Pacific salmon (Oncorhynchus spp.) utilize three or four different habitats during their life: the subsurface of streams and lakes as embryos and larvae, streams and lakes as young juveniles, and estuaries and the ocean as older juveniles. Variability in survival in each of these habitats will contribute to recruitment variation, as will spawner abundance. Here I briefly review the relative roles of each factor in determining adult abundance.

Pacific salmon are relatively unique among nontropical commercial species in that their semelparous life history and often simple age structure results in many spawning populations being dominated by a single cohort of recruits. Thus there is often great variation in spawner abundance from year to year. Further, the low fecundity of salmon compared to many marine fish implies that egg-recruit mortality for salmon is much lower (M = 6 to 7; Bradford 1995) than for many commercial marine species (M = 11 to 14; Koslow 1992). Interannual variability in mortality is correlated with the mean (Bradford 1992, Bradford and Cabana 1996), implying that variation in egg-recruit survival in salmon is probably lower than that for marine fish, depending on the strength and timing of density-dependent mortality.

The result of these observations is that there is often a relatively strong relationship between stock and recruitment for Pacific salmon compared to similar data for other species (Fig. 1), and in many cases recruitment can be forecasted from stock sizes with precision not available for marine fish. However, because the harvested stock consists only of recruits, and harvest rates in excess of 70% are common, forecasts of recruitment from stock size are usually not precise enough for modern-day management. Nonetheless, the point I wish to highlight is that in many cases a significant portion of variability in adult returns can be related to parent stock size, irrespective of survival conditions in the various habitats.

Salmon use the substrates of rivers and lakes for incubation, and some species use freshwater habitats for rearing. Survival from egg deposition to fry emergence the following spring averages about 10%, although there are some data that suggest survival might be higher for coho (O. kisutch) and chinook (O. tshawytscha) salmon eggs and alevins (Bradford 1995). The coefficient of variation (CV) for interannual survival is about 30%, averaged over all species. Density effects due to crowding on spawning grounds seem common for species that spawn in dense aggregations (i.e., pink (O. gorbuscha), sockeye (O. nerka), and chum (O. keta) salmon; Foerster 1968).

Coho, sockeye, and some chinook salmon populations spend a year or more in fresh water, and survival rates for this year range from 5% to 25%. In virtually all cases examined, there is strong evidence for density-dependent population regulation in coho salmon (Table 1), probably due to limited amounts of suitable rearing habitat (Bradford et al. in press). Density-dependent survival has also been observed in lake-rearing sockeye salmon juveniles (reviewed by Hume et al. in press). In some cases, the density effects are strong enough to virtually eliminate the effects of parent stock size on smolt abundance, except at very low stock size.

While the roles of interannual variation in the physical environment (i.e., weather, hydrology) and density on survival have been fairly well documented, there are few data sets that allow analysis of decadal changes in freshwater survival due to low-frequency climate change or habitat degradation. One exception is the long series of egg-smolt survival rates for Chilko Lake, B.C. sockeye salmon (Fig. 2). Visual examination of these data suggests an approximately 15-year cycle in survival, although these data have not been quantitatively examined. It is not clear whether the overall downward trend in freshwater survival is real or has resulted from changes in the methods for estimating spawner abundance. Lower survival rates in the most recent years may also be related to record high spawner abundances that occurred in the 1980s (Hume et al. in press).

Most estimates of marine survival are based on smolt abundances made some distance inland from the sea and, therefore, include mortality during downstream migration and residence in the estuary. The survival of salmon in the ocean ranges from less than 1% to more than 10%, depending on the species, the size at ocean entry, and the length of time spent in the ocean (Bradford 1995). Interannual variability in survival can be quite high (Fig. 2). Although it might be expected that species that enter the ocean at small size are more susceptible to ocean conditions, and have more variable survival rates, the evidence for this is equivocal (Bradford 1995).

In contrast to the freshwater environment, the evidence for density-dependent survival in the ocean is not as clear (Pearcy 1992). The analysis of such data is often difficult because of measurement error (Peterman 1982) and time trends in abundance data, hatchery releases, and oceanic conditions. In most analyses, density-dependent effects contribute only a small fraction of the total variability in marine survival rates.

A question that is sometimes posed is what is the relative importance of freshwater, estuarine, or marine habitats for recruitment variation? It is difficult to comment on the role of the estuary on variation in adult abundance due to the technical problems in accurately estimating survival in this habitat, so it is necessary to consider the estuary with the marine environment.

For the other two major habitats, at the species level, empirical analysis indicates mortality is roughly equally divided between the two habitats (Fig. 3), and we expect that the variability in mortality would also be divided similarly (Bradford and Cabana 1996). Available data suggest that this is true (Bradford 1995), and each habitat contributes significantly to recruitment variation, although the exact distribution of mortality among habitats varies by species.

The preceding is based on the naive assumption that mortality in each habitat varies randomly and is independent of events in the other habitat. The sometimes strong density-dependent mortality that occurs in fresh water means that variation in the number of smolts produced by a stream or lake is lower than might be expected based on estimates of the variability in survival or initial egg abundance. In such cases the role of the marine environment in determining recruitment strength will be greater.

Perhaps of greater interest for conservation purposes is determining the relative roles of long-term trends in survival in marine and freshwater habitats. A number of recent analyses have shown decadal trends in adult salmon abundance, which appear to be related to changes in oceanographic conditions that occur at similar scales (e.g., Hare and Francis 1995). Nonetheless, at the moment it is difficult to apportion these trends in abundance to changes in survival in the marine or freshwater environment, or to changes in the number of parent spawners caused by harvest management. The only way to directly address this issue is with long-term data sets where all major stages of the life cycles are accurately enumerated. With these data we may be able to determine the effect that long-term changes in climate have on productivity of both marine and freshwater environments.

Chilko Lake sockeye salmon time series is an example of such data (Fig. 2). From the smoothed trends in freshwater and marine survival rates there might be some evidence for interdecadal cycles in survival. However, the cycles (if real) appear to be out of phase with each other, and there is little correlation in survival between environments. Coupling of freshwater and marine survival rates may be more likely for coastal stocks (e.g., Cooney et al. 1995).

In summary, partitioning survival to different parts of the salmon's life cycle is somewhat scale dependent. At the annual scale, we can generalize that survival and its variability are roughly equally divided between freshwater and marine habitats. However, there are only a few instances where we can examine the effects of interdecadal-scale climate forcing on freshwater and marine survival simultaneously. Such analyses will be difficult because density-dependent effects and anthropogenic habitat changes in fresh water will tend to obscure climate effects. Nonetheless, these analyses are essential to correctly interpret changes in salmon production at decadal or longer scales.

Bradford, M. J. 1992. Strength and precision of recruitment predictions from the early life stages of marine fishes. Fish. Bull., U.S. 90:439-453.

Bradford, M. J. 1995. Comparative analysis of Pacific salmon survival rates. Can. J. Fish. Aquat. Sci. 52:1327-1338.

Bradford, M. J., G. C. Taylor, and J. A. Allan. In press. Empirical review of coho salmon smolt abundance and the prediction of smolt production at the regional level. Trans. Am. Fish. Soc.

Bradford, M. J., and G. Cabana. 1996. Interannual variation in stage-specific survival rates and the causes of recruitment variation. In R. C. Chambers and E. A. Trippel (editors), Early life history and recruitment in fish populations, p. 597-630. Chapman and Hall, New York.

Cooney, R. T., T. M. Willette, S. Sharr, D. Sharp, and J. Olson. 1995. The effect of climate on North Pacific pink salmon (Oncorhynchus gorbuscha) production: Examining some details of a natural experiment. In R. J. Beamish (editor), Climate change and northern fish populations, p. 475-482. Can. Spec. Publ. Fish. Aquat. Sci. 121.

Foerster, R. E. 1968. The sockeye salmon. Bull. Fish. Res. Board Can. 162.

Hare, S. R., and R. C. Francis. 1995. Climate change and salmon production in the Northeast Pacific Ocean. In R. J. Beamish (editor), Climate change and northern fish populations, p. 357-372. Can. Spec. Publ. Fish. Aquat. Sci. 121.

Hume, J. M. B., K. S. Shortreed, and K. F. Morton. In press. Juvenile sockeye rearing capacity of three lakes in the Fraser River system. Can. J. Fish. Aquat. Sci.

Koslow, J. A. 1992. Fecundity and the stock-recruitment relationship. Can. J. Fish. Aquat. Sci. 49:210-217.

Pearcy, W. G. 1992. Ocean ecology of North Pacific salmonids. Univ. Washington Press, Seattle, 179 p.

Peterman, R. M. 1982. Non-linear relation between smolts and adults in Babine Lake sockeye populations and implications for other salmon populations. Can. J. Fish. Aquat. Sci. 39:904-913.

Table 1. Density-dependent mortality of juvenile coho salmon. Shown are regression slopes (b) and standard error (SE) for regression of log_e (smolts) on log_e (adults) for streams with more than 10 years of data. Slopes significantly less than 1 indicate survival decreases with increasing abundance.

Table of Contents