Bering Sea sea ice during winter 2017–2018 was the lowest ever recorded. Ecosystem effects of low ice have been observed in the southeastern Bering Sea, but never in the northern Bering Sea. Observations in both systems included weakened water column stratification, delayed spring bloom, and low abundances of large crustacean zooplankton. Summer Cold Pool presence was extremely limited. Young walleye pollock production and condition were similar to prior warm years, though catches of other pelagic forage fishes were low. Summer seabird die‐offs were observed in the northern Bering Sea, and to lesser extent in the southeastern Bering Sea, and reproductive success was poor at monitored colonies. Selected bottom‐up responses to lack of sea ice in the north were similar to those in the south, potentially providing environmental indicators to project ecosystem effects in a lesser studied system. Results offer a potential glimpse of the broader Bering Sea pelagic ecosystem under future low‐ice projections.

In 2014, the Bering Sea shifted back to warmer ocean temperatures (+2 oC above average), bringing concern for the potential for a new warm stanza and broad biological and ecological cascading effects. In 2015 and 2016 dedicated surveys were executed to study the progression of ocean heating and ecosystem response. We describe ecosystem response to multiple, consecutive years of ocean warming and offer perspective on the broader impacts. Ecosystem changes observed include reduced spring phytoplankton biomass over the southeast Bering Sea shelf relative to the north, lower abundances of large-bodied crustacean zooplankton taxa, and degraded feeding and body condition of age-0 walleye pollock. This suggests poor ecosystem conditions for young pollock production and the risk of significant decline in the number of pollock available to the pollock fishery in 2–3 years. However, we also noted that high quality prey, large copepods and euphausiids, and lower temperatures in the north may have provided a refuge from poor conditions over the southern shelf, potentially buffering the impact of a sequential-year warm stanza on the Bering Sea pollock population. We offer the hypothesis that juvenile (age-0, age-1) pollock may buffer deleterious warm stanza effects by either utilizing high productivity waters associated with the strong, northerly Cold Pool, as a refuge from the warm, low production areas of the southern shelf, or by exploiting alternative prey over the southern shelf. We show that in 2015, the ocean waters influenced by spring sea ice (the Cold Pool) supported robust phytoplankton biomass (spring) comprised of centric diatom chains, a crustacean copepod community comprised of large-bodied taxa (spring, summer), and a large aggregation of midwater fishes, potentially young pollock. In this manner, the Cold Pool may have acted as a trophic refuge in that year. The few age-0 pollock occurring over the southeast shelf consumed high numbers of euphausiids which may have provided a high quality alternate prey. In 2016 a retracted Cold Pool precluded significant refuging in the north, though pollock foraging on available euphausiids over the southern shelf may have mitigated the effect of warm waters and reduced large availability of large copepods. This work presents the hypothesis that, in the short term, juvenile pollock can mitigate the drastic impacts of sustained warming. This short-term buffering, combined with recent observations (2017) of renewed sea ice presence over southeast Bering Sea shelf and a potential return to average or at least cooler ecosystem conditions, suggests that recent warm year stanza (2014–2016) effects to the pollock population and fishery may be mitigated.

From 2007 to 2013, the southeastern Bering Sea was dominated by extensive sea ice and below-average ocean temperatures. In 2014 there was a shift to reduced sea ice on the southern shelf and above-average ocean temperatures. These conditions continued in 2015 and 2016. During these three years, the spring bloom at mooring site M4 (57.9°N, 168.9°W) occurred primarily in May, which is typical of years without sea ice. At mooring site M2 (56.9°N, 164.1°W) the spring bloom occurred earlier especially in 2016. Higher chlorophyll fluorescence was observed at M4 than at M2. In addition, these three warm years continued the pattern near St. Matthew Island of high concentrations (>1 μM) of nitrite occurring during summer in warm years. Historically, the dominant parameters controlling sea-ice extent are winds and air temperature, with the persistence of frigid, northerly winds in winter and spring resulting in extensive ice. After mid-March 2014 and 2016 there were no cold northerly or northeasterly winds. Cold northerly winds persisted into mid-April in 2015, but did not result in extensive sea ice south of 58°N. The apparent mechanism that helped limit ice on the southeastern shelf was the strong advection of warm water from the Gulf of Alaska through Unimak Pass. This pattern has been uncommon, occurring in only one other year (2003) in a 37-year record of estimated transport through Unimak Pass. During years with no sea ice on the southern shelf (e.g. 2001–2005, 2014–2016), the depth-averaged temperature there was correlated to the previous summers ocean temperature.