Biologic invasions are increasingly recognized as one of the major threats to biodiversity worldwide (Mooney and Cleland 2001; Clavero and García-Berthou 2005). This is particularly true on islands where recent invasions have led to the extinction of many endemic species (Blackburn et al. 2004; Sax and Gaines 2008). Invasive species can have dramatic effects through competition with or predation of native organisms and disturbance of ecosystem functioning (Wilcove et al. 1998; Davis et al. 2005; Richardson and Pysek 2006). Invasive populations are often thought to originate from a very limited number of individuals and therefore challenge the idea that populations going through bottlenecks should suffer from inbreeding and reduced fitness (Facon et al. 2011). To better understand the process of an invasion of a new territory, determining the origins of invasive species as well as the past and incipient evolutionary processes is essential. Several recent studies suggest that exotic species success is dependent on variable trait combinations, which makes it difficult to identify general determinants of invasiveness (Facon et al. 2006; Van Kleunen et al. 2010; Gurevitch et al. 2011). Specific studies are thus required to understand the recent evolutionary history of invasive species.
The olive tree (Olea europaea L., Oleaceae, hereafter ‘the olive’) is often associated with Mediterranean regions, but it is also known to be invasive and disruptive to the local flora, particularly in South Australia, New South Wales (NSW), Norfolk Island, northern New Zealand (e.g. Kermadec, Auckland Bay), the Hawaii archipelago and Saint Helena (Spennemann and Allen 2000; Cuneo and Leishman 2006; GISD 2010). Two olive subspecies have been spread by humans (Cuneo and Leishman 2006; Besnard et al. 2007a): Olea europaea subsp. europaea (the Mediterranean olive) and O. e. subsp. cuspidata (Wall. ex G. Don) Cif. (the African olive). The subspecies show a distinct native regional distribution (Green 2002; Besnard et al. 2012) – O. e. europaea is a characteristic taxon of the Mediterranean Basin, while O. e. cuspidata is distributed from southern and eastern Africa to southern Asia (Pakistan, India, Iran and China). The two subspecies are generally easy to distinguish based on morphological traits (Médail et al. 2001; Green 2002; Cuneo and Leishman 2006), and their long geographical isolation also led to a high, significant genetic divergence (Rubio de Casas et al. 2006; Besnard et al. 2007b). The Mediterranean olive was one of the first woody crops and was spread by human cultivation during the last six millennia (Kaniewski et al. 2012). Unlike the europaea subspecies, O. e. cuspidata's fruit has no commercial value, but the African olive has been exploited for its hard and durable wood and can be used as a rootstock, ornamental or hedging plant (Spennemann and Allen 2000; Starr et al. 2003).
The history of both invasive olive subspecies is only partially documented (Dellow et al. 1987; Cuneo and Leishman 2006). The agricultural development of Australia gained momentum in the early 1800s and coincided with the introduction of many plants from Africa and the Mediterranean that were climatically suited to Australia. The Mediterranean olive tree was one of the earliest plant introductions into Australia by agricultural pioneer John Macarthur in 1805. Since then, multiple clones have been introduced, and more than 100 olive varieties are presently reported (Sweeney and Davies 1998). During the mid-1800s, the Macarthur family operated a large nursery at the famous Camden Park estate in south-west Sydney, NSW, and shipped potted plants throughout the colony. Plant listings in the 1843 Camden Park Nursery catalogue include a number of introduced plants that have since become environmental weeds, including African olive, which was established at this time. Isolated trees of African olive were also reported in the Adelaide region, South Australia (Shepherds Hill; Cuneo and Leishman 2006). In contrast to continental Australia, the origins of invasive olive in the oceanic islands and archipelagos are not clearly documented (GISD 2010). The infestation on Norfolk Island by the African olive is probably relatively old (during the 19th century; Cuneo and Leishman 2006), while the first records on Maui (Hawaii) and Saint Helena date back to the 1960s and early 2000s, respectively (Starr et al. 2003; GISD 2010).
While human activities contribute to transcontinental dispersal of O. europaea, birds are responsible for its local spread (Spennemann and Allen 2000). The dispersal range and the amount of seeds dispersed depend on the animal species but probably also on the size of the fruits (Alcantara and Rey 2003). After dispersal and establishment, the olives outcompete the native vegetation (such as eucalypts) by preventing regeneration. Olea europaea forms a crown under which olive seedlings can grow, but most native flora cannot (Cuneo and Leishman 2006; Cuneo et al. 2010; Major 2010). For example, the formation of African olive canopy in the Cumberland plain woodland resulted in a 78% reduction in native understory plant richness (Major 2010). In addition, the establishment of African olive can affect the local fauna by changing the vegetation structure and fruit availability. The speckled warbler has been shown to be negatively affected by the African olive invasion, while nonindigenous bird species such as the common starling and Eurasian blackbird are attracted by the presence of the African olives. This further encourages the displacement of the native fauna (DECC 2007). Not only can olive trees thrive in dry woodlands, they are also highly invasive in coastal regions. Hence, olives are considered as a serious threat to the biodiversity of Australia (Manders and Richardson 1992; Tozer 2003; Cuneo and Leishman 2006; GISD 2010).
The use of genetic data can be useful in reconstructing the history and hence identifying the source of invasions and documenting the population dynamics of invaders (Estoup et al. 2004; Bonhomme et al. 2008; Wilson et al. 2009; Ascunce et al. 2011; Lander et al. 2011; Ndlovu et al. 2013). Such information not only increases our understanding of the ecological constraints of the native habitat of the invader (by comparing the invasive and native habitats), but it can also help unravel evolutionary changes that have occurred since it was introduced (Prentis et al. 2008; Dlugosch and Parker 2008; Rey et al. 2012). Previous genetic characterizations of invasive Olea, using both plastid DNA and nuclear markers, have located the potential geographical origins of these invasive populations (Besnard et al. 2007a). Populations near Adelaide (subsp. europaea) showed high genetic similarities with central and western Mediterranean cultivars, while Hawaii and NSW populations (subsp. cuspidata) showed a genetic affinity with southern African populations. An event of early admixture between europaea and cuspidata subspecies was reported (Besnard et al. 2007a), indicating that hybridization could have played a role in the invasion of the two olive taxa. Yet, this hypothesis was strongly criticized by other authors (Breton et al. 2008), who argued that these two subspecies are not in contact in the invasive range. Furthermore, Besnard et al. (2007a) found that the NSW cuspidata population displayed reduced genetic diversity compared with a native population from South Africa, suggesting a strong bottleneck during the introduction in Australia. The genetic diversity found in the Hawaiian population was even lower than in NSW. The hypothesis of sequential introductions was stated but still needs to be tested. Under this scenario, the first introductions may have occurred from southern Africa to NSW, and then, NSW may have been a source of invaders for other regions such as the Hawaii archipelago (Spennemann and Allen 2000; Starr et al. 2003; Besnard et al. 2007a).
Recent advances in population genetics have generated methods to reconstruct the past demographic history of species. Coalescent theory and Bayesian analysis have provided a major framework that led to the development of several inferential methods to study changes in population size (Hudson 1990; Beaumont 1999). In many cases, the methods were computationally very demanding even for simple models (full-likelihood methods). More recently, an alternative framework, called approximate Bayesian computation (ABC; Beaumont et al. 2002), has emerged. This flexible framework has been particularly successful for the estimation of population parameters under complex demographic histories, especially to investigate the recent colonization history of invasive species (Pascual et al. 2007; Beaumont 2010; Csilléry et al. 2010; Estoup and Guillemaud 2010; Estoup et al. 2010; Lombaert et al. 2011; Sousa et al. 2012).
The purpose of our study was to examine the origins and spread of invasive olives in Australia and Hawaii using both plastid and nuclear markers. A large sample of invasive and native accessions was characterized and provided strong evidence for the Mediterranean (europaea) and African (cuspidata) origins of Australian invasive olives as well as for putative admixture between the subspecies. Then, we used an ABC framework to identify the most probable among different colonization scenarios and to infer several key parameters of the foundation history of olives in East Australia and Hawaii (e.g. duration of bottlenecks, effective number of founders). Because of the multiple introductions of Mediterranean cultivars (clones) from various geographic origins to Australia, probably in numerous sites over the last 200 years, it appears difficult at this stage to model the complex origin of invasive European olive. The presented ABC analyses were thus only applied to reconstruct the demographic history of the invasive African olive in Australia and Hawaii, which seems to be much simpler, based on the results from the previous and present studies.