Phenotypic plasticity and plant invasions: is it all Jack?


*Correspondence author. E-mail:


  • 1Phenotypic plasticity has long been suggested to facilitate biological invasions where a general purpose genotype or ‘Jack-of-all-trades’ strategy could facilitate invasion should native species be, on average, more specialized. Current understanding of the importance of phenotypic plasticity is limited by methodological difficulties yet glasshouse experiments, paralleling those used in evolutionary biology, are increasingly being used to assess whether invasive species have high phenotypic plasticity or not. Unfortunately, these studies have several major limitations.
  • 2In general, glasshouse experiments have quantified relatively labile means of expressing phenotypes, such as plant biomass, rather than changes in plant development. As some environments favour plant growth and reproduction more than others, simply quantifying allometric or physiological responses relating to differential plant growth may not reveal much about phenotypic plasticity.
  • 3Plasticity is often a comparative rather than an absolute measure. A range of comparators have been used, for example, unrelated natives, congeneric natives, other congeneric aliens or conspecifics from the source region. Thus, unlike many life-history traits used in analyses of invasion, phenotypic plasticity is strongly context-dependent and limits comparison across different studies.
  • 4Phenotypic plasticity is assumed to lead to a greater breadth of environmental conditions across which a species can maintain positive population growth and increase the likelihood of invasiveness. Yet, most studies examine only a partial subset of the full environmental range experienced by the species. If the reaction norm of a target species and its comparator vary independently across an environmental gradient, this partial approach can present different interpretations of phenotypic plasticity.
  • 5Rather than simply quantifying greater phenotypic plasticity in invasive species, research questions should be directed at better understanding its role in the geographic distribution, successful colonization, population persistence and/or high local abundance of invasive species in the introduced range. These issues require integrated measures of plant performance rather than crude differences in individual traits across an environmental gradient. As yet, there is only limited appreciation of the role of phenotypic plasticity in any one of these areas and there is a need to extend studies beyond glasshouse experiments.


Phenotypic plasticity has long been suggested to facilitate biological invasions (Daehler 2003; Rejmánek et al. 2005; Pyšek & Richardson 2007) under the assumption that a general purpose genotype or ‘Jack-of-all-trades’ strategy would facilitate invasion, where native species were more specialized cf. ‘Jack of all trades, master of none, though ofttimes better than master of one’. It is clearly intuitive that the greater the breadth of environmental conditions across which a species can maintain positive population growth, the greater the likelihood it has of being invasive. The extent of such ‘niche breadth’ may reflect adaptive homeostasis (e.g. physiological tolerance), local adaptation and/or phenotypic plasticity (Dudley 2004). The study of phenotypic plasticity has in large part attempted to disentangle these three factors, focusing on the genotype–environment interaction (DeWitt & Scheiner 2004). The fundamental advances in both evolutionary and functional ecology that have arisen from such studies are relevant irrespective of whether plants are invasive or not (DeWitt & Scheiner 2004; Pigliucci 2004). Introduced plants do pose interesting opportunities to further the study of phenotypic plasticity but the extent to which these studies may improve understanding and predictions of invasions is less clear (Ghalambor et al. 2007). This is due to the different aims in plant invasions research, being primarily to determine whether one or more life-history traits can robustly discriminate between potentially invasive and non-invasive species, explain the success of biological invaders or predict their impact (Hulme 2003, 2006). These are challenging aims and establishing clear links to phenotypic plasticity have been limited due to the difficulties in quantifying this trait (Rejmánek et al. 2005). Increasingly, glasshouse and/or common garden experiments have attempted to quantify phenotypic plasticity in invasive species and the evidence to date is, at best, equivocal with only a similar number of studies finding as failing to detect significant trends (Bossdorf et al. 2005; Richards et al. 2006). However, while such studies may be appropriate for disentangling genotype–environment interactions, they may not fit the broader aims of invasion research particularly well. These limitations, their consequences and a possible range of options to address them are outlined below.

Challenges in linking phenotypic plasticity and plant invasiveness

A common definition of phenotypic plasticity is the environmentally sensitive production of alternative phenotypes by given genotypes. Phenotypic plasticity in plants has largely focused upon developmental processes rather than other labile means of expressing phenotypes, such as physiological or allometric shifts (DeWitt & Scheiner 2004). There are certainly some examples where introduced species exhibit dramatic adaptive differences in morphology and resource allocation across environmental gradients (Williams, Mack & Black 1995; Spector & Putz 2006). However, the majority of studies examining invasive plant species assess plasticity in more labile attributes such as plant biomass, growth rate, nutrient content, photosynthesis, etc. (Richards et al. 2006). A confounding factor in understanding phenotypic plasticity in these studies is that some environments favour plant growth and reproduction more than others and simply quantifying allometric or physiological changes relating to differential plant growth may not reveal much about phenotypic plasticity or adaptive responses (Dudley 2004). If allocation to plant structures involved in resource capture or reproduction is allometric, then any environmental factor that affects plant size will alter allocation patterns. ‘Apparent plasticity’ may be observed when experimental conditions only affect plant growth rate such that comparisons after a fixed time period give the impression of different allocation strategies although allocation at a given size is the same in different environments (Weiner 2004).

Most studies attempting to examine the evolutionary significance of phenotypic plasticity have been performed under controlled glasshouse or common garden conditions, and this is equally true for the approaches used in biological invasions (Richards et al. 2006; Ghalambor et al. 2007). While evolutionary studies focus on intraspecific relationships and compare the responses of multiple genotypes of the same taxon across environmental gradients, the choice of comparator is less clear in invasions. Various methodological approaches have been employed including comparing the response to environmental gradients of aliens with unrelated natives, congeneric natives, other congeneric aliens or conspecifics from the region of origin (Richards et al. 2006). Only in the latter case is the choice of comparator constrained in a similar way to evolutionary studies of phenotypic plasticity (Ghalambor et al. 2007). While comparative experiments of closely-related species are indeed often powerful tools, their power is derived in part by the appropriate choice of comparators. Phylogenetic controls are important when assessing differences in plant life-histories but to assess likelihoods of invasion risk, it is equally important to ensure that the native or non-native congenerics co-occur (or could potentially) in similar ecosystems (Lambdon & Hulme 2006). Assessments of other plant life-history attributes highlight that the relative importance of any single trait is probably dependent on the invaded ecosystem (Lloret et al. 2005). Consistent with this view are theoretical models that suggest the degree of phenotypic plasticity of recipient ecosystems may in turn determine the importance of phenotypic plasticity in the invasive species (Peacor et al. 2006). Comparisons with native species, even if not congeners, may therefore be appropriate. Thus unlike many of the traits used in comparative analyses of invasion risk (e.g. seed size, life-form, etc.) measures of phenotypic plasticity are strongly context-dependent and may not be comparable across different studies where different gradients and/or comparators have been used.

As a result, a fundamental difference between phenotypic plasticity and many other life-history traits is the difficulty in obtaining an absolute standardized measure. At least 17 different indices have been employed as measures of phenotypic plasticity and most cannot be standardized across traits or compared among different species (Valladares, Sanchez-Gomez & Zavala 2006). Furthermore, most of these measures refer to comparison of a single plant trait yet, rather than a single trait, variables that integrate different aspects of plant performance may more accurately describe invasiveness (Zou, Rogers & Siemann 2007). Examples include an ‘invasion criterion’ that reflects the mean population growth rate (Willis & Hulme 2002), or phenotypic inertia that combines plant survival and performance (Milberg, Lamont & Pérez-Fernández 1999). Unfortunately, such measures are no easier to calculate or compare but are probably more comprehensive assessments of fitness than changes in plant morphology, biomass or physiology.

Glasshouse studies are often constrained by the breadth of the environmental gradient they can simulate, and many only examine two environmental states (low- vs. high-nutrients, light, etc.). Reaction norms for continuously distributed traits, such as biomass, are usually described as a line or curve on a plot of the environmental value vs. the phenotypic value (Richards et al. 2006; Ghalambor et al. 2007). Evidence for plasticity is assessed in relation to the difference in the slope of the relationship between the target species (or genotype) and its comparator(s). Based on the relative differences in the slope of the reaction norms, Richards et al. (2006) put forward three phenotypic plasticity strategies that may characterize alien plant performance. The ‘Jack-of-all-trades’ strategy describes cases where phenotypic plasticity assists in maintaining constant fitness (measured as population growth rate or inferred from changes in reproduction or survival rates) across an environmental gradient. The ‘Master-of-some’ strategy reflects situations where phenotypic plasticity enables fitness to be higher under certain environmental conditions. The third strategy, ‘Jack-and-master’ combines elements of the two other strategies. However, from an invasions perspective, what is more pertinent in terms of plasticity is the breadth of environments over which positive fitness is maintained, e.g., fitness homeostasis. Examining only a partial subset of the full environmental range experienced by the species will not fully describe the reaction norm across the environmental gradient.

Some of the problems in this ‘partial’ approach can be illustrated by examining hypothetical reaction norms of taxa across a common environmental gradient. Imagine two species that differ in terms of both the breadth of environmental conditions across which they maintain positive fitness as well as the maximum fitness achieved (Fig. 1). Although a range of reaction norms are possible, the overwhelming evidence indicates that the response under the fullest breadth of environmental conditions under which a plant species occurs is likely to be unimodal (Austin 1999). When examining the entire environmental gradient, it might be concluded that Species 2 exhibits greater phenotypic plasticity as it spans the wider range of environments. Now, suppose three studies attempt to compare Species 1 and 2, but examine different regions of the same environmental gradient (Fig. 1). In study A, Species 1 although having initially lower fitness, exhibits a stronger response to environmental conditions than Species 2 and mirrors the ‘Master of some’ strategy. In B, Species 1 has higher fitness across the environmental conditions and if these are assumed ‘favourable’ this would qualify the strategy as ‘Jack-and-master’. Finally in C, Species 1 hardly responds to the environmental gradient examined but Species 2 does and this conforms to the expectation of the ‘Jack-of-all-trades’ strategy. While the precise details of the schematic are not important it illustrates the risk that when slopes are compared the same species assessed along the same environmental gradient could be classed differently depending on where along the continuum the studies took place.

Figure 1.

Schematic representing two hypothetical species that differ in terms of both the breadth of environmental conditions across which they maintain positive fitness as well as the maximum fitness achieved. Imagine three studies (A, B and C) each comparing Species 1 and 2, but examining different regions of the same environmental gradient. The perceived strategies of each species are then contingent on where along the environmental gradient the study took place.

Phenotypic plasticity in plants is known to occur in response to gradients in temperature, light, water, nutrients, CO2, inter- and intraspecific competition, type of soil, herbivory and mechanical perturbation (Dudley 2004). Each of these gradients has been shown to facilitate plant invasions (Hulme 2006) and thus the role of phenotypic plasticity cannot be disregarded. However, given the number of potential environmental gradients and the fact that they are likely to covary, and even have synergistic effects on the level of plasticity observed (Portsmuth & Niinemets 2007), assessments along a single gradient may be over simplistic. Plant strategies identified along one gradient may differ along another and identifying a particular strategy across multidimensional niche space may not be straightforward. Under such a complex scenario, it is essential to frame the right question so as to focus on the relevant gradient(s).

How should phenotypic plasticity be addressed in invasion studies?

Plant invasiveness is poorly defined but includes wide geographic distribution, successful colonization of new areas, population persistence and/or high local abundance in the introduced range. As yet, there is only limited appreciation of the role of phenotypic plasticity in any one of these areas, but it may be possible to identify useful avenues for more targeted studies on phenotypic plasticity. Geographic distribution will ultimately be determined by climatic limits and phenotypic plasticity is commonly found for plants in relation to temperature response which is often an important determinant of range size (Atkin et al. 2006). Persistence in a variable environment (whether temporal or spatial) is likely to be facilitated by phenotypic plasticity and this could be especially important in successful colonization events (Ghalambor et al. 2007). Such gradients occur at a relatively fine spatial and temporal scale and may include nutrients, water or light. However, plasticity may play a lesser role in the high local abundance of alien plants if a trade-off exists between the average abundance and the degree of phenotypic plasticity. This would result in less plastic species attaining a higher abundance under their optimum conditions than more plastic species (Bradshaw 1965; Balaguer et al. 2001).

Surprisingly few studies have examined phenotypic plasticity, even if only assessed at the population level, in relation to the geographical distribution of alien plant species. These few studies support the view, albeit predictable, that a broad climatic tolerance is an essential component to a widespread geographic distribution (Williams et al. 1995; Willis & Hulme 2002). However, most glasshouse experiments focus on relative performance of alien plants across nutrient, water or light gradients (Bossdorf et al. 2005; Richards et al. 2006). Unsurprisingly, many of these studies confirm what is often observed regarding the abundance of species across different habitats. An index of potential phenotypic plasticity could be gained far more rapidly and for many more species by assessing habitat breadth from floristic accounts in the native range. Indeed, probably the most interesting aspect of such experiments would be to compare expectations from the native habitat distribution with performance across environmental gradients in the invaded region. Phenotypic plasticity may enable species to colonize environments not experienced in the native range, for example, invasion of alpine tussock grasslands of New Zealand by European weeds (Radford, Dickinson & Lord 2007) or novel anthropogenic habitats (Mihulka et al. 2006). To test the role of phenotypic plasticity in colonization success would perhaps invoke studies comparing species with different life-history strategies along a competition-colonization gradient. Although studies on native species reveal contradictory trends, for example, short-lived tropical pioneer trees which are the primary colonists of treefall gaps show low levels of plasticity compared to shade-tolerant canopy trees (Rozendaal, Hurtado & Poorter 2006). Few studies attempt to tease apart the role of phenotypic plasticity in population persistence in a variable environment. If persistence is to be assessed, the environmental gradients may need to be more subtle, for example, shade vs. sunfleck and the response should include a measure of population survival rather than plant biomass. It should be evident that depending on the aspect of invasion of interest, different types of comparator may be more appropriate. Rather than simply assessing whether plasticity occurs, the questions need to be refined to address the particular aspect of invasion that is of interest.

It is therefore worth considering whether invasion biologists should explore alternatives to glasshouse experiments when assessing phenotypic plasticity. There is a long history of exploring reaction norms of species within plant communities along environmental gradients (Austin 1999). Statistical tools have been developed to fit reaction norms and thus quantitatively describe individual species abundance across complex environmental gradients (Oksanen & Minchin 2002) but as yet have not been used to address issues relating to biological invasions. These studies might reveal relationships between average response breadth of the recipient community and the reaction norms of individual alien species. The traditional comparative approach examining life-history traits could incorporate information on the variance as well as mean value for traits and would identify whether the coefficient of variation in a trait was more important than the mean for colonizing species. The output of both reaction norm and variation analyses would not be able to distinguish among adaptive homeostasis, local adaptation and/or phenotypic plasticity but may highlight suitable species, traits and environmental gradients along which to explore the importance of plasticity further. There may even exist plant traits that can be used as proxies for phenotypic plasticity in a similar way as brain size has been used to identify behavioural plasticity as a key trait in bird invasions (Sol, Timmermans & Lefebvre 2002).


The foregoing highlights several differences between evolutionary ecology and invasion biology perspectives regarding phenotypic plasticity. The former has a stronger focus on developmental vs. labile traits, genotype vs. population level analyses and the slope rather than breadth of the reaction norm. While these two fields come together when exploring the evolution of invasiveness in introduced species, they diverge when exploring the wider implications of biological invasions. For example, the ‘general purpose genotype’ in an evolutionary context describes a genotype that is superior in all environments (Ghalambor et al. 2007), yet invasion ecologists use this term to describe a species that performs adequately across a range of environments but is not superior in any, for example, a ‘Jack-of-all-trades’ (Mihulka et al. 2006). As a result, the general purpose genotype is an example of low plasticity in one field (Ghalambor et al. 2007) and high plasticity in another (Rejmánek et al. 2005). Phenotypic plasticity suffers from too broad a definition as almost all biological processes are influenced by the environment (DeWitt & Scheiner 2004). Attempts to classify the plastic response of plants highlight a hierarchy of responses that range from allometric growth (e.g. apparent plasticity), to local physiological adaptation (e.g. the production of sun vs. shade leaves) at the scale of the individual plant organ and up to integrated physiological differentiation (e.g. induced plant defences) at the scale of the whole plant (Weiner 2004). Thus, under such circumstances it may be best for invasion ecologists to restrict their use of phenotypic plasticity to occasions where clear developmental or morphological adaptations to new environments occur. This has not been the case historically, and phenotypic plasticity has often been invoked following superficially assessment. However, a species such as Schinus terebinthifolius, that depending on the habitat can be found growing as a tree, a vine or a scrambling subshrub (Spector & Putz 2006), poses quite different questions regarding plasticity than Alliaria petiolata that simply doubles its biomass when grown in light rather than shade (Richards et al. 2006). Situations where variation is found in more labile components of life-history may be better described as fitness homeostasis. The terms ‘general purpose genotype’ and ‘Jack-of-all-trades’ are phenomenological descriptions of observed patterns but should not be used as synonyms for phenotypic plasticity. Finally, evolutionary ecologists have tended to study phenotypic plasticity using suitable model systems with the implicit recognition that they ‘will not likely yield results that can be generalized beyond fairly narrow taxonomic or life-history categories’ (Pigliucci 2004). This may be fine when teasing apart genotype–environment interactions. However, the aims of invasion ecologists are often the opposite, attempting to find generalities on the basis of specific studies. Thus, different approaches may be required in these two fields. At the end of the day any plant trait proposed to explain biological invasion should be testable and enable predictions of the potential risk of species possessing that trait. At present, while some studies validate their glasshouse observations with information on distribution or status of the species in the wild, no study has attempted to make predictions for species not examined in the experiments. This is a crucial step in invasion risk assessment and should be a priority of research on phenotypic plasticity of invasive species.