Contact heterogeneity, rather than transmission efficiency, limits the emergence and spread of canine influenza virus

Abstract

Host-range shifts in influenza virus are a major risk factor for pandemics. A key question in the study of emerging zoonoses is how the evolution of transmission efficiency interacts with heterogeneity in contact patterns in the new host species, as this interplay influences disease dynamics and prospects for control. Here we use a synergistic mixture of models and data to tease apart the evolutionary and demographic processes controlling a host-range shift in equine H3N8-derived canine influenza virus (CIV). CIV has experienced 15 years of continuous transfer among dogs in the United States, but maintains a patchy distribution, characterized by sporadic short-lived outbreaks coupled with endemic hotspots in large animal shelters. We show that CIV has a high reproductive potential in these facilities (mean R(0) = 3.9) and that these hotspots act as refugia from the sparsely connected majority of the dog population. Intriguingly, CIV has evolved a transmission efficiency that closely matches the minimum required to persist in these refugia, leaving it poised on the extinction/invasion threshold of the host contact network. Corresponding phylogenetic analyses show strong geographic clustering in three US regions, and that the effective reproductive number of the virus (R(e)) in the general dog population is close to 1.0. Our results highlight the critical role of host contact structure in CIV dynamics, and show how host contact networks could shape the evolution of pathogen transmission efficiency. Importantly, efficient control measures could eradicate the virus, in turn minimizing the risk of future sustained transmission among companion dogs that could represent a potential new axis to the human-animal interface for influenza.

Figure 1. Phylogenetic trees of HA1, NP and M sequences for EIV (black) and CIV (colors).

Boxes surround CIV clades comprising two or more samples from the same US state. Branches leading to CIV samples from the same location are colored by location (New York, blue; Pennsylvania, orange; Colorado, purple. Branches leading to CIV samples from multiple locations are colored grey.

Figure 2. Demography of dogs in US animal shelters.

(A) Cumulative distribution of median population size in each shelter (black dashed line) compared to a negative binomial distribution fitted to the data (solid red line), and a fitted Poisson distribution (dotted blue line). (B) Intake rate as a function of population size. Points show the median value for each shelter and vertical lines enclose the interquartile range. Line shows fit by linear regression to log-transformed median intake rates. (C) Length of stay as a function of shelter size. The 95% confidence interval on the slope of the dashed line includes 0. (D) Cumulative distribution of length of stay across all shelters (bars) compared to an exponential distribution with mean rate 1/9.88 days⁻¹ (solid line).

Figure 3. Seroprevalence, R₀ and R_e for CIV, estimated from host demographic data, seroprevalence data, and molecular data.

(A) Saturating relationship between seroprevalence and R₀ in a stochastic SIR framework, parameterized from the shelter intake and output data. Red line shows equilibrium seroprevalence predicted by the mean-field model. Points show point seroprevalence estimates from the stochastic simulations, where 74 dogs are sampled at random in a shelter with an average dog population of 134, corresponding to . (B) Deviations of point seroprevalence estimates from the long-term average (bars) compared to a normal distribution (line). (C) Posterior distribution of R₀ based on an observed seroprevalence of 0.42 in . (D) and (E) R_e for CIV, estimated by fitting a birth-death skyline phylodynamic model to HA1 gene sequences. The black line shows the mean estimate while the grey shaded shows the highest posterior density (HPD) range, encompassing 95% of the credible set of sampled values.

Figure 4. Demographics, persistence, spread rate and possible eradication of CIV.

(A) Dog population sizes in animal shelters and within-shelter spread rates at which CIV can persist for at least 100 days according to present intake and output rates. The surface shows a smoothed (kriged) version of the outcome (persistence for at least 100 days) of 1000 simulations conducted at uniform random points within the plane described by the figure. Darker shades correspond to higher probabilities of persistence. Red symbols show features of the distribution for dog population sizes (estimated from the demographics data) and the posterior distribution of R₀ in large shelters (estimated from seroprevalence data by MCMC; see Figures 2 and 3), including the median (hollow circle), mean (filled circle), 2.5th percentile (minus sign) and 97.5th percentile (plus sign). (B) Results of an intervention that reduces the arrival rate of susceptible individuals at a shelter to 1/<R₀> its current value, where <R₀> is the mean posterior distribution of R₀ for CIV estimated from all available data. A kernel density estimate for the distribution of shelter sizes in the demographic data is shown below (A) and (B) to illustrate the scarcity of shelters large enough to support CIV in an endemic state.

Figure 5. Predicted performance of a control program administered to dogs on arrival in US animal shelters.

(A) A vaccination (or other control measure) that removes individuals from the chain of transmission with 85% probability (κ = 0.15) within 24 h (α = 1 day) is predicted to eradicate CIV from shelters within six months. The simulations used 100 shelters with dog population size, intake rate, and outtake rate jointly sampled with replacement from the shelter demographics data, and R₀ = 3.9. White lines show medians and shaded areas enclose the 5^th to the 95^th percentiles of the simulation data. (B) Decreasing control efficacy to 75% can still achieve eradication in isolated shelters (blue region, solid line), however shelters that transfer dogs amongst themselves at the observed mean rate of τ = 0.1 would preserve CIV in a few shelters despite the vaccination program (red region, dashed line). (C) Further decreases in vaccine efficacy make eradication significantly less likely, particularly if shelters are connected through the transfer of dogs.

Figure 6. A simulation of CIV invasion over multiple shelters, starting with an infection in a single large shelter.

(A) Each vertex represents an animal shelter with dog population size proportional to the area of the circle. Edges show transfer of infection from shelter to shelter over time through the movement of infected dogs. Edge lengths are arbitrary. The data for this figure were produced by simulating the metapopulation stochastic SIR model with 100 shelters for 100 days, starting with a single infection in the largest shelter. Population sizes were sampled with replacement from the shelter data. R₀ = 3.9. Transfer probability is set to the mean observed value of τ = 0.1. (B) Large shelters tend to receive the infection earlier (and more often) following an outbreak at another shelter. (C) Probability that CIV will persist for 100 days in a shelter of a given size following the introduction of a single infected individual to an otherwise susceptible population. The plateau on the curve arises for populations sufficiently large that early depletion of susceptibles is not an important factor in the probability of an outbreak: rather than population size, this probability is determined by R₀.