We aimed to select species that had shown potential for rapid evolution through postintroduction morphological change. We chose species with restricted native and introduced ranges so that we could comprehensively sample across their distributions. Annual or short-lived perennial species with sexual reproduction were selected, because these species have had more generations since introduction, increasing the opportunity for evolution to occur in the introduced range. We avoided selecting crop and pasture species that were likely to have been introduced many times.
Arctotheca populifolia (Fig. 1), chosen based on the findings of rapid morphological change in introduced populations (Buswell et al. 2011), is in the Asteraceae and is a perennial, herbaceous succulent native to South Africa and introduced to Australia (Harden 1992). The first records of this species in Australia occurred in the 1930s on both the east and west coasts (AVH Database, 2012). The Australian distribution encompasses coastal environments from Geraldton (Western Australia) to northern New South Wales (Heyligers 1998). The Global Compendium of Weeds lists A. populifolia as an agricultural and environmental weed (GCW database 2012).
The second study species was Petrorhagia nanteuilii, which is an annual, herbaceous plant in the Caryophyllaceae. It is native to western Europe and western North Africa (Ball and Heywood 1964) and introduced to Asia, Australia, North America, South America, and Macaronesia (GRIN database 2012). This species was first recorded in Australia in 1882, and, currently, the Australian distribution is restricted to the southeast, ranging from Brisbane to Adelaide (AVH Database, 2012). Petrorhagia nanteuilii is also listed in the Global Compendium of Weeds as an agricultural and environmental weed (GCW database 2012).
Similar to A. populifolia, P. nanteuilii also showed evidence of morphological change over time since introduction (see 'Results', below). This was determined using herbarium specimens following methods described by Buswell et al. (2011). We measured height on all available specimens at the National Herbarium of Victoria (MEL) at the Royal Botanic Gardens, Melbourne. This gave data for 184 plants from 56 herbarium sheets ranging in collection date from 1882 to 1998. No leaf traits were measured because leaves do not press well in this species. All plants had grown in the range of the introduction, in Victoria and New South Wales in Australia. We ran a general linear model including region and year as predictors and log10-transformed height as a dependent variable. The term for region was included to prevent the possibility that a population expansion along an environmental gradient would be mistaken for adaption to the native range across time (Buswell et al. 2011). To do this, we recorded the region of origin for each sample. Because most regions were represented by relatively few specimens, we pooled bioregions to construct four broad climate regions: (1) humid coast and hinterlands (including East Gippsland, Victoria, and the New South Wales Central Coast and South Coast), (2) humid highlands (including Eastern Highlands, the Snowfields, and the Southern Tablelands), (3) subhumid slopes (including the Victorian Midlands and Riverina, and the New South Wales South West Plains, South West Slopes, and North West Slopes), and (4) semi-Mediterranean (including the Victorian Volcanic Plain, the Grampians and Wannon). In order to acknowledge the nonindependence of plants from the same herbarium sheet, we weighted individuals according to the number measured on the herbarium sheet such that the weights for all the plants on each sheet sum to one. For example, a single plant on a sheet received a weight of one, while two individuals on the same sheet each received a weight of 0.5. Analyses were performed in JMP, version 5 (SAS Institute, Cary, NC).
For both species, we also measured plant height over time in the native range, using the methods described above. This was done in order to determine whether any changes identified in the introduced range were concurrently occurring in the native range, perhaps as a result of global climate change. For these data, region was not included as a term due to the small number of samples available for each region. In total, 52 herbarium samples from 28 sheets were measured from the native range of A. populifolia and 86 samples from 26 sheets for native range P. nanteuilii.
We sampled leaves from 348 A. populifolia plants from 10 sites covering the native range (N = 188; Fig. 2A) and seven sites across the introduced range in Australia (N = 160; Fig. 2B, triangles). For P. nanteuilii, we sampled a total of 345 plants, including those from 12 sites in the native range (N = 282; Fig. 2C) and two sites across the introduced range in Australia (Fig. 2B, squares). Two attempts were made to sample this species in the vicinity of Adelaide, South Australia, at the westernmost reported extreme of the Australian distribution, and in the vicinity of Sydney where P. nanteuilii has also been reported; however, on all occasions, none were present. Leaves were placed in vials containing a solution of 40% sodium chloride, 4% sodium ascorbate, 4% silica, and 3% cetyltrimethylammonium bromide (Thompson 2002) and stored at 4°C. To prepare samples for extraction, leaves were removed from the preservative, washed in Milli-Q water, patted dry, and frozen at −70°C prior to freeze drying. Freeze-dried samples were crushed and DNA was extracted using a NucleoSpin 96 Extraction II Kit (Macherey-Nagel, Düren, Germany).
Figure 2. Sampled areas with place name abbreviations and number of individuals sampled in parentheses. (A) native range samples of Arctotheca populifolia. (B) Australian introduced range samples of A. populifolia (triangles) and Petrorhagia nanteuilii (squares). (C) native range samples of P. nanteuilii. Genetic groups are indicated by bars labeled with group name (i.e., A1; see 'Results'). Note that group assignment of P. nanteuilii samples PM and GI is ambiguous (see Figs 3 and 4).
Download figure to PowerPoint
Microsatellites were developed using next-generation sequencing on the GS-FLX 454 platform (Roche, Manheim, Germany) following methods described by Abdelkrim et al. (2009). QDD v 0.9.0.0 Beta (Meglécz et al. 2010) was used to identify microsatellites, and primers were designed using the program Primer 3 (Rozen and Skaletsky 2000). A panel of polymorphic markers was chosen for each species (A. populifolia, seven microsatellite loci; P. nanteuilii, 12 microsatellite loci; Table S1). Using universal primers (Neilan et al. 1997) having four differently colored fluorescent labels, we multiplexed PCRs within label color and multiloaded all loci for each species into a single reaction per individual. The step-down PCR protocol consisted of ten cycles each at the following annealing temperatures: 70°C, 64°C, 58°C, 54°C, 50°C. Samples were genotyped using an ABI 3730 (Applied Biosystems, Foster City, CA) using GS-500 (Liz) in each capillary as a size standard. Allele sizes were estimated on GeneMapper, version 3.7 (Applied Biosystems).
Statistical analyses of genetic data
We tested microsatellite data for departures from Hardy–Weinberg and linkage equilibrium in Arlequin, version 22.214.171.124 (Excoffier et al. 2005), and P-values were Bonferroni corrected. We used Structure, version 2.2 (Pritchard et al. 2000; Falush et al. 2003), to determine whether multiple genetic groups were present across the range of each species and to determine the native source of introduced populations. For this analysis, we used the admixture model with correlated allele frequencies and tested the number of genetic groups (K) for each value of K between one and ten. We ran ten replicates for each value of K, each run having a burn-in period of 100,000 Markov chain Monte Carlo steps followed by 106 iterations. The most likely number of genetic groups was inferred using Evanno et al.'s (2005) ΔK method. We determined group membership assignment of each sample using the highest proportion of membership across all ten runs of Structure. Principal coordinate analysis (PCoA) conducted in GenAlEx v. 6.3 (Peakall and Smouse 2006) was used to visualize genetic distances (Nei 1972) between populations.
Many authors have stressed that a spectrum of diversity measures gives the best summary of diversity (Pielou 1966; Hill 1973). Therefore, we used measures closely related to each of Hill's first three diversity orders: zero (number of alleles, NA; allelic richness, R), unity (Shannon's Index, SH), and two (Hardy–Weinberg expected heterozygosity, HE). To calculate NA, R, and HE for each sample, we used FSTAT, version 126.96.36.199 (Goudet 1995, 2002), and SH was calculated using GenAlEx. For greatest utility in future comparisons, we also convert diversity orders 1 and 2 into their effective number equivalents, which avoid many well-known problems of diversity measures (Jost et al. 2010; Leinster and Cobbold 2011). The respective effective numbers equivalents are 1Dwithin = 2^SH, and 2D = 1/(1−HE).
We used nonparametric Mann–Whitney U-tests to compare within-population diversity levels between the samples identified as sources for Australian introductions of both species because these data could not be made normal by transformation. Three approaches were used to assess diversity between populations. Pairwise FST values were calculated in Arlequin for comparison with other studies that quote this measure. Pairwise values for Shannon's mutual information index (SHUA) were calculated in GenAlEx. Compared with FST, mutual information is known to be more robust to a wide range of population sizes and dispersal rates (Sherwin et al. 2006; Dewar et al. 2011); additionally, the mutual information index can be converted to a numbers equivalent (1Dbetween), which avoids some serious problems that occur with other between-population measures (Jost et al. 2010).
We surveyed the literature regarding HE measured from polymorphic microsatellite data in species from both families containing our study taxa, Asteraceae and Caryophyllaceae, to determine whether HE estimates generated from native populations in this study were congruent with those from other members of the same family. This search was conducted in Google Scholar using the family name as a search term in conjunction with the terms “microsatellite” and “heterozygosity” in August 2012. Where data were given for multiple populations within a study, a mean value of HE was used. We avoided including estimates generated from introduced ranges, those of populations suspected of hybridization, and those of cultivated populations. Then, we surveyed the literature for examples of species showing evolutionary change in their introduced range, where genetic diversity had been estimated in both the native and introduced ranges. This search was conducted in Google Scholar in October 2012 using the terms “introduced” and “heterozygosity” in conjunction with either “rapid evolution” or “contemporary evolution.” Additionally, we included studies referenced in a review of genetic variation across native and introduced ranges (Dlugosch and Parker 2008a) showing evidence of morphological change in the introduced environment. We calculated the ratio of diversity found in the introduced range to that found in the native range (RHE), which gives an estimate of diversity retained after introduction, assuming no changes in diversity have occurred in the introduced range. For this calculation, we only used estimates of HE generated from microsatellite data because the absolute values of diversity estimates differ according to the marker used, and we wanted to directly compare these results to those generated in the current study.