Guizhou, Guangxi, and Hunan provinces in China have reported the highest incidence of human rabies in recent years and were selected for national epidemic surveillance in 2005. According to national data on rabies surveillance, domestic dogs are the major infection source for humans and other domestic animals (3).

Phylogenetic analysis (Figure 2) showed that human RABV specimens generally belong to the same branch as dog RABVs from the same region, as exemplified by specimens HN0633D and HN0610H (subgroup IIa) and GZ0518H and GZ0505D (group III). This finding provides virologic evidence that domestic dogs are major reservoirs of human rabies epidemics (2). In addition, these results support the inference that RABVs from the same regions tend to group together and represent the same rabies focus or same geographically circumscribed outbreak (20–22).

We observed an infection rate of 2.3% in apparently healthy dogs from 15 cities in the 3 provinces. Previous surveys (23,24) in regions of high incidence of rabies showed different rates, ranging from 3.9% to 17.9% in dogs. Discrepancies in infection rates may be related to the sampling method (including nonclinically suspicious rabid dogs), different sample sizes, detection methods, and particular local variations. Regardless of the infection rate, the data indicate that rabies in dogs is prevalent in these regions and represents a cause of the high incidence of rabies in humans.

Phylogenetic analysis of specimens collected in the 3 provinces clearly shows that RABVs have a distinctive geographic correlation, in which group I was found predominantly in Hunan Province, group II in Guangxi Province, and group III in Guizhou Province (Figure 2). From an epidemiologic perspective, these groups may be interpreted as ongoing independent foci of dog rabies (or outbreaks).

Identity levels of deduced amino acid sequences of the 60 specimens ranged from 95.4% to 99.6%, which are higher than similarity values for corresponding nucleotide sequences. This finding indicates that most nucleotide variations in N gene coding sequence were synonymous mutations compatible with ongoing natural selection. Samples from different provinces had identical nucleotide sequences, which indicated that these viruses may have originated from the same outbreak or that a rabies focus had been spread between provinces (13,14).

Members of groups I and II were found in all 3 provinces selected for this study (Figure 2), which indicates that they have been spread throughout the region over time. The fact that subgroups Ic and IIa are integrated with virus samples from provinces distinct from their actual representative groups suggests that RABV translocation events are occurring between provinces (25). Phylogenetic inferences and epidemiologic and historical data suggest that ongoing rabies foci in Guangxi and Guizhou provinces, represented by subgroup Ic, had their origins in Hunan Province. Statistically supported independence of subgroup Ic within group I also suggests that such translocation events may have occurred in an isolated manner and may implicate dog RABV no longer circulating in Hunan Province. Similarly, subgroup IIa represents a rabies focus that had its origins in Guangxi Province and was then translocated to Hunan Province. Recent translocations from Hunan Province to Guangxi Province are represented by taxa CGX0612D, CGX0607D, CGX0510D, CGX0606D, and CGX0619D. Recent translocations from Guangxi Province to Hunan and Guizhou provinces represented by taxa CHNO635H, CGZ0514D, CGZ0506D, CGZ0512D, and CGZ0516D (Figure 2). Our results show trends in virus spread from current and historical occurrences in neighboring provinces.

Transprovincial spread of RABVs may be one of the forces responsible for exacerbating the dog rabies epizootic, as reflected by a human rabies epidemic. Specimen CGX0617D from Liuzhou in Guangxi Province (group III; Figure 2) is 100% identical to specimens from Guizhou Province, a finding that supports the suggestion that RABV was introduced into Liuzhou from Guizhou Province during the human rabies outbreak in 2004. Our findings are supported by previous studies in Guangxi Province (26,27), which suggested that different RABV lineages cocirculating in this province may correlate with an increased number of human rabies cases during 2003–2004 in Guangxi Province.

We sought to determine how transprovincial spread of RABVs occurred. The geographic nature of these rabies foci or groups suggests that dogs are not moving, per se, but that human-related activities may account for these phenomena. Persons in southern China are accustomed to eating dog meat because they believe it helps them resist dampness and cold in winter. In addition, many restaurants in cities and regions of southern China sell dog meat, which increases the demand for this type of meat. In rural areas of southern China, nearly every family owns several dogs, most of which are free-roaming, without special diets, and unvaccinated against rabies (to save costs). Sale of these dogs to restaurants can increase a farmer’s income (average 12–15 US dollars/dog). Free-roaming and unvaccinated dog populations may increase the likelihood of transprovincial spread of RABVs. Other human-related activities, such as persons migrating with their dogs may also contribute to long-distance spread of rabies. Dogs purchased by restaurants are soon killed for consumption. With the exception of butchers, there would be insufficient time to transmit RABV to other dogs and humans. Until now, persons in China who eat dog meat have not been not considered at risk for rabies because no related infections have been reported.

Specimens belonging to groups I and II, which are the predominant RABVs distributed across the 3 provinces, the Chinese vaccine strain CTN and strains from Southeast Asia form an isolated phylogenetic cluster. This finding demonstrates that strains from China and Southeast Asia may have the same origin (11,14,28–30). Southwestern China borders Vietnam, Laos, and Myanmar, countries into which persons in China have migrated over the past 2,000 years. Thus, dissemination of RABV has been historically linked across Southeast Asia.

Conversely, natural barriers such as the Himalayas and large rivers may have prevented RABV dissemination from the Indian subcontinent and central Asia into China (20). Our data indicate that RABV from these regions has not been recently transmitted into China. Enzootic dog RABV lineages from these regions (Indian dog and Arctic-like RABVs) are phylogenetically distinct from lineages in China, which supports this concept (31,32).

Groups I and II appear to have evolved from a common, long-term dog rabies enzootic in China. Group III appears to be more closely associated with widely spread dog-related RABVs represented by vaccine strains and other lineages established in wildlife in other regions (Figure 3). These results suggest 2 origins for rabies in dogs in the 3 provinces in southern China: an autochthonous origin and a more recent origin by introduction of a cosmopolitan dog RABV variant into China.

Approximately 57 years ago, Johnson (33) speculated that RABV strains from Europe were transmitted into China through Hong Kong and Shanghai. The attenuated 3aG strain, which was isolated in Beijing in 1931, and the DRV strain, which was isolated in Jilin Province in 2002, are closely related to group III. This finding implies that a group of viruses that originated in Europe is present in China and is still circulating. The hosts of this group include not only domestic dogs but also other mammals likely infected by rabid dogs (DRV strain). Alternatively, the similarity among some RABVs circulating in dogs in China and international vaccine strains (34) should motivate health authorities in China to revisit quality standards and adequacy for use of attenuated rabies vaccines to ensure that vaccine-related cases do not occur.

In conclusion, spread of RABVs from high-incidence regions, particularly by the long-distance migration or transprovincial movement of dogs caused by human-related activities, may be one of the causes of recent massive human rabies epidemics in southern China. We identified only 1 rabies case in a deer in China; it had evidence of spillover infection from a rabid dog. Although no evidence was found of independent rabies enzootics sustained by wildlife in China, multidisciplinary studies are required to improve rabies surveillance to identify potential alternative sources of human infection. Creation of a sensitive laboratory-based surveillance system for animal rabies detection is a prerequisite for enabling effective dog vaccination campaigns. Integration of enhanced surveillance, molecular analysis, local canine population management, and efficient rabies vaccination coverage in dogs are essential to prevent rabies from being transmitted by rabid dogs to humans.