Jose L. Silva-Lugo, George W. Tanner
The Journal of the Torrey Botanical Society 137 (2), 263-276, (1 April 2010) https://doi.org/10.3159/09-RA-058R1.1
KEYWORDS: cluster analysis, control and treatment sites, discriminant analysis, F-ratio test, fire effects, scrubby flatwoods, site analysis
To study the effects of prescribed burning on long-unburned scrubby flatwoods in Cedar Key Scrub State Reserve, FL, USA, we carried out a site analysis to determine if two controls and two treatment sites at preburn conditions did not differ regarding soil, slope, fuel characteristics, plant species structure and composition, and fire history. Sites had the following characteristics: (a) three sites had the same type of soil and the other site a similar one, (b) 0–2% slope, (c) they had not been burned since 1955, and (d) they did not receive mechanical treatment. Therefore, only fuel characteristics and species structure and composition needed to be tested. We installed a grid of 150 × 150 m in each site to quantify moisture content, abundance, and cover. Samples of dry and live fuel were taken in 20 points selected at random to estimate moisture content. We also recorded the abundance and cover of 31 herb and woody species in 50 quadrats (4 m2) selected at random in each site. Moisture content of dry and live fuel did not have significant differences between control and treatment sites. Species richness, diversity, evenness, and species-area curves indicated structural and compositional difference among sites. Similarity coefficients showed some structural similarities among sites. Since these indices drew contradictory results, we performed a cluster analysis, a discriminant analysis, and an F-ratio test to study whether there were differences or not among sites. We found that control and treatment samples grouped in one cluster. Out of 52 variables, only the median abundance of Vaccinium myrsinites Lam. was significantly different between control and treatment sites. Therefore, control and treatment sites were estimated to be ecologically similar, and this methodology could be used for testing control sites for fire ecology research.