Herbivory tolerance and coevolution: an alternative to the arms race?


  • Arthur E. Weis,

    1. Department of Ecology and Evolutionary Biology 321 Steinhaus Hall, University of California, Irvine, CA 92697-2525, USA (Author for correspondence: tel +1 949 824 2241; fax +1 949 824 2181; email aeweis@uci.edu)
    Search for more papers by this author
  • Steven J. Franks

    1. Department of Ecology and Evolutionary Biology 321 Steinhaus Hall, University of California, Irvine, CA 92697-2525, USA (Author for correspondence: tel +1 949 824 2241; fax +1 949 824 2181; email aeweis@uci.edu)
    Search for more papers by this author

Herbivores are important in most terrestrial ecosystems and can reach outbreak proportions, causing spectacular levels of damage to many plant species. It is thus widely believed that herbivory is important in plant population dynamics and evolution. Furthermore, plants have a wide array of defenses against herbivores. Plant defense systems are broadly comprised of two components: resistance and tolerance. Resistance traits reduce damage levels either by lowering the probability of herbivore attack or by decreasing the amount of tissue removed. Because resistance factors can affect herbivore fitness, an evolutionary increase in resistance is expected to select for herbivore counter-measures. Tolerance traits, the second type of defense, minimize the adverse effects of damage by enabling the plant to survive, regrow and reproduce after an attack occurs. While it was previously assumed that an evolutionary increase in tolerance has no effect on herbivore fitness (Stowe et al., 2000; Tiffin, 2000), it has been suggested that tolerance traits may impose selection on herbivores (Stinchcombe, 2002). In this issue, Espinosa & Fornoni (pp. 609–614) present the first experimental test of this hypothesis. They found that increasing tolerance levels in the annual plant Datura stramonium (Solanaceae) has no detectable effect on fitness components of its defoliating enemy, the leaf beetle Lema trilineata (Coleoptera: Chrysomelidae). This finding lends credence to the conventional wisdom that tolerance does not result in an evolutionary ‘arms race’, and also supports the notion that tolerance is a more evolutionarily stable form of defense than resistance.

‘Tolerance can thus modify the plant–herbivore arms race, but can tolerance directly influence the evolution of herbivores?’

The arms race metaphor

Evolutionary ecology has made much progress over the past four decades in understanding resistance, but the study of tolerance gained momentum only in the mid-1990s (Stowe et al., 2000). Perhaps this is because evolutionary ecologists have been fond of the arms race metaphor for the evolution of plant defense and herbivore counter-defense. If a mutation reduces herbivore attack, that mutation should spread (provided any adverse effects of the mutation are small relative to the benefit of reduced damage). Consequently, the herbivore's food source is diminished. However, a subsequent mutation in the herbivore could allow it to counter plant resistance. This new herbivore allele would then spread, increase herbivore load, and thus favor new plant resistance mutations. And so plant and enemy advance up a coevolutionary spiral.

The arms race metaphor became popular with the publication of Ehrlich & Raven's (1964) paper on the coevolution of butterflies and their host plants. They proposed that the evolutionary diversification of defensive secondary chemicals in a plant clade is followed by diversification in any butterfly clade that evolves the proper adaptations to deal with the chemicals. As a result, phylogenetically related caterpillars attack phylogenetically related plants. Many saw an easy connection between this macroevolutionary scenario and microevolution. However, tolerance does not fit neatly into this frame of thought.

Resistance and tolerance: alternative strategies?

Several models, inspired by the suggestion that resistance and tolerance are alternative forms of defense (Van der Meijden & Verkaar, 1988), have examined their joint evolution. The intensity of selection on resistance should depend on the level of tolerance, and vice versa (Abrahamson & Weis, 1997). If a plant is completely resistant, there is no selective advantage to increased tolerance as there is no damage from which to recover. A completely tolerant plant gains no selective advantage by reducing damage through resistance, because damage does not lower fitness. Several models suggest that resistance and tolerance represent alternative peaks on the plant's adaptive landscape (Fineblum & Rausher, 1995; Mauricio et al., 1997; Roy & Kirchner, 2000). Selection pushes the population towards one peak or the other depending on the relative costs and benefits of the two defenses and on the expected frequency and intensity of attack. However, other models show that both defense types can persist under some conditions. Strongly negative genetic correlations between resistance and tolerance can promote polymorphism in each (Tiffin, 2000). Mixed defense strategies, with resistance and tolerance both maintained at intermediate levels, are possible when the cost of each defense rises disproportionately with its effectiveness (Fornoni et al., 2004).

Tolerance can thus modify the plant–herbivore arms race, but can tolerance directly influence the evolution of herbivores? It has been assumed that tolerance will not exert selective pressure on herbivores because tolerance is the ability of the plant to recover from damage and, unlike resistance, is not related to traits that would obviously influence herbivore fitness. However, a model by Restif & Koella (2003) suggests that tolerance can impose selection on the plant's enemies, at least in the case of microbial pathogens. Assuming that tolerance is costly in terms of plant fecundity, they show that a pathogen will be under selection for lower virulence as tolerance increases. This is because new, uninfected hosts are produced at a slower rate, so it pays for the pathogen to extend the life span of its current host until a new one comes along. Given its basis in microbial life histories, Restif and Koella's model probably has limited application to insect and vertebrate herbivores. But are there other ways in which increased tolerance can affect enemy fitness?

Stinchcombe (2002) suggested two ways in which tolerance could influence herbivore evolution. First, the same genes that code for resistance traits may also code for traits that influence tolerance (pleiotropy), which would mean that a change in these genes would simultaneously affect resistance and tolerance in the plant as well as the response of the herbivore. Second, traits that allow tolerance may directly influence herbivore fitness, mainly through influencing the quality, quantity or availability of tissue used as food by the herbivores. Traits that can increase tolerance include altered resource allocation, architecture, physiological processes such as photosynthesis, and changes in phenology (Strauss & Agrawal, 1999; Stowe et al., 2000; Tiffin, 2000). A widely cited example of tolerance is release from apical dominance and the production of lateral buds (Paige & Whitham, 1987). Suppose that the resprouted tissue is just as likely to be eaten, and in the same quantity as the original tissue (no change in resistance), but that this tissue is of lower quality than the original leaves. If herbivores that have higher rates of consumption or use this resource more efficiently have higher fitness than herbivores without these attributes, then these traits may be under selection, and tolerance has the potential to cause evolutionary changes in the herbivores.

Experiments on tolerance: practical challenges and results

The relationship between tolerance and selection on herbivores may remain largely unexplored because of both theoretical biases, and practical difficulties in disentangling this relationship from other factors. A strong genetic correlation between resistance and tolerance would make it difficult to detect their individual direct effects. Stinchcombe (2002) suggested a straightforward way of circumventing this problem by creating lines of genetically related organisms that are equally resistant, but that differ systematically in tolerance. This breaks the genetic covariance between resistance and tolerance that would otherwise confound attempts to examine tolerance directly.

Espinosa & Fornoni implemented this protocol with lines of the plant D. stramonium that were similar in resistance, but varied in tolerance to damage by the leaf beetle L. trilineata. They measured fitness-related traits in beetles fed on plants differing in tolerance, thus estimating the degree to which tolerance can impose selection on the beetles. They also measured the degree of genetic variation in the beetle traits. Such genetic variation is necessary for evolution to occur in the presence of selection. While they found genetic variation in survival, they did not find genetic variation in the other traits measured. Most importantly, they found no evidence that the host plant's level of tolerance has any measurable effect on the herbivore's fitness.

The work by Espinosa & Fornoni suggests that theoreticians have been on the right track in assuming that tolerance imposes no selection on nonmicrobial herbivores. Of course, they worked with only one species of plant and herbivore, so counter-examples may be out there. It is clear that, like Stinchcombe (2002), we continually need to re-examine basic assumptions and become more aware of potential biases that could obscure our ability to depict and understand the complex relationships between plants and their natural enemies.


The zeitgeist of the cold war, which was in full swing when Ehrlich & Raven (1964) published their coevolution paper, no doubt lent a special resonance to the arms race metaphor. Although the metaphor has limitations (Thompson, 2005), the growth in our understanding of plant resistance owes much to the inspiration it gave to a generation of investigators. But the cold war also offered a metaphor that could be applied to plant tolerance –‘civil defense’ (Karban & Baldwin 1997). A nation could develop infrastructure that allows rapid economic recovery in the event of a first strike. If governments had invested as heavily in civil defense as in strategic defense, would evolutionary ecologists have turned their attention to tolerance decades earlier? The general lesson here is that metaphors in ecology and evolution can be heuristic, but can also introduce hidden biases and assumptions of which we all should be aware.


We acknowledge the financial support from NSF grant DEB-0345030.