Unnatural selection

Fishing and hunting by humans are the main causes of mortality in many populations of wild animals. The consequence is that large and rapid changes occur in certain characteristics that far exceed changes due to other agents.

Online collection.

Harvesting by humans exerts tremendous pressure on wild populations, resulting in both ecological and evolutionary change. Writing in Proceedings of the National Academy of Sciences, Darimont et al.1 demonstrate the consequences of harvesting on changes that occur in observable traits (phenotypes) in populations experiencing one of three different types of selection.

Using a meta-analysis of previously published data, the authors compared the rates of phenotypic change in 40 populations subject to human harvesting with the rates seen in 20 systems that experienced selection from natural forces (for example, Darwin's finches) and with the rates in 25 systems that experienced other human disturbance (for example, pollution). The human-harvested organisms included fish, ungulates, invertebrates and even plants, examples of these being respectively sockeye salmon, bighorn sheep, marine snails and gingseng.

The recorded rates of change in harvested populations were astonishingly high, outpacing changes arising from natural agents by 300% and those from other anthropogenic causes by 50%. The changes were both rapid and dramatic in magnitude: declines in morphological traits (such as body size) averaged about 20%, and shifts in life-history traits (such as reproductive age) averaged nearly 25%. Darimont et al. conclude with a thoughtful discussion about the effects of such changes on harvestable biomass, and whether populations will show a similarly swift return to previous trait distributions if recovery is a goal of resource management.

Darimont and colleagues1 argue that large and rapid phenotypic changes have profound implications, regardless of whether they are due to genetic (evolutionary) or plastic (purely ecological) responses. Although a plastic response will alter some aspects of population dynamics, genetic changes have further effects that, from a management perspective, suggest that it is essential to identify them. Most notably, a population is expected to recover more slowly from genetic than from plastic changes (Fig. 1a), and genetic change might even be irreversible2. Given the high harvesting pressure on larger and older individuals, the result might be 'juvenescence' of populations, possibly leading to increased variability in abundance and reduced genetic variability3. The evolution of traits related to growth and reproduction could also lead to reductions in the yield and profits derived from exploited populations.

Figure 1: Genetic (evolutionary) and plastic (ecological) responses to mortality from human harvesting and natural forces.

a, Harvesting-induced evolution (of, for example, age at sexual maturation) is often faster than evolution caused by natural mortality. Ecological changes induced by harvesting pressure occur more quickly than evolutionary changes, but the latter are expected to be longer lasting when harvesting-induced mortality is removed. Natural mortality could also cause a rapid ecological response, but this is not shown. b, Harvesting pressure — from such activities as trawling or trophy hunting — operates on a trait such as body size in the opposite direction to mortality from natural sources7.

Why do harvesting-induced changes outpace those caused by other sources? The response, R, of a trait to selection is determined by the breeder's equation, R = h2S, where h2 is heritability of the trait (the ratio of additive genetic variance to phenotypic variance) and S is the selection differential (a measure of the strength of selection)4. Assuming that heritabilities are not systematically different among the various populations included in Darimont and colleagues' survey, it seems that the selection differential generated from harvesting is higher than that from other sources. An obvious reason is that the mortality rates produced from human fishing and hunting are often far greater than natural mortality rates.

Most notably, mortality arising from commercial fishing can exceed natural mortality by as much as 400% (ref. 5). In the years immediately preceding the collapse of stocks of Atlantic cod off Newfoundland and Labrador, the likelihood6 of an individual of harvestable size being caught was as high as 89%. The strength of the selective pressures generated by such high rates of mortality no doubt contributes to the observed faster rates of change generated by human harvesting than by other sources of trait change.

Besides resulting in high rates of mortality, the selective pressures induced by harvesting typically act in the opposite direction to those caused by natural sources of mortality7 (Fig. 1b). Natural mortality rates of young or small individuals are usually much higher than for old or large individuals, for example because predators are often limited by the size of their gape. Human predators are different, usually preferring to hunt and capture large prey for sport, profit or sustenance. Mortality from trawl fishing increases with a fish's body size because the smallest fish can escape through the trawl mesh. Similarly, many recreational fisheries have size limits that ban the taking of small individuals, and trophy hunters often preferentially target large game with, for example, impressive antlers.

Natural mortality is, of course, always acting in the background. As the different sources of selection often oppose one another7, this might produce intermediate rates of change. But harvesting mortality is usually far stronger than natural mortality. So selection from the former would be expected to swamp that due to the latter, and rates of change may still be fairly high and faster than those induced by natural mortality alone.

One commonly observed trend, and one for which Darimont et al.1 obtained data, is that fishing causes the evolution of sexual maturation at younger ages and smaller sizes8. In the absence of fishing, it pays to be big, both because you can escape predation and because there is a positive relationship between body size and fecundity. Small females that reproduce early in life will still gain some reproductive advantage, but it will not be as strong as for those females that reproduce when large and highly fecund. In the presence of fishing, being large becomes a liability, as large individuals are targeted by the fishery. In this case, females that delay maturation until they are larger could have a total lifetime reproductive fitness of zero, making the effect of selection much stronger, because the fish will be killed before having the chance to reproduce. Therefore, natural selection favouring large size and delayed maturation is often weaker than the selection induced by fishing, possibly contributing to the disparate rates of change reported by Darimont and colleagues1.

One implication of this study1 is that natural-resource management should include consideration of evolutionary responses to fishing and hunting. Evolutionary-impact assessments can be used for this purpose, where predictive models are used to evaluate the potential evolutionary impact of alternative management actions8. The underlying model should include both ecological and evolutionary parameters (and the interaction between them), as well as 'utility' measures such as yield and profit. The evidence for harvesting-induced evolution remains under debate, and the relative importance of ecological and evolutionary contributions to observed change remains contentious9,10,11,12. Nonetheless, a precautionary approach13 requires that management strategies be designed under the assumption that harvesting-induced evolution might occur. The potential costs of ignoring that possibility are severe — most notably, the resulting changes may be difficult or impossible to reverse.


  1. 1

    Darimont, C. T. et al. Proc. Natl Acad. Sci. USA 106, 952–954 (2009).

  2. 2

    Stenseth, N. C. & Rouyer, T. Nature 452, 825–826 (2008).

  3. 3

    Anderson, C. N. K. et al. Nature 452, 835–839 (2008).

  4. 4

    Falconer, D. S. & Mackay, T. F. C. Introduction to Quantitative Genetics (Longman, 1996).

  5. 5

    Mertz, G. & Myers, R. A. Can. J. Fish Aquat. Sci. 55, 478–484 (1998).

  6. 6

    Myers, R. A. et al. Mar. Ecol. Prog. Ser. 138, 293–308 (1996).

  7. 7

    Carlson, S. M. et al. Ecol. Lett. 10, 512–521 (2007).

  8. 8

    Jørgensen, C. et al. Science 378, 1247–1248 (2007).

  9. 9

    Hilborn, R. Fisheries 31, 554–555 (2006).

  10. 10

    Law, R. Mar. Ecol. Prog. Ser. 335, 271–277 (2007).

  11. 11

    Browman, H. I. et al. Science 320, 47 (2008).

  12. 12

    Kuparinen, A. & Merilä, J . Science 320, 47–48 (2008).

  13. 13

    Food and Agriculture Organization Precautionary Approach to Capture Fisheries and Species Introductions FAO Tech. Guidelines for Responsible Fisheries No. 2 (FAO, 1996).

Download references

Author information

Additional information

For more on Darwin, see http://www.nature.com/darwin .

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Stenseth, N., Dunlop, E. Unnatural selection. Nature 457, 803–804 (2009) doi:10.1038/457803a

Download citation

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.