Biological Invasions

, Volume 11, Issue 5, pp 1107–1119

The role of intraspecific hybridization in the evolution of invasiveness: a case study of the ornamental pear tree Pyrus calleryana


    • Department of Biological SciencesUniversity of Cincinnati
  • Nicole A. Hardiman
    • Department of Biological SciencesUniversity of Cincinnati
Original Paper

DOI: 10.1007/s10530-008-9386-z

Cite this article as:
Culley, T.M. & Hardiman, N.A. Biol Invasions (2009) 11: 1107. doi:10.1007/s10530-008-9386-z


Hybridization between genetically distinct populations of a single species can serve as an important stimulus for the evolution of invasiveness. Such intraspecific hybridization was examined in Pyrus calleryana, a Chinese tree species commonly planted as an ornamental in residential and commercial areas throughout the United States. This self-incompatible species is now escaping cultivation and appearing in disturbed habitats, where it has the potential to form dense thickets. Using genetic techniques incorporating nine microsatellite markers, we show that abundant fruit set on cultivated trees as well as the subsequent appearance of wild individuals result from crossing between genetically distinct horticultural cultivars of the same species that originated from different areas of China. We conclude that intraspecific hybridization can be a potent but little recognized process impacting the evolution of invasiveness in certain species.


Callery PearEvolutionIntraspecific hybridizationPyrus calleryanaSelf-incompatibility


Hybridization is a strong evolutionary force that can potentially reshape the genetic composition of populations and create novel genotypes that facilitate adaptation to new environments (Stebbins 1950; Anderson and Stebbins 1954; Arnold 1997). The importance of hybridization in evolutionary processes such as speciation has long been acknowledged (Darlington 1940; Stebbins 1959, 1969), but its application to the field of invasion biology has only more recently been discussed (Abbott 1992; Ellstrand and Schierenbeck 2000; Cox 2004; Schierenbeck and Aïnouche 2006), as has the larger role of evolution itself (Lee 2002; Lavergne and Molofsky 2007; Novak 2007). Hybridization between genetically distinct taxa has been proposed as a mechanism for the evolution of invasiveness in introduced and native species (Ellstrand and Schierenbeck 2000). Most well-known examples involve interspecific or intergeneric processes (Ellstrand and Schierenbeck 2000), as in Spartina (Aïnouche et al. 2003; Cox 2004) or Senecio (Abbott 1992).

Less well studied has been intraspecific hybridization, defined as successful matings between individuals from well differentiated populations originally isolated from one another and consequently with different gene frequencies (Stebbins 1950). As such, this process does not pertain to crosses between individuals from the same gene pool that possess different alleles (Arnold 1997). If resulting F1 hybrid individuals or later generation hybrids are fertile, recombination may lead to novel genetic rearrangements which can allow hybrids to expand their ecological tolerance and invade new niche environments (Stebbins 1959; Arnold 1997). Intraspecific hybridization can also result in increased genetic variance, altered epistatic interactions, masking or unloading of deleterious alleles, and/or transfer of favorable genes (Lee 2002). Alternatively, such hybridization events can also produce outbreeding depression by disrupting co-adapted gene complexes and local adaptation in established species (Arnold 1997). Intraspecific hybridization has been carried out artificially for centuries to improve agriculturally or horticulturally important plant species (Khanduri and Sharma 2002; Johnston et al. 2003), but it has rarely been examined in natural populations, with few exceptions (Hufford and Mazer 2003; Erickson and Fenster 2006; Johansen-Morris and Latta 2006).

Within the context of invasion biology, intraspecific hybridization could potentially explain the recent spread of certain species, such as those with extended lag periods (Sakai et al. 2001), during which time hybridization and selection may act to create invasive genotypes (Ellstrand and Schierenbeck 2000). Intraspecific hybridization is most likely to follow multiple introductions of a species (Kolbe et al. 2004), especially non-native species that are transplanted from different parts of their native range into a new locality. Multiple introductions may be done deliberately, for example, when plant species are imported for horticulture (Reichard and White 2001; Burt et al. 2007) or accidentally as when seeds are introduced as contaminants during shipping (Sakai et al. 2001). Once introduced, genetically distinct individuals may cross-pollinate to create novel genotypes through admixture that otherwise would never have been possible in the native environment (Arnold 1997; Novak and Mack 2005; Roman and Darling 2007). In this way, among-population variation present in the native range is transformed into within-population variation in the introduced area (Kolbe et al. 2004). Although most novel genotypes may be inappropriate in the new environment, production of individuals combined with novel selection (e.g. release from native competitors, predators or pathogens) may strongly favor certain genotypes. Within native species, intraspecific hybridization can also occur if new genotypes are introduced from different parts of the native range. However, not all cases of intraspecific hybridization will lead to invasiveness (Ellstrand and Schierenbeck 2000; Wolfe et al. 2007), but only when the right combination of novel genetic rearrangements match with the appropriate introduced environment in which invasive traits can be selected.

The purpose of this paper is to examine the role of intraspecific hybridization in the evolution of invasiveness in plant species. To do so, we first define the process of intraspecific hybridization as it relates to invasive species, citing specific examples in the literature. Second, we focus on a case study of the Callery Pear (Pyrus calleryana; Rosaceae), an ornamental Asian tree that is increasingly invading sites throughout the United States. Finally, we conclude with suggestions for future research. Ultimately, we maintain that intraspecific hybridization is a potentially important and often overlooked stimulus for facilitating invasiveness in certain species given the right conditions.

Evidence of intraspecific hybridization

The process by which intraspecific hybridization can lead to invasiveness (Wolfe et al. 2007) requires four steps. If any of these steps do not hold, invasiveness may not evolve through this mechanism. First, founders from at least two genetically divergent populations of a species must be introduced into the same area. Evidence of such multiple introductions consist of known introductory history of the species (e.g., Luken and Thieret 1996) or genetic data tracing back the origin of introduced populations to different parts of the native range (e.g., Williams et al. 2005). Second, individuals from these differentiated populations must cross and produce fertile offspring. In some species, this may create novel recombinant genotypes (i.e. admixture) with a selective advantage in the new range. The offspring may also exhibit elevated levels of genetic variation, especially if the parental populations each experienced separate genetic bottlenecks or founder events that lowered levels of variation within their respective populations (Husband and Barrett 1991; Cox 2004; Novak and Mack 2005; but see Roman and Darling 2007). Third, recombinant hybrids must be fit, generate offspring and able to persist in new environments by possessing invasive characters that allow them to exploit resources in current or new habitats. In some F1 hybrids, increased fitness may be evident because of heterosis due to overdominance (Facon et al. 2005) and not necessarily to recombination. Finally, natural selection in the new environment must favor certain gene combinations in hybrid individuals and consequently their traits will persist and spread in populations.

There are several cases of invasive species in which some or all of these steps are present. In the tree species Schinus terebinthifolius for example, there is evidence of multiple introductions into North America from genetically different source populations along with post-introduction recombination events (Williams et al. 2005, 2007). The same is also true for the grass Phalaris arundinacea (Lavergne and Molofsky 2004, 2007), which at one time had been suggested to be dominated by European cultivars that escaped cultivation (see Lavoie and Dufresne 2005). Although multiple introductions have been indicated in Alliaria petiolata (Durka et al. 2005), Ambrosia artemisiifolia (Genton et al. 2005), Bryonia alba (Novak and Mack 1995), and Hirschfeldia incana (Lee et al. 2004), it remains unclear whether hybridization has subsequently occurred because these studies were not designed to examine recombination events.

Current behavior and past history of an invasive species can also superficially resemble intraspecific hybridization. For example, common reed (Phragmites australis) is native to North America where it is now spreading and forming dense monocultures (Orson 1999; Wilcox et al. 2003). Recent genetic evidence suggests that these invasive populations are composed primarily of a single introduced Eurasian genotype, which instead of hybridizing with native genotypes, is now outcompeting and displacing them (Saltonstall 2002; Wilcox et al. 2003; Lelong et al. 2007). A similar process may also be occurring in ornamental fountain grass (Pennisetum setaceum), an introduced Eurasian perennial in which invasive populations in Hawaii and Arizona contain the same genotype (Poulin et al. 2005). This genotype was also found in noninvasive California populations, indicating phenotypic plasticity in invasiveness within the species (Poulin et al. 2005). In another case, Silene latifolia has undergone multiple introductions into the United States from genetically structured European populations, but artificially crossing plants from different source populations did not increase hybrid reproductive output or survival in a common garden (Wolfe et al. 2007). Consequently detailed examinations of additional species are needed to fully understand the impact of intraspecific hybridization in the evolution of invasiveness.

Case study of the Callery Pear

Pyrus calleryana is an ornamental tree species from Asia that is in the early stages of spread in the United States (Vincent 2005; Culley and Hardiman 2007; Hardiman and Culley 2007). The species, commonly known as the Callery Pear, is a popular cultivated tree often planted in commercial and residential areas, where it is prized for its early spring flowers, rapid growth, and fall color. Until recently, the species was considered unable to escape from cultivation or to naturalize because of self-incompatibility, vegetative propagation, and rare fruit production (Gilman and Watson 1994). The species is currently recognized as invasive because volunteer populations have been reported with increasing frequency over the last 5 years in at least 26 states (Vincent 2005; Culley and Hardiman 2007), concurrent with recent observations of abundant fruit set in cultivated and escaped individuals. Because of its present spread, the species is now listed by the United States Fish and Wildlife as a plant invader of Mid-Atlantic natural areas (Swearingen et al. 2002) and is considered either invasive or watch-listed in ten states (Culley and Hardiman 2007).

The reproductive biology of Pyrus calleryana is conducive to its ability to invade new areas. In early spring before leaves appear, abundant flowers are produced with 6–12 flowers per inflorescence (Cuizhi and Spongberg 2003). Each flower contains 20 stamens and 2–5 fused carpels with two ovules per locule, giving a maximum seed number of 10 (Jackson 2003). Pollen is dispersed by several generalist pollinators, including honeybees (Apis mellifera L.), bumble bees (Bombus terrestris L.) and hover flies (Farkas et al. 2002). Fruits mature in autumn and are dispersed by animals such as European starlings, American Robins, and squirrels (Gilman and Watson 1994). Pyrus calleryana exhibits gametophytic self-incompatibility (Zielinski 1965) in which compatible crosses are only possible between haploid pollen and diploid pistil tissue that do not share a self-incompatibility allele. In this system, crosses can result in full compatibility, partial compatibility, or complete incompatibility, depending on the genotypes of the two individuals being crossed. In invasive populations, the self-incompatibility (SI) system acts to maximize outcrossing and hence hybridization events (Culley and Hardiman 2007) but its effectiveness depends on the number of SI alleles present within populations. In new populations, the SI system may contribute to an Allee effect and slow invasion (Taylor et al. 2004; Taylor and Hastings 2005) because low density of individuals may limit the number of compatible genotypes present and therefore their ability to reproduce with one another (i.e. decrease fitness). However, the Allee effect can be overcome with time as the number of introductions increase and gene flow is facilitated by a variety of pollen and seed dispersers, which in turn increases diversity of SI alleles.

Here we provide evidence that the invasive nature of P. calleryana has evolved via intraspecific hybridization, as outlined by some of the criteria above. Namely, introduced populations of P. calleryana consist of cultivars (i.e. cultivated varieties that have been artificially selected for horticulturally import ant traits) that represent native genotypes from different areas of the Asian range. In addition, crossing between these genetically distinct cultivars has created recombinant hybrid genotypes that comprise invasive populations. Currently we are examining the fitness of these genotypes relative to the parental cultivars to determine if they exhibit a fitness advantage in field conditions. Because several characteristics of P. calleryana described below are also present in other introduced species, the case of the Callery Pear can serve as a model for other potentially invasive species.

History of multiple introductions

Pyrus calleryana was originally introduced to breed fire blight resistance and provide compatible rootstock for Pyrus communis, the common edible pear (Culley and Hardiman 2007). The species was imported into the United States beginning in the early 1900s, primarily by the USDA plant explorer, Frank Meyer and plant breeder Frank Reimer, both of whom collected seed in various regions in China, Japan and Korea in 1918. According to Meyer’s (1918) correspondence, the species was found growing in a wide variety of habitats in China where it had a thorny phenotype and sparsely occurred in small populations. Meyer obtained P. calleryana seed from at least five different geographic locations while Reimer also collected seed in Korea and Japan (Cunningham 1984), although Reimer’s seed collections were not maintained separately. Following importation to the United States, the species was primarily maintained and tested at the USDA Introduction Station in Glenn Dale, Maryland and at Corvallis, Oregon where large numbers of seedlings were planted and monitored for fireblight testing and as a source of rootstock for economically important Pyrus species.

The species was first cultivated as an ornamental flowering tree several decades later, beginning with an attractive non-thorny tree found growing at the Glenn Dale site and first sold commercially in 1962 as the ‘Bradford’ cultivar (Whitehouse et al. 1963). As a result of its widespread popularity nationwide, several additional cultivars were subsequently introduced through the latter half of the twentieth century. Many of these cultivars are derived directly from different introductions of Asian seed collected in the earlier part of the century (Table 1). For example, ‘Bradford’ was selected from a seedlot sent to the USDA in 1919 from Nanking, China while ‘Autumn Blaze’ originated from Reimer’s 1918/1919 collection. A limited number of cultivars are also potentially of hybrid origin; for example, ‘Whitehouse’ presumably resulted from a cross between ‘Bradford’ and an unknown P. calleryana (Accession information for PI420995 at the National Clonal Germplasm Repository; To maintain the uniform characteristics of each cultivar, trees are vegetatively propagated by grafting the desired cultivar (the scion) onto P. calleryana rootstock; thus, all individuals of a given cultivar should consist of genetically identical scions, although the rootstock genotypes often vary (T.M. Culley, N.A. Hardiman, unpublished data).
Table 1

List of cultivars of Pyrus calleryana, including the year each became commercially available, the site of origin and the source and/or parentage of the cultivar, if known


Approx. year

Site of origin




Independence, KY

Chinese seed collected by Meyer; selected from P. calleryana seedlings in 1969

Autumn Blaze


Corvallis, OR

Parent originated from Chinese seed from Reimer’s 1917 or 1919 collection

Avery Park


Corvallis, OR

From a population of P. calleryana seedlings planted in Avery Park, Corvallis, OR



Glenn Dale, MD

Chinese seed purchased in Nanking, China in 1919; original tree planted at the USDA station (Santamour and McArdle 1983)

Bursnozam (Burgandy Snow™)


Perry, OH



Abt 2003

Cambridge City, IN




Washington, D.C.

‘Bradford’ × unknown P. calleryana parent

Chanticleer® (Cleveland Select, Stone Hill, Select, Glenn’s Form)


Olmsted Falls, OH & Corvallis, OR

Original tree planted in Cleveland, OH was derived from commercial seed purchased in 1946 (Santamour and McArdle 1983)

Cleprizam (Cleveland Pride®)


Perry, OH




Vincennes, IN

‘Bradford’ × unknown pollen parent

Edgewood® (Edgedell)


DuPage County, IL

P. calleryana × P. betulifolia

Fronzam (Frontier™)


Perry, OH


Gladzam (Galdiator®)


Perry, OH


Grant St. Yellow

Abt 1980



Jaczam (Jack®)


Perry, OH


Jilzam (Jill™)


Perry, OH


Mepozam (Metropolitan™)


Perry, OH


New Bradford® (Holmford)


Boring, OR




Olmsted Falls, OH




Olmsted Falls, OH




South Brunswick township, NJ

‘Bradford’ × unknown pollen parent

Trinity® (XP-005)


Portland, OR

Purchased seed

Valzam (Valiant®)


Perry, OH

‘Cleveland Select’ × unknown pollen parent


Abt 2004

Visalia, CA

‘Aristocrat’ × P. kawakammii unknown cultivar



Glenn Dale, MD

‘Bradford’ × unknown P. calleryana at USDA Station

Cultivars analyzed in the genetic study of differentiation (Hardiman and Culley 2007) are italicized

Today at least 25 different cultivars are available (Table 1) with more being developed and the species remains extremely popular among nurserymen and horticulturalists. Over 1.5 million Callery Pear trees were sold in 1998 alone totaling over $30 million dollars (Li et al. 2004), and the ‘Chanticleer’ cultivar was named the Urban Street Tree of the Year in 2005 (Phillips 2004). Consequently, the popularity of this species among the general public combined with its commercialization has led to a situation where different Asian genotypes (i.e. cultivars) have purposely been introduced multiple times across the country and are still planted today.

Genetic differentiation of source populations

Another criterion for intraspecific hybridization leading to invasiveness is that the introductions must be from genetically differentiated source populations. This seems likely in Pyrus calleryana given that most cultivars originated as seeds collected from populations in different regions of Asia that appear to be genetically divergent (N.A. Hardiman, T.M. Culley, unpublished data). It is thus possible to confirm the genetic differentiation of cultivars (i.e. native genotypes). To do so, we used nine microsatellite markers that were originally designed for closely related P. communis and Malus domestica (Yamamoto et al. 2002; Gianfranceschi et al. 1998), and which successfully amplify in P. calleryana (Hardiman and Culley 2007). To examine genetic differentiation among cultivars, we acquired samples from each of eight commercially available cultivars in Southwestern Ohio (2–22 individuals sampled per cultivar; Fig. 1) and individual samples from seven additional cultivars from the National Clonal Germplasm Repository (NCGR; Table 1). Genotypes of the multiple samples for each of the eight cultivars were also compared to test whether individuals within each cultivar were genetically identical.
Fig. 1

Principle coordinates analysis based on pair-wise genetic distance (calculated according to Smouse and Peakall 1999) showing genetic differentiation among cultivars of P. calleryana sample sizes are given in the legend

Most cultivars were genetically differentiated from one another and identifiable based on their multilocus genotypes. In a Principle Coordinates Analysis (PCoA) based on all possible pairwise genetic distances calculated according to Smouse and Peakall (1999; Fig. 1), each cultivar generally clustered away from all others, indicating that it is genetically distinct. ‘Bradford’, ‘Valzam’, ‘Whitehouse’, and ‘Capital’ each contained a unique private allele and ‘Grant St. Yellow’ contained two private alleles. However, ‘Chanticleer’, ‘Cleveland Select’ and ‘Stone Hill’ cultivars were all genetically identical, which is consistent with anecdotal accounts that they are derived from the same street tree in Cleveland, Ohio (Hardiman and Culley 2007). Within each cultivar, individuals were genetically identical, except for ‘Redspire’, ‘Autumn Blaze’ and one ‘Cleveland Select’ sample. These anomalous individuals differed by only a single allele at one to four loci, and were obtained from the same nursery indicating potential contamination or mutation within the growers stock. In addition, an AMOVA provided evidence for significant genetic differentiation among cultivars, with the majority of variation explained by genetic structuring among cultivars (ΦPT = 0.961, P < 0.001) rather than within cultivars. These data do not include rootstock genotypes, which in preliminary analyses are always genetically different than the scions with which they are paired (T.M. Culley, N.A. Hardiman, unpublished data); thus the rootstock has the potential to cross with the scions if allowed to sprout and flower. Consequently the introduced populations of cultivars of P. calleryana in the United States represent a mixture of genetically different Asian genotypes.

Hybridization and genetic recombination

Given that genetically different cultivars are frequently planted in residential and commercial areas across the United States, there is potential for these cultivars to naturally outcross-pollinate and produce fertile hybrids. This is critical because as a self-incompatible species, fruits cannot be produced in P. calleryana through selfing or cross-pollination of individuals of the same cultivar. The ability of cultivars to successfully hybridize with one another was examined two different ways.

First, we performed hand-pollinations in a common garden over 3 years comparing fruit set among reciprocal crosses of four common cultivars: ‘Bradford’, ‘Chanticleer’, ‘Aristocrat’ and ‘Redspire’. Hand-pollinations were performed on multiple individuals of each cultivar, using emasculated flowers on days 3–4 of anthesis. Each cross combination between cultivar pairs was replicated at least twice. Self-pollinations or crosses within each cultivar were also preformed and occasionally resulted in fruit formation, but no viable seeds were obtained. Reproductive success across all cultivar cross combinations was high, with four of the cross combinations resulting in 100% fruit set and an average percent fruit set of 75% (Table 2). Overall, few differences were found among cross combinations, indicating that cultivars are capable of freely crossing with one another. The single exception was the ‘Bradford’ × ‘Chanticleer’ cross with ‘Bradford’ as the maternal parent in which no fruits were formed, but this was based on a small sample size. Differences in fruit set across cross combinations may be primarily driven by the gametophytic SI system in P. calleryana, which is currently being tested at the genetic level. Fruits on average, yielded approximately 2 seeds (range: 1–4), with over 87% seed germination expressed in most crosses (N.A. Hardiman, T.M. Culley, unpublished data); seedlings are now being monitored in a common garden to quantify early establishment and photosynthetic performance. Generalist pollinators were also observed moving frequently between unmanipulated flowers of different cultivars with fruits developing soon thereafter, suggesting that hybridization events are likely under natural conditions. Overall, these results indicate that with few exceptions, most cultivars are cross-compatible and capable of producing viable hybrid offspring.
Table 2

Percent fruit set resulting from a 3 year hand-pollination study of Pyrus calleryana cultivars in a common garden

Maternal source

Paternal source







100% (3)

67% (21)

60% (5)


75% (12)


0% (4)

100% (4)


58% (24)

81% (16)


70% (10)


90% (10)

100% (8)

100% (7)


The upper diagonal represents fruit set with the given cultivar as a pollen donor and the lower diagonal represents the given cultivar as the pollen receiver. Sample sizes indicating the number of crosses are shown in parentheses next to each percentage value

The ability of cultivars to cross-fertilize was also examined a different way by focusing on the parentage of existing invasive individuals. If these individuals result from recent hybridization between nearby cultivars as proposed (e.g. Vincent 2005), the genetic contribution of each cultivar should be evident in the invasive genotypes. Using nine microsatellite loci (Hardiman and Culley 2007), invasive individuals were genotyped in three populations in Ohio (OH), Tennessee (TN), and Maryland (MD) representing different ages of invasion. The Cincinnati, OH population recently formed in the last 7 years and is hypothesized to contain mostly F1 hybrids while the older Nashville, TN population is expected to contain more advanced generation hybrids. Because the oldest population occurs in Glenn Dale, MD where the species was first introduced in the early 1900s, this population is expected to consist largely of advanced-generation hybrids with parentage reflecting both original Asian genotypes and cultivars in the neighboring area. To quantify cultivar composition at each site (i.e. putative parents), samples were also collected from cultivars planted in residential and commercial areas surrounding each invasive population.

Cultivar identification of neighborhood trees in the residential and commercial areas near each site was fist assigned with GeneClass2 (Piry et al. 2004) using Rannala and Mountain’s (1997) Bayesian method. Results indicated that neighborhood trees always consisted of a mixture of cultivars but the exact combination differed between sites (Fig. 2). For example, ‘Bradford’ was the most common cultivar in the OH (55.1%), TN (79.4%), and MD (8.3%), while ‘Aristocrat’ was the second most popular tree in OH (15.4%) but was absent from the TN and MD sites where it is more susceptible to fireblight infection. There was also a higher proportion of unknown cultivars in the MD neighborhood (61.1.9%) than in OH (2.6%) or TN (11.9%), most likely reflecting the greater diversity of Chinese genotypes historically planted around the USDA station (Culley and Hardiman 2007).
Fig. 2

Proportion of cultivars growing in residential and commercial areas surrounding invasive populations in (a) Ohio, (b) Tennessee, and (c) Maryland. Sample sizes are shown below each graph. Unknown individuals could not be matched to genotypes of 13 reference cultivars and may represent cultivars yet to be identified

The relative genetic contribution of these cultivars as well as admixture within invasive populations was next examined at the three sites using the Bayesian genotype clustering program Structure v2.0 (Falush et al. 2003). This technique determines the most likely number of genetic populations (K) given the observed data and assigns individuals to those populations based on their multilocus microsatellite genotypes. The most likely value of K is that which maximizes the log-likelihood of obtaining the observed sample of multilocus genotypes. Using the admixture model with correlated allele frequencies, we ran 20,000 steps with a burn-in of 30,000 for each K tested. The highest model log-likelihood was obtained with K = 11, which corresponded to the eight cultivars and populations at the three sites. The analysis confirmed that cultivars are genetically distinct (Fig. 3) and that the invasive populations in Ohio, Tennessee and Maryland each consist of a mixture of cultivar genotypes. Invasive individuals of a single cultivar genotype were never detected, indicating that cultivars themselves are not escaping into natural sites. As expected, there was a large number of F1 hybrids (i.e. possessing approximately 50% of two cultivar genotypes) in the youngest Ohio population and greater recombination and advanced generation hybrids (containing genotypic contributions from 3 or more cultivars) in the older populations. In Ohio, most of the F1 individuals resulted from crosses between ‘Bradford’ and ‘Aristocrat’, which were also the most common cultivars planted in the surrounding residential neighborhood (Fig. 2). In Tennessee, many hybrids exhibited ‘Bradford’ parentage, consistent with the popularity of that cultivar in the surrounding neighborhood. Maryland populations contained a greater number of unknown genotypes, which may represent additional unknown cultivars or Asian genotypes (Culley and Hardiman, 2007). Overall, these data indicate that a diverse combination of cultivar genotypes contribute to the invasive populations, consistent with intraspecific hybridization.
Fig. 3

Graphical output from structure in which each vertical bar represents an individual tree for (a) multiple individuals of known cultivars and (b) invasive populations in Ohio (N = 102), Tennessee (N = 60) and Maryland (N = 97). The color of the bar indicates the cultivar group to which an individual has been placed, and the extent of the color is the percent of the genotype attributable to the corresponding group. Individuals containing approximately half of each genotype of two cultivars are considered F1 plants. Cultivars include ‘Aristocrat’ (A), ‘Bradford’ (B), ‘Redspire’ (R), ‘Capital’ (C), ‘Chanticleer’ (Ch), and ‘Autumn Blaze’ (AB)

Conclusions and future directions

As indicated by the our ongoing study of the Callery Pear, intraspecific hybridization can be an important stimulus for the evolution of invasiveness provided that specific conditions are met. Namely, at least two introductions of genetically distinct populations must occur in the same locality with subsequent hybridization of different individuals, resulting in novel genotypic combinations with adaptive potential in the new environment. Just as only a small number of introduced species become invasive, intraspecific hybridization will not lead to invasiveness in every case (e.g. Wolfe et al. 2007). With the increasing globalization of our world today, however, introductions of new populations and species continue, thus increasing the probability of future hybridization events. Given the high economic and ecological costs associated with only a few invasive species (Pimentel et al. 2000, 2005), it is crucial that we identify evolutionary processes that facilitate invasiveness so as to prevent and control future problem species.

We are still at an early stage in understanding how frequently intraspecific hybridization leads to invasiveness but the cases identified so far (Lavergne and Molofsky 2004, 2007; Williams et al. 2005, 2007; Culley and Hardiman 2007) indicate that it can occur in several unrelated species. Even studies in which evidence does not support intraspecific hybridization (e.g. Wolfe et al. 2007) are still valuable as they provide the context for establishing the overall frequency. Additional investigations are now needed to determine the extent and effect of intraspecific hybridization in introduced and native plant taxa. These studies must take into consideration that intraspecific hybridization may proceed alongside other processes promoting invasiveness such as polyploidy (Schierenbeck and Aïnouche 2006), interspecific hybridization (Ellstrand and Schierenbeck 2000) and escape from native predators or competitors (Sakai et al. 2001). In invasive Spartina in California for example, both intra- and interspecific hybridization have been documented (Aïnouche et al. 2003; Bando 2005). Some introduced species also may possess preadaptive traits that are not a product of hybridization per se but rather act to enhance invasiveness once hybridization produces recombinant genotypes. For example, P. calleryana in its native Asian range is tolerant of diverse soil moisture conditions, which is consistent with its ability to invade wet, mesic or dry sites in the United States (Culley and Hardiman 2007). In addition, reed canary grass undergoes intraspecific hybridization (Merigliano and Lesica 1998; Lavergne and Molofsky 2007) and typically exhibits high competitive growth especially in nutrient rich habitats (Lavergne and Molofsky 2004).

The propensity for a species to undergo intraspecific hybridization will depend in part on traits and processes that promote outcrossing. Breeding systems such as dicliny, heterostyly or self-incompatibility maximize fertilization between genetically distinct individuals and thus increase the potential for hybridization. Although such systems may induce an Allee effect in an initial population, the effect may quickly disappear as outcrossing occurs between populations, especially with the contribution of multiple introductions. For example, two plant species with strong evidence of intraspecific hybridization are dioecious (Schinus terebinthifolius; Williams et al. 2005, 2007) or self-incompatible (Pyrus calleryana), traits also noted in several perennial weeds (Price and Jain 1981). Such obligatory outcrossing is in contrast with the traditional characterization of invasive species as self-compatible (Baker 1974; Price and Jain 1981; Roy 1990; but see Novak and Mack 2005). This suggests that species in which outcrossing is actively promoted may be more likely to become invasive through intraspecific hybridization than selfing species. In P. calleryana, different bee species indiscriminately visit flowers and often carry pollen between neighboring cultivars, resulting in hybrid fruit set. Finally, intraspecific hybridization will also be enhanced by animal-mediated seed dispersal which often results in seeds being carried long distances (Schiffman 1997), thus promoting movement of novel recombinant genotypes across the landscape. In P. calleryana, most seedlings bear no genetic similarity to nearby mature trees, presumably because these seedlings originated from seeds defecated indiscriminately as birds forage across an area. Overall, traits that promote outcrossing and gene flow have the potential to produce intraspecific hybridization, if combined with multiple introductions of genetically distinct individuals. Consequently, it is important to closely examine the reproductive biology of introduced species when determining their potential for spread.

Future investigations of intraspecific hybridization should especially consider the impact of horticulture and agriculture on plant invasions, which can facilitate multiple introductions of cultivated plant species. Compared to natural processes, commercialization allows these species to spread more quickly and extensively because genetically differentiated cultivars can be mass-produced and sold nationwide to gardeners and landscapers who plant them in combination within a variety of locations. In addition, the horticultural industry is largely driven by consumer demand for unique and novel plant species, which in turn facilitates introduction of non-native species. Although the vast majority of horticulturally important plant species never become invasive (Reichard and White 2001), there are cases of invading species that have a horticultural origin, including honeysuckle (Amur spp.; Luken and Thieret 1996; Schierenbeck 2004), English Ivy (Hedera spp.; Clarke et al. 2006) and Brazilian peppertree (Schinus terebinthifolius; Williams et al. 2005, 2007). In some cases, invasive genotypes may originate unintentionally after crossing occurs between different cultivars planted in the landscape, as in P. calleryana (described above) and Lythrum salicaria (Anderson and Ascher 1993). Cultivar selection prior to introduction itself can also increase invasiveness, as with selection for showy appearance and dense foliage of Japanese Ardisia crenata in the United States (Kitajima et al. 2006). Consequently, some plant breeders have begun examining individual cultivars for invasive traits (Anderson et al. 2006), such as abundant seed set and high seed germination (e.g. Lehrer et al. 2006; Wilson and Knox 2006). The next step is to determine if crossing between cultivars of particular species results in potentially invasive genotypes before they are released. In some cases, sterile cultivars of highly popular species are being developed (Li et al. 2004) and voluntary initiatives have been proposed to minimize plant invasions (Burt et al. 2007).

Finally, the importance of intraspecific hybridization in species invasions has several implications for management. Control plans of invasive species must first determine if intraspecific hybridization is present and if so, every effort must be made to prevent introduction of new genotypes into an area. Because genotypes may be morphologically indistinguishable from one another, land managers need to work closely with scientists to use genetic techniques to identify problematic genotypes for early removal. In cases where different genotypes have already been widely released into the environment, as with the popular Callery Pear, it is unrealistic that the invasive species will ever be completely eradicated and therefore control measures that prevent formation of new populations will be most effective. These should involve the following: (1) quick removal of invasive populations in natural areas after their detection; (2) replacement of cultivated parental plants with non-invasive species whenever possible, especially in locations near natural areas; (3) consideration of voluntary self-regulation or legislative measures that minimize introduction of new, compatible genotypes; (4) education of the general public on the importance of using suitable alternatives. To this end, the development of completely sterile cultivars of highly popular horticultural species may reduce the number of parental genotypes capable of spawning invasive populations while still providing a profitable alternative for the nursery industry. Ultimately, understanding how invasiveness may evolve in light of intraspecific hybridization is of paramount importance to preventing or controlling invasive species before they exert substantial ecological and economic impacts on the environment.


The authors thank D. Ayers, N. Ellstrand and K. Schierenbeck for organizing the symposium that led to this special issue, as well as enlightening discussions and comments on the manuscript. K. Manbeck provided an invaluable perspective from the green industry while M. Klooster, S. Rogstad and two anonymous reviewers provided helpful suggestions that greatly improved the manuscript. This research was supported by a grant from the US Department of Agriculture, Cooperative State Research, Education, and Extension Service, to T.M.C. (USDA CREES 06-35320-16565).

Copyright information

© Springer Science+Business Media B.V. 2008