Colonization of Australia by house mice

All Australian D-loop sequences examined belonged to the subspecies Mus musculus domesticus. This fits with the taxonomic status based on morphology [16], [17] and lymphocyte antigens [18]. This does not mean that other subspecies are not present. Some of our samples were deliberately chosen as domesticus, but in most cases the specimens we used were classified as ‘Mus musculus’ with no attempt to define subspecies, yet all were characterized as domesticus on D-loop sequencing. Thus, from our wide-ranging studies, it appears that M. m. domesticus is the predominant subspecies of house mouse in Australia. This accords well with the hypothesis tested here that house mice arrived from within the range of the western European domesticus subspecies [17], either on Dutch or British ships, and not from the closer potential source region (Southeast Asia) where a different subspecies is present, Mus musculus castaneus [19].

Looking in fine detail at the D-loop clades and haplotypes present, the specific link with the British Isles becomes clearer. The two most widely distributed clades in Australia are also the two most widely distributed clades in the British Isles: clades E and F. The haplotype of clade E that has been found in most localities in Britain (nine sites in southern Scotland, Wales and southern England [6], [7]) is also the most widespread and frequent haplotype of that clade in Australia (AUSTRALIA.01), but has only been detected in one out of 17 individuals from the Netherlands. Clade F is well documented around the northern and western periphery of the British Isles and the widespread clade F haplotype in Australia (AUSTRALIA.05) has been also found there (in a museum specimen dating from 1914 from the Isle of Lewis off the coast of Scotland [7]). On the other hand, clade F was not present among the Dutch mice we examined, nor has it been detected in the surrounding regions of continental Europe previously sampled in France, Germany and Denmark [7]–[9], [13], except for three specimens out of 43 collected in northern France (taken from a single location in Abbeville [7]). The low frequency of AUSTRALIA.01 ( = NETHERLANDS.02) and absence of clade F in our sample of contemporary Dutch mice does not fully exclude the Netherlands as a possible source area for the Australian house mouse. However, pending larger sample sizes and more detailed genetic data, a British Isles origin of Australian mice is the most reasonable interpretation of our results.

Therefore, our results agree with the historical routes of colonization, further suggesting that house mice were brought to Australia from at least two parts of the British Isles (northwestern Britain and/or Ireland [clade F] and somewhere in southern Britain [clade E]) probably early in the settlement of Australia. The early historical links between Britain and Australia date back to the claiming of Australia as part of the British Empire by James Cook in 1770. Although New Holland had already been discovered and named by the Dutch in 1606 and the west coast visited on a number of occasions, no trade or settlement was attempted at this time. The 11 ships that constituted the British First Fleet arrived in January 1788 at Botany Bay, near the present site of Sydney, to establish the first European penal colony in Australia. During the following years two other convoys also comprising numerous ships carrying settlers and supplies arrived in the newly founded colony of New South Wales. Throughout these early stages of establishment, thousands of English, Scottish and Irish convicts were transported to Australia leaving mainly from southern England and Ireland, bringing provisions and livestock with them [20]. Therefore, it seems reasonable to believe that the first mouse colonists, survivors of the months-long journey, arrived on this occasion, rapidly expanding their range throughout Australia, spreading the genetic signature of their geographic origin – clades E and F. Mice of other clades are restricted to coastal areas of Australia, near major harbors: Perth, Adelaide, Hobart, Brisbane and Darwin (Figure 2) and therefore most probably represent pockets of secondary colonization events. Thus, the genetic signature of the primary colonization appears to be retained in the extant mouse populations with a limited impact of subsequent introductions. This same phylogeographic pattern is found recurrently when house mouse populations from small isolated islands are surveyed [e.g. 4], [14], [21]–[23] but is particularly striking for a landmass as huge as Australia. Although Australia is nearly as large as Europe in area, the dynamics of arrival of house mice is still effectively like that of a small island, i.e. with a limited number of entry points and expansion from those.

The two most common haplotypes in Australia have also been found in New Zealand, with AUSTRALIA.01 being the most widespread and abundant haplotype there (scored as NZ.4) [4]. Given the shared history of activities involving British ships during settlement of the two countries this is not surprising [4], [24]. It is interesting that genetic diversity within mitochondrial clades was higher for New Zealand than for Australia. That may be because New Zealand is a multitude of islands and the sampling reflected that [4]. In particular, among the small islands of the archipelago, the mice tended to differ in D-loop haplotype, increasing overall diversity, and with the inference that the different islands had different colonization histories [4]. The sampling of Australia was dominated by specimens from the mainland where haplotypes AUSTRALIA.01 and AUSTRALIA.05 managed to spread far and wide. The observed low haplotype diversity on the Australian landmass is either the reflection of low diversity among the founders or small numbers of them, or post-colonization population bottlenecking.

Overall then, the mitochondrial data presented here provide valuable information on the colonization history of Australia by the house mouse. There are strong indications that mitochondrial DNA is a particularly useful marker for first colonization [6], [14]. However, the sequence that we used, the D-loop, is only a small part of the mitochondrial genome. More precision on the source and timing of initial colonization of Australia by house mice, associated population sizes, and the detail of secondary colonization events, will be provided by studies involving larger numbers of individuals, complete mitochondrial genome sequencing, and substantive analysis of the nuclear genome [5], [7], [25]. Nuclear genome data has the potential to provide considerably more information than currently available, including the perspective of both male and female colonization (as the maternally inherited mitochondrial genome only reflects female colonization).

The house mouse as a model organism in invasion biology

Studies on species-invasion have provided invaluable insights into ecological and evolutionary processes, highlighting the utility of alien species as model organisms [26]. In a mammalian context the house mouse is an iconic example of a successful invader [27], capable of thriving in the most remote and inhospitable locations on the planet, particularly on islands. Given that house mice have colonized islands across all climatic and biogeographic regions, island invasions constitute valuable experiments to test for local adaptation to different biotic and abiotic factors. Australia, a continent with characteristics similar to an oceanic island in terms of its invasibility, constitutes one of the most dramatic house mouse invasion zones worldwide. Knowing that Australian house mice likely came from the British Isles allows comparison of populations from the source and colonized areas, with all the genomic tools available for this model species. This could result in a better understanding of characteristics of Australian house mice such as their distinctive predisposition to plagues [2].

It has been claimed that through phylogeographic analysis house mice are good proxies for human movement patterns and settlement history over the last few thousand years because of their remarkable tendency to be transported wherever humans go [27]. The close match between post-European human settlement history and house mouse phylogeography in Australia supports this contention. It is noteworthy that house mice did not colonize from the closest geographical region, Southeast Asia, even though house mice (castaneus subspecies) have long been present there [19]. There is evidence of a long pre-European trade period between Australia, Indonesia and the Asian mainland, perhaps three to four centuries prior to the arrival of the first Europeans [28]. Apparently, despite this early nautical and commercial activity, propagule pressure of castaneus mice was not sufficient to allow them to invade and spread throughout Australia. It appears that only with the arrival and establishment of the British colony the invasion risk became significantly high, allowing the successful introduction of mice.

Samples

We obtained a total of 77 house mouse tissue and DNA samples representing 38 locations throughout mainland Australia, one from Kangaroo Island and six from Tasmania (Figure 2). The samples were provided by museums and private collectors (see Table S1 for details).

During the last two decades hundreds of D-loop sequences of Mus musculus domesticus have been published from locations all over Western Europe, including the British Isles, and our analysis includes all those available at the time (Figure S1). However, there were no sequences from the Netherlands and therefore, considering its potential as a source area for Australia, an additional set of 17 tissue samples from 10 Dutch localities were obtained from the Zoological Museum of Amsterdam and sequenced (Table S1).

Molecular methods

Genomic DNA was extracted from all samples using the Qiagen DNeasy Blood & Tissue extraction Kit according to the manufacturer's guidelines. For the Australian samples, complete mitochondrial D-loop and flanking regions were amplified by PCR with the primer pair L15774 - 5′-TGAATTGGAGGACAACCAGT-3′ and H2228 - 5′-TTATAAGGCCAGGACCAAAC-3′, using standard concentrations of DNA and reagents as described in [6]. The DNA recovered from the Netherlands museum samples was degraded and was amplified in three overlapping fragments with the following primer pairs (designed here): dloopF1-5′-GCACCCAAAGCTGGTATTCT-3′ and dloopR1-5′-TTGTTGGTTTCACGGAGGAT-3′, dloopF2 - 5′-ACTATCCCCTTCCCCATTTG - 3′ and dloopR2 - 5′-GATTGGGTTTTGCGGACTAA-3′, dloopF3-5′-ATAGCCGTCAAGGCATGAAA-3′ and dloopR3-5′-GCTTTGCTTTGTTATTAAGCTACA-3′.

Data analysis

All haplotypes that were obtained are deposited in GenBank (JF277281-JF277300). These were combined with 367 previously published haplotypes of Mus musculus domesticus for phylogenetic analysis (Figure S1). For the broadest possible comparison, the sequences were truncated to a standard length of 853 base pairs, including the whole D-loop. Based on the DNA substitution model retrieved from jModelTest0.1.1 [29], the dataset was analyzed using the TIM2+I+Γ model for Bayesian MCMC inference [30] through two simultaneous runs of five chains each, during 10 million generations sampled every 1000 steps. After a 30% burn-in, both runs were used to generate a 50% majority rule consensus tree. Mus musculus musculus and M. m. castaneus sequences were used as outgroups. In comparisons involving Australian mice, haplotype (h) and nucleotide (π) diversities were determined using Arlequin 3.11 [31].