Soil specialists can have restricted distributions, and effective management in the face of land use change depends on a thorough understanding of the ecology of these unique plants. We investigated the ecology of Eriogonum tiehmii Reveal, a rare soil specialist, using pitfall traps, flower observations, and pollinator exclusion to assess arthropod communities in E. tiehmii habitat, the most common visitors to E. tiehmii flowers, and the importance of pollination for seed set. We collected soil from 21 occupied and unoccupied sites for analysis of physical and chemical characteristics, and conducted a greenhouse soil preference experiment to test how seeds and seedlings respond to soil variation. We also monitored tagged plants in extant populations, measuring survival, size, and reproductive output. Arthropod communities within and around E. tiehmii sites were abundant and diverse; each sample site contained numerous unique species, and there was high turnover in arthropod community composition over time. Open pollination significantly increased seed production, with beetles, wasps, and flies the most likely important pollinators.Reproductive output and size class distributions varied among wild populations, and soil properties differed between occupied and unoccupied sites. Occupied sites were, on average, lower in sulfur, zinc, potassium, and magnesium and, on average higher in boron, pH, and silt, among other differences, though there was high variation in these characteristics among sites.Seedlings demonstrated sensitivity to individual soil properties, and seedlings grown in soils from occupied sites had, on average, higher total biomass and higher root allocation than seedlings grown in soils from unoccupied sites. These preference tests suggest a "specialist" model of soil specialization rather than a "refuge" model, indicating that plants are not simply highly stress-tolerant, but are specifically adapted to their native soil types. While some unoccupied sites we tested were favorable for some life history stages, we did not identify unoccupied sites that could support both establishment and growth of E. tiehmii seedlings. Future work could consider the effects of biotic interactions such as plant competition and herbivores on plant growth, and employ