Interest is growing in the question of whether the ocean as well as fresh water can limit Pacific salmon (Oncorhynchus spp.) production. Both the North Pacific marine science organization, PICES, and the North Pacific Anadromous Fish Commission (NPAFC) have on their research agendas proposals to develop joint international efforts to try to answer this question. It is likely that there are few more technically difficult questions to answer in fisheries oceanography and few that are of potentially greater importance to salmon-producing countries.

The primary effects of the ocean on salmon productivity involve both growth and survival of salmon. The decline in salmon survival in Washington and Oregon since 1977 is probably caused by as yet poorly understood processes in the marine (as opposed to freshwater) environment. The need to understand the causes of survival variations in fish are apparent; however, comparatively little effort has been focused on understanding the changes in growth that have occurred. Variation in growth has been assumed to be of relatively little importance to the regulation of productivity in fish populations in general. Yet, body size at maturity directly affects the meat yield of salmon and indirectly, and probably even more important, the biological productivity: changes in fecundity, average age at maturity, and ability to swim upriver and prepare and defend nest sites (Forbes and Peterman 1994).

Salmon are fundamentally marine animals whose reproduction is tied to fresh water. Viewed as a whole, salmon spend most of their lives getting out of fresh water. Most species probably do not even remain in the coastal environment for any significant time, instead using the coastal environment as a transit zone during their passage offshore and again on the return trip inshore to spawn.

From a Darwinian viewpoint, this behavior is interesting. It suggests that there are great benefits to getting to the offshore pelagic zone as quickly as possible. Because our most productive species are pink (O. gorbuscha), chum (O. keta), and sockeye (O. nerka) salmon (which spend the smallest amount of time in coastal waters), this suggests that the offshore open ocean is the most productive environment. However, since the productivity of the offshore is much lower than the coastal region per unit area, salmon must move offshore to either reduce rates of predation or increase growth rates because of density-dependent interactions in the coastal zone, or both.

In the following sections, I review some of the evidence that a) salmon show sharp thermal limits which appear to be the result of reducing their basal metabolic rates in times of low food abundance, b) significant changes in rates of annual growth occur in the ocean, c) density-dependent changes in growth are limited to the first and last years of life in the ocean when salmon must pass through the coastal zone to the offshore, and d) evolutionary selection has resulted in extremely sensitive responses by salmon to factors affecting their growth rates in the offshore areas.

Between 1955 and the late 1960s, Canada, Japan, and the United States mounted an intensive ocean research program on the distribution of Pacific salmon. It was during this time that North American scientists first discovered that salmon undertook extensive ocean migrations and were not simply residents of the nearshore coastal zone during the marine phase of their life history.

Although substantial effort was put into ocean surveys of salmon, there was little evidence of strong limits to the distribution of salmon until recently. However, evidence for quite strong thermal limits on their ocean distributions was found during a survey of salmon distributions in the spring of 1990 (Welch et al. 1995). Based on these preliminary results, Canada and Japan mounted a series of 11 one-month surveys in 1992 to further examine the limits on the salmon distribution in different seasons and areas of the North Pacific. In addition, we supplemented these surveys by combining all of the available Canadian and U.S. salmon surveys for the 1950s and 1960s, and the more limited data collected by Canada in the 1980s and 1990s. An extensive data set was also available from Japan, which has been conducting substantial salmon surveys since 1972. (Surveys for prior years are available, but they are not amenable to computer analysis.)

The geographic distribution of this data set, which comprises some 20,636 observations on ocean salmon abundance, is shown in Figure 1. At each location, the catch of each species of salmon is recorded, along with the amount of fishing effort and a number of oceanographic variables, chiefly sea surface temperature (SST).

Our current findings indicate that the primary control on salmon distribution is temperature, but that the upper thermal limit varies throughout the year. To illustrate this point, Figures 2a, 2b, and and 2c shows the distribution of log(salmon catch) against SST, but with the catches split out in a trellis plot by five decades (rows) and four geographic regions (columns) for three seasons of the year: winter, spring, and summer. (We actually plot 1n(catch+1) to preserve the zeros in the log transformation.) Figures 2a, 2b, and and 2c refers specifically to sockeye salmon, but our general findings also apply to all the other species of Pacific salmon. Although the temperature defining the thermal limit in any given month differs significantly between species in the spring and summer, and shows some smaller variation between decades and areas of the ocean, the existence and remarkable sharpness of this limit is the same in all species.

With the exception of October, when we have no data, a thermal limit is evident in all months. For the "winter" months of November through March, sockeye are confined to ocean regions of <7°C. These thermal limits then begin to change, with the thermal limits increasing from April through September, when they reach a maximum of about 15°C. Thermal limits then begin to decline again until they finally reach the 7°C level by November. Unfortunately, the paucity of data in the autumn prevents us from fully defining how salmon respond to temperature during the fall, as their thermal limits must cycle back to the winter temperature limits.

What is most remarkable about this extensive data set is just how sharp these thermal limits are, particularly when we consider that the data extend over five decades and include data from all parts of the North Pacific Ocean and the Okhotsk Sea. (Temperatures in the Bering Sea never increased to levels of thermal limits recorded in the other regions of the North Pacific.)

In British Columbia (BC) alone, there are approximately 3,150 genetically separate stocks of the five most abundant species of salmon (Healey 1982). British Columbia formerly produced about 15% of the total salmon from the North Pacific Ocean, which makes for a rough estimate of 21,000 genetically separate salmon stocks occupying the North Pacific. As the thermal limits are extremely sharp for all species, with abundances dropping by one to two orders of magnitude in approximately 1°C, essentially all genetically separate stocks must be responding to the same thermal limit—otherwise, such a sharp drop in abundance would not be observed across so many populations. Whatever the specific causes of the thermal limits are, their effects on productivity and survival for each species must be very large, or such uniformity in the response to temperature would not be observed.

What is not clear is how such a sharp response can also exist if the different stocks have different patterns of spatial distribution, since this would presumably make their thermal limits different. Perhaps the stocks differentially distribute themselves within the salmon distribution not only relative to temperature, but also in such a way as to minimize their competition for food and thereby maximize their growth rates (see below), thereby smoothing out local variation in the relative abundance of different stocks.

The sharpness of the roll-off in salmon abundance with temperature is best shown by example (Fig. 3, from Welch et al. 1995). Applying the observed roll-off in abundance with temperature to the observed temperature field present in the spring of 1990, all species of salmon show extremely sharp declines in abundance with temperature. (The survey area shown covered most of the central and southern Gulf of Alaska, out to about 300 nautical miles west of Station Papa, or 2,000 km from east to west, and 1,300 km from north to south.)

General circulation models of the world's atmosphere predict that the oceans will warm by 2°-3°C as the carbon dioxide content of the atmosphere doubles. What will happen to the available thermal habitat (ATH) for salmon under a doubling of atmospheric CO₂? The predicted effect of this level of warming is quite surprising. The ATH is predicted to shrink dramatically in both summer and winter (Fig. 4), raising the prospect that salmon will be pushed out of the subarctic Pacific, increasing trophic competition and potentially reducing growth rates. Thus, although much attention has been paid to the possibility that some stocks of salmon near the southern end of their freshwater distribution may be disrupted (Fraser River sockeye being a prime candidate (Levy 1992)), events happening in the marine phase may be even more disruptive. At present, we have no evidence for BC salmon entering the Bering Sea, apart from some stocks of chinook.

If the oceans warm past these sharply maintained thermal limits, BC, Washington, and Oregon salmon would appear to have three choices: 1) develop (or express) the ability to migrate into the Bering Sea (and return), 2) begin to vertically migrate to stay below the permanent thermocline at about 100 m, or 3) simply incur the energetic or other penalties that they now strongly avoid by remaining at higher temperatures.

We have no way of establishing at present which of these three mechanisms may occur, but it seems safe to assume that since such behavior does not seem to occur now, expressing this behavior in the future will have negative effects on salmon production. And as BC salmon are distributed in the Gulf of Alaska, where there is no route into the Bering Sea except at the farthest western extent of their distribution, it would seem reasonable that further study of these possibilities is called for. It is also of some interest that marine survival of Atlantic salmon (Salmo salar) has been correlated with interannual variation in the size and position of 4°-8°C water in the North Atlantic (Friedland et al. 1993).

The observed thermal limits are, however, qualitatively consistent with that predicted by Welch et al. (1995), who suggested that salmon could maximize their growth rates by lowering their body temperature and reducing basal metabolic rates in those seasons when food levels are lower. By doing so, salmon increase the amount of food energy left for growth, and thereby increase their somantic growth rates to a greater extent than would otherwise be possible if they remained at constant temperature (Brett 1956, 1971). There is also substantial evidence that fish placed in thermal gradients in the laboratory move to that temperature yielding maximum growth rates (reviews by Beitinger and Fitzpatrick 1979, Brett et al. 1969).

Unfortunately, this neat qualitative picture tends to fall apart in the details—the field data are in some ways too good. Graphs of growth rate vs. temperature show a roughly parabolic and symmetric response of growth to environmental temperature (Brett et al. 1969), so a 1°C increase or decrease in temperature from that giving the maximum growth rate should cause a similar reduction in growth rate. The response of each species of salmon to temperature is clearly a step-function—temperature has no measurable effect on abundance in the interior of the distribution (away from the edge), yet all individuals are stopping in a fraction of a degree Celsius at the edge. The sharpness of the edge is inconsistent with growth maximization, although the qualitative prediction is correct. It may be that salmon simply distribute themselves in areas where they do not incur a growth deficit; that is, as long as growth rate is positive, salmon forage independently of temperature. However, I think that the key point for this workshop is that if all species and genetic populations of salmon are responding with such exquisite sensitivity to factors affecting growth, then such factors may play a very large role in the evolutionary biology, and therefore productivity.

Almost all salmon growth is completed in the ocean, and the final size of salmon at return can be measured relatively accurately. In addition, the growth of scales leaves a permanent record of the amount of growth achieved in different years at sea and, therefore, allows us to partition the integrated growth of the body between years at sea. In essence, although we do not know what the migration pathways of salmon are at sea in any detail, the scale growth records allow us to treat salmon as biological conductivity, temperature, and depth (CTD) sensors which go out to sea and record growth conditions in their scales. If we cannot identify the influential oceanographic factors for growth, it seems unlikely that we will ever be able to identify important factors for survival, given the difficulty with accurately measuring the latter.

Figure 5 shows an idealized input-output sketch of the relationship between salmon growth and a variety of underlying factors in the ocean, viewing the ocean as a chemostat. Fish growth will not reach its maximum potential if food density (food available divided by ocean volume) is insufficient to provide the maximum daily ration. If this critical level of food is not exceeded, then the potential for the oceans to limit salmon growth exists. As the examples shown later demonstrate, food availability in the ocean appears to be limiting growth because there are large year-to-year variations in salmon size at maturity.

At food levels below critical, a general framework can be erected to describe how salmon growth can be impacted. First, physical factors may change the productivity of the oceans from year to year, changing the amount of food available. Second, if the geographic distribution of salmon is limited, physical forcing of the southern (or northern) boundaries can change the area available for grazing. Variation in physical variables can then impact salmon growth by changing the amount of available habitat from year to year. Third, ocean temperature may affect growth by increasing basal metabolic rates at higher temperatures, resulting in less energy being available for growth. Finally, the number of salmon entering the ocean may impact growth through competition for the available food.

All of these factors may impact growth. However, only the number of salmon entering the ocean is amenable to human intervention (through manipulation of adult escapement and ocean ranching), although it is an important variable. Increases in salmon abundance by one country may reduce the growth rates of both their own and other countries' salmon. The scientific issue is to disentangle and evaluate the relative contributions of these variables on ocean salmon growth.

Accurate predictions of the future effect of climate warming on salmon are difficult. One possible approach is to examine the response of salmon to changes in ocean climate through retrospective analyses. The size of salmon at return to the coast varies substantially between years (Ricker 1995). However, the effects of ocean climate on growth are confounded with time because salmon spend 2 to 4 years at sea before their return, depending upon species and life history type (Healey 1986). When is this variation in size expressed?

Measurements of annual growth rings on sockeye salmon scales archived at the Department of Fisheries and Oceans and the Pacific Salmon Commission suggest that substantial growth variation occurs at all ages (Welch 1994). However, in those stocks we have examined so far, variations in the amount of salmon scale growth achieved in any 1 year by a given fish is uncorrelated with the amount of growth observed at other ages for the same fish (e.g., Figs. 6 and 7). Thus, there does not appear to be any evidence that consistently faster- or slower-growing fish of a given stock occur, nor that the factors affecting scale growth in different years are correlated—the spectrum for the ocean factors influencing scale growth appears to be white rather than red on short-time scales.

Several general effects on scale growth do appear to be noteworthy, however. First, long-term declining trends in scale growth appear to be largely confined to those ages when a given life history type transits the coastal zone on its way offshore or on the return trip to spawn. This would be consistent if it is the coastal zone that is primarily food limited, and the recent increases in salmon abundance have resulted in increased trophic competition while in coastal waters. Supporting this conjecture, long-term trends in scale growth appear to be largely absent for those ages when all growth occurs in offshore regions (M2 (second marine year) growth for age 4.2 and 5.3 sockeye, and M2 and M3 growth for age 5.2 sockeye).

As the previous section shows, there are large inter-annual changes in final body size, suggesting that food is limiting at some point in the life history (otherwise salmon should always grow at their maximum growth rates). These variations in growth also raise the possibility of conducting retrospective analyses to statistically identify those oceanographic variables that influence growth.

One such factor is the area of the North Pacific Ocean thermally available to salmon in different years. The ATH for the Gulf of Alaska can be defined as that region of the offshore lying north of some specified isotherm, and east of 160°W. My colleagues at the University of British Columbia (Keith Thomson, Paul LeBlond, and Ian Jardine) and I have begun calculating these areas, using the Comprehensive Ocean-Atmosphere Data Set (COADS) and, in more recent years, the Advanced Very High Resolution Radiometry (AVHRR) data base.

The early results are encouraging as they suggest that since 1948 the winter ATH has varied by a factor of 1.5 or so, while spring and summer ATHs have varied by a maximum of slightly more than 2 (Fig. 8). Although we have not yet begun the analysis of whether fluctuations in ocean area in conjunction with variations in salmon abundance influence salmon growth, this avenue seems to offer a promising approach to assessing the question of whether or not the carrying capacity of the oceans is sufficiently limited to impact the productivity of salmon populations.

The question of whether the North Pacific Ocean has a carrying capacity sufficiently limited that this fact should be taken into account in our salmon management plans is still a number of years away from being adequately answered. However, the exquisitely sharp response of the ocean distribution of salmon to temperature, presumably related to growth, strongly suggests that natural or artificially induced changes in the growth rates of salmon will have important effects on their productivity and, thus, on the success of our salmon fisheries. The changes in ocean climate that are predicted to accompany climate warming may have massive impacts on the biology of salmon. It is important that we develop a deeper and more quantitative understanding of how growth and survival of salmon in the ocean will change, and to clearly define the link between these biological changes and the sustainable productivity of the fisheries that depend on them.

Beitinger, T. L., and L. C. Fitzpatrick. 1979. Physiological and ecological correlates of preferred temperature in fish. Am. Zool. 19:319-329.

Boer, G. J. , N. A. McFarlane, and M. Lazare. 1992. Greenhouse gas-induced climate change simulated with the Canadian Climate Centre's second-generation general circulation model. J. Climate 5:1045-1077.

Brett, J. R. 1956. Some principles in the thermal requirements of fishes. Quart. Rev. Biol. 32(2):75-87.

Brett, J. R. 1971. Energetic responses of salmon to temperature: A study of some thermal relations in the physiology and freshwater ecology of sockeye salmon (Oncorhynchus nerka). Am. Zool. 11:99-113.

Brett, J. R., J. E. Shelbourn, and C. T. Shoop. 1969. Growth rate and body composition of fingerling sockeye salmon, Oncorhynchus nerka, in relation to temperature and ration size. J. Fish. Res. Board Can. 26:2363-2394.

Forbes, L. S., and R. M. Peterman. 1994. Simple size-structured models of recruitment and harvest in Pacific salmon (Oncorhynchus spp.). Can. J. Fish. Aquat. Sci. 51:603-616.

Friedland, K. D., D. G. Reddin, and J. F. Kocik. 1993. Marine survival of North American and European Atlantic salmon: Effects of growth and environment. ICES J. Mar. Sci. 50:481-492.

Healey, M. C. 1982. Multispecies, multistock aspects of Pacific salmon management. In M. C. Mercer (editor), Multispecies approaches to fisheries management advice. Can. Spec. Publ. Fish. Aquat. Sci. 59:119-126.

Healey, M. C. 1986. Optimum size and age at maturity in Pacific salmon and effects of size-selective fisheries. In D. J. Meerburg (editor), Salmonid age at maturity. Can. Spec. Publ. Fish. Aquat. Sci. 89:39-52.

Levy, D. A. 1992. Potential impacts of global warming on salmon production in the Fraser River watershed. Can. Tech. Rep. Fish. Aquat. Sci. 1889:1-108.

McFarlane, N. A., G. J. Boer, J. P. Blanchet, and M. Lazare. 1992. The Canadian Climate Centre second-generation general circulation model and its equilibrium climate. J. Climate 5:1013-1044.

Ricker, W. E. 1995. Trends in the average size of Pacific salmon in Canadian catches. Can. Spec. Publ. Fish. Aquat. Sci. 121:593-602.

Welch, D. W. 1994. Variation in marine growth rates of British Columbia pink and sockeye salmon stocks. Proceedings of the Annual Meeting of the North Pacific Anadromous Fish Commission, Vladivostok, Russia, October 1994. NPAFC Document 94:1-30.

Welch, D. W., A. I. Chigirinsky, and Y. Ishida. 1995. Upper thermal limits on the oceanic distribution of Pacific salmon (Oncorhynchus spp.) in the spring. Can. J. Fish. Aquat. Sci. 52:489-503.

Welch, D. W., Y. Ishida, and K. Nagasawa. In prep. Thermal limits and ocean migrations of Pacific salmon: Long-term consequences of global warming. (Available from Ocean Sciences and Productivity Division, Department of Fisheries and Oceans, Pacific Biological Station, Nanaimo, BC, Canada V9R 5K6.)

Table of Contents