The history of Coast Salish "woolly dogs" revealed by ancient genomics and Indigenous Knowledge

Abstract

Ancestral Coast Salish societies in the Pacific Northwest kept long-haired "woolly dogs" that were bred and cared for over millennia. However, the dog wool-weaving tradition declined during the 19th century, and the population was lost. In this study, we analyzed genomic and isotopic data from a preserved woolly dog pelt from "Mutton," collected in 1859. Mutton is the only known example of an Indigenous North American dog with dominant precolonial ancestry postdating the onset of settler colonialism. We identified candidate genetic variants potentially linked with their distinct woolly phenotype. We integrated these data with interviews from Coast Salish Elders, Knowledge Keepers, and weavers about shared traditional knowledge and memories surrounding woolly dogs, their importance within Coast Salish societies, and how colonial policies led directly to their disappearance.

Figure 1. Domestic dogs in the culture and society of Indigenous Coast Salish peoples.

1A. Coast Salish ancestral lands include the inner coastal waterways of Salish Sea in southwest British Columbia and Washington State. Archaeological woolly dog data are from (2). Distribution of the Coast Salish languages in the 19th century as indicated by colored areas. The map is modified from https://commons.wikimedia.org/wiki/File:Coast_Salish_language_map.svg and licensed under CC BY-SA 4.0. 1B. Woven Skokomish/Twana basket with woolly dog iconography, depicted with upturned tails. Woolly dog puppies are inside pens represented by diamond shapes (10) (courtesy of Burke Museum, Catalog number #1-507). 1C. Forensic reconstruction of a woolly dog based on Mutton’s pelt measurements and archaeological remains (9). Sketches of Arctic and spitz dog breeds are shown for scale and comparison of appearance, and do not imply a genetic relationship.

Figure 2. Genetic ancestry of woolly dogs.

2A. mtDNA tree of 207 dogs with A2b (Mutton) and A1a (SB Dog) haplotypes expanded. Map points correspond to colored tree tips for the most similar archaeological and historic dog mtDNAs, highlighting the subclades of interest and the broader haplotypes. Samples used are listed in DataS1. 2B. Outgroup-f3 statistics (f3(GrayFox; Mutton, B) or estimation of shared drift between Mutton and 229 other dogs reveals that Mutton has highest similarity to PCDs. Black point estimates indicate ancient genomes. 2C. D-statistics (((PCD, Mutton), Test Dog), Gray Fox) consistent with gene flow into Mutton’s background, with European breeds appearing the most likely contributors to Mutton’s non-PCD ancestry. 2D. f4-ratio tests (f4(A, Out; Mutton, AL3194-PortauChoix): f4(A, Out; B, AL3194-PortauChoix)) to estimate the proportion of European settler dog ancestry in Mutton’s background using six modern European breeds as proxies for Mutton’s European ancestry component.

Figure 3. Genomic outcomes of management and selection.

3A. Global heterozygosity and long runs of homozygosity over transversions in Mutton compared to modern dogs and the ancient Port au Choix dog. All dogs have been downsampled to Mutton’s coverage level for analysis. 3B. Tree schematic used in dN/dS analysis to identify genes under selection in Mutton compared to other canids. Branching order after (50). dN/dS estimates were done separately including one of the four dogs plus all other canids. Genes with elevated dN/dS_Genome values in multiple dogs could reflect more ancient shared selection before the separation of the woolly dog lineage. Therefore, likely candidates for selection in woolly dogs were conservatively assessed where dN/dS_Genome>1.5 in Mutton (9), but dN = 0 in the other three dogs, including one PCD. 3C. Genes with an excess of non-synonymous mutations in Mutton. Black points are the 125 selection candidates on the basis of dN/dS_genome ≥1.5 in Mutton but dN=0 in three other dogs including one PCD (9). Several genes with high dN/dS_genome in Mutton (shown in gray) are excluded as selection candidates because they carry at least one non-synonymous mutation in other dogs. This approach is designed to conservatively highlight genes where selection is more likely specific to Mutton’s lineage rather than during dog domestication or in the common ancestors of PCDs. Candidate genes discussed in text are indicated.