The role of Ecological System mapping for assessment and stewardship applications in Texas

Background/Question/Methods

We are developing a vegetation map for Texas based on, and extending, the Ecological System Classification of NatureServe. We have currently completed more than half the state. The types are mapped using image objects, derived from NAIP imagery resampled to 10m resolution and using eCognition software. Landcover is developed using a decision tree classification applied against LandSat Thematic Mapper imagery from 3 dates along with abiotic variables including slope, aspect, landscape position, and solar insolation. Landcover is then applied to the 10m image objects along with edaphic information derived from SSURGO soils layers and landscape variables including slope and landscape position. From knowledge of the landcover occupying particular abiotic site types, image objects are attributed with appropriate vegetation types.  Currently we have mapped about 280 vegetation types for approximately the eastern half of Texas. These vegetation types include what might be considered “subsystems,” in that they represent the various landcovers that are aggregated within the concept of a give Ecological System. Additionally, semi-natural or ruderal types dominated by native and/or non-native species are mapped.

Results/Conclusions

This layer has proved useful for several ecological assessments ranging from endangered species habitat assessments, to targeting areas for evaluation of restricted habitat types, to evaluation of fuel models over extensive landscapes. The layer is also available for regional assessments such as Ecoregional Assessments where limited data has previously been available.

The fine thematic resolution of the data allowed us to discriminate between juniper (Juniperus ashei) dominated evergreen woodlands and evergreen woodlands dominated by live oak (Quercus fusiformis). This distinction may prove important, since the endangered Golden-cheeked Warbler has a close association with juniper, but not necessarily with live oak. We have also used change detection over a 5 yr period to quickly update models to reflect recent clearing.

Acquisition and assembly of soil and landform data has allowed us to identify areas where restricted habitats may be present, allowing prioritization of search effort. As an example, we were able to identify ravines in East Texas that harbor occurrences of the West Gulf Coastal Plain Mesic Hardwood Forest.