Biological Invasions

, Volume 15, Issue 5, pp 1067–1087 | Cite as

Comparative phylogeography of invasive Rattus rattus and Rattus norvegicus in the U.S. reveals distinct colonization histories and dispersal

  • Justin B. Lack
  • Meredith J. Hamilton
  • Janet K. Braun
  • Michael A. Mares
  • Ronald A. Van Den Bussche
Original Paper

Abstract

Invasive Rattus are arguably the most costly and destructive invasive species on the planet, but little is known concerning their invasion history and population structure in the U.S. We utilized both nuclear microsatellites and mitochondrial DNA sequences (mtDNA) to compare the colonization history, patterns of gene flow, and levels of genetic diversity of Rattus rattus and R. norvegicus in the U.S. Analyses of mtDNA suggest R. rattus is characterized by a single rapid expansion into the U.S. from one or two very closely related mtDNA lineages or geographic sources. For R. norvegicus, mtDNA analyses suggest at least four invasions distinct in space and/or time have occurred to establish its distribution in the U.S. Microsatellite analyses suggest for R. rattus that dispersal is characterized by an isolation-by-distance pattern, suggesting a relatively low frequency of long distance dispersal, and low levels of establishment for novel propagules. In contrast, microsatellite analyses of R. norvegicus suggest high frequencies of long distance dispersal and essentially panmixia among nearly all sampled populations, as well as a high frequency of novel propagules entering at the east and west coasts and assimilating into established populations. We discuss these results in the context of invasive Rattus management in the U.S. and their implications for invasive species in general, as well as the implications for managing the spread of rat-borne pathogens.

Keywords

Rattus rattus Rattus norvegicus Colonization history Interspecific interaction Dispersal 

Introduction

In studying biological invasions, population genetic analyses can provide valuable information concerning colonization history, population demographics, and patterns of gene flow in both the native and invaded range (Lee 2002; Le Roux and Wieczorek 2009). These parameters are useful for identifying source populations of the original colonization and for contemporary dispersal into the invaded range, and are typically used to model a species’ invasion, allowing for extrapolations that can be used to predict routes of dispersal and future spread. If management plans are to be effective, it is vital that we understand from where invasive species are entering, as well as understand the ecological factors affecting population connectivity. Population genetics provides a powerful tool for improving the effectiveness and sustainability of these plans (Abdelkrim et al. 2005).

Although the use of population genetic analyses to indirectly quantify critical components of invasions has become a mainstay of invasion biology, genetic diversity is itself a vital component of the colonization process. The adaptability of a population to a novel environment is largely a function of the standing genetic diversity. The invasion process typically results in a genetic bottleneck from founder effects (Prentis et al. 2008), and reduced genetic diversity and small population size leads to a significant increase in extinction risk (Frankham and Ralls 1998; Allendorf and Lundquist 2003). In spite of this, many introduced species overcome the initial colonization to establish and spread (termed the “genetic paradox” of invasions; Sakai et al. 2001; Allendorf and Lundquist 2003). Studies have shown that this is likely a result of multiple geographically distinct source populations undergoing admixture in the invaded range (Kolbe et al. 2004, 2008), resulting in novel genetic combinations and elevated levels of genetic diversity. The time required for admixture to occur (and genetic diversity to rise) is often invoked as an explanation for the time lag typically observed between the initial invasion and the onset of exponential population growth and geographic spread.

Rattusrattus and R. norvegicus are arguably the most successful invasive species on the planet. Through their commensal relationship with humans they have spread to almost every corner of the globe, with R. rattus occurring on all continents and R. norvegicus excluded only from Antarctica (Nowak 1999; Aplin et al. 2003; Musser and Carleton 2005). In the U.S., R. rattus are restricted to the south-central and coastal lower 48 states with a single known population in Alaska. Their numbers in coastal areas can grow quite large due to a close association with ships and the shipping industry (i.e., shipping containers and warehouses). At least in the U.S., R. rattus are more common in the upper stories of buildings and in trees, rarely occurring on the ground or in open fields (Schwartz and Schwartz 1981; Caire et al. 1989). The generalist R. norvegicus occurs throughout the country, including Alaska and Hawaii. Suitable habitat includes the ground floor of buildings and dwellings, around sewers and garbage dumps, and can even be found in open fields around rural areas (Schwartz and Schwartz 1981; Caire et al. 1989). Where R. rattus and R. norvegicus co-occur, it has been noted that the more aggressive R. norvegicus excludes R. rattus from favorable habitat (Foster 2010; Yom-Tov et al. 1999). In essentially every habitat invaded, Rattus have had severe negative impacts on natural diversity. Of the approximately 123 island groups worldwide, about 82 % have been invaded by R. norvegicus, R. rattus, or the Polynesian rat (R. exulans; Courchamp et al. 2003), and recent reports estimated that introduced rats have been responsible for 40–60 % of all bird and reptile extinctions since 1600 (Island Conservation 2006).

In addition to damaging natural ecosystems, a more universal concern is the economic and human health impacts of invasive Rattus. Within the U.S., invasive rats are responsible for approximately 19 billion dollars in annual economic loss through the transmission of disease, structural damage to buildings, and contamination and destruction of food supplies (Pimentel et al. 2000). From an epidemiological standpoint, Rattus are known to spread many zoonoses including bubonic plague, murine typhus, rat-bite fever, Salmonella food poisoning, leptospirosis, listeriosis, chagas, trichinosis, tularemia, hantavirus and schistosomiasis (Gratz 1984). Over the last millennium, rat-borne pathogens are estimated to have killed more people than all wars and revolutions combined (Meerburg et al. 2009), and rat-borne pathogens are still a serious concern. As an example, plague (Yersinia pestis) is often thought of as a pathogen relevant only to the European Dark Ages, but the recent discovery of antibiotic resistant strains and the increased incidence of infections suggests otherwise (Galimand 1997; Keeling and Gilligan 2000). Drug-resistant Y. pestis has even been suggested as a potential bioterrorism weapon (Inglesby et al. 2000).

Despite the substantial economic and human health impacts of invasive Rattus, essentially nothing is known of their colonization history and population structure in the U.S. Information concerning the colonization history, genetic diversity, and gene flow of R. rattus and R. norvegicus within the U.S. can be used to inform eradication efforts, but more importantly can be used to assess the relative risk each species poses in terms of the import and spread of infectious disease. Invasive Rattus are notorious for their ability to utilize human transportation vectors (i.e., ships) to disperse (Sullivan 2004). Given this ability and the high volume of international shipping entering at the U.S. coastlines (6 million shipping containers annually; Frittelli et al. 2005), it is possible that propagules of both species are entering the U.S. in high numbers and assimilating with already established populations.

We utilized both mitochondrial and nuclear DNA to assess colonization history, partitioning of genetic diversity, and patterns of gene flow in the U.S. for R. rattus and R. norvegicus. Prior genetic analyses outside the U.S. revealed several divergent mitochondrial DNA lineages with geographic structure for R. rattus (Hingston et al. 2005; Tollenaere et al. 2010; Aplin et al. 2011) and R. norvegicus (Bastos et al. 2011), although there is less data available for R. norvegicus. Therefore, we can predict that mitochondrial diversity will reflect the diversity in source populations for colonizing individuals. In addition, if individuals are continuing to enter the U.S. from international localities, we predict genetic diversity will be highest in coastal localities. Finally, if individuals of either species are dispersing throughout the U.S. using human transportation vectors (i.e., trucks and trains), we predict patterns of genetic differentiation will reflect frequent long-distance dispersal and will therefore not fit a model of isolation-by-distance (IBD) across a relatively continuous landscape.

Materials and methods

Population sampling

We obtained tissues from museum collections in addition to individuals collected through trapping efforts for localities across the U.S. as well as from international localities. For R. rattus we were able to obtain tissues from 231 individuals from 24 localities (18 localities in the U.S.). For R. norvegicus we obtained 212 tissues from 26 localities (23 localities in the U.S.). Sample sizes for populations varied significantly for both species, ranging from one to 51 collected individuals. A complete list of all collected individuals, source localities, population sample sizes, and loaning institutions are given in Table S1.

MtDNA sequence generation and analyses

The complete mitochondrial cytochrome b gene (1140 bp) was amplified using primers (RattusCytbF 5′-TGACATGAAAAATCATCGTTGTAAT-3′; RattusCytbR 5′-GGTTTACAAGACCAGAGTAAT-3′) designed from an alignment of the complete mtDNA genome sequences of R. rattus (NC012374), R. norvegicus (AY769440), R. tanezumi (NC011638) and Mus musculus (NC006915) obtained from GenBank. PCR amplifications were carried out in 30 μl reactions containing 200–500 ng of DNA, 0.12 μl of 5 U/μl GoTaq Flexi DNA polymerase, 0.50 μl of each 10 μM primer, 2.4 μl Bovine Serum Albumin (0.01 g/ml), 2.4 μl of 25 μM MgCl2, 6.0 μl 5X Green GoTaq Flexi Buffer, 4.2 μl of a 10 μM nucleotide mixture, and 8.88 μl of double distilled water (ddH2O). The thermal profile consisted of an initial denaturation of 94 °C for 4 min, followed by 35 cycles of 94, 42, and 72 °C for 1 min each. A final elongation at 72 °C for 7 min ensured all reactions went to completion. Double-stranded products were purified using the Wizard SV Gel and PCR Clean-Up System (Promega Corporation, Madison, WI).

Both strands of the purified PCR products were sequenced using BigDye chain terminators following the manufacturer’s protocol (Applied Biosystems) using the PCR amplification primers as well as two internal primers (RattusCytbIntF 5′- GGCTTCTCAGTAGACAAAGC-3′; RattusCytbIntR 5′- TTTGATCCTGTTTCGTGGAGGAA-3′) designed using the mtDNA genome alignment generated above. DNA sequencing reactions were electrophoresed on a 3130 Genetic Analyzer (Applied Biosystems). Contigs were assembled and edited using Geneious v 5.5 (Drummond et al. 2010).

In addition to the complete cytochrome b sequences generated above, we obtained all available R. rattus and R. norvegicus cytochrome b sequences from GenBank (a complete list of sequences obtained from GenBank with sampling localities is available in Table S2), resulting in 275 and 229 cytochrome b sequences for R. rattus and R. norvegicus, respectively. The sequences obtained from GenBank ranged from 713 to 1,140 basepairs. Initial phylogenetic analysis (conducted in Lack et al. 2012) and haplotype network analyses (conducted herein) were performed on both the full-length sequences only and the total dataset, and results were identical concerning clade assignment and haplotype network structure, and were also consistent with previous mtDNA analyses (Robins et al. 2007; Aplin et al. 2011). Therefore, subsequent mtDNA analyses included all sequences and only these results are presented. Sequences were aligned using the Geneious v5.5 aligner (Drummond et al. 2010) and edited using MacClade v4.08 (Maddison and Maddison 2000).

To verify field identifications and GenBank records, a Bayesian phylogenetic analysis of all cytochrome b sequences was previously conducted by Lack et al. (2012). This analysis revealed all individuals previously identified as R. norvegicus to be correctly identified. However, several individuals obtained from the San Francisco Bay Area, CA and Panama City, FL were members of the cryptic species R. tanezumi, and therefore excluded from further mtDNA analysis. After confirming species identifications, we generated unrooted haplotype networks for each species (R. rattus and R. norvegicus) using TCS (Clement et al. 2000).

We also conducted several population level analyses for samples from within the U.S. We estimated haplotype diversity (h) and nucleotide diversity (π) with the software package DnaSP (Rozas et al. 2003). To investigate the demographic history of each species in the U.S. we generated mismatch distributions of pairwise differences among all individuals using DnaSP and assessed the fit of the observed distribution to a model of sudden population expansion using 1,000 bootstrap replicates (Rogers and Harpending 1992). In addition, we estimated Fu’s FS, D*, and F* (Fu 1997) as a measure of selective neutrality and population expansion, and assessed significance using coalescent simulations in DnaSP. A significant FS in the absence of significance for D* and F* suggests recent population expansion, while the opposite scenario suggests background selection is responsible for the observed pattern of genetic variation (Fu 1997).

Microsatellite data generation and analyses

We genotyped all individuals collected from localities in the U.S. at nine microsatellite loci (loci names and primers given in Table S3). Microsatellites were amplified by PCR in 15 μl reactions containing 9 μl of True Allele PCR Premix (Applied Biosystems, Inc., Foster City, CA), 4 μl ddH2O, 0.5 μl of each primer (10 μM), and 1 μl template DNA with the following conditions: an initial denaturation of 95 °C for 12 min, 35 cycles of 94 °C for 40 s, 57 °C for 40 s, and 72 °C for 30 s; and a final elongation of 72 °C for 4 min. Then 0.5 μl of product was added to 9.5 μl of loading mix containing a size standard (ROX 400HD; Applied Biosystems, Inc., Foster City, CA) and this mixture was analyzed using an ABI 3130 Genetic Analyzer and GeneMapper 3.7 to visualize microsatellite alleles and determine genotypes. All genotypes were scored twice, and anonymously randomized for the second scoring to ensure no bias was present in the final dataset.

As mentioned above, several individuals from the San Francisco Bay and Panama City populations possessed the mtDNA of the cryptic species R. tanezumi (Lack et al. 2012). While these individuals were excluded from mtDNA analyses, previous analysis of nuclear data for these populations indicated that R. tanezumi and R. rattus have undergone extensive hybridization with introgression, and measures of population structure indicated nuclear genome panmixia among these mtDNA lineages at each population (Lack et al. 2012; Conroy et al. 2012). Therefore, we conducted initial microsatellite analyses (measures of diversity and clustering analysis in Structure v2.3.2; Pritchard et al. 2000) excluding individuals with R. tanezumi mtDNA as well as using the entire dataset. Consistent with previous analyses, measures of diversity (e.g., expected and observed heterozygosity, gene diversity) and clustering analysis gave essentially the same results for both datasets (an identical optimal K, K = 4, was selected for both datasets). Therefore, all analyses and results presented here were conducted on the total dataset.

Genepop v4.0 (Raymond and Rousset 1995) was used to test for deviations from Hardy–Weinberg equilibrium (HWE) by conducting global heterozygosity excess and deficit tests for each locality and species. Each test was run for 10,000 dememorization steps followed by 100 batches of 5,000 steps each. Additionally, we used Genepop to conduct a composite linkage disequilibrium (LD) test (Weir 1996). Significance was assessed with the same MCMC settings used for the heterozygosity tests, and a Bonferroni correction was used to correct for multiple comparisons. To assess genetic diversity, we calculated the corrected gene diversity with rarefaction correction for unequal sample sizes in FSTAT v2.9.3 (Goudet 2001), excluding populations with less than 5 genotyped individuals. We estimated observed and expected heterozygosity (HO and HE, respectively) in Genalex v6.41 (Peakall and Smouse 2006). Mantel tests (Mantel 1967) of significance of regression between pairwise genetic distance (DEST, calculated using SMOGD; Crawford 2010) and straight-line geographic distance were conducted using Genalex v6.41, and this analysis excluded populations with fewer than 5 genotyped individuals. A significant positive correlation is taken to indicate an IBD pattern of divergence.

To examine fine-scale structure we used two approaches, both of which included all sampled individuals. We conducted a principle coordinate analysis (PCoA) on a pairwise genetic distance matrix for individuals and populations in Genalex v6.41. In addition, we used the Bayesian model-based clustering approach implemented in Structure v2.3.2 (Pritchard et al. 2000) to infer population structure. This approach gives the probability of assignment of each individual to postulated clusters independent of the location they were sampled, allowing for the identification of individuals with ancestry attributable to multiple populations. For each value of K (we ran independent analyses from K = 1 to K = N, where N equals the number of sampled populations in the microsatellite dataset for each species), we ran five independent analyses of 500,000 generations following 250,000 generations of burn-in under the admixture model and with the assumption that allele frequencies among populations are correlated. Convergence was evaluated by plotting likelihoods throughout the run and comparing likelihood values and population assignments between duplicate runs. We calculated the optimal number of clusters for our data using the ΔK statistic (Evanno et al. 2005). The five replicates for each K were then combined using Clumpp (Jakobsson and Rosenberg 2007), which identifies common modes among replicates runs, and resulting outputs were then used to construct graphs of Structure results in Distruct (Rosenberg 2004).

To estimate recent migration rates among sampled populations of each species, we used BayesAss v3.0.1 (Wilson and Rannala 2003). This analysis estimates proportions of non-migrants and the source of migrants for each sampled population over the last several generations (Wilson and Rannala 2003), which is an ideal approach for invasive species because invasions are such recent events (hundreds of years or less). For this analysis, only populations with at least five genotyped individuals were included for each species. Initial runs consisted of 10 million generations sampled every thousand generations and a burnin of 1 million generations. To ensure parameter space is adequately explored, Wilson and Rannala (2003) suggest acceptance rates between 20 and 40 %. Following the preliminary analyses, proposal step length was increased to 0.3 for the migration parameter (m), the allele frequencies (a), and inbreeding coefficients (f) in order to reduce acceptance rates into the recommended range. Using the updated search settings, we ran 5 replicate runs of 100 million generations sampling every 1,000 generations and 10 million burnin generations. Convergence was assessed by comparing migration estimates across replicate runs and by examining the log probability of each analysis in Tracer v1.5.

Results

Due to differences in the quantity and quality of tissue and DNA extractions, we were unable to generate both cytochrome b sequence and microsatellite genotypes for several individuals. Therefore, sample sizes differed slightly for several localities between the mtDNA and microsatellite datasets for each species. Sample sizes for each species, population, and dataset are given in Tables 1, 2, 3, 4.
Table 1

Descriptive statistics and population sample sizes (n) for the R. rattus mtDNA dataset, illustrating the number of haplotypes (h) and nucleotide diversity (π) for population with multiple samples

Locality

n

Haplotypes

h (SD)

π (SD)

Shemya Is., AK

23

1

Great Sitkin Is., AK

1

1

Little Rock, AR

6

2

0.333 (0.215)

0.00058 (0.00038)

San Francisco Bay Area, CA

23

8

0.791 (0.063)

0.00128 (0.00021)

Tehama, CA

17

1

Gainesville, FL

21

2

0.257 (0.110)

0.00023 (0.00010)

San Diego, CA

2

1

Panama City, FL

4

1

Miami, FL

3

1

Key Largo, FL

10

2

0.533 (0.095)

0.00327 (0.00058)

Baton Rouge, LA

11

2

0.182 (0.144)

0.00016 (0.00013)

Jefferson Davis, MS

1

1

Brownsville, TX

1

1

Houston, TX

7

1

Weatherford, TX

1

1

San Angelo, TX

14

6

0.681 (0.132)

0.00120 (0.00028)

Austin, TX

13

4

0.769 (0.072)

0.00157 (0.00015)

Seattle, WA

5

4

0.900 (0.161)

0.00246 (0.00058)

Total

163

21

0.806 (0.019)

0.00180 (0.00023)

Table 2

Descriptive statistics and population sample sizes (n) for the R. norvegicus mtDNA dataset, illustrating the number of haplotypes (h) and nucleotide diversity (π) for population with multiple samples

Locality

n

Haplotypes

h (SD)

π (SD)

Adak Is., AK

41

2

0.290 (0.078)

0.00204 (0.00055)

Attu Is., AK

2

1

Great Sitkin Is., AK

5

1

Sedanka Is. AK

1

1

Revillagigedo Is., AK

2

1

Fairbanks, AK

1

1

Little Rock, AR

3

2

0.667 (0.314)

0.00058 (0.00028)

San Diego, CA

11

4

0.491 (0.175)

0.00108 (0.00053)

Chicago, IL

11

3

0.473 (0.162)

0.00159 (0.00067)

Spencer, IN

10

1

Baltimore, MD

20

2

0.1 (0.088)

0.00018 (0.00015)

Albuquerque, NM

2

1

New York City, NY

1

1

Oklahoma City, OK

2

1

Corvalis, OR

4

2

0.500 (0.265)

0.00395 (0.00209)

Union, PA

42

1

Memphis, TN

16

3

0.342 (0.140)

0.00049 (0.00021)

Austin, TX

1

1

San Angelo, TX

1

1

Seattle, WA

1

1

Monroe, WV

7

5

0.905 (0.103)

0.00560 (0.00537)

Total

184

11

0.638 (0.028)

0.00294 (0.00020)

Table 3

Descriptive statistics and population sample sizes (n) for the R. rattus microsatellite dataset, illustrating the mean number of alleles across all loci (Na), the number of private alleles (NP), the observed (HO) and expected (HE) heterozygosity, and gene diversity for each population

Sampling locality

n

Na

NP

HO

HE

Gene diversity

Austin, TX

13

6.667

3

0.624

0.723

0.757

Shemya Is. AK

24

4.778

7

0.486

0.541

0.554

San Francisco Bay, CA

29

8.667

6

0.678

0.76

0.775

San Diego, CA

2

2.667

7

0.611

0.556

Panama City, FL

31

7.111

0

0.616

0.673

0.685

Key Largo, FL

10

5.444

0

0.711

0.699

0.737

Miami, FL

3

1.889

0

0.407

0.265

Gainesville, FL

21

3.556

2

0.574

0.469

0.488

Houston, TX

7

4.222

2

0.46

0.59

0.649

Little Rock, AR

6

3.667

8

0.519

0.557

0.617

Baton Rouge, LA

9

5.444

1

0.704

0.662

0.702

Jefferson Davis, MS

1

1.444

0

0.444

0.222

Tehama, CA

17

4.111

1

0.531

0.61

0.632

Brownsville, TX

1

1.667

0

0.667

0.333

San Angelo, TX

14

6.333

1

0.635

0.67

0.697

Weatherford, TX

1

1.444

0

0.444

0.222

Seattle, WA

5

5.222

1

0.711

0.693

0.781

Table 4

Descriptive statistics and population sample sizes (n) for the R. norvegicus microsatellite dataset, illustrating the mean number of alleles across all loci (Na), the number of private alleles (NP), the observed (HO) and expected (HE) heterozygosity, and gene diversity for each population

Sampling locality

n

Na

NP

HO

HE

Gene diversity

Austin, TX

1

1.444

0

0.444

0.222

Adak Is., AK

51

6.333

8

0.621

0.696

0.704

Attu Is., AK

2

1.889

0

0.556

0.375

Douglas Is., AK

1

1.111

1

0.111

0.056

Fairbanks, AK

1

1.778

1

0.778

0.389

Kagalaska Is., AK

1

1.556

0

0.556

0.278

Revillagigedo Is., AK

2

2.333

1

0.5

0.444

Sedanka Is., AK

1

1.333

0

0.333

0.167

Sitkin Is., AK

4

1.889

0

0.278

0.26

Baltimore, MD

30

7.889

7

0.522

0.684

0.699

San Diego, CA

11

6.111

9

0.525

0.7

0.744

Chicago, IL

11

3.667

2

0.434

0.459

0.483

Spencer, IN

10

2.333

1

0.589

0.425

0.439

Little Rock, AR

3

1.556

0

0.259

0.216

Memphis, TN

16

3.111

3

0.417

0.401

0.414

Albuquerque, NM

2

2

0

0.389

0.403

New York City, NY

1

1.444

0

0.444

0.222

Oklahoma City, OK

2

1.778

0

0.278

0.264

0.389

Corvalis, OR

4

2.667

0

0.444

0.458

0.537

Union, PA

42

3.111

0

0.329

0.358

0.362

San Angelo, TX

1

1.333

0

0.333

0.167

Seattle, WA

1

1.444

1

0.444

0.222

Monroe, WV

6

5

4

0.444

0.659

0.746

MtDNA analyses

Despite collection localities spanning essentially the entire U.S., mtDNA diversity for individual populations was unexpectedly low for both species (Tables 1, 2). For R. rattus, we detected 21 haplotypes from 163 individuals collected from 18 U.S. localities (Table 1), but nearly all haplotypes were very closely related (π = 0.00180). Within populations of R. rattus, we recovered either 1 or 2 haplotypes for the majority of localities, and this appeared to be independent of sample size. In addition, mtDNA diversity for R. rattus did not appear to exhibit any geographic pattern, with relatively high and comparable haplotype diversity values detected in both coastal (e.g., San Francisco Bay, h = 0.791) and central U.S. localities (e.g., Austin, TX, h = 0.769 and San Angelo, TX, h = 0.681). For R. norvegicus the pattern was very similar in terms of haplotype diversity, with the vast majority of localities consisting of 1 or 2 haplotypes (Table 2). In contrast, nucleotide diversity was higher for R. norvegicus (π range was 0.00018–0.00560 for populations, total π = 0.00294) than R. rattus (π range was 0.00016–0.00327 for populations, total π = 0.00180).

Haplotype networks revealed distinct colonization histories for R. rattus (Fig. 1) and R. norvegicus (Fig. 2). For R. rattus, nearly all haplotypes recovered in the U.S. (and Mexico, Central America, and Argentina as well) were no more than 2 mutations removed from the 2 most common and geographically widespread haplotypes (labeled Rr1 and Rr2 in Fig. 1). The only exceptions to this were two divergent haplotypes recovered in south Florida. One of these haplotypes was shared with multiple localities in South Africa and was closely related to several other haplotypes recovered in southeast Africa and the Middle East. The other haplotype was recovered from no other locality, but was closely related to several haplotypes recovered from South Africa, southeast Africa, the Middle East, India, and Indonesia. Haplotype Rr1 was recovered from the Aleutian Islands, San Francisco Bay, CA, central Texas, and multiple localities in Florida within the U.S., and outside the U.S. was recovered from Mexico, Central America, Egypt, the Lesser Antilles, East Asia, and several localities in Africa. Haplotype Rr2 was recovered in Washington, two California localities (San Francisco Bay and San Diego), central Texas, Louisiana, Arkansas, and all four Florida localities within the U.S., and outside the U.S. was recovered from East Asia and South Africa.
Fig. 1

Statistical parsimony haplotype network generated from cytochrome b sequences for the R. rattus. Colored nodes on the haplotype networks correspond to unique haplotypes, and the size of the circle corresponds to the frequency of the haplotype. The color and size of the pie for each haplotype corresponds to the sampling locality indicated in the map by a colored dot and the frequency of that haplotype at that locality, respectively. The small black nodes represent extinct or unsampled haplotypes, and each uninterrupted straight line (independent of line length) corresponds to a single mutational step

Fig. 2

Statistical parsimony haplotype network generated from cytochrome b sequences for the R. norvegicus. Colored nodes on the haplotype networks correspond to unique haplotypes, and the size of the circle corresponds to the frequency of the haplotype. The color and size of the pie for each haplotype corresponds to the sampling locality indicated in the map by a colored dot and the frequency of that haplotype at that locality, respectively. The small black nodes represent extinct or unsampled haplotypes, and each uninterrupted straight line (independent of line length) corresponds to a single mutational step

For R. norvegicus we recovered 4 divergent but relatively common and widespread haplotypes in the U.S. (labeled Rn1–Rn4 in Fig. 2), with the remaining haplotypes consisting primarily of singletons clustered around one of the widespread haplotypes (Fig. 2). Haplotype Rn1 was recovered within the U.S. from the Aleutian Islands, Washington, southern California, Arkansas, northern Illinois, and Pennsylvania, and was recovered outside the U.S. from East Asia and Central America. Haplotype Rn1 gave rise to haplotypes found in Arkansas, southern California, Mexico, and the Lesser Antilles. Haplotype Rn2 was recovered within the U.S. from the Alexander Archipelago (southeast Alaska) and mainland Alaska, southern California, New Mexico, Memphis TN, Indiana, Chicago, IL, Baltimore, MD, and West Virginia, and was recovered outside the U.S. from the Lesser Antilles, East Asia, and Argentina. Haplotype Rn2 gave rise to haplotypes found in West Virginia, Memphis, TN, Baltimore, MD, South Africa, Argentina, and the Lesser Antilles. Haplotype Rn3 was recovered within the U.S. from the Aleutian Islands, Oregon, Oklahoma, central Texas, and Indiana, and was recovered outside the U.S. from East Asia. Haplotype Rn3 gave rise to haplotypes detected only in East Asia. Haplotype Rn4 was recovered within the U.S. from Chicago, IL, Oregon, and New York, and was not recovered outside the U.S. Haplotype Rn4 gave rise to haplotypes from Chicago, IL and Denmark. The remaining haplotypes were recovered from East Asia.

Tests for historical population expansion indicate the mtDNA diversity of R. rattus in the U.S. is the result of a single rapid expansion. Fu’s FS was significantly negative (FS = −8.997, P < 0.001) while D* (D* = −0.8296, P > 0.1) and F* (F* = −1.5018, P > 0.1) were both non-significant, indicating the deviation from selective neutrality observed for this species is the result of expansion and not background selection. The mismatch distribution also supported expansion for R. rattus (Fig. 3a), with an essentially unimodal distribution of pairwise differences and low raggedness index (r = 0.0757) and R2 (R2 = 0.0317) values, which indicate the observed data fit the sudden expansion model. In contrast, demographic analyses for R. norvegicus reject the expansion model. All measures of selective neutrality were non-significant (FS = 1.845, P = 0.088; D* = −1.147, P > 0.10; F* = −0.937, P > 0.10), and the mismatch distribution (Fig. 3b) exhibited a multimodal distribution of pairwise differences (r = 0.288; R2 = 0.0893).
Fig. 3

Mismatch distribution showing the frequency of pairwise differences in cytochrome b sequence for all sampled R. rattus (a) and R. norvegicus (b) in the U.S. Observed distributions are represented by the black line, and the expected distribution under the sudden expansion model are represented by the grey line

Microsatellite analyses

For the microsatellite data, no locality exhibited significant deviation from HWE or LD for either species and all loci were polymorphic for both species. For R. rattus, measures of diversity exhibited no obvious patterns among populations (Table 3). For populations with at least 5 sampled individuals, gene diversity values ranged from 0.488 to 0.781. For R. norvegicus, measures of diversity appeared to be highest in coastal populations and lowest in populations in the center of the country (Table 4). For populations with at least 5 samples, gene diversity values ranged from 0.362 to 0.746, with the highest values in populations in close proximity to either coast; the only exception to this was the rural Pennsylvania population.

The Mantel test of IBD revealed significantly different patterns for each species in the U.S. (Fig. 4). The Mantel test for R. rattus (Fig. 4a) indicated a significant positive relationship between geographic and genetic distance (P = 0.049). This suggests a pattern of gene flow conforming to the IBD model. In contrast, we detected no significant relationship between geographic and genetic distance (P = 0.353) for R. norvegicus (Fig. 4B), rejecting a model of IBD and indicating frequent long-distance dispersal events.
Fig. 4

Resulting Mantel tests of a significant relationship between of genetic distance (DEST) based on the microsatellite dataset and straight-line geographic distance for R. rattus (a) and R. norvegicus (b). Only populations from the U.S. and with ≥5 individuals sampled were included

For the individual-based analyses of genetic structure, both species exhibited relatively little population structure. The PCoA of R. rattus (Fig. 5; 26.7 and 20.3 % of variation explained by axis 1 and 2, respectively) indicated the Shemya Is., AK and Gainesville, FL populations were relatively distinct from all other sampled populations. Interestingly, the single individual obtained from Great Sitkin Is., AK grouped with the main cluster of samples, and not with the more distinct Shemya Is., AK population. A second PCoA analysis of R. rattus with the Gainesville, FL and Shemya Is., AK populations removed revealed no additional samples or populations were distinct (results not shown), and all samples formed essentially a single cluster. For the R. norvegicus PCoA (Fig. 5; 37.8 and 22.8 % of variation explained by axis 1 and 2, respectively), the Aleutian Is., AK populations were distinct from all other sampled populations, as in the R. rattus analysis. Also, the Alexander Archipelago (southeast Alaska) samples and single individual from mainland Alaska (Fairbanks, AK) grouped relatively close to the main cluster of samples. The rural Pennsylvania population also was distinct from all other sampled individuals. In a subsequent analysis with the Aleutian Is. and Pennsylvania populations removed, no additional populations became distinct (results not shown).
Fig. 5

Principal coordinate analysis (PCoA) of genetic variation based on the microsatellite dataset for R. rattus (26.7 and 20.3 % of variation explained by axis 1 and 2, respectively) and R. norvegicus (37.8 and 22.8 % of variation explained by axis 1 and 2, respectively

For the Bayesian assignment approach implemented in Structure, the results were largely congruent with the PCoA, but with increased resolution. For R. rattus, the ΔK statistic (Fig. S1) suggested the number of clusters present in the data to be four (Fig. 6). As in the PCoA, the Shemya Is., AK and Gainesville, FL populations each formed distinct groups. A third cluster of individuals (blue in Fig. 6) consisted primarily of the Gulf Coast populations (Louisiana and Florida), but with intergradation into central Texas and the San Diego, CA and San Francisco Bay, CA populations. The fourth cluster (green in Fig. 6) consisted primarily of western U.S. populations (with the exception of Shemya Is., AK), but with intergradation into central Texas, gulf coast Texas, and Arkansas. The Greater Sitkin Is., AK individual grouped with this western cluster and not with the Shemya Is., AK population. For R. norvegicus, the ΔK statistic (Fig. S2) suggested the number of clusters present in the data to be three (Fig. 6). The first cluster (green in Fig. 6) consisted almost entirely of Aleutian Islands populations, as well as some mixed assignments in the Alexander Archipelago, AK. The second distinct cluster (green in Fig. 6) consisted entirely of the rural Pennsylvania population. The third cluster (red in Fig. 6) consisted of all remaining populations with little intergradation from the other two clusters. The Alexander Archipelago, AK and the single Fairbanks, AK (mainland Alaska) individual also grouped more closely with the large cluster (red in Fig. 6) than with the Aleutian Islands populations.
Fig. 6

Results of the Bayesian clustering analysis implemented in Structure under the admixture model for R. rattus and R. norvegicus based on the microsatellite dataset. Vertical columns represent the assignment probabilities to each of the inferred clusters identified for the optimal value of K (K = 4 for R. rattus and K = 3 R. norvegicus)

Assessment of pairwise population migration rates revealed several differences between R. rattus and R. norvegicus. Duplicate runs for each species produced essentially identical results, and Tables 5 and 6 present the mean across the 5 replicates for R. rattus and R. norvegicus, respectively. The overall mean pairwise rate between sampled populations was higher for R. norvegicus (0.0327) than for R. rattus (0.0195). For R. rattus, pairwise migration rates (Table 5) appeared to support the results of the Mantel test (Fig. 4), with the majority of populations exhibiting higher rates of gene flow between geographically more proximate populations. As an example, the Austin, TX and Houston, TX populations exhibited an approximately 10-fold higher migration rate with San Angelo, TX than all other sampled populations. Moreover, the majority of gene flow among these Texas populations was occurring into the more rural San Angelo, TX population and not in the other direction. For Seattle, WA migration again was highest for the relatively close San Francisco Bay, CA and Tahoma, CA populations. There were two exceptions to this pattern, where the highest migration rates into San Francisco Bay, CA were from Key Largo, FL and Baton Rouge, LA, illustrating the dispersal potential of this species. For R. norvegicus, we did not observe a pattern of migration rates correlated with geographic proximity (Table 6). For example, the highest rate of gene flow into the Chicago, IL population originated from the San Diego, CA population, while the highest rate of gene flow into the Baltimore, MD population originated from the more rural Spencer, IN population. The only clear exception of this was the elevated migration rate from the Monroe, WV population into the nearby Baltimore, MD population.
Table 5

Pairwise proportion of ancestry estimates based on the analysis of the microsatellite dataset in BayesAss

 

Austin, TX

Shemya Is., AK

San Francisco, CA

Panama City, FL

Key Largo, FL

Gainesville, FL

Houston, TX

Little Rock, AR

Baton Rouge, LA

Tehama, CA

San Angelo, TX

Seattle, WA

Austin, TX

0.6809

0.0136

0.0148

0.0142

0.0135

0.0135

0.0136

0.0136

0.0135

0.0136

0.1817

0.0136

Shemya, AK

0.0102

0.8951

0.0088

0.0087

0.0103

0.0087

0.0101

0.0103

0.0102

0.0086

0.0087

0.0104

San Francisco, CA

0.0079

0.0078

0.8802

0.0081

0.0079

0.0156

0.0078

0.0078

0.0079

0.0083

0.0327

0.0079

Panama City, FL

0.0073

0.0073

0.0086

0.9161

0.0073

0.0087

0.0073

0.0074

0.0073

0.0075

0.0078

0.0074

Key Largo, FL

0.015

0.0151

0.1508

0.0309

0.6827

0.0151

0.015

0.0151

0.0151

0.015

0.0152

0.015

Gainesville, FL

0.0094

0.0094

0.0096

0.0098

0.0095

0.8957

0.0094

0.0095

0.0095

0.0094

0.0094

0.0095

Houston, TX

0.0174

0.0293

0.0176

0.0173

0.0174

0.0174

0.6852

0.0175

0.0174

0.0176

0.1285

0.0174

Little Rock, AR

0.0184

0.0185

0.0232

0.0185

0.0185

0.0213

0.0185

0.6866

0.0185

0.0185

0.1211

0.0185

Baton Rouge, LA

0.0158

0.0158

0.1425

0.0297

0.0158

0.0158

0.0159

0.0159

0.6834

0.0158

0.0177

0.0159

Tehama, CA

0.0108

0.0108

0.0109

0.0108

0.0109

0.0109

0.0109

0.0108

0.0109

0.8804

0.0109

0.0109

San Angelo, TX

0.0126

0.0128

0.0179

0.021

0.0126

0.0128

0.0126

0.0126

0.0126

0.0144

0.8453

0.0128

Seattle, WA

0.0198

0.0196

0.0405

0.0209

0.0196

0.0196

0.0196

0.0198

0.0197

0.065

0.0196

0.6886

Only populations with ≥5 individuals sampled were included

Table 6

Pairwise proportion of ancestry estimates based on the analysis of the microsatellite dataset in BayesAss

 

Adak, AK

Baltimore, MD

San Diego, CA

Chicago, IL

Spencer, IN

Memphis, TN

Corvalis, OR

Pennsylvania

West Virginia

Adak, AK

0.9545

0.0058

0.0057

0.0057

0.0057

0.0057

0.0056

0.0057

0.0056

Baltimore, MD

0.0091

0.9175

0.0102

0.0092

0.0111

0.0121

0.0112

0.0085

0.0111

San Diego, CA

0.0166

0.0165

0.7135

0.1476

0.0237

0.0166

0.0245

0.0166

0.0245

Chicago, IL

0.0165

0.0181

0.0165

0.8659

0.0164

0.0169

0.0165

0.0169

0.0164

Spencer, IN

0.0168

0.1974

0.0168

0.0167

0.6854

0.0167

0.0167

0.0168

0.0167

Memphis, TN

0.0133

0.0138

0.0133

0.0134

0.0133

0.8928

0.0133

0.0133

0.0133

Corvalis, OR

0.0257

0.0259

0.026

0.1208

0.0259

0.0265

0.6976

0.0258

0.0259

Pennsylvania

0.0067

0.0067

0.0067

0.0067

0.0067

0.0067

0.0067

0.9464

0.0067

West Virginia

0.0235

0.05

0.0227

0.1053

0.0227

0.0373

0.0227

0.0227

0.693

Only populations with ≥5 individuals sampled were included

Discussion

Rattus colonization history in the U.S

Through archeological evidence and historical record, we have some understanding of the early colonization of the U.S. by R. rattus and R. norvegicus. R. rattus appears to have been the first to arrive in North America, with archeological evidence placing them on the island of Hispaniola with Columbus in 1492, and with established populations on the east coast of the continental U.S. by the mid 1500 s (Armitage 1993). In contrast, R. norvegicus arrived in the U.S. in the mid 1700 s with the massive wave of British immigrants that continued into the late 1700s (Armitage 1993). By the early 1800s the larger and more aggressive R. norvegicus had caused a drastic reduction in R. rattus numbers (MacGillivray 1838). This scenario is strikingly similar to the spread of R. rattus from its native range on the Indian subcontinent beginning in the first millennium BC, and reaching essentially every corner of the Old World inhabited by humans by the second century AD. In the eighteenth century, R. norvegicus rapidly expanded out of central Asia into Europe, displacing R. rattus in much of the newly invaded range (Twigg 1975).

Although information gleaned from the historical record is important in understanding the early introduction of Rattus in the U.S. and elsewhere, it typically cannot deliver insight at the resolution required to understand the biological properties of the invasion, providing little concerning the source and diversity of propagules that went on to establish and produce the massive invasive rat populations present today. However, the combination of these data sources can be used to draw more powerful inferences, as has been shown in the study of R. rattus in the Mediterranean (Ruffina and Vidal 2010) and the commensal relationship between humans and the Polynesian rat (R. exulans) in the South Pacific (Matisoo-Smith et al. 1998; Matisoo-Smith and Robins 2004). Our mtDNA analysis provides insight into the distinct differences between R. rattus and R. norvegicus in their colonization of the U.S., even in the absence of a comprehensive global sampling.

It is clear from our haplotype network (Fig. 1), as well as previous studies (Hingston et al. 2005; Tollenaere et al. 2010; Aplin et al. 2011), that considerable mtDNA diversity exists for R. rattus at the global scale. In spite of this, we detected only a subsampling of this genetic diversity within the U.S. Overall nucleotide diversity for R. rattus that was nearly half that of R. norvegicus and essentially all U.S. haplotypes forming a single star-shaped cluster (Fig. 1); the only exceptions were two divergent haplotypes that were recovered in coastal south Florida. In addition, the mismatch distribution (Fig. 3a) and neutrality statistics suggest a single rapid expansion best explain these data. This pattern suggests several possible scenarios for the colonization history of the U.S. by R. rattus.

Although it is possible our geographic sampling was too sparse to be representative of the overall mtDNA diversity of R. rattus in the U.S., this is unlikely. Our knowledge of the current distribution of R. rattus in the U.S. suggests it has been reduced to populations only in the southeast, the Gulf Coast, and the Pacific coast (Jackson 1982), and our sampling includes multiple localities in each of these regions. Alternatively, it is possible R. rattus mtDNA diversity in the U.S. was historically much higher, and representative of the mtDNA diversity of the native range. Subsequent invasion and competitive exclusion by R. norvegicus did lead to a widespread and documented bottleneck for R. rattus in the U.S. (MacGillivray 1838; Armitage 1993), eliminating much of the mtDNA diversity of the original R. rattus populations. However, the bottleneck scenario is an unlikely explanation for the observed pattern because, if divergent mtDNA haplotypes had been present at a significant frequency, we would not expect the same haplogroup to persist in R. rattus populations from the Alaska to Florida. In addition, evidence suggests many coastal populations of R. rattus were never adversely affected by the later invasion of R. norvegicus (Silver 1927; Ecke 1954; Landon 1991). Rattus rattus and R. norvegicus are sympatric over most of their global range, including South Africa. In comparative mtDNA analyses of Rattus in South Africa, Bastos et al. (2011) found considerable mtDNA diversity for R. rattus, suggesting a complex colonization history with multiple global sources of divergent mtDNA lineages and indicating interspecific interactions were most likely not obscuring colonization history.

As the most likely explanation for the distribution of R. rattus mtDNA diversity in the U.S., we suggest colonization occurred by two closely related mtDNA lineages (Rr01 and Rr02), and subsequent colonizers from divergent mtDNA lineages have not invaded and spread. In an examination of global R. rattus mtDNA diversity, Aplin et al. (2011) noted that the initial expansion out of the Indian subcontinent appeared to originate from a single mtDNA lineage, which they termed the “ship rat” lineage, that then spread across the globe. The haplotypes in the main cluster of U.S. R. rattus recovered in our study are members of this “ship rat” lineage (Aplin et al. 2011), indicating the initial colonization of the U.S. and any subsequent invasions were essentially all members of this group. The presence of two common and widespread haplotypes in the U.S. (Rr1 and Rr2) suggests either two distinct waves of invasion or a single invasion by both mtDNA lineages spread across the country. Initial introductions of R. rattus into the U.S. undoubtedly came from Europe onto the east coast and spread west following human colonization (Armitage 1993). In contemporary times, as international trade and shipping have rapidly expanded, the vast majority of incoming shipping on the west coast originates in Asian ports, while the majority of cargo entering the U.S. on the east coast originates in Europe, Africa, and the Middle East (Kaluza et al. 2010). It is evident in this and previous analyses of black rat mtDNA (Aplin et al. 2011; Bastos et al. 2011), that R. rattus in the source localities for much of the cargo transported to the U.S. possess mtDNA haplotypes divergent from the U.S. haplogroup, but these haplotypes are simply not entering the U.S. While Rr1 and Rr2 are each found outside of the Americas at a high frequency, the only haplotypes derived from Rr1 and Rr2 that were detected outside the Americas were three haplotypes detected in South Africa and a single haplotype from Reunion, an island near Madagascar. The two divergent haplotypes recovered in coastal Florida are members of an mtDNA haplogroup recovered in East Asia, Africa, the Middle East, and India. The presence of these haplotypes in coastal Florida suggests mtDNA lineages distinct from the “ship rat” lineage are being readily spread, but are not being incorporated into the majority of U.S. populations and do not appear to be spreading from the coastal localities where they are initially introduced.

The distribution of mtDNA diversity for R. norvegicus suggests a different colonization history relative to that of R. rattus. The mismatch distribution (Fig. 3b) and measures of selective neutrality all reject a single rapid expansion for this species and instead support a more complex scenario of multiple introductions from multiple mtDNA stocks. We detected four relatively widely distributed and frequent haplotypes in the U.S., suggesting at least four distinct invasions. Moreover, the geographic distribution of these four haplotypes in the U.S. gives some indication of where these invasions originated, although without a more extensive global sampling any conclusions are speculative. For haplotype Rn1, the distribution included both western and eastern U.S. populations, but it and the haplotypes it produced were more frequently detected in the west, and it was also detected in East Asia. This suggests an introduction from Asia that then spread east. Haplotype Rn2 was most frequently detected on the east coast and appeared to decrease in frequency moving west across the U.S. In addition, the more rare haplotypes closely related to Rn2 were all located in eastern U.S. localities, in addition to being recovered in the Lesser Antilles, Argentina, and South Africa. This suggests Rn2 may represent an east coast invasion. Haplotype Rn3 was recovered in the Aleutian Islands, AK, Oregon, central Texas, Oklahoma, and West Virginia in the U.S., as well as from East Asia. This haplotype distribution in the U.S. appears equivocal in supporting an east or west coast invasion, and therefore requires further sampling, although its high frequency in the relatively isolated Aleutian Islands populations suggests invasion from the west. Finally, haplotype Rn4 was recovered almost exclusively on the east coast, with one individual from Oregon, and the remaining nine individuals from Chicago (8 individuals) and New York (1 individual), suggesting an origin on the east coast of the U.S. The presence of a closely related haplotype recovered from Denmark and another closely related singleton from Chicago, IL also supports a colonization from the east coast for this lineage.

While the lack of a thorough global sampling for either R. rattus and R. norvegicus hinders our ability to make strong inferences concerning some aspects of their invasion history, we can make relative comparisons between the two species. It is clear that R. norvegicus in the U.S. have originated from a diversity of source populations, with at least four divergent mtDNA lineages detected at high frequency (Fig. 2). Furthermore, the significantly higher gene diversity detected for R. norvegicus nuclear microsatellites in the coastal localities relative to the localities in the center of the U.S. suggests there is still a high influx of individuals that are integrating into the established coastal populations, a pattern also observed for invading cheat grass and Japanese oyster drill (Martel et al. 2004; Novak and Mack 2001). In contrast, R. rattus in the U.S. appear to be almost entirely derived from the same mtDNA lineage (the “ship rat” lineage; Aplin et al. 2011) that initially expanded out of India to colonize most of the planet. Furthermore, the lack of a clear difference in nuclear gene diversity between coastal populations and more inland populations, as well as the presence of divergent mtDNA haplotypes in coastal Florida but nowhere further inland (including the other two Florida localities), suggests that individuals that arrive at coastal populations from localities outside North America are not likely to integrate into already established R. rattus populations.

Genetic structure and dispersal in the U.S.

For species dispersing without the aid (or inhibition) of humans, it is expected that geography will be the overriding factor and that gene flow across a continuous landscape will roughly fit a pattern of isolation by distance (Skellman 1951; Kimura 1953). For organisms that are commensal with humans, the frequency of long-distance dispersal events increases dramatically (Suarez et al. 2001). The patterns detected for dispersal and population structure for R. rattus and R. norvegicus have implications in both rat management and zoonotic disease epidemiology, and suggest distinct patterns for each species.

For R. rattus, the Mantel test supported an overall pattern of IBD (Fig. 4a), suggesting long-distance dispersal events are rare and gene flow is likely occurring through more natural mechanisms of dispersal as opposed to the utilization of human transportation vectors. In contrast, we detected no IBD for R. norvegicus, with the relationship between genetic and geographic distance essentially flat (Fig. 4b). A similar lack of IBD has been detected for the Quagga mussel (Dreissena bugensis) in the Great Lakes area of the U.S. and this was similarly attributed to a high frequency of long-distance dispersal due to human-aided dispersal (boat transportation; Wilson et al. 1999). This suggests that R. norvegicus may be utilizing human transportation vectors for frequent long-distance dispersal events much more frequently than R. rattus. Although life-history similarities between species can result in similarities in their population biology (e.g., population size, dispersal rate, etc.), the striking similarities in life-history strategies for R. rattus and R. norvegicus may be driving competitive interactions and therefore resulting in distinct patterns in their invaded range. The Norway rat has long been regarded as the much more aggressive of the pair, displacing R. rattus in most of the localities they co-inhabit (although for exceptions see Silver 1927; Ecke 1954), and typically forcing R. rattus into less desirable habitat (Foster 2010). With the possible exception of southern Florida (Armitage 1993) and a few islands, R. norvegicus has invaded all U.S. localities where R. rattus occurs or previously occurred. The absence of R. norvegicus in coastal Florida and the fact that the Key Largo, FL and Miami, FL populations were the only U.S. populations where R. rattus mtDNA haplotypes divergent from the main cluster were detected lends further support to the proposed role of R. norvegicus in limiting establishment and dispersal of R. rattus in the U.S., and possibly even replacing R. rattus in many localities.

The overall degree of genetic differentiation between populations appears to be very low. The PCoA (Fig. 5) indicated the Aleutian Islands populations for both species are genetically distinct from all other populations, which is not surprising given the extreme isolation of these islands. However, for R. rattus in Alaska, all but one sample was obtained from a single island, Shemya Is., that is uninhabited by humans, and these samples represented the distinct cluster of Alaskan samples in both the PCoA (Fig. 5) and the Structure analysis (Fig. 6). The single individual from Great Sitkin Is., an island located closer to the Alaskan mainland and occurring on the Great Circle trading route that has high ship activity (approximately 3,100 ships annually) between East Asia, Alaska, and the west coast of the contiguous 48 states, actually grouped within the main cluster of individuals. Similarly, R. norvegicus collected from the Alexander Archipelago in southeastern Alaska and the single individual from Fairbanks, AK on the Alaskan mainland grouped more closely with the main R. norvegicus cluster than the vast majority of the Aleutian Is. R. norvegicus in the PCoA and Structure analyses. This suggests that, while the isolation of Alaska has led to some distinction between both species in the contiguous 48 states and their conspecifics in Alaska, international trade and travel is still transporting these species and allowing for a detectable level of gene flow.

In addition to the divergent Alaskan populations, both species exhibited additional population structure, but diversity estimates suggest these may be the result of sampling artifact. For R. rattus, the Gainesville, FL population was distinct in both the PCoA (Fig. 5) and the Structure analysis (Fig. 6), which was unexpected given its location in close proximity to the other three Florida populations we sampled, its central location in the state with the most significant north/south interstate in Florida passing through it, and the population size (>250,000) of the Gainesville metropolitan area. The R. rattus collected from Gainesville were all taken from a single locality and may represent a unique population within the city, with an origin distinct from other Florida R. rattus. It is interesting to note that this population had low mtDNA haplotype diversity (Table 1) and the lowest microsatellite gene diversity (Table 3) of all sampled R. rattus populations, further attesting to its uniqueness and justifying further study. For R. norvegicus, the only population found to be distinct from the main U.S. cluster (aside from the Alaskan individuals) was collected from a single farm in rural Pennsylvania. This population consisted of a single haplotype for 42 individuals (Table 2), and the lowest microsatellite gene diversity of any sampled R. norvegicus population. The presence of a single mtDNA haplotype among such a large sample size and the very low nuclear genetic diversity suggests this population may represent a single family group, and its distinction from the remainder of the U.S. R. norvegicus is likely a sampling artifact. However, it is important to note that neither the Gainesville, FL R. rattus population nor the Union, PA R. norvegicus population exhibited significant deviations from HWE.

With the exception of the populations mentioned above, our analyses suggested very little genetic distinction among all other sampled populations. The PCoA of either species with the aforementioned populations removed resulted in a single cluster of individuals with no discernible structure (Not Shown). In the model-based approach of Structure, the optimal K for R. norvegicus was three (Fig. 6), with near complete uniformity of assignment for all individuals to a single cluster (red in Fig. 6) to the exclusion of the Aleutian Islands populations and the rural Pennsylvania population. The Structure analysis of R. rattus microsatellites did detect some genetic structure not obvious in the PCoA. The optimal K for R. rattus was four with the Gainesville, FL and Aleutian Islands populations distinct from all others, and with the remaining two clusters (green and blue in Fig. 6) corresponding roughly to an east/west gradient, but with significant intergradation in the Texas and Arkansas populations. This is again consistent with the IBD pattern suggested by the Mantel analysis, with relatively natural, stepwise dispersal occurring from coastal populations into the interior of the U.S.

Rattus management implications

Given the tremendous ecological, economical, and human health impacts of invasive Rattus, management of these costly invasive species is extremely important. Successful management and eventual eradication has been achieved on many islands, and these techniques are increasing in efficacy allowing for success at expanding geographic scales for rats as well as other organisms (Howald et al. 2007; Simberloff 2009). Our sampling regime did not permit a high-resolution examination of complex gene flow scenarios (i.e., Guillemaud et al. 2010), which is important information for understanding the role of various dispersal pathways in maintaining population connectivity and identifying eradication units (Robertson and Gimmell 2004). Nonetheless, our analysis does provide insight into the overall extent of gene flow among the sampled populations, as well as relative differences in the extent and diversity of incoming propagules. In terms of propagule pressure from international localities, our data suggests significantly higher rates of establishment and genetic diversity of new individuals for R. norvegicus than for R. rattus. Therefore, a management plan for R. rattus would be less likely to require a strong focus on monitoring incoming ships and freight, and focus the majority of resources on simply eliminating the already established populations, while an R. norvegicus plan would likely need to focus on eliminating incoming propagule pressure prior to any attempt to eliminate the already established populations. Recolonization has been a common issue in past rat eradication programs for the primary reason that adequate attention is not always devoted to eliminating the source of colonizing individuals (Abdelkrim et al. 2005). In terms of dispersal within the U.S., R. norvegicus appears to be exhibiting a relatively high frequency of long-distance movements and the lower 48 states essentially represents a single panmictic unit. For R. rattus, patterns of gene flow fit an IBD model and clustering analysis detected slight differentiation between the west and the east, but gene flow again appears to be high among populations in the lower 48 states. With these high levels of gene flow at such a large geographic scale, a concerted nation-wide effort would be necessary to effectively manage invasive Rattus in the U.S. and prevent recolonization of eradicated areas from unmanaged populations.

In addition to overall propagule pressure and dispersal, interspecific interactions and ecological context must also be considered in the management of Rattus in the U.S. If interaction between R. rattus and R. norvegicus is a key factor in limiting the ongoing establishment of new R. rattus individuals, as has been previously suggested (Foster 2010; Yom-Tov et al. 1999), then the patterns of propagule pressure and gene flow detected for R. rattus in this study are only applicable in a context where these two species co-occur. Management plans in the U.S. should therefore be developed jointly for these two species. Also, because R. norvegicus appear to be stifling R. rattus colonization and dispersal, management efforts should target R. rattus first, and only target R. norvegicus after successful removal of R. rattus. This type of relationship between R. novegicus and R. rattus has also been observed for R. rattus and Mus musculus, where successful eradication of R. rattus on some islands has resulted in a population explosion of Mus musculus (Witmer et al. 2007). Within the U.S., removal of R. norvegicus prior to R. rattus may allow for a rapid increase in R. rattus population size or a significant influx of R. rattus propagules.

Disease ecology implications

Because of their unparalleled role in the spread of zoonotic disease (Gratz 1984), it is important to consider our analysis in the context of rat-borne pathogens. In the study of an infectious disease, the ability to predict the spread of the causative agent is extremely valuable in stopping disease transmission and minimizing negative impacts. Population genetic analysis of the host can provide insight into the rate and geographic extent of transmission for a pathogen (Holmes 2008; Biek and Real 2010), and these approaches have been effective in understanding the epidemiology and transmission history of many zoonotic pathogens including rabies in raccoons (Cullingham et al. 2009) and dengue fever and its mosquito vector, Aedes aegypti (Urdaneta-Marquez and Failloux 2010).

For the U.S., our analyses suggest R. norvegicus presents a greater risk than R. rattus in their ability to bring a diversity of pathogens from various international sources and spread them across the U.S. The mean pairwise migration rate among our sampled localities was higher for R. norvegicus than for R. rattus, suggesting gene flow among populations is in general higher for the Norway rat, and migration rates, cluster analyses, and Mantel test suggest long-distance dispersal events are more frequent for R. norvegicus. Under this scenario, an infectious pathogen found in R. norvegicus can not only spread rapidly, but it is extremely difficult to predict patterns of dispersion when long-distance dispersals are frequent, as has been observed for some bat-borne pathogens (Breed et al. 2010). Another issue with rat-borne pathogens in the U.S. is the potential diversity of source locations from which incoming rats may originate. Whereas our analyses suggest R. rattus are not entering the U.S. and assimilating with the established populations, R. norvegicus does appear to be entering the U.S. on both coasts, and likely from multiple distinct source localities. A major concern in disease epidemiology is the potential for increased virulence to arise through recombination among distinct strains or genotypes (e.g., Gibbs et al. 2001; Grigg et al. 2001; He et al. 2009). If R. norvegicus are entering the U.S. from distinct source localities on opposite coasts and then spreading across the U.S., as our analyses suggest, potentially distinct pools of rat-borne zoonotic pathogens may be coming into contact within the borders of the U.S. Given the human health risk, this certainly warrants further study.

Conclusions

This study represents a first approximation of colonization history and contemporary dispersal for invasive Rattus at a large geographic scale and in a country with a complex transportation infrastructure. While these scenarios for the origin of each haplogroup are speculative, they represent starting hypotheses for future study and further geographic sampling. We detected clear differences in colonization history and contemporary patterns of gene flow and propagule pressure for invasive Rattus in the U.S., suggesting differences in the ability of these species to spread zoonotic pathogens. Moreover, we found evidence that ecological interactions between R. rattus and R. norvegicus may be driving the contemporary distribution of genetic diversity for R. rattus, illustrating the importance of considering both population genetics and ecological parameters in modeling species invasions and developing effective management approaches. For R. rattus, it appears that the presence and high propagule pressure of R. norvegicus may be prohibiting new propagules from establishing and even replacing already-present R. rattus. Our analyses, while informative, require a much more thorough sampling of international localities to understand the importance of various potential sources in generating current patterns of genetic diversity; this is especially true for European populations, where historical records indicate both the R. rattus and R. norvegicus lineages that invaded the U.S. originated. Finally, a more thorough sampling within the U.S. is needed to increase resolution in terms of migration rates, allowing us to identify the major routes of dispersal for these species (e.g., the Mississippi River waterway vs the I-35 transportation corridor for north/south human-aided dispersal).

Supplementary material

10530_2012_351_MOESM1_ESM.eps (645 kb)
Supplementary material 1 (EPS 644 kb)
10530_2012_351_MOESM2_ESM.eps (654 kb)
Supplementary material 2 (EPS 654 kb)
10530_2012_351_MOESM3_ESM.docx (204 kb)
Supplementary material 3 (DOCX 203 kb)
10530_2012_351_MOESM4_ESM.docx (92 kb)
Supplementary material 4 (DOCX 91 kb)
10530_2012_351_MOESM5_ESM.docx (86 kb)
Supplementary material 5 (DOCX 86 kb)

References

  1. Abdelkrim J, Pascal M, Calmet C, Samadi S (2005) Importance of assessing population genetic structure before eradication of invasive species: examples from insular Norway rat populations. Conserv Biol 19:1509–1518CrossRefGoogle Scholar
  2. Allendorf FW, Lundquist LL (2003) Introduction: population biology, evolution, and control of invasive species. Conserv Biol 17:24–30CrossRefGoogle Scholar
  3. Aplin KP, Chesser T, ten Have J (2003) Evolutionary biology of the genus Rattus: profile of an archetypal rodent pest. In: Singleton GR, Hinds LA, Krebs CJ, Spratt DM (eds) Rats, mice, and people: rodent biology and management. The Australian Centre for International Agricultural Research (ACIAR), Canberra, pp 487–498Google Scholar
  4. Aplin KP, Suzuki H, Chinen AA et al (2011) Multiple genetic origins of commensalism and complex dispersal history of black rats. PLoS ONE 6:e26357PubMedCrossRefGoogle Scholar
  5. Armitage PL (1993) Commensal rats in the New World, 1492–1992. Biologist 40:174–178Google Scholar
  6. Bastos AD, Nair D, Taylor PJ et al (2011) Genetic monitoring detects an overlooked cryptic species and reveals the diversity and distribution of three invasive Rattus congeners in South Africa. BMC Genet 12:26PubMedCrossRefGoogle Scholar
  7. Biek R, Real LA (2010) The landscape genetics of infectious disease emergence and spread. Mol Ecol 19:3515–3531PubMedCrossRefGoogle Scholar
  8. Breed AC, Field HE, Smith CS, Edmonston J, Meers J (2010) Bats without borders: long-distance movements and implications for disease risk management. EcoHealth 7:204–212PubMedCrossRefGoogle Scholar
  9. Caire W, Tyler JD, Glass BP, Mares MA (1989) Mammals of Oklahoma. University of Oklahoma Press, Norman, OKGoogle Scholar
  10. Clement M, Posada D, Crandall K (2000) TCS: a computer program to estimate gene genealogies. Mol Ecol 9:1657–1660PubMedCrossRefGoogle Scholar
  11. Conroy CJ, Rowe KC, Rowe KMC, Kamath PL, Aplin KP, Lucia H, James DK, Moritz C, Patton JL (2012) Cryptic genetic diversity in Rattus of the San Francisco Bay region, California. Biol Invasions. doi:10.1007/s10530-012-0323-9
  12. Courchamp F, Chapuis JL, Pascal M (2003) Mammal invaders on islands: impacts, control, and control impact. Biol Rev 78:347–383PubMedCrossRefGoogle Scholar
  13. Crawford NG (2010) SMOGD: software for the measurement of genetic diversity. Mol Ecol Res 10:556–557CrossRefGoogle Scholar
  14. Cullingham CI, Kyle CJ, Pond BA, Rees EE, White BN (2009) Differential permeability of rivers to raccoon gene flow corresponds to rabies incidence in Ontario, Canada. Mol Ecol 19:3592–3602Google Scholar
  15. Drummond AJ, Ashton B, Cheung M et al (2010) Geneious, version 5.0. http://www.geneious.com
  16. Ecke DH (1954) An invasion of Norway rats in southwest Georgia. J Mamm 35:521–525CrossRefGoogle Scholar
  17. Evanno G, Regnaut S, Goudet J (2005) Detecting the number of clusters of individuals using the software STRUCTURE: a simulation study. Mol Ecol 14:2611–2620PubMedCrossRefGoogle Scholar
  18. Foster SP (2010) Interspecific competitive interactions between Rattus norvegicus and R. rattus. Unpublished Thesis. University of Waikato, New ZealandGoogle Scholar
  19. Frankham R, Ralls K (1998) Conservation biology: inbreeding leads to extinction. Nature 392:441–442CrossRefGoogle Scholar
  20. Frittelli JF, Lee MR, Medalia J, O’Rourke R, Perl R (2005) Port and maritime security: background and issues for congress. Novinka Books, New York City, NYGoogle Scholar
  21. Fu YX (1997) Statistical tests of neutrality of mutations against population growth, hitchhiking and background selection. Genetics 133:693–709Google Scholar
  22. Galimand M (1997) Multidrug resistance in Yersinia pestis mediated by a transferable plasmid. N Engl J Med 337:677–680PubMedCrossRefGoogle Scholar
  23. Gibbs MJ, Armstrong JS, Gibbs AJ (2001) Recombination in the hemagglutinin gene of the 1918 “Spanish flu”. Science 293:1842–1845PubMedCrossRefGoogle Scholar
  24. Goudet J (2001) FSTAT, a program to estimate and test gene diversities and fixation indices (version 2.9.3). Department of Ecology and Evolution, University of Lausanne, Switzerland. Available at: http://www2.unil.ch/popgen/softwares/fstat.htm. Accessed 10 January 2012
  25. Gratz NG (1984) The global public health importance of rodents. In: Dubock AC (ed) Proceedings of a conference on the organization and practice of vertebrate pest control. World Health Organization, Geneva, Switzerland, pp 413–435Google Scholar
  26. Grigg ME, Bonnefoy S, Hehl AB, Suzuki Y, Boothroyd JC (2001) Success and virulence in Toxoplasma as the result of sexual recombination between two distinct ancestries. Science 294:161–165PubMedCrossRefGoogle Scholar
  27. Guillemaud T, Beaumont MA, Ciosi M, Cornuet JM, Estoup A (2010) Inferring introduction routes of invasive species using approximate Bayesian computation. Heredity 104:88–99PubMedCrossRefGoogle Scholar
  28. He CQ, Xie ZX, Han GZ et al (2009) Homologous recombination as an evolutionary force in the avian influenza A virus. Mol Biol Evol 26:177–187PubMedCrossRefGoogle Scholar
  29. Hingston M, Goodman SM, Ganzhorn JU, Sommer S (2005) Reconstruction of the colonization of southern Madagascar by Rattus rattus. J Biogeogr 32:1549–1559CrossRefGoogle Scholar
  30. Holmes EC (2008) Evolutionary history and phylogeography of human viruses. Annu Rev Microbiol 62:307–328PubMedCrossRefGoogle Scholar
  31. Howald G, Donlan CJ, Galvan JP et al (2007) Invasive rodent eradication on islands. Conserv Biodiv 21:1258–1268Google Scholar
  32. Inglesby TV, Dennis DT, Henderson DA et al (2000) Plague as a biological weapon. J Am Med Assoc 283:2281–2290CrossRefGoogle Scholar
  33. Island Conservation (2006) Evaluation of primary and secondary exposure risks to land bird species for: experimental use permit application for field efficacy trial of 0.0025% brodifacoum broadcast bait to eradicate introduced rats from Aleutian Islands in the Alaska Maritime National Wildlife Refuge, July 20, 2006. Island Conservation, Center for Ocean Health, Santa Cruz, CAGoogle Scholar
  34. Jackson WB (1982) Norway rat and allies. In: Chapman JA, Feldhammer GA (eds) Wild Mammals of North America. Johns Hopkins University Press, Baltimore, MD, pp 1077–1088Google Scholar
  35. Jakobsson M, Rosenberg NA (2007) Clumpp a cluster matching and permutation program for dealing with label switching and multimodality in analysis of population structure. Bioinformatics 23:1801–1806PubMedCrossRefGoogle Scholar
  36. Kaluza P, Kolzsch GastnerMT, Blasius B (2010) The complex network of global cargo ship movements. J R Soc Interface 7:1093–1103PubMedCrossRefGoogle Scholar
  37. Keeling MJ, Gilligan CA (2000) Metapopulation dynamics of bubonic plague. Nature 407:903–906PubMedCrossRefGoogle Scholar
  38. Kimura M (1953) “Stepping-stone” model of population. Annu Rep Natl Inst Genet Jpn 3:62–63Google Scholar
  39. Kolbe JJ, Glor RE, Schettino LR, Lara AC, Larson A, Losos JB (2004) Genetic variation increases during biological invasion by a Cuban lizard. Nature 431:177–181PubMedCrossRefGoogle Scholar
  40. Kolbe JJ, Larson A, Losos JB, de Queiroz K (2008) Admixture determines genetic diversity and population differentiation in the biological invasion of a lizard species. Biol Lett 4:434–437PubMedCrossRefGoogle Scholar
  41. Lack JB, Greene DU, Conroy CJ, Hamilton MJ, Braun JK, Mares MA, Van Den Bussche RA (2012) Invasion facilitates hybridization with introgression with introgression in the Rattus rattus species complex. Mol Ecol 21:3545–3561Google Scholar
  42. Landon DB (1991) Zooarchaeology and urban foodways: a case study from eastern Massachusetts. Unpublished Dissertation, Boston UniversityGoogle Scholar
  43. Le Roux J, Wieczorek AM (2009) Molecular systematics and population genetics of biological invasions: towards a better understanding of invasive species management. Ann Appl Biol 154:1–17CrossRefGoogle Scholar
  44. Lee CE (2002) Evolutionary genetics of invasive species. Trends Ecol Evol 17:386–391CrossRefGoogle Scholar
  45. MacGillivray W (1838) A history of British quadrupeds. WH Lizars, Edinburgh, ScotlandGoogle Scholar
  46. Maddison DR, Maddison WP (2000) MacClade 4: analysis of phylogeny and character evolution. Sinauer Associates, Inc., Sunderland, MassachusettsGoogle Scholar
  47. Mantel N (1967) The detection of disease clustering and a generalized regression approach. Cancer Res 27:209–220PubMedGoogle Scholar
  48. Martel C, Viard F, Bourguet D, Garcia-Meunier P (2004) Invasion by the marine gastropod Ocinebrellus orinata in France I. Scenario for the source of introduction. J Exp Mar Biol Ecol 305:155–170CrossRefGoogle Scholar
  49. Matisoo-Smith E, Robins JH (2004) Origins and dispersals of Pacific people: evidence from mtDNA phylogenies of the Pacific rat. Proc Natl Acad Sci USA 101:9167–9172PubMedCrossRefGoogle Scholar
  50. Matisoo-Smith E, Roberts RM, Irwin GJ, Allen JS, Penny D, Lambert DM (1998) Patterns of prehistoric human mobility in Polynesia indicated by mtDNA from the Pacific rat. Proc Natl Acad Sci USA 95:15145–15150PubMedCrossRefGoogle Scholar
  51. Meerburg BG, Singleton GR, Kijlstra A (2009) Rodent-borne diseases and their risks for public health. Crit Rev Microbiol 35:221–270PubMedCrossRefGoogle Scholar
  52. Musser GG, Carleton MD (2005) Family Muridae. In: Wilson DE, Reeder DM (eds) Mammal species of the world, a taxonomic and geographic reference. Smithsonian Institute Press, Washington, DC, pp 1189–1531Google Scholar
  53. Novak SJ, Mack RN (2001) Tracing plant introduction and spread: genetic evidence from Bromus tectorum (cheatgrass). Bioscience 51:114–122CrossRefGoogle Scholar
  54. Nowak R (1999) Walker’s Mammals of the world, 6th edn. John’s Hopkins University Press, Baltimore, MarylandGoogle Scholar
  55. Peakall R, Smouse PE (2006) Genalex 6: genetic analysis in Excel. Population genetic software for teaching and research. Mol Ecol Notes 6:288–295CrossRefGoogle Scholar
  56. Pimentel D, Lach L, Zuniga R, Morrison D (2000) Environmental and economic costs of nonindigenous species in the United States. Bioscience 50:53–65CrossRefGoogle Scholar
  57. Prentis PJ, Wilson JRU, Dormontt EE, Richardson DM, Lowe AJ (2008) Adaptive evolution in invasive species. Trends Ecol Evol 13:288–294Google Scholar
  58. Pritchard JK, Stephens M, Donnelly P (2000) Inference of population structure using multilocus genotype data. Genetics 155:945–959PubMedGoogle Scholar
  59. Raymond M, Rousset F (1995) GENEPOP (version 1.2): population genetics software for exact tests and ecumenicism. J Heredity 86:248–249Google Scholar
  60. Robertson BC, Gimmell NJ (2004) Defining eradication units in pest control progammes. J Appl Ecol 41:1032–1041CrossRefGoogle Scholar
  61. Robins JH, Hingston M, Matisoo-Smith E, Ross HA (2007) Identifying Rattus species using mitochondrial DNA. Mol Ecol Notes 7:717–729CrossRefGoogle Scholar
  62. Rogers AR, Harpending H (1992) Population growth makes waves in the distribution of pairwise genetic differences. Mol Biol Evol 9:552–569PubMedGoogle Scholar
  63. Rosenberg NA (2004) Distruct: a program for the graphical display of population structure. Mol Ecol Notes 14:4299–4312Google Scholar
  64. Rozas J, Sanchez-DelBarrio JC, Messeguer X, Rozas R (2003) DnaSP. DNA polymorphism analyses by coalescent and other methods. Bioinformatics 19:2496–2497PubMedCrossRefGoogle Scholar
  65. Ruffina L, Vidal E (2010) Early colonization of Mediterranean islands by Rattus rattus: a review of zooarcheological data. Biol Invasions 12:2389–2394CrossRefGoogle Scholar
  66. Sakai AK, Allendorf FW, Holt JS et al (2001) The population biology of invasive species. Annu Rev Ecol Syst 32:305–332CrossRefGoogle Scholar
  67. Schwartz CW, Schwartz ER (1981) The wild mammals of Missouri. University of Missouri Press, Columbia, MOGoogle Scholar
  68. Silver J (1927) The introduction and spread of house rats in the United States. J Mamm 8:58–60Google Scholar
  69. Simberloff D (2009) We can eliminate invasions or live with them. Successful management projects. Biol Invasions 11:149–157CrossRefGoogle Scholar
  70. Skellman JG (1951) Random dispersal in theoretical populations. Biometrika 38:196–218Google Scholar
  71. Suarez AV, Holoway DA, Case TJ (2001) Patterns of spread in biological invasions dominated by long-distance jump dispersal: insights from Argentine ants. Proc Natl Acad Sci USA 98:1095–1100PubMedCrossRefGoogle Scholar
  72. Sullivan R (2004) Rats: observations on the history and habitat of the city’s most unwanted inhabitants. Bloomsbury Publishing, New York City, NYGoogle Scholar
  73. Tollenaere C, Brouat C, Duplantier JM et al (2010) Phylogeography of the introduced species Rattus rattus in the western Indian Ocean, with special emphasis on the colonization history of Madagascar. J Biogeogr 37:398–410CrossRefGoogle Scholar
  74. Twigg G (1975) The brown rat. David and Charles, LondonGoogle Scholar
  75. Urdaneta-Marquez L, Failloux AB (2010) Population genetic structure of Aedes aegypti, of the principle vector of dengue viruses. Infect Genet Evol 11:253–261PubMedCrossRefGoogle Scholar
  76. Weir BS (1996) Genetic data analysis II: methods for discrete population genetic data. Sinauer Associate, Inc., Sunderland, MAGoogle Scholar
  77. Wilson GA, Rannala B (2003) Bayesian inference of recent migration rates using multilocus genotypes. Genetics 163:1177–1191PubMedGoogle Scholar
  78. Wilson AB, Naish KA, Boulding EG (1999) Multiple dispersal strategies of the invasive quagga mussel (Dreissena bugensis) as revealed by microsatellite analysis. Can J Fish Aquat Species 56:2248–2261CrossRefGoogle Scholar
  79. Witmer GW, Boyd F, Hillis-Star Z (2007) The successful eradication of introduced roof rats (Rattus rattus) from Buck Island using diphacinone, followed by an irruption of house mice (Mus musculus). Wildl Res 34:108–115CrossRefGoogle Scholar
  80. Yom-Tov Y, Yom-Tov S, Moller H (1999) Competition, coexistence, and adaptation amongst rodent invaders to Pacific and New Zealand Islands. J Biogeography 26:947–958CrossRefGoogle Scholar

Copyright information

© Springer Science+Business Media Dordrecht 2012

Authors and Affiliations

  • Justin B. Lack
    • 1
    • 4
  • Meredith J. Hamilton
    • 1
  • Janet K. Braun
    • 2
  • Michael A. Mares
    • 2
    • 3
  • Ronald A. Van Den Bussche
    • 1
  1. 1.Department of ZoologyOklahoma State UniversityStillwaterUSA
  2. 2.Sam Noble MuseumUniversity of OklahomaNormanUSA
  3. 3.Department of ZoologyUniversity of OklahomaNormanUSA
  4. 4.Department of ZoologyOklahoma State UniversityStillwaterUSA

Personalised recommendations