Simple models of plant response to warming climates predict vegetation moving to cooler and/or wetter locations: in mountainous regions shifting upslope. However, species-specific responses to climate change are likely to be much more complex. We re-examined a recently reported vegetation shift in the Santa Rosa Mountains, California, to better understand the mechanisms behind the reported shift of a plant distribution upslope. We focused on five elevational zones near the center of the gradient that captured many of the reported shifts and which are dominated by fire-prone chaparral. Using growth rings, we determined that a major assumption of the previous work was wrong: past fire histories differed among elevations. To examine the potential effect that this difference might have on the reported upward shift, we focused on one species, Ceanothus greggii: a shrub that only recruits post-fire from a soil stored seedbank. For five elevations used in the prior study, we calculated time series of past per-capita mortality rates by counting growth rings on live and dead individuals. We tested three alternative hypotheses explaining the past patterns of mortality: 1) mortality increased over time consistent with climate warming, 2) mortality was correlated with drought indices, and 3) mortality peaked 40-50 years post fire at each site, consistent with self-thinning. We found that the sites were different ages since the last fire, and that the reported increase in the mean elevation of C. greggii was due to higher recent mortality at the lower elevations, which were younger sites. The time-series pattern of mortality was best explained by the self-thinning hypothesis and poorly explained by gradual warming or drought. At least for this species, the reported distribution shift appears to be an artifact of disturbance history and is not evidence of a climate warming effect.