Climate Change Tree Atlas & Bird Atlas

The model inputs are pre-defined by the researchers, so no user-collected data is required. For the Tree Atlas, these defined inputs include data on current tree species distributions from the Forest Inventory and Analysis (FIA) program, as well as datasets that fall under four main categories: climate, elevation, soil characteristics, and land use. For the Bird Atlas, inputs include data on current bird distribution from the Breeding Bird Survey as well as data on climate, elevation, and tree species distributions. See the Climate Change Atlas website for a full list of inputs.

Tool outputs primarily include maps (and numerical tables) of current and future suitable species habitat. Measurements of suitable habitat distribution and change are given in importance value (trees) and incidence (birds). Importance value is a measure of abundance that accounts for both tree basal area and number of stems, ranging from 0 - 100. Incidence is calculated from the number of years (from 1981-1990) that a species was observed on the Breeding Bird Survey routes selected by the USGS. Incidence value ranges from 0 to 1.

The Atlas should be used at an appropriate scale; analyses are done at the level of 20x20 km cells, so fine scale interpretations may be inappropriate. It is also important to note that there are several uncertainties inherent in this type of species distribution model. For example, there is a considerable range in projected future climate conditions depending on which climate model is used and which emissions scenario is chosen to represent future conditions. The Atlas deals with this by providing projected suitable habitats for a range of different climate models and emissions scenarios.

In addition, the model predicts suitable habitat better for some species than for others. A reliability index is included for tree and bird species to reflect this. The reliability scores are based on the statistical techniques used to create the model, but basically take into account how good the predictor variables are at describing where species are located. Projections for species with low reliability scores should be interpreted with caution.

Finally, the model is only predicting where suitable species habitat may be in the future. It cannot predict where species themselves will be located, since that depends on factors such as species migrations, land use changes, biological factors (e.g. regeneration, dispersal, competition) and disturbances (e.g . fire, insects, pollution), all of which are difficult to quantify. Researchers are currently working on both quantitative and qualitative tools that help to account for some of these uncertainties.

To see a full list of strengths and limitations for the DISTRIB model used in the Atlas, please see this list.

Researchers at the USDA Forest Service Northern Research Station (NRS) first developed a statistical model in 1996 to assess potential changes in habitat for common tree species in the eastern United States. Since then, the model has gone through several improvements, leading to the development of the DISTRIB model which is the basis for the Climate Change Atlas. See the Atlas publications page for a list of documents related to the development and use of this model.