David Greenland

When temporally smoothed data are used for the period 1925 to 1985 there is a close inverse statistical relationship acting at an inter-decadal timescale between the Pacific Northwest (PNW) air temperatures and Coho salmon catch off the coast of Washington and Oregon. This relationship is now well known, although not fully explained, but at the time of its discovery in 1994 it was part of advances being made by several research groups on interdecadal-scale climate/ecological changes in the PNW (Greenland 1995). The discovery and later, related findings may be usefully examined within the context of the framework questions of this book (see chapter 1) because it provides a very interesting example of climate variability and ecosystem response found, in part, by Long-Term Ecological Research (LTER) investigators. The logical progression for this chapter is first to review a little of the relationship between Coho salmon and climate and then to explain how a study at one LTER site led to a finding with regional implications. An update of the findings at interdecadal-scale climate/ecological changes in the PNW is then appropriate, followed by a discussion of the topic with the framework questions of this book. The PNW is defined, for the purposes of this chapter, as the area of Washington and Oregon west of the crest of the Cascade Range. The term decadal is used loosely in this chapter to refer to changes that focus on time periods of about 10 to 30 years in length.

Salmon and Climate

Salmon live part of their lives in terrestrial, freshwater environments and part in marine, saltwater environments. The salmon life history starts with fertilized eggs remaining in gravel in freshwater stream beds and hatching after 1-3 months. One to five months later, fry emerge in the spring or summer. Juvenile fish are in freshwater from a few days to 4 years, depending on species and locality. After the juveniles change to smolts, they can migrate to the ocean, usually in spring or early summer, often taking advantage of streamflows driven by snowmelt. The fish spend 1-4 years in the ocean and then return to their freshwater home stream to spawn and die. More specifically, the typical life cycle for Oregon Coho spans 3 years (18 months in freshwater and 18 months in the ocean). Climatic factors affect the fish at all stages of their fascinating life history, and Greenland (1995) has reviewed some of these factors. However, it is believed that salmon are most vulnerable to climate variations when they are at the migrating smolt stage (Pearcy 1992). This stage is also where decadal-scale climate variability plays a role.

There are many other factors besides climate that control salmon survival rates. Indeed the "cyclic" variability described here is played out against a background of a century-long decrease in the salmon population in the PNW—a decrease resulting mainly from a suite of direct and indirect human-caused factors. The following account relates only to Coho salmon, although somewhat similar or inverse patterns have also been found with Sockeye, Pink, and Chinook salmon in, and north of, the PNW (Mantua et al. 1997). Where the term salmon is used herein without a qualifier, the reference is implicitly to Coho salmon, although the point being made may well apply to other species as well.

We must first examine how research results at an individual LTER site give rise to further discoveries at a regional scale.

From the Andrews Forest to the Pacific Ocean

An important part of the LTER program is its interdisciplinary nature and how research findings in one subdiscipline stimulate discoveries in another. This connectivity was particularly important in the results reported in this chapter, and the steps of the connection are worth recording as well as the relationship to which they led.

In the early 1990s, I completed an introductory analysis of the climate of the H. J. Andrews LTER site (HJA) that is located in the foothills of the Cascade Range in Oregon (Greenland 1994; see also chapter 19). Several times I presented the results of the climate analysis to the HJA LTER research group. One of the results of my climate study indicated the relationship of the El Niño-Southern Oscillation (ENSO) phenomenon to the Andrews climate. One graphic showed that the value of the Southern Oscillation Index (SOI) for 1982-1983 was markedly displaced from an otherwise clear linear relationship between SOI and winter water year precipitation. This graphic reminded Dr. Stanley Gregory, a stream ecologist, of an analogous graphic of the relationship between the number of previous year Coho jacks (early maturers) versus an overall Oregon Production Index (Pearcy 1992, fig. 5.5). The 1982-1983 data point was also an outlier in the Coho jack relationship. Outliers in both graphs are a function of the very intense El Niño event of that year.

I was then asked to present papers at Fishery Science meetings on the relationship between climate and salmon within the context of the effect of climate on, and the linkages between, the terrestrial and oceanic part of the salmon habitat. In preparation for these presentations, I found an important paper by Frances and Sibley (1991) on decadal variation of Coho salmon catch in the northeastern Pacific Ocean. Frances and Sibley (1991) had previously reported a close relationship between winter (November to March) air temperatures at Sitka, Alaska, and surface water temperature at Langara Spit, Queen Charlotte Island, British Columbia, and the catch of Pink salmon in the Gulf of Alaska. They had also reported an inverse relationship between the catch of Pink salmon in the Gulf of Alaska and the catch of Coho salmon off the coast of Washington and Oregon. Both relationships covered the period 1925 to 1985 and were found when the data were normalized and subjected to a 7-year weighted filter (R. C. Frances, pers. comm., 1994). Given these relationships I reasoned that it would be likely that there would be an inverse relationship between air temperatures in the PNW and the catch of Coho salmon off the coast of Washington and Oregon.

I used a 5-year unweighted filter of the annual mean air temperatures at the H. J. Andrews Long-Term Ecological Research site. The filter was applied to values of Andrews' temperatures that were normalized to the long-term mean for the 19251985 period. It had been shown elsewhere that the Andrews' temperatures are well related to those of western Oregon in general (Greenland 1994). To a large extent the temperatures of western Oregon are also related to those of the broader PNW region. Coho Salmon catch data were extracted from the graphs of Frances and Sibley and a close inverse relationship was indeed found (figure 13.1). This relationship and time series suggest an approximate 20 year "cycle" in both air temperatures in the PNW and the catch of Coho salmon off the coast of Washington and Oregon. The term "cycle" is placed in quotes because the data sets are not of sufficient length to determine the existence of true cycles. However, recent tree-ring studies by Biondi et al. (2001) for the southwestern United States and Gedalof and Smith (2001) for northwestern North America have extended the length of the time series to about 400 years. A similar decadal-scale variability shows up throughout this 400-year record. The interdecadal variation in the tree rings of Interior Alaska (chapter 12) are also related to PDO changes.

In summary, when temporally smoothed data are used for the period 1925 to 1985, there is a close inverse statistical relationship between the PNW air temperatures and Coho salmon catch off the coast of Washington and Oregon. To attempt to explain this inverse relationship, we must explore the large spatial and long-term aspects of atmospheric and ocean currents as well as further thinking of Frances and Sibley and their colleagues. Although there are biophysical variations of a variety of time and space scales that affect salmon catch, we are mostly interested here in the decadal-scale changes that are inevitably linked with Pacific basinwide spatial variations in the atmosphere and ocean.

A Possible Cascade

A cascading system is generally regarded as one that exhibits flow of material, energy, or information. The following cascade is proposed to partially explain the re-

Figure 13.1 Five-year moving average of annual mean temperature at the H. J. Andrews LTER site (open circles) and seven-year moving average of Coho salmon catch off the coast of Washington and Oregon (filled circles). With permission from the California Department of Water Resources. (Salmon data from Frances and Sibley, 1991.)

Figure 13.1 Five-year moving average of annual mean temperature at the H. J. Andrews LTER site (open circles) and seven-year moving average of Coho salmon catch off the coast of Washington and Oregon (filled circles). With permission from the California Department of Water Resources. (Salmon data from Frances and Sibley, 1991.)

lationship between the PNW air temperatures and Coho salmon catch off the coast of Washington and Oregon and the resulting variability at the interdecadal timescale. A five-level cascade actually starts with a coupled ocean/atmosphere variability now called the Pacific Decadal Oscillation (PDO) (Level 1). The two major modes of the PDO affect the air circulation over the northeastern Pacific Ocean area, and this leads to changes in the direction of the air currents and the intensity of the ocean currents (Level 2). These changes in the direction of the air currents give rise to the changes in the air temperatures of the PNW. The air temperatures are a by-product in this cascade and have little to do with salmon catch. It is hypothesized that the operation of the ocean current leads to the provision of greater or lesser nutrients (Level 3) that in turn work their way up the food chain and eventually result in a variable abundance of salmon (Level 4) and subsequent salmon catch (Level 5). There can be several more levels in this cascade, depending on the degree of detail at which the food chain is resolved.

Attempts to specify the system cascade in such a way have the advantage, among other things, of identifying topics that require further investigation. In this cascade the atmospheric changes are known in some detail. The related provision, or lack of provision, of nutrient-rich ocean water under different atmospheric conditions also seems understandable, although it is known in less detail. However, there are major uncertainties about the linkages between the lower and upper parts of the food chain, as well as aspects of salmon ecology such as the comparative role of wild and hatchery-bred salmon. Besides uncertainties in food chain linkage, there are other possibilities for the "cascade." For example, an alternative explanation is that changes in the physical habitat, such as temperature changes, alter the distribution and effectiveness of predators as suggested by Fisher and Pearcy (1988). The establishment and further development of this cascade provides us with a fertile research agenda.

Advances in Atmospheric and Fisheries Science

Since the link between the climate at HJA and Coho fish catch off the coast of the PNW was discovered, there have been many advances in our knowledge of the climate and fisheries of the region.

At the outset it must be explained that the basic geography of surface ocean currents in the northeastern Pacific Ocean consists first of the subarctic current moving water eastward across the North Pacific Ocean at about 45-50° N, the approximate latitude of the cities of Seattle and Vancouver. As the water nears the North American continent, it bifurcates. Part of it flows northward near the Canadian and Alaskan coastlines and into the Gulf of Alaska to help form the Alaskan gyre. Another part of the water from the subarctic current flows southward, forming the Californian current.

Even at the time that Frances and Sibley were reviewing the inverse relation between Gulf of Alaska and PNW salmon catch, Frances (1993) quoted the ideas of Hollowed and Wooster (1991) to explain the oceanographic and atmospheric differences in the two modes of the (then unnamed) PDO. It was argued that the inverse relationship in catch in the two ocean areas might be explained by the north or south movement of the divergence, or bifurcation, zone between the Alaskan and California currents. The suggested model is bimodal and postulates two states or modes of operation of the ocean currents (figure 13.2). When the bifurcation zone is more to the north (PDO cold phase), the Aleutian low-pressure zone in the atmosphere is weak and more cold subarctic current water is taken into the Cali-fornian current and the upwelling of nutrient-rich water off the Oregon and Washington coasts is enhanced. When the bifurcation zone is farther to the south (PDO warm phase), the Aleutian low-pressure zone is deep, and it swings winds, rain-bearing storms, and more cold subarctic current water into the Alaskan current, leaving the water off the Oregon and Washington coasts relatively warmer. The reality is more complex than simply changing water and air temperatures by altering water transport. The Aleutian Low pressure zone forces changes the upper ocean temperatures in the northeastern Pacific by a combination of surface heat fluxes, vertical mixing, and Ekman transports (Miller et al. 1994).

The Hollowed and Wooster model may be placed in a larger atmospheric context by noting its relationship to the synoptic climatology and teleconnective index values (figure 13.2). What later became known as the PDO cold phase is associated with a weak Aleutian low pressure with winds coming more directly from the west

PDO - Cold Phase

Weak Aleutian Law. High CNP. Negative PNA. Northerly current bifurcation. El Nino unlikely. Negative W Coast SST Anomalies

Weak Aleutian Law. High CNP. Negative PNA. Northerly current bifurcation. El Nino unlikely. Negative W Coast SST Anomalies

PDO - Warm Phase

Strong Aleutian Low. Low CNP Positive PNA. Southerly current bifurcation. El Nino possible. Positive W Coast SST Anomalies

PDO - Warm Phase

Strong Aleutian Low. Low CNP Positive PNA. Southerly current bifurcation. El Nino possible. Positive W Coast SST Anomalies

Figure 13.2 A schematic of the surface atmospheric pressure distribution accompanying the two main modes of the Pacific Decadal Oscillation. Solid black lines represent isobars. Gray arrows represent hypothesized accompanying bifurcation of the subarctic ocean current into the Alaskan current and the Californian current.

Figure 13.2 A schematic of the surface atmospheric pressure distribution accompanying the two main modes of the Pacific Decadal Oscillation. Solid black lines represent isobars. Gray arrows represent hypothesized accompanying bifurcation of the subarctic ocean current into the Alaskan current and the Californian current.

across the Pacific at the latitudes of Washington and Oregon. The more northerly bifurcation of the subarctic current pushes more water into the Californian current and gives rise to negative sea surface temperature (SST) anomalies. These circumstances are also associated with surface offshore wind flow and ocean upwelling that brings food sources to the marine food chain near the coast. These conditions are not generally consistent with intense El Niño conditions. At the opposite mode of the suggested model, the PDO warm phase is associated with a strong Aleutian low pressure and enhanced southwesterly winds in the northeast Pacific. The more southerly subarctic current bifurcation enhances northward ocean flow into the Alaskan current, giving rise to positive SST anomalies in the eastern part of the northeastern Pacific. These conditions are consistent with the results of El Niño events. The relative values of other atmospheric indices of pressure distribution and implied atmospheric motion such as the Central North Pacific (CNP) index and the Pacific North American (PNA) index are also shown in figure 13.2.

The most important discovery was the codification of the interdecadal-scale climate regime shifts in the northeastern Pacific Ocean and the northwestern North American continent now known to be related to changes in the ocean and named the Pacific Decadal Oscillation (PDO). University of Washington researchers noted that the PDO was in a cool phase from about 1900 to 1925 and from 1945 to 1977 (Mantua et al. 1997). The PDO was in a warm phase from 1925 to 1945 and after 1977. Another phase change may have occurred in the mid-1990s. Additional work has attempted to identify the terminal points of the regime shifts. Overland et al. (1999) found that since the turn of the century, 37% of the winter interannual variance of the Aleutian low is at timescales greater than 5 years. An objective algorithm detected zero crossings of Aleutian low central pressure anomalies in 1925, 1931, 1939, 1947, 1959, 1968, 1976, and 1989. Ware and Thomson (2000) noted that the climate of the northeastern Pacific Ocean has oscillated at three dominant timescales over the last 400 years: the well-known 2- to 8-year El Niño-Southern Oscillation (ENSO) timescale, a 20- to 40-year interdecadal timescale, and a 60- to 80-year multidecadal timescale. The latter oscillation has been the dominant mode of air temperature variability along the west coast of North America during the last 400 years. During this period, there have been conspicuous temporal modulations of the ENSO and the interdecadal signals. Low-frequency temperature oscillations at periods greater than 10 years in the northeast Pacific have been significantly coherent and in-phase from southern California to British Columbia. However, with the exception of the ENSO signal, higher frequency variability has been weakly coherent along the west coast. Biondi et al. (2001) found the PDO is closely matched by the dominant mode of tree-ring variability that provides a preliminary view of multiannual climate fluctuations spanning the past four centuries. A reconstructed PDO index features a prominent bidecadal oscillation, whose amplitude weakened in the late l700s to mid-1800s. A comparison with proxy records of ENSO suggests that the greatest decadal-scale oscillations in Pacific climate between 1706 and 1977 occurred around 1750, 1905, and 1947.

Undoubtedly, more investigation of the various timescales of variability will be made, but the focus in this account remains on the decadal-scale variability. Studies at this scale have extended our knowledge further. Climatic factors influence the type, distribution, and abundance of predators of young salmon. During warm coastal ocean years (PDO and ENSO warm phase), migrating predatory fish such as Pacific hake and Pacific mackerel arrive in the coastal ocean of the PNW earlier in the year and are closer to the shore (JISAO CIG 1999). Large changes in abundance of predators and prey, as well as changes in species composition of the zooplankton community off Newport, Oregon, and Vancouver Island, have been reported in recent years (Peterson and Mackas 2001). Stratification of ocean water has also been noted as a factor. The lower abundance of salmon in the PNW during warm phases of the PDO is believed to have occurred in more stratified coastal ocean water that is less nutrient-rich from upwelling during this phase. University of Washington scientists have also suggested why the decadal-scale relationship between climate and salmon abundance is stronger than the year-to-year ENSO relation (Hare et al. 1999). First, because not all fish spend the same number of years in the ocean, the year classes that may have been affected differently by interannual climate variation are smeared together. Second, an individual fish may feel beneficial effects of ENSO one year and detrimental effects during the opposite ENSO phase, but the PDO tends to have year-to-year persistence.

Overall, the more we learn about these decadal-scale sets of events, the more fascinating they become.

The Framework Questions and Future Research

I now address some of the framework questions of the book (see chapter 1). We have already seen in this chapter how the climate phenomenon of the PDO gives rise to a cascade of ecological events, resulting in greater or lesser fish abundance. The cascading principle is thus directly applicable in this example. However, there are other events taking place outside the principal coastal ocean cascade that may play a role in the overall system. For example, the changing availability of nutrients in the coastal ocean currents that is affected by changes from one PDO mode to the other is only one of the manifestations of the PDO modal shift. Another important resulting change is that, during the PDO cold phase, terrestrial snowpacks and later meltwater are relatively high, causing bountiful stream water flow conditions in the PNW for salmon during their downstream migration. The opposite is true during the PDO warm phase. Thus, there is also a terrestrial component acting outside the coastal ocean cascade that is the focus of this account. A more complete cascade/model would include both the oceanic and the terrestrial part and their potential interaction in the final determination of the size of the salmon populations. An even more complete consideration would also include the deep, as well as the coastal, ocean. The subarctic current itself is relatively nutrient rich, so another research topic is to determine the relative size of the nutrient pool provided by up-welling in the coastal ocean compared to that provided by the import at the surface from the subarctic current. In the example of this chapter, a more profound consideration of the principal cascade leads to the recognition of the importance of its consideration as an open system.

The issue of preexisting conditions has at least two major implications for salmon survival in the PNW. These relate to cyclic changes superimposed on longer term trends. First, if the human-related insults to salmon populations such as water pollution and habitat destruction are not reversed or removed, the populations will disappear regardless of any natural decadal-scale cycles that may exist. Long-term trends provide a new set of preexisting conditions for the start of each decadal-scale "cycle." Second, for the future, climate models suggest increased winter flooding and decreased spring and summer streamflows along with elevated stream and estuary temperatures that, singly and together, will be detrimental to salmon stocks even before they enter the ocean. If such climate changes are indeed realized, the prospect for PNW salmon stock has been described as "bleak" (JISAO CIG 1999).

As far as the decadal-scale "cycle" goes, and ignoring longer term trends, the ecological effects, operating through the cascade, are mostly complete by the time the next climate cycle begins. This is because the timescale of an Oregon Coho salmon cohort, of about 3 years, is much shorter than the approximate 15- to 20-year decadal climate regime shift.

The example of decadal-scale change in PNW salmon abundance does have at least an implied upper and lower limit. These limits are identified by the two major modes of climate variability (figure 13.2) and the values of climatic and other variables that result from these modes. As a working hypothesis we may suggest that, in the absence of other factors operating at other timescales, the cascading effects of the climatic variation will stop at these extreme points and not proceed in the sense of a continuing positive feedback. In addition, in this example the climatic event or episode does appear to reverse to some "original" or at least repeated state because of its cyclic nature. The observed historic record does not show evidence of the climate state going beyond an original position, but tree-ring records have been useful in better defining the size and intensity of the climate regime shifts during a 400-year period (Biondi et al. 2001; Gedalof and Smith 2001). We may also hypothesize that the cascade in this example does reverse in a more or less direct manner that does not display some kind of hysteresis. This reversal occurs in the sense that once the physical climatic conditions that give rise to a lack of oceanic water nutrients and to low fish abundance have changed to the other extreme, if all other factors remain the same, nutrient availability and fish abundance returns. The phytoplankton at the lower end of the food chain have not been irreversibly destroyed by adverse climatic conditions in the warm phase of the PDO—they have simply been made unavailable to the coastal oceanic salmon. The broad timing of these decadal-scale events is fairly well known, so as far as this is concerned, a simulation model of the system could be initiated. However, there is a lack of information on the smaller timescale events such as the operation of the food chain within any given spring or summer.

The "if all other factors remain the same" in the previous discussion is very important and has not been realized since human activities have been carried on in-tensively—at least during the twentieth century. A reviewer of this chapter points out that "while it seems likely that the impacts of decadal climate cycles are "reversible" in many ways, the potential for extinction of wild Coho is very real. Losing locally adapted breeding populations may not be reversible, and this is a critical issue that has caused a great deal of concern in restoration circles. Clearly, PDO climate cycles haven't been the sole cause of Coho extinctions or threats to extinction, but in cases where populations are severely reduced in number by habitat loss and degradation, overfishing, and/or poor management practices, a string of years with unfavorable climate conditions may be the last straw."

We might speculate that further insights into the decadal-scale changes of climate and salmon abundance might be obtained by examining the system as if it were a chaotic one. Certainly there are many nonlinear changes in the various components of the system. Even more intriguingly, the two modal positions of the climate regimes (figure 13.2) give the appearance of being attractors for the oceanic/climatic part of the system. Further specification of the nature of the chaos, if it exists, will await the quantification of the system.

Acknowledgment This chapter is based on work performed in association with the H. J. Andrews LTER group that is funded in part by grants from the National Science Foundation Division of Environmental Biology, Long-Term Studies. I am very grateful for extremely helpful comments and contributions provided by Dr. Nathan Mantua of the University of Washington.


Biondi, F., A. Gershunov, and D.R. Cayan. 2001. North Pacific decadal climate variability since 1661. Journal of Climate. 14:5-10.

Fisher, J., and W. Pearcy. 1988. Growth of juvenile Coho salmon off Oregon and Washington, USA, in years of differing coastal upwelling. Canadian Journal of Fisheries and Aquatic Science. 45:1036-1044.

Francis, R.C. 1993. Climate change and salmonid production in the North Pacific Ocean. Pages 33-42 in Proceedings of the Ninth Annual Pacific Climate (PACLIM) Workshop. April 21-24, 1992. Asilomar, California. California Department of Water Resources. Sacramento.

Francis, R.C., and T.H. Sibley. 1991. Climate change and fisheries: What are the real issues? The Northwest Environmental Journal. 7:295-307.

Gedalof, Z., and D. Smith, 2001. Interdecadal climate variability and regime-scale shifts in Pacific North America. Geophysical Research Letters. Vol. 28:1515-1518.

Greenland, D. 1994. The Pacific Northwest regional context of the climate of the H. J. Andrews Experimental Forest Long-Term Ecological Research Site. Northwest Science 69:81-96.

Greenland, D. 1995. Salmon Populations and Large Scale Atmospheric Events. Proceedings of American Fisheries Society 1994 Northeast Pacific Chinook and Coho Salmon Workshop. Salmon Ecosystem Restoration: Myth and Reality. Nov. 7-10. Pages 103-114. Republished by request under the title "Offshore Coho Salmon Populations near the Pacific Northwest and Large Scale Atmospheric Events" in Proceedings of Twelfth Pacific Climate (PACLIM) Workshop, May 2-5, 1995. C.M. Isaacs and V.L. Tharp (eds). 1996. California Department of Water Resources. Interagency Ecological Studies Program, Technical Report 46, pp. 109-120.

Hare, S.R., N.J. Mantua, and R.C. Francis. 1999. Inverse production regimes: Alaskan and West Coast Pacific Salmon. Fisheries. Vol. 21, No. 1:6-14.

Hollowed, A.B., and W.S. Wooster. 1991. Year class variability of marine fishes and winter ocean conditions in the Northeast Pacific Ocean. ICES Variability Symposium. June 5-7, 1991. Mariehamn, Finland. Paper No. 33. Session 3.

JISAO CIG. 1999. Impacts of climate variability and change in the Pacific Northwest. Pacific Northwest Contribution the National Assessment in the Potential Consequences of Climate Change for the United States. The Joint Institute for the Study of Atmosphere and Ocean Climate Impacts Group, Box 354235. University of Washington, Seattle, Washington. JISAO contribution #715. 109 pp.

Mantua, N.J., S.R. Hare, Y. Zhang, J.M. Wallace, and R.C. Francis. 1997. A Pacific interdecadal climate oscillation with impacts on salmon production. Bulletin of the American Meteorological Society. 78:1069-1079.

Miller, A.J., D.R. Cayan, T.P. Barnett, N.E. Graham, and J.M. Oberhuber. 1994. The 1976-7 climate shift of the Pacific Ocean. Oceanography. 7:21-26.

Overland, J.E., J. Miletta Adams, and N.A. Bond. 1999. Decadal Variability of the Aleutian Low and Its Relation to High-Latitude Circulation. Journal of Climate. 12:1542-1548.

Pearcy, W.G. 1992. Ocean Ecology of North Pacific Salmonids. University of Washinsgton Press, Seattle.

Peterson, W., and D. Mackas, 2001: Shifts in zooplankton abundance and species composition off central Oregon and southwestern British Columbia. PICES Press, North Pacific Marine Science Organization. Vol. 9:28-31.

Ware, D.M., and R.E. Thomson. 2000. Interannual to Multidecadal Timescale Climate Variations in the Northeast Pacific. Journal of Climatology. 13: 3209-3220.

0 0

Post a comment