- Top of page
- Materials and Methods
- Supporting Information
Adaptation is a process of genetic change that increases the likelihood of survival and reproduction in a given environment (Endler, 1986). An evolutionary response to novel selective pressure on traits such as phenology, biomass allocation, growth rate and life history can allow an organism to persist in newly encountered environments. Understanding the potential for and constraints on adaptation to altered environments has important implications in the areas of conservation biology, global change biology and the evolution of species’ ranges.
The study of adaptation requires that we relate genotypic and associated phenotypic variation to the environment in which variants have evolved or survived. A cline in genetically based variation in ecologically important traits, when associated with an environmental gradient, is considered to be strong evidence of adaptation to geographically varying selection (Mayr, 1956; Endler, 1977, 1986; Caicedo et al., 2004; Stinchcombe et al., 2004; Stillwell et al., 2007). Altitudinal gradients have been commonly used to study clinal adaptation (Clausen et al., 1940; Williams & Black, 1993; Oleksyn et al., 1998; Angert & Schemske, 2005; Brennan et al., 2009; Gonzalo-Turpin & Hazard, 2009). One of the main advantages of altitudinal gradients is that they offer steep environmental gradients across short spatial distances, representing exciting biological experiments in nature (Körner, 2007). Of course, it is not altitude itself to which organisms adapt. Instead, a suite of environmental gradients accompanies changes in altitude. By using altitudinal gradients it is possible, within a circumscribed region, to study a species’ response to a suite of climatic conditions that are representative of the broader latitudinal range of a species.
An explanation of how a pattern of clinal adaptation arose can often lead to a broader understanding of adaptation and its constraints. This requires both an understanding of the network of molecular, physiological and morphological mechanisms that underlie adaptive phenotypic evolution (Lepetz et al., 2009) and the functional effects of variation in these mechanisms in the environmental context in which the adaptations arise. Although studies of local adaptation abound (see Linhart & Grant, 1996; Reznick & Ghalambor, 2001), an understanding of the underlying mechanisms driving adaptation lags behind. This is partly a result of the difficulty of integrating multidisciplinary knowledge from the internal biology with an understanding of the natural environments in which the species has evolved.
For the model plant, the mouse-ear cress, Arabidopsis thaliana (L.) Heyhn. (Brassicaceae) (hereafter A. thaliana), detailed knowledge of internal biology (Meyerowitz & Somerville, 1994) and the molecular genetic basis of this biology is accumulating with extraordinary rapidity (reviewed in Sheldon et al., 2000; Blazquez et al., 2006; Holdsworth et al., 2008; Ishida et al., 2008), making this species ideal for the study of adaptive polymorphisms (Tonsor et al., 2005; Metcalf & Mitchell-Olds, 2009). Further, latitudinal clines have been noted in functional genetic variation underlying several traits, including flowering time (Stinchcombe et al., 2004; Wilczek et al., 2009), heat shock protein expression (Tonsor et al., 2008), response to vernalization (Hopkins et al., 2008), freezing tolerance (Hannah et al., 2006; Zhen & Ungerer, 2008), and size and growth rate (Li et al., 1998). However, the study of the natural variation in ecologically important traits in the field (Donohue et al., 2005a,b; Handley et al., 2005; Hannah et al., 2006; Boyd et al., 2007; Lundemo et al., 2009) and its ecology in natural populations (Arany et al., 2005; Montesinos et al., 2009; Bomblies et al., 2010) is just beginning. An understanding of the process of adaptive evolution in A. thaliana in nature requires the identification and linkage of important adaptive traits to the key environmental features that have historically acted as selective sieves (Metcalf & Mitchell-Olds, 2009).
Interest in adaptation has stimulated efforts at inferring A. thaliana’s phylogeography across its native range in western Eurasia (Sharbel et al., 2000; François et al., 2008), its rapid worldwide expansion (Provan & Campanella, 2003), and has uncovered the extent of admixture from multiple Pleistocene refugia in much of its range (Mitchell-Olds & Schmitt, 2006). To understand the nature and causes of adaptation to divergent environments, it is highly desirable to work in those areas in which recent migration and admixture are minimal.
The Iberian Peninsula served as the Mediterranean’s Pleistocene refuge and A. thaliana has been present there for more than 10 000 yr (Sharbel et al., 2000). Arabidopsis thaliana in Iberia shows strong genetic isolation by distance among regions associated with major geographic barriers that divide the Peninsula (Picóet al., 2008). Populations in north-eastern Spain belong to a distinct genetic group and are strongly differentiated from populations in the other Iberian geographic regions, interpreted as a common ancestry within the region and a history of isolation from other regions (Picóet al., 2008). Within the north-eastern region, isolation by distance is not detectable (Montesinos et al., 2009).
In the altitudinal gradient of this study, the climatic conditions are based on the confluence of Atlantic and Mediterranean influences, as well as on elevation itself (Del Barrio et al., 1990; Martínez et al., 2007). Low altitudes are expected to be associated with moderate precipitation and cool temperatures in the winter and low precipitation and high maximum temperatures in spring. By contrast, higher altitudes are associated with increased precipitation and lower minimum temperatures in the winter and a prolonged cool and moist spring.
We predict that, at low altitudes, selection will favor the maximization of winter growth, as winter rosettes of many species are capable of net carbon gain when above freezing in winter (Regehr & Bazzaz, 1976). We also expected selection at low altitudes to favor mechanisms that increase allocation to roots as a response to low spring moisture, and the avoidance of or tolerance to heat and drought during late spring through early flowering. By contrast, at high altitudes, we hypothesized that selection would favor the maximization of growth at the earlier onset of favorable conditions in the autumn, tolerance of extreme cold during winter, and prolonged growth and delayed flowering in the spring.
Here, we report our test for the gradual quantitative genetic differentiation of 17 natural A. thaliana populations collected along an altitudinal gradient from 100 to 1600 m above sea level (asl). We grew plants under controlled conditions that recreated the average temperatures and photoperiod across the altitudinal gradient during the growing period of the species. Because we were interested in adaptation in both geographic and climatic terms, we structured the test in three sequential steps: (1) we characterized the altitudinal gradient based on seasonal precipitation and temperature; (2) we tested for a relationship between genetically based trait variation and the altitude of each genotype’s origin; and (3) we described the relationship between genetically based trait variation and the climatic gradient associated with altitude.
In this approach, a significant relationship between phenotype in the common environment and the altitude/environment of origin is sufficient evidence of a nonrandom evolutionary genetic cause for the cline in phenotype (i.e. not caused by genetic drift). However, two causes are hypothetically possible, one historical accident and one adaptive.
The first posits an historic artifact in the founding of the populations and consequent isolation by distance. In this scenario, high- and low-altitude sites could have been colonized by distinct lineages that were differentiated at the time of colonization. Colonization would have to have been followed by spread to intermediate altitudes from the high- and low-altitude colonization sites, followed by continued isolation-by-distance gene flow. In this scenario, natural selection would have played no role in establishing the relationship between altitude and phenotype.
By contrast, the entire landscape may have been colonized by either a single lineage or by multiple lineages. In this scenario, there would have been no relationship between altitude and the distribution of genetically based variation in phenotype at the time of colonization. Only over time, through the action of natural selection, would genotypes have sorted out along the altitudinal gradient, or have differentiated de novo along the gradient.
In this article, we measured phenotypes under controlled environmental conditions, testing for a relationship between multivariate trait variation and the altitude of origin of the lineages. In the discussion, we consider the two possible causes for the patterns outlined above, asking what our previously published patterns in neutral genetic variation can tell us about the first potential cause, and then placing the relationship between trait values and environment in the context of previous ecological work to determine how our growth chamber measures can be interpreted in terms of previously observed patterns of adaptive variation.
- Top of page
- Materials and Methods
- Supporting Information
In north-eastern Spain, the altitude of our A. thaliana study population sites of origin is significantly associated with maximum temperatures in spring, minimum temperatures in winter and precipitation during the growing season. Biomass allocation, fecundity, phenology and vegetative growth for 111 genetic lineages from these sites exhibit gradual changes with altitude of origin under controlled environmental conditions. Using CCAs that account for interdependences among traits and climatic variables, we tested directly for a relationship of trait values with the environmental conditions in the sites of origin. In cool and wet environments with cold winters and warm late springs, plants tend to reduce rosette growth during cold and extend their development time (days to flower and to produce fruit). Although high-altitude plants are genetically differentiated to produce fewer, heavier seeds, their greater overall size results in a larger number of heavier seeds. At low altitudes, plants allocate a greater percentage of biomass to roots, and shorten the number of days to flower and produce fruits. Thus, changes in allocation and the timing of life history transitions appear to be associated with adaptation to the climatic gradient associated with altitude.
The association of genetically based clines in trait values with altitude or other environmental gradients can result from adaptive clinal differentiation (Endler, 1986). However, an alternative explanation must also be considered.
As only a single regional gradient in altitude was included in this study, it is possible that two different lineages initially colonized low altitudes (say lineage 1) and high altitudes (say lineage 2). With time, these lineages could have spread, eventually coming into contact in mid-altitudes on our gradient. With continued gene flow, it is possible that a genetic cline would have been established with more or less pure ‘lineage 1’ at low altitudes, gradually transitioning to pure ‘lineage 2’ at high altitude, with the genotypes in between exhibiting characteristics that are dependent on the percentage ancestry from each of these founding lineages. In this scenario, there would be no necessary adaptive value to the cline. If this scenario were the cause of the pattern, we would see evidence of isolation by distance among the populations, and a gradual transition in neutral genetic marker frequencies from low to high elevation. Montesinos et al. (2009) reported population genetic structure for 10 of our 17 study populations. In that study, there was no evidence for isolation by distance based on spatial autocorrelation. Montesinos et al. (2009) used AMOVA to partition genetic variation between montane vs coastal regions (7%), populations within regions (31%) and variation within populations (62%). This pattern is consistent with high levels of isolation and drift among populations (Hartl & Clark, 2007). Furthermore, 15–18% of loci were singletons within populations. As no multilocus SNP-based genotype was found in more than one population, the rate of migration between the populations considered in this study is likely to be considerably lower than the rate of mutation at the 118 SNP loci. Thus, there is little evidence in support of the notion that dispersal and mixing of predifferentiated colonists at low and high elevation has led to the observed pattern of clinal variation in genetically based multitrait phenotypes. The remaining explanation is that the cline results from adaptive differentiation along the altitudinal gradient.
We do not pretend that the conditions in our controlled environment chambers mimic the conditions in any specific population site. Neither do we assume that the phenotypes we have measured in the growth chambers are the phenotypes that would be manifested in the field. Furthermore, in order to expose every plant to the same conditions, we germinated all seeds in the simulated autumn that a winter annual would have experienced, then exposing rosettes to vernalization. Therefore, we make no claims regarding the fitness consequences in the wild that would be associated with the differentiation we observed in controlled environments. The syndromes described under controlled conditions can, however, be ecologically interpreted in the light of the environmental conditions of their sites of origin. Below, we interpret the observed association of genetic differentiation with environment from our experiments as evidence for adaptation, suggesting that the observed differences associated with the populations’ locations along this gradient could be described as ecotypic differentiation.
Our observed altitudinal and climatic cline is broadly consistent with our a priori expectations. Regarding trait variation, our a priori expectation that low altitudes would favor increased allocation to roots, higher rates of winter growth, and avoidance of or tolerance to heat and drought during late spring, whereas high altitudes would favor high growth rates in the autumn, tolerance of winter cold and prolonged growth in the spring, was broadly consistent with the results. This suggests that divergent life history strategies have differentiated along the altitudinal gradient (Table 3 and Fig. 4).
Plants from lower altitudes experience moderate Mediterranean winters, allowing vegetative growth of the rosette during winter. Our results suggest that low-altitude populations have adapted to the mild winters by increasing growth at low temperatures relative to high-altitude populations. After the winter, the low-altitude plants accelerate their reproductive cycle, with fewer days to bolt, flower and produce fruits. We suggest that this adjustment avoids the onset of heat and drought in the late spring. This rapid life cycle in a warmer drier growing season and the negative correlation between greater allocation to roots and more limited above-ground productive capacity suggest an explanation for the lower total biomass observed in low-altitude populations.
By contrast, plants from higher altitudes of origin show greater prewinter growth and less vegetative growth during the winter months. They have slower developmental timing, and bolt later. Later bolting can potentially reduce the risk of frost damage to flowers and fruits in early spring and allow more leaves to be added to the rosette before bolting. Later bolting under benign conditions can lead to higher total fecundity (e.g. Tonsor & Scheiner, 2007). High-altitude plants can prolong growth in the moister and cooler late spring, and thus show relatively greater biomass accumulation overall than low-elevation populations. These hypothetical strategies are supported by field censuses of flowering phenology in nine of our study populations (Montesinos et al., 2009). The combination of rapid growth and early flowering as a response to higher temperature and lower precipitation is a widely documented pattern in the context of climate change (i.e. Dech & Nosko, 2004; Bertin, 2008; Gordo & Sanz, 2009). In another annual species, Brassica rapa, early flowering has been shown to be selected rapidly under these conditions (Franks et al., 2007; Franks & Weis, 2008).
We observed lower lifetime accumulation of biomass in populations from low altitudes relative to high altitudes when grown under common conditions in our controlled environment experiments. This result is consistent with the suggestion that a reduction in total biomass could be an adaptive response to selection favoring reduced transpirational water loss in dry environments (Herms & Mattson, 1992). Interactions between phenology and biomass-related traits have been reported previously. Mitchell-Olds (1996) observed a trade-off between flowering time and plant size at reproduction in A. thaliana, and McKay et al. (2003) showed that two flowering time-associated genes, FRI and FLC, have epistatic effects on water use efficiency.
Surprisingly, in the field, our low- vs high-elevation study populations exhibited no statistically significant differences in plant size using fruit number as a proxy for size (Montesinos et al., 2009). This suggests that the growth environment at high elevations limits the genetic potential of high-elevation plants to accumulate more biomass, compared with the uniform environmental conditions that all populations received in our experiment.
The increase in biomass allocation to roots in low-elevation populations relative to high-elevation populations is in keeping with previous studies. In Polygonum persicaria (Heschel et al., 2004) and Boechera holboellii (Knight et al., 2006), individuals from sites with low soil moisture exhibited adaptation to drought through an increased allocation to roots.
The lower rate of rosette growth during the winter for plants from high altitude may result from a balance between tolerance to below-freezing temperatures and the ability to grow at low temperatures (Körner, 2003). Physiologically, A. thaliana genotypes that exhibit enhanced freezing tolerance show correlated down-regulation of photosynthesis, and cold tolerance is associated with the latitude of genotype origin (Hannah et al., 2006). In addition, Li et al. (1998) found that growth rates decrease with latitude in A. thaliana.
Arabidopsis thaliana individuals in this study showed the classic negative genetic correlation between seed mass and seed number (Harper, 1977; De Jong & Klinkhamer, 2005), here associated with the altitudinal gradient and, more specifically, with temperature and conditions during autumn–winter. There is an apparent conflict between the observed trade-off between seed number and seed size from the CCA and the regressions shown for seed number and seed mass in Fig. 4, in which both seed size and seed number increase with altitude. This conflict is resolved by recognizing that plant mass increases with the altitude of origin and seed number is highly correlated with plant mass. This is the same issue as illustrated in Sterns (1992) (Fig. 4). An increase in size results in increases in both seed number and seed size, disguising the underlying negative correlation. A genetic correlation between seed mass, seed number and flowering time was reported for the Landsberg erecta × Cape Verde Islands recombinant inbred line population (Alonso-Blanco et al., 1999). In the gradient we studied, seed mass increases with altitude, whereas seed number decreases. An increase in seed mass has been suggested as a strategy for survival under intense seedling competition (Salisbury, 1942; Turnbull et al., 2004). Theory also predicts that seed mass will be higher and seed number lower in regions in which unpredictable periods of stressful conditions make increased dormancy advantageous (Venable & Brown, 1988; Pake & Venable, 1996). Previous field studies of our populations revealed that high-altitude populations show higher dormancy in autumn, higher densities of seedlings in October and higher mortality during the winter (Montesinos et al., 2009). Therefore, the production of fewer but heavier seeds at high altitudes could be an adaptive mechanism to facilitate dormancy, survive intense competition in populations with high densities of seedlings germinating in the autumn, or provide larger seedlings with greater tolerance to autumn and winter stresses experienced at high altitude. These scenarios are not mutually exclusive.
As altitude increases, plants exhibit faster developmental rates in early stages and slower rates in later stages, matching our hypothesis of a benefit of accelerating prewinter growth and lengthening the period of growth before flowering in the cooler springs of higher altitudes. However, the variance explained by the timing of development in this experiment is small. The low canonical correlations of developmental timing traits are likely to have been influenced by our experimental design in which we forced all seed to germinate in the simulated autumn, constraining them to a winter annual life history. In the population sites, we see germination in pulses starting in autumn and ending in spring. This variation in germination is a major factor determining the timing of life history transitions (Boyd et al., 2007; Chiang et al., 2009).
Although we have presented strong plausibility arguments for an adaptive interpretation of the clinal patterns observed in our controlled environment studies, field studies will be necessary to resolve the adaptive effects of the field phenotypes associated with the observed clines in seed, morphological and life history traits reported here.
The use of individual lineages as replicates in multivariate analysis was justified on the basis of the high level of genetic variation within populations shown in previous studies, but there is still the possibility of some pseudoreplication in these analyses. However, the univariate regressions presented in the figures are based on population means. The significance of these regressions, in general, makes it clear that the significant pattern observed in the multivariate analyses is not simply the result of pseudoreplication, but represents a real pattern of clinal differentiation.
Arabidopsis thaliana shows clinal variation associated with altitude. Altitude is a proxy for a climatic gradient: precipitation increases with altitude; minimum winter temperature and maximum spring temperature decrease with altitude. Vegetative growth during winter and the number of seeds produced decrease in higher altitude sites where the minimum temperatures in autumn–winter are lower. Above-ground mass, number of leaves in the rosette at bolting, age at flowering and seed weight increase in sites with low minimum temperatures in autumn–winter and higher rain in autumn–winter, associated with high altitudes. Smaller plant size, early flowering and greater root allocation have been associated with adaptations to drought stress in other ecological studies, and high maximum temperatures in spring–summer and less rain during autumn–winter are associated with low elevation in the altitudinal gradient studied. Decreased vegetative growth during winter, delayed flowering time and heavier seeds have been associated with overwintering dormancy and high densities of seedlings, conditions that are found in sites with low temperatures in autumn–winter and spring–summer, associated with high altitude in this gradient. Some of these traits have been shown to have partially shared molecular genetics causes. In this study, we have identified how multiple traits correlate with an environmental gradient, suggesting that certain combinations of traits have been favored by natural selection, locally adapting plants to particular environments. The results presented here indicate that this study system could be useful for improving our understanding of adaptation to environment across a species’ range. The multidisciplinary knowledge accumulated in model species, such as A. thaliana, combined with ecological studies in natural environments, can contribute to exciting increases in the understanding of the underlying mechanisms driving adaptation in the field.