There is growing recognition that restoring species diversity is crucial to maintaining ecological functions and services. Increasing the diversity of species used in restoration programs has placed greater emphasis on determining the seed transfer needs for a wider array of plants. However, many plants, outside of commercial forestry, lack information that would provide guidance on seed transfer for current or future climates. Generalized seed transfer approaches use climate partitioning to approximate adaptive differentiation among populations and provide an estimation of seed transfer distance for such species.
Herein, we describe a generalized seed transfer approach that uses Euclidean distance of 19 climate variables within North America (from northern Honduras to the Arctic). Euclidean distances are used to identify climate analogs from vegetation databases of about 685,000 plots, an average density of 1 plot per 32 km2. Analogs are classified into three thresholds (strong, moderate, and weak) that correspond to altitudinal climate gradients and are guided by the scientific literature of observed adaptive variation of natural tree populations and seed transfer limits.
For strong threshold observations, about 97% of the analogs had climate distances equivalent to ≤300 m elevation, whereas for the weak threshold observations, 53% had an elevation equivalence of ≤300 m. On average 120, 267, and 293 m elevation separated two points under strong, moderate, and weak thresholds, respectively. In total, threshold classification errors were low at 13.9%.
We use examples of plot data identified from a reference period (1961–1990) and mid-century (2056–2065) analogs across North American biomes to compare and illustrate the outcomes of projected vegetation change and seed transfer. These examples showcase that mid-century analogs may be located in any cardinal direction and vary greatly in spatial distance and abundance from no analog to hundreds depending on the site. The projected vegetative transitions will have substantial impacts on conservation programs and ecosystem services. Our approach highlights the complexity that climate change presents to managing ecosystems, and the need for predictive tools in guiding land management decisions to mitigate future impacts caused by climate change.