month, with 95% confidence intervals (CI) from a bootstrap over 10-degree blocks. WOD accession 0000416 (Japanese data, 1935-1993) holds 37% of Secchi casts; 99,058 of its values are 1, 2 or 3 m, also in the open ocean with blue water colour, where other sources report a median of 20 m. Their share changes with time, and they alone change the Secchi trend from -1.5%/decade (CI -3.8 to +0.6) to -7.6%/decade (CI -8.8 to -4.5). Without them, the open-ocean Secchi trend is +0.2%/decade (CI -0.4 to +1.0). A 1%/yr chlorophyll decline (Boyce et al. 2010) needs about +4.3%/decade in Secchi depth and -9.6%/decade in chlorophyll; the cleaned Secchi trend and the bottle chlorophyll trend (+3.3%/decade, CI +0.3 to +6.5) exclude these values. When Secchi-derived chlorophyll is blended with bottle chlorophyll to 2008 in 10-degree cells, the median cell trend is -0.50%/yr (CI -0.69 to -0.28), but a separate mean for each measurement type gives +0.24%/yr (CI -0.00 to +0.56); the Falkowski-Wilson conversion gives 1.58 times the bottle value in 18,853 paired casts. Limits: uneven coverage (67% of Secchi casts are Japanese), unknown cause of the 1-3 m values, no correction for bottle chlorophyll method changes, and we do not know whether Boyce et al. used the suspect values.

Secchi depth · chlorophyll · phytoplankton · World Ocean Database · data quality · ocean transparency · Forel-Ule · long-term trend · method bias · fixed effects