“…However, there are multiple sources of uncertainty inherent in transferring the paired Mg/Ca‐δ 18 O measurements into deconvolved time series of temperature, δ 18 O sw , and salinity. Propagating analytical, sampling [ Thirumalai et al ., , ], and calibration errors [ Khider et al ., ] along with age uncertainty [ Trauth et al ., ; Blaauw and Christen , ] into a climatic time series is complex [ Rohling and Bigg , ; Schmidt , ; Rohling , , ] due to the nonlinear nature of the associated Mg/Ca and δ 18 O relationships [ Dekens et al ., ; Anand et al ., ], the potential effect of salinity and postdepositional effects on Mg/Ca [ Ferguson et al ., ; Kısakürek et al ., ; Mathien‐Blard and Bassinot , ; Arbuszewski et al ., ; Dueñas‐Bohórquez et al ., ; Hertzberg and Schmidt , ; Hönisch et al ., ; Khider et al ., ], the effect of ice volume on δ 18 O sw [ Shackleton , ; Mix and Ruddiman , ; Shackleton , ; Elderfield et al ., ], and the potential nonstationarity of the spatially dependent δ 18 O sw ‐salinity relationship [ Rohling and Bigg , ; Bigg and Rohling , ; Leduc et al ., ; Caley and Roche , ; Holloway et al ., ]. Each of these factors can complicate the interpretation of downcore foraminiferal geochemistry, depending on the signal‐to‐noise ratio being reconstructed.…”