Relationships between fox populations and rabies virus spread in northern Canada

Abstract

Rabies spreads in both Arctic (Vulpes lagopus) and red foxes (Vulpes vulpes) throughout the Canadian Arctic but limited wildlife disease surveillance, due to the extensive landmass of the Canadian north and its small widely scattered human population, undermines our knowledge of disease transmission patterns. This study has explored genetic population structure in both the rabies virus and its fox hosts to better understand factors that impact rabies spread. Phylogenetic analysis of 278 samples of the Arctic lineage of rabies virus recovered over 40 years identified four sub-lineages, A1 to A4. The A1 lineage has been restricted to southern regions of the Canadian province of Ontario. The A2 lineage, which predominates in Siberia, has also spread to northern Alaska while the A4 lineage was recovered from southern Alaska only. The A3 sub-lineage, which was also found in northern Alaska, has been responsible for virtually all cases across northern Canada and Greenland, where it further differentiated into 18 groups which have systematically evolved from a common predecessor since 1975. In areas of Arctic and red fox sympatry, viral groups appear to circulate in both hosts, but both mitochondrial DNA control region sequences and 9-locus microsatellite genotypes revealed contrasting phylogeographic patterns for the two fox species. Among 157 Arctic foxes, 33 mitochondrial control region haplotypes were identified but little genetic structure differentiating localities was detected. Among 162 red foxes, 18 control region haplotypes delineated three groups which discriminated among the Churchill region of Manitoba, northern Quebec and Labrador populations, and the coastal Labrador locality of Cartwright. Microsatellite analyses demonstrated some genetic heterogeneity among sampling localities of Arctic foxes but no obvious pattern, while two or three clusters of red foxes suggested some admixture between the Churchill and Quebec-Labrador regions but uniqueness of the Cartwright group. The limited population structure of Arctic foxes is consistent with the rapid spread of rabies virus subtypes throughout the north, while red fox population substructure suggests that disease spread in this host moves most readily down certain independent corridors such as the northeastern coast of Canada and the central interior. Interestingly the evidence suggests that these red fox populations have limited capacity to maintain the virus over the long term, but they may contribute to viral persistence in areas of red and Arctic fox sympatry.

Fig 1. Map of Canada showing localities of fox sampling for population studies.

Samples were collected between 1993–2017 with the majority recovered between 2012–2014. Each marker represents a coordinate corresponding to one or several individuals. Major sampling localities are shown by large circles for red foxes, and large squares for Arctic foxes, while those localities represented by 1–3 foxes are small circles or squares respectively.

Fig 2. Phylogeny of the Arctic rabies virus lineage using samples collected between 1977–2017.

This ML tree was generated in MEGA v7 using an alignment of 1350 positions of the N gene for 278 RABV samples and a member of the distantly related Arctic-like lineage as outgroup. Bootstrap values expressed as a percentage are indicated at major nodes. The host species from which each sample was recovered is shown according to the host key provided in the inset. The major sub-lineages are identified to the right of the tree.

Fig 3. A time-scaled maximum clade credibility tree produced using WGS of 208 Arctic lineage RABVs collected between 1977–2017.

The tree, which was generated using the BEAST v1.7 software package, depicts the four major sub-lineages (A1 to A4). Distinct A3 groups are colour coded and identified to the right. The predicted evolutionary time scale is shown below the tree. The inset to the right further resolves the A3-17 and A3-18 groups into several sub-groups each of which is supported by a posterior probability of ≥ 0.99.

Fig 4. Distribution of A3 sub-lineage RABVs.

Maps show the locations where viral samples of the A3 sub-lineage were recovered over three time periods: A, 1990–1999; B, 2000–2009; C 2010–2017. Points identify both the viral group, according to color, consistent with the color coding of the phylogeny shown in Fig 3, and the host species by shape as illustrated. Multiple cases of the same viral type in the same host in the same location are represented by a single point to preserve location accuracy.

Fig 5. Median spanning network of 18 red fox mitochondrial control region haplotypes generated using network 5.0.

Circles representing haplotypes are scaled according to frequency, with proportional membership in sampling localities as indicated in the key. Sample locality codes are given in Table 1.

Fig 6. Median spanning network of 33 Arctic fox mitochondrial control region haplotypes generated using network 5.0.

Circles representing haplotypes are scaled according to frequency, with proportion membership in sampling localities as indicated in the key. Sample locality codes are given in Table 1.

Fig 7. Structure analysis of red foxes.

A. Plot of individual membership in K = 2 and K = 3 genetic clusters using the default admixture model with 100,000 burn-in and 500,000 retained iterations. Sample locality codes are given in Table 1. Tree plot of the relationships among the K = 3 genetic clusters under this model. B. A. Plot of individual membership in K = 3 genetic clusters using the admixture model with LOCPRIOR, with 100,000 burn-in and 500,000 retained iterations. Sample locality codes are given in Table 1. Triangle plot of individual association with each of the K = 3 genetic clusters under the admixture with LOCPRIOR model. Individuals are colour-coded as indicated in A and B. The yellow dots represent the individuals not in one of the six main sampling localities, as indicated in Table 1. Tree plot of the relationships among the K = 3 genetic clusters, under the admixture with LOCPRIOR model.