Plant colonization after complete and partial removal of ... - Springer Link

WETLANDS, Vol. 23, No. 4, December 2003, pp. 1015–1029 䉷 2003, The Society of Wetland Scientists

PLANT COLONIZATION AFTER COMPLETE AND PARTIAL REMOVAL OF DISTURBED SOILS FOR WETLAND RESTORATION OF FORMER AGRICULTURAL FIELDS IN EVERGLADES NATIONAL PARK George H. Dalrymple1, Robert F. Doren2, Nancy K. O’Hare1, Michael R. Norland3, and T. V. Armentano3 1 Everglades Research Group, Inc. 21425 S.W. 368 Street Homestead, Florida, USA 33034 2

National Park Service, Southeast Environmental Research Program Florida International University Miami, Florida, USA 33199 3

South Florida Natural Resources Center Everglades National Park Homestead, Florida, USA 33034

Abstract: The Hole-in-the-Donut is a 4000-ha region of former farmlands within Everglades National Park that is dominated by a monoculture of the non-indigenous pest plant Schinus terebinthifolius (Brazilian pepper). Prior to extensive farming in the region, the area consisted of short hydroperiod graminoid wetlands and mesic pine savannah. Rock plowing in preparation of these lands for farming created an artificial soil layer that broke up the limestone substrate, mixed and aerated the native marl soil layer with the broken limestone, and elevated the surface slightly. Farming practices also included the use of chemical fertilizers and pesticides. The modified soil substrate quickly became dominated by S. terebinthifolius when farming ceased in 1975, despite efforts to control its establishment, such as prescribed fire, herbicide treatment, and mowing. Preliminary evidence indicated that soil removal would prevent re-invasion by S. terebinthifolius and could lead to colonization by native wetlands plants. Two trials, a partial soil removal (PSR) and a compete soil removal (CSR), were performed on a pilot test site beginning in 1989 to determine whether all or only a portion of this modified soil substrate needed to be removed to attain desired results. Removal of rock-plowed surface material lowered elevation in both treatments. While the PSR treatment did show an increase in the number and coverage of hydrophytes for a few years, it did not prohibit re-colonization and re-establishment of a canopy of S. terebinthifolius, and by 1996, the site was again dominated by a monoculture of S. terebinthifolius. By contrast, the CSR treatment showed rapid colonization by hydrophytes and no successful re-colonization by S. terebinthifolius. Lowering elevations by 15 to 45 cm allowed for longer periods of flooding and rapid colonization by hydrophytes on both sites. After the sites were cleared, the average difference in elevation between the two treatment areas was less than a tenth of a meter, but this resulted in a slightly shorter hydroperiod on the PSR site. The small amount of residual rock-plowed soil with high levels of nutrients, along with its slightly shorter hydroperiod on the PSR site, appear to have contributed significantly to the success of S. terebinthifolius in re-colonizing this treatment area. Key Words: wetlands, restoration, agricultural fields, soil removal, Everglades National Park, exotic plants, Schinus terebinthifolius, Brazilian pepper

INTRODUCTION

Ewel et al. 1982, Krauss 1987). Mechanical rock plowing of farm fields was common after 1950 and was used to clear approximately 2400 ha of this region (Ewel et al. 1982, Krauss 1987). This produced a better substrate for farming because the broken rock surface reduced flooding, increased aeration, and al-

Farming in the Hole-in-the-Donut (HID) region of Everglades National Park (ENP), Florida, USA between 1916 and 1975 altered approximately 4000 ha of natural vegetation, including short hydroperiod prairies and pinelands (Loope and Dunevitz 1981, 1015

1016 lowed increased levels of chemical pesticides and fertilizers to be applied for row crop agriculture (Orth and Conover 1975, Orth 1981). When farming practices ended in 1975, the non-native pest plant Schinus terebinthifolius Raddi (Brazilian pepper) aggressively colonized the more intensively farmed portions of the HID. Currently, S. terebinthifolius forms an almost impenetrable thicket over most of the former agricultural lands (Ewel et al. 1982, Krauss 1987, Doren et al. 1990). Hydroperiod and substrate modifications, including nutrient enrichment of the farmed soils, facilitated the invasion of S. terebinthifolius in this area (Loope and Dunevitz 1981, Ewel 1986, Li and Norland 2001). Schinus terebinthifolius is a tree or shrub that grows to 12 m in height, with multiple arching branches. The plants are usually wider than tall. They are usually dioecious with alternate odd-pinnate compound leaves. The plants develop bright red clusters of fruit during the months of November to January (Ewel et al. 1982, Doren and Whiteaker 1990a). The species is listed as facultative wetland (i.e., is equally likely to occur in wetlands or uplands; Reed 1988). Ewel et al. (1982) did extensive studies of seed dynamics, germination rates, seedling translocation, seed dispersal, and seed banks. The majority of germination took place in January and February. Germination rates were as high as 60%, and seedling survival ranged from 30% to 66% in field studies of tagged plants (Ewel et al. 1982). High rates of seed germination were found in all age classes of stands, including young successional stands (10 years or less since abandonment of farming), old successional stands (11 to 20 years since abandonment), and even in mature stands (more than 20 years since abandonment; Ewel et al. 1982, Doren and Whiteaker 1990a, b). Seedling densities ranged from ⬍ 10 to ⬎ 200/m2 (Ewel et al. 1982). In seedling transplantation studies, seedling survival was more than 50% at eight of ten sites, and on some of these sites seedlings increased six- to ten-fold in height in only two years following transplantation (Ewel et al. 1982). In their seed bank study, Ewel et al. (1982:115) found ‘‘ . . . no evidence that schinus (sic) retains its viability for many months after it reaches the soil . . . ’’. They noted no germination of S. terebinthifolius from soil seed bank samples after three months and emphasized the importance of annual dispersal of seeds, primarily as seed drop and dispersal by birds (especially robins) and raccoons, for the presence of high levels of new seedlings. Attempts at chemical control of S. terebinthifolius stands alone and together with prescribed burning resulted in some short term successes, but abundant reproduction and high survival rates of seedlings allowed S. terebinthifolius to maintain its position on

WETLANDS, Volume 23, No. 4, 2003 sites (Doren and Whiteaker 1990a). Since dense canopies of S. terebinthifolius in old successional and mature stands reduced understory plant cover to very low values (Ewel et al. 1982, Doren and Whiteaker 1990a, b), the sub-canopy fuel loads were too low to carry fires into these stands. Live fine fuel moisture levels of 50% to 60% and a high percentage of non-volatile live fuels made it extremely difficult to use fires to burn S. terebinthifolius stands effectively (Doren and Whiteaker 1990b). In many cases, fires had to be restarted several times, and often, the fires did not penetrate far into S. terebinthifolius stands (Doren and Whiteaker 1990b, pers. obs.). Even in those cases where backing fires with a higher level of dead fuels resulted in effective burns, rapid and high rates of seed germination and high levels of seed survival saw these sites rapidly re-develop into S. terebinthifolius stands (Doren and Whiteaker 1990b). The resistance to chemical treatment and prescribed fire, as well as the effective seed dispersal, high seed germination rate, high survival rate of seedlings, and rapid growth of seedlings support the conclusions of Loope and Dunevitz (1981), Ewel et al. (1982), and Krauss (1987) regarding recolonization and self-maintenance by S. terebinthifolius in this area (Doren and Whiteaker 1990a:33). All previous attempts to remove S. terebinthifolius were of limited success, or were economically infeasible to continue on a larger scale (Doren and Whiteaker 1990a, b, Doren et al. 1990). Even Hurricane Andrew in 1992 had only minor short-term effects on existing stands of S. terebinthifolius, and these stands recovered rapidly after that storm (Armentano et al. 1995). Preliminary findings in both the Rocky Glades region of the eastern Everglades and in some old abandoned fields in the HID indicated that ‘‘scraping’’ a site (i.e., mechanical removal of existing S. terebinthifolius and underlying rock-plowed rubble and substrate) resulted in the natural recolonization of native wetland plant species (Doren et al. 1990, Dalrymple et al. 1993). These findings led to a cooperative agreement between Miami-Dade County and ENP to use mitigation funds from Miami-Dade County for a pilot restoration project in the HID to evaluate the feasibility of this approach to wetland restoration and mitigation. The results from the pilot project, reported on herein, led to a cooperative agreement to apply the technique to all of the S. terebinthifolius-dominated, rock-plowed lands in the HID as part of a long-term wetland restoration program using wetland mitigation impact fees collected in portions of adjacent MiamiDade County. One administrative stipulation in the design of the pilot project was that the chosen site be far removed from visitor activities, which led to the choice of a remote site that was completely surrounded by S. tere-

Dalrymple et al., WETLAND RESTORATION IN THE FLORIDA EVERGLADES

1017

Figure 1. Map showing location of study site in relation to the remainder of the Hole-in-the-Donut, and surrounding habitats in Everglades National Park.

binthifolius forest. This was a compromise because it had been recognized that a site with a direct connection to the natural short hydroperiod prairies would probably increase the rate at which native plants would colonize the site. However, the pilot project was finally planned to address only the question of soil removal and not the effects of adjacent vegetation types. An additional stipulation was that the pilot project include two distinct soil removal treatments: complete soil removal (CSR) from 18 ha and partial soil removal (PSR) from the remaining 6 ha. The PSR was a cost-conscious alternative requested by park administrators to determine whether a partial lowering of elevation by PSR would be sufficient to prevent S. terebinthifolius re-establishment and permit natural colonization by native wetland plants. While direct observation of the plants and wildlife using the site convinced local land managers to agree to the wide scale application of the scraping technique for restoration of the HID, no detailed analyses of the eight years of data collected on this pilot site have previously been made available. Analyses of the data from the pilot site should assist us in identifying trends in plant colonization that are considered beneficial and desirable and help us to develop new sampling and experimental designs that will permit a hypothesis-

testing framework for evaluating succession theory in the HID (Tilman 1988, Kent and Coker 1994, Bazzaz 1996). This is particularly relevant, given the largescale restoration efforts now underway in the HID. METHODS Study Site The 24-ha study site was approximately 650 m from north to south and 450 m from east to west. It was located 335.3 m ⫾ north of the southeast corner of Section 36, Township 58S, Range 36E. The site was in the vicinity of site 10 of the S. terebinthifolius forest studied by Ewel et al. (1982). However, they did not list geographic coordinates, so it was not possible to map this older study site. Ewel et al. (1982) described the area as rock-plowed farmland abandoned in 1973, with ground-surface elevations between 1.2 m and 1.52 m NGVD 1929. The disturbed soil depths on this site were considered to be between 15 and 48 cm (Doren and Whiteaker 1988). The site was unequally divided into two treatments (Figure 1). The northern 18 ha had all vegetative cover and underlying rockplowed substrate removed down to bedrock (complete soil removal or CSR treatment) in early 1989. The

1018

WETLANDS, Volume 23, No. 4, 2003

Figure 2.

Aerial photograph of the study site taken in July 1992.

southern 6 ha had all existing vegetative cover removed; however, a thin layer of the rock-plowed substrate was left intact on the site (partial soil removal or PSR treatment). This was accomplished by setting the bulldozer blade approximately 6 cm higher than was used for the complete soil removal treatment. The site was surrounded by S. terebinthifolius on all sides and had no direct contact with undisturbed, natural vegetation (Figure 1 and Figure 2). Plant colonization occurred by natural recruitment, and no seeding or planting of any species was done. With the exception of hand removal of S. terebinthifolius saplings once in 1995, no management of undesirable plant species has occurred. Sampling Procedures Plot Layout. A series of 63 10 m ⫻ 10 m uniformly spaced plots was established to monitor plant colonization. The plots were 60.9 m apart and arranged in a grid of 9 rows (north to south), with each row having seven plots equally spaced east to west. Forty-nine of the plots (the northern seven rows) were located on

the CSR site. Fourteen plots (the remaining two southern rows) were located on the PSR site. Soil Depth and Elevation. The exact amount of soil on the sites before soil removal and the amount left on after treatment were not measured directly. Instead, after all the soil was removed from the CSR treatment, the bulldozers blades were set approximately 6 cm higher for partial soil removal on the PSR site. Likewise, ground-surface elevation was not measured before soil removal began. After treatment, a single point elevation was measured at the center of the northern edge of each plot. Elevation was determined using traditional land survey methods. Hydrologic Data. Study sites were far removed from any flood-control structures, including water-management canals, and therefore, the only source of water for surface flooding for the site was from seasonal rainfall. In this report, annual patterns of hydrologic variables were analyzed for hydrologic years, which run from June 1st to May 31st of the following calendar year. A well with a solar-powered recording gauge was

Dalrymple et al., WETLAND RESTORATION IN THE FLORIDA EVERGLADES Table 1. Percent cover ranges of the Daubenmire scale (cf. Mueller-Dombois and Ellenberg 1974:63) and midpoint values used in recording and quantifying relative cover in each height class. Percent Cover 1% 5% 25% 50% 75% 95%

to to to to to to

5% 25% 50% 75% 95% 100%

Midpoints 2.5% 15.0% 37.5% 62.5% 85.0% 97.5%

1019

study, all plant taxonomic names were updated to follow Wunderlin (1998).The Federal list of wetland plants (Reed 1988) was followed to categorize the plant species as hydrophytes (i.e., plants that are listed as Obligate (Obl) or Facultative Wetland (FacWet) species) for comparisons of the two treatments for success at re-establishing a wetland flora. The percent of total cover by hydrophytes per plot was calculated by dividing the percent cover of these wetland categories by the total cover per plot and multiplying by 100. Statistical Analyses

located at the center of the CSR site in late 1989 to monitor daily elevation of the water table, regardless of whether it was above or below the ground-surface elevation. Elevation of the water table was used relative to the average plot elevations on CSR and PSR to estimate duration and depth of surface flooding on sites each hydrologic year from June 1, 1990 to May 31, 1997. Vegetation Sampling. Plant community associations in the plots were evaluated each fall (August) from 1989 to 1996 using the Braun-Blanquet method (Mueller-Dombois and Ellenberg 1974). Each species was recorded in a series of five height classes, and an estimate of coverage for the species was recorded in each height class it occupied. Plants were assigned to height classes based upon their life form. The standard reference was Reed (1988), which lists the life form of wetland plants. Height class 1 was for species described as ‘‘submerged,’’ and height class 2 was for species described as ‘‘vines.’’ All other species must occur in height classes 3, 4, or 5. Height class 3 was for plants less than or equal to 1 meter in height, class 4 for heights up to 2 meters, and height class 5 for plants over 2 meters tall. Cover codes chosen for data collection were midpoints of the Daubenmire scale (Table 1; MuellerDombois and Ellenberg 1974). The sampling method required in this project was for the cover in every height class to be quantified. A plant that was not assigned to either the submersed or vine height classes could theoretically occur in all remaining height classes. In each plot, the amount of coverage the plant occupied in each height class it occupied was quantified, not just its coverage in the highest height class it occupied. This resulted in sums of coverage values for taxa that may exceed 100 percent. In this respect, the coverage values should be seen as a relative index of total coverage by plant taxa. Taxonomic Names and Wetland Status. Because numerous taxonomic name changes occurred during the

Data from the annual Braun-Blanquet sampling were used for all analyses of vegetation. Species richness and abundance, as well as calculation and description of Whittaker curves (or dominance-diversity curves; see Whittaker 1975, Colinvaux 1986) were used to evaluate patterns of dominance and diversity, both in the overall data sets and in individual vegetative height classes or layers within the coverage (submerged, vine, less than one meter, etc.). Statistical testing followed standard procedures (Sokal and Rohlf 1995, Zar 1996) using ‘Statistica’ (StatSoft 1994). Whittaker Curves for both partial soil removal (PSR) and complete soil removal (CSR) were generated and compared within and between years. These curves were calculated for the entire data set per treatment per year and by separate height classes to identify shifts in species ranking in different height classes among the plots. Individual species’ ranking in these curves was listed and compared between treatments and between years to evaluate how much species turnover and shift in dominance ranking had occurred. The details of these patterns are described and discussed in the following section. Species abundances or Whittaker curves were based on the Importance Values (IV) defined as: IV ⫽ (Cover sum ⫻ Plot frequency) ⫼ (Maximum cover ⫻ Number of plots) where: Cover sum ⫽ sum of cover values recorded for a taxa or height class in all the plots, Plot frequency ⫽ total number of plots in which the taxa occurred, Maximum cover ⫽ maximum possible total cover a taxon could obtain using modified Daubenmire scale (97.5%), and Number of plots ⫽ number of plots on treatment (49 plots on CSR; 14 plots on PSR). The importance values were a useful method of correcting for variation in the number of plots occupied by taxa, and they give a more realistic measure of dominance. In other words, the importance value corrects for the fact that some species may have high total coverage values but occur in few plots.

1020


Distribution by height class and percent cover codes for selected species were evaluated in relationship to plot location and soil treatment. Correlation coefficients and simple linear regressions on elevation of individual species by height class were calculated to determine how many species showed significant positive and negative associations between coverage and elevation. Finally, a more advanced method of ordination by factor analyses provided an additional independent measure of the degree of difference among vegetation plots based upon the raw data for coverage by every taxa height class. Data for factor analysis was the nonstandardized percent coverage of each taxa height class in each plot for each year. The method was simple principal components analysis (with no rotation of the factors) using either the squared Euclidean distances or percent disagreement distances. These analyses also served as an effective evaluation of spatial associations (Gauch 1982, Pielou 1984). Species-area curves were generated for each sampling year to permit comparisons of the total number of species observed in the total number of plots each year. Logarithmic transformations (log base 10) were used to calculate species area curves based on the standard formula, S ⫽ cAZ, where: S ⫽ number of species and A ⫽ area in meters (see MacArthur and Wilson 1967). To generate these estimates of total number of species accumulated as sampling area increases, and to avoid the potential for any bias from spatial autocorrelation, plots were added to the curve in a random selection order. If the y-intercept ‘‘c’’ remains a constant, then the higher the value of ‘‘Z’’, the lower the amount of land area required to double the number of species. Alternatively, if the y-intercept increases with time, the number of species in the smallest theoretical plot is greater. Curves were also used to compare soil treatments for species richness at any given sampling area. For example, the CSR area included forty-nine 100-square-meter plots, while the PSR area included only 14 plots; so the curves permit the evaluation of whether the total number of species expected on PSR and CSR would be the same for any given standard sampling area. All data have been entered into a permanent database maintained by Everglades National Park’s South Florida Natural Resources Center.

Figure 3. Histogram of plot elevations for the Complete Soil Removal (CSR) and Partial Soil Removal (PSR) treatments.

0.869 m (s. e. ⫽ 0.027), and the median was 0.904 m. There was a significant difference between these elevation values (as tested by both independent t-test: t ⫽ ⫺0.201, p ⫽ 0.0488; and Mann-Whitney U test: U ⫽ 166.0, adj. Z ⫽ ⫺2.926, p ⫽ 0.003). Plot 53, on the PSR treatment, was much lower in elevation than all of the others (Figure 3). This plot was at an elevation of 0.56 m. This plot sat at a low spot on the edge of a deep pond that was historically present in this area as part of a former small slough with cypress trees. The depth of this pond was actively maintained by adult alligators. If this plot was excluded from the calculations, mean and median elevation values for the PSR plots were higher: 0.893 m and 0.904 m, respectively. In general, while average and median values for elevation for the sites were significantly different statistically, the overall difference in elevation between the two treatments was only about 4.5 cm to 7 cm, depending upon whether plot 53 is included in the calculation. Unfortunately, depth of residual soil on the PSR treatment was not directly measured in previous studies. Based upon construction methods, residual soil on PSR should have been approximately 6 cm. This estimated soil depth is similar to the difference in elevation between the two treatments. Lowering both sites by approximately 15 to 45 cm in elevation allowed for much longer hydroperiods on both sites.

RESULTS Water Levels Elevation and Soil Depths on Treatments Mean elevation (NGVD 1929) of the 49 CSR plots was 0.824 m (s. e. ⫽ 0.009), and the median was 0.833 m (Figure 3). Mean elevation for the 14 PSR plots was

A small but significant difference in mean groundsurface elevation between the treatments was reflected in duration of the standing water, or hydroperiod, on the two sites. Hydroperiod was measured as the num-

Dalrymple et al., WETLAND RESTORATION IN THE FLORIDA EVERGLADES Table 2. Duration of hydroperiod on Complete Soil Removal (CSR) and Partial Soil Removal (PSR) treatments for hydroyears 1990–1991 through 1996–1997. Hydroyear

CSR

PSR

1990–91 1991–92 1992–93 1993–94 1994–95 1995–96 1996–97

130 169 196 208 244 253 173

107 148 166 172 215 228 142

ber of days per year with water at or above mean surface elevation on the two soil treatments (excluding plot 53 from the PSR treatment). In the driest year on record for the project (1990–1991), water was at or above the surface for only 130 days on CSR and 107 days for PSR, while in the wettest year (1995–1996), there was water at the surface for 253 days on CSR and 228 days on PSR. Mean duration of the annual hydroperiod on the CSR site was 196 days (s. e. ⫽ 16.41, n ⫽ 7), and on the PSR site, it was 168 days (s. e. ⫽ 15.91, n ⫽ 7; Table 2). The 28-day difference in the mean duration of standing water on the two sites was not significantly different statistically (t ⫽ 1.22, d.f. ⫽ 12, p ⫽ 0.25), but it appears to have been biologically important for control of S. terebinthifolius and colonization of some hydrophytic species (see below). Complete Soil Removal Treatment Average number of species per plot on the CSR treatment increased from 23.3 to 33.6 from 1989 to 1996 (Table 3). There was an appreciable drop in the average value for the August sampling in 1995 due to a prescribed fire on the site in May 1995. Average total coverage per plot increased from 84.6 in 1989 to 214.6 in 1996 (Table 4). The layers of vegetation (submerged, vine, etc.) were dominated by species in height class 3 (less than 1 m), with a few vines and some submerged vegetation present. The submerged layer became more abundant in 1995 (Figure 4). The 1995 fire led to a significant drop in coverage (down to 129.4), but the site recovered to nearly the 1994 level by 1996. Fire succeeded in reducing coverage by some woody taxa considered less desirable including Ludwigia octovalvis (Jacq.) Raven (Mexican primrose willow), L. peruviana (L.) H. Hara (Peruvian primrose willow), and Baccharis glomeruliflora Pers. (Silverling). Immediately following the burn, there was a slight increase in coverage by Typha domingensis Pers. (Southern cattail), a significant increase in the sun-loving annual Sesbania herbacea (Mill.) McVaugh (Dan-

1021

Table 3. Mean number of species per plot for Complete Soil Removal and Partial Soil Removal treatments from 1989 to 1996, with sample size (N) and standard error (s.e.). Complete Soil Removal

Partial Soil Removal

Year

N

Mean

s.e.

N

Mean

s.e.

1989 1990 1991 1992 1993 1994 1995 1996

49 49 49 49 49 49 49 49

23.3 25.8 31.0 27.8 32.6 31.9 26.1 33.6

0.66 0.56 0.65 0.63 0.57 0.58 0.60 0.66

14 14 14 14 14 14 14 14

22.1 32.3 20.7 20.1 25.1 19.7 17.2 21.9

1.05 1.38 1.05 0.94 1.39 1.34 0.94 1.13

glepod), and an increase in coverage by Muhlenbergia capillaris (Lam.) Trin. (Hairawn muhly grass). Hydrophytes (Facultative and Obligate wetland species) became progressively dominant through time on the CSR site. Average percent of total coverage composed of hydrophytes was 63.7% in 1989 and increased to 74.2% by 1996 (Table 5). Dominant wetland grasses and sedges in the nearby, undisturbed short hydroperiod prairies were Cladium jamaicense Crantz (Sawgrass), Rhynchospora microcarpa Baldwin ex. A. Gray (Southern beaksedge), Schoenus nigricans L. (Black bogrush), Muhlenbergia capillaris, Schizachyrium scoparium (Michx.) Nash (Little bluestem), and Andropogon glomeratus (Walter) Britton et al. (Bushy bluestem) (Olmsted and Loope 1984 and unpublished data). Many of these species showed marked increase in frequency of occurrence in plots, as well as coverage, as the study progressed. Cladium jamaicense was not found in any plots in 1989, 10 plots in 1990, and 21 plots in 1996. Rhynchospora microcarpa was not found in any plots in 1989, 10 plots in 1992, and 36 plots in 1996. Muhlenbergia capillaris was not found in any plots in 1989, 7 plots in 1991, and 29 of the 49 plots by 1996. Schizachyrium Table 4. Mean total cover per plot by all plant taxa for Complete Soil Removal (CSR) and Partial Soil Removal (PSR) from 1989 to 1996, with sample size (N) and standard error (s.e.). Complete Soil Removal


Year

N

Mean

s.e.

N

Mean

s.e.

1989 1990 1991 1992 1993 1994 1995 1996

49 49 49 49 49 49 49 49

84.6 156.2 246.3 207.1 280.3 242.7 129.4 214.6

4.28 5.81 8.96 7.87 9.56 8.25 5.19 6.45

14 14 14 14 14 14 14 14

197.9 238.4 321.9 230.9 252.1 245.4 254.1 232.3

13.67 14.27 20.54 11.59 7.91 9.36 19.90 14.64

1022


Figure 4. Percent of total coverage in plots by taxa in five Layers or Height Classes (Lyr ⫽ Layer) for CSR (4A) and PSR (4B) from 1989 to 1996.

scoparium was not found in any plots in 1989, 29 plots in 1991, and 42 of the 49 plots by 1996. Andropogon glomeratus was the dominant grass on the CSR treatment and was found in 47 of 49 plots by 1991. In contrast, Schoenus nigricans was never found on the site. Schinus terebinthifolius was found in 8 plots in 1989, 3 in 1990, and only 2 of 49 plots in 1996 (Figure 5). In all cases, the plants were seedlings or saplings. Very few seedlings survived due to annual flooding conditions. Due to the drought conditions at the commencement of this project (see above), many more seedlings germinated and survived flooding in the first Table 5. Percentage of total cover per plot by hydrophytes for Complete Soil Removal (CSR) and Partial Soil Removal (PSR) from 1989 to 1996, with sample size (N) and standard error (s.e.). Complete Soil Removal


Year

N

Mean

s.e.

N

Mean

s.e.

1989 1990 1991 1992 1993 1994 1995 1996

49 49 49 49 49 49 49 49

63.7 68.3 77.4 81.2 80.2 79.3 82.2 74.2

1.61 1.42 0.91 1.14 0.87 1.30 1.29 1.28

14 14 14 14 14 14 14 14

63.7 68.6 78.2 81.8 80.4 80.9 80.7 70.9

5.53 2.40 2.64 2.81 3.30 2.12 3.76 4.15

two years than was expected. When higher rainfall and water levels returned in hydrologic year 1991–1992, many seedlings were large enough to tolerate inundation. During the dry season of 1994–1995, prior to prescribed burn of the site, an attempt was made to locate every seedling or sapling S. terebinthifolius on the CSR site, regardless of whether they were in or outside of plots. A total of 505 plants were found. This was equivalent to a density of 0.003 per m2. The plants had an average height of 1.1 m (s.e. ⫽ 0.017) and canopy width of 1.07 m (s.e. ⫽ 0.03). In general, they were in poor condition, being chlorotic, with few branches and few leaves per branch. These plants were removed, and no additional S. terebinthifolius plants have been found on the site since 1995. Species abundance or Whittaker curves (see Methods section) are shown for the first year (1989) and last year (1996) of the study in Figure 6. Plots of importance values (IV, see Methods) permitted graphical comparison of number of taxa height classes and their abundance and importance for the first and last years of the study. Over eight years, there was a shift in the dominant plants. In 1989, dominant species were Sesbania herbacea, Ludwigia octovalvis, the calcareous alga Chara sp. (Muck-grass), and Spermacoce floridana Urb. (Florida false buttonweed) . In 1996, dominant species were L. microcarpa (Smallfruit primrose willow), Andropogon glomeratus, Schizachyrium sco-


1023

Figure 6. Whittaker or Dominance-Diversity curves for CSR treatment in 1989 and 1996. Plots show rank order scores of taxa height classes by their Importance Values (see text for calculation of Importance Value).

Figure 5. Number of plots (5A) and percent of plots (5B) with Schinus terebinthifolius in CSR and PSR treatments.

parium, and Bacopa monnieri (L.) Pennell (Herb-ofgrace). The 1996 curve was much flatter than the 1989 curve, reflecting an increase in evenness over time. Number of taxa by height class increased from 83 in 1989 to 133 in 1996, reflecting an increase in species richness and vertical complexity in the community. Partial Soil Removal Treatment Average number of species per plot on the PSR treatment showed a dramatic increase after one year, changing from 22.1 species per plot in 1989 to 32.3 in 1990. However, the average fell back to a lower level for the remainder of the study (Table 3). Average percent coverage per plot rose from 197.9 in 1989 to 321.9 by 1991, then fell to 230.9 in 1992 and remained at this level through 1996 at 232.3 (Table 4). While attempts were made to burn this site in May 1995, fire did not effectively penetrate this site and had no significant effect on percent cover in the plots.

Average percent of total coverage by hydrophytes was 63.7% in 1989 (identical to the value for the CSR treatment) and increased to 70.9 % by 1996 (Table 5). The vine layer (Layer 2) was much more abundant on this soil treatment than the CSR treatment (Figure 4). Layers greater than 1 m in height were also becoming increasingly abundant by 1991. The ineffective 1995 fire did not reduce these higher layers. The trend was for increasing coverage by taller woody species, dominated by S. terebinthifolius. Schinus terebinthifolius covered almost every square meter of the site by the dry season of 1994–1995. Average height of 20 randomly selected trees on the PSR site was about 3.18 m (s.e. ⫽ 0.246), and canopy width was 3.53 m (s.e. ⫽ 0.042), compared to 1.1 m for average height and 1.07 m for average canopy width on the CSR site. While C. jamaicense, M. capillaris, and Schizachyrium scoparium all showed increasing frequencies of occurrence in the CSR plots through time (see above), these species were extremely rare in the PSR plots. Cladium jamaicense was found in only one of 14 PSR plots during the study in 1995 and was missing again in 1996. Rhynchospora microcarpa was found in two plots in both 1993 and 1996. Schoenus nigricans was absent from the PSR site. Muhlenbergia capillaris was found in only one plot in 1996. Schizachyrium scoparium was never found in a PSR plot. Schinus terebinthifolius was found in seven plots in 1989, 10 plots in 1990, and 12 of the 14 plots in 1996 (Figure 5). Whittaker curves were generated for 1989 and 1996 (Figure 7). The degree of dominance (maximum IV value) within the plant community diminished somewhat through time. The dominant species in 1989 were Sesbania herbacea, Ludwigia microcarpa, and Spermacoce floridana, as was the case for the CSR treat-

1024

WETLANDS, Volume 23, No. 4, 2003 Table 7. Results of independent t-tests on mean total cover of all plants per plot for Complete Soil Removal (CSR, n ⫽ 63) versus Partial Soil Removal (PSR, n ⫽ 14) for 1989 to 1996; degrees of freedom (df) ⫽ 61; n.s. ⫽ not significant at 0.05. Mean

Figure 7. Whittaker or Dominance-Diversity curves for PSR treatment in 1989 and 1996. Plots show rank order scores of taxa height classes by their Importance Values (see text for calculation of Importance Value).

Year

CSR

PSR

t-Value

1989 1990 1991 1992 1993 1994 1995 1996

84.6 156.2 246.3 207.1 280.3 242.7 129.4 214.6

197.9 238.4 321.9 230.9 252.1 245.4 254.1 232.3

⫺10.513 ⫺6.210 ⫺3.787 ⫺1.485 1.526 ⫺0.163 ⫺8.730 ⫺1.232

p-Value ⬍0.001 ⬍0.001 ⬍0.001 0.143 0.132 0.871 ⬍0.001 0.223

n.s. n.s. n.s. n.s.

There were significant differences in average number of species per plot between the soil treatments for all years except the first, 1989 (Table 6). In 1990, there were more species per plot on the PSR treatment, but in subsequent years, there were more species per plot

on the CSR treatment. There was significantly more average cover per plot by plants on the PSR versus CSR site for the years 1989–1991 and for 1995 (following the fire; Table 7). It became increasingly obvious that coverage by vines and plants over one meter in height (height classes 4 and 5) were becoming more dominant on the PSR as time went on (Figure 4). Average percent of total coverage by hydrophytes showed no significant difference between the CSR and PSR treatments within or between years (Table 8). PSR plots had high levels of coverage by S. terebinthifolius, but this pest plant was almost completely absent from the CSR treatment (Figure 5). Schinus terebinthifolius was well-established and common in height class ⬎ 2m on PSR by 1996. In general, PSR plots had less submersed vegetation, more vine coverage, and more woody species coverage in higher height classes than CSR plots. Species composition of CSR plots was much more similar to the natural vegetation than PSR plots, with Cladium jamaicense, Schizachyrium scoparium, and Muhlenbergia capillaris becoming more common by 1996 in CSR plots.

Table 6. Results of independent t-tests on mean number of species per plot for Complete Soil Removal (CSR, n ⫽ 63) versus Partial Soil Removal (PSR, n ⫽ 14) for 1989 to 1996; degrees of freedom (df) ⫽ 61; n.s. ⫽ not significant at 0.05.

Table 8. Results of independent t-tests on mean total cover of hydrophytes per plot for Complete Soil Removal (CSR, n ⫽ 63) versus Partial Soil Removal (PSR, n ⫽ 14) for 1989 to 1996; degrees of freedom (df) ⫽ 61; n.s. ⫽ not significant at 0.05.

ment (above). By 1996, the dominant plants were Baccharis glomeruliflora, Sarcostemma clausum (Jacq.) Roem. & Schult. (White twinevine), and Schinus terebinthifolius. The 1996 curve is flatter than the 1989 curve, reflecting an increase in evenness over the eight years of the study. The number of height classes increased, reflecting the growth of woody species into taller height classes and an overall increase in species richness. By 1996, all height classes of B. glomeruliflora and S. terebinthifolius made up six of the top 10 taxa height classes by percent coverage and IVs on the site. Comparison of the Two Treatments

Mean

Mean

Year

CSR

PSR

t-Value

1998 1990 1991 1992 1993 1994 1995 1996

23.3 25.8 31.0 27.8 32.6 31.9 26.1 33.6

22.1 32.3 20.7 20.1 25.1 19.7 17.2 21.9

0.862 ⫺5.065 7.669 6.033 5.726 9.405 7.190 8.580

p-Value 0.392 n.s. ⬍0.001 ⬍0.001 ⬍0.001 ⬍0.001 ⬍0.001 ⬍0.001 ⬍0.001

Year

CSR

PSR

t-Value

1989 1990 1991 1992 1993 1994 1995 1996

63.7 68.3 77.4 81.2 80.2 79.3 82.2 74.2

63.8 68.6 78.3 81.8 80.4 80.9 80.7 70.9

⫺0.012 ⫺0.080 ⫺0.411 ⫺0.207 ⫺0.070 ⫺0.643 0.478 1.019

p-Value 0.990 0.936 0.683 0.837 0.944 0.522 0.635 0.312

n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s.


1025

Figure 8. Scatterplots of scores of individual plots (P1-P63) on first 2 Factors (Principal Components) from Factor Analyses for data from CSR and PSR plots in 1990 (8A) and 1996 (8B). Plots 1–49 were on the CSR treatment and plots 50–63 were on PSR treatment. In the 1996 plot, CSR plots separated from PSR plots (a circle was drawn around them for clarity). Note the distinct location of Plot 53 in 1996, this is the single very low-lying plot on PSR, which was located at the edge of a permanent pond (see text).

Besides the importance of monitoring S. terebinthifolius on the two treatments, three other woody species that were common to both treatments were Ludwigia octovalvis, L. peruviana, and Baccharis glomeruliflora. While L. octovalvis and B. glomeruliflora are native species, L. peruviana is not (Wunderlin 1998). None of these woody species is typical of the vast expanses of open short hydroperiod prairie that surrounds much of the HID area, and their presence in large coverages in higher height classes was not desirable in this restoration effort. The prescribed fire of May 1995 was expected to burn back much of this taller woody growth on the CSR treatment. The 1995 fire did effectively reduce the taller woody plants on the CSR treatment. By 1996, the cover per plot on the CSR treatment had increased, but the amount of L. octovalvis, L. peruviana, and B. glomeruliflora was greatly reduced in frequency and height. However, the fire was not successful on the PSR treatment, since it hardly penetrated the much denser and taller woody cover on the PSR treatment. The inability to control woody cover with fire successfully on the PSR site was expected based

upon all the previous studies of fire effects in S. terebinthifolius stands. The decision to terminate the PSR study and concentrate on continued use of the CSR treatment for S. terebinthifolius removal was not dependent upon the results of the May 1995 fire. Factor analysis by Principal Components Analysis (using taxa height classes for all species by percent cover per plot) showed no separation of the two treatments in 1990 but a clear separation of treatments by 1996 (Figure 8). The one plot that was most unlike the remaining PSR treatments was plot 53, which had the lowest elevation of any PSR plot. This plot was on the edge of a deep depression and had much less coverage by S. terebinthifolius. In Figure 8B, Plot 53, while separate from the remaining PSR plots, was not closely associated with the CSR treatment plots either. Whittaker curves for the two treatments for 1989 and 1996 (Figures 6 and 7) showed that, in both years, there were more taxa height classes and greater coverages on CSR versus PSR treatment. This was in large part a function of the larger area and number of plots studied on the CSR treatment (see results of species-area curves below). Comparison between treat-

1026

WETLANDS, Volume 23, No. 4, 2003 DISCUSSION Elevation, Hydrology, Soil, and Schinus terebinthifolius

Figure 9. Species-Area curves for CSR and PSR treatments in 1989 and 1996. Plots show cumulative number of species recorded at increasing cumulative area sampled.

ments using Whittaker curves was difficult because of the difference in size of the sites. However, within both treatments, there were more species height classes and a less steep curve in 1996 in comparison to 1989. In both treatments, there was an increase in species (taxa height class) diversity and a reduction in dominance (evenness increased) as the eight years of the study progressed. Both treatments showed the same basic trends. Cumulative species-area curves for the two treatments made it very clear that there were not only more species on the CSR treatment because, it was larger, but also because, for any given sampling area size, more species occupied this soil treatment (Figure 9). The 49 plots on the CSR treatment had a cumulative 80 species in 1989 and 91 species in 1996, an increase of 11 species. By comparison, the species area curves for the PSR treatment changed little between 1989 and 1996. There were a cumulative 58 species in 14 plots on PSR in 1989 and 64 species in 1996, for an addition of only six species in eight years. Using 14 randomly selected plots for the CSR site, there were 59 species in 1989 and 75 species in 1996, a total of 16 more species. Based upon this comparison of PSR and CSR (14 plots or 1,400 m2), there were between two and three times as many species added per unit time on the CSR versus PSR site. In 1997, the PSR site was cleared of vegetation and residual soil when it was scraped down to bedrock. The PSR site became a portion of a new, larger restoration site that has a large interface to natural, shorthydroperiod prairie to the east and south. During the succeeding years, the area once known as the PSR site has quickly taken on the characteristics of a wetland, with no successful colonization by S. terebinthifolius to date.

Mean elevations of the two sites were significantly different, and this resulted in about a 28-day difference in the duration of flooding on the two sites. This difference in hydrology did not result in a difference in average percentage cover by hydrophytes on the two sites, but it did allow many more S. terebinthifolius seedlings to germinate and flourish on the drier PSR site. In addition to the difference in hydrology, the small amount of residual rock- plowed soil left on site in the PSR treatment appears to have contributed significantly to the success of S. terebinthifolius in re-colonizing this treatment. Recent studies identified plantavailable nutrients (phosphorus, zinc and copper) in the rock-plowed soils that were in 3 to 10 times greater concentrations than in naturally occurring soils, and P levels in S. terebinthifolius leaves were 3 to 5 times greater than in the native sedge Cladium jamaicense (Li and Norland 2001). ‘‘The high correlation coefficients between leaf P and total and plant-available P in soils indicate that P enrichment in farmed soils facilitated the invasion of Brazilian pepper in this area’’ (Li and Norland, 2001:400). It is likely that some S. terebinthifolius seeds remained in the soil that was left on the PSR site, although studies of seed banks by Ewel et al. (1982) indicated that S. terebinthifolius seeds usually do not remain viable in the soil for more than three months. The site and soil clearing took place from January through April, which is the time of year with the highest rates of S. terebinthifolius seed dispersal (Ewel et al. 1982). Since both of these small sites were completely surrounded by old successional and mature S. terebinthifolius forests, and the rain of seeds onto the sites was high (pers. obs.), the seed bank was not the sole, nor the most important, source of new seedling germination on the sites. Those S. terebinthifolius seedlings that did germinate on the CSR site were observed to drown and die as water levels rose in the early summer (also see Ewel et al. 1982). The resulting scarcity of seedlings on the CSR treatment and the rapid domination by S. terebinthifolius on the PSR treatment indicate that the characteristics of the partial soil layer, including slightly higher elevation, aeration, and a higher remnant nutrient level, were beneficial to S. terebinthifolius germination, growth, and survival. Observations on Plant Colonization While the partial soil removal treatment did increase the number and coverage of hydrophytes, it did not

Dalrymple et al., WETLAND RESTORATION IN THE FLORIDA EVERGLADES prohibit re-colonization and re-establishment of a canopy of S. terebinthifolius. By 1996, S. terebinthifolius was abundant and common in the ⬎ 2m height class in 12 of the 14 partial soil removal plots, while it was found only as ephemeral seedlings in two of the 49 complete soil removal plots. In general, the partial soil removal treatment had high coverage values for vines and woody plants in height classes greater than 1 meter, which is not typical of short hydroperiod wetlands of this region. While elevations of the two soil treatments were similar, the partial layer of rock-plowed soil left on the partial soil removal treatment significantly reduced the site’s ability to inhibit re-colonization by S. terebinthifolius. Early dominance by weedy sun-loving species was common to both soil treatments in the first years of the study. The dominance of some of these early colonizers may be related to past farming practices. For example, Sesbania herbacea (Danglepod) is a domesticated annual legume subshrub species (Correll and Correll 1982). In southern Florida, this species was commonly planted on farm fields after harvest (pers. obs.), in part because of its nitrogen-fixing ability and its ability to tolerate wet conditions (Long and Lakela 1971, Godfrey and Wooten 1981, Wunderlin 1998). The species’ seeds are thick coated and survive long periods of dormancy. It is common for three to six months on previously farmed areas that have experienced a reduced vegetative cover. This pattern was observed at the inception of this project in 1989 and again after the prescribed burn in 1995 on the study area. Native species that were common in the first year tended to be weedy species that could tolerate lack of soil on the sites, such as Spermacoce floridana. Correll and Correll (1982:1422) describe this species as occurring on ‘‘raw soil in open thinly grassed area.’’ Chara sp. was another common native plant during the first year. The combination of full sun light, hot shallow water, and high calcium carbonate-based marl dust on scraped sites appears to present ideal conditions for this calcareous algal taxa. As the ground cover became thicker and more diverse, these early colonizers became less common. The presence of Chara in later years was negatively correlated to the development of a periphyton mat during periods of standing water (pers. obs. and unpublished data). A combination of Ludwigia microcarpa, several grasses, especially Andropogon glomeratus and Schizachyrium scoparium, as well as Bacopa monnieri replaced these early dominant sun-loving species on the CSR site. On the PSR site, these species were replaced by Ludwigia peruviana, L. microcarpa, Baccharis glomeruliflora, and S. terebinthifolius and prohibited these sun-loving species from establishing on the PSR site.

1027

Early grass invaders, such as Leptochloa fascicularis (Lam.) A. Gray (Bearded sprangletop) and Panicum dichotomiflorum Michx. (Fall panicgrass), remained on the sites through 1996 but became less abundant with time. These annual grasses have long-lived seeds and occur almost exclusively in disturbed sites in the region (pers. obs.). Gross (1987:182) found P. dichotomiflorum and Panicum capillare L. (Witchgrass) to have similar habits in a restoration study in Michigan. Caespitose grasses and rhizomatous sedges and grasses became more common with time. By 1996, the caespitose (tufted) grasses Andropogon glomeratus, Schizachyrium scoparium, Eragrostis elliottii S. Watson (Elliott’s lovegrass), Muhlenbergia capillaris, and the rhizomatous sedges Cladium jamaicense and Rhynchospora microcarpa were becoming common on the CSR site. A consolidator strategy is used by clones of caespitose grasses to accumulate resources just below the soil surface (Briske and Derner 1998). The same strategy is used by the perennial clonal C. jamaicense (Steward and Ornes 1975), which accumulates nutrients in below-ground rhizomes. This strategy may play an important role in these herbaceous species’ continued success on the CSR site. ‘‘The most plausible interpretation of the ecological success of the caespitose growth form has been attributed to effective resource monopolization within the immediate environment of clones’’ (Briske and Derner 1998:123). Greater pools of soil organic carbon and nitrogen have been found in soil beneath clones of Schizachyrium scoparium, and this nutrient accumulation was restricted to the upper 5 cm of the soil (Derner et al. 1997). Accumulation of nutrient storage in soil below caespitose grass clones and in rhizomes of perennial sedges helps to reduce above-ground nitrogen loss from herbivory, fire, and volatilization during periods of limited water availability (Steward and Ornes 1975, Briske and Derner 1998). This should permit these species to continue to expand their coverage on the CSR treatment. The degree of interclonal competition within and between the caespitose grasses and competition between them and perennial sedges may result in a shared dominance in future years on the CSR treatment. However, on the CSR site, once C. jamaicense reached a certain level, it began to expand rapidly through rhizomatous growth and addition of new culms. In the 1999 sampling of 20 randomly selected plots on the CSR site, C. jamaicense ranked third in total coverage (unpublished data), indicating a tendency for C. jamaicense to replace the other species. The relatively low coverage of woody species on the CSR versus PSR site was very similar to observations on many of the older abandoned fields along the edges of the HID where rock plowing was not employed (pers. obs.). On such sites, the dominant

1028


vegetation is composed of native sedges, grasses, and forbs, but woody species, including Ludwigia octovalvis, L. peruviana, Baccharis glomeruliflora, and S. terebinthifolius, are not uncommon. Lack of a rockplowed and aerated modified soil layer in these older farming areas may have prevented widespread colonization by S. terebinthifolius, even though these areas were reported to have nutrient levels similar to rockplowed areas (Li and Norland 2001). Succession after farming resulted in the rapid re-establishment of native wetland plants but with a higher complement of woody and weedy species. Prescribed fire was successful at controlling growth of woody species in the absence of Schinus terebinthifolius. However, fire was not effective in controlling woody species in areas dominated by S. terebinthifolius. Previous studies have shown that fires are frequently unable to penetrate deep into these stands (Doren and Whiteaker 1990b). We recommend the use of prescribed fire on CSR sites at three- to five-year intervals to mimic fire frequency in natural vegetation. This would not only reduce coverage by woody shrubs but would also promote expansion of desirable native plant species such as Cladium jamaicense. As with many wetland mitigation permits, the primary success criteria was coverage by hydrophytes. However, wetland restoration sites may have extensive coverage by hydrophytes, yet still not have a community structure that resembles a native vegetation community. While species dominance may differ between restored sites and native vegetation, an additional success criterion should be the development of a vegetation community that is similar in vertical and horizontal structure to natural vegetation. Moreover, the vegetation community should be self-maintaining, such that natural processes promote the growth of desirable species and suppress undesirable species. CONCLUSIONS The main objective of the two-treatment pilot project was to determine whether it was necessary to remove all soil down to consolidated bedrock to inhibit S. terebinthifolius re-colonization and permit an extended hydroperiod to induce natural colonization by native wetland plants. The CSR treatment had more herbaceous species per plot, while the PSR treatment had a higher canopy dominated by S. terebinthifolius and other woody species in the 1 to 2 m and ⬎ 2 m height classes. In both treatments, there was an increase in species (taxa height class) richness and evenness as succession continued through eight years of the study, but the greater coverage of S. terebinthifolius in all height classes led to a much lower overall diversity and evenness on the PSR. The eight-year

study of two soil treatments showed that complete soil removal (CSR) of rock-plowed substrate produced by farming was necessary to prevent re-establishment of S. terebinthifolius and to promote natural re-colonization by native wetland plants. The results of the study also demonstrated that a wetland restoration may have extensive coverage by hydrophytes but also have an undesirable exotic pest plant load. ACKNOWLEDGMENTS Raw data for the two soil treatments were collected by the late Mark McMahon from 1989 to 1996. Sally Black, Keith Bradley, and Roger Hammer provided very useful discussions of taxonomic issues. Dick Reimus of the South Florida Natural Resources center provided an updated species list for the park and was always willing to discuss plant identifications. Walter Meshaka provided regular access to the park’s herbarium. G. F. (Stinger) Guala and the staff of the herbarium at Fairchild Tropical Garden were helpful with taxonomic issues and access to the garden’s collections. Staff of Metropolitan Dade County’s Department of Environmental Resources Management, initially George Molnar and later Jean Evoy and Kathy Fanning, were always available for advice on the project. Rachel Budelsky and two anonymous reviewers provided thoughtful comments and suggestions to improve the manuscript. LITERATURE CITED Armentano, T. V., R. F. Doren, W. J. Platt, and T. Mullins. 1995. Effects of Hurricane Andrew on coastal and interior forests of southern Florida: overview and synthesis. Journal of Coastal Research 21:111–144. Bazzaz, F. A. 1996. Plants in Changing Environments. Cambridge University Press, Cambridge, UK. Briske, D. D. and J. D. Derner. 1998. Clonal biology of caespitose grasses. p. 106–135. In G. P. Cheplick (ed.) Population Biology of Grasses.Cambridge University Press, Cambridge, UK. Colinvaux, P. 1986. Ecology. John Wiley and Sons, New York, NY, USA. Correll, D. S. and H. B. Correll. 1982. Flora of the Bahama Archipelago. Lubrecht Cramer Limited, Port Jervis, NY, USA. Dalrymple, N. K., G. H. Dalrymple, and K. A. Fanning. 1993. On the vegetation of restored and unrestored rock-plowed wetlands of the East Everglades of southern Florida. Restoration Ecology 1:220–225. Derner, J. D., D. D. Briske, and T. W. Boutton. 1997. Does grazing mediate soil carbon and nitrogen accumulation beneath C4 perennial grasses along an environmental gradient? Plant and Soil 191: 147–156. Doren, R. F. and L. D. Whiteaker. 1988. Proposal for mitigation and monitoring of secondary successional communities in ENP: Holein-the-Donut. Everglades National Park, Homestead, FL, USA. Doren, R. F. and L. D. Whiteaker 1990a. Comparison of economic feasibility of chemical control strategies on differing age and density classes of Schinus terebinthifolius. Natural Areas Journal 10: 28–34. Doren, R. F. and L. D. Whiteaker. 1990b. Effects of fire on different

Dalrymple et al., WETLAND RESTORATION IN THE FLORIDA EVERGLADES size individuals of Schinus terebinthifolius. Natural Areas Journal 10:106–113. Doren, R. F., L. D. Whiteaker, G. Molnar, and D. Sylvia. 1990. Restoration of former wetlands within the Hole-in-the-Donut in ENP. p. 33–50. In F. J. Webb, Jr. (ed.) Proceedings of the Seventh Annual Conference on Wetland Restorations and Creation. Hillsborough Community College, Plant City, FL, USA. Ewel, J. J. 1986. Invasibility: lessons from south Florida. p. 214– 230. In A. H. Mooney and J. J. Drake (eds.) Ecology of Biological Invasions of North America and Hawaii. Springer Verlag, New York, NY, USA. Ewel, J. J., D. S. Ojima, D. A. Karl, and W. F. Debusk. 1982. Schinus in successional ecosystems of Everglades National Park. South Florida Research Center, Everglades National Park, Homestead, FL, USA. Technical Report T-676. Gauch, H. G., Jr. 1982. Multivariate Analysis in Community Ecology. Cambridge University Press, New York, NY, USA. Godfrey, R. K. and J. W. Wooten. 1981. Aquatic and Wetland Plants of Southeastern United States: Dicotyledons. University of Georgia Press, Athens, GA, USA. Gross, K. L. 1987. Mechanisms of colonization and species persistence in plant communities. p. 173–188. In W. R. Jordan, M. E. Gilpin, and J. D. Aber (eds.) Restoration Ecology: a Synthetic Approach to Ecological Research. Cambridge University Press, Cambridge, U K. Kent, M. and P. Coker. 1994. Vegetational Description and Analysis. Wiley and Sons, New York, NY, USA. Krauss, P. 1987. Old field succession in Everglades National Park. South Florida Research Center Report, Everglades National Park, Homestead, FL, USA. Technical Report SFRC-87/03. Li, Y. and M. Norland. 2001. The role of soil fertility in invasion of Brazilian pepper (Schinus terebinthifolius) in Everglades National Park, Florida. Soil Science 166:400–405. Long, W. L. and O. Lakela. 1971. A Flora of Tropical Florida. University of Miami Press, Coral Gables, FL, USA. Loope, L. L. and V. L. Dunevitz. 1981. Investigations of early plant succession on abandoned farmland in Everglades National Park.

1029

South Florida Research Center, Everglades National Park, Homestead, FL, USA. Technical Report T-644. MacArthur, R. M. and E. O. Wilson. 1967. The Theory of Island Biogeography. Princeton University Press. Princeton, NJ, USA. Mueller-Dombois, D. and H. Ellenberg. 1974. Aims and Methods of Vegetation Ecology. John Wiley and Sons, Inc, New York, NY, USA. Olmsted, I. and L. L. Loope. 1984. Plant communities in Everglades National Park. p. 167–184. In P. J. Gleason (ed.) Environments of South Florida Present and Past II. Miami Geological Society, Miami, FL, USA. Orth, P. G. 1981. Fertility management of Dade County soils. Soil Crop Science Society Florida Proceedings 40:1–3. Orth, P. G. and R. A. Conover. 1975. Changes in nutrients resulting from farming the Hole-in-the-Donut, Everglades National Park. Proceedings of Florida State Horticultural Society 28:221–225. Pielou, E. C. 1984. The Interpretation of Ecological Data. John Wiley and Sons, New York, NY, USA. Reed, P. B., Jr. 1988. National List of Plant Species That Occur in Wetlands: Southeast Region (Region 2). National Wetlands Inventory, U.S. Fish and Wildlife Service, Washington, DC, USA. Biological Report 88(26.2). Sokal, R. R. and F. J. Rohlf. 1995. Biometry. W. H. Freeman, New York, NY, USA. StatSoft. 1994. Statistica for Windows. StatSoft, Inc. Tulsa, OK, USA. Steward, K. K. and W. H. Ornes. 1975. The autecology of sawgrass in the Florida Everglades. Ecology 56:162–171. Tilman, D. 1988. Plant Strategies and the Dynamics and Structure of Plant Communities. Princeton University Press, Princeton, NJ, USA. Whittaker, R. H. 1975. Communities and Ecosystems, second edition. MacMillan, New York, NY, USA. Wunderlin, R. P. 1998. Guide to the Vascular Plants of Florida. University Press of Florida, Gainesville, FL, USA. Zar, J. H. 1996. Biostatistical Analysis, third edition. Prentice-Hall, Englewood Cliffs, NJ, USA. Manuscript received 23 August 2002; revisions received 17 March 2003 and 2 June 2003; accepted 20 August 2003.