Interpretation and design of ocean acidification experiments in upwelling systems in the context of carbonate chemistry co-variation with temperature and oxygen

Coastal upwelling regimes are some of the most productive ecosystems in the ocean but are also among the most vulnerable to ocean acidification (OA) due to naturally high background concentrations of CO2. Yet our ability to predict how these ecosystems will respond to additional CO2 resulting from anthropogenic emissions is poor. To help address this uncertainty, researchers perform manipulative experiments where biological responses are evaluated across different CO2 partial pressure (pCO(2)) levels. In upwelling systems, however, contemporary carbonate chemistry variability remains only partly characterized and patterns of co-variation with other biologically important variables such as temperature and oxygen are just beginning to be explored in the context of OA experimental design. If co-variation among variables is prevalent, researchers risk performing OA experiments with control conditions that are not experienced by the focal species, potentially diminishing the ecological relevance of the experiment. Here, we synthesized a large carbonate chemistry dataset that consists of carbonate chemistry, temperature, and oxygen measurements from multiple moorings and ship-based sampling campaigns from the California Current Ecosystem (CCE), and includes fjord and tidal estuaries and open coastal waters. We evaluated patterns of pCO(2) variability and highlight important co-variation between pCO(2), temperature, and oxygen. We subsequently compared environmental pCO(2)-temperature measurements with conditions maintained in OA experiments that used organisms from the CCE. By drawing such comparisons, researchers can gain insight into the ecological relevance of previously published OA experiments, but also identify species or life history stages that may already be influenced by contemporary carbonate chemistry conditions. We illustrate the implications co-variation among environmental variables can have for the interpretation of OA experimental results and suggest an approach for designing experiments with pCO(2) levels that better reflect OA hypotheses while simultaneously recognizing natural co-variation with other biologically relevant variables.