Abstract
Climate change models predict that sites suitable for vineyards will shift over the next 50 years. In part through rootstocks and hybrid scions bred from North American Vitis species, cultivated grapevines can thrive in a variety of environments. However, relatively little is known about the range of climatic conditions under which natural populations of native North American Vitis species occur. In this study, we use geographic information system data to describe the climatic niches of 13 North American Vitis species that are distributed east of the Rocky Mountains and to compare climatic niches in a phylogenetic context. We extracted bioclimatic data from thousands of locality points and performed univariate and multivariate statistical analyses for each Vitis species. We then used these data to determine if phylogenetic relatedness was associated with similarities or differences in climatic niches. Finally, we performed Maxent modeling and multivariate environmental similarity surfaces analysis to identify suitable climate space for each species throughout the United States. Our results indicate Vitis species occur under a wide range of climatic conditions, including warm and wet (V. shuttleworthii), warm and dry (V. mustangensis), cold and wet (V. labrusca), and cold and dry (V. riparia) climates. Some clades within Vitis exhibit similarities in their climatic niches (niche conservatism), while others include species with notably different climatic niches (niche divergence). Additionally, we identify novel geographic locations where some species could potentially flourish. These data provide valuable insights into the abiotic environments occupied by natural populations of North American grapevines and may be useful in the development of rootstock genotypes to suit specific climates now and in the future.
Climate change poses a serious threat for vineyards throughout the world, as grape crops are historically sensitive to extreme weather fluctuations (Jones et al. 2005, White et al. 2006). It has been suggested that some Vitis vinifera L. varieties will experience a shift in suitable growing regions over the next 35 years (Hannah et al. 2013) and will require adaptive management strategies to ensure long-term vineyard health (van Leeuwen et al. 2013). One important way to manage grapevine growth and survival is through the development of other Vitis species as hybrid scions and rootstocks. An important component of many Vitis hybrid scions and rootstocks are native North American Vitis species, which may confer increased abiotic and biotic stress tolerance to the scion. As such, these species have been used for centuries to facilitate the persistence and expansion of vineyards into a wide range of locations. Despite their importance for the global grape industry, relatively little is known about the environmental variables characterizing the natural geographic distributions of native North American Vitis species.
North American Vitis species are valuable sources for rootstock breeding. Today, rootstocks derived from North American Vitis species are widely used in some of the most productive grapegrowing regions of the world, including Europe and California (Arrigo and Arnold 2007). North American Vitis species used in rootstock breeding include V. cinerea var. berlandieri (Planch.) Comeaux, V. riparia Michx., and V. rupestris Scheele (Table 1). As rootstocks, these species provide resistance to phylloxera (Daktulosphaira vitifoliae Fitch) and other pathogens, confer abiotic stress tolerance to the scion via drought tolerance, facilitate survival in unique soil types, and/or affect canopy configuration and nutrient uptake (Keller et al. 2012). Interestingly, the majority of rootstock genotypes used today have been bred from relatively few individuals and are reproduced clonally (e.g., 1109P, 3309C, SO4; Galet 1979); consequently, the full range of potentially valuable rootstock material found in natural populations has not been fully explored (Pavek et al. 2003, Pap et al. 2015).
Habitat, geographic range, and life history traits for 13 Vitis species examined in this study (Moore and Wen, in press), as well as number of individuals sampled for climatic data (n).
Climatic niche modeling is an important tool for characterizing the abiotic environment in which a species occurs (Nakazato et al. 2010), and can provide valuable insights for agricultural systems (Hijmans and Spooner 2001, Miller and Knouft 2006, Beck 2013, Khoury et al. 2015). Climatic niche modeling uses geographic information system (GIS) data (temperature and precipitation variables) extracted at multiple, specific latitude/longitude points where a species is known to occur. Taken together, these data are used to construct a model of the climatic conditions that characterize a species’ geographic range (Soberón and Nakamura 2009). In this context, one potential application of climatic niche modeling is that it may be useful in identifying geographic areas where climates are currently suitable for growing. Characterizing the climatic niches of crop wild relatives can offer valuable insights into the range of climates occupied by members of primary and secondary gene pools (sensu Harlan 1976), and can inform conservation of valuable germplasm resources (Beck 2013, Khoury et al. 2015).
The goal of this study is to provide a comprehensive assessment of the climatic niches of Vitis species native to eastern and central North America, with a particular focus on two species that have been important contributors to rootstock development (V. riparia and V. rupestris). We use climate data extracted from thousands of Vitis locality points to: 1) characterize the climatic niches of 13 North American Vitis species in univariate and multivariate space, and 2) compare climatic variables in a phylogenetic context to test the hypothesis of niche conservatism (i.e., more closely related species have similar climatic niches; Soberón and Nakamura 2009). To our knowledge, this work represents the first attempt to describe the climatic attributes of North American Vitis species using GIS data and provides an important contribution to current understanding of native grapevine diversity.
Materials and Methods
Sampling
In this study, we focused on native North American Vitis species that occur east of the Rocky Mountains, as the majority of Vitis species currently used in rootstock breeding are native to this geographic area. We also included closely related congeners from the central and eastern United States to understand climatic niches in a phylogenetic context. Latitude/longitude locality data for 13 native North American Vitis species (Table 1) were obtained from TROPI-COS (www.tropicos.org) and Global Biodiversity Information Facility Data Portal (www.data.gbif.org, accessed 30 June 2015). Duplicate entries and coordinates not falling within the boundary of the United States were removed. To account for spatial autocorrelation, when any two localities of a single species were located within 1 km of each other, one of them was randomly removed. In total, 7181 unique localities were used for this study (Appendix S1). The number of points per species ranged from seven for V. shuttleworthii to 1518 for V. aestivalis, with an average of 552 localities per species.
To generate background locality data used in subsequent analyses (see below), buffer zones of ~100 km were drawn around each Vitis occurrence, following Callen and Miller (2015), and merged to form a map reflecting the distribution of Vitis (Supplemental Figure 1). This map was used to randomly sample 10,000 background points, with each point separated from neighboring ones by ≥1 km, in ArcGIS v.10.1 (ESRI 2011).
Climate data were compiled by extracting climate values at each Vitis locality and background points from 19 bioclimatic layers obtained from WorldClim at 30 arc-seconds (~1 km2) resolution: annual mean temperature (°C), mean diurnal range (°C), isothermality, temperature seasonality, maximum temperature warmest and coldest months (°C), temperature annual range (°C), mean temperature wettest and driest quarters (°C), mean temperature warmest and coldest quarters (°C), annual precipitation (mm), precipitation of wettest and driest months and quarters (mm), precipitation seasonality, and precipitation of warmest and coldest quarters (mm) (www.worldclim.org; Hijmans et al. 2005). Extracted climate data were log10-transformed to meet assumptions of normality for statistical analyses. Although soils are likely an important environmental correlate of species distributions in North American Vitis, GIS data used here come exclusively from aboveground climate variables. Based on our field observations of native Vitis populations, soil types exist in a fine-scale mosaic shifting from dry, rocky creek beds to moist, rich, forest soils and back over the course of a few meters. We were unable to find appropriate fine-resolution GIS layers of soil composition to improve niche models generated in this study (see Discussion for more details). One possible work-around is to collect and analyze soil samples for each of the species occurrence points; however, all of our locality data were obtained from online databases that did not include soil-sample information.
Univariate comparisons
One-way analyses of variance (ANOVA) assuming unequal variances (cf. Welch approximation) were conducted on each climate variable to test for a species effect. Multivariate analyses of variance (MANOVA; Wilk’s λ) were conducted between each Vitis species pair across all climate variables to test if species significantly differ in their overall climatic conditions. To account for multiple pairwise comparisons (78 tests), a Bonferroni correction was applied (α = 0.05) for assessing significance (Holm 1979). To examine where specific differences in climatic conditions exist between species, pairwise Student t-tests assuming unequal variances were performed to compare means, and F tests were used to compare variances for each climate variable between each species pair. Significance was determined after Bonferroni correction (α = 0.05; 1482 comparisons; Holm 1979). All statistical analyses were implemented in R v. 3.1.2 (R Development Core Team 2014).
Effect of sample size on univariate comparisons
A bootstrapping procedure was conducted to explore the effect of sample size on statistical comparisons of means and variances (Miller and Knouft 2006). The transformed climate data set of the species with the greatest number of localities (V. aestivalis, n = 1518) was randomly resampled until the resulting climate data set contained the same number of values as the data set of the species with the smallest number of localities (V. shuttleworthii, n = 7). Mean and variance of this resampled data set were calculated for each bioclimatic variable. This resampling process was repeated 1000 times, generating a null distribution of mean and variance values for each V. aestivalis climate variable to which the mean and variance of V. shuttleworthii climate variables were compared.
Multivariate comparisons
Given that climate emerges from the interaction of many different components, we visualized Vitis climatic niches in multivariate space. Many of the 19 bioclimatic variables used here are unique estimates of similar features, such as precipitation of wettest month and wettest quarter. To minimize redundancy in our data set, we reduced the number of climate variables by first performing a principal components analysis (PCA) and a Pearson’s correlation test, and then removing variables that were both highly correlated with other variables (Pearson’s r > 0.75) and explained less variation on the first two principal component (PC) axes. The resulting data set consisted of eight climate variables: mean diurnal range, temperature seasonality, mean temperature wettest and driest quarters, precipitation seasonality, precipitation of wettest month, and precipitation of warmest and coldest quarters.
To characterize the climatic niches of Vitis species in multidimensional space, we performed a PCA-env, which maximizes separation along PCs based on background climate, on the correlation matrix (Broennimann et al. 2012). After calibrating the PCA-env on background climate, scores for each species occurrence were calculated and projected onto a gridded PCA space of 100 × 100 cells, and occurrence and background climate densities were smoothed using a Gaussian kernel function (sensu Broennimann et al. 2012). Ninety-five percent confidence ellipses were drawn around each species cloud. Correlation circles were used to visualize the relative contribution of each climate variable to the first two PC axes. PCA-env was performed utilizing the ‘ecospat’ (Broennimann et al. 2012) and ‘ade4’ (Dray and Dufour 2007) libraries in R, and ellipses were drawn using the ‘vegan’ package (Oksanen et al. 2016).
Similarity among Vitis climatic niches was assessed by comparing niche breadth, niche position, and niche overlap (Schoener’s D; Schoener 1970) between each species pair from PCA-env (Broennimann et al. 2012). Niche breadth and niche position were calculated along the first two PC axes, and Schoener’s D, which ranges from 0 (no overlap) to 1 (complete overlap), was calculated on the smoothed density of occurrences using ‘ecospat.’
Niche hypothesis testing
To investigate if climatic niches of closely related Vitis species are more or less similar than those of more-distantly related species (i.e., niche conservatism or niche divergence, respectively), we interpreted the climatic data in a phylogenetic context following the most recent evolutionary analyses of Vitis (e.g., Zecca et al. 2012, Miller et al. 2013, Wan et al. 2013, Liu et al. 2016, J. Londo, unpublished data, 2016). We used these existing Vitis phylogenies to determine that the Vitis species included in our study fall largely within three distinct groups (Figure 1; Wan et al. 2013). The first clade includes V. rotundifolia, the only member of the subgenus Muscadinia and the earliest branching among extant North American Vitis species. The second group represents a clade that includes V. cinerea, V. palmata, V. mustangensis, and V. shuttleworthii. Relationships within this group can be further dissected into (1) a subclade that includes V. palmata and V. cinerea, and (2) a subclade that includes V. mustangensis and V. shuttleworthii. A third group includes V. aestivalis, V. labrusca, and V. vulpina. A fourth clade includes V. acerifolia, V. arizonica, V. riparia, and V. rupestris. In some phylogenetic analyses, V. monticola is also included within this clade (e.g., Miller et al. 2013, J. Londo, unpublished data, 2016); however, in Wan et al. 2013 (Figure 1), V. monticola is at the base of a large clade that includes V. acerifolia, V. arizonica, V. riparia, and V. rupestris, as well as other North American and Asian species. While V. rotundifolia diverged from other North American Vitis species sometime between the late Eocene and early Miocene, the other clades examined here appear to have last shared a common ancestor sometime during the Miocene (Wan et al. 2013; Figure 1), suggesting these clades are relatively similar in age.
Chronogram of Bayesian divergence time estimates of Vitis diversification based on 27 concatenated nuclear gene fragments inferred using BEAST software (taken from Wan et al. 2013, Figure 3). Gray bars represent 95% highest posterior density intervals of nodal age in million years. Black text represents outgroups and Vitis species not used in this study. Purple, blue, gray, and orange text and boxes represent Vitis species and their assigned clades.
Within this evolutionary context, we performed two niche hypothesis tests: niche equivalency and niche similarity. Niche equivalency, the most stringent test of niche conservatism, assesses whether or not the niches of two species are identical (Warren et al. 2008). Niche similarity examines whether or not two niches are more or less similar to each other than random, based on the availability of background conditions (Warren et al. 2008). For both tests, observed D values are compared against null distributions of simulated D values. For equivalency, the null distribution of Schoener’s D is generated by calculating D between two data sets created from randomly resampling the pooled (between two focal species) climate data n times, where n equals the number of occurrences of each species (Broennimann et al. 2012). The null distribution of Schoener’s D for similarity tests is generated by comparing overlap between the niche of one focal species and simulated niches, which are created by randomly shifting the center of the second species’ niche on the background space and weighting the niche space based on background densities (Broennimann et al. 2012). Niche tests were conducted between each species pair using ‘ecospat,’ with simulations for each test repeated 100 times.
Climatic niche modeling
Species and background locality and climate data were imported into Maxent, a machine-learning approach for predicting species distributions with presence-only data (Phillips et al. 2006), to examine current versus potential geographic distributions of each Vitis species. Maxent was run with default parameters (except maximum number of iterations was set to 5000), 70% of the data was used to train the model and 30% was used for testing model accuracy, and jackknifing was applied. Model accuracy was assessed using the continuous Boyce index (CBI), calculated utilizing ‘ecospat’ in R. Multivariate environmental similarity surfaces (MESS) analysis was implemented in Maxent to examine geographic areas of climatic similarity/novelty relative to each species distribution model prediction (Elith et al. 2010).
Results
Characterization of climatic niches
Univariate analyses
Each Vitis species examined in this study is characterized by a unique set of climatic conditions. Significant species effects were detected for each of the 19 bioclimatic variables (Welch’s ANOVA; p < 0.0001; Supplemental Table 1). All pairwise species comparisons on the combined set of climate variables significantly differed after Bonferroni correction (Wilk’s λ; p < 0.0001; Supplemental Table 2). On average, ~14 (74%; range: 2 to 18) climate variable means (Supplemental Table 3) and nine (47%; range: 2 to 14) climate variable variances (Supplemental Table 4) significantly differed among Vitis species pairs. Species pairs V. rupestris/V. vulpina, V. arizonica/V. monticola, and V. acerifolia/V. monticola demonstrated a relatively high degree of similarity in their climatic conditions, as few pairwise comparisons of means and variances were significantly different. Bootstrapping results for V. aestivalis/V. shuttleworthii (Supplemental Table 5) were congruent with results from Student t-tests and F tests for this species pair, suggesting sample-size differences did not likely affect univariate comparisons. Taken together, these results indicate Vitis species occur under a wide range of distinct climatic conditions, including warm and wet (V. shuttleworthii), warm and dry (V. mustangensis), cold and wet (V. labrusca), and cold and dry (V. riparia) climates (Supplemental Tables 6 and 7).
Multivariate analyses
Vitis species differ in their climatic niches in multivariate space. The first two PCs of PCA-env explained 67.29% of the overall variance in the data (PC1 = 41.56%, PC2 = 25.73%). Vitis climatic niches were differentiated by a precipitation/mean diurnal temperature range gradient (PC1) and by a temperature gradient (PC2; Figure 2). Observed niche overlap among species pairs ranged from 0.0003 (nearly no overlap) for V. shuttleworthii/V. monticola to 0.7705 (high overlap) for V. vulpina/V. aestivalis, with a mean overlap of 0.2192 (Table 2). Except for V. mustangensis, overlap was generally lower between V. rotundifolia, the most ancestral species in this group, and more derived Vitis species. V. rotundifolia is the only member of subgenus Muscadinia; it has a different chromosome number from all other Vitis species and is considered the basal taxon in the genus (Wan et al. 2013). Further, in this study, Vitis species in the same clade generally exhibited greater overlap with each other (i.e., are more similar) than they did with those outside the clade (Figure 2, Table 2). However, aside from V. arizonica and V. monticola, which both displayed consistently low overlap relative to all other Vitis species, species in the V. arizonica/V. acerifolia/V. monticola/V. riparia/V. rupestris group typically demonstrated lower overlap (i.e., were more different) with each other than with other Vitis species (Table 2).
Climatic niches of 13 native North American Vitis species from principal component analysis (PCA-env), grouped according to clades (see Zecca et al. 2012, Miller et al. 2013, Wan et al. 2013). Colored points signify each Vitis locality in both climate and geographic (see insets of the United States) spaces. Ellipses represent 95% confidence intervals surrounding each species’ niche space. Correlation circles illustrate the relative contribution of each climate variable to the first two principal component axes. Solid lines indicate 100% of the climatic background, with dashed lines delimiting 75, 50, and 25% of the background space.
Overlap values.
Vitis species’ niches are further distinguished from each other by size and location in climate space. V. shuttleworthii, a Florida endemic, has the narrowest climatic niche across both PC axes (Figure 2, Table 3), with populations restricted to warm, wet climates. In contrast, V. riparia and V. rupestris niches demonstrate the largest climatic widths across PC1 and PC2, respectively (Figure 2, Table 3). Generally, Vitis species vary in position along PC1 relative to their phylogenetic relationships, whereas niche position along PC2 tends to shift upward from warmer to cooler climates as species become more distantly related (Figure 2, Table 3). An exception to this pattern is V. arizonica, which is evolutionarily distant from V. rotundifolia but has the warmest niche position (Figure 2, Table 3).
Position breadth.
Niche hypothesis testing
The strictest interpretation of niche conservatism is that the climatic niches of different species are the same, thus reflecting phylogenetic relatedness. Our niche equivalency tests indicate that no two climatic niches of North American Vitis species are identical (Supplemental Table 8; p = 0.0198), which might suggest rejection of niche conservatism in North American Vitis. However, tests of niche similarity reveal that differences in the climatic niches of Vitis species can be attributed to regional differences in each species’ available background climate. Pairwise tests of similarity between most (80%) Vitis climatic niches were not significant (Supplemental Table 9), suggesting that deviations from niche conservatism between most species pairs is as expected due to background climatic differences. Of the niche-similarity tests that were significant (20%), all niche pairs were more similar than expected by chance given their background climatic conditions. In these cases, climatic niche similarity generally occurred between species that were relatively closely related (e.g., V. aestivalis/V. labrusca/V. vulpina; V. arizonica/V. monticola; V. shuttleworthii/V. rotundifolia), indicating some degree of niche conservatism among them.
Climatic niche modeling
Climatic niche models performed moderately well to very well (CBI 0.362–0.996; Table 4). Variables that contribute highly to climate models of particular species are shown in bold in Table 4. Generally, mean temperature of the driest quarter contributes highly (>24%) to all Vitis species climate models, except those of V. aestivalis, V. labrusca, V. monticola, and V. shuttleworthii, suggesting that for most species, temperature during the driest quarter is an important variable explaining their contemporary distribution. In contrast, mean temperature of the wettest quarter explains little of the species distributions (Table 4), suggesting temperature during the wet season does not greatly affect where Vitis occurs. However, overall, each Vitis climatic niche model is built upon a distinct combination of climate variables. For example, while most precipitation variables contribute similarly to the models of V. arizonica and V. rupestris, they are distinguished by differences in most temperature variables and by precipitation of the wettest month. Moreover, although some degree of overlap exists in Maxent-predicted regions of high climatic suitability between each Vitis species model (e.g., eastern United States was predicted suitable for V. vulpina, V. labrusca, and V. aestivalis), predicted areas of suitability vary between models and even between species within the same clade (Figure 3). MESS analysis indicated model predictions occur in regions of climatic similarity (i.e., positive MESS values denote climatic similarity relative to species distribution model predictions) from the eastern to midwestern United States (Supplemental Figure 2), with lower portions of Alaska (not pictured) and coastal Washington modeled to have novel climatic conditions.
Modeling summary.
Climatic niche models of 13 native North American Vitis species produced by Maxent. Climate suitability ranges from no suitability (0; green) to high suitability (1; red). Purple, blue, gray, and orange boxes group species of the same clade.
Discussion
Vitis, which includes the most economically important berry crop in the world (Cooper et al. 2012), is a near-ubiquitous and charismatic component of the North American flora, with at least 15 species and three named hybrids representing two subgenera, subgenus Muscadinia (V. rotundifolia, 2n = 40) and subgenus Vitis (all other species, 2n = 38) (Moore and Wen, in press). Although most cultivated grapevines are primarily the European species V. vinifera ssp. vinifera, many vines grown in eastern and central North America are “French-American hybrids” (i.e., crosses between V. vinifera ssp. vinifera and North American Vitis species). In addition, North American Vitis species are truly the roots of the global grape industry, as many V. vinifera varietals are grafted to rootstocks developed from native North American Vitis species. In this study, we provide the first detailed assessment of the climatic niches of native North American Vitis species from eastern and central North America, including species used for rootstocks. These data may be useful to breeders interested in expanding grapevine cultivation by exploring more abiotic stress-tolerant scions and rootstocks.
North American Vitis species occupy unique climatic niches
Univariate and multivariate analyses generated here demonstrate that North American Vitis species typically occupy distinct climatic niches. In the context of our current understanding of the Vitis phylogeny (Figure 1), early branching lineages are characterized by greater amounts of precipitation, less temperature seasonality, and warmer temperatures. For example, the climatic environment of V. rotundifolia is much hotter and wetter than that of species that appear to have diverged later in the evolutionary history of North American Vitis, such as V. riparia, which roughly tolerates the coldest (along with V. labrusca) and driest (along with V. arizonica) environments (Figure 4).
Summary of native North American Vitis species in climate (i.e., multivariate) space. Colored points illustrate the centroids of each species’ climatic niche. Purple, blue, gray, and orange ellipses encircle the extent of the niche for each clade, with species that belong to each clade delimited by the same colored text. Colored arrows link the centroids of one clade to the next that contains more-recently derived species.
Niche conservatism is a clade-specific phenomenon in North American Vitis
Theory and empirical evidence suggest that more-closely related species often demonstrate a high degree of similarity in their climatic niches compared to more distantly related species (i.e., phylogenetic niche conservatism, particularly when extant species occur in allopatry; Peterson et al. 1999, Wiens and Graham 2005). Despite this expectation, several recent studies of niche evolution have demonstrated patterns of niche divergence among closely related species (Nakazato et al. 2010) and populations (Miller and Knouft 2006). In this study, we used recently constructed phylogenies for the genus Vitis (Zecca et al. 2012, Miller et al. 2013, Wan et al. 2013, J. Londo, unpublished data, 2016) to examine climatic niche data in the context of evolutionary history. Our results suggest both niche conservatism and niche divergence are associated with the geographic distributions of these 13 North American Vitis species.
We find evidence of niche conservatism among members of some Vitis clades where closely related species (members of the same clade) tend to occupy similar climatic niches relative to other species not included in that clade. We interpret this finding to be the result of shared common ancestry and a shared common ancestral niche (Wiens and Graham 2005; Figure 4). For example, V. aestivalis/V. labrusca/V. vulpina + V. cinerea/V. mustangensis/V. palmata/V. shuttleworthii, two subclades of a well-supported group, are found in geographic regions that experience relatively high precipitation, warm temperature gradients, and a low range of diurnal temperatures. Niche similarity between the V. aestivalis/V. labrusca/V. vulpina + V. cinerea/V. mustangensis/V. palmata/V. shuttleworthii clade and V. rotundifolia, the earliest diverging North American relative, is particularly evident in Figure 4, in which the aforementioned species niches’ are mostly encompassed by the V. rotundifolia climatic niche. This pattern is congruent with these close relatives’ tendency towards similar habitats.
Theory predicts that no two species can occupy the same habitat at the same time (i.e., no two species can occupy the same niche; Gause 1934). When the climatic tolerances of two species overlap, one species may be excluded from the area of overlap due to biotic interactions (Hutchinson 1959) or other abiotic factors. Although similarity in climatic niches among some closely related Vitis species reflects similarity in their geographic ranges, their coexistence in sympatry is likely the result of the evolution of unique ecological and morphological features that distinguish closely related species at finer scales. One example of biotic factors that may drive species differences in similar environments comes from V. cinerea and V. palmata. Although the geographic distribution of V. palmata is within that of its close relative, V. cinerea (Figure 2), V. cinerea flowers and fruits earlier in the growing season (Table 1; Moore and Wen, in press). Similarly, flowering and fruiting phenology differences among close relatives V. aestivalis/V. labrusca/V. vulpina (Table 1) may reduce competition among these species that appear to occupy similar climatic niches. On the abiotic side, soil almost certainly plays a role in driving divergence among Vitis species that occupy similar climatic niches. One striking example of this is V. riparia and V. rupestris. In some portions of their geographic ranges, for example in Missouri, V. riparia and V. rupestris occur in the same area—often along the same creek. In these situations, V. rupestris occurs in rocky, dry creek beds, and V. riparia occurs in rich, moist soil along the banks of the creek. Characterizing soil associated with native Vitis populations is an important direction for future research.
While niche conservatism characterizes some aspects of all the clades of North American Vitis (see above), other clades exhibit notable signatures of divergence in climatic niche. For instance, Figure 4 illustrates a large departure in the climatic niches of close relatives V. acerifolia/V. arizonica/V. monticola/V. riparia/V. rupestris (orange) relative to that of V. rotundifolia (purple). This shift is manifested in these species’ overall preferences for drier, cooler climates with greater seasonality. Compared with V. riparia, which has an expansive northeastern distribution across the United States and, consequently, a broad climatic niche, V. arizonica and V. monticola are distributed in the southwestern United States and occupy narrower niches that are drier and warmer (Figure 2). Similarly, the closest relatives of V. riparia, V. acerifolia and V. rupestris, also have much smaller geographic distributions that overlap to a limited extent with V. riparia, as they are primarily distributed in the midwestern and central United States, respectively (Figure 2). This slight overlap in geographic range is reflected in these species’ climatic niches in multidimensional space (Figure 2), where both V. acerifolia and V. rupestris climatic niches are partly nested within climatically rarer portions of the niche of V. riparia. However, overall, the climatic niches of both V. acerifolia and V. rupestris are smaller in relation to that of V. riparia, and their centroids are shifted either towards the boundary of or completely outside the V. riparia climatic niche, respectively. Consequently, given slight overlap in geographic distributions but contrasting climatic preferences, these patterns suggest a potential role for niche conservatism (i.e., sharing of climatic conditions) in the divergence (i.e., niche contraction and niche shift to other climates) of these Vitis niches.
In addition to differences among the climatic niches and range distributions of V. acerifolia, V. riparia, and V. rupestris, these three species exhibit concomitant differences in life history and morphological traits. While all three members share similar flowering times in late spring, as well as a preference for habitats in the vicinity of stream banks, V. riparia has a very different growth habit compared with its close relatives V. acerifolia and V. rupestris. V. riparia is a high climbing vine, while V. acerifolia and V. rupestris are both shrub-like sprawling vines with short-lived or absent tendrils. Additionally, all three species have unique leaf features. V. acerifolia has densely pubescent leaves, an adaptation to drier environments (Ehleringer and Mooney 1978). V. riparia leaves are cordate in outline and shallowly lobed with coarsely biserrate margins. Such toothed margins are suggested to be advantageous in colder climates for increased transpiration and respiration (Peppe et al. 2011). V. rupestris leaves are glabrous, square-shaped, have coarsely toothed margins, and are often conduplicate (folded); this may also be advantageous for alleviating stress from intense, direct sunlight (Feild et al. 2001). Differences among these close relatives have likely facilitated their expansion into new climatic niches.
Implications of niche analysis for viticulture in a changing climate
The climatic niches of V. riparia and V. rupestris, together with V. cinerea ssp. berlandieri, are of particular importance because, to date, they have been the primary species utilized as rootstocks in viticulture. Rootstocks derived from North American Vitis species, which have historically been used to convey abiotic and biotic stress tolerance to V. vinifera scions, may contribute to combatting the predicted effects of climate change on V. vinifera (Hannah et al. 2013). For instance, rather than shifting the distribution of vineyards in some regions to track current suitable climatic conditions, vineyards may be able to exchange current rootstocks for those already adapted to warmer and drier future environments. Our multivariate and MESS analyses suggest that, of the current sources of rootstocks, V. rupestris may be more important for viticulture under the warmer and drier future climatic environment than V. riparia. Of those species that have not traditionally been used as rootstocks, our data suggest their close relative V. acerifolia, which generally tolerates warm and dry environments, could be explored as a viable source of rootstock for vineyards impacted by climate change, assuming cloning and grafting is possible for V. acerifolia, as not all Vitis species exhibit the same level of graft compatibility. Further, it will be important to incorporate information about rootstock–soil compatibility. Areas for future work include characterizing soils, both in natural areas where species used for rootstocks occur, as well as in agricultural regions used for viticulture. These data will provide critical insights into interactions among soil, rootstock, scion, and local climate, and will have important applications in selecting rootstock–scion combinations that maximize yield and grape quality in a range of different environments.
Conclusion
Native North American grapevine species are important resources for breeding hybrid scions and rootstocks to support viticulture across diverse climates. Data presented here provide a quantitative assessment of climatic characteristics of 13 native North American Vitis species. We demonstrate that each of the 13 Vitis species investigated here displays a unique climatic niche. Patterns congruent with niche conservatism define some closely related species that have similar climatic niches (V. aestivalis/V. labrusca/V. vulpina); in contrast, other clades include closely related species whose climatic niches have diverged from one another (V. acerifolia/V. arizonica/V. monticola/V. riparia/V. rupestris). These data can be used to inform rootstock-breeding programs in the grape industry aimed at adopting viticulture for changing climates. Future research directions include comparing climatic tolerances of North American Vitis species to that of the European V. vinifera in a global context, identifying rootstock species that are most appropriate for different grape growing regions, understanding plant–soil–environment interactions in nature and in vineyards, and exploring rootstock impacts on scion phenotypes in different climates.
Acknowledgments
Saint Louis University Graduate Research Assistantships to STC and LLK supported this work. The authors are grateful to Jason Londo and members of the Miller Lab for valuable feedback on previous versions of the manuscript and to the Saint Louis University Center for Sustainability for providing access to ArcGIS.
Footnotes
Supplemental data is freely available with the online version of this article at www.ajevonline.org.
- Received November 2015.
- Revision received April 2016.
- Accepted April 2016.
- Published online December 1969
- ©2016 by the American Society for Enology and Viticulture
Literature Cited
Vol 67 Issue 3
Jump to section
Related Articles
Cited By...
More from this TOC section
Similar Articles
