Smith ScholarWorks

These spatial data layers were derived using the simulation modeling work of Duveneck and Thompson (2017) for all of New England. The input data were 250m resolution simulations of above-ground biomass (g/m2) for each tree species across the landscape, using a simulation model (LANDIS-ii with PnET-Succession) that incorporates forest ecosystem processes such as seed dispersal, growth, and mortality. Climate change was included in the forest modeling, but was represented with limited spatial variability (i.e., 25 separate climate regions across New England (Duveneck and Thompson 2017, Figure 1).

We used three primary data sources from Duveneck and Thompson (2017): l ) estimates of above-ground biomass in 2010, 2) simulations of above-ground biomass for 2100 under a continuation of recent climate conditions (i.e., static climate), and 3) simulations of above-ground biomass for 2100 under climate change specifically using the representative concentration pathway (RCP) 8.5. Duveneck and Thompson (2017) simulated RCP 8.5 using four separate Global Circulation models (GCMs). We specifically used results from their work based on the Hadley Global Environment Model v2-Earth System (HADLEY) GCM, and the Community Climate System Model v4.0 (CCSM4) GCM. All simulations presented here represent a ‘potential' landscape at 2100 that includes forest dynamics and succession, but does not include any land-use change, timber harvesting, insect defoliation, agriculture, or other management activities. Separate ongoing research by Duveneck and Thompson address some of the interacting effects of these other disturbances. We categorized climate change refugia as any cell where a species’ (non-zero) above-ground biomass in 2100 under RCP 8.5 was expected to be greater than or equal to 2100 values under simulations with no climate change. More details on the modeling can be found in the publication link below.

There are some important assumptions and caveats to keep in mind when using these data. First, the simulations do not include any land-use or management activities – factors that can interact and influence forests as much if not more than climate change by itself. Second, these simulations were based on calibrations that considered forest processes across all of New England, not just Maine. Thus, these results represent simulations based on broad-scale patterns and may be best suited for assessing broad spatial patterns. Third, the climate variation used to parameterize the simulation models in Duveneck and Thompson (2017) was not free to vary across every cell and thus may not fully represent the full suite of factors that can influence climate change refugia. The authors represented climate variability with 25 climate regions, representing distinct areas of monthly temperature and precipitation; these climate regions may represent coastal and broad elevational influences on climate change refugia well, but not other factors that occur at the data’s spatial scale of 250m. Furthermore, the simulated conditions in each 250 m cell represents the overall pattern simulated. Finally, these results do not represent or express any measure of certainty. As such the results should not be taken as absolute prediction, rather as a tool for assessing the potential relative occurrence for these species in the future.

DOI for Duveneck & Thompson Refugia Products collection: https://doi.org/10.35482/ccranp.002.2019