<%BANNER%>

Longleaf Pine Sandhill Seed Banks and Seedling Emergence in Relation to Time Since Fire

Permanent Link: http://ufdc.ufl.edu/UFE0021193/00001

Material Information

Title: Longleaf Pine Sandhill Seed Banks and Seedling Emergence in Relation to Time Since Fire
Physical Description: 1 online resource (84 p.)
Language: english
Creator: Parks, Geoffrey R
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2007

Subjects

Subjects / Keywords: aristida, bank, beyrichiana, canopy, disturbance, emergence, fire, germination, groundcover, laevis, leaf, litter, longleaf, oak, palustris, pine, pinus, quercus, restoration, sandhill, savanna, seed, soil, southeast, stratification, stricta, suppression, wiregrass
Botany -- Dissertations, Academic -- UF
Genre: Botany thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Longleaf pine (Pinus palustris) sandhill is a fire-dependent ecosystem with high groundcover species richness that endemic to the southeastern United States. The extent of longleaf pine sandhills has been greatly reduced in recent decades, and much of what remains is degraded by lack of fire. Herbaceous species such as wiregrass (Aristida stricta) which are important as a fuel source for frequent ecosystem-maintaining fires, decline when fires are suppressed. There is much interest in restoring groundcover to degraded longleaf pine sites. If seeds of the native groundcover species are abundant in the soil seed bank, then restoration efforts might benefit from treatments that promote their germination and establishment. I examined soil seed banks in longleaf pine sandhill sites with a range of time since the most recent fire. I also used field experiments to assess how different environmental conditions influence seedling establishment in sandhills. I collected soil from the top 2 cm of soil at sites at the Ordway-Swisher Biological Station in Florida, USA, that had burned 1, 5, 10, and 35 years prior to the study, and observed seedling emergence from the samples for 6 months in the greenhouse. In all, 1399 seedlings of 60 taxa emerged from the soil samples. Seed density varied from 372 seeds m-2 to 1200 seeds m-2. Cold stratification of the samples reduced the number of seedlings emerging, but some species, most notably legumes, were detected only in stratified samples. Although seedlings of some characteristic sandhill plants emerged from the samples, most seedlings were of weedy species, and the seed bank had low similarity to sandhill vegetation. To examine the factors that affect seedling emergence, I observed emergence of dicot seedlings in response to litter removal, midstory removal, and soil disturbance at a regularly burned site and a site that had not been burned for 35 years. Dicot seedling emergence in the unburned site was negligible (0.3375 seedlings/m2 year), regardless of treatment, while dicot seedlings emerged at a rate of 3.96 seedlings/m2 year in the burned site. A total of 344 seedlings of 27 species emerged during the study. All seedlings were of species found in the vegetation. Soil disturbance and litter removal increased seedling emergence relative to controls. Removal of midstory hardwoods did not influence seedling emergence. Although a buried seed pool is present in longleaf pine sandhills, this seed bank has only a limited capacity to contribute to the restoration of diversity or cover of desirable herbaceous species in degraded sandhills. Restoration of groundcover will require planting of transplants or seeds of native species.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Geoffrey R Parks.
Thesis: Thesis (M.S.)--University of Florida, 2007.
Local: Adviser: Putz, Francis E.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2009-08-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2007
System ID: UFE0021193:00001

Permanent Link: http://ufdc.ufl.edu/UFE0021193/00001

Material Information

Title: Longleaf Pine Sandhill Seed Banks and Seedling Emergence in Relation to Time Since Fire
Physical Description: 1 online resource (84 p.)
Language: english
Creator: Parks, Geoffrey R
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2007

Subjects

Subjects / Keywords: aristida, bank, beyrichiana, canopy, disturbance, emergence, fire, germination, groundcover, laevis, leaf, litter, longleaf, oak, palustris, pine, pinus, quercus, restoration, sandhill, savanna, seed, soil, southeast, stratification, stricta, suppression, wiregrass
Botany -- Dissertations, Academic -- UF
Genre: Botany thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Longleaf pine (Pinus palustris) sandhill is a fire-dependent ecosystem with high groundcover species richness that endemic to the southeastern United States. The extent of longleaf pine sandhills has been greatly reduced in recent decades, and much of what remains is degraded by lack of fire. Herbaceous species such as wiregrass (Aristida stricta) which are important as a fuel source for frequent ecosystem-maintaining fires, decline when fires are suppressed. There is much interest in restoring groundcover to degraded longleaf pine sites. If seeds of the native groundcover species are abundant in the soil seed bank, then restoration efforts might benefit from treatments that promote their germination and establishment. I examined soil seed banks in longleaf pine sandhill sites with a range of time since the most recent fire. I also used field experiments to assess how different environmental conditions influence seedling establishment in sandhills. I collected soil from the top 2 cm of soil at sites at the Ordway-Swisher Biological Station in Florida, USA, that had burned 1, 5, 10, and 35 years prior to the study, and observed seedling emergence from the samples for 6 months in the greenhouse. In all, 1399 seedlings of 60 taxa emerged from the soil samples. Seed density varied from 372 seeds m-2 to 1200 seeds m-2. Cold stratification of the samples reduced the number of seedlings emerging, but some species, most notably legumes, were detected only in stratified samples. Although seedlings of some characteristic sandhill plants emerged from the samples, most seedlings were of weedy species, and the seed bank had low similarity to sandhill vegetation. To examine the factors that affect seedling emergence, I observed emergence of dicot seedlings in response to litter removal, midstory removal, and soil disturbance at a regularly burned site and a site that had not been burned for 35 years. Dicot seedling emergence in the unburned site was negligible (0.3375 seedlings/m2 year), regardless of treatment, while dicot seedlings emerged at a rate of 3.96 seedlings/m2 year in the burned site. A total of 344 seedlings of 27 species emerged during the study. All seedlings were of species found in the vegetation. Soil disturbance and litter removal increased seedling emergence relative to controls. Removal of midstory hardwoods did not influence seedling emergence. Although a buried seed pool is present in longleaf pine sandhills, this seed bank has only a limited capacity to contribute to the restoration of diversity or cover of desirable herbaceous species in degraded sandhills. Restoration of groundcover will require planting of transplants or seeds of native species.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Geoffrey R Parks.
Thesis: Thesis (M.S.)--University of Florida, 2007.
Local: Adviser: Putz, Francis E.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2009-08-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2007
System ID: UFE0021193:00001


This item has the following downloads:


Full Text

PAGE 1

LONGLEAF PINE SANDHILL SEED BANK S AND SEEDLING EMERGENCE IN RELATION TO TIME SINCE FIRE By GEOFFREY RICHMOND PARKS A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2007

PAGE 2

2007 Geoffrey R. Parks

PAGE 3

Dedicated to the memory of my father, Geor ge Richard Parks, who cultivated in me a love and wonder for the natural world.

PAGE 4

iv ACKNOWLEDGMENTS This material is based upon work s upported under a National Science Foundation Graduate Research Fellowship. In addition, I am grateful to a great many people for advice, help, and support during the process of completing this projec t. Steve Coates was helpful with many aspects of this resear ch, and provided tools, equipment, and information. Dick Franz helped me locate study sites. Bil Grauel, Sarah Bray, Katie Baumann, and Morgan Varner braved hot weat her, ticks, and stinging caterpillars to assist in the field. Richard Abbott and Barry Davis assisted with plant identification. Diana Alvira checked seedling flats in the greenhouse when I was out of town. Alison Fox and Lisa Huey generously provided gr eenhouse space when it seemed that there was none available. Mel Sunquist provided loca l rainfall data when the Ordway weather station was disabled by light ning. Discussions with Amy Miller Jenkins helped my thinking about seed banks and methodological issues. Xueli Liu and Aixin Tan advised on statistical matters. My committee memb ers Doria Gordon, Kaoru Kitajima, and Francis Putz patiently guided me through the process and provided useful insights at all stages of my studies. Fellow graduate stude nts Lou Santiago, Geoff Blate, Tova Spector, Katia Silvera, and Adam Watts made my time at UF, on and off campus, much more enjoyable. Finally, none of this would ha ve been possible without the support and encouragement of my family, particularly my wife, Carol, whose faith in me was unswerving.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iv LIST OF TABLES............................................................................................................vii LIST OF FIGURES.........................................................................................................viii ABSTRACT.......................................................................................................................ix CHAPTER 1 POTENTIAL CONTRIBUTIONS OF SOIL SEED BANKS TO GROUNDCOVER RESTORATION IN FIRE-SUPPRESSED LONGLEAF PINE SANDHILLS......................................................................................................1 Ecology and Conservation Status of Longleaf Pine Communities...............................1 Sandhill Soil Seed Banks and their Potential Use in Groundcover Restoration...........2 Methodological Issues in Seed Bank Research............................................................4 Factors Limiting Seedling Establishment in Fire-Suppressed Longleaf Pine Sites, and Possible Restoration Actions.............................................................................6 Study Goals...................................................................................................................8 2 LONGLEAF PINE SANDHILL SOIL SE ED BANKS IN RELATION TO FIRE HISTORY.....................................................................................................................9 Introduction...................................................................................................................9 Methods......................................................................................................................10 Study Sites...........................................................................................................10 Sample and Data Collection................................................................................12 Groundcover Vegetation Sampling.....................................................................13 Statistical Analyses..............................................................................................13 Results........................................................................................................................ .15 Timing of Seedling Emergence...........................................................................15 Seedling Density..................................................................................................15 Taxa Represented................................................................................................16 Species Richness.................................................................................................17 Similarity of Seeds in Soil Samples to Field Vegetation....................................18 Discussion...................................................................................................................18

PAGE 6

6 3 DICOT SEEDLING EMERGENCE IN BURNED AND UNBURNED LONGLEAF PINE SANDHILLS IN RE LATION TO OVERSTORY, LITTER, AND SOIL DISTURBANCE.....................................................................................40 Introduction.................................................................................................................40 Methods......................................................................................................................41 Study Site.............................................................................................................41 Study Design.......................................................................................................42 Data Collection....................................................................................................43 Seedling surveys...........................................................................................43 Canopy structure..........................................................................................44 Vegetation composition................................................................................44 Weather data.................................................................................................44 Comparison with seedling emer gence immediately postfire.......................45 Statistical Analyses..............................................................................................45 Results........................................................................................................................ .46 Seedling Density..................................................................................................46 Effects of Soil and Canopy Treatments...............................................................47 Seasonal and Weather-Related Patterns..............................................................47 Seedling Taxa......................................................................................................48 Similarity Between Emerging Seedings and Aboveground Vegetation..............48 Comparison with Seedling Emergence from Soil Samples.................................49 Comparison with Seedling Emergence Immediately Following Fire.................49 Discussion...................................................................................................................50 Relation of Seedling Emergence in th e Field to Aboveground Vegetation and Seeds in Soil Samples......................................................................................50 Climate Effects....................................................................................................51 Effects of Soil Treatments, Canopy Treatments, and Site...................................52 Fire and Seedling Emergence..............................................................................54 Implications for Restoration and Management...................................................55 LIST OF REFERENCES...................................................................................................66 BIOGRAPHICAL SKETCH.............................................................................................74

PAGE 7

vii LIST OF TABLES Table page 2-1 Fire history and vegetation characteristics of four sandhill sites in Putnam County, Florida.........................................................................................................26 2-2 Analysis of Variance for effects of site and cold stratification on estimated seed density in soil samples taken from sandhill sites with different fire histories.........28 2-3 Seeds found in stratified and fresh soil samples taken from 4 sandhill sites with different fire histories...............................................................................................31 2-4 Analysis of Variance using SAS pr oc GLM testing effects of site and soil stratification on the species composition of seeds detected in soil samples taken from 4 sandhill sites with different fire histories.....................................................37 2-5 Incidence-based Jaccard estimator of similarity among vegetation, seeds detected in fresh soil, and seeds detected in stratified soil from 4 sandhill sites with different times since fire...................................................................................38 3-1 Vegetation structure and fire hist ory of a frequently burned and an unburned sandhill site in Putnam County, Florida...................................................................56 3-2 Number of seedlings by species emerging over 1 year in two longleaf pine sandhill sites with diffe rent fire histories.................................................................61 3-3 Incidence-based Jaccard similarity estimator values comparing seedlings emerging over one year and vegetation at two longleaf pine sa ndhill sites with different fire histories...............................................................................................62 3-4 Abundance-based Jaccard similarity estimators and observed and estimated shared species between seedlings emer ging in two longleaf pine sandhill sites and seeds from soil samples taken from the same sites...........................................63 3-5 Species that were found as newly-em erged seedlings in a burned longleaf pine sandhill site that were also found as s eeds in soil samples taken from the same site........................................................................................................................... .64

PAGE 8

viii LIST OF FIGURES Figure page 2-1 Mean ( S.E.) density of viable seeds in soil samples taken from four sandhill sites with different fire histories...............................................................................27 2-2 Density of seeds (Mean S.E.) found in fresh and stratified soils taken from the top 2 cm of soil in four sites with diffe rent fire histories and germinated in a greenhouse................................................................................................................29 2-3 Mean species richness and overlap of species between vege tation and seeds from soil samples taken from four sandhill sites..............................................................33 2-4 Coleman rarefaction curves and es timated species richness for seeds found in soil samples taken from sandhill site s with different fire histories..........................35 2-5 Mean field frequency of seeds detect ed in stratified and fresh soil samples from sandhill sites with differing fire histories.................................................................36 3-1 Number of seedlings emerging over one year in two longleaf pine sandhill sites in Putnam County, Florida.......................................................................................57 3-2 Effect of hardwood removal on pe rcent canopy openness in frequently-burned and unburned longleaf pine sandhill in Putnam County, Florida.............................58 3-3 Seasonal pattern of emergence of s eedlings of different plant families in longleaf pine sandhills............................................................................................................59 3-4 Seedling emergence in two longleaf pine sandhill sites in Putnam County, Florida, in relation to rainfall...................................................................................60 3-5 Total density of seedlings emerging within 2 weeks of pr escribed fire in a longleaf pine sandhill in comparison to s eedlings emerging during the same time period in an adjacent sandhill that had not been burned for 5 years........................65

PAGE 9

ix Abstract of Thesis Presen ted to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science LONGLEAF PINE SANDHILL SEED BANK S AND SEEDLING EMERGENCE IN RELATION TO TIME SINCE FIRE By Geoffrey Richmond Parks August 2007 Chair: Francis E. Putz Major: Botany Longleaf pine (Pinus palustris) sandhill is a fire-dependent ecosystem with high groundcover species richness that endemic to th e southeastern United States. The extent of longleaf pine sandhills has been greatly re duced in recent decades, and much of what remains is degraded by lack of fire. Herbaceous species such as wiregrass ( Aristida stricta ) which are important as a fuel source for frequent ecosystem-maintaining fires, decline when fires are suppressed. There is much interest in restoring groundcover to degraded longleaf pine sites. If seeds of the native groundc over species are abundant in the soil seed bank, then restoration efforts might benefit from treatments that promote their germination and establishment. I examined soil seed banks in longleaf pine sandhill sites with a range of time since the most rece nt fire. I also used field experiments to assess how different environm ental conditions influence seedling establishment in sandhills.

PAGE 10

x I collected soil from the top 2 cm of soil at sites at the Ordway-Swisher Biological Station in Florida, USA, that had burned 1, 5, 10, and 35 years prior to the study, and observed seedling emergence from the samples for 6 months in the greenhouse. In all, 1399 seedlings of 60 taxa emerged from the soil samples. Seed density varied from 372 seeds m-2 to 1200 seeds m-2. Cold stratification of the samples reduced the number of seedlings emerging, but some species, most notably legumes, were detected only in stratified samples. Although seed lings of some characterist ic sandhill plants emerged from the samples, most seedlings were of weedy species, and the seed bank had low similarity to sandhill vegetation. To examine the factors that affect seed ling emergence, I observed emergence of dicot seedlings in response to litter removal, midstory remova l, and soil disturbance at a regularly burned site and a site that had not been burned for 35 years. Dicot seedling emergence in the unburned site was negligible (0.3375 seedlings/m2 year), regardless of treatment, while dicot seedlings em erged at a rate of 3.96 seedlings/m2 year in the burned site. A total of 344 seedlings of 27 species emerged during the study. All seedlings were of species found in the vegetation. Soil distur bance and litter rem oval increased seedling emergence relative to controls. Removal of midstory hardwoods did not influence seedling emergence. Although a buried seed pool is present in l ongleaf pine sandhills, this seed bank has only a limited capacity to contribute to the rest oration of diversity or cover of desirable herbaceous species in degraded sandhills Restoration of groundcover will require planting of transplants or seeds of native species.

PAGE 11

1 CHAPTER 1 POTENTIAL CONTRIBUTIONS OF SOIL SEED BANKS TO GROUNDCOVER RESTORATION IN FIRE-SUPPRESSED LONGLEAF PINE SANDHILLS Ecology and Conservation Status of Longleaf Pine Communities Longleaf pine (Pinus palustris) ecosystems of the southeastern United States are maintained by frequent low intensity fire s (e.g., Christensen 1981). Over the past two centuries, loss of longleaf pine habitats has been extensive, with less than 2% of the original acreage of th ese systems existing today (Noss et al. 1995); much of what remains is degraded by fire suppression (Outcalt 2000). Fire suppression in longleaf pine communities results in substantial change s in community structure and composition, including increases in midstory and ca nopy density (Heyward 1939, Gilliam et al. 1993, Provencher et al. 2001), and reductions in c over and diversity of herbaceous vegetation (Provencher et al. 2001). The reduction of herbaceous vegetation in fire-suppressed longleaf pine communities is of particular concern. Thes e communities are among the most diverse in North America (Walker and Peet 1983, Drew et al. 1998), with most of the species richness in the herbaceous groundcover. In addition, the groundcover, dominated by bunchgrasses such as wiregrass (Aristida stricta) and Andropogon spp., plays a critical role in propagating fire across the land scape (Heyward 1939, Clewell 1989, Noss 1989), affecting the fire regimes of adjacent co mmunities as well. Recently there has been increasing interest in restora tion of longleaf pine habitats, with many of the restoration efforts focused on groundcover specie s (e.g., Brockway et al. 1998, Seamon 1998,

PAGE 12

2 Mulligan and Kirkman 2002, Provencher et al 2001, Jenkins 2003). Restoration of this diverse and functionally important groundcover in degraded sites could be facilitated if many species are represented in a persistent soil seed bank. Sandhill Soil Seed Banks and their Potential Use in Groundcover Restoration Soil seed banks, consisting of viable but ungerminated seeds, are found in a wide variety of plant communities. Dormant seeds, which can remain viable for decades or even centuries (reviewed Priestly 1986), can insure against environmental unpredictability, and thus serve as a bet-hedg ing strategy for plants. When environmental conditions suitable for survival, recruitmen t, or reproduction vary over time, seed dormancy increases the probability that some seeds will survive to germinate during favorable periods, thus fostering dispersal in time. Therefore, maintaining a seed bank may be an advantageous strategy for speci es in variable envi ronments (Cohen 1966). If there is a persistent seed bank of native species in degraded longleaf pine systems, such a seed bank could reduce the costs associated with groundcover restoration. Restoration projects us ing direct seeding have been eff ective, but seeds of the potentially hundreds of groundcover species are generally not available commercially. Even when seeds or seedlings of desired species are availa ble, the genotypes of the material available may not be appropriate for the site (Gor don and Rice 1998). Although local colle ction of seeds is sometimes possible from nearby donor s ites, collection of seeds from the entire assemblage of groundcover species requires multiple methods due to morphological differences and must be done at different times due to phenological differences among species. The presence of existing seed banks at degraded sites could circumvent many of these drawbacks. Numerous researchers have suggested that seed banks could be of use in restoration (Jefferson and Usher 1987, Be llemakers et al. 1993 cited in Bakker 1996,

PAGE 13

3 Zhang et al. 2001), so if seed banks persist in degraded longleaf pine sites, they could represent a valuable asset for land manage rs seeking to restore these communities. There is reason to believe that seed banking may be adaptive for pl ants that live in sandhills, although investigations of whether there are soil seed banks in other firedependent pine communities ha ve had mixed results (Malaika l et al. 2000, Cohen et al. 2004, Jenkins 2003). Following fire s, conditions for plant grow th can be favorable in a number of ways (Keeley a nd Fotheringham 2000): fire increases light availability by removing competing vegetation, and increase s the availability of soil nutrients by converting nutrients into biol ogically available forms. Sp ecies with propagules that survive fire and germinate rapidly can take advantage of these favorable postfire conditions, and may be at an advantage relativ e to other species. In some other firedependent plant communities such as Califor nia chaparral (Keeley 1991), Spanish pine forest (Ferrandis et al. 1996) and Australian eucalypt fore st (Read et al. 2000) and savanna (Read et al. 2000, Williams et al. 2005) seed banks have been shown to be an especially important source of postfire recruitment, but this is not always the case (Shrubland in the New Jersey pine barrens, Ma tlack et al. 1993; Florida sand pine scrub, Carrington 1996). Previous studies of seed banks in ot her longleaf ecosystems and in other endangered plant communities have encounter ed mixed results. Hopes for large seed banks of native species in degraded exampl es of other rare or endangered ecosystems have largely been disappoi nted (e.g., Baptista and Shum way 1988, McCall and Gibson 1999). Studies in mesic and wet longleaf pine sites have reached differing conclusions about the quality of the seed bank present. Mala ikal et al. (2000) l ooked at seed banks in

PAGE 14

4 mesic to wet pine sites in Florida with resp ect to time-since-fire, and found little evidence that desirable groundcover spec ies maintain persistent seed banks. Similarly, Jenkins (2003) examined soil seeds in former mesic l ongleaf pine sites that had been converted to non-native pasture and found s eeds of numerous species, th ough seeds the most common and characteristic species of intact systems we re absent. In contrast Cohen et al. (2004) found seeds of several species considered char acteristic of high quality communities in degraded wet longleaf pine sites in North Carolina. Although soil seed banks can poten tially facilitate restoration, they may not always be useful for restorationists. For exampl e, it has long been k nown that species of disturbed habitats often ha ve long-lived seeds (e.g., Harper 1977), a strategy which enables them to quickly colonize gaps when th ey become available. Thus seeds of ruderal species are often overrepresented in seed ba nks. For this and other reasons, previous studies of soil seed banks in a wide array of ecosystems have found that seed banks are often dissimilar to aboveground vegetation and to seed rain (e.g., Rabinowitz 1980, Rabinowitz and Rapp 1980). Alt hough differences between cu rrent vegetation and seed bank composition may be informative about the hi storic vegetation of the site, this history may be biased, insofar as it disproportionately reflects historic dist urbance events (e.g., Livingston and Allessio 1968). Fi nally, some component of the soil seed bank may not represent local vegetation at al l, but may instead result from dispersal of seeds from other communities in the landscape (Trabaud 1994, Poiani and Dixon 1995). Thus seed bank composition may have little similarity either to current or to historic vegetation at a site. Methodological Issues in Seed Bank Research While some of the disparities in sp ecies composition between seed bank and vegetation are likely due to plant life histories, some observed differences may be

PAGE 15

5 attributable to the methodology used to char acterize the seed bank. One of the principal methods for characterizing soil seed banks is the emergence method, in which soil samples are placed in a greenhouse or other suitable conditions and emerging seedlings noted (e.g., Livingston and A llessio 1968, Baptista and Shum way 1998). The results of this approach are dependent on the germin ation conditions provi ded: species with germination requirements that are not met are not detected. Modifications to the emergence method to provide a broader rang e of germination conditions can overcome this drawback somewhat, but not complete ly. Alternatives to the emergence method include manually, mechanically, or chemically extracting seeds from soil samples. Although these methods are not sensitive to differences in the germination requirements of individual species, they can be time-cons uming and can potentially misrepresent the seed bank by counting seeds that are not viable, and by underestimating the number of seeds of hard-to-detect sp ecies (Gross 1990, Jenkins 2003). Although the emergence method and other t echniques for characterizing seed banks can yield information about what seeds are pr esent at a given site at a given time, the relationship between the seed bank compos ition as determined by these methods and patterns of seedling emergence in the field is unclear. Given the multitude of factors that may affect germination, emergence, and surviv al of seedlings in nature, the predictive value of the seed bank for pl ant recruitment is unknown (H yatt 1999). Few attempts have been made to link soil seeds with seedling emergence under field conditions, so there is currently little information about how the seed bank--as characterized using typical greenhouse or laboratory methods--relates to emergence of seedlings under field conditions.

PAGE 16

6 Factors Limiting Seedling Establishment in Fire-Suppressed Longleaf Pine Sites, and Possible Restoration Actions If groundcover species of longleaf pine co mmunities are well represented in a persistent seed bank, there is still the questi on of how best to enc ourage their germination when restoring fire-suppressed areas. Seeds ge rminate in response to a diverse array of stimuli (Baskin and Baskin 1998), so specific conditions may be necessary to encourage germination of desirable speci es rather than und esirable ones (e.g., Bellemakers et al. 1993, cited in Bakker et al. 1996). Both su ccessional changes in fire-suppressed pinelands and the lack of fire -related germination cues coul d inhibit seedli ng recruitment. Although fire or smoke directly stimulates seed germination in many plant species (Keeley and Fotheringham 2000), as yet there is apparently of a strong seed germination response to fire in xeric pi ne communities in Florida (Carrington 1999, Whelan 1985). If seedling recruitment in sandhills is affected by fi re, it therefore is likely that indirect fire effects, rather than direct effects, are responsible. When longleaf pine communities are degraded due to fire suppression, seed ge rmination and seedling emergence may be reduced due to a variety possible mechanisms. Several environmental conditions that can affect seedling establishment, including light environment, diurnal temperature fluctu ations, and surface s ubstrates all change markedly in fire dependent communities when fires are suppressed. Soil disturbance (e.g., Isselstein et al. 2002), removal of comp eting vegetation (e.g., Isse lstein et al. 2002), and removal of litter (e.g., Maret and Wilson 200 5), by mimicking the effects of fire, may all create conditions favorable for ge rmination and seedling emergence. In the absence of fire, there is a buildup of organic material on the surface that can reduce seedling emergence in several ways (Facelli and Pickett 1991, Guzman-Grajales

PAGE 17

7 and Walker 1991, Maret and Wils on 2005). First, litte r can prevent contact between seeds and the mineral soil, preventing seeds from im bibing moisture or in creasing mortality of seedlings that germinate in or on top of the litter layer (Fowle r 1988, Rotundo and Aguiar 2005). Additionally, litter can be a physical barrier to the emergence of seedlings germinating from seeds with insufficient stored reserves to emerge from beneath a deep litter layer (Fenner 1985, Molofsky and A ugspurger 1992, Leishman et al. 2000). The presence of litter also changes the light environment (Pons 2000, Jensen and Gutekunsk 2003) and the chemical environment (Bosy and Reader 1995, Preston and Baldwin 1999) in ways that can inhibit seed germination. A number of other germination cues can be influenced by the increase in vegetation cover in the absence of fire. For example, ve getation structure affect s diurnal patterns of soil temperature (e.g., Breshears et al. 1997). Given that temperatur e fluctuations can influence germination (Probert 2000), attenu ation of daily soil temperature extremes could reduce these stimuli to levels below those necessary to stimulate germination (Rice 1985). Changes in canopy structure also affect light quality and li ght quantity, both of which may play roles in stimulating germination (Roy and Sonie 1992, Pons 2000). During the successional changes that accomp any fire suppression in longleaf pine communities, there are a number of change s that could theoretically reduce seed germination and seedling emergence, both from seed rain and from the seed bank. Returning fire to the system, by changing th e light environment, removing litter, and exposing mineral soil, shoul d theoretically increase s eedling emergence in firesuppressed systems. Unfortunately for rest orationists, increase d litter deposition by broad-leafed trees and the reduction in herb aceous groundcover make prescribed fire a

PAGE 18

8 challenge in long-unburned systems, as hardw ood litter carries fire poorly. Additionally, duff accumulation after years without fire can smolder for prolonged periods, resulting in abnormally high mortality of longleaf pine trees when fire is reintroduc ed, particularly in old-growth stands (Varner et al 2005). Treatments that mimic some of the effects of fire could therefore hold some promise in restor ing groundcover, particularly if a seed bank were present. Study Goals While it is clear that there is a need to restore groundcover vegetation in degraded longleaf pine communities, we currently lack a thorough understanding of soil seed dynamics of groundcover species, and the factors that affect recruitment of these species from seed. My general goals in this study were twofold: To characterize the soil seed banks of longleaf pine sandhills with varying fire histories, to compare seed banks with current vegetation, and to evaluate the potential contribution of soil seed banks to restoration. To characterize patterns of seed germination in longleaf pine sandhills, and to examine how these patterns are affected by litter, canopy structure, and soil disturbance.

PAGE 19

9 CHAPTER 2 LONGLEAF PINE SANDHILL SOIL SEED BANKS IN RELATION TO FIRE HISTORY Introduction The longleaf pine communities endemic to the southeastern United States once covered an estimated 37 million ha from southeastern Virginia south to peninsular Florida and west to east Texas (Frost 1993) Plant species richness (Walker and Peet 1983) and levels of endemism (Peet and Allard 1993) in these commun ities are both very high. Lamentably, over 98% of the presettleme nt acreage of longleaf pine ecosystems is estimated to have been lost over the past two centuries (Noss et al. 1995). Further compounding this loss is the fact that fire suppression has degraded many remnant longleaf pine areas; these co mmunities are highly dependent on frequent, low intensity fire (Heyward 1939, Christensen 1981), and onl y approximately half of the remaining acreage is regularly managed with fire (Outcalt 2000). As a result, longleaf pine communities are considered to be among the mo st endangered in the United States (Noss et al. 1995). Due to concern over the loss and degradati on of longleaf pine forests, woodlands, and savannas, numerous efforts to restore th ese ecosystems are underway. While many of these restoration projects primarily involve tree planting, efforts ar e increasingly focused on the herbaceous groundcover (e.g., Brockw ay et al. 1998, Seamon 1998, Provencher et al. 2001, Mulligan and Kirkman 2002, Jenkins 2003, Cox et al. 2004). Attention to restoring the groundcover is warranted both becau se of its diversity, and because of the

PAGE 20

10 key role of wiregrass (Aristida stricta) and other native bunchgr asses for carrying the frequent, low intensity fires that are nece ssary to the health of these communities (Clewell 1989, Noss 1989). Many of the remaining longleaf pine commun ities are in need of restoration, often due to the effects of years of fire suppressi on. Lack of fire in longleaf pine systems is typically accompanied by increases in fire-i ntolerant trees (Heyward 1939) and declines in herbaceous cover and sp ecies richness (Lewis and Harshberger 1976, Myers 1985). The decrease in herbaceous species is especial ly troubling, given that seeds or plants of few of the potentially hundreds of groundcove r species are commerci ally available in quantities necessary to restore these co mmunities. Clearly, if herbaceous groundcover plants maintain a dormant seed bank in fi re-suppressed longleaf pine communities, such a seed bank could be useful for restoration. Th e goal of this study was to characterize the soil seed banks of xeric longleaf pine comm unities with different fire histories. Methods Study Sites Ordway-Swisher Biological Station (her eafter, Ordway) of the University of Florida is an approximately 3800 ha preserve in western Pu tnam County, Florida, USA. Average annual rainfall at the site is 1 432 mm (Readle 1990), with the majority of rainfall resulting from convective storms during the summer months. Average annual temperature is 20 C (Readle 1990). Topography of the site is gently rolling and soils are nutrient-poor, excessively drai ned typic quartzipsamments in the Candler series (Readle 1990). The dominant vegetation is sandhill (Florida Natura l Areas Inventory 1990); also referred to as high pine (Myers 1990), c onsisting of widely-spa ced longleaf pines (Pinus palustris) with an open to moderately dense midstory of hardwoods, most commonly

PAGE 21

11 turkey oak (Quercus laevis) but occasionally other woody species such as sand live oak (Q. geminata) and Florida rosemary (Ceratiola ericoides) Groundcover on wellmaintained sandhills consists largely of wiregrass ( Aristida stricta var. beyrichiana ) and a diverse assemblage of other grasses, legumes, and forbs. The longleaf pine savannas at Ordway, like most of those in the region, were turpentined, logged, and used for livestock grazing during the 19th and early 20th centuries (Frost 1993, Earley 2003). However, the presen ce of wiregrass, which does not tolerate soil disturbance and does not readily recolonize sites on ce extirpated (Clewell 1989), coupled with the lack of non-native forage gra sses on the study sites, i ndicate that historic land use at the study sites was not intensive. Sandhill areas in four management units with varying fire histories were selected for this study (Table 2-1), which was conducted between August 2001 and October 2002. The -year site was set aside as an unbur ned control at the tim e the preserve was established in the early 1980s and had not been burned since 1966. The -year site was also initially part of the unburned c ontrol, but was burned by prescription in 1991. The -year site was managed under an appr oximately 3-year burn rotation, with fires occurring in 1984, 1986, 1989, 1992, and most recently in 1996. The -year site also had a history of being managed with fire, having been burned 1988, May 1996, and February of 2000. Although the exact burn hi stories prior to the 1980s are not known, during the 1950s and 1960s, sandhills on the prop erty were burned on an approximately 2-year rotation, according to th e recollections of the propert ys caretaker during that time (T. Perry, personal communication). During the period of this study, all sites had

PAGE 22

12 vegetation typical of xeric sa ndhill sites in the region, althou gh differences due to recent fire history were a pparent (Table 2-1). Sample and Data Collection Data were collected from ten 31 x 17 m plot s in each of the four sites. Two 17 x 13 m subplots were established in each plot as part of a rela ted study of seedling emergence. Two soil samples from each of eight random point s within each subplot were collected in September 2001 using a 5.46 cm diameter PVC t ube to remove the top 2 cm of soil. Following collection, samples were placed in cl osed paper bags to minimize exposure to light. Half of the samples from each plot (749 cm3 total volume per plot; 7492 cm3 per site) were combined and spread in a layer 0.39 cm deep over a mix of potting soil and sand in flats in the greenhouse within 10 days of collection. These samples are hereafter referred to as the fresh soil samples. The remaining eight samples from each plot (749 cm3 total volume per plot; 7492 cm3 per site; hereafter the stratified samples) were combined, moistened with 5 ml distilled water, placed in plastic bags, and stored in the dark in a refrigerator at 7o C for 6 mo (i.e., until April, 2002). Because most groundcover plants in this community flower in fall and disperse seeds in late fall and early winter, sampling during September should minimize th e influence of recently-dispersed seeds that remain in the soil for only a short period (the transient seed bank, Thompson and Grime 1979) so that the data as closely as possible represent the persistent seed bank (i.e., seeds which persist in the soil for more than one year, Thompson and Grime 1979). Seedling emergence from fresh and stra tified soil samples was observed for 6 months under greenhouse conditions. Fresh soil samples were exposed to natural daylight conditions in a temperature-controlled gree nhouse and watered every other day. In April 2002, the stratified soil samples were potted in the same manner as the fresh samples.

PAGE 23

13 These samples were kept in a separate temp erature-controlled greenhouse, with natural light augmented with artifici al lighting to mimic a natural light levels and daylength schedule. All flats were monitored for new seedlings daily for the first 2 weeks after being placed in the greenhouse, and weekly thereafter for six months. Seedlings were removed as they were identified (follo wing Wunderlin 1998) or were repotted and allowed to grow until identification was possi ble. In each greenhouse, 20 flats containing a layer of fine sand over the potting soil-sand mix were used as controls to detect contaminant seedlings. Seedlings in the treatment flats were assumed to be contaminants if the same taxon was found in the control flat s. Flat locations were shifted at random every two weeks. Groundcover Vegetation Sampling Presence/absence of all herbaceous species in each plot was recorded by searching the plots during October-November 2001, May 2002, and August-September 2002. A prescribed fire in the 10-year site in summer 2002 prevented collection of complete vegetation data from 3 of the 20 subplots, so those subplots were omitted from analyses when necessary. Statistical Analyses Statistical analyses were conducted usi ng SAS software (SAS Institute 2000). Counts of total seedlings emerging were s quare root transformed (Underwood 2000) and percentages were arcsin-square root transforme d prior to analysis. Analyses of variance were conducted using SAS proc ANOVA. Simila rity analyses were calculated using incidence-based Jaccard simila rty estimators (Chao et al. 2005) found in the EstimateS 7.51 software (Colwell, R. K. 2004. ESTI MATES: Statistical Estimation of Species Richness and Shared Species from Samples, Version 7.5. University of Connecticut,

PAGE 24

14 Storrs, Connecticut, USA. http://viceroy.eeb.uc onn.edu/estimates Persistent URL http://purl.oclc.org/estimates. Last accesse d May 2007.). This similarity estimator has been shown to be less sensitive to sample size than classic methods of calculating similarity (Chao et al. 2005), in that it comp ensates for the downward bias in similarity which results from the tendency of rare spec ies note being shared among samples due to sampling error. For comparing species richne ss among sites and stratif ication treatments, two methods were used. Coleman rarefacti on (Colwell and Gotelli 2001) was used to rescale the samples to remove the influe nce of varying numbers of individuals on estimates of species richness. Species ric hness was also estimated using the incidencebased coverage estimator of species ric hness (ICE; Colwell 2004). A comparison of several diversity sta tistics found that ICE was less influenced by sample size and patchiness than other commonly-used estimat ors (Chazdon et al. 1998). An added feature of ICE is that it is designed to estimate the true species richness of an assemblage under the assumption that some species are missing from the dataset due to sampling (Colwell 2004). I explored whether seedlings represente d species that were common in the sandhill vegetation by calculating the proportion of plots where adults of each species were encountered (field frequency). The mean fi eld frequency for each plot is therefore the mean of field frequency values for all seedli ngs that emerged from soil samples from that plot. To compare effects of site and stratification on fiel d frequency, SAS proc GLM was used, because missing field vegetation data from 3 subplots resulted in an unbalanced design. Because fire history is not a replicated treatment in th is study, statistical tests cannot test for an effect of fire management per se (Hurlbert 1984). Instead, differences

PAGE 25

15 among sites will be referred to throughout as si te effects. Significant site effects may be related to a variety of fire-related variable s, or to other unknown variables that differ among the study sites. Results Timing of Seedling Emergence In both fresh samples (maintained in the greenhouse between September 2001 and March 2002) and stratified samples (maintai ned in the greenhouse from April-October 2002), the cumulative number of seedlings emerging increased roughly linearly for the first three weeks. Following the initial 3 w eeks, emergence rates declined, but small numbers of seedlings continued to appear throughout the study During December 2001 through March 2002, a pronounced second pulse of seedling emergence occurred from the fresh samples, consisting primarily of speci es that had not previ ously been recorded. Species appearing during this later period included Aureolaria pedicularia var. pectinata, Agalinis filifolia, Linaria c anadensis, Ceratiola ericoides Paronychia patula, and Habenaria repens Emergence of several of these species (Agalinis, Aureolaria, Paronychia, Linaria) coincided with their emergence in the field (pers. obs), suggesting that germination in these species occurs in resp onse to cues that were still present in the controlled environment of the greenhouse or resulted from endogenous rhythmicity. No second pulse of germination was obs erved in the stratified samples. Seedling Density A total of 932 seedlings emerged over 6 months from the fresh soil samples. The mean density of viable seeds ra nged from a low of 576 seeds/m2 in the 1-year site to a high of 1200 seeds/m2 in the 10-year site (Figure 2-1) From the stratified soil samples, 467 seedlings emerged over 6 months. Buried seed density as estimated from stratified

PAGE 26

16 soil samples were again lowest in the 1-year site at 372 seeds/m2 and highest in the 10year site at 620 seeds/m2. The number of seedlings emerging varied significantly among sites (Table 2-2), with fewer seedlings emerging from the 1-year site relative to all other sites except the 35-year site. Stratification reduced seedli ng emergence by 48.9% overall. There was no interaction between stratification and site. Taxa Represented Most seedlings (94.5 % fresh; 87% of stra tified) developed sufficiently before the end of the study to be identified to species or genus (Table 2-3). Due to space limitations, not all individuals of the most common species (e.g., Gnaphalium, Eupatorium, Bulbostylis) were grown to maturity; although multip le species within each of these genera are known to have occurred, results are here given at the genus level. One vegetatively distinctive morphospecies in the Cyperaceae was only iden tifiable to family. The term species as used below refers to species when known, and to genus or morphospecies as noted above. A total of 43 species were recorded from fresh soil samples, while 39 species emerged from strati fied samples. Twenty-two species found in the fresh samples were not observed in the stratified samples. Conversely, thirteen species in the stratified samp les were not found in the fresh samples (Table 2-3). The most abundant seedlings in each st udy site are shown in Figure 2-2. Ruderal taxa (Eupatorium, Gnaphalium ) and wetland taxa ( Juncus Xyris ) were prominent in all study sites. The few typical sandhill speci es that occurred consistently (e.g., Bulbostylis spp., Stipulicida setacea ) are short-lived species genera lly found in the most open, sandy microsites.

PAGE 27

17 Soil sample stratification reduced the do minance of some of the most abundant species, while apparently stimulating germina tion of a few species not observed emerging from fresh soil. Strikingly, no seedlings of the Fabaceae emerged from the fresh samples, whereas 25 individuals from 5 Fabaceae species emerged from stratified soil samples. For example, Dalea pinnata, a common sandhill legume, was among the most frequently found species in the stratified samples from the 5-year site (found in 11 of 20 subplots), but did not emerge from the fr esh samples from that site. A few sandhill species were common in the fresh samples but not in the stratified samples. These species included Agalinis filifolia and Aureolaria pedicularia var. pectinata Balduina angustifolia and Polanisia tenuifolia which are occasional to common components of the sandhill ve getation at Ordway. Additionally, Ceratiola ericoides a shrub species which is more characteristic of Florida scrub communities (Myers 1990) but occurs in sandhills in the region, emerged only from fresh soil samples. Species Richness Seedling species richness in the soil sample s was low, ranging from an average of 2.9 species per plot in the stratified samples fr om the 35 year site to 4.9 species per plot in the fresh samples from the 10 year site (F igure 2-3). Mean species richness per plot did not vary across the study sites or stratific ation treatment. Species richness estimated by Chaos Incidence-based Coverage Estimator for species richness (ICE) varied widely among and within sites and stratif ication treatments, from a low of 17 species in stratified soil from the 5-year site to a high of 55 speci es in the fresh samples from both the 1and 5-year sites (Figure 2-4). Rarefaction curves (F igure 2-4) show that, with the exception of fresh soils from the 10-year site and stratifie d soils from the 5-year site, the number of

PAGE 28

18 species encountered had not reac hed an asymptote at the leve l of sampling used in this study. Similarity of Seeds in Soil Samples to Field Vegetation Incidence-based Jaccard similarity be tween the soil seed community and the aboveground vegetation was low; similarity among seedlings from soils taken from different sites (Jaccard estimator 0.421-1.0) was greater than similar ity of soil seedling composition to vegetation (Jaccard estimator 0.034-0.377, Table 2-4). Only a small proportion of the species found in the soil samp les were also recorded in the aboveground vegetation in the same study plot, and conve rsely only a small proportion of the species observed in field plots emerged as seedlings from the soil samples. The average number of species in common between the field pl ots and their respective soil samples was similar between fresh and stratified soil, despite the fact that fewer seedlings emerged per sample from the stratified soil. To further ev aluate the similarity between the seed bank samples and aboveground vegetation, I determ ined the frequency of all species that emerged as seedlings from the soil sample s in the aboveground vegetation by calculating the percent of field plots in each study s ite in which a species was found. Mean field frequency of emerging seedlings varied by st udy site and stratificat ion treatment (Figure 2-5), but there was no interaction (Table 25). Post-hoc mean separation using Tukeys Studentized Range Test indicat e that the mean field freque ncy of seeds emerging from the 5-year site was higher than the field fr equency of seeds emerging from the 35-year site; all other means we re indistinguishable. Discussion Although there was a moderately high density of soil seeds in all sites, one group of species, native bunchgrasses, including wiregr ass, were notably lacking. This result is

PAGE 29

19 similar to results of previ ous studies in longleaf pine communities (Jenkins 2003, Cohen et al. 2004). Although others have observed that wire grass appears to maintain at least a short-term seed bank (Seamon 1998, Mulligan and Kirkman 2002, Cox et al. 2004), no wiregrass seedlings emerged from the soil sa mples. Wiregrass is dependent on growingseason fire for seed production (Clewell 1989, Outcalt 1994), and the most recently any of the plots had received a growing season burn was in 1997, four years before sample collection. These data suggest th at wiregrass seeds do not persis t longer than 4 years, or that the density of surviving s eeds was too low to detect. Alte rnatively, it is possible that wiregrass seeds persisted in soil but did not germinate under the conditions in the greenhouse. Whatever the explanation, a lack of a long-term wiregrass seed bank is consistent with the observation that this sp ecies does not readily ecolonize sites from which it has been extirpated (Clewell 1989). Although the vast majority of herbaceous species in the sandhill community are perennials, most of the seedlings of typical sandhill plants that germinated from the soil seed bank were from the handful of annual species (e.g., Agalinis filifolia, Polanisia tenuifolia) and biennial species (Balduina angustifolia) that occur at the site. This finding is likely related to the importa nce of reproduction from seed in the life history of species with short-lived adults. As is commonly found in seed bank studies, the sandhill seed bank was dominated by ruderal species and was di ssimilar in species composition to the aboveground vegetation (e.g. Rice 1989). It is un clear to what extent the seed bank at these sites is the result of seed input from surrounding area s, or the legacy of past disturbances occurring at the site (Dav ies and Waite 1998). Many of the seedlings

PAGE 30

20 emerging from the soil seed bank were of species commonly found in much wetter habitats (e.g., Typha sp., Eupatorium capillifolium and Juncus spp.), suggesting that nearby wetlands contribute substantially to the seed banks of upland sites. Given the xeric conditions on sandhills, it is unlikely that many of these species would survive or reproduce onsite even if their seeds germinat ed.The dominance of short-lived and ruderal species is not surprising from a populati on biology standpoint. Natural Selection on short-lived species to produce dormant s eeds should be relatively strong, because otherwise an entire populati on could be eliminated by a single year of unfavorable recruitment conditions (Cohen 1966). Additionally, the availability of recruitment sites for many ruderal species is spatially and temporally variable, and accordingly, their production of large numbers of often well-di spersed seeds is well documented (Harper 1977). By contrast, the groundcover of longleaf pine sandhill communities is dominated by perennials which generally survive and r eadily resprout following fires. Seedling recruitment is likely to be more frequent a nd more demographically important for ruderal and short-lived species than for the long-lived perennials that compose the bulk of this community. The seed bank at these sites was domina ted by species that are not considered characteristic of sandhill ecosystems. This finding points to a weakness in defining a community on the sole basis of the speci es observed aboveground. Others (Major and Pyott 1966) have argued that species present in the seed bank of a site should be considered part of the community because th ey contribute to the potential vegetation of the site. Changing conditions over time in th ese sites could make them conducive to the growth and reproduction of some of the species represented by seeds in the soil; even if

PAGE 31

21 seedlings of these species have low survival in the field, large scale disturbance could result in their dominance, simply due to the large number of seeds present. The seed bank density recorded at the st udy sites falls within the range of that found in other natural communities, particularly fire-dependent grasslands and savannas, and communities in the southeastern Unite d States (Table 2-6). In making such a comparison, however, there are several compli cating issues to keep in mind. For one, many studies fail to make a distinction betw een transient and pers istent seed banks (Thompson and Grime 1979; Baskin and Baskin 1998) and thus may count large numbers of seeds that are only in the soil for a shor t time. Transient seed banks and the operation of other processes such as pr edation and germination over the course of a year contribute to intrayear variation in seed bank density (e.g., Gashaw et al. 2002, Williams et al. 2005) that can be substantial, making careful choice of sampling season an important consideration (Baskin and Baskin 1998). Fina lly, methodological di fferences can cause differences among studies: there is little stan dardization in sampli ng depths across seed bank studies, and when the seedling emergence method is used to assay for soil seeds, environmental factors may influence which seeds germinate. The data presented here suggest that soil seed densities in sa ndhills initially are low after fire, then increase for a few years post-fire before eventually declining. Studies of the relationship between di sturbance and soil seed bank density in other ecosystems report a similar pattern ov er the course of succession (e.g., Dupuy and Chazdon 1998, but see Zammit and Zedler 1994). Although mo st of these studies focused on canopy disturbances in forest ecosystems, the data presented here suggest that soil seed dynamics follow similar patterns during postfire succession in open-canopied communities.

PAGE 32

22 Nevertheless, because the reproductive res ponse to fire by many of the common pine savanna species is rapid (Abrahamson 1984, Platt et al. 1988, Clewell 1989), it is somewhat surprising that the 1-year post fire study unit had the lowest soil seed density, despite having more than a full growing season after fire for seeds to accumulate. Several factors may explain this resu lt. First, although many species (particularly wiregrass) produce abundant seed during the first growing season after fire, others (e.g., Dalea pinnata and Sorghastrum secundum pers. obs.) have a postfir e reproductive peak in the second season after fire. Seeds of these specie s would not be expected in large numbers in soil from the 1-year site. Given that few soil seeds are attributab le to present local vegetation, it is possible that if heat from fires kills some soil seeds as has been suggested in other plant communities (Holl et al. 2000), soil seed density may be low immediately after fire and increase over time as seed s dispersing from surrounding areas accumulate. Although there are clearly draw backs to using germinati on as a seed bank assay (Gross 1990), these disadvantages can be mitig ated. The differences observed in seedling species composition between fresh and stratified soil samples in this study indicate that detection of some species that are relativ ely common is improved by using a variety of conditions during germination trials (Gross 1990, Whittle et al. 1997). Most notably in this study, legume seedlings were absent from fresh soil samples, wh ile stratification of soil taken from the same sites resulted in a bundant germination of five legume species. Stratification of soil samples augmented the total number of species detected in the germination assays. Nonetheless, the lower seed density estimated from the stratified samples suggests that many seeds died or went into deep dormancy during storage. Thus, using stratified samples alone may result in inaccurate estimates of soil seed density, but

PAGE 33

23 their use in tandem with fresh samples may gi ve a more complete picture of soil seed bank composition. A number of factors (e.g., se ason and seed age) were confounded with the stratification treatment in this study. Nonetheless, whet her stratification or another factor caused the differences between the tw o treatments, the overall conclusionthat a variety of germination conditi ons allows a more complete glimpse of the buried seed poolis no less true. The finding that cold stratification of soil affected seed germination of in a region where winter temperatures seldom fall to n ear freezing is somewhat surprising, as it would not be expected that these species woul d have the opportunity to evolve a response to cold. However, the increase in legumes from stratified samples may be explained by the fact that the species pres ent in this study ar e temperate represen tatives of a plant family that originated in the tropics. It is possible that Fabaceae ta xa that were able to spread into temperate zones were those that could time their germination to avoid being killed by cold weather. Alternatively, the di fference may result from a more generalized temperature response. Temperature changes often break dormancy in physically dormant seeds (Baskin and Baskin 1998); it may be th at both high and low temperature extremes are capable of bringing about the physical changes needed to allow seeds to imbibe moisture and germinate. Although most seedlings emerged from the soil samples within a short time after being placed in the greenhouse, the data sugge st that even under controlled conditions, seasonal patterns of germination were st ill present. Nearly synchronous, delayed germination by several species, coinciding with the emergence of these species in the field (Chapter 3; pers. obs.), suggests that cues triggering or delaying germination still

PAGE 34

24 affect germination patterns in the greenhouse. So me of the species that only occurred in fresh samples (e.g. Agalinis filifolia, Paronychia patula, and Aureolaria pedicularia var. pectinata) only emerged during winter and early spring (mostly December through March). Since seedling emergence from the st ratified samples was only monitored from April until November, seeds of these species may still have been present and viable in the stratified samples, but dormant during the peri od of observation. While virtually all seed bank studies that use the emergence method sh ow that the vast majority of seedings emerge quickly after being placed in germination flats, the difference in composition between earlyand later-emerging seedlings in the fresh samples suggests that observing seedling emergence for a full year may give a more complete picture of the soil seed pool. Although the density of buried seeds in the sandhill surface soils was high, the composition of the seed bank suggests that bur ied seeds can not play a large role in ecosystem restoration. Notably, the most dom inant and characteristic species of this community--bunchgrasses and perennial forbswe re nearly absent, as reported in other studies of seed banks in longleaf pine communities (Jenkins 2003, Cohen 2004). That said, several species were represented only by single individuals, suggesting that with more extensive sampling, more species would be encountered. While the density of seeds of characteristic species would not be enough to restore function to a degraded sandhill, over a large area, a relatively diverse asse mblage of species could nonetheless be recovered from the soil seeds if conditions were favorable. An evolutionary question raised by thes e results is why, when species in many other fire-dependent communities build up substantial seed banks, is this strategy so

PAGE 35

25 uncommon among dominant longleaf pine sa ndhill species? A potentia l answer to this question is that sandhills may provide a re latively predictable environment. Seed dormancy should be an advantageous stra tegy when environmental conditions are unpredictablein such cases where the re duction in fecundity resulting from seed dormancy is offset by the reduced risk of lo sing all of ones offspring in a bad year. Sandhill fire regimes are relatively frequent, with fire occurring every 2-5 years. If fires are frequent enough to be predictable, there might not be selection to produce dormant seeds, and thus a seed bank would not be expected to develo p. Alternatively, as Thompson (2000) argues, several traits other than seed dormancy, including large seed size, seed dispersal, and long adult lifespan, also can act as buffers against environmental unpredictability. Since the vast majority of sandhill species are perennials, selection for seed dormancy may be largely absent because established adults of most species have a high probability of surviving to produce seeds during favorable times.

PAGE 36

26 Table 2-1. Fire history and vegetation characte ristics of four sandhill sites in Putnam County, Florida. 1 year 5 year 10 year 35 year Years since previous Firea 5935 Unknown Number of burns since 1984 351 0 Longleaf pine density (# trees 10cm dbh/ha ) b 316.7.85119.9.60142.5.08 153.8.21 Longleaf pine basal area ( m2/ha ) 11.6.665.8.906.9.33 6.6.82 Turkey oak density (# trees 10 cm dbh/ha ) 40.7.0058.8.5299.5.49 239.8.41 Turkey oak basal area ( m2/ha ) 2.5.564.7.015.2.29 10.8.63 Graminoids (% cover) 51.3.8746.3.0133.3.86 12.8.95 Litter depth (cm) 1.5.372.1.792.9.79 2.7.61 Bare ground (% cover) 22.9.5312.1.565.2.90 2.7.79 a Number of years since the s econd most recent fire in the study sites at the time of the study (2001-2002). b N=10 plots per site for vegetation data.

PAGE 37

27 Figure 2-1. Mean ( S.E.) density of viable seeds in soil samples taken from four sandhill sites with different fire hist ories. Seed density was determined by observing seedlings emerging in a greenhouse over 6 months.

PAGE 38

28 Table 2-2. Analysis of Variance table for e ffects of site and cold stratification on estimated seed density in soil samples ta ken from sandhill sites with different fire histories. Source DF Sum of squares Mean square F Pr > F Site 311.963.996.7 0.001 Stratification 133.5033.5056.6 <0.001 Site x stratification 32.460.821.4 0.262 Plot(site) 3622.300.621.1 0.444 Replicate(plot) 7972.460.921.6 0.071 Seed density was estimated by observing seedling emergence from soil samples in a greenhouse for 6 months.

PAGE 39

29 Figure 2-2. Density of seeds (Mean S.E.) f ound in fresh and stratified soils taken from the top 2 cm of soil in four sites with di fferent fire histories and germinated in a greenhouse. N=10. A) 1 year since fire B) 5 years since fire. C) 10 years since fire. D) 35 years since fire. Gnaphalium spp. includes Pseudognaphalium falcatum P. pennsylvanicum and P. purpureum Eupatorium includes E. compositfolium and E. capillifolium Cyperus retrorsus includes C. retrorsus and C. croceus

PAGE 40

30 Figure 2-2. Continued

PAGE 41

31Table 2-3. Seeds found in stratified and fresh soil samples ta ken from 4 sandhill sites with different fire histories. Taxon Family Stratified samples Fresh samples 1-year 5-year 10-year 35-year 1-year 5-year 10-year 35-year Typha sp. (w) Typhaceae 651 6 Andropogon virginicus (s,r) Poaceae 11 Aristida sp. (s, r) Poaceae 23 3131 Dichanthelium aciculare (s, r) Poaceae 12 1113 Digitaria filiformis (s,r) Poaceae 19 15 Eragrostis elliottii (s, r) Poaceae 1 Bulbostylis (s)a Cyperaceae 132215 142315206 Cyperus filiculmis (s) Cyperaceae 6128 1333 Cyperus plukenetii (s) Cyperaceae 1 2 Cyperus retrorsus (s, r, w) Cyperaceae 122 82194 Cyperus sp. Cyperaceae 132 Cyperaceae sp. Cyperaceae 2 Xyris sp. (w) Xyridaceae 132 123 Lachnocaulon sp. (w) Eriocaulaceae 1 Cuthbertia rosea (s) Commelinaceae 21 Juncus elliotii (w) Juncaceae 1 11 Juncus marginatus (w) Juncaceae 24 2295 Juncus tenuis (w) Juncaceae 11 3155 Juncus sp. 4 (w) Juncaceae 1 Juncus sp. 5 (w) Juncaceae 1 Hypoxis sp. (s) Hypoxidaceae 4 Habenaria repens (w) Orchidaceae 2 Polygonella gracilis (s) Polygonaceae 1 Drymaria cordata (r) Caryophyllaceae 1 Paronychia patula (s) Caryophyllaceae 2 Stipulicida setacea (s) Caryophyllaceae 3523 102147222 Polanisia tenuifolia (s) Brassicaceae 161 Rubus cuneifolius (r) Rosaceae 2 Crotalaria rotundifolia (s, r) Fabaceae 2

PAGE 42

32Table 2-3. Continued Taxon Family Stratified samples Fresh samples 1-year 5-year 10-year 35-year 1-year 5-year 10-year 35-year Crotalaria spectabilis (r) Fabaceae 1 Dalea pinnata (s) Fabaceae 19 Sesbania macrocarpa (r) Fabaceae 2 Tephrosia chrysophylla (s) Fabaceae 1 Oxalis corniculata (r) Oxalidaceae 1 Tragia urens (s) Euphorbiaceae 1 Ceratiola ericoides (s) Ericaceae 1610 Eryngium aromaticum (s) Apiaceae 2 1111 Polypremum procumbens (r) Tetrachondraceae 12 Linaria canadensis (r) Scrophulariaceae 12 Scoparia dulcis (w) Scrophulariaceae 121 Agalinis filifolia (s) Orobanchaceae 1106243 Aureolaria pedicularia var. pectinata (s) Orobanchaceae 5 Hedyotis procumbens (s, r) Rubiaceae 11 Hedyotis uniflora (w) Rubiaceae 11 221 Wahlenbergia marginata (r) Campanulaceae 1 Baccharis halmifolia (r, w) Asteraceae 11 11 Balduina angustifolia (s) Asteraceae 31 Conyza canadensis (r) Asteraceae 12 Erechtites hieracifolia (r) Asteraceae 544 133 Eupatorium spp. (r) Asteraceae 13613 1623122950 Gnaphalium spp. (r) Asteraceae 91020 1561578875 Gnaphalium sp 4 (r) Asteraceae 1 Pluchea rosea (w) Asteraceae 1 Pterocaulon pycnostachyum (s, r) Asteraceae 4 3 Solidago fistulosa (w) Asteraceae 1 Soil samples were taken from top 2 cm of soil; total sample volume was 7492 cm3 per site for each stratification treatment; total sample area was .375 m2 per site for each stratification treatment. a Typical habitat for each species, according to Wunderlin (1998). S=sandhill; w=wetland; r=ruderal.

PAGE 43

331 year since fire 5 years since fire Figure 2-3. Mean species richness and overlap of species between vegetation and s eeds from soil samples taken from four sandhil l sites. Fresh soil Field vegetation 44.9 3.87 species 3.4 0.87 species Stratified soil 3.0 1.25 species 0.5 0.47 species 0.5 0.47 species 1.4 0.61 species Fresh soil Field vegetation 41.2 10.97 species 4.1 1.65 species Stratified soil 3.2 1.16 species 1.2 1.03 species 1.4 0.63 species 0.8 0.42 species

PAGE 44

34 3410 years since fire 35 years since fire Figure 2-3. Continued 1.5 0.67 species Fresh soil Field vegetation 36.6 7.84 species 4.9 1.53 species Stratified soil 3.4 1.93 species 1.6 1.33 species 1.1 0.89 species Fresh soil Field vegetation 27.0 3.07 species 4.8 1.03 species Stratified soil 2.9 1.10 species 0.8 0.75 species 0.4 0.41 species 1.2 0.67 species

PAGE 45

35 Figure 2-4. Coleman rarefaction curves and es timated species richness for seeds found in soil samples taken from sandhill sites with different fire histories. A) Fresh soil. B) Stratified soil. Samples were taken from the top 2 cm of soil; 7492 cm3 total volume, .375 m2 total area sampled per site per stratification treatment. Species richness is estimat ed using incidence-based coverage estimates (ICE, Chazdon et al. 1998) and shown at the righ t margins of the graph.

PAGE 46

36 Figure 2-5. Mean field frequenc y of seeds detected in stra tified and fresh soil samples from sandhill sites with differing fire histories. Field frequency is the proportion of field plots in which an adult of each species was recorded. N=10.

PAGE 47

37 Table 2-4. Analysis of Variance using SAS pr oc GLM testing effects of site and soil stratification on the species composition of seeds detected in soil samples taken from 4 sandhill sites with different fire histories. Source DF Sum of squares Mean square F Pr > F Site 3352711768.81 <0.001 Stratification 11893189314.19 <0.001 Site x stratification 310223412.55 0.063 Plot(site) 3652171451.09 0.444 Replicate(plot) 4046011150.86 0.691 The response variable, field frequency, was dete rmined for each seed as the proportion of field plots where an adult of that species was found.

PAGE 48

38Table 2-5. Incidence-based Jaccard estimator of similarity among vegetation, seeds detect ed in fresh soil, and seeds detected i n stratified soil from 4 sandhill sites with different times since fire. Vegetation Fresh soil Stratified soil 5-year 10-year 35-year 1-year 5-year 10-year 35year 1-year 5-year 10-year 35-year 1-year vegetation .930 .885.869.269.236.034 .172.094.095.111.081 5-year vegetation .847.802.259.307.056 .118.077.139.113.087 10-year vegetation .870.377.341.093 .249.108.137.167.159 35-year vegetation .205.134.043 .109.046.116.064.082 1-year fresh soil .907.842 .841.785.780.855.985 5-year fresh soil .753 1.000.610.561.652.811 10-year fresh soil .794.531.421.613.684 35-year fresh soil .611.484.790.963 1-year stratified soil .6661.0001.000 5-year stratified soil .592.881 10-year stratified soil 1.000

PAGE 49

39 Table 2-6. Comparison of estimated soil seed de nsities in selected temperate/subtropical fire-dependent communities. Seeds per square metera Plant community Location Source 20 Sand pine scrub Florida, USA Carrington 1996 100-4,700 Afromontane grassland/savanna/woodland Ethiopia Gashaw et al. 2002 186 Fire-suppressed Ponderosa pine South Dakota, USA Wienk 2004 300-1,300 Pinus halepensis woodland Israel Neeman and Izhaki 1999 372-1,200 Longleaf pine sandhill Florida, USA This study 500 Quercus-Ilex woodland Spain Trabaud 1994 700-800 Wet prairie Florida, USA Wetzel et al. 2001 11,200 Quercus suber woodland Southern Spain Diaz-Villa et al. 2003 14,125 Quercus canariensis woodland Southern Spain Diaz-Villa et al. 2003 16,690 Pine flatwoods Florida, USA Jenkins 2003 31,800 Grassland Southern Spain Diaz-Villa et al. 2003 43,032 Pasture undergoing restoration to pine flatwoods Florida, USA Jenkins 2003 aSoil seed density was estimated in each study by counting seedlings that emerged from soil samples.

PAGE 50

40 CHAPTER 3 DICOT SEEDLING EMERGENCE IN BURNED AND UNBURNED LONGLEAF PINE SANDHILLS IN RELATION TO OVERSTORY, LITTER, AND SOIL DISTURBANCE Introduction Soil seed banks, documented in a wide variety of natural communities worldwide, occur when ungerminated seeds persist in th e soil. Seed banks benefit plants when conditions favorable for germination, recruitm ent, or survival are temporally variable (Cohen 1966, Thompson 2000). Because persistent seeds may preserve elements of the vegetation of a site after e nvironmental conditions change, seed banks have the potential to contribute to restoration of degraded natural communities (Bakker et al. 1996). Longleaf pine (Pinus palustris) savannas, endemic to the southeastern United States, are an example of a community that c ould benefit from the use of soil seed banks in restoration. Once occupying over 37 million hect ares of the southeastern coastal plain (Frost 1993), longleaf pine communities have been reduced to less than 2% of their former extent (Noss et al. 1995). These co mmunities are among the most biologically diverse plant assemblages in North America (Walker and Peet 1983), due primarily to a species-rich herbaceous groundcover. Frequent, lo w intensity fires are characteristic of these habitats (Christensen 1981) and in their absence, the density and diversity of the herbaceous groundcover decline dramaticall y, while hardwood canopy cover increases. Many of the remaining examples of longleaf pine communities are now degraded due to fire suppression (Outcalt 2000; Va rner et al. 2005). However, there has been a great deal of recent interest in restoring these co mmunities (e.g., Brockway et al. 1998, Seamon

PAGE 51

41 1998, Mulligan and Kirkman 2002, Provencher et al. 2001, Jenkins 2003, Cox et al. 2004). If herbaceous plants in longleaf pine communities are represented in a persistent seed bank, restoring canopy structure and the fire regime may result in the restoration of groundcover composition without more active replanting efforts. Although soil seed banks have been studied in a wide variety of plant communities, only a few studies (e.g., Kitaji ma and Tilman 1996)have as sessed the connection between the composition of soil seed banks and the environmental conditions that determine whether or not soil seeds germinate. Litter, so il disturbance, and light environment (Pons 2000) are among a host of environmental fact ors that affect seed germination and seedling emergence in nature; attempts to ta ke advantage of soil seed banks in fire suppressed longleaf pine habita ts would benefit from information on the importance of these factors for emergence of seedlings. This study had three goals: to study patterns of dicot seed ling emergence in burned and fire-suppressed longleaf pine sandhills; to determine how soil disturbance, litter, and canopy structure affect emergence of seed lings in degraded and healthy sandhill communities; and to compare seedling emergence in the field with the soil seed bank as characterized by seedlings that emerged from soil samples taken at the same sites. Methods Study Site This study was conducted at the Ordway-Swi sher Biological Station (Ordway), an approximately 3800 ha preserve owned by the Un iversity of Florida in Putnam County, Florida (N 29 41, W 81 74). Rainfall aver ages 1432 mm per year, falling mostly due to convective storms from June to Sept ember. Mean annual temperature is 20 C.

PAGE 52

42 Data were collected between 2001 and 2002 in two sites with different fire histories. The first site (her eafter the burned site) had been burned frequently since the establishment of the preserve in the early 1980s, and most recently 5 years prior to the initiation of the study in 2001. The second site (the unburned site) is adjacent to the burned site, but had not been burned si nce 1966. Soils in both sites are typic quartzipsamments of the Candler series (Read le 1990), and vegetation in both sites was typical of xeric longleaf pine sandhills, consisting of an ove rstory of scattered longleaf pine with an open to moderately dense midstory of turkey oak (Quercus laevis) and a groundcover consisting of bunchgrasses and forb s. Although dominant plants were the same at both sites, vegetation structure differe d due to variation in past fire management (Table 3-1). Study Design In each site, 10 pairs of 17 x 13 m plot s were established at random locations. Within each plot, four 1x1 m subplots were established. Between July and August 2001, all midstory hardwoods and shrubs (predomin antly turkey oak) were removed from one randomly selected plot from each pair. Pines we re left standing. Trees were felled with chainsaws and removed by hand from the plots in such a way as to minimize soil disturbance, and precautions were taken to fell trees away from the sampling subplots. Resprouts were periodically rem oved over the course of the study. Each of the four subplots in each plot were randomly allocated to one of four soil treatments (soil disturbance, litter removal, st erilization, and unma nipulated control). Plots in the soil disturbance treatment were raked thoroughly to a depth of 5 cm with a council rake. All litter was remove d from the litter plots using a leaf rake, with care taken to leave live vegetation and the underlying mine ral soil intact. For th e soil sterilization

PAGE 53

43 treatments, the top 3 cm of soil was removed fr om the plots and placed in an autoclave at 100 C for 30 minutes. Woven geotextile weed barrier cloth was used to line the 1x1 m depression to prevent sprouting from roots le ft in the ground, and th e sterilized soil was replaced and leveled. The intent of this treat ment was to kill any buried seeds present so that sterilized plots would serve as an assay of seed rain. Although autoclaving may affect soil chemistry, Williams-Linera and Ew el (1984) showed that autoclaved soil did not affect germination, growth, and survival of seedlings of severa l tree species from several familes. Greenhouse tests of a variety of sandhill species determined that seeds of all species tested germinated and grew in autoclaved sandhill soils. Finally, litter, soil, and vegetation of control subplots were left intact. Data Collection Seedling surveys Sample plots were surveyed every 15 days for one year beginning in September 2001. During each survey, all newly-emerged dicot seedlings were recorded and marked with numbered toothpicks for later relocati on. When possible, seedlings were identified using a reference collection grown from s eed in the greenhouse. Plants that were suspected to be seedlings but that lacked obvious cotyledons were excavated to determine whether they were connected to nearby plants or established belowg round structures. Due to the difficulty in distinguishing monocot s eedlings from new ramets of existing plants, monocots were not included in this study. Fi eld observation and greenhouse germination of field-collected seeds showed that seedlings of virtually all of th e common dicots at the site with the exception of Rhynchosia spp. (Fabaceae) could be readily identified as seedlings based on the presen ce of visible cotyledons.

PAGE 54

44 Canopy structure To calculate the effects of hardwood removal on canopy openness, hemispheric canopy photos were used. In October 2001, photos were taken in from the center of each plot using a Nikon Coolpix digi tal camera with hemisperic lens positioned on a 30 cm tall tripod. Canopy openness in each photo was cal culated using GLA software (Frazer, G. W., Canham, C. D., and K. P. Lertzman. 1999. Gap Light Analyzer (GLA), Version 2.0: Imaging software to exract canopy structure a nd gap light transmission indices from truecolor fisheye photographs. Simon Fraser University, Burnaby, British Columbia, Canada. http://www.rem.sfu.ca/forestry/dow nloads/gap_light_analyzer.htm Last accessed: July 2007). Vegetation composition Presence of all herbaceous species in each plot was recorded by searching the plots during October-November 2001, May 2002, and August-September 2002. Species identification followed Wunderlin (1998). Weather data Between August 2001 and July 2002, climate data were recorded by an automated weather station at Ordway. In July 2002, the station was disabled by a lightning strike; rainfall data during the remainder of th e study (July and August 2002) were obtained from a regularly-monitored rain gauge located adjacent to the preserve. Daily values of the Keetch-Byram Drought Index (KBDI; K eetch and Byram 1968) for Putnam County were acquired from the Florida Division of Forestry. KBDI uses precipitation and maximum temperature to predict soil moistu re. This index, a broad-scale estimate of available soil moisture, is primarily used to predict wildfire risk, but it is correlated with water content of herbaceous plants (Dimitrakopoulos and Bemmerzouk 2003).

PAGE 55

45 Regression analysis was used to determine the effect on seedling emergence of mean KBDI in the 22 days prior to each census mean daily KBDI trend during the 22 days prior to each census, and total rainfall during the 22 days prior to each census. 22 days was chosen to include the 15 days between cen suses plus 7 days because of the generally good correlation between seed emergence a nd weather variables 7 days prior to emergence (Grundy and Mead 2000). Comparison with seedling em ergence immediately postfire Although fire was not applied as an experime ntal treatment in this study, prescribed fires in mid-July 2002 in two sandhill areas adjacent to the burned site offered the opportunity to compare the composition of s eedlings emerging immediately following a prescribed fire (between the burns on Ju ly 17-18 and data collection on August 1) to seedlings emerging in the study sites during the same time period. All dicot seedlings were recorded in 90 1.35m2 circular plots (121.5 m2 total sampling area) along randomlyplaced transects in each of the two units burned in mid-July. These data cannot be statistically compared to the seedlings found in the experimental plots due to different sampling methods, so only qualita tive comparisons are presented. Statistical Analyses Statistical analyses were conducted usi ng SAS software (SAS Institute, 2005). Analysis of treatment effects was done us ing Proc GLIMMIX, which is capable of analysis of mixed models using non-normal da ta. In this case, count data were modeled using a log-link function. Similarity between seedlings in the fiel d and field vegetation and between field seedlings and seedlings that germinated from soil samples taken from the same sites were calculated using EstimateS version 7.51 (Colwe ll, R. K. 2004. ESTIMATES: Statistical

PAGE 56

46 Estimation of Species Richness and Shared Sp ecies from Samples, Version 7.5. Available at http://viceroy.eeb.uconn.edu/estimates Persistent URL http://purl.oclc.org/estimates. Last accessed May 2007.). Methods for soil sample collection and germination are described in Chapter 2. Jaccard similarity estimators (Chao et al. 2005) correct for underestimation of similarity due to the fact that rare species are likely to be absent from samples. Therefore most similarity indices are biased downward due to the undercounting of shared rare species. Because vegetation da ta consisted of species lists from the field plots (i.e., replicated presen ce-absence data), the incidenc e-based Jaccard similarity estimator was used. Abundance data were available for both field and greenhouse seedlings, so similarity analysis among thes e data sets used the abundance-based Jaccard similarity estimator (Chao et al. 2005). Results Seedling Density Over the course of the study, a total of 344 dicot seedlings emerged in the 160 m2 of study plots. There was a marked difference in the density of seedlings emerging from the burned and unburned sites, with 91.5% fewer seedlings emerging in the unburned study site relative to the burned site (Figure 3-1). Overall seedling emergence in the burned site corresponds to an av erage rate of 3.96 seedlings m-2 year-1. Seedling emergence measured at the unburned site corr esponds to an average of 0.338 seedlings m2 year-1. 79.1% of seedlings survived long enough to be identified to at least family level, and of those, most were identfied to genus or species. Seedlings identified consisted largely of Fabaceae (31.1%), Asterac eae (25.3%), Polygonaceae (7.8%), and Caryophyllaceae (5.2%).

PAGE 57

47 Effects of Soil and Canopy Treatments Experimental manipulation of the canopy and soil had mixed effects on seedling emergence. Due to the small number of seedlings emerging in the unburned unit, statistical analyses of soil and canopy eff ects are limited to the burned site. Hardwood removal in the burned site measurably in creased canopy openness (F igure 3-2), but did not affect seedling emergence (Figure 3-1; F=0.32; df=1,54; p=.57). Seedling emergence at the burned site was affected by soil treatment (F=29.31; df=3,54; p<0.0001), with higher seedling emergence in litter removal a nd soil disturbance plots relative to control and sterilized plots. Seasonal and Weather-Related Patterns Seedling emergence rates varied subs tantially among sample periods, with pronounced peaks occurring in early December, the latter half of January, and July (Figure 3-3). The timing of seedling emer gence of individual species generally corresponded with their flowering and fru iting phenology. For example, seedlings of Asteraceae species that fruit in late fall (particularly Balduina angustifolia and Pityopsis graminifolia ) were most abundant in winter. A se condary peak in Asteraceae seedling emergence in summer was due almost entirely to seedlings of Pterocaulon pycnostachyum which flowers and fruits in sprin g. Seedlings of most members of the Fabaceae, which generally flower and fruit in the spring and summer, were also most frequent in summer. In contrast, seedlings of Dalea pinnata unusual among Fabaceae in that it flowers in late fall, were mo st frequent during winter censuses. Timing of seedling emergence was correlated with only one of the meteorological variables tested. Emergence wa s not significantly correlated with precipitation during the 22 days prior to each censu s period (P=0.249; Figure 3-4) There was no relationship

PAGE 58

48 between seedling emergence and the mean KBDI during 15 days prior to each census (P=0.93). In contrast, there was a negative relationship (P=0.043) between the average daily KBDI trend preceding a census and seedling emergence: emergence was generally lower following drying periods and higher following wetting periods. Seedling Taxa Among the seedlings identified to genus or species, 27 taxa from 13 families were represented (Table 3-2). The most abundant species found in the burned site was Balduina angustifolia (Asteraceae), which accounte d for 46 of 317 (14.5%) of all seedlings found in the burned site. No B. angustifolia seedlings were recorded in the unburned site. The most abundant seedli ng found in the unburned site was Stylisma abdita (Convolvulaceae), which accounted for 4 of 27 seedlings (14.8%) found at that site. Only one seedling of this species was f ound in the burned site. Seedlings of 6 taxa occurred in both sites, 3 taxa were unique to the unburned site, a nd 18 taxa unique to the burned site. No seedlings of non-native speci es were recorded in the study plots. Similarity Between Emerging Seed ings and Aboveground Vegetation For seedlings that could be identified to genus or species, 207 of 245 seedlings (84.5%) emerged from plots where a conspeci fic adult was present. This proportion was higher in the burned site (196 of 227 seedlings 86.3%) than in the unburned site (11 of 18 seedlings, 61.1%). Similarities between seedli ngs appearing in the field during the 1 yr census period and established vegetation, cal culated using the Incidence-Based Jaccard Similarity Estimators, are shown in Table 3-3. All seedlings identified to genus or species were members of taxa found as adults in at least one plot.

PAGE 59

49 Comparison with Seedling Emergence from Soil Samples Dicot seedlings that emerged in the fiel d had low to moderate similarity (Jaccard <0.5) to dicot seedlings that emerged in th e greenhouse from soil samples taken from the plots used in this study (Table 3-4; Ch apter 2). There was no overlap in species composition of seedlings that emerged in th e unburned site and seedlings that emerged from the soil samples, although samples sizes we re small. In the burned site, seedlings in the field were more similar to seedlings that emerged from fresh so ils than to seedlings emerging from stratified soils. In the burned site, 9 taxa, 8 of them characteristic sandhill plants, were found in both fresh soil samples and field germination trials (Table 3-5). This group of shared taxa accounted for 27.8% of dicot seedlings observed in the field and 88.4% of dicot seedlings that emerged from fresh soil samples. Only 4 taxa, 3 of them characteristic of sandhills, were shared between field seedlings and stratified soil seedlings (Table 3-5). These four taxa account ed for 14.5% of dicot seedlings in the field and 62.5% of dicot seedlings that occurred in the soil samples. Despite the dissimilarity of field seedlings to seedlings from stratified soil samples, there are some countervail ing trends. Most importantly, weedy Asteraceae, (e.g., Gnaphalium spp. and Eupatorium spp.), were virtually absent in the field, and were much less common as germinants in stratified than fresh soil samples. Legumes, which were among the most common seedlings in the fiel d, germinated only from stratified soil samples. Comparison with Seedling Emergen ce Immediately Following Fire Seedling density and composition immediately after prescribed fires differed from that recorded in the experimental plots during the same time period (Figure 3-5). In the postfire sites, 137 seedlings were found, corre sponding to a density of 1.13 seedlings m-2.

PAGE 60

50 In contrast, in the experimental plots over the same two week period, only 20 seedlings (0.25 seedlings m-2) emerged. Dalea pinnata was the most abundant seedling immediately postfire, where it accounted fo r 61% of all seedlings and occurred at a density of 0.69 seedlings m-2, as compared to 18% of seedlings and 0.025 seedlings m-2 in the study plots. Seedlings of Rhus copallina and of Tephrosia sp. were common in the postfire sites, each contributi ng 11% of the seedlings and occurring at densities of 0.12 seedlings m-2. No seedlings of these two species were found in the experimental plots during the late July census. Seedlings of Chapmannia floridana which were the most common seedlings in the study plots duri ng that period (30% of seedlings; 0.075 seedlings m-2), were not found immediately postfire. Discussion Relation of Seedling Emergence in the Fi eld to Aboveground Vegetation and Seeds in Soil Samples All seedlings that emerged in the field we re of species found as adults in sandhills at Ordway. Common sandhill groundcover sp ecies emerged from both unburned and burned sites, but no species missing from the vegetation in the unburned site germinated in response to soil or canopy tr eatments. Consistent with ot her studies (Mal aikal et al. 2000, Jenkins 2003) there is little indication th at a soil seed bank in this site would contribute to restoring the ove rall diversity of the area. Seedlings emerging over the one year mon itoring period in sandhill sites were dissimilar to seedlings that emerged in gr eenhouses from soil samples taken from the same sites. This finding is not entirely unexpected given the generally low similarity between soil seeds and vegetation found in ma ny soil seed bank studies. However, one striking result is that se veral weedy species (e.g., Eupatorium compositifolium and E.

PAGE 61

51 capillifolium Gnaphalium spp.) were absent or near ly absent from the suite of species that germinated in the field, even in disturbe d soil, but their seeds were present in large quantities in the soil. This contrast suggests that, at le ast under the weather conditions during the period of this study, conditions fo r germination of these species were not favorable in sandhi lls at Ordway. Climate Effects Although there were pronounced synchronous peaks in seedling emergence, there was no correlation between seedling emergence and rainfall in this study. The lack of clear correlation between r ecent rainfall and seedling emergence in such a xeric community seems counterintuitive, but there ar e reasons to expect a lack of a rainfall effect. Observed seedling emergence at the co mmunity level is a function of the behavior of numerous different species, each of which may respond to climate variables differently (Baskin and Baskin 1998, Grundy and Mead 200 0). In addition, the probability of observing seedlings during a given census is lik ely to be affected by seasonal variation in availability of seeds in the soil, the dorman cy state of the seeds, and the probability of survival between germination and being censused. As each of these varies from species to species as well, any real eff ects of weather on germination of individual species could very easily be masked at the community level. While average KBDI did not correlate w ith seedling emergence, decreasing KBDI values were related to incr eases in germination. KBDI on a given day is dependent on prior drought conditions (Keetch and Byram 1968 ), so average KBDI following a rainfall event during the dry season would be lower th an the average KBDI following an identical rainfall event during the rainy season. Daily change in KBDI should be a better indicator of short-term moisture conditions that are important to seed germ ination. Periods of

PAGE 62

52 increasing soil moisture, whether they occur dur ing dry or wet times in the annual cycle, appear to be favorable for seed germination. Effects of Soil Treatments, Canopy Treatments, and Site The rate of seedling emerge nce in the unburned site was low, despite relatively high soil seed density, and despite soil dist urbance and litter removal treatments which were successful at enhancing seedling emergence in the burned site. Greater sampling intensity could overcome the low density of s eedlings and provide bett er statistical power to test for effects of midsto ry removal and soil treatments. Litter removal and soil disturbance both enhanced seedling emergence in the burned site, a result that could be due to a variety of mechanisms. Litter and burial can both affect the light reaching seeds; removal of litter or moving seeds to the surface by disturbing the soil could remove this inhibito ry effect of low light levels (Pons 2000). Furthermore, leachates from litter can i nhibit germination (Bosy and Reader 1995, Preston and Baldwin 1999). Alternately, bot h litter removal and soil disturbance treatments exposed areas of bare mineral soil that are generally favor able microsites for germination of newly-dispersed seeds. The enhanced germination observed over controls may therefore be a combination of increased germination from the seed bank and from seed rain. It is worthy of note that litter re moval in the burned site stimulated enhanced germination after only 5 years without fire. The low numbers of seedlings emerging from sterilized soil subplots when compared with soil disturbance or litter re moval subplots suggest that few seedlings were attributable to the current seed rain. This finding suggests the existence of a persistent seed bank exists for several sandhill dicots at Ordway, and is in ag reement with the fact that many of the species that germinated in the field were found in at least small numbers

PAGE 63

53 in soil samples. It is also consistent w ith studies in mesic (J enkins 2003) and hydric (Cohen et al. 2004) longleaf pine sites th at report seed banking by some herbaceous species. Despite the lack of seedlings emerging from the sterilized soil in the field, several lines of evidence suggest that many of the s eedlings observed in this study were from newly-dispersed seeds. Seedling emergence in the field was tightly coupled with the reproductive phenology of individual species, as would be observed if seedlings germinated soon after dispersal. Furthermore, there was some direct evidence suggesting that seedlings in the field were from recentlydispersed seeds, such as the pappus bristles still attached to germinated achenes of Pityopsis graminifolia Seedling density in the unburned site was much lower than in the bur ned site, despite a high estimated soil seed density in both; this could be partially due to differences in seed rain. Fire suppression in sandhills reduces herbaceous plant density, wh ich was reflected in lower percent cover and lower species richness of herbaceous plants in the unburned site relative to the burned site. In addition, fire suppression reduces the probabil ity of reproduction in many sandhill species (Clewell 1989, Brewer 1995), and is associated with higher levels of predispersal seed predation (Hiers et al. 2000, Vickery 2002). Thus, the seed rain of herbaceous plants would be expected to be lo wer in fire-suppressed sandhills such as the 35-year site in this study. Fi nally, the likelihood that many s eedlings resulted from seeds dispersed during the study is further supported by the tre nd towards reduced seedling emergence in the soil disturbance plots rela tive to the litter removal plots. Both treatments opened up sandy microsites that presumably favor seed germination, but

PAGE 64

54 reproductively mature plants were destroyed by the soil disturbance treatment, thereby eliminating sources of local seed rain. If local seed rain during the course of the study c ontributed substantially to the seedling emergence observed, then seedlings sh ould have been common in the sterilized control plots. Germination of seeds of Chapmannia floridan a (Fabaceae) experimentally placed in sterilized soil subplots was high, sugge sting that sterilizati on of soil itself did not reduce germination, at least of this species. There is ev idence that placement of the weed barrier cloth under the st erilized soil could have redu ced water infiltration, leading to increased runoff and to seeds being washed away. By the end of the study, the sterilized subplots generally appeared to have less soil than initially, and sterilized soil eroded from the plots was observed. Based on these observations, it is expected that future studies that address these issues w ill show that a larger proportion of seedling germination was due to seed rain than the current results suggest. Fire and Seedling Emergence There is little information in the literature on the germina tion responses of seeds of sandhill plants to fire. Obse rvations that many seedlings emerged within a month after fire, and that these seedlings differed in co mposition from seedlings that emerged in an adjacent healthy sandhill site, suggest that at least some seeds survive fire, and that germination in response to fire is a strategy used by some species in this community. Not surprisingly, seeds of congeners of three of the most common species emerging in the postfire sites (Rhus copallina, Tephrosia sp ., and Opuntia humifusa) have dormancy that can be broken by heat, fire, or smoke (Baski n and Baskin 1998). Further study of postfire seedling responses in this community is warranted.

PAGE 65

55 Implications for Restoration and Management The results of this study suggest a numb er of recommendations for managers who seek to improve the groundcover in fire -suppressed sandhills. Although a few sandhill species may have seeds that pe rsist in soil, treatments su ch as litter removal, soil disturbance, or midstory cleari ng are likely to do little to incr ease the diversity or density of herbaceous plants in fire-suppressed sa ndhill. No germination response to these treatments was observed in fire suppressed sandhills, and all of the species that germinated in the field were still present as adults in the vegetation, even after 35 years of fire suppression. Effective restoration of thes e systems will require planting of seeds or transplants, rather than relying on seed banks or seed rain. Nonethele ss, a wide variety of herbaceous plants do appear capab le of surviving even long periods of fire suppression in xeric sandhill sites, and these will contribute to the diversity of restored s ites if care is taken to protect them during restoration.

PAGE 66

56 Table 3-1. Vegetation structure and fire history of a fre quently burned and an unburned sandhill site in Putnam County, Florida. Burned site Unburned site Longleaf pine density (# of trees >10 cm dbh/ha)a 119.9.60153.8.21 Turkey oak density (# of trees >10 cm dbh/ha) 58.8.5239.8.41 Herbaceous plants (% cover)b 49.3.3313.8.48 Oak litter (% cover) 42.2.8988.5.57 Litter depth (cm)c 2.1.792.7.61 Bare ground (% cover) 12.1.562.7.79 Canopy openness (%) 41.2.4524.8.06 Date of most recent fired 19961966 Number of fires since 1984 50 N=10. All vegetation data are re ported as mean SD. a Tree density was measured by counting trees in 10 pairs of 11 x 13 m plots in each site. b Herbaceous plant cover, litter cover, and bare ground cover were measured as the percent cover along 40 10 cm line transects placed along a randomly located lin e through 10 paired plots at each site. cLitter depth was measured at 20 points along a ra ndomly located line through each of 10 paired plots in each site. Dead leaves that were still connected to living plants were counted as litter. dAs of the time of this study (2001-2002)

PAGE 67

57 Figure 3-1. Number of seedlings emerging ove r one year in two l ongleaf pine sandhill sites in Putnam County, Florida. A) Seedlings emerging from plots in a frequently-burned site (5 years since last fire). B) Seedlings emerging from plots in a long-unburned site (35 years since last fire). N=10.

PAGE 68

58 Figure 3-2. Effect of hardwood removal on percent canopy openness in frequently-burned and unburned longleaf pine sandhill in Putnam County, Florida. The frequently-burned site had last been burned 5 year s prior to the study; the unburned site had not been burned for 35 years. Results are shown as mean S.D.; N=10.

PAGE 69

59 Figure 3-3. Seasonal pattern of emergence of seedlings of different plant families in longleaf pine sandhills. Data are pooled across all soil, canopy, and site treatments (160 x 1 m2 plots).

PAGE 70

60 Figure 3-4. Seedling emergence in two longleaf pine sandhill sites in Putnam County, Florida, in relation to rainfall. Rainfa ll was measured during the 22 days prior to each seedling census. Seedling count is the total of all plots pooled (total area sampled=160 m2).

PAGE 71

61 Table 3-2. Number of seedlings by species emerging over 1 year in two longleaf pine sandhill sites with differe nt fire histories. Species Family Burned sitea Unburned siteb Balduina angustifolia Asteraceae 46 0 Dalea pinnata Fabaceae 38 0 Chapmannia floridana Fabaceae 26 1 Polygonella gracilis Polygonaceae 25 0 Crotalaria rotundifolia Fabaceae 22 1 Pterocaulon pycnostachyum Asteraceae 21 0 Paronychia patula Caryophyllaceae 15 0 Opuntia humifusa Cactaceae 0 4 Pityopsis graminifolia Asteraceae 9 3 Chamaecrista spp. Fabaceae 9 1 Polanisia tenuifolia Brassicaceae 7 0 Hieracium gronovii Asteraceae 6 1 Stylisma abdita Convolvulaceae 1 4 Tephrosia chrysophylla. Fabaceae 4 0 Aureolaria pedicularia var. pectinata Orobanchaceae 4 0 Stipulicida setacea Caryophyllaceae 3 0 Eryngium aromaticum Apiaceae 3 0 Agalinis filifolia Orobanchaceae 3 0 Tragia urens Euphorbiaceae 0 2 Eriogonum tomentosum Polygonaceae 2 0 Houstonia procumbens Rubiaceae 2 0 Croton agyranthemum Euphorbiaceae 1 0 Rhus copallina Anacardiaceae 0 1 Lespedeza hirta Fabaceae 1 0 Stylosanthes biflora Fabaceae 1 0 Asclepias verticillata Apocynaceae 1 0 Gnaphalium sp Asteraceae 1 0 a Site had been burned 5 years prior to the study and had burned 5 times in 17 years preceding the study. b Site had not been burned for 35 years. Total area sampled=80 m2 per site.

PAGE 72

62 Table 3-3. Incidence-based Jaccard similar ity estimator values comparing seedlings emerging over one year and vegetation at two longleaf pine sa ndhill sites with different fire histories. Vegetation, burned sitea Vegetation, unburned siteb Seedlings, burned site 0.4680.349 Seedlings, unburned site 0.2900.283 aThe burned site had been burned 5 times in the 17 years preceding the study, and most recently 5 years prior to the study. bThe unburned site had not been burned for 35 years.

PAGE 73

63Table 3-4. Abundance-based Jaccard similarity estimators and observed and estimated shar ed species between seedlings emerging i n two longleaf pine sandhill sites and seeds from soil samples taken from the same sites. Burned sitea Unburned siteb Soil samples Similarity to field seedlings Shared species (observed) Shared species (estimated)c Similarity to field seedlings Shared species (observed) Shared species (estimated) Burned site (fresh) .443 916000 Burned site (stratified)d .203 49000 Unburned site (fresh) .673 715000 Unburned site (stratified) .017 22000 aThe burned site had been burned 5 years prior to th e study, and 5 times in the 17 years prior to the study.bThe unburned site had not been burned for 35 years.cEstimated shared species were calculated using Chaos shared species estimator (Colwell 2004). dSeed germination from stratified and fresh soil samples was observed in the greenhouse for 6 months. Stra tified soils were stored at 7C for six months before being placed in the greenhouse.

PAGE 74

64 Table 3-5. Species that were found as newly-emerged seedlings in a burned longleaf pine sandhill site that were also found as s eeds in soil samples taken from the same site. Species # of seeds # of seeds # of seedlings (fresh soil)a (stratified soil) Agalinis filifolia 31060 Aureolaria pedicularia var. pectinata 450 Balduina angustifolia 4630 Dalea pinnata 38019 Eryngium aromaticum 310 Gnaphalium sp. 15710 Houstonia procumbens 210 Paronychia patula 1520 Polanisia tenuifolia 710 Stipulicida setacea 3145 Tephrosia chrysophylla 401 aSoil seeds were detected by germination of seedlings for six months in a greenhouse (Chapter 2).

PAGE 75

65 Figure 3-5. Total density of s eedlings emerging within 2 week s of prescribed fire in a longleaf pine sandhill in comparison to seedlings emerging during the same time period in an adjacent sandhill that had not been burned for 5 years.

PAGE 76

66 LIST OF REFERENCES Abrahamson, W. G. 1984. Species responses to fire on the Florida Lake Wales Ridge. American Journal of Botany 71 :35-43 Bakker, J. P., P. Poschlod, R. J. Strykstra, R. M. Bekker, and K. Thompson. 1996. Seed banks and seed dispersal: important topics in restoration ecology. Acta Botanica Neerlandica. 45 :461-490. Baptista, T. L., and S. W. Shumway. 1998. A comparison of the seed banks of sand dunes with different disturbance historie s on Cape Cod National Seashore. Rhodora 100 :298-313 Baskin, C. C., and J. M. Baskin. 1998. Seeds: ecology, biogeogra phy, and evolution of dormancy and germination. Academic Pr ess, San Diego, California, USA. Bosy, J. L., and R. J. Reader. 1995. Mech anisms underlying the suppression of forb seedling emergence by grass (Poa pratensis) litter. Functional Ecology 9 :635-639. Breshears, D. D., P. M. Rich, F. J. Ba rnes, and K. Campbell. 1997. Overstory-imposed heterogeneity in solar radiation and soil moisture in a semiarid woodland. Ecological Applications 7 :1201-1215. Brewer, J. S. 1995. The relationship between soil fertility and fi re-stimulated floral induction in two populations of grass-leaved golden aster, Pityopsis graminifolia. Oikos 74 :45-54. Brockway D. G., K. W. Outcalt, and R. N. Wilkins. 1998. Restoring longleaf pine wiregrass ecosystems: plant cover, di versity, and biomass following low-rate hexazinone application on Florida sandhi lls. Forest Ecology and Management 103 :159-175. Carrington, M. E. 1996. Postfire recruitment in Florida sand pine scrub in comparison with California chaparral. Dissertation, University of Florida, Gainesville, Florida, USA. Carrington, M. E. 1999. Post-fire seedling es tablishment in Florida sand-pine scrub. Journal of Vegetation Science 10 :403-412. Chao, A., R. L. Chazdon, R. K. Colwell, and T.-J. Shen. 2005. A new statistical approach for assessing similarity of species composition with incidence and abundance data. Ecology Letters 8 :148-159.

PAGE 77

67 Chazdon, R. L., R. K. Colwell, J. S. Denslow, and M. R. Guariguata. 1998. Statistical methods for estimating species richness of woody regeneration in primary and secondary rain forests in NE Costa Rica. Pages 285-309 in F. Dallmeier and J. A Comiskey, editors. Forest biodiversit y research, monitoring and modeling: conceptual background and old world case studies. Parthenon Publishing, Paris, France. Christensen, N. L. 1981. Fire regime s in southeastern ecosystems. Pages 112-136 in H. A. Mooney, T. M. Bonnicksen, N. L. Christen esen, J. E. Lotan, and W. A. Reiners, editors. Fire regimes and ecosystem pr operties. USDA Forest Service General Technical Report WO-26., Washington, D. C., USA. Clewell, A. F. 1989. Natu ral history of wiregrass ( Aristida stricta Michx., Gramineae). Natural Areas Journal 9 :223-232. Cohen, D. 1966. Optimizing reproduction in a randomly varying environment. Journal of Theoretical Biology 12 :119-129. Cohen, S., R. Braham, and F. Sanchez, 2004. Seed bank viability in disturbed longleaf pine sites. Restoration Ecology 12 :503-515. Cox, A.C., D. R. Gordon, J. L. Slapcins ki, and G. S. Seamon. 2004. Understory restoration in longleaf pine sandhills. Natural Areas Journal 24 :4-14. Davies, A., and S. Waite. 1998. The persiste nce of calcareous grassland species in the soil seed bank under developing and established scrub. Plant Ecology 136 :27-39. Diaz-Villa, M. D., T. Maranon, J. Arroyo, and B. Garrido. 2003. Soil seed bank and floristic diversity in a forest-grassland mosaic in southern Spain. Journal of Vegetation Science 14 :701-709. Dimitrakopoulos, A. P., and A. M. Bemmerzouk. 2003. Predicting live herbaceous moisture content from a seasonal drought index. Internat ional Journal of Biometeorology 47 :73-79. Drew, M. B., L. K. Kirkman, and A. K. Gholson, Jr. 1998. The vascular flora of Ichauway, Baker County, Georgia: a remn ant longleaf pine/wiregrass ecosystem. Castanea 63 :1-24. Dupuy, J. M., and R. L. Chazdon. 1998. Long-term effects of forest regrowth and selective logging on the seed bank of tropi cal forests in NE Costa Rica. Biotropica 30 :223-237. Earley, L. S. 2004. Looking for longleaf: the fall and rise of an American forest. University of North Carolina Press, Chapel Hill, North Carolina, USA. Facelli, J. M., and S. T. A. Pickett. 1991. Plant litter: its dynamics and effects on plant community structure. Botanical Review 57 :1-32.

PAGE 78

68 Fenner, M. 1985. Seed ecology. Chapman a nd Hall Ltd. New York, New York, USA. Ferrandis P., J. M. Herranz, and J. J. Martinez -Sanchez. 1996. The role of soil seed bank in the early stages of plan t recovery after fire in a Pinus pinaster forest in SE Spain. International Journal of Wildland Fire 6 :31-35. Florida Natural Areas Inventory and Florida Department of Natural Resources. 1990. Guide to the natural communities of Florida. Tallahassee, Florida, USA. Fowler, N. L. 1988. What is a safe site? Ne ighbor, litter, germination date, and patch effects. Ecology 69: 947-961. Frost, C.C. 1993. Four centuries of changi ng landscape patterns in the longleaf pine ecosystem. Pages 17-43 in Proceedings of the 18th annual Tall Timbers fire ecology conference. Tall Timbers Research Station, Tallahassee, FL. Gashaw, M., A. Michelsen, M. Jensen, and I. Friis. 2002. Soil seed bank dynamics of fire-prone wooded grassland, woodland, and dry forest ecosystems in Ethiopia. Nordic Journal of Botany 22 :5-17. Gilliam, F. S., B. M. Yurish, and L. M. Goodwin. 1993. Community composition of an old growth longleaf pine forest: relationshi p to soil texture. Bulletin of the Torrey Botanical Club 120: 287-294. Gordon, D. R. and K. J. Rice. 1998. Pa tterns of differentiation in wiregrass (Aristida beyrichiana) : Implications for restorati on efforts. Restoration Ecology 6 :166-174. Gross, K. L. 1990. A comparison of met hods for estimating seed numbers in soil. Journal of Ecology 78 :1079-1093. Grundy, A. C. and A. Mead. 2000. Modeling weed emergence as a function of meteorological records. Weed Science 48 :594-603. Guzman-Grajales, S. M, and L. R. Walker. 1991. Differential seedlin g responses to litter after Hurricane Hugo in the Luquillo experi mental forest, Puerto Rico. Biotropica 23 :407-413 Harper, J. L. 1977. Population biology of plants. Academic Press, London, UK. Heyward, F. 1939. The relation of fire to stand composition of longleaf pine forests. Ecology 20 :287-304. Holl, K. D., H. N. Steele, M. H. Fusari, and L. R. Fox. 2000. Seed banks of maritime chaparral and abandoned roads: potential for vegetation recovery. Journal of the Torrey Botanical Society 127 :207-220. Hurlbert, S. H. 1984. Pseudoreplication and the design of ecological field experiments. Ecological Monographs 54 :187-211.

PAGE 79

69 Hyatt, L. A. 1999. Difference between seed bank composition and field recruitment in a temperate zone deciduous forest. American Midland Naturalist 142 :32-38. Isselstein, J., J. R. B. Tallowin, and R. E. N. Smith. 2002. Factors affecting seed germination and seedling establishmen t of fen-meadow species. Restoration Ecology 10 :173-184. Jefferson, R. G., and M. B. Usher. 1987. The seed bank in soils of disused chalk quarries in the Yorkshire wolds, Engl and: implications for conservation management. Biological Conservation 42 :287-302. Jenkins, A. M. 2003. Seed banking and ve sicular-arbuscular mycorrhizae in pasture restoration in Central Florida. Thesis, Univ ersity of Florida, Gainesville, Florida, USA. Jensen, K., and K. Gutekunsk. 2003. Effects of litter on establishment of grassland plant species: the role of seed size and successional status Basic and Applied Ecology 4 :579-587. Keddy, P.A., and A. A. Reznicek. 1982. The ro le of seed banks in the persistence of Ontarios coastal plain flora. American Journal of Botany 69 :13-22. Keeley, J. E. 1991. Seed germination a nd life-history syndrome s in the California chaparral. Botanical Review 57 :81-116. Keeley, J. E., and C. J. Fotheringham. 2000. Role of fire in regeneration from seed. Pages 311-330 In M. Fenner, editor. Seeds: the ecology of regeneration in plant communities. CABI publishing, Wallingford, Oxfordshire, UK. Keetch, J. J., and G. M. Byram. 1968. A drought index for forest fire control. United States Department of Agriculture Fo rest Service, Research Paper SE-38. Washington, D. C., USA. Kitajima, K., and D. Tilman. 1996. Seed banks and seedling establishment on an experimental productiv ity gradient. Oikos 76 :381-391. Leishman, M. R., I. J. Wright, A. T. Mo les, and M. Westoby. 2000. The evolutionary ecology of seed size. Pages 31-58 In M. Fenner, M., editor. Seeds: the ecology of regeneration in plant communities. CA BI publishing, Wallingford, Oxfordshire, UK. Lewis, C. E. and T. J. Harshberger. 1976. Shrub and herbaceous vegetation after 20 years of prescribed burning in the South Carolina coasta l plain. Journal of Widlife Management 29 :13-18. Livingston, R. B., and M. L. Allesio. 1968. Bu ried viable seeds in successional field and forest seres, Harvard Forest, Mass. Bulletin of the Torrey Botanical Club 95 :58-69.

PAGE 80

70 Major, J. and W. T. Pyott. 1966. Buried vi able seeds in Californ ia bunchgrass sites and their bearing on the definiti on of a flora. Vegetatio 13 :253-282. Malaikal S. K., E. S. Menges, and J. S. Denslow. 2000. Community composition and regeneration of Lake Wales Ridge wiregr ass flatwoods in relation to time-sincefire. Journal of the To rrey Botanical Society 127 :125-138. Maret, M. P., and M. V. Wilson. 2005. Fire and litter effects on seedling establishment in Western Oregon upland pr airies. Restoration Ecology 13 :562-568. Matlack, G. R., D. J. Gibson, and R. E. Good. 1993. Regeneration of the shrub Gaylussacia baccata and associated species after low intensity fire in an Atlantic coastal plain forest. American Journal of Botany 80 :119-126. McCall, R. K., and D. J. Gibson. 1999. Th e regeneration potential of a threatened southern Illinois shale barren. Journa l of the Torrey Botanical Society 126 :226-233. Myers, R. L. 1985. Fire and the dynamic relationship between Florida sandhill and sand pine scrub vegetation. Bulletin of the Torrey Botanical Club 112 :241-252. Myers, R. L. 1990. Scrub and high pine. Pages 150-193 In Myers, R.L. and J. J. Ewel, eds. Ecosystems of Florida. University of Press of Florida, Gainesville, Florida, USA. Molofsky, J., and C. K. Augspurger. 1992. Th e effect of leaf lit ter on early seedling establishment in a tropical forest. Ecology 73 :68-77. Mulligan, M. K., and L. K. Kirkman. 2002. Burning influences on wiregrass (Aristida beyrichiana) restoration plantings: Natural s eedling recruitment and survival. Restoration Ecology 10 :334-339. Neeman, G., and I. Izhaki. 1999. The effect of stand age and microhabitat on soil seed banks in Mediterranean Aleppo pine forests after fire. Plant Ecology 144 :115-125. Noss, R. F. 1989. Longleaf pine and wiregr ass: keystone components of an endangered ecosystem. Natural Areas Journal 9 :211-213. Noss, R. F., E. T. LaRoe, and J. M. Sco tt. 1995. Endangered ecosystems of the United States: a preliminary assessment of loss a nd degradation. United States Department of the Interior National Bi ological Service Biological Report 28. Washington, D.C., USA. Outcalt, K. W. 2000. Occurrence of fire in lo ngleaf pine stands in the southeast United States. Pages 178-182 in Proceedings of the 21st annual Tall Timbers fire ecology conference. Tall Timbers Research St ation, Tallahassee, Florida, USA Platt, W. J., G. W. Evans, M. M. Davis. 1998. Effects of fire season on flowering of forbs and shrubs in longleaf pine forests. Oecologia 76 :353-363.

PAGE 81

71 Poiani, K. A., and P. M. Dixon, 1995. Seed banks of Carolina bays: Potential contributions from surrounding landscape ve getation. American Midland Naturalist 134 :140-154. Pons, T. J. 2000. Seed responses to light. Pages 237-260. In M. Fenner, editor. Seeds: the ecology of regeneration in plant co mmunities. CABI Publishing, Wallingford, Oxfordshire, UK. Preston, C. A., and I. T. Baldwin. 1999. Positive and negative signals regulate germination in the post-fire annual Nicotiana attenuata Ecology 80 :481-494. Priestly, D. A. 1986. Seed aging: implicati ons for seed storage a nd persistence in the soil. Comstock Publishing Associates, Ithaca, New York, USA. Probert, R.J. 2000. The role of temperatur e in the regulation of seed dormancy and germination. Pages 261-292 In M. Fenner, editor. Seeds: the ecology of regeneration in plant communities. CA BI Publishing, Wallingford, Oxfordshire, UK. Provencher, L., B. J. Herring, D. R. Gordon, H. L. Rodgers, K. E. M. Galley, G. W. Tanner, J. L. Hardesty, and L. A. Br ennan. 2001. Effects of hardwood reduction techniques on longleaf pine sandhill vegeta tion in Northwest Florida. Restoration Ecology 9 :1-15. Rabinowitz, D. 1980. Buried viable seeds in a North American tallgrass prairie: The resemblance of their abundance and co mposition to dispersing seeds. Oikos 36 :191-195. Rabinowitz, D. and J. K. Rapp. 1980. Seed ra in in a North American tallgrass prairie. Journal of Applied Ecology 17 :793-802. Read, T. R., S. M. Bellairs, D. R. Mulligan, and D. Lamb. 2000. Smoke and heat effects on soil seed bank germination for re-establis hment of a native forest community in New South Wales. Austral Ecology 25 :48-57. Readle, E. L. 1990. Soil survey of Putnam County area, Florida. US Department of Agriculture, Soil Conservation Se rvice, Washington, D.C., USA. Rice, K. J. 1985. Responses of Erodium to varying microsites: the role of germination cueing. Ecology 66 :1651-1657. Rice, K. 1989. Impacts of seed banks on grassland community structure and population dynamics. Pages 211-229 in Leck, M. E., V. T. Parker, and R. L. Simpson, editors, Ecology of soil seed banks. Academic Press, San Diego, California, USA. Roy, J. and L. Sonie. 1992. Germination and population dynamics of Cistus spp. in relation to fire. Journal of Applied Ecology 29 :647-655.

PAGE 82

72 Rotundo, J. L., and M. R. Aguiar. 2005. L itter effects on plant regeneration in arid lands: A complex balance between seed re tention, seed longevity and soil-seed contact. Journal of Ecology 93 :829-838. Seamon, G. 1998. A longleaf pine sandhill restoration in north -west Florida. Restoration and Management Notes 16 :46-50. Thompson, K. 2000. The Functional Ecology of Soil Seed Banks. Pages 215-236 in M. Fenner, editor. Seeds: the ecology of re generation in plant communities. CABI publishing, Wallingford, Oxfordshire, UK. Thompson, K., and P. Grime. 1979. Seasonal va riation in the seed banks of herbaceous species in 10 contrasting habitats. Journal of Ecology 76 :393-421. Trabaud, L. 1994. Diversity of the soil seed bank of a Mediterranean Quercus-Ilex forest Biological Conservation 69 :107-114 Underwood, A. J. 2000. Experiments in ecology: Their logical design and interpretation using analysis of variance. Cambridg e University Press, Cambridge, UK. Varner, J. M., D. R. Gordon, F. E. Putz, and J. K. Hiers. 2005. Restoring fire to longunburned Pinus palustris ecosystems: Novel fire effects and consequences. Restoration Ecology 13 :356-544. Vickery, P. D. 2002. Effects of the size of pres cribed fire on insect predation of northern blazing star, a rare grassland pe rennial. Conservation Biology 16 :413-421. Walker, J., and R. J. Peet. 1983. Compositi on and species diversity of pine-wiregrass savannas of the Green Swamp, North Carolina. Vegetatio 55 :163-179. Wetzel, P. R., A. G. Van Der Valk, and L. A. Toth. 2001. Restoration of wetland vegetation on the Kissimmee River floodplai n: Potential role of seed banks. Wetlands 21 :189-198. Wienk, C. L., C. H. Sieg, and G. R. McPh erson. 2004. Evaluating the role of cutting treatments, fire, and soil seed banks in an experimental framework in Ponderosa pine forests of the Black Hills, South Dakota. Forest Ecology and Management 192 :375-393. Whelan, R. J. 1985. Patterns of recruitment to plant populations after fire in Western Australia and Florida. Proceedings of the Ecological Society of Australia 14 :169178. Whittle, C. A., L. C. Duchesne, and T. Needham. 1997. Comparison of emergence methods to evaluate viable plant propagules in forest soils following fire. Canadian Field Naturalist 111 :436-439.

PAGE 83

73 Williams, P. R., R. A. Congdon, A. C. Grice, and P. J. Clarke. 2005. Germinable soil seed banks in a tropical savanna: Seasona l dynamics and effects of fire. Austral Ecology 30 :79-90. Williams-Linera, G., and J. J. Ewel. 1984. Effe ct of autoclave steril ization of a tropical andept on seed germination and seedling growth. Plant and Soil 82 :263-268. Wunderlin, R. P. 1998. Guide to the vascular plants of Florida. University Press of Florida, Gainesville, Florida, USA. Zammit, C., and P. H. Zedler. 1994. Orga nization of the soil seed bank in mixed chaparral. Vegetatio 111 :1-16. Zhang, Z. Q., W. S. Shu, C. Y. Lan, and M. H. Wong. 2001. Soil seed bank as an input of seed source in revegetation of lead /zinc mine tailings. Restoration Ecology 9 :378-385.

PAGE 84

74 BIOGRAPHICAL SKETCH Geoffrey Parks was born in Rhode Isla nd in 1971. After graduating from high school in Maine, he attended Wesleyan Univer sity in Middletown, CT, where he earned a bachelors degree in biology. Following graduati on from college, he worked as a field biologist in Arizona, Califo rnia, and South Carolina. From 1997 to 1999 he did postbaccalaureate study at the University of Mi ssouri in the Department of Biological Sciences Avian Ecology Lab. Geoff married Ca rol Church in 2002 and is the father of Eleanor Parks-Church, born in 2004. Since 2002 he has worked for the City of Gainesville, Floridas Nature Operations Di vision, where his responsibilities include wildlife and vegetation monitoring, manageme nt planning, ecological restoration, and invasive plant control. He plans to conti nue managing, conserving, and restoring natural communities.