Spatial structure influences ecological interactions but it is not well understood how ecological interactions drive spatial patterns across scales of observation. We combined spatial models at two scales to understand how spatial patterning results from host-pathogen interactions in the Douglas-fir tussock moth, \textit{Orgyia pseudotsugata}, and what the consequences of spatial structure are for transmission rates and host heterogeneity. We first combine spatial infection rate data from multiple tussock moth populations with spatial transmission models to understand how limited dispersal drives pathogen dynamics and informs the extent of insect tree damage. We then assess how tree damage patterns observed at larger spatial scales can be explained by eco-evolutionary dynamics over longer time frames. We found that locally, patchy pathogen distributions generate hotspots of transmission that shape overall pathogen dynamics and increase the severity of tree damage caused by Douglas-fir tussock moth larvae. We observed that this spatial structure constitutes the majority of infection risk variation, but that accounting for the heritable proportion of infection risk variation in eco-evolutionary models was necessary to explain the accelerating waves of multiple interacting \textit{O. pseudotsugata} populations over larger spatial scales. Therefore, both small-scale diffusion between trees and heritable host infection risk are important for determining the intensity and location of insect outbreaks at meta-population scales. Our research, founded in model comparisons that confront theory with data, represents a novel synthesis on Douglas-fir tussock moth disease ecology at two spatial scales and provides insight into general host-pathogen theory for understanding pathogen dynamics in nature.