Plants are being moved across the globe at an increasing rate both within and outside their current distributional range. Research in the past few decades has focused intensely on invasive exotic plant species that are moved outside their distributional range. Such invaders can have dramatic impacts on indigenous communities through altered biotic interactions with competitors, mutualists and antagonists (Mitchell et al., 2006). But likewise, translocation within the distributional range exposes nonnative plant genotypes to interactions with nonlocally coevolved competitors, mutualists and antagonists. A major concern in restoration ecology is that translocation of seeds from foreign seed sources introduces plant genotypes that are maladapted to local conditions and subsequently hybridize with locally adapted conspecifics, decreasing mean population fitness (Hufford & Mazer, 2003). Interestingly, while local adaptation of plants to their abiotic environment is well documented, warranting the concern, studies of plant local adaptation to their biotic environment have shown less unequivocal results, ranging from local adaptation to local maladaptation for interactions with antagonists (Greischar & Koskella, 2007). A large reciprocal transplant field study by Crémieux et al., reported in this issue of New Phytologist (pp. 524–533), offers a prime example of the diversity of outcomes of altered biotic interactions that we can observe following translocation of plants within their current distributional range even within a single community. In one plant species, local plant genotypes were more resistant to the local demes of a specialist antagonist than nonlocal plant genotypes, suggesting plant local adaptation to important aspects of the biotic environment. However, in another plant species, the reverse pattern was observed. The take-home message from this study is a teeth-grinder for restoration ecologists: it is hard to make a general prediction as to whether foreign seed provenances pose a risk of introducing alleles causing low biotic resistance in the restoration area or not.
Plant local adaptation to the abiotic vs biotic environment
What do we know about the relative contributions of adaptation to abiotic versus biotic aspects of the environment to patterns of plant local adaptation? The surprising answer is: very little. Plant local adaptation can be defined as the higher fitness of local plants at their home site compared with that of nonlocal plants. Performance at the home site is governed by the ability to cope with local abiotic as well as biotic conditions. Moreover, responses to abiotic and biotic aspects of the environment may be contingent upon each other (Fig. 1). First, host condition often affects susceptibility to antagonists, so differentiation at loci involved in coping with important aspects of the abiotic environment may result in local plants attaining larger size, having different chemistry, or capturing more resources for defense than nonlocal plants, affecting their interactions with competitors, mutualists, and antagonists (Fig. 1, bold arrows). Secondly, abiotic conditions can affect the expression of genetic differences in resistance among plants and host exploitation between antagonists (Fig. 1, dashed arrow) even to the extent of altering the direction of local adaptation in plant–antagonist interactions (Laine, 2008). Plant evolutionary responses to environmental factors that do not coevolve with the plant, such as abiotic conditions, are expected to result in patterns ranging from neutral to plant local adaptation. Indeed, studies emphasizing abiotic aspects of the environment have generally shown plants to be locally adapted, although this pattern is not universal (Hufford & Mazer, 2003). By contrast, plant evolutionary responses to biotic factors that continuously coevolve to maximize plant exploitation could conceivably result in patterns ranging from local adaptation to local maladaptation, and indeed, this whole range is observed for interactions with antagonists (Greischar & Koskella, 2007). It is therefore curious to note that research on plant local adaptation has seldom followed an integrated approach of assessing the relative contributions of abiotic and biotic adaptation to overall local adaptation. On the one hand, studies have addressed local adaptation at large, typically focusing on local adaptation along abiotic environmental gradients without explicitly considering the contribution of adaptation to the biotic environment. On the other hand, local adaptation in plant–antagonist studies is often studied from the antagonist's perspective (Kawecki & Ebert, 2004), in work inspired by questions about the evolution of parasites in host–parasite interactions, not surprisingly without considering its contribution to host local adaptation at large. Indeed, an integrated approach would require extensions of traditional reciprocal transplant studies to include treatments that manipulate, control or otherwise factor out particular aspects of the biotic environment, for instance using exclosures or fungicide/insecticide treatments (Fine et al., 2004; Abdala-Roberts & Marquis, 2007). Such studies can offer important insights but are still rare.
Plant local (mal)adaptation to the biotic environment: what should we expect?
The study by Crémieux et al. shows plant local adaptation in a plant–insect herbivore interaction and plant local maladaptation in a plant–fungal pathogen interaction. Should we expect such widely divergent outcomes? And what do we know about mechanisms causing adaptation versus maladaptation in biotic interactions? Divergent outcomes are not unexpected. Most host–parasite models assume that parasites specialize on locally common host genotypes, creating a fitness advantage for rare, resistant, genotypes, leading to time-lagged cycles of host resistance and parasite infectivity alleles. A crucial assumption is that parasites have an evolutionary advantage over their hosts by virtue of their shorter generation times, higher mutation/recombination rates, and/or higher relative migration rates. As a consequence, they can closely track their locally common host genotypes, and are expected to be better at infecting local than nonlocal hosts which are likely to be in a different phase of the coevolutionary cycle. Thus, depending on where in the cycle we sample hosts and parasites, we could observe the whole spectrum from host local adaptation to host local maladaptation, but the latter should prevail (Kaltz & Shykoff, 1998). This is indeed what experimental studies show (Greischar & Koskella, 2007). As predicted by theory (Lively, 1999), the rare cases of host local adaptation are more frequently observed for interactions with parasites that have low migration rates relative to their hosts (Greischar & Koskella, 2007; Hoeksema & Forde, 2008). It would be interesting to see whether the fungal pathogen used by Crémieux et al. is characterized by a high and the insect herbivore by a low relative migration rate, as this could be one of the explanations for the patterns of parasite local adaptation and maladaptation that they observed in these cases, respectively. Opposite patterns for insect herbivores and fungal pathogens may also follow from differences in underlying interaction mechanisms. Kniskern & Rausher (2001) contrasted two basic interaction mechanisms, toxin–detoxifier and elicitor–receptor systems. Toxin–detoxifier systems (Fig. 2a) give rise to a true arms race: plants produce toxins and local antagonists evolve ways to detoxify them, prompting plants to evolve modified toxins, and so on. By contrast, elicitor–receptor systems (Fig. 2b) are based on recognition of antagonists by the plant, required to mount defenses. This gives rise to an information race: antagonists produce molecules (elicitors) and local plants evolve receptors that can recognize them, prompting antagonists to evolve modified elicitors evading recognition, and so on. As Kniskern and Rausher point out, if local populations follow different coevolutionary trajectories, the two systems lead to opposite predictions for local adaptation. Plants translocated to nonlocal populations will fail to recognize elicitors of many novel antagonists (Fig. 2a) but will possess toxins that many novel antagonists cannot deal with (Fig. 2b). Thus, toxin–detoxifier systems, considered as an important mechanism in plant–insect interactions, predict an overall pattern of plant local maladaptation whilst elicitor–receptor systems, considered as an important mechanism in plant–pathogen interactions, predict an overall pattern of plant local adaptation to antagonists. Unfortunately, these predictions are exactly opposite to the patterns observed by Crémieux et al., who showed plant local adaptation in the plant–herbivore and plant local maladaptation in the plant–pathogen system. Clearly, while underlying interaction mechanisms may be important determinants of patterns of local adaptation, studies clarifying their actual involvement in specific plant–herbivore or plant–pathogen interactions are needed to test such ideas.
Understanding patterns of plant local adaptation necessitates a better integration of studies on plant adaptation to their abiotic environment with studies on reciprocal adaptation of plants and their biotic environment. The study by Crémieux et al. is illustrative in this respect. While velvet grass (Holcus lanatus) is locally maladapted to one of its important fungal antagonists, in a related paper (Bischoff et al., 2006) the authors show that local plants of this species nevertheless attain higher fitness than nonlocal conspecifics. Thus, even when plants are locally maladapted to important antagonists, this maladaptation may or may not be strong enough to override local adaptation to the abiotic environment, and an overall pattern of plant local adaptation can still be found. One of the current challenges is to understand the varied patterns observed in such biotic interactions, ranging from local adapation to local maladaptation. While theoretical studies are increasingly providing directions for focusing research on potentially important underlying mechanisms (the impact of relative migration rates, transmission modes, interaction mechanisms, etc.) recent meta-analyses make clear that a major bottleneck in detecting the impact of such factors on patterns of local adaptation is the number of empirical studies available for such analyses.