VETTER: NATURAL MORTALITY IN FISH STOCKS 



As discussed previously, these different esti- 

 mates can lead to at least as great a difference in 

 results produced by fishery analyses such as yield 

 models or stock reconstruction analyses (Sec- 

 tion II). 



Reported differences in estimates of M for 

 whitefish stocks living in Shakespeare Lake com- 

 pared with other relatively small lakes (e.g., Lake 

 McDonald, Table 3) are particularly significant. 

 Because both stocks are of the same species and 

 living in more or less similar environments 

 (small lakes), one might easily (and incorrectly) 

 assume that both have the same rate of natural 

 mortality; but they did not. The lower rates oc- 

 curred in the stock occupying a small lake with no 

 predators. This is a clear example of the effect 

 that environment, particularly the predator envi- 

 ronment, can have on the realized rate of natural 

 mortality in a fish stock. Obviously, choosing a 

 single appropriate constant for this species would 

 be difficult. Choosing an appropriate species- 

 specific constant for some of the other species 

 with multiple estimates might be difficult as well 

 (e.g., rock bass, lake trout, perch, roach, tarakihi, 

 or menhaden. Table 3). 



None of these studies from either unexploited 

 or exploited stocks support the assumption that M 

 is constant for any given stock or species, nor are 

 these within-stock ranges particularly narrow. In 

 addition, treatment of the original catch data 

 may have in some cases obscured the "true" vari- 

 ability. Ricker (1947) and Kennedy (1953, 1954, 

 1963), for example, use a 3-yr smoothing tech- 

 nique to reduce the effects of unequal recruit- 

 ment; this also serves to reduce variability that 

 may actually be due to differences in natural mor- 

 tality. Also, single estimates from data collected 

 during only one or two years of sampling (e.g., 

 Wohlschlag 1954; Williams 1967; Mann 1973; 

 Vooren 1977) can be seriously biased by annual 

 changes in either recruitment or mortality rates. 

 If the estimates reported above are even approxi- 

 mately accurate, it is apparent that the range of 

 possible values for M is wide, and that variability 

 can be considerable even within single stocks. 



A solution to this problem of choosing a reason- 

 able value for M, at least for long-lived fish, is 

 suggested by the possibility that variation in M 

 (not just the mean value) may be related to max- 

 imum lifespan. Fish that live for many years 

 must naturally have lower mortality rates than 

 more short-lived fish. These lower rates may also 

 be less variable in the longer lived stocks, if as in 

 many other biological processes, variability is 



proportional to the mean. This could account for 

 the ubiquity and apparent effectiveness of the 

 constant 0.2 year"\ used almost universally for 

 the long-lived (20 to 30 years) and well-studied 

 fish stocks from northern European seas (e.g., 

 Beverton 1964). If so, assuming a constant M 

 might be valid for these longer lived stocks. 



Unfortunately, the few studies cited above do 

 not support this attractive idea. Although in gen- 

 eral, mortality rates decease as lifespan in- 

 creases, the variability in estimates does not ap- 

 pear to follow the same trend. This may be due 

 partially to the relatively similar lifespans (10 to 

 20 years) for most of the species for which esti- 

 mates exist. But the apparent range in rates for 

 the shortest lived species cited above (Ahrenholz 

 1981, Brevoortia patronus, ages 1 to 3 years, M 

 range 0.7 to 1.6 year"^) is certainly not greater 

 than ranges reported for the longer lived white- 

 fish (Henderson et al. 1983, Coregonus clu- 

 peaformis, ages 10+ years, M range 0.34 to 1.67 

 year"^). 



VI. SUMMARY AND 

 RECOMMENDATIONS 



Thus it appears that rates of M, or at least rates 

 of M derived by existing estimation methods, do 

 in fact vary widely within many fish stocks. Be- 

 cause the variations appear to be considerable 

 and because the results from fishery models can 

 be sensitive to large variations in M , one must 

 conclude that assuming constancy without proof 

 can have serious consequences for fishery man- 

 agement. 



A better approach may be to discard the notion 

 that a single "best" estimate of M can be found, 

 and instead try to tailor estimates of M to local 

 groups, based on some combinations of the meth- 

 ods discussed in Section III. Obviously, practical 

 considerations of time and resources will limit the 

 accuracy and precision with which M can be esti- 

 mated. Also, the estimates in the studies re- 

 viewed here are prone to all the artifacts men- 

 tioned in the previous sections. True rates of 

 natural mortality, and their variability, are still 

 very poorly known for even the great stocks of 

 commercial fish in temperature regions that have 

 been subject to continuous exploitation for 

 decades. Careful, repeated tagging experiments 

 probably hold the most promise for determining 

 with any reasonable degree of accuracy, rates of 

 natural mortality in fish stocks. But even these 

 have inherent problems that are not easily 



39 



