Twenty three years ago Dr. Bill Platt and others at Tall Timbers Research Station and St. Marks National Wildlife Refuge started a study to test hypotheses about effects of biennial fires applied at different seasons in sandhill and flatwoods longleaf pine woods in north Florida. The experiment includes eight season-of-burn treatments spaced at equal intervals throughout the year. There are two replicates per habitat and treatments were randomly assigned to 2-5 ha plots within habitats. Various types of data have been collected, but the most comprehensive with respect to ground layer composition involved species frequency data obtained from randomly located, permanently marked, 0.25 m2 circular subplots. One hundred of these subplots were located in each fire treatment plot for a total of 3200 total subplots, 1600 in each habitat.
The subplots in the sandhill habitat were first sampled during the growing season of 1988, eight years after the study was initiated. Sampling consisted simply of listing all the vascular plant species occurring in each subplot. The abundance of any given plant species for a particular burn treatment plot was then determined by the number of subplots that it occurred in. Results of this original census, published in the 1993 Tall Timbers Fire Ecology Proceedings, showed little, if any, effect of the seasonal burn treatments on ground-layer vegetation composition in either habitat. This was not too surprising, however, since results of the initial survey were confounded by differences in pre-existing vegetation. Furthermore, since most of the plant species are long-lived perennials that resprout following fires, we hypothesized that long-term observations might be necessary to document the expected trends.
We re-sampled the sandhill subplots during the 2000 and 2001 growing seasons, 12-13 years after the original survey. Results are presented herein for the 25 most abundant sandhill perennial species: 5 grasses, 11 forbs, 5 legumes, and 4 woody species including three small shrubs and bluejack oak sprouts. Our results continue to confound conventional wisdom. The first surprising result is the overall stability, at a plot or experiment scale, of most of the species. Of the 25 species, 17 had not changed significantly in overall frequency (i.e., averaged across burn treatment plots) during the 12-13 years between surveys. In the case of certain species, this overall stability reflected an actual lack of change. This was exemplified by wiregrass (Aristida beyrichiana), the dominant grass in our plots. In 1988, wiregrass occurred, on average, in 82 of 100 subplots. In 2001 this figure was unchanged. Furthermore, there was little recruitment or disappearance of wiregrass from individual subplots. This was true regardless of burn season.
Bluejack oak (Quercus incana) was another species for which overall stability was related to very low small-scale turnover. However, for most of the other species, stability at the population level was a function of balanced recruitment and extinction processes at either the subplot or plot scale. That is, most of these species, especially the forbs and legumes, were very dynamic, but were nevertheless in equilibrium at the level of the study as a whole.
Of the eight species that changed significantly in overall abundance during the course of the study, six declined and two increased. The declining species included one grass (Schizachyrium scoparium), one small rhizomatous shrub (Gaylussacia dumosa), one legume (Galactia floridana), and three forbs (Chrysopsis gossypina, Pityopsis flexuosa, and Phlox floridanas). These declining species are somewhat of concern since three of them have restricted ranges (Pityopsis flexuosa is limited to a handful of counties in north Florida. G. floridana and P. floridana occur in north FL and southern GA) and are consequently conservation priorities. The Galactia and Phlox declines were moderate (i.e., 2-3 subplots/plot) and probably primarily represent effects of the severe drought of the mid-late 1990ís. The Pityopsis flexuosa decline was more severe, i.e., an average loss of 7 of 18 subplots or 39% of the 1988 population. The worst decline was in the July treatment plots where Pityopsis flexuosa disappeared from an average of 14 subplots. Though not quite statistically significant as an effect of burn season due to high overall variation, this drastic loss of Pityopsis flexuosa from the July plots is worrisome and requires further investigation. It may be that for Pityopsis flexuosa, at least, frequent uninterrupted growing season fires represent a threat to long-term population viability.
The two increasing species included Quercus minima, a rhizomatous shrub, and Stylosanthes biflora, a small legume. The legume was not surprising, since frequent fires that volatilize nitrogen should favor legumes. In fact several other legume species increased as well, though not quite significantly. A fairly substantial increase in Quercus minima (from 16 subplots on average in 1988 to 20 in 2001) was unexpected and indicates the very high fire tolerance of this runner oak.
Out of the 25 species, only two responded significantly to burn season. Like Pityopsis flexuosa, Elephantopus elatus declined sharply in the July burn treatment plots (mean disappearance rate=11 subplots) producing a significant overall season of burn effect in the ANOVA for that species. Unlike Pityopsis flexuosa, Elephantopus increased in a sufficiently large number of other plots to counterbalance the negative impact of July burning; consequently the overall (i.e., experiment-wide) population of Elephantopus did not decline significantly.
Lastly, Liatris tenuifolia and other Liatris spp. (mostly elegans and provincialis) both declined sharply (-13, -11, respectively) in the April-burn plots, producing a significant overall season of burn effect for both of these species.