This is the author's manuscript for a work that has been accepted for publication. Changes resulting from the publishing process, such as copyediting, final layout, and pagination, may not be reflected in this document. The publisher takes permanent responsibility for the work. Content and layout follow publisher's submission requirements.

Empirical statistical downscaling (ESD) methods seek to refine global climate model (GCM) outputs via processes that glean information from a combination of observations and GCM simulations. They aim to create value-added climate projections by reducing biases and adding finer spatial detail. Analysis techniques, such as cross-validation, allow assessments of how well ESD methods meet these goals during observational periods. However, the extent to which an ESD method's skill might differ when applied to future climate projections cannot be assessed readily in the same manner. Here we present a "perfect model" experimental design that quantifies aspects of ESD method performance for both historical and late 21st century time periods. The experimental design tests a key stationarity assumption inherent to ESD methods -namely, that ESD performance when applied to future projections is similar to that during the observational training period. Case study results employing a single ESD method (an Asynchronous Regional Regression Model variant) and climate variable (daily maximum temperature) demonstrate that violations of the stationarity assumption can vary geographically, seasonally, and with the amount of projected climate change. For the ESD method tested, the greatest challenges in downscaling daily maximum temperature projections are revealed to occur along coasts, in summer, and under conditions of Climatic Change (2016)

Spatial distributions of marine fauna are determined by complex interactions between environmental conditions and animal behaviors. As climate change leads to warmer, more acidic, and less oxygenated oceans, species are shifting away from their historical distribution ranges, and these trends are expected to continue into the future. Correlative Species Distribution Models (SDMs) can be used to project future habitat extent for marine species, with many different statistical methods available. However, it is vital to assess how different statistical methods behave under novel environmental conditions before using these models for management advice, and to consider whether future projections based on these techniques are biologically reasonable. In this study, we built SDMs for adults and larvae of two ecologically important pelagic fishes in the California Current System (CCS): Pacific sardine (Sardinops sagax) and northern anchovy (Engraulis mordax). We used five different SDM methods, ranging from simple [thermal niche model (TNM)] to complex (artificial neural networks). Our results show that some SDMs trained on data collected between 2003 and 2013 lost substantial predictive skill when applied to observations from more recent years, when ocean temperatures associated with a marine heatwave were outside the range of historical measurements. This decrease in skill was particularly apparent for adult sardine, which showed non-stationary relationships between catch locations and sea surface temperature (SST) through time. While sardine adults and larvae shifted their distributions markedly during the marine heatwave, anchovy largely maintained their historical spatiotemporal distributions. Our results suggest that correlative relationships between species and their environment can become unreliable during anomalous conditions. Understanding the underlying physiology of marine species is therefore

No abstract

No abstract