Emergence of multiple ocean ecosystem drivers in a large ensemble suite with an Earth system model

摘要

Marine ecosystems are increasingly stressed by human-induced changes. Marineecosystem drivers that contribute to stressing ecosystems – includingwarming, acidification, deoxygenation and perturbations to biologicalproductivity – can co-occur in space and time, but detecting their trends iscomplicated by the presence of noise associated with natural variability inthe climate system. Here we use large initial-condition ensemble simulationswith an Earth system model under a historical/RCP8.5 (representativeconcentration pathway 8.5) scenario over 1950–2100 to consider emergencecharacteristics for the four individual and combined drivers. Using a1-standard-deviation (67% confidence) threshold of signal to noise todefine emergence with a 30-year trend window, we show that oceanacidification emerges much earlier than other drivers, namely during the20th century over most of the global ocean. For biological productivity, theanthropogenic signal does not emerge from the noise over most of the globalocean before the end of the 21st century. The early emergence pattern for seasurface temperature in low latitudes is reversed from that of subsurfaceoxygen inventories, where emergence occurs earlier in the Southern Ocean. Forthe combined multiple-driver field, 41% of the global ocean exhibitsemergence for the 2005–2014 period, and 63% for the 2075–2084 period.The combined multiple-driver field reveals emergence patterns by the end ofthis century that are relatively high over much of the Southern Ocean, NorthPacific, and Atlantic, but relatively low over the tropics and the SouthPacific. For the case of two drivers, the tropics including habitats of coralreefs emerges earliest, with this driven by the joint effects ofacidification and warming. It is precisely in the regions with pronouncedemergence characteristics where marine ecosystems may be expected to bepushed outside of their comfort zone determined by the degree of naturalbackground variability to which they are adapted. The results underscore theimportance of sustained multi-decadal observing systems for monitoringmultiple ecosystems drivers.