Subgenus Mus. Schwarz and Schwarz (1943) provided a revision that was followed with minor changes by Ellerman and Morrison-Scott (1951). This arrangement was criticized by Jones and Johnson (1965:394) who found that specimens from Asia that they studied "bear little or no relation to this idealized classification." Subsequent treatments of this group were presented by J. T. Marshall, Jr. (1977b, 1981, 1986, 1998) and J. T. Marshall, Jr. and Sage (1981). The most recent classification combined biochemical analyses of European, Asian, and African mice (Bonhomme et al., 1984; Boursot et al., 1993, 1996; Prager et al., 1998). The translation of these results, as well as incorporation of morphological data, into a new classification of Mus musculus and its allies (J. T. Marshall, Jr, 1998), and the allocation of the many names to M. musculus, are surprisingly concordant with the treatment of Schwarz and Schwarz (1943). Some of the scientific names listed under M. musculus by those authors, but now placed elsewhere, are here associated with M. spicilegus, M. spretus, M. macedonicus, and M. caroli (see those accounts).

There are three distinctive and well known groups of house mice (Bonhomme et al., 1994; Boursot et al., 1993, 1996; Din et al., 1996; J. T. Marshall, Jr., 1998; Sage et al., 1993; Yonekawa et al., 1994), a fourth not as well documented (Din et al., 1996), and a fifth recently identified (Prager et al., 1996). Among Mus researchers, the three usual complexes (castaneus, domesticus, and musculus) are either recognized as species (J. T. Marshall, Jr., 1998; J. T. Marshall, Jr. and Sage, 1981; Prager et al., 1998; Sage et al., 1993), or subspecies as we list them here (see discussions in Boursot et al., 1993, 1996; Yonekawa et al., 1994; Din et al., 1996; Prager et al., 1998); the fourth group (bactrianus) is generally treated as a subspecies (Din et al., 1996); the fifth (gentilulus) is considered to be a possible species (Prager et al., 1998). Analyzing mtDNA by restriction enzymes and sequencing in worldwide samples of M. musculus led Yonekawa et al. (1994) to recognize the same three subspecies described above and a fourth, M. m. bactrianus, in N India, Pakistan, and Afghanistan. They assigned subspecific rather than specific status to these four geographic entities "because they have not been sexually isolated from each other under natural conditions" (p. 34). Presence of four identifiable geographic groups connected by significant gene flow among them is also indicated by analysis of variation of ribosomal RNA (Suzuki and Kurihara, 1994). Whether viewed as species or subspecies, each group is characterized by distinctive genetic and morphological traits. In their phylogenetic analyses of sequences from the mtDNA control region, Prager et al. (1998) identified four major lineages with M. m. domesticus as sister group to M. m. castaneus, M. m. musculus, and M. m. gentilulus, and the latter being sister to M. m. castaneus and M. m. musculus.

Mus musculus musculus ranges from C Europe and Scandinavia through E Europe, Ukraine, Turkmenistan, SW Georgia, NC Iran, through N Afghanistan (north of the Hindu Kush) and N Asia to Manchuria, Korea, and Japan (Prager et al., 1998). Introduced pockets of this form occur within the ranges of the other two groups (in the Nile delta, for example; J. T. Marshall, Jr., 1998).

Mus musculus domesticus occurs in N Africa and ranges in Eurasia from W Europe (and most Mediterranean islands) through S Eurasia to the Caucasus and eastward through Iran, Afghanistan and Pakistan to N India and Nepal. It is this form that was spread inadvertently by European colonization throughout the New World, numerous islands, Australia, and probably southern Africa (see Meester et al., 1986, for discussion about the possible subspecies in southern Africa). Auffray et al. (1988) hypothesized the origin of commensalism in M. m. domesticus to have taken place in the Fertile Crescent. Prager et al. (1998) identified samples from Afghanistan as musculus and castaneus, and specimens from Pakistan, N India, and Nepal as castaneus; the actual range of domesticus in these regions requires clarification.

Mus musculus castaneus extends from Central Asia (C Afghanistan south of Hindu Kush, Pakistan, peninsular and NE India, E and C Nepal) to throughout SE Asia; it has spread to Taiwan and eastward through the Moluccas to New Guinea; also on the Mariana Isls (J. T. Marshall, Jr., 1998; Prager et al., 1998). Populations on Japan (traditionally referred to as molossinus) consist of either musculus, castaneus, or hybrids between the two (Yonekawa et al., 1988).

The fourth and lesser known group, M. m. bactrianus from Afghanistan, was first identified by Yonekawa et al. (1981) and Bonhomme et al. (1984) using mitochondrial and nuclear gene sequences. In their analysis of nuclear gene sequences of several samples from the Indian subregion, Din et al. (1996:535) would restrict the range from that proposed by Yonekawa et al. (1994), writing that "M. m. bactrianus defined on the basis of the different mitochondrial. . . and nuclear genes. . . from the region of Kabul in Afghanistan. . . has a very distinct nuclear gene composition. . . and probably belongs to a geographically confined group of populations living in the valleys of Afghanistan where contact with the neighbouring regions is limited by the high mountain ranges which surround the country" and "can no longer be considered to be representative of the other Mus musculus populations from the central area of the species range [Pakistan and peninsular India]." Prager et al. (1998:835) noted that M. m. bactrianus is "the least well defined and characterized" of the five groups "and it is not known whether it is a cohesive genetic entity." Kandahar in SE Afghanistan is type locality of bactrianus.

The fifth group, M. m. gentilulus, was recently identified by analyses of sequences from the mtDNA control region (mtDNA) in a sample from Yemen; the mice are also distinctive in their small body size (Prager et al., 1998). This form may have a more expansive distribution and Prager et al. (1998:856) noted the desirability of genetically sampling populations from elsewhere on the Arabian Peninsula, "all along the northern shores of the Persian Gulf and the Gulf of Oman, and also the Horn of Africa and adjacent regions." Similar mitochondrial sequences and Y chromosomal traits that characterize M. m. gentilulus have recently been discovered in samples of M. musculus from Madagascar (Duplantier et al., 2002). Those authors suggest this kinship supports a hypothesis of colinization by human importation from the Arabian Peninsula to islands along East Africa, including Madagascar, a colonization route also claimed for the shrew Suncus murinus (Hutterer and Tranier, 1990).

Mus m. musculus and M. m. domesticus meet and hybridize through a narrow zone (less than 50 km) in C Europe (from the Jutland Peninsula to Bulgaria), which has been the focus of extensive genetic, parasitological, and some morphological and behavioral studies (Agulnik et al., 1993; Alibert et al., 1994; Auffray et al., 1996; Britton-Davidian et al., 2002; Garagna et al., 1992; Lazarová, 1999; Macholán and Zima, 1994; Macholán et al., 2003; Moulia et al., 1991; Munclinger et al., 2002; Niwa-Kawakita, 1994; Prager et al, 1993; Sage et al., 1986b; Smadja and Ganem, 2002; Turker et al., 1993; also see the reviews and citations referenced by Boursot et al. [1993], J. T. Marshall, Jr. [1998], Mezhzherin et al. [1998], and Sage et al. [1993]; Mitchell-Jones et al. [1999] provided vivid depictions of the European ranges of domesticus and musculus), and in a much wider zone (more than 300 km) in the Transcaucasus region (Mezhzherin et al., 1998; Orth et al., 1996). The hybrid zone between M. m. domesticus and M. m. castaneus may be in NW India (J. T. Marshall, Jr., 1998), but that contact requires critical analysis (Agrawal [2000:121] listed the Indian populations of M. m. domesticus, which he referred to as praetextus and homourus, as "outdoor subspecies" and M. m. castaneus as an "indoor subspecies"). In their analysis of nuclear gene sequences, Din et al. (1996) documented samples from Tehran in N Iran as having a nuclear gene pool containing elements of domesticus and mice from N India and Pakistan, which strengthens the hypothesis of a series of intergrading populations between N India and the periphery of the domesticus range in Europe. Mus m. musculus and M. m. castaneus contact each other in a wide zone of intergradation in N China (more than 1000 km in latitude), in the main Japanese Isls, and on Okinawa (Bonhomme et al., 1989; Boursot et al., 1993, and references cited therein; J. T. Marshall, Jr., 1998; Mezhzherin et al., 1998; Yonekawa et al., 1994).

Samples of M. musculus have been the focus of numerous morphometric, chromosomal, and molecular studies undertaken in context of phylogenetic inquiries bearing upon discrimination and relationships among populations of M. musculus and between it and other species of Mus as well as general treatises on speciation and phylogeny. Some of the contributions published during the last two decades that also summarized and cited earlier reports are Evans (1981), Martin et al. (2000), Foster et al. (1981), Capanna (1982), Bonhomme et al. (1984), Potter et al. (1986), Giagia et al. (1987), Hubner (1988), Nishioka (1987), Winking et al., (1988), Britton-Davidian (1990), Corti and Ciabatti (1990), Garagna et al. (2002), Gerasimov et al. (1990), She et al. (1990), Capanna and Corti (1991), Searle (1991), Viroux and Bauchau (1992), Karn and Dlouhy (1991), Bush and Paigen (1992), Fraguedakis-Tsolis (1992), Mathias and Ramalhino (1992), Mezhzherin and Kotenkova (1992), Ritte et al. (1992), Scriven and Bauchau (1992), Ganem (1993), Garagna et al. (1993), Hauffe and Searle (1993), Korobitsyna et al. (1993), Said et al. (1993), Searle et al. (1993), Capanna and Redi (1994), Capanna et al. (1994), Giménez and Bidau (1994), Miyashita et al. (1994), Nagamine et al. (1994), Ohtsuka et al. (1996), Tokumitsu and Ogawa (1994), Chondropoulos et al. (1996), Ganem et al. (1996), Macholán (1996c), Naruse et al. (1996), Smets et al. (1996), Garagna et al. (1997), Said et al. (1993, 1999), Britton-Davidian et al. (2000), Gündüz et al., 2000; Huang et al. (2001), Leamy et al., (2002), Karn et al. (2002), and Wallace et al. (2002).

A highlight among the many chromosomal studies is the exceptionally variable karyotype characteristic of M. m. domesticus throughout its range compared with other subspecies, especially M. m. musculus (Britton-Davidian et al., 2000; Capanna, 1982; Capanna et al., 1994; Chondropoulos et al., 1996; Hauffe and Searle, 1993; Muñoz-Muñoz et al., 2003; Said et al., 1993, 1999; Nachman and Searle, 1995, provided an excellent synopsis). In addition to the standard karyotype of all acrocentric chromosomes (2n = 40), numerous populations of domesticus exhibit metacentric chromosomes derived from Robertsonian fusion mutations. Such mutations are absent from most populations of M. m. musculus, and when present occur at very low frequencies (Zima and Macholán, 1989; Zima et al., 1990).

Recently, initial sequencing and comparative analysis of the mouse genome has been documented using laboratory strain C57BL/6J, and revealing about 30,000 genes, with over 90% of mouse and human genomes "reflecting segments in which the gene order in the most recent common ancestor has been conserved in both species" (Mouse Genome Sequencing Consortium, 2002).

Analysis of DNA sequences from a set of genes places M. musculus in the same clade as M. spretus, M. macedonicus, and M. spicilegus (Graur, 1994; Larizza et al., 2002; Lundrigan et al., 2002; Martin et al., 2000; Prager et al., 1996, 1998). Combined analyses of morphological traits, DNA/DNA hybridization, and mitochondrial 12S rRNA sequences align M. musculus with M. spicilegus, M. macedonicus, and M. spretus in a clade separate from an Asian clade composed of M. caroli, M. cervicolor, and M. cooki (Auffrey et al., 2003; Chevret et al., 2003; also supported by cranial morphology, J. T. Marshall, Jr., 1977b, and in litt., 2004). Osteometric comparisons between subfossil remains from Lavezzi Isl off the coast of S Corsica and recent Corsican populations reported by Vigne et al. (1993). Analyses of mtDNA variation among samples from Japanese islands reveal a polyphyletic origin of Japanese M. musculus derived from musculus, castaneus, and domesticus strains (Bonhomme et al., 1989; Yonekawa et al., 1994; also confirmed by specimens in USNM and original descriptions, J. T. Marshall, Jr., in litt.). These results are also supported by a morphometric study of insular variation of M. musculus from various Japanese islands (Takada et al., 1994). Morphometric variation in samples from the Japanese Izu Isls south of Tokyo Bay documented by Takada et al. (1999), and their relationship to populations on the Ogasawara Isls elucidated by Takada et al. (2002). Morphometric and immunological relationships among samples from Greek Isls in the Ionian and Aegean seas indicate that all Greek mainland and insular samples are M. musculus domesticus, the same subspecies that occurs in W Europe, North Africa, Turkey, and eastward (Chondropoulos et al., 1995). A taxonomic assessment of house mice from the Ukraine, Moldavia, and European sector of Russia by Zagorodnyuk (1996b) reinforced the specific status of M. spicilegus (including sergii and makovensis) and recognized the following subspecies of Mus musculus: musculus (which includes hapsaliensis, polonicus, borealis, and funerus), formosovi (includes tataricus), wagneri (including nogaiorum, bicolor, wagneri, and sareptanicus), and hortulanus (includes nordmanni and arenarius). Geographical distribution of hemoglobin Hbb haplotypes in samples from E Asia and correlation with subspecies defined by morphology discussed by Kawashima et al. (1995). A comprehensive summary of genetic map for Mus musculus and its current applications and future prospects presented by Copeland et al. (1993). Taxonomic results of morphometric and genetic characteristics of samples from China, Korea, Taiwan, and Japan identify several geographic groups of M. musculus in the region (Tsuchiya et al., 1994).

Two important reviews are critical to understanding evolution of house mice. Boursot et al. (1993) discussed phylogenetic position of M. musculus within the genus, origin and radiation of the species, range expansions and secondary contacts of subspecies, origin and consequences of commensalism, and chromosomal evolution. Sage et al. (1993) summarized biosystematics and evolution of the house mouse complex, and use of house mice as models in studies of hybrid zone biology, chromosomal evolution and speciation, and testing phylogenetic reconstruction. In other pertinent reviews, Thaler (1986) gave an appraisal of fossil evidence and morphological traits of M. musculus and other European species of Mus. Kotenkova and Bulatova (1994) edited contributions covering evolution and systematics of house mice, geographic distributions, isolating mechanisms, intraspecific genetic polymorphisms, karyotype variability, and behaviour. Bonhomme et al. (1994) presented a synthetic view of genetic differentiation within M. musculus and the hypothesis that the complex formed a ring species (Din et al., 1996, did not think the evidence sufficient to test this idea). Moriwaki et al. (1994) documented genetic variation and its geographical distribution, immunology, chromosomes and genetic mapping, and establishment of wild-derived laboratory strains. Silver (1995) provided a useful compendium for systematists and laboratory researchers. Finally, Berry (1984) reviewed general biology, domestication, history in biological research, and other topics.

Mitochondrial DNA and nuclear gene sequences have been used to determine the evolutionary origin and radiation of M. musculus, and results are the basis of two models. One identifies the N Indian subcontinent as the place where M. musculus likely evolved (Boursot et al., 1996; Din et al., 1996), a region also known for the greatest diversity of species in subgenus Mus recorded by Plio-Pleistocene fossils (Patnaik et al., 1993, 1996). Genetic data indicated that ranges of musculus, castaneus, and domesticus likely "correspond to three distinct possible paths of expansion from the Indian cradle" (Boursot et al., 1993:128, and references cited therein), a conclusion based on the greater genetic variability in samples from the Indian subcontinent compared with the genetic variation in the peripheral regions of its range (Din et al., 1996). Boursot et al. (1996) also championed this view (their scenario, summarized in fig. 4, p. 405, exemplifies the hypothesis), and they look upon M. m. bactrianus as an isolate in mountainous Afghanistan derived from the ancestral population in the N Indian subcontinent. The other model was proposed by Prager et al. (1998:835): "an origin in west-central Asia . . . and the sequential spreading of mice first to the southern Arabian Peninsula, thence eastward and northward into south-central Asia, and later from south-central Asia to north-central Asia (and thence into most of northern Eurasia) and to southeastern Asia."

 "One of the most characteristic features of the house mouse life history is probably its commensalism in relation to humans. The worldwide colonization by this species is mainly due to passive transport by humans and is a consequence of its ecological dependence on them" (Boursot et al., 1993:135). Mus m. castaneus is found only in buildings but the other two subspecies occur in a wide variety of anthropogenic and wild environments. Analyses, using palaeontological and archaeozoological approaches, of the colonization process of W Eurasia and the origin of commensalism were presented by Auffray et al. (1988, 1990a, c) and Auffray and Britton-Davidian (1992). Cucchi et al. (2000) documented unintentional anthropogenic introduction to Cyprus during the 9^th and 8^th millenia BC. Jaarola et al. (1999) outlined colonization history of Fennoscandia by house mice. The living population of M. m. domesticus in the Maderian Arch., 600 km off the Atlantic coast of NW Africa, is also represented by subfossil specimens (Mathias and Mira, 1992; Pieper, 1981) indicating presence before Portuguese discovery of the archipelago in 1419. Analyses of mitochondrial D-loop sequences by Gündüz et al. (2001) indicated the source area for the mice was probably the population of domesticus from N Europe; the Vikings may have inadvertently brought the mice in the ninth century, although no historical evidence exists of their visit. Their study is an example of molecular data providing clues to human historical colonization events rather than the opposite — using human history to infer colonization history of commensals.

Populations in nearly all of the United States descended from the European M. musculus domesticus, probably in the 16^th century, but a population around Lake Casitas, about five miles (8 km) from the Pacific coast in S California, possesses certain genic traits and retroviruses not found in M. m. domesticus but identical to those in populations of the Asian M. m. castaneus (Gardner et al., 1991). Immigrant Chinese laborers who worked on railroads and ranches in the Lake Casitas area in the middle 1800s were likely accompanied by castaneus on the voyages from China to S California (Gardner et al., 1991).

The commensal proclivity of M. musculus ultimately led to its domestication by humans, first as pets and later as laboratory animals. The mitochondrial genome and most of the nuclear genome of classical laboratory inbred strains of M. musculus originated from the European M. m. domesticus (see review and references in Berry, 1984, Bonhomme et al., 1987, Flegr et al., 1994, and Nadeau, 2002; and J. T. Marshall, Jr., in litt., 2004, notes that laboratory mice match W European domesticus in body size, tail length and conformation of zygomatic plate) but questions still exist about genealogy of the strains and their relationships to one another and to wild forms. This is a significant problem because laboratory strains constitute the "universal mammalian model" (Flegr et al., 1994: 33) or "the mammalian archetype in genetic studies" (Bonhomme et al., 1987:52), and knowledge of their origins and genetic relationships among them is critical to interpreting experimental results in laboratories or phylogenetic comparisons between inbred strains and wild populations of M. musculus and other species. Bonhomme et al. (1987:53) discussed the polyphyletic origin of laboratory mice and noted that "any one of the alleles present in inbred strains could have come from domesticus, musculus or castaneus," and they are "a mosaic of genomes coming from different taxonomic origins," a conclusion echoed by Sage et al. (1993; see also Blank et al., 1986; Nishioka, 1987). In a phylogenetic analysis of mitochondrial D-loop sequences in laboratory and wild strains, Flegr et al. (1994) reported that some of the strains they used were more similar to M. m. musculus than to M. m. domesticus. Moriwaki (1994) provided an informative diagram illustrating the genetic differentiation and geographic ranges of subspecies groups of M. musculus and their relevance to the origin of laboratory strains. According to Moriwaki (p. xxiii) "the genetic background of today’s laboratory mice is mainly derived from that of the European domesticus subspecies group and a little from that of some Asian mice, probably Japanese fancy mice belong to the musculus subspecies group." Aligning the genome sequence from the C57BL/6J strain with sequences from other laboratory strains suggested that the genomes of inbred strains used "are mosaics with the vast majority of segments derived from domesticus and musculus sources" (Wade et al., 2002; Nadeau, 2002).

J. T. Marshall, Jr. (1998) is the only systematist who has studied all available holotypes and read all original descriptions of taxa associated with house mice (except arenarius, which is included in musculus by Zagorodnyuk, 1996b, and modestus, listed as a synonym by Meester et al., 1986). Allocation of synonyms follows J. T. Marshall, Jr. (1998), whose report should be consulted for type locality, assignment, and remarks for each name, and a map of 77 continental type localities. Some of the synonyms may eventually be differently allocated once the range of domesticus and status of bacterianus are resolved (see below). Two taxa have been erroneously associated with Mus and the identity of another is uncertain. Musser and Carleton (1993) listed molissimus (Dehne, 1855) in the synonymy of M. musculus, but J. T. Marshall, Jr. (1998) claimed the original description is that of a dormouse, Eliomys or Muscardinus. Hodgson’s (1845:267) "Mus? hydrophilus" has been listed as incertae sedis near Mus musculus accounts (Ellerman and Morrison-Scott, 1951; Ellerman, 1961), but Hodgson described an animal much too large to be any species of Mus known to occur in Nepal, and he grouped it with the species of rats, not the mice he described. Identity of the Taiwanese Mus kagii Kuraoka, 1912, has yet to be determined (Kaneko and Maeda, 2002).

Three names listed in the synonymy were proposed for fossils. Young (1934:80) described "Mus musculus var. bieni" from "Pleistocene" of Choukoutien, China. Pei (1936) studied more than 200 samples he identified as M. musculus from another Choukoutien locality and remarked that Young’s specimens fell well within the size range of typical M. musculus. Late Pleistocene Hungarian fossils were described by Kretzoi as the species synanthropus (in Kretzoi and Vértes, 1965a) and solymarensis (probably proposed by Kretzoi and listed in Jánossy, 1986) in subgenus Budamys of Mus. Both names are listed as synonyms of M. musculus by Kowalski (2001).