Associated habitat and suitability modeling of goldenseal (Hydrastis canadensis L.) in Pennsylvania: explaining and predicting species distribution in a northern edge of range state

Abstract

Goldenseal (Hydrastis canadensis L.) is a well-known perennial herb indigenous to forested areas in eastern North America. Owing to conservation concerns including wild harvesting for medicinal markets, habitat loss and degradation, and an overall patchy and often inexplicable absence in many regions, there is a need to better understand habitat factors that help determine the presence and distribution of goldenseal populations. In this study, flora and edaphic factors associated with goldenseal populations throughout Pennsylvania--a state near the northern edge of its range--were documented and analyzed to identify habitat indicators and provide possible in situ stewardship and farming (especially forest-based farming) guidance. Additionally, maximum entropy (Maxent) modeling was applied to better predict where suitable habitat might be encountered more broadly and explain species absence from regions of the state (and the northeastern states). Habitat study results identified rich, mesic, woodland sites as being suitable for goldenseal. The most prevalent overstory tree associates on such sites were tulip-poplar (Liriodendron tulipifera L.) and sugar maple (Acer saccharum Marshall), and the most common understory associates were spicebush (Lindera benzoin L.), Virginia creeper (Parthenocissus quinquefolia (L.) Planch.), Jack-in-the-pulpit (Arisaema triphyllum (L.) Schott), mayapple (Podophyllum peltatum L.), wood fern (Drypoteris marginalis (L.) A. Gray), and rattlesnake fern (Botrypus virginianus (L.) Michx.). Loam soils were the most common textural class (average sand, silt, and clay ratio of 50:30:20) with an average pH of 6.2 and high variation in macronutrients. While such sites are widespread in the state, Maxent modeling suggested the present distribution in Pennsylvania is largely restricted by winter temperatures and bedrock type. The latter of these, in turn, is correlated in part with land use legacy (e.g., clearing for farming or livestock grazing), especially in southeast portions of the state.