Wednesday, August 5, 2009 - 4:20 PM

COS 81-9: Historical disturbance driven dynamics in Central Texas woodlands

Darrel B. Murray1, Carl Schwope2, and Joseph D. White1. (1) Baylor University, (2) Balcones Canyonlands National Wildlife Refuge

Background/Question/Methods

Spatially extensive analysis of landcover change over time is difficult to assess due to data limitations.  For ecosystem management, landcover change is important as a measure of landscape variability, habitat persistence, and vegetation successional dynamics.  In this study, we used historical aerial photographs to estimate landcover change from the 1930’s to the present for Central Texas woodlands.   We also showed that these data can be used to infer disturbance source based on resultant patch size and shape changes over time.  For the Balcones Canyonlands National Wildlife Refuge, a 7000 ha reserve in Central Texas, aerial photographs from 1939, 1951, 1964, 1980, 1995, and 2004 were digitized, georeferenced, normalized, and classified into woodland/non-woodland areas.  Landscape metrics calculated from these data included total woodland area, number of patches, mean patch size, perimeter to area ratio, and patch edge complexity.  This analysis showed that disturbance histories that impact woodland cover are variable in temporal intensity and extent over the refuge. 

Results/Conclusions Analysis indicated a moderate increase in woodland area from 62.0% to 64.2% (156 ha) from 1939 to 2004.  This also showed a general trend of increasing landscape stability. From 1939 to 1951, 63% of the study area remained the same cover type and 37% changed to a different type, while during the time period from 1995 to 2004, 76% remained the same and 24% converted to a different cover type.  Of the woodlands present in 1939 (4407.2 ha), 29.1 % persisted throughout the study period as woodlands. Refuge tracts with different treatments of cutting followed by burning and burning only were utilized as training sites to determine if landscape metrics could be used to infer past disturbance type.  Perimeter to area values were found to be significantly related with values above and below 24500 represented cutting/burning and burning respectively.   Back application of these metrics to all photos showed that burning was the dominant disturbance type between 1939 and 1951 affecting 30% of the refuge tracts.  Cutting and burning occurred with the highest frequency between 1951 and 1964 with 70% of the refuge tracts being affected.  A trend of generally less disturbance followed 1964 until the time period between 1995 to 2004, when management activities resulted in increased burning and cutting, with a combined incidence of 60%.