Heredity (2010), 1–13 & 2010 Macmillan Publishers Limited All rights reserved 0018-067X/10 $32.00 www.nature.com/hdy ORIGINAL ARTICLE Patterns of hybridization and asymmetrical gene flow in hybrid zones of the rare Eucalyptus aggregata and common E. rubida DL Field1, DJ Ayre2, RJ Whelan2 and AG Young3 1Department of Ecology and Evolutionary Biology, University of Toronto, Toronto, Ontario, Canada; 2Institute for Conservation Biology, School of Biological Sciences, University of Wollongong, Wollongong, New South Wales, Australia and 3CSIRO Plant Industry, Canberra, Australian Capital Territory, Australia The patterns of hybridization and asymmetrical gene flow of long-term migration rates also indicate asymmetric gene among species are important for understanding the pro- flow, with higher migration rates from E. aggregata to hybrids cesses that maintain distinct species. We examined the compared with E. rubida. Taken together, these results potential for asymmetrical gene flow in sympatric populations indicate a greater genetic input from E. aggregata into the of Eucalyptus aggregata and Eucalyptus rubida, both long- hybrid populations. This asymmetry probably reflects differ- lived trees of southern Australia. A total of 421 adults from ences in style lengths (E. rubida: B7 mm, E. aggregata: three hybrid zones were genotyped with six microsatellite B4 mm), which can prevent pollen tubes of smaller-flowered markers. We used genealogical assignments, admixture species from fertilizing larger-flowered species. However, analysis and analyses of spatial genetic structure and spatial analyses of fine-scale genetic structure suggest that localized distribution of individuals, to assess patterns of interspecific seed dispersal (o40 m) and greater clustering between gene flow within populations. A high number of admixed hybrid and E. aggregata individuals may also contribute to individuals were detected (13.9–40% of individuals), with directional gene flow. Our study highlights that floral traits hybrid populations consisting of F1 and F2 hybrids and back- and the spatial distributions of individuals can be useful crosses in both parental directions. Across the three sites, predictors of the directionality of interspecific gene flow in admixture proportions were skewed towards the E. aggregata plant populations. genetic cluster (x ¼ 0.56–0.65), indicating that back- Heredity advance online publication, 10 November 2010; crossing towards E. aggregata is more frequent. Estimates doi:10.1038/hdy.2010.127 Keywords: hybridization; introgression; gene flow; mating patterns; fine scale spatial genetic structure; hybrid zones Introduction towards the rare species, can further exacerbate their decline through a combination of pollen swamping and Natural interspecific hybridization occurs commonly in genetic assimilation (Rhymer and Simberloff, 1996). flowering plants (Mallet, 2005), and is thought to have an Determining which reproductive barriers influence important role in their evolutionary history (Grant, 1981). asymmetrical gene flow is therefore important for In many cases, however, hybridization is asymmetrical, understanding both the processes that facilitate specia- and one species is more often the maternal parent of tion and maintain species identities. hybrid offspring (Tiffin et al., 2001). In some cases, Given the importance of asymmetrical hybridization repeated backcrossing between hybrids and one of the and introgression, there has been a long standing interest parental species can lead to unidirectional introgression in understanding the factors which drive these patterns (Bacilieri et al., 1996). In particular, the directional bias in in plants. Some of these underlying factors include: gene flow and infiltration of genes from one species into mating system variation (SI Â SE rule; Lewis and Crowe, another can be important in determining the direction of 1958), the relative proportions of parent species (Burgess evolutionary change and species succession (Bacilieri et al., 2005) and differences in the fitness of reciprocal et al., 1996; Petit et al., 2004). Asymmetrical hybridization crosses (Tiffin et al., 2001). Flower size can also be a major is also important in a conservation context, especially structural barrier to hybridization that is often unilateral when hybrids exhibit high fitness and rare species exist (Williams and Rouse, 1988; Gore et al., 1990). This trait is in small populations. Under these conditions, frequent especially important because flower size often differs among hybridization followed by unidirectional backcrossing hybridizing species. The genus Eucalyptus (Myrtaceae), which forms a dominant vegetative component of the Correspondence: Dr DL Field, Department of Ecology and Evolutionary Australian continent, provides a good model system for Biology, University of Toronto, 25 Willcocks street, Toronto, Ontario, investigating hybridization dynamics in relation to flower Canada M5S 3B2. size. On the basis of morphological patterns, interspecific E-mail: david.fi[email protected] Received 6 February 2010; revised 17 August 2010; accepted 6 hybridization is considered widespread in Eucalyptus September 2010 (B50% of species reportedly hybridize; Griffin et al., 1988) Asymmetrical gene flow in Eucalyptus hybrid zones DL Field et al 2 and several molecular studies have detected broad-scale substantial overlap in flowering times (October–January introgression of maternally inherited markers among for E. rubida, December–February for E. aggregata) likely numerous sympatric species (McKinnon et al., 2001, provides ample opportunities for interspecific pollen 2010). In this genus, smaller flowered species are often flow. Morphological and molecular evidence has identi- unable to cross-fertilize with larger flowered species. fied substantial hybrid frequencies in the seed arrays of This is because the pollen tubes of small flowered E. aggregata where the two species occur in sympatry species are unable to grow the full length of the styles (Mean 8.9%; Field et al., 2009), and the presence of both of larger flowered species (Gore et al., 1990). This F1 and later generation hybrids. As the more common potential difference in fertilization success provides a E. rubida has larger flowers (style length B7 mm) than testable prediction regarding the expected direction of E. aggregata (style length B4 mm), and both species are hybridization and introgression in natural hybrid zones pollinated by generalist insects (Field et al., 2008), floral consisting of small and large flowered Eucalyptus species. morphology may be an important driver of asymmetrical The availability of highly polymorphic genetic mar- gene flow. Despite the widespread occurrence of hybri- kers (for example, microsatellites, amplified fragment dization in Eucalyptus, little is known about the fine-scale length polymorphisms) and powerful statistical analyses patterns of hybridization and the factors driving asym- (for example, Bayesian clustering) has greatly improved metrical gene flow in natural hybrid zones. the detection of first generation and later generation To assess the extent and direction of interspecific gene hybrids. Apart from some notable exceptions (for flow we use molecular markers and Bayesian analyses to example, Chung et al., 2005), few studies have utilized examine the genomic composition of three hybrid zones these statistical procedures to examine hybrid popula- consisting of E. rubida, E. aggregata and putative hybrids. tions in the spatial context of their parental species. In A prevalence of F1 hybrids would indicate little particular, for plants with limited seed dispersal, this opportunity for gene exchange, whereas a high fre- information can be important for inferring the direction quency of backcross hybrids would suggest a high of gene flow because early generation hybrids are potential for introgression. Moreover, a skewed fre- expected to be located near their maternal parent. quency of hybrid backcrosses in one parental direction Moreover, greater spatial clustering of hybrids around would indicate the directionality of introgression. We one of the parental species may be important for also used coalescent-based methods to estimate long- reinforcing directional backcrossing and can contribute term historical levels of gene flow between parental and to subsequent asymmetrical introgression. When asses- hybrid populations. Given the differences in flower size, sing patterns of hybridization and introgression, the we predict that hybrid populations will exhibit direc- spatial analysis of reproductive adult genotypes has tional gene flow towards E. aggregata. From these several advantages over direct methods including predictions we asked the following questions: (i) what reciprocal pollinations and assessments of introgression is the frequency of introgression between the two in open pollinated progeny arrays. First, for tall trees species? and (ii) is asymmetrical gene flow in the with a long generation time and time to first flowering direction of the smaller-flowered E. aggregata? In addi- such as Eucalyptus, crossing experiments are often tion, we used spatial analyses and spatial autocorrelation impractical. Second, both direct techniques do not methods to assess both the degree of spatial clustering provide information regarding the establishment of between parental species and hybrids and the scale of backcrossed hybrids, a necessary pre-requisite for seed dispersal. Given that seed dispersal is locally potential introgression between species. restricted in Eucalyptus