Data and code for: Disease outbreaks select for mate choice and coat color in wolves

We know much about pathogen evolution and the emergence of new disease strains but less about host resistance and how it is signaled to other individuals and subsequently maintained. The cline in frequency of black-coated wolves across North America is hypothesized to result from a relationship with canine distemper virus (CDV) outbreaks. We test this hypothesis using cross-sectional data from wolf populations across North America that vary in the prevalence of CDV and the allele that makes coats black, longitudinal data from Yellowstone National Park, and modeling. The frequency of CDV outbreaks generates fluctuating selection that results in heterozygote advantage that in turn impacts the frequency of the black allele, the optimal mating behavior, and the black wolf cline across the continent.

Methods
Technical details about the analyses and the models are provided in the Material and Methods section of the article.

1 Survival analysis
Data can be found in file 'dataYNPsurvivalanalysis340.RData', codes and step-by-step guidance are available to run the analysis and visualize results (file : SurvivalAnalysis_Cubaynesetalarticle.Rmd.

We used demographic data that included individual survival and immigration records, age, pack, and coat color phenotype and genotype for 340 wolves equipped with radio-collars in Yellowstone National Park, monitored between 1998 to 2020. Of these wolves, 181 were grey and 159 black (144 black heterozygotes, 10 black homozygotes, 5 of unknown genotype). We censored records at their last live sighting within the boundary of the park. 105 wolves were censored at the point they permanently dispersed outside of the National Park (n=101) or were killed by humans (n=4), deliberately or accidentally, within Yellowstone National Park. 

During annual capture and radio-collaring efforts, the National Park Service routinely collects blood serum for serological analysis. All serum samples (n=425) were analyzed for CDV antibodies using a serum neutralization assay at Cornell University’s Animal Health Diagnostic Center (22,47). CDV exposure was a binary variable representing previous exposure (1, i.e., seropositive) or not (0, i.e., seronegative). A result was seropositive when the titer dilution was equal or greater than the standard titer cutoff provided by the assay manufacturer. All wolf samples analyzed for this study were previously collected the National Park Service; no wolf was captured or handled for the purposes of this study.

CDV outbreaks have occurred in 1999, 2005, 2008, 2017 and 2018 in the Yellowstone wolf population. Reported exposure rate of wolves to CDV was near zero in non-outbreak years, and between 60% and 80% in outbreak years. We assumed no exposure occurred in non-outbreak years, unless a wolf seroconverted in that year. Given CDV’s extreme transmissibility within a pack, individuals were determined positive for CDV exposure if they, or at least one pup in their pack, seroconverted during an outbreak year. Given high exposure rates in outbreak years, we assumed that individuals that never tested negative for CDV at or after collaring were previously exposed to CDV if they were alive during an outbreak year. Introduced wolves that received a CDV vaccine as part of the release program were considered to have never been exposed to CDV (n=12). 

In a year following first exposure to CDV, a wolf can either die or gain lifelong immunity. In our model, a wolf had two fates following its first CDV exposure: 1) die within the year of exposure, or 2) gain lifelong CDV immunity.

We combined demographic, genetic and serological data to obtain the encounter histories of the 340 wolves alive or dead of natural causes within Yellowstone National Park between 1998 and 2020. 

2 Analysis across North American wolf populations
Details about the data and sampling protocols are available in E.E. Brandell, et al. Patterns and processes of pathogen exposure in gray wolves across North America. Scientific reports, 11,1-14 (2021). 

Data can be found in file 'dataset_binary_reduced.csv', codes and step-by-step guidance are available to run the analysis and visualize results (files: 'RunBayes_models_noCPV_allPops.r' and 'WriteBayes_models'.

3 Projection
Step-by-step guidance to run the population projection model is provided in the file Script_ProjectionModel_fromCubaynesetal2022.Rmd