Breakout Session 2 -- Regime Shifts and Decadal Shifts

Regime and decadal scale shifts: can they be detected, what is their impact, are they predictable?

Discussion Leaders: Tom Royer and Anne Hollowed
Participants: Vera Alexander, Tim Baumgartner, Dan Cayan, Bruce Frost, Nick Graham, Scott Hatch, Jim Ingraham, Nate Mantua, Pete Rand, Gary Sharp, Beth Sinclair.

Long term variations in ocean conditions appear to occur at different time scales and the biological responses appear to differ in magnitude. The temporal periods most commonly mentioned are: a) decadal and bi-decadal scale shifts, 6-12 year warm cool eras (Trenberth and Hurrell, 1994; Hollowed and Wooster, 1992), b) cyclic phenomenon such as the 18.6 year pattern in sea surface temperature (Royer, 1993), and c) regime shifts that are 30-60 year cycles and appear to generate measurable ecosystem responses (Francis and Hare, 1994; Baumgartner et. al., 1992, Kawasaki. 1991). Many studies indicate that a marked change in ocean conditions occurred in the late 1970s which some believe was a regime shift, that is, a rapid change from one stable condition to another. Some effort should be devoted to distinguishing the similarities or differences between these three longer term processes.

1) Can they be detected?

Indications of decadal scale and regime shifts might be detected at high latitudes in the northern North Pacific in both the atmosphere and ocean. Trenberth and Hurrell (1994) found the North Pacific winter sea level pressure between 30° and 65°N and 160°E and 140°W shows a mean pressure of about 1010 mb from 1946 to 1977, changing to about 1007 mb from 1977 to 1988 when it changed back to about 1010 mb (Fig. 8a). The timing of blocking marine ridges has changed from being primarily in winter in the early 1970s to primarily in fall in the late 1970s (Salmon, 1992). In contrast to this potential regime shift, the low frequency components of the sea level pressure since 1925 suggest a more cyclic pattern of variability (Trenberth and Hurrell 1994) (Fig. 8b). Cayan (In Prep.) shows pseudo stress (m s-2) anomalies obtained from the differences in states existing before and after the shift of 1976-77 which suggests an intensification of cyclonic winds in the Gulf of Alaska between these two periods (that is a warming) (Fig. 9). Ingraham (Pers. Comm., Alaska Fisheries Science Center, Seattle, WA) used the Ocean Surface Current Simulations (OSCURS) model to generate wind driven surface drift trajectories initiated during winter months (Dec.-Feb.) from Station P for the period 1946 to the present. The endpoints of these 3-month drift trajectories shifted in a bimodal pattern to the north and south around the mean. Thus, the winter flow during each year is persistent enough to, result in a large displacement of surface mixed layer water. The displacement also varies in a decadal pattern. Using the rule that the present mode is maintained until 3 years in a row of the opposite mode occur (Fig. 10) four mode shifts were suggested; a south mode from 1946 to 1956, a north mode from 1957 to 1963, a south mode from 1964-1974, and a north mode 1975 to 1994.

Temperature anomalies in the Gulf of Alaska and Bering Sea illustrate a relative warm period in the late 1950s followed by cooling especially in the early 1970s followed by a rapid temperature increase in the latter part of that decade. Since 1983, the Gulf of Alaska and Bering Sea have undergone different temperature changes. The sea surface temperature in the Gulf of Alaska were generally above normal and those in the Bering Sea were below normal. The temperature differences between the two bodies of water have jumped from about 1.1°C to about 1.9°C (Figure 11). Subsurface temperature anomalies for the coastal Gulf of Alaska (GAK1, 60°N, 149°W) also show a change from the early 1970s into the 1980s similar to that observed in the sea surface (Fig. 12). In addition, high latitude temperature responses to ENSO events can be seen especially at depth in 1977, 1982, 1983, 1987 and in the 1990s.

2) Potential Causes of Climate Decadal Scale Variability

The high latitude ocean temperature variability results from a number of possibly independent physical processes such as the seasonal, 3-4 year and 6-7 year wind stress, El Niño-Southern Oscillation and lunar nodal tidal (18.6 year) signals. Trenberth and Hurrell (1994) argue that the shift in the atmospheric circulation over the North Pacific in the middle 1970s can be attributed to the onset of a period of enhanced El Niño activity. Graham (1994) argues that this shift is more associated with an increase in the background mean SST in the tropical Pacific. A host of studies (e.g., Horel and Wallace 1981) have documented the atmospheric link between the tropical and North Pacific. On the other hand, recent GCM results have shown that the North Pacific ocean-atmosphere climate system can oscillate on time scales of approximately 20 years, independent of tropical forcing (Latif and Barnett 1994). The evidence of these causal mechanisms is principally found in modeling and comparative studies. For example, a proxy of ocean temperature over the last 166 years has been constructed using air temperatures at Sitka, Alaska. This proxy time series contains lower frequency variations that are possibly a response to lunar nodal tide forcing (18.6 year) (Royer, 1993) (Fig. 13). Research that will improve our understanding of potential forcing mechanisms influencing longer term variations are an important topic for U.S. GLOBEC investigations.

3) Biological Impacts of Climate Variability

Decadal scale changes in the distribution and abundance of marine populations have occurred coincident with shifts in the physical environment. Francis et al. (in review) review the impacts of the most recent regime shift through lower, secondary and top trophic levels of the North Pacific marine ecosystem. Some of the following summaries are taken from this review.

Population trends of zooplankton in the Subarctic Pacific and California Current show opposite trends in recent years. Evidence of lower trophic level responses to decadal scale climate change is provided in the increase of subarctic zooplankton abundance from the 1950s to the 1980s reported by Brodeur and Ware (1992) (Fig. 14). Roemmich and McGowan (1995) used the CalCOFI time series to document declines in zooplankton biomass between the 1950s and 1980s off southern California.

Abundances of higher trophic level consumers are available for commercially important fish and shellfish stocks (shrimp and crab). Most of the biological time series for these species are comparatively short (they span only two physical states: cool and warm). Notable exceptions are the Pacific halibut and Pacific herring recruitment time series, the Pacific salmon catch, and northern fur seal population data collected on rookeries. These time series extend back to the early 1900s. Parker et. al. (1995) show marked similarities between time series of the lunar nodal tidal cycle, and recruitment patterns of Pacific halibut. Hollowed and Wooster (1995) examined time series of marine fish recruitment and observed that some marine fish stocks exhibited an apparent preference (measured by the probability of strong year classes and average production of recruits during the period) for a given climatic regime. Hare and Francis (1995) found a striking similarity between large scale atmospheric conditions and salmon production in Alaska. Quinn and Niebauer (1995) studied the Bering Sea pollock population and found that high recruitment coincided with years of warm ocean conditions (above normal air and bottom temperatures and reduced ice cover). This fit was improved by accounting for density dependent processes. Livingston and Methot (in review) estimated the abundance of age-1 pollock in the Bering Sea before predation (primarily cannibalism) occurred. Using this abundance index, a marked shift in the mean recruitment was observed before and after the regime shift (Fig. 15).

On a larger scale, evidence of biological responses to decadal scale changes in climate are also found in the coincidence of global fishery expansions or collapses of similar species complexes. Examples include the recent increase in the South American and Japanese sardine stocks and their subsequent fisheries expansions (Kawasaki 1991).

Climate effects on seabirds are primarily indirect, affecting the availability of preferred prey and composition of marine fish stocks. Attempts to correlate seabird responses (breeding productivity in particular) with simple physical parameters such as sea surface temperature have given mixed results. Hunt et. al. (in press) provide an example of the potential impact of climate change on prey availability of seabirds on the Pribilof Islands in the Bering Sea. These authors found the proportion of age-1 walleye pollock in seabird diets decreased between the 1970s and 1980s, while during the same period, the proportion of age-1 pollock in NMFS bottom trawl hauls decreased in the vicinity of the Pribilof Islands. Since the decrease in age-1 pollock in the diet was observed in both surface and subsurface feeding seabirds, the authors conclude that the change resulted from a horizontal change in the distribution of age-1 pollock in the vicinity of the Pribilof Islands.

Hatch and Sanger (1992) and Springer et. al. (1986) examined the potential impact of changes in the abundance of seabird prey. Springer et. al. (1986) found the reproductive success of kittiwakes in the Bering Sea was related to indices of juvenile pollock abundance. However, more recent analyses of Decker et al. (1995) and Hunt et al. (1995) suggest that the availability of forage fish is more important than the abundance of pollock in determining the reproductive success of kittiwakes on the Pribilof Islands. Hatch and Sanger (1992) found the importance of juvenile pollock in the diet of tufted puffins in the Gulf of Alaska was positively related to estimates of yearÐclass strength of juvenile pollock.

Piatt and Anderson (1995) provide evidence of possible changes in prey abundance due to decadal scale climate shifts. These authors examine relationships between significant declines in marine birds in the northern Gulf of Alaska during the past 20 years and found significant declines in common murre populations occurred between the mid to late 1970s and the early 1990s. Piatt and Anderson (1995) found marked changes in diet composition of five seabird species collected in the Gulf of Alaska between 1975-78 and 1988-91. There was a shift in diet from one dominated by capelin in the late 1970s to one where capelin was virtually absent in the later period.

Evidence of direct linkages between changes in the physical environment and survival of marine organisms exist for several species in the North Pacific and Bering Sea. For example, York (1991) examined the relationship between sea surface temperature and survival of male northern fur seals from the Pribilof Islands. After accounting for age-specific harvest of fur seals, York's analysis showed strong correlations between SST and survival. She suggested that the relationship could be related to the influence of SST on the availability of food for young fur seals.

Pinnipeds are considered specialized generalists feeding on a variety of prey types with notable preferences for particular types of prey. If the number of alternative prey types is low, then prey availability may have a significant impact on survival. The species composition of the diet of marine mammals before and after the late-1970s regime shift was different. Merrick and Calkins (in press) examined the prey items of Steller sea lions during 1975-1978 and 1985-1986. This study showed walleye pollock were the most common prey item in the diet in virtually all seasons and areas of the Gulf of Alaska before and after the regime shift. Their work showed an increase in the proportion of sampled animals that consumed pollock from the 1970's to the 1980's despite a decrease in the abundance of 2-3 year-old pollock. They show that while juvenile pollock declined in abundance, a simultaneous decline in the abundance of alternative prey may have occurred. They suggest that this simultaneous decline in abundance may have contributed to sea lion population declines in this region.

4) Paleoceanographic Record

Paleoceanographic data are an important element of research on longer term climate affects on marine ecosystems. High-resolution paleoceanographic data from sites with detailed chronologies of annual to near-annual precision provide a long-term historical context of the interdecadal variability in ecologically and economically important populations as well as the background ocean environment. The record of the Santa Barbara Basin, off southern California is the most complete so far, yielding a history of the sardine, anchovy and hake populations slightly longer than the past 1500 years. Currently under development are time series of oxygen isotope history from planktonic foraminifera (for ocean temperatures), analysis of the diatom and radiolarian assemblages, as well as radiocarbon measurements of a planktonic pteropod as a measure of the age of the near-surface waters (an indication of large-scale upwelling). The 1500-year record of the sardines and anchovies supports the existence of basin-scale interdecadal shifts in ocean climate. A major objective of the investigation of these records is to determine the characteristic time scales of variability to aid in understanding the nature of the decadal scale regimes including the biological response, and to likewise aid in predicting their duration.

Although sites such as the Santa Barbara Basin which can be used for high-resolution paleoceanographic reconstruction are rare (resulting from anaerobic bottom conditions found along continental margins), the interest generated by such programs as U.S. GLOBEC and PICES has stimulated the search for sites in the Subarctic Pacific so that the information from the Santa Barbara Basin can be linked to a more complete geographic network which we hope will include the region of the Alaska Peninsula (Skan Bay, Unalaska Is.) plus selected fjords of British Columbia (e.g., west coast of Vancouver Is.) in which signals from the open ocean may be preserved. The Skan Bay site has been confirmed but not exploited, and a cruise is planned for December, 1995, to explore promising sites off Barkley and Nootka Sounds on Vancouver Island. The Vancouver Island sites are of particular interest because of their potential to register the northern expansion of the Pacific sardine during periods of warming along the west coast of North America. Such an expansion occurred during the 1930s when sardines were abundant from Baja California, Mexico, to Vancouver Island, and a significant fishery was established along the west coast of the island.

5) Need for Additional Research

The previous section presented compelling evidence of a coupling between large scale, multi-year climate variability and ecosystem responses. However, additional research is required to improve our understanding of the mechanisms underlying regime shifts in the physical environment, and the response of the ecosystem to regime shifts. Knowledge of the functional relationships between climate forcing and the physical and biological system is needed.

If the past reflects the future, it is possible to speculate potential impacts of future changes in climate on marine ecosystems. If the climatic regime returned to a state similar to that observed in the early 1970s, one might expect zooplankton production in the Subarctic Gyre might move offshore and overall production would decrease. A marked shift in the production of commercially exploited fish stocks in the North Pacific would be expected. Alaskan salmon and Bering Sea pollock production would be expected to decrease in the cooler regime.

6) Are they predictable and can they be simulated?

Indications of decadal scale shifts might be detected in time series of some or all of the following physical variables: sea level pressure, sea level pressure gradients, upper level pressures and gradients, cloudiness, turbidity, precipitation, sea surface temperature, temperature gradients, sea surface salinity, stratification, sea level, coastal and altimeter measurements of sea height, currents, transport and mixed layer depth.

Biological indicators of climate change include: abrupt changes in recruitment or in the frequency of extreme year classes, changes in the vertical or horizontal distribution of pelagic species, changes in trophic phasing (match-mismatch), changes in species dominance or species mix, changes in somatic growth, changes in bioenergetics, shifts in predator and/or prey.

North Pacific basin physical modeling results show promise in simulating and explaining monthly and decadal fluctuations of the thermal and dynamic properties over the mid-latitudes over relatively coarse scales. However, elements of the freshwater flux forcing (evaporation-precipitation, runoff) have not been incorporated and validated. These elements affect the mixed layer and are key in the structure of the upper layers in the subarctic region north of 40°N. Modeling efforts of the 1970-88 period by Miller (1994) successfully simulated the 1976-77 shift in the upper ocean temperature over the Pacific. Diagnosis of the model results show that this shift was a basin-wide phenomenon which occurred as a transition between two relatively stable but distinct regimes of ocean-atmosphere coupling. This change produced a reorganization of the structure and circulation of both the ocean and atmosphere. Models such as this may be useful in predicting future changes in the North Pacific system. Regional and mesoscale ocean models that are either driven directly by observed forcing or nested within larger scale ocean general circulation models (GCM) have been developed for the Gulf of Alaska through the FOCI program but they need to be developed in other regions of the North Pacific and expanded to include other species. Short term projections (1-2 years) of fish production based on observed ocean characteristics show promising results (Megrey et. al. in review).