A genetics-based Universal Community Transfer Function for predicting the impacts of climate change on future communities

Authors


Summary

  1. Although the genetics of foundation plant species is known to be important drivers of biodiversity and community structure, and climate change is known to have ecological and evolutionary consequences for plants, no studies have integrated these concepts. Here we examine how their combined effects are likely to affect the diversity of future communities.
  2. We draw on several complimentary fields (community ecology, landscape genetics and biogeography) to model how climate change will alter productivity of foundation plant species and their associated communities. We focus on three issues: (i) genetic variation of foundation species influences community diversity; (ii) gene-by-environment interactions define associated communities; and (iii) relationships between productivity and species diversity follow predictable patterns.
  3. For many foundation species, responses to climate are population specific because populations are often genetically differentiated and locally adapted. Thus, biological models that examine the effects of climate change on species distribution, forest productivity, community structure or function, should incorporate population effects. Our genetics-based Universal Community Transfer Function (UCTF) provides a method to integrate climate-based population differences into community diversity models.
  4. Several major findings emerged: (i) using the UCTF, we found that genetics-based differences between populations play an important role in defining future communities. (ii) The shape of the productivity/diversity relationship (e.g. humpbacked versus linear) dramatically affects future communities making it essential to quantify this relationship. (iii) Climate change will impact the community differently at leading, continuous and rear edges of a species' distribution, but diversity at the rear edge will suffer most.
  5. Genetics-based approaches are important to understand the ecological and evolutionary consequences of climate change on future communities and ecosystems. Such modelling can assist in identifying populations of foundation species of special value based on their sensitivity to climate change, future biodiversity and potential to support high biodiversity with assisted migration.

Introduction

Recent reviews have demonstrated that intraspecific genetic variation in plants results in predictable community and ecosystem phenotypes (Genung et al. 2011; Whitham et al. 2012). Central to these studies is the finding that different genotypes support different communities of arthropods, soil microbes, epiphytes, endophytes, trophic interactions and ecosystem processes. Common garden studies with replicated plant genotypes have shown that community phenotypes are quantifiable and heritable (Johnson & Agrawal 2005; Shuster et al. 2006; Schweitzer et al. 2008; Keith, Bailey & Whitham 2010).

Variation in community and ecosystem phenotypes among plant genotypes are driven by the combined suite of multivariate plant traits interacting with both biotic and abiotic components of the environment. For example, Barbour et al. (2009) showed genetics-based morphological and phytochemical traits differed between individual genotypes of Eucalyptus globulus and were strongly correlated with the differences in the arthropod and fungal communities supported by these genotypes. Furthermore, trees that were genetically similar in either neutral markers or quantitative traits (e.g. phytochemistry, morphology) also supported similar communities of arthropods and fungi. Thus, genetic distance between individual genotypes can determine diversity at the community level (Bangert et al. 2006).

Although genetic variation in key quantitative plant traits is likely to affect the communities associated with most plants, these community effects are likely to be most pronounced for foundation species because they are recognized as being ‘drivers’ of their respective ecosystems (Whitham et al. 2006). We also include other species of large effect, such as keystone species, dominant species and ecosystem engineers. Changes in the genetic structure of these species brought about by natural or anthropogenic causes are expected to have cascading consequences on the composition and function of their associated communities and ecosystems.

An underlying principle linking plant genetics to associated communities is the fact that genetics defines functional plant traits, which in turn can affect many other species. For example, Compson et al. (2011) demonstrated that the functional traits mediating sink–source relationships were genetically based and greatly influenced the resistance of narrowleaf cottonwood (Populus angustifolia) to the gall-forming aphid (Pemphigus betae). In turn, the presence and abundance of this aphid defined a much larger community of arthropods, fungi and birds (Keith, Bailey & Whitham 2010 and references therein). Thus, the functional traits for sink–source relationships were found to have important community consequences. Community responses to genetically based functional traits (e.g. sink–source relationships, drought tolerance, phytochemistry) of foundation species are illustrated in Fig. 1a.

Figure 1.

The influence of multiple factors mediating the composition and structure of future associated communities of foundation species. (a) A community genetics perspective in which genes define functional traits, which in turn affect associated communities. (b–c) Climate can have independent and interactive effects on all components and pathways. Dependent communities are influenced by climate, functional traits of the foundation species and productivity (mediated by underlying functional traits). (d) Pathways to incorporate in future models are the potential for: (1) plant functional traits to interact with an associated community, changing the composition of other community members, (2) communities to respond directly to climate and (3) associated communities feedback to affect the underlying genetic structure of the foundation species (i.e. community evolution). See text for examples of each category.

Because climate change has been shown to be an agent of selection on diverse plant species from biennials (Franks, Sim & Weis 2007) to long-lived trees (Allen et al. 2010), selection on functional traits will affect community phenotypes. For example, record droughts over the American Southwest have resulted in high, nonrandom tree mortality. Mature Pinus edulis trees genetically resistant to insect attack suffered 68% mortality, whereas genetically susceptible trees suffered only 21% mortality (Sthultz, Gehring & Whitham 2009a). Because resistant and susceptible trees support different communities, climate-driven changes in the frequency distribution of resistant phenotypes are shifting the composition of the associated communities they support (Sthultz et al. 2009b). Climate-based selection on different genotypes, which subsequently affects the community, is illustrated in Fig. 1b.

Alternatively, the associated community may be keying in to changes in the productivity of foundation species. For example, maladaptation of foundation plant species to rapid climate change results in highly stressed plants that are less productive and support lower species richness and abundance (Swaty et al. 2004; Stone, Gehring & Whitham 2010). Thus, even if stressed genotypes survive on the landscape, their ability to support a diverse community will be greatly diminished. Conversely, in other regions of a plant's distribution where conditions become more favourable with climate change (e.g. cold extremes become more moderate), release from stress and increased growth could result in increased diversity. Thus, differences in productivity, mediated by underlying functional traits, can alter biodiversity (Fig. 1c). These various scenarios of how climate can affect or interact with genes, traits and/or productivity illustrate why it is important to integrate both genetics and environmental-based factors that can act independently or in concert to affect biodiversity in climate change models.

The goal of our synthesis is to develop conceptual links unifying genetic, environmental and geographic perspectives into model predictions of the impact of climate change on plant-associated communities. First, we summarize research linking climate change to changes in the genetic structure of plants, community composition and ecosystem function. Secondly, we develop a conceptual framework based on these relationships to guide research and mitigation of climate change impacts on future communities. Thirdly, we present the Universal Community Transfer Function (UCTF), a genetics-based model for predicting changes in community diversity due to climate change. The UCTF is based on the findings that plant populations tend to be locally adapted and that different populations and communities will respond differently to changes in climate. Fourthly, we outline how these relationships are expected to result in specific predictions for communities within the context of leading, continuous and rear portions of species' distributions (Hampe & Petit 2005).

Foundation species in a changing climate

Although numerous studies have assessed the impact of climate change on plant species' distributions (Walther et al. 2002; Parmesan & Yohe 2003), and many have focused on foundation species (Ellison et al. 2005; Allen et al. 2010), we are unaware of any that have addressed how these changes will affect their associated communities. We use the term foundation species as ‘a single species that defines much of the structure of a community by creating locally stable conditions for other species and by modulating and stabilizing fundamental ecosystem processes’ (Ellison et al. 2005). Regionally common and locally abundant, foundation species exert a large influence on their environments in response to environmental, climatic and anthropogenic factors. Due to their key functional role, climate change impacts are predicted to be greatest in ecosystems dominated by foundation species (Gitlin et al. 2006; Anderegg, Kane & Anderegg 2012). While many conservation efforts focus on rare species (sensu, Bangert et al. 2013), we argue that it is especially important to identify and focus research efforts on foundation species. Because they define and support whole communities, understanding the links between climatic stress and foundation species' responses will facilitate predictions of changes in biodiversity and ecosystem services.

There is considerable evidence of northern hemisphere species' distributions shifting northward and up in elevation as they track shifting climate niches in response to anthropogenic climate change (Walther et al. 2002; Parmesan & Yohe 2003), resulting in expansion towards the northern edge of the distribution and range contraction at the trailing edge (Hampe & Petit 2005). Where rates of migration or adaptation are insufficient to keep pace with climate change, maladaptation is likely and extinction is possible (Aitken et al. 2008). This may be particularly important at the rear edge of a species' range (Fig. 2a), where regions tend to be characterized by warmer and drier conditions (Davis & Shaw 2001).

Figure 2.

The effect of climate change on foundation species and their associated communities will be dependent upon population location within the species' range (a, Hampe & Petit 2005). Responses to climate can be partitioned into genetic and environmental effects. Effects were derived from the Universal Response Function developed from a Pinus contorta provenance trial (b, Wang, O'Neill & Aitken 2010). Positive environmental responses in cold locations (i.e. at the leading edge of the range) depicted on the left end of the x-axis indicate that increases in site temperature will result in increased growth, whereas negative environmental responses in warm locations (i.e. at the rear edge of the range) depicted on the right end of the x-axis indicate that increases in site temperature will result in reduced growth. Genetic responses to warming follow those of environmental responses but are more moderate.

Provenance tests – common garden trials, in which populations from a range of climates are grown in one or more test climates, are emerging as ideal climate change laboratories (Mátyás 1994; Rehfeldt et al. 2003; Savolainen, Pyhäjärvi & Knürr 2007; O'Neill, Hamann & Wang 2008; Grady et al. 2011). They provide insight into the nature and magnitude of climate change impacts on foundation species and their interactions with dependent organisms. They reflect the combined in situ effects of multiple, interacting biotic and abiotic environmental factors, including a range of climate extremes, through the use of large, long-term studies in natural environments (Mátyás 1994). Transfer functions developed from provenance test data relate population productivity to the climate distance trees are transferred (i.e. test site climate minus home site climate), to reveal impacts of climate change (Howe et al. 2003; Jump & Peñuelas 2005) on biological systems.

As climate change decouples populations of foundation species from the local climates to which they are adapted, provenance test results indicate that maladaptation will become increasingly prevalent (St. Clair & Howe 2007). However, the responses of populations to climate change also vary among species and may depend on the evolutionary history of an organism (Rehfeldt et al. 2003; Savolainen, Pyhäjärvi & Knürr 2007). For example, in a common garden trial located at the thermal maximum of the distribution of three species, growth rates of Fremont cottonwood (Populus fremontii) were shown to be highly correlated to thermal variation among populations, while two species of willow exhibited intermediate (Salix gooddingii) and low (Salix exigua) variation in growth rates (Grady et al. 2011). These results indicate that the strength of selection pressures mediated by temperature varies between species. Furthermore, intraspecific responses to novel climates can also vary with evolutionary history. Two populations from similar climates of origin may respond differently due to unique trajectories of drift, migration or natural selection (Wang, O'Neill & Aitken 2010; Leites et al. 2012).

To provide a more realistic picture of how climate change will affect species' distributions, it is necessary to incorporate the genetics of the foundation species (i.e. among population variation) and biotic interactions that span multiple trophic levels (Hamann & Aitken 2013). Correlative models, such as species distribution models, have been used to identify areas of range expansion and contraction under future climate change scenarios across broad geographic scales (Pearson & Dawson 2003). An assumption of these models is that all populations share the same climate niche (Aitken et al. 2008), but many species are locally adapted in response to climate-related selection pressures (e.g. Epperson 2003; Howe et al. 2003). Different populations of foundation species are likely to exhibit diverse responses to climate, to other foundation species and to other community members within sites or across their range. These interactions can alter community composition, richness, abundance and stability (e.g. Gilman et al. 2010; Keith, Bailey & Whitham 2010; Smith et al. 2011). Here we propose the UCTF to scale climate change impacts from populations to communities. The mechanism is based on the Universal Transfer Function (UTF), which merges aspects of population genetics and ecology by predicting population productivity as a function of population and test climate; it is ‘universal’ in the sense that it can predict productivity of any population (e.g. from any climate) growing in any current or future climate (O'Neill, Hamann & Wang 2008).

The basis of a Universal Community Transfer Function

The lack of fit between the assumption that all populations share the same climate niche, and abundant experimental evidence to the contrary, indicates that new, genetics-based models are needed to more accurately predict future climate change impacts on species and communities. Here we present the conceptual underpinnings of the UCTF, illustrating how climate change can impact associated communities, through changes in both foundation species productivity and underlying functional traits that mediate productivity. We address how the genotypic, environmental and gene-by-environment interactions affect: (i) plant functional traits; (ii) productivity of foundation species; and (iii) how interactions between the two will impact the associated community.

Plant Traits Mediate Plant Performance with Climate Change

Climate change can influence plant phenotype over ecological time, by altering physiology, morphology or behaviour of individual genotypes, as well as over evolutionary time, by altering population structure through selection. Provenance trials have demonstrated phenotypic alteration (i.e. plasticity), wherein the phenotype of a genotype varies between environments (Wang, O'Neill & Aitken 2010; Smith et al. 2011; Grady et al. 2013). Population differentiation is strong and closely related to population climate origin in many broadleaf and conifer species, evidence of evolutionary alteration of phenotype through natural selection for climate. Grady et al. (2013) found population leaf economic traits (specific leaf area, stomatal conductance, net photosynthetic rate, leaf water-use efficiency) and productivity of P. fremontii were strongly related to population climate origin when populations were grown in a hot common garden at the rear edge of the species' distribution. Genotypes from warm provenances exhibited traits associated with conservative resource use (e.g. low stomatal conductance and photosynthetic rates) and the highest growth rates, suggesting that conservative traits were adaptive to high temperatures. These results also suggest two outcomes: (i) where plant stress is likely to increase, trees with conservative traits will outperform trees with acquisitive traits (i.e. traits which allow for the rapid acquisition of nutrients) and (ii) direct selection favours conservative phenotypes. Thus, strong effects of genotype, environment and gene-by-environment interactions influence plant functional traits, which can affect associated communities.

Changes in Productivity Across the Landscape

Productivity is a useful composite metric because it integrates multiple functional traits into one value for assessing ecosystem function. Provenance trials allow population-specific responses of functional traits, such as productivity, to be related to test site climate, facilitating spatially explicit predictions of the effects of climate change and assisted migration on productivity and other functional traits.

Landscape mortality events also alter population and ecosystem productivity. Extreme climate, particularly drought or high temperatures can dramatically reduce productivity by killing off large segments of a population and virtually eliminating the growth of the survivors (Allen et al. 2010). For example, drought-induced mortality of Populus tremuloides in Western Canada exceeded 35% in areas (Michaelian et al. 2011), and drought-triggered mountain pine beetle attacks have altered lodgepole pine forests in North America, impacting approximately 47 million ha in Western Canada in the last 10 years (Raffa et al. 2008).

Climate change can differentially affect landscape-level productivity depending on the location of a genotype or population along the leading, continuous or rear edge of the species' distribution. Plant stress is likely to diminish on the leading edges of a species distribution and increase on the rear edges (Fig. 2a; Hampe & Petit 2005), although extreme climate change may render even leading edge populations maladapted (O'Neill, Hamann & Wang 2008). This pattern has been demonstrated repeatedly in provenance trials of Larix and Pinus species (Rehfeldt et al. 2003; Savolainen, Pyhäjärvi & Knürr 2007; Wang, O'Neill & Aitken 2010). On the leading edge of the distribution, gene flow from central populations is more adapted to arriving climates, whereas on the rear or ‘xeric’ edge, gene flow from more central populations compromises adaptation to warming climates (Davis & Shaw 2001). This is demonstrated in Fig. 2b, which illustrates that populations of Pinus contorta from the warmest (rear) edge of the species distribution respond negatively to warming environments, while populations from the coldest (leading) edge respond positively to warming. Thus, increased productivity on the leading edge and decreased productivity on the rear edge are expected to have predictable consequences on dependent communities.

Shifts in the Community Across the Landscape

The communities of tomorrow are influenced by climate and two genotype-by-environment interactions. In the first genotype-by-environment interaction, climate can influence the phenotypic frequency of a functional trait, which in turn influences the community (Fig. 1b). For example, Sthultz, Gehring & Whitham (2009a) showed that mortality associated with a record drought changed the frequency of faster-growing moth resistant and slower-growing susceptible pinyon pine (Pinus edulis) trees on the landscape from 3:1 to a 1:1 ratio. In a follow-up study, Sthultz et al. (2009b) found that the ectomycorrhizal community differed on moth resistant and susceptible trees, demonstrating the potential for climate change to alter the genetic composition of a foundation tree, which in turn can impact the associated community, as mediated by changes in phenotypic frequencies.

Climate can also influence the expression of functional traits that contribute to productivity, altering the productivity level and thus the associated community (Fig. 1c). In this second interaction, population genetic structure of foundation species may remain unchanged, yet stress can alter productivity levels within individual trees such that they are no longer able to support their associated communities. In the pinyon pine system, multiple studies point to changes in both arthropod and mycorrhizal community structure as a result of drought-related stress in the tree. For example, by quantifying tree stress within a single site, Stone, Gehring & Whitham (2010) determined that both arthropod species richness and abundance declined dramatically along an increasing stress gradient (Fig. 3a). Additionally, ectomycorrhizal communities also shifted in response to drought, but abundance followed a unimodal rather than a linear pattern with increasing stress (Swaty et al. 2004; Fig. 3b).

Figure 3.

Two reported relationships between productivity (inverse of stress) and community diversity in Pinus edulis. (a) Arthropod species richness was negatively correlated with tree stress (Stone, Gehring & Whitham 2010). (b) Ectomycorrhizal colonization showed a humpbacked or quadratic relationship with tree stress ranking (Swaty et al. 2004).

Populations of foundation species which are genetically differentiated are likely to be functionally diverse, and will vary both in their response to climate change as well as in their ecosystem services to the associated community. As populations of foundation species shift on the landscape in response to climate change, it is likely that changes in genetic structure will redefine community structure in other ways not described by the abiotic environmental interactions detailed above. Because different genotypes of foundation species have different biotic interactions with other foundation species, such as herbivores and pathogens (e.g. Smith et al. 2011), community structure can be altered by a change in these interactions. Together these represent gene-by-biotic and environment-by-abiotic environment interactions that define new communities as a result of climate change and may not be functionally equivalent to the older communities. Thus, just as gene-by-environment interactions affect the traditional phenotypes of individuals and populations (e.g. morphology, size, chemistry), so too can they affect their community phenotypes.

The above examples illustrate how climate impacts on foundation species can alter the associated community and ecosystem phenotypes. The next section provides a mechanistic tool to predict how locally adapted populations of foundation species will respond to climate change.

A genetics-based Universal Community Transfer Function as a tool to predict diversity: scaling community dynamics to landscapes

Recognizing the importance of genetic-based differences in productivity between populations of a foundation species in response to climate change, the UTF (O'Neill, Hamann & Wang 2008) was developed to predict population-specific productivity responses to seed transfer or climate change. Our UCTF extends the UTF to predict changes in the associated communities. Fig. 4a illustrates use of the UTF (simplified for graphical purposes to include only a single climate variable) to predict productivity impacts associated with population movement or climate change of a foundation tree species. Here we use relationships between foundation species productivity and dependent community diversity to predict dependent community diversity in new climates from UTF predictions of population productivity in new climates (Fig. 4b). We also extend this tool to predict changes in species diversity at the leading and rear edges of foundation species' distributions. Given that a major impact of anthropogenic climate change can be the precipitous decline of biodiversity (Barnosky et al. 2011), research examining relationships between climate and diversity through a genetically based trait like productivity, provides an important conservation tool.

Figure 4.

Universal Transfer Function (UTF) provides a scalable relationship to predict future productivity, diversity and ecosystem functioning in a changing climate. (a) Transfer function showing observed productivity of Pinus contorta populations from a wide range of climates when grown at a single common garden (Dog Creek) (circles), and a transfer function fitted to these data (solid line) (from O'Neill, Hamann & Wang 2008). (b1) Schematic illustration of UTF for productivity developed using transfer functions where the same populations are grown at multiple test sites. A response surface is fitted to predict population productivity as a function of environment (site mean coldest month temperature (MCMT) and provenance MCMT). The solid diagonal line illustrates productivity of local populations in current climates. A shift in climate will have predictable consequences for productivity. For example, the top horizontal arrow shows that productivity of a population from −15 °C MCMT is 80 m3ha−1 in its current climate (i.e. at −15 °C provenance MCMT), but is predicted to increase to 100 m3ha−1 when MCMT is 5 °C warmer (i.e. at −10 °C site MCMT). If a population from a location slightly colder than the planting location (i.e. from −19 °C MCMT) were planted at a −15 °C MCMT site, productivity is expected to decrease from 40 to 20 m3ha−1 when MCMT is 5 °C warmer (bottom arrow). The Universal Transfer Function can be scaled to predict community diversity as a function of productivity (b2) and ecosystem functioning as a function of community diversity (b3).

The relationship between ecosystem productivity and species diversity is widely studied in ecology. The iconic ‘humpback’ or unimodal relationship was first reported by Grime (1973); later studies reported U-shaped curves in addition to both positive and negative linear associations (Waide et al. 1999; Mittelbach et al. 2001). Variation in the shape has been hypothesized to stem from scale-dependent relationships (Chase & Ryberg 2004) to the taxonomic group in question (Waide et al. 1999). In a recent meta-analysis, no clear relationship was observed between productivity and fine-scale richness within sites, regions or across the globe (Adler et al. 2011), although the analysis has been criticized for several reasons, including lack of high-productivity sites in the sample data sets (Fridley et al. 2012; Pan, Liu & Zhang 2012). Mittelbach et al. (2001) reviewed the literature regarding the productivity–diversity relationship and found that for vascular plants, a hump-shaped relationship was predominant, and new studies continue to support this finding (Michalet et al. 2002; Wang et al. 2013). For the purpose of this synthesis, we focus on unimodal and positive linear functions due to their prevalence in a wide range of ecosystems.

To illustrate the UCTF, we used productivity data from O'Neill, Hamann & Wang (2008) to model future diversity patterns using linear (y = β0 + β1x) and quadratic (unimodal) (y = β0 + β1x + β2x2) productivity–diversity functions, where x is productivity, y is the associated diversity and β's are the intercept and regression coefficients. We use species richness as the metric of biodiversity, but any diversity metric could be used. The functions were parameterized to have the same maxima, range, and y-intercepts to allow for direct comparison of the two models, and were applied to predicted productivity data (O'Neill, Hamann & Wang 2008) over four time points (1975, 2025, 2055 and 2085) to forecast species richness. We present a conceptual framework for this process in Fig. 5 for a single time period (see legend for explanation) and across all time periods in Fig. 6.

Figure 5.

Schematic illustration showing use of a Universal Community Transfer Function to estimate future community diversity within the range of Pinus contorta var. latifolia in British Columbia, Canada. Changes in productivity due to climate change (first panel) are applied to productivity–diversity relationships in middle panel to predict impacts on diversity. Red arrows in the middle panel illustrate a decline in productivity results in contrasting predictions depending upon the productivity–diversity relationship (decrease under a linear function and increase under a quadratic). Diversity maps on the right combine the first two panels to generate a ‘Universal Community Transfer Function’. Areas of high diversity are indicated in blue and low diversity in red.

Figure 6.

Community diversity predicted through time as a function of productivity, following the same approach as described in Fig. 5. Areas of high diversity are indicated in blue and low diversity in red. Note that alternative productivity–diversity relationships (linear and quadratic) generate strong differences in diversity maps.

Comparisons of the effects of the two productivity–diversity models (Figs 3 and 4) on predicted species richness yield two main findings. First, the landscape-level temporal trajectory of predicted diversity differs greatly between the two models (Figs 5 and 6). The linear productivity–diversity model predicts biodiversity will track changes in productivity, whereas in the quadratic model predicts that biodiversity depends on both changes to productivity and initial productivity. Secondly, the predicted rate of change in biodiversity is higher under the quadratic model for high- and low-productivity areas, but lower in mid-productivity regions, resulting in a more moderate decline in predicted diversity compared with the linear model. These two models predict markedly different future diversity scenarios, indicating that future diversity will be largely dependent upon the shape of the productivity–diversity relationship for individual communities. We emphasize that these models are based on real communities that differ in their response functions; for example, humpbacked for mycorrhizae and linear for arthropods, which are typical of many other communities

By placing UCTF model results in a biogeographical framework, we can generate hypotheses of how climate change may alter species diversity across the range of a foundation species distribution. We predict that (i) at the leading edge of the distribution where productivity is lower; diversity will most likely increase (due to an increase in primary productivity from climate warming) regardless of the function characterizing the productivity–diversity relationship. Furthermore, as gene flow from central populations moves northward, within-population genetic diversity will increase in northern populations that will subsequently increase the diversity of dependent species (e.g. Genung et al. 2010; Ferrier et al. 2012). (ii) Near the centre of the distribution, climate change impacts may be small (Wang, O'Neill & Aitken 2010) and variable and will depend on the shape of the productivity–diversity relationship. (iii) Populations at the rear edge are often characterized by low productivity, which is expected to decline with climate change. Both linear and quadratic productivity–diversity functions predict community diversity will decline in these populations. However, rear-edge populations have also been found to harbour the greatest amount of genetic diversity (Hewitt 2004), which could mitigate the impact of climate change on community diversity.

Although simplistic, the UCTF provides a starting point to incorporate further complexities. We note three main interactions not addressed in our conceptual model summarized in Fig. 1d. First, the associated community can interact with plant functional traits to change the composition of other community members. For example, the rhizobium Bradyrhizobium japonicum is known to affect leaf quality in soya beans (Glycine max) (Harris, Pacovsky & Paul 1985), which in turn affects herbivores (Barber & Marquis 2011). Thus, changes in one community member can indirectly affect other community members. Secondly, communities may respond directly to climate, independently of either functional traits or productivity of foundation species (Chapman et al. 2012). Thirdly, the associated community may influence the fitness of a particular foundation genotype (or population), which introduces a pathway that allows for community evolution (reviewed in Post & Palkovacs 2009). Because these additional interactions are often genetically based, they further support a genetics approach to understand how climate change will affect the communities of tomorrow.

Conclusions

Five basic findings have emerged from the literature. (i) Community structure can be strongly influenced by the genotypes of individual plants and populations. (ii) Climate change is an agent of selection, resulting in evolution in plant populations at local and regional levels. (iii) Plant stress negatively affects most community members. (iv) Plant productivity represents an intermediate mechanism or link between the plant and the associated community that is influenced by both genetic and environmental interactions. (v) Because plants are often locally adapted, it is important to incorporate genetics into climate change models to predict future plant and community distributions in the face of climate change. While support for these five factors varies from moderately to extensively documented, few studies have merged two or more factors to examine their combined, cumulative effects on associated communities. Fig. 1 represents a conceptual visualization of the expected gene-by-environment interactions that will define the distributions of foundation plant species and their dependent communities in the future. This integration represents a step forward in predicting the adverse effects of climate change on biodiversity and offers new tools to inform assisted migration and help prioritize populations whose associated future biodiversity is at risk.

The UCTF, together with provenance trials in multiple climates, provides a quantitative tool to help predict community diversity under a wide range of climate change scenarios. Towards this end, studies of communities associated with foundation species in extant and new provenance trials should be conducted to test predictions and validate models of the relationships developed herein. By incorporating genetics into these models, we can better estimate the ecological and evolutionary consequences of climate change on the communities of tomorrow. Importantly, these same facilities also allow us to identify superior genotypes and source populations of foundation species that might best cope with these changes and support the highest biodiversity.

Acknowledgements

This research was supported by NSF-IGERT Fellowships (DH Ikeda, HM Bothwell, MK Lau), NSF GK-12 Fellowship and NPS George Melendez Wright Climate Change Fellowship (HM Bothwell), Science Foundation of Arizona Fellowship Award (KC Grady), Northern Arizona University Technology Research Initiative Fund (TRIF) grant (D.H. Ikeda, K.C. Grady, T.G. Whitham) the Bureau of Reclamation Grants CESU-06FC300025 (T.G. Whitham), 04FC300039 (T.G. Whitham) and NSF FIBR grant DEB-0425908 (T.G. Whitham). The Illingworth lodgepole pine provenance trial was established, maintained and collected by the BC Ministry of Forests, Lands and Natural Resource Operations. We thank the Whitham laboratory group, in addition to CS Tysor, JR desLauriers, Drs. SM Shuster, LM Evans, R. Michalet and BJ Butterfield, the Southwest Experimental Garden Array (SEGA), the IGERT ‘Genes-to-Environment’ course of Northern Arizona University and two anonymous reviewers for their valuable feedback.

Ancillary