Wood Thrush has already been covered, with this animated map and discussion. This feature will explore some of the differences apparent in visualizing the data at different scales.
The STEM models produce data–vast amounts of data. These data can be summarized in many different ways (for example, we also sometimes produce graphs showing habitat relationships, or graphs exploring the relative importance of different landscape variables in predictions in different areas.) Our favorite way to explore these results is on a map, and these are the orange animations on the black maps that you have seen in these stories.
Of course, there are lots of different ways to create a map from the data. The predictions are made for 130,000 random points across the country, and these can be aggregated at any scale. Our typical maps, like the one above for Wood Thrush and this one for Brown-headed Nuthatch, aggregate the data at a 30km scale. Below is an example, also from 28 June, at the 30 km scale.
Summer (28 Jun) — click image for larger version
The fine scale image, by contrast, is an experimental run that aggregates the data at a 3km scale. This gets much closer to the scale at which birds make decisions about individual territories, and overall gives much better resolution to our maps. Here are a few things to notice that indicate that the maps is very accurate at these fine scales:
1) Look at the cites: notice how well defined the urban zones are for Boston, Hartford, Springfield (MA), New York, Baltimore, Washington D.C., Richmond, Charlotte, etc. Wood Thrushes do not fare well in urban landscapes, since what woodlots remain are often too fragmented for successful breeding.
2. Agricultural zones. Notice how most of northern Indiana and northern Illinois show up as dark, since those are regions of large-scale agriculture with limited forest. Some of the areas with woodlands do appear as brighter spots in this sea of darkness, as seen in southwestern Wisconsin.
3. Look at western Virginia, eastern West Virginia, western Maryland, and eastern Pennsylvania. Several strips of dark show up like rivers flowing through the dense orange. In western Maryland, the more western valley is the Hagerstown Vally, and the more easterly one is the Frederick Valley. This is the “Ridge and Valley” province, where the remnants of the ancestral Appalachians show up as narrow linear (forested) ridges, and valley floors that are mostly cleared for agriculture. This geological formation is reflected in the occurrence of Wood Thrush, since they remain common on the ridges but no longer occur in the valleys since there is a paucity of habitat. Similar patterns can be seen in central Massachusetts and Connecticut, where the urban and agricultural Connecticut River Valley cuts a swath of degraded Wood Thrush habitat through the center of those two states.
4. Notice how areas of extensive pine forest, like much of Cape Cod and the New Jersey Pine Barrens, show areas of lower predicted occurrence for Wood Thrush. Wood Thrushes really prefer deciduous habitat, so these are in fact areas where Wood Thrushes are much less common.
In the future we hope to be able to generate these maps for more species, but they are more difficult and costly to produce. We present this one here as an example of what may become possible in the future. Research into these models is ongoing at Cornell, and we hope you share our excitement.