Viruses of free-roaming and hunting dogs in Uganda show elevated prevalence, richness and abundance across a gradient of contact with wildlife

Abstract

Domestic dogs (Canis lupus familiaris) live with humans, frequently contact other animals and may serve as intermediary hosts for the transmission of viruses. Free-roaming dogs, which account for over 70% of the world's domestic dog population, may pose a particularly high risk in this regard. We conducted an epidemiological study of dog viromes in three locations in Uganda, representing low, medium and high rates of contact with wildlife, ranging from dogs owned specifically for traditional hunting in a biodiversity and disease 'hotspot' to pets in an affluent suburb. We quantified rates of contact between dogs and wildlife through owner interviews and conducted canine veterinary health assessments. We then applied broad-spectrum viral metagenomics to blood plasma samples, from which we identified 46 viruses, 44 of which were previously undescribed, in three viral families, Sedoreoviridae, Parvoviridae and Anelloviridae. All 46 viruses (100 %) occurred in the high-contact population of dogs compared to 63 % and 39 % in the medium- and low-contact populations, respectively. Viral prevalence ranged from 2.1 % to 92.0 % among viruses and was highest, on average, in the high-contact population (22.3 %), followed by the medium-contact (12.3 %) and low-contact (4.8 %) populations. Viral richness (number of viruses per dog) ranged from 0 to 27 and was markedly higher, on average, in the high-contact population (10.2) than in the medium-contact (5.7) or low-contact (2.3) populations. Viral richness was strongly positively correlated with the number of times per year that a dog was fed wildlife and negatively correlated with the body condition score, body temperature and packed cell volume. Viral abundance (cumulative normalized metagenomic read density) varied 124-fold among dogs and was, on average, 4.1-fold higher and 2.4-fold higher in the high-contact population of dogs than in the low-contact or medium-contact populations, respectively. Viral abundance was also strongly positively correlated with the number of times per year that a dog was fed wildlife, negatively correlated with packed cell volume and positively correlated with white blood cell count. These trends were driven by nine viruses in the family Anelloviridae, genus Thetatorquevirus, and by one novel virus in the family Sedoreoviridae, genus Orbivirus. The genus Orbivirus contains zoonotic viruses and viruses that dogs can acquire through ingestion of infected meat. Overall, our findings show that viral prevalence, richness and abundance increased across a gradient of contact between dogs and wildlife and that the health status of the dog modified viral infection. Other ecological, geographic and social factors may also have contributed to these trends. Our finding of a novel orbivirus in dogs with high wildlife contact supports the idea that free-roaming dogs may serve as intermediary hosts for viruses of medical importance to humans and other animals.

Keywords: Africa; Uganda; ecology; epidemiology; free-roaming dogs; viruses; zoonoses.

Fig. 1.. Map of study locations, showing three districts in Uganda chosen to span a gradient of contact between dogs and wildlife (Bundibugyo District, high contact; Kabarole District, medium contact and Wakiso District, low contact).

Fig. 2.. Maximum-likelihood phylogenetic tree of anelloviruses based on an alignment (455 positions) of partial ORF1 amino acid sequences. Taxa are colour coded to indicate clades corresponding to the recognized genera. The tree is outgroup rooted. Taxon names are followed by host, country, year and GenBank accession number in parentheses, where available. The taxon names of viruses identified in this study are in bold. Numbers beside nodes indicate bootstrap values (per cent; only values ≥50 % are shown); the scale bar indicates nucleotide substitutions per site.

Fig. 3.. Maximum-likelihood phylogenetic trees of parvoviruses based on alignments ((a) 636 positions, (b) 406 positions and (c) 593 positions) of partial NS1 amino acid sequences. Clades corresponding to recognized genera are indicated. Each tree is outgroup rooted. Taxon names are followed by host, country, year and GenBank accession number in parentheses, where available. The taxon names of viruses identified in this study are in bold. Numbers beside nodes indicate bootstrap values (per cent; only values ≥50 % are shown); scale bars indicate nucleotide substitutions per site.

Fig. 4.. Maximum-likelihood phylogenetic tree of orbiviruses based an alignment (815 positions) of partial VP3 (T2) amino acid sequences. The tree is outgroup rooted. Taxon names are followed by host, country, year and GenBank accession number in parentheses, where available. The taxon name of the virus identified in this study is in bold. Numbers beside nodes indicate bootstrap values (per cent; only values ≥50 % are shown); the scale bar indicates nucleotide substitutions per site.

Fig. 5.. Viral prevalence (proportion of dogs infected; 46 viruses), richness (number of viruses per dog; 48 dogs) and abundance (log₁₀vRPM/kb; 44 dogs in which at least one virus was detected) in three populations of dogs in Uganda. The three populations were selected to span a gradient of contact with wildlife, from Wakiso (low contact; n=12) to Kabarole (medium contact; n=12) to Bundibugyo (high contact; n=24). Lines and error bars are means and standard deviations, respectively.