Document: DOR-3-99-130

Comparing restoration methods using similarity metrics in Florida longleaf pine sandhills.

GORDON, D.R.* ^1,3, L.PROVENCHER ², A.R.LITT ², G.W.TANNER ^1,3, L.A.BRENNAN ⁴ and J.L.HARDESTY ^1,3

The Nature Conservancy, Gainesville, FL 32611 USA ¹
The Nature Conservancy, Niceville, FL 32588-0875 USA ²
University of Florida, Gainesville, FL 32611 USA ³
Tall Timbers Research Station, Tallahassee, FL 32312 USA ⁴

Abstract:
Experimentally evaluating the success of restoration techniques against a reference condition is not directly possible because reference sites are not randomized or replicated, and therefore are not within the experimental design. We addressed this issue in fire-suppressed longleaf pine, Pinus palustris, sandhill communities using a large-scale experimental block design. The experiment involved comparing the effectiveness of three hardwood removal techniques (fire, herbicide, and mechanical) for restoration of understory and habitat components from 1994 and 1999. We measured soil, vegetation, and animal variables in restoration treatment and reference sites. Similarity of variables in restored plots to those in the reference sites was evaluated using proportional similarity and endpoint difference, a similarity metric we developed that is based on the t-statistic. The two similarity metrics will be illustrated using data from the sandhill plots. These metrics can be used to distinguish among species or variables that dominate similarity indices and less common species that strongly associate with certain conditions, such as high fire frequencies.

Keywords: restoration

This abstract is being presented at: 11:30 AM in session:
Oral Session #60: Forest Restoration.