Genetic characterization of Okinawan black rats showing coat color polymorphisms of white spotting and melanism

We examined pelage color variation in wild populations of black rats (the Rattus rattus species complex) in the Yambaru forest area, northern Okinawa Island, Ryukyu Archipelago, Japan. Our field study revealed that 8.7% (38/438) and 0.2% (4/2500) of rats exhibited two types of coat color: white spotting and melanism, respectively. Using 34 representative animals, the phylogeography of the population was inferred using a nuclear gene marker, i.e., sequences (954 bp) of the melanocortin-1 receptor (Mc1r) gene responsible for the melanistic form in black rats. Four sequences from Okinawa were characterized as R. tanezumi, the Asian strain of black rat. Notably, neither of the phenotypic characters of white spotting or melanism was associated with the Mc1r haplotypes. Analysis of mitochondrial cytochrome b (Cytb) sequences (1140 bp) revealed that four haplotypes recovered from Okinawa clustered with the clade of R. tanezumi and differed by one or more bases from haplotypes at other localities in Japan and Asian countries. Thus, both variants may have arisen in the native rat population of Okinawa without interaction with the lineage of R. rattus, which exhibits a worldwide distribution and displays such coat color variants. The Yambaru population of black rats has thus experienced its own evolutionary history in allopatry for a substantial period of time (e.g., 10,000 years), which has preserved valuable genetic polymorphisms and will be useful for assessing the ecological consequences of genetic variation in natural populations.


INTRODUCTION
Many animal species exhibit variation in pigmentation (Jones, 1923), which clearly plays important roles in camouflage, social communication, and sexual displays (e.g., Hoekstra et al., 2004;Brunberg et al., 2006).Variation in pigmentation is mediated not only by changes in the color itself, but also by changes in patterns, such as spotting and barring.Intensive studies using model organisms and domestic animals, such as mice, rats, and chickens, have improved our understanding of the genetic systems affecting pigmentation variation, such that hun-dreds of relevant genes have been identified (e.g., Kijas et al., 1998;Brunberg et al., 2006;Candille et al., 2007;Montoliu et al., 2009; see also Osawa, 2009, for review).Notably, many contrasting coat color phenotypes are caused by single mutations of single genes (e.g., Gratten et al., 2007).Nonetheless, studies assessing the ecological significance of color variation are rather limited, in part because polymorphic states of color variation in natural populations have been less intensively surveyed in certain nocturnal animals.A well-known example involves natural selection on coat color variation in pocket mice, in which coat color variation is correlated with soil color in their habitat (Hoekstra et al., 2004(Hoekstra et al., , 2005)).As this case exemplifies, rodent species have great potential as subjects for such studies because they occur in broad geographic regions, including insular domains, and use a wide variety of habitats.Rodents can thus provide opportunities to test various ecological factors that shape coat color, such as predation.
Black rats (the Rattus rattus species complex), which are globally distributed because of their capacity for overseas colonization (Patton et al., 1975;Aplin et al., 2003aAplin et al., , 2003bAplin et al., , 2011;;Robins et al., 2007;Pagès et al., 2010), are good subjects for examining coat color evolution.Their close phylogenetic relationship with the well-known model animal R. norvegicus is also advantageous for identifying genes responsible for characters of interest.In fact, black rats display polymorphic states in coat color.European populations, for example, show color variation in both dorsal (agouti and melanistic) and ventral (white, gray, melanistic) areas (e.g., Ondrias, 1966).Melanocortin-1 receptor ( Mc1r) and agouti signaling protein (Asip) (formerly genes E and A, respectively) are known to cause the melanistic form in a dominant and recessive manner, respectively (Feldman, 1926;Tomich and Kami, 1966).Both genes reportedly affect coat color changes in a variety of mammalian species (e.g., Vage et al., 1997;Eizirik et al., 2003;Hosoda et al., 2005;Kingsley et al., 2009).In a previous study of rats from the Otaru port on Hokkaido, where the dorsal agouti and black polymorphism exist, we found that the Otaru rats have the dominant system of melanism in which a single-nucleotide polymorphism (SNP) of Mc1r, i.e., a change from G to A at site 280 that leads to the replacement of glutamic acid with lysine in the amino acid sequence, causes the melanism (Kambe et al., 2011).To date, the molecular details related to the recessive system of melanism in black rats have not yet been determined.
Because black rats have one of the greatest negative impacts on nature and humans, great efforts have been made to survey and remove this species from a variety of locations, including ports, airports, natural forests, and insular areas in need of conservation from this introduced animal.This situation thus provides a good opportunity for a large-scale survey of phenotypic variation, including coat color, in this species.In fact, Yabe (2010) documented the occurrence of a melanistic form of rat with white spots on its head, neck, and tail tip on a small Ogasawaran island (Mukojima), where an eradication program had once successfully removed black rats entirely (Hashimoto, 2010).Meanwhile, melanistic black rats (with white necks) were observed in the natural forests (Yambaru) of Okinawa Island, Ryukyu Archipelago, Japan, in 2010 and 2011 (K.Nakata, personal observation).These observations were made during activities related to introduced species' removal (including the mongoose) from the northern part of Okinawa where Yambaru forests have been preserved.
Given the importance of controlling black rat popula-tions in island ecosystems, it is necessary to assess the phylogenetic background of Okinawan rats and the genetic nature of coat color variants to better understand the ecological significance of such variation.Until recently, mitochondrial cytochrome b (Cytb) gene sequences (e.g., Chinen et al., 2005;Robins et al., 2007;Pagès et al., 2010;Aplin et al., 2011) have been used for phylogenetic inference in the R. rattus species complex.Such analyses have revealed the presence of the two most well-known groups (Musser and Carleton, 2005), R. rattus (Lineage I) and R. tanezumi (Lineage II), in addition to four other groups (Lineages III-VI) in Southeast Asia (Pagès et al., 2010;Aplin et al., 2011).The Asian black rat (R. tanezumi) is distributed in Southeast and East Asia and likely evolved there (Yosida, 1980;Aplin et al., 2011), whereas R. rattus originates from the Indian subcontinent and has now spread to the rest of the world.R. tanezumi invaded the Japanese Islands during prehistoric times (Kowalski and Hasegawa, 1976;Kawamura, 1989).Likewise, R. rattus dispersed to Europe in prehistoric times and has spread from there through global exploration and trade (Yosida, 1980;Aplin et al., 2011).R. rattus has been introduced more recently to Japan and has hybridized with the native lineage of R. tanezumi.The introduction of genetic components of R. rattus has been inferred from cytogenetic and molecular analyses (Suzuki et al., 2001;Chinen et al., 2005;Kambe et al., 2011).Such introductions have occurred throughout Japan in a variety of environments, including ports (Otaru and Kagoshima), large towns (Tokyo), and insular regions (the Ogasawara Islands).The Mc1r sequences of six individuals from Okinawa have indicated that they are R. tanezumi (Kambe et al., 2011); however, intensive studies assessing the phylogenetic background of Okinawan rats has not yet been conducted.
In the present study, we conducted a large-scale survey to determine the frequencies of the coat color variants of white spotting and melanism in Yambaru forests.Using nuclear and mitochondrial markers of the Mc1r (954 bp) and Cytb (1140 bp) gene sequences, we then examined the phylogenetic background of the Okinawan black rat population to potentially reveal imprints of the introduced lineage of the black rat, R. rattus.In addition, we addressed whether Mc1r is responsible for the coat color variants observed in the Yambaru forests, particularly the melanistic form.Such assessment of the mode of introduction of the exotic R. rattus lineage into natural forests will provide useful information for conservation efforts on Okinawa Island, where the issue of introduced species, including R. rattus, remains a serious problem.

Study area and samples
The study area was located in near-climax forests on the northern part of Okinawa White spotting in Okinawan black rats Island, locally called Yambaru.In the forests, evergreen oak (Castanopsis sieboldii) dominates the tree community (Ito, 1997), and endangered native species with ancient lineages, including mammals such as the Okinawan spiny rat (Tokudaia muenninki) and the Ryukyu long-haired rat (Diplothrix legata), can be found (Suzuki et al., 2000;Endo and Tsuchiya, 2006;Suzuki, 2009;Murata et al., 2011).The ecology of the Yambaru forest is threatened by intensive colonization by an introduced animal, the small Indian mongoose (Herpestes auropunctatus), which preys on various native mammals, ground-nesting birds, amphibians, and reptiles (Yamada, 2002;Yamada and Sugimura, 2004;Yamada et al., 2010).
In the northern part of Okinawa Island, the Java Mongoose Capturing Project has been conducted by the Ministry of the Environment, Japan and the Okinawa Prefecture Government, for a decade, and traps have been set to remove mongooses and black rats (Ministry of the Environment, 2009; Fig. 1).The 438 black rats used in this study were captured in the Yambaru area in February 2009 (n = 11; exact locality is unknown) and April to August 2010 (n = 427).The colors of their coats were examined, and the presence or absence of white spotting on the dorsal side of each rat was determined by visual inspection.The smallest area of spotting detected was 5 mm in the longest axis.Regardless of white spotting, all of the black rats examined had a pelage of agouti back and a gray venter, as is typical in the Asian black rat, R. tanezumi.None of the rats had white-tipped tails.Representative specimens of rats with (n = 21) and without (i.e., wild-type, n = 16) white spotting were kept in the laboratory and subjected to further analysis (Table 1).Flat-skin specimens (n = 29,(141)(142)(143)(144)(145)(146)(147)(148)(202)(203)(204)(205)(206)(207)(208)(209)(210)(211) were prepared and examined to determine the size and shape of areas with white spotting.Representative individuals were subjected to gene sequence analysis.Mc1r sequences were determined for 11 specimens captured in 2009, 6 of which had been previously analyzed (Kambe et al., 2011).
The melanistic rat was first observed in the Yambaru forest of Okinawa in June 2010.During a 1-year study period from July 2010 to June 2011, we found the melanistic form in four individuals among ca.2500 captured rats.These four individuals were used for further molecular analyses.For comparison, a flat-skin specimen (HTS230) from Mukojima Island, Ogasawara Islands, with background coloration of melanism and white spotting on the head, tail tip, and ventral portion (Yabe, 2010), was subjected to DNA extraction and sequence determination for the Mc1r gene sequence.
To address phylogenetic position of the Okinawan rats, we newly determined the Cytb sequences in specimens from Chitose (HS4527), Niigata (HS4117), and Osaka (HS4216) in addition to 30 individuals from Okinawa Island (Table 1).
Polymorphic Mc1r sequences with more than one heterozygous site in heterozygous individuals were resolved using SNP-specific sequence primers (14 mers), as described previously (Kambe et al., 2011), to separate haplotype sequences.

Appearance of white spotting and melanism
Among the 438 individuals examined, 38 black rats (8.7%) from Okinawa had white spotting on their head and/or neck.Rats with spotting were captured throughout the study area (Fig. 1).Flat-skin specimens were examined to determine the size and shape of areas with white spotting (n = 29; HTS125-135, HTS141-148, HTS202-211; Table 1).The size, shape, number, and position (neck or head) of white spotting varied among individuals (Fig. 2, for example).Most of the white spotting appeared on the neck (n = 19).The size of white spotting patches varied from 5 mm to 5 cm in diameter (Fig. 2), and the shapes of spotting patches were irregular.Larger spotting sometimes accompanied small spotting.Spotting on the neck could be classified into four types, taking into account the length of the long axis of the largest patch: small (< 1 cm, type a; n = 5), middle (1-3 cm, type b; n = 10), large (> 3 cm, type c; n = 3), and extended (d; n = 1).The type d individual had an extended area of white hairs that produced a patch that looked like a white collar.Two individuals had white spotting on their forehead (type e).Age (or body mass) or gender did not appear to affect the appearance of white spotting in rats.
In the survey for melanism, one of the melanistic rats was collected from PB3 and the other three were found in Btk (Fig. 1).Mc1r sequences The nucleotide sequences of the entire coding of Mc1r (954 bp) was compared among 34 rats from Okinawa: 16 wild-type and 18 white-spotted individuals (Table 1).Together with 11 haplotype sequences (hap 1-3 of R. tanezumi and hap 4-11 of R. rattus; Kambe et al., 2011), phylogenetic analyses revealed a total of 12 haplotypes in this data set (Fig. 3).Four haplotypes (hap 1, 2, 3, and 12) were recovered from Okinawa.Hap 12 (accession number AB62660) was first found in this study, as was an SNP from site 212 (Fig. 3).Sequences from three individuals with this SNP were subjected to haplotype separation using two sequence primers specific to the SNP from site 212.A statistical parsimony network of the 12 haplotypes, in conjunction with the 4 haplotypes from Okinawa and reference sequences for the two species, R. rattus and R. tanezumi (Kambe et al., 2011), produced two well-distinguished clusters representing each of the two species (Fig. 3A).All of the haplotypes from Okinawa fell into the R. tanezumi cluster.
The individual black rat harboring both phenotypic variants of melanism and white spotting from the Ogasawara Islands (Yabe, 2010) was homozygous for hap 5, representing the dominant allele A at site 280, which Fig. 3.A network of the 12 Mc1r gene haplotypes (A).Numbers along branches indicate nucleotide positions.Two distinct haplotypes represent the two species, R. tanezumi and R. rattus (Kambe et al., 2011).The four haplotypes with asterisks (1, 2, 3, and 12) were detected in the current data set with 34 rats from the Yambaru forest, and the sizes of circles in the network reflect haplotype frequency.Sequence comparison for haplotypes of Mc1r (954 bp) detected in black rats from Okinawa (B).Representative sequences for the two species, R. tanezumi (haplotypes 1-3, 12) and R. rattus (haplotypes 5 and 10), are used for the comparison (Kambe et al., 2011).Only variable sites are shown, with predicted ancestral nucleotides.Nonsynonymous sites are shown in bold letters.Fig. 4. Neighbor-joining tree based on Cytb gene sequences (1140 bp) from Okinawan black rats, with representative sequences of the R. rattus species complex (see Aplin et al., 2011), with R. argentiventer and R. norvegicus included as an outgroup (A).Numbers at nodes denote bootstrap values (> 50%).The species complex has six distinct mitochondrial sequences (Lineages I-VI) including those representing R. rattus (Lineage I) and R. tanezumi (Lineage II).A network of the five Cytb haplotypes recovered from Japan and Vietnam (members of the sublineage IIa, Aplin et al., 2011) was constructed (B).Numbers along branches indicate nucleotide positions.The four haplotypes with asterisks (IIa-1, IIa-3, IIa-4, and IIa-5) were detected in the current data set with 30 rats from the Yambaru forest, and the sizes of circles in the network reflect haplotype frequency.White spotting in Okinawan black rats is responsible for the melanism (Kambe et al., 2011).The four rats with melanism from Okinawa, however, were found to have the R. tanezumi haplotypes (haps 1 and 2; Table 1), and the allele at site 280 was G and not A.

Cytb sequences
We recovered four Cytb haplotypes, designated IIa-1, IIa-3, IIa-4 (accession number AB689861) and IIa-5 (AB689862), from the Okinawan rats (n = 30, Table 1) and two haplotypes from the other localities of Japan; IIa-1 from Niigata and Osaka, and IIa-6 (AB689863) from Chitose.NJ tree construction together with haplotypes representing the six lineages of the R. rattus species complex (Lineages I-VI; Aplin et al., 2011) and those from Japan (present study; Chinen et al., 2005) revealed that the four haplotypes from Okinawa belonged to the cluster of R. tanezumi (Lineage II; Fig. 4A), further exhibiting integration of them into a subclade ("IIa", see Aplin et al., 2011) accompanied by haplotypes from Japan and Vietnam (IIa-1, IIa-2, IIa-6) with a high supporting value (100%).The predominant haplotype in Okinawa was IIa-3 (80%, Table 1), which differed from the other Okinawan haplotypes at one site (Fig. 4B), in contrast to the fact that IIa-3 has never been recovered from other localities of Japan, including Hokkaido (Chitose), Honshu (Niigata, Tokyo, Osaka), Kyushu (Shibushi, Kagoshima), and Ogasawara Islands.The four haplotypes from Okinawa, except IIa-1, are likely unique to Okinawa or the Ryukyu Archipelago, given that the haplotype IIa-3 has been recovered from Amamioshima Island, the Ryukyu Archipelago (Robins et al., 2007).

DISCUSSION
Phylogenetic nature of the Okinawan rats From the current phylogenetic characterization using Mc1r and Cytb sequences (Figs. 3 and 4), the Okinawa population appears to be purely R. tanezumi and is free from introgression of R. rattus.These results contradict our initial presumption that the polymorphic state of coat color variation stemmed for multiple colonization events with different genetic backgrounds, as has been reported from the Galapagos Islands (Patton et al., 1975).Genetic admixture of the widely distributed lineage of R. rattus and the Asian lineage of R. tanezumi is common throughout Japan, e.g., in Otaru (Hokkaido), Tokyo (Honshu), Kagoshima (Kyushu), and the Ogasawara Islands (Chinen et al., 2005;Kambe et al., 2011).Thus, the black rat population of the Yambaru forests may harbor certain features that have prevented the introduction of the lineage of R. rattus.
These observations echo those of studies of the house mouse M. musculus, in that new individuals arriving to an established population are generally unable to survive or gain mates and consequently do not contribute to the population's gene pool (Gabriel et al., 2010;Hardouin et al., 2010;Jones et al., 2011).Thus, it is reasonable that Okinawan black rats have been preserved with a substantial population size for a long evolutionary time, perhaps due to specific habitat conditions of the Yambaru forests, such as the subtropical climate, abundant natural resources, and lack of efficient predators.
To support this view, our data for Mc1r and Cytb exhibit unique genetic components.In the Mc1r analysis, the examination of 36 individuals from Okinawa resulted in the identification of four haplotypes with six variable sites, two (hap 3 and 12) of which were likely to be Okinawa-specific.Accordingly, after examining available Cytb sequences in the literature (Chinen et al., 2005;Robins et al., 2007;Aplin et al., 2011), we conclude that the three Cytb haplotypes (IIa-3, IIa-4 and IIa-5) from the Okinawa black rats are also likely unique to the islands and perhaps adjacent islands of the Ryukyu Archipelago.These results indicate that the Okinawan population has retained a certain level of genetic variation, perhaps reflecting newly arisen mutations in addition to mutations within the genetically diverse initial lineage of R. tanezumi that colonized Okinawa from somewhere on the Asian continent several thousands of years ago, most likely during Neolithic times (Kowalski and Hasegawa, 1976;Kawamura, 1989).Future phylogeographic studies using finer-resolution markers and a sufficient number of samples from adjacent areas are necessary to support these predictions.
Evolutionary implications of white spotting and melanism Through a large-scale survey of coat color variation in hundreds to thousands of rats, we demonstrated that pelage variation exists in natural black rat populations on Okinawa Island.Because the phenotypes of white spotting and melanism have never been reported as features of genetic variation in black rats from East Asia, except for introduced rats such as those on the Ogasawara Islands (Yabe, 2010) where R. rattusspecific sequences were recovered (Mc1r for Fig. 3; Cytb and Irbp for Chinen et al., 2005), it is reasonable to presume that both of the mutations responsible for white spotting and melanism emerged on Okinawa Island.
Accounting for the frequency of phenotype observation (ca.9%) and the rather broad range of the spatial distribution of the Yambaru forests, we conclude that the black rat population harbors the white spotting variant as a polymorphism.If we assume the melanism observed in Okinawa (4 of 2500 rats) is a recessive trait, the frequency of the responsible, as-yet-unidentified allele can be estimated to be 4%, suggesting some spread of the mutation in the Yambaru population.These polymorphic states indicate that Okinawan rats provide an opportunity to assess the ecological meaning of the emergence of both the melanistic form and white spotting under natural conditions.
The question then arises of why Okinawan black rats exhibit such peculiar variation in coat color.Aberrant coloration is conspicuous and generally considered deleterious for survival, particularly for small mammals.The genes involved in pigmentation variation likely evolved under purifying selection in small rodents, as exemplified in Mc1r sequences from Mus and Rattus (Shimada et al., 2009;Kambe et al., 2011).In this context, it is surprising to see such peculiar coat color variants of white spotting and melanism in Okinawa over a short study period.One possible explanation is that these coat color variants are maintained by sexual selection, as is predicted for the white-spotted tail of barn swallows Hirundo rustica (Kose and Moller, 1999), as no efficient predators exist on Okinawa and constraints on pelage changes would be rather weak.Alternatively, the peculiar coat variants may be solely explained by the fact that we conducted such a large-scale survey on the island, where thousands of rats were trapped annually within the entire area of the Yambaru forests.Such coat color variation may be evolving in a neutral manner and may be found in any population of any species when carefully and intensively examined (e.g., Iwasa, 2004;Ishiguro et al., 2008).For example, Japanese wood mice sometime exhibit coat color variants, such as small white spottings and tails with white tips (K.Tsuchiya, personal communication).Future large-scale surveys should assess which ecological and genetic factors affect the appearance of coat color variants in a variety of species.

Possible responsible genes for coat color variants
We sought the causative genes for white spotting and melanism in Okinawan black rats and thus examined nucleotide variation in Mc1r, which is related to coat color variation.Four SNPs were found in the coding region of Mc1r in the Okinawan rats; one SNP was a synonymous mutation, whereas the other three were nonsynonymous (Fig. 3).However, no recovered SNPs were associated with the polymorphism of both white spotting and melanism (Table 1), indicating that genes other than Mc1r are responsible for the coat color variants.
White spotting tends to be confined to the head and neck.Mutations at a number of loci, such as the Kit and Kitl genes, are known to give rise to white spotting of the coat (e.g., Bennett and Lamoreux, 2003;Baxter et al., 2004;Osawa, 2009).For example, mutations of the Kit gene are shown to cause a failure in migration of melanoblasts, the precursors to melanocytes, and thus result in the white spotting phenomena in R. norvegicus.The Kit gene encodes a receptor tyrosine kinase protein.Individuals with a heterozygous mutant Kit gene locus (Ws/+) show ventral white spotting, and homozygous rats for the mutated Kit (Ws/Ws) show severe anemia and mast cell depletion (Niwa et al., 1991).
For the melanistic phenotype, given that Mc1r is not the causative gene, the other obvious candidate gene is Asip, which causes the melanistic form in a recessive manner in variety of animals, including black rats (Feldman, 1926;Tomich and Kami, 1966).Finally, we would like to emphasize that such rare phenotypic mutations arising in Okinawan rats might indicate that Okinawa provides a unique opportunity to study the evolution of the phenotypes and genes responsible for coat color variation.Hopefully, these findings will stimulate studies of intraspecies and interspecies coat color variation in Rattus and highlight the evolutionary implications of variation in pigmentation.

Fig. 1 .
Fig. 1.Collection points of black rats from the Japanese Islands (A) and study area in the northern part of Okinawa Island (locally called Yambaru), Ryukyu Archipelago (B).On the map of Yambaru, green circles (sometimes appearing as solid lines) indicate trapping sites, and black solid circles indicate locations where black rats were captured.Stars and closed square symbols indicate sites where rats with white spotting and melanism were captured, respectively.The dotted line indicates the southern border of an area where anti-mongoose nets have been set as part of the Java Mongoose Capturing Project conducted by the Ministry of the Environment and the Okinawa Prefecture Government.

Fig. 2 .
Fig. 2. Rats from Okinawa, showing variant coat colors of white spotting (A) and melanism (B).White spotting patterns can be divided into five types in terms of size and position.Spotting on the neck is classified into four types according to size: a single white spot (the largest on the individual) less than 1 cm (a), 1-3 cm (b), and > 3 cm (c) in the long axis, and a white stripe around the neck (d).The fifth type is that on the head (e).A wildtype phenotype without white spotting is also shown (Wt).Arrows indicate the position of white spotting.