Browsing by Author "Cramb, Nicholas"

Now showing 1 - 1 of 1
  • Thumbnail Image
    ItemOpen Access
    Forest Cover Influences the Predictions Made by Species Distribution Models: A Case Study of American hazelnut (Corylus americana)
    (2024-10-11) Cramb, Nicholas; Vamosi, Jana; Dawson, Andria; Galpern, Paul; Yeaman, Samuel
    Species distribution models can be used to predict climate impacts on biodiversity and guide conservation efforts. However, they may not fully represent biological reality when they entirely rely on climate variables alone and neglect biological interactions. My research objective is to test if including forest cover as a quantitative proxy for shrub-canopy interactions improves the predictive ability of species distribution models. My work focuses on American hazelnut (Corylus americana) as a case-study to test this framework. American hazelnut is a good candidate species because it is widely distributed throughout the eastern temperate and northern forests of North America and is likely to interact with the full gradient of 0 to 100% forest cover. The species remains understudied regarding the determinants of its distribution, despite an excellent fossil pollen record. This project used the hazelnut fossil pollen record to model the influence of canopy cover on distribution through time and to test whether hazelnut niche has been stable over time. For distribution models of American hazelnut, I found no difference in model performance when land cover variables are included at a continental extent. However, at a regional extent I found significant increases in model performance when forest cover was included. These findings suggest that land cover can more precisely define where species habitat exists at a local level compared to climate variables alone. Distribution models developed using fossil pollen occurrences and climate simulations indicate that hazelnut has maintained a consistent niche space over the past 11,000 years before present. However, the limitations of fossil pollen data created less certainty in these results. Future work involves testing this framework on additional species to determine if these patterns are consistent across woody taxa, as well as testing if land cover datasets that are more categorically resolved and include human impacts are able to improve distribution models.