RESEARCH ARTICLE ABSTRACT: Southeastern habitats dominated by (Pinus palustris Miller) and associated species have declined dangerously. Conservation of rare and common of longleaf pine habitats may be aided by starting new popula- tions in the field. We review methods for initiating plant populations and integrate information from our studies of rare and common longleaf pine ground-layer plants of the outer Coastal Plain. In our experience it is possible to start new populations of most longleaf pine ground-layer plants, including rare species if (1) seeds Starting New are collected from frequently burned sites with reasonably large populations of desired species; (2) appropriate media are used for seedling propagation; (3) outplanting of nursery grown seedlings or direct seeding is done during periods of sufficient soil Populations of moisture; and (4) introduction sites properly match habitat requirements (inferred from indicator plants) of desired species, and the sites can be managed with frequent prescribed Longleaf Pine fire. 3-ound-layer Plants Index terms: Longleaf pine, Pinus palustris, rare plants, reintroduction, southeastern United States in the Outer

Coastal Plain of ranges (see Morse 1996, Gaston 1997 and references therein). These species typical- INTRODUCTION ;outh Carolina, USA ly appear on federal or state rare lists, and receive low G-ranks from The Nature Longleaf pine (Pinus palustris Miller) sa- Conservancy. Hardin and White (1989) vannas and woodlands are characterized and Walker (1993) compiled and published by a sparse tree canopy and a very rich Jeff S. Glitzenstein lists of longleaf pine associated species ground layer usually dominated by herbs that meet these traditional criteria for rar- Donna R. Streng (Bridges and Orzell 1989, Peet and Allard ity. Walker’s (1993) list included 389 spe- 1993). Decline of these biologically rich Tall Timbers Research Station cies, 187 of which were considered rare habitats, which once encompassed vast 13093 Henry Beadel Drive range-wide and vulnerable to extinction areas of the southeastern United States, is Tallahassee, FL 32312 USA (G-ranks = l-3). Most are herbaceous pe- well documented (Frost 1993, Harcombe rennials that depend on frequent natural et al. 1993). Only about 3% of an estimat- Dale D. Wade fire. More than half (65%) occur in moist ed 37 million ha remains. Contributing USDA Forest Service longleaf pine habitats (i.e., seasonally factors include logging, clearing, inten- Southern Research Station flooded flatwoods and savannas, depres- sive forestry, and, perhaps most important, 320 Green Street sions, shrub bog ecotones, seepage slopes, fire exclusion. Although precise data are Athens, GA 30602 USA pond margins), although many (45 spe- not available for many species, most her- cies, 24.2% of the total) occur in typical baceous plants in longleaf pine dominated John Brubaker mesic-dry upland habitats (Walker 1993). habitats apparently require frequent (i.e., An update of Walker’s (1993) list, avail- Medical University of South Carolina mean fire return intervals less than 5 years), able online at http://www.talltimbers.org/ Department of Pharmacology low-intensity fires for population mainte- research.html, includes 256 additional spe- Charleston, SC 29403 USA nance (Lemon 1949, Platt et al. 1991, cies. This large increase reflects, in part, Waldrop et al. 1992). (Plant nomenclature the endangerment of ever-larger numbers follows Kartesz [1994] unless otherwise of ground-layer plants of longleaf pine l indicated.) habitats. Given the continuing loss and endanger- The second group of at-risk species includes ment of high quality longleaf pine habitats formerly dominant, but still often locally (e.g., Percher et al. 1999), many of their abundant, “climax” grasses of North Amer- characteristic plants could, or should, be ican prairies and savannas. These grasses considered rare or endangered. Two groups tend to be shallow-rooted and are highly l Correspollding authorauthor e-mail: of species may be at greatest risk. The first [email protected] susceptible to even light to moderate distur- group includes species that fit the tradi- bance of upper soil horizons such as me- tional definition of rarity-species with chanical site preparation for artificial regen- Natural Areas Journal 21:89-110 small population sizes, few populations, eration of pine stands (e.g., Schultz 1976). or very limited geographic or ecological

Volume 21 (I), 2001 Natural Areas journal 89 Wind-pollinated and limited in dispersal, has been going on for centuries, and his- mendations on seed sampling for genetic these plants may also be particularly vulner- torical evidence suggests that many cur- diversity, see Center for Plant Conserva- able to habitat fragmentation and declining rently rare species were more common tion (1991). population size (Tilman et al. 1997). prior to recent antbropogenic declines (El- liott 18161824, Percher 1848, Ravenel Fire history is an important factor in site Ecologists generally agree that conserva- 1876, Weakley 1999); and (3) if current quality for ground-layer plants of longleaf tion is a more effective tool for protecting trends continue, today’s common species pine habitats. Frequently burned sites have and maintaining biodiversity than restora- may be rare in the near future. In the first reduced competition from hardwood tion (Falk et al. 1996). Fortunately, some section of this paper we discuss seed is- shrubs and sprouts (Waldrop et al. 1992), large tracts of properly managed, high sues, including choice of collection sites resulting in more vigorous ground-layer - quality longleaf pine habitat remain on and dates, seed processing and storage, plants with larger and great- private and public lands. Nevertheless, to and germination tests. The second section er seed production. On sites with a history counteract continuing loss of biodiversity, covers seedling issues and includes con- of frequent fire, canopy density can have _ it may be necessary to start new popula- siderations of seedling propagation, plant- an influence on seed viability. Means tions of the most threatened species (Falk ing sites, and seedling performance after (1997) found a negative correlation be- et al. 1996, Pavlik 1996). There are three outplanting. This section also contains re- tween density of overstory slash pine (Pi- basic strategies for starting plant popula- sults from a small translocation experi- nus elliottii) and seed viability of Aristida tions: translocation (i.e., moving individu- ment we conducted with Agrimonia in- beyrichiana Trinius and Ruprecht. Fire als or parts of individuals directly from cisa, a globally rare forb of subxeric history also influences seed predation, field populations), outplanting of nursery longleaf pine habitats. The third section which can have a major impact on seed grown seedlings or cuttings, and direct discusses direct seeding methods, and com- availability. Hiers et al. (2000) found that seeding (Guerrant 1996). Each strategy pares results from two of our own studies approximately 68% of Tephrosia virgini- involves several procedural issues. For of direct seeding that varied with respect ana pods in unburned control plots con- example, to successfully grow and out- to competition control. Finally, we briefly tained seed predators compared to only plant seedlings one must collect and pro- discuss native plant gardens and the po- 23% in burned plots. cess the seed, germinate it, select a growth tential role of gardens as refuges and source medium, choose a planting site, and man- populations. For each topic, we first pro- Seed production of most longleaf pine age the new population. Direct seeding vide a review of pertinent literature, and ground-layer species tends to decline with also involves seed collection and germina- when appropriate, present our own unpub- time since burning. Seeds are most abun- tion and requires information on the envi- lished data. We present data suggesting dant during the year of the bum, or, if the ronmental and biotic conditions needed to that it is not difficult to start new popula- bum occurs late in the year, the following maximize germination and establishment tions of longleaf pine ground-layer plants, growing season. However, some species, under field conditions. Translocation may including rare species, if the introduction such as ambiguus , Aristida be the easiest initiation technique, although site is located in an appropriate habitat and purpurascens, and Muhlenbergia capil- it is often hard to justify ethically, and can be managed with frequent fire. laris, and produce abundant seed choice of planting site and transplant con- for several years following fire (Pfaff and ditions are critical concerns (Hall 1986). Gonter 1996). It is not known whether I: SEED ISSUES seed quality in these species declines with This paper reviews methods for starting time since fire. Seed Collection Sites and Dates populations of ground-layer plants of longleafpine savannas and woodlands, and Finding a quality seed collection site is Season of burning can affect seed produc- includes some of our own unpublished usually the first step in starting a new plant tion. Many pine savanna grasses, and some data. We emphasize species additions to population. Unfortunately, there is little composites, flower more prolifically after relatively intact communities. In contrast, information on collection site quality for growing-season bums (Streng et al. 1993). Pfaff and Gonter (1996) discussed com- longleaf pine ground-layer species. For Efficiency of wind and insect pollination munity restoration on badly degraded or rare species generally, seeds should be tends to increase with larger numbers of reclaimed land. We discuss results from collected from large populations to ade- flowering stems (Kunin 1997), and it might - our studies of both rare and common quately capture the range of genetic vari- be expected that growing-season bums longleaf pine ground-layer plants because ability present in the population and to would lead to increased seed production. (1) many rare plant species have not been minimize inbreeding depression (Center However, Hiers et al. (2000) did not find a well studied, and studies of related com- for Plant Conservation 1991, Zettler and consistent, positive effect of growing-sea- mon species might provide important hints McInnis 1992). However some seeds also son burns on sexual reproduction (flower about common biological characteristics should be collected from small popula- and fruit production) of legume species and propagation strategies (e.g., Buchele tions because interpopulation genetic vari- common in dry longleaf pine woodlands. et al. 1991); (2) the process of endanger- ation can be high (Center for Plant Con- Rather, species responded individualisti- ment of longleaf pine ground-layer plants servation 1991). For further recom- tally to burn season. Further, there ap-

90 Natural Areas Journal Volume 21 (l), 2001 peared to be little overall effect of burn Table 1. Results of germination tests 19961999. Letter code below species name is a collection site season on seed production in the three label. Collection dates are next to the site labels. Burn codes are as follows: D= dormant season focal species. burn, G = growing season burn, G -1 = growing season burn in the previous year; except for G - 1 all burns occurred less than 1 year prior to the date of seed collection. “Germ tray” indicates We collected and tested (see “Germination germination tests carried out in closed plastic germination boxes with seeds on moist blotting paper. Tests” for germination methods) seed of Tests were performed on a bench placed next to a window at room temperature with no attempt to precisely control light or temperature. “Grow tray” is for outside germination tests carried out in 42 ground-layer species from 15 longleaf standard horticultural trays Blled with moist sand. For these latter tests seed was placed on top of pine sites (Table 1) in or near the Francis cells within 1 month after it was collected from the Beld sites. Thus these seeds were exposed to a Marion National Forest (FMNF), north- regime of natural light and temperature fluctuations. Numbers under column headings are germi- east of Charleston, South Carolina. Except nation percentages. If different from n=lOO, sample size is listed in parentheses after the gennina- for two roadside sites that were maintained tion result. For tomentosus stage of seed maturity is indicated after each line as follows: NE = pappus not expanded, E = pappus expanded, but heads not entirely opened, G = by mowing, all sites had a recent history of pappus expanded, heads opened, seeds arranged in a 360 degree pattern around the heads, thus tire and a relatively open longleaf pine appearing globose e preceding name indicates that it is locally rare). Globally rare species (Walker canopy. Although most collections had 1993), caroliniana, Plantago sparsifiora, aristosa, Tridens ambiguus and Sporobo- germination rates > 30% (Table 1, Figure luspinetorum Weakley & P.M. Peterson are preceded by **. Results for P. caroliniuna are given for l), there was considerable variability in covered (cov.) and uncovered (uric.) seeds (see also Figure 5). percent germination, even within the same species. Collection Burn Germ Grow Date Season Tray Tray The literature suggests that information on bum season and time-since-bum can help A. APIACEAE predict quality seed collection sites. To Oxypolis jiliformis test the effect of bum season, we selected (AS) 10/05/99 G 75.3(150) 11 species for which we had collected seed from both dormant-season bum sites B. and growing-season bum sites. Data were Amica acaulis compared using a paired t-test. Consistent w-) 05128196 G 90.0 64.3( 115) with Hiers et al. (2000), mean seed germi- (SO) 05/I 2198 G 48.7( 150) nation for the two treatments were similar (SCTS) 05124199 D 57.3( 150) (growing season = 36.7% germination; Aster concolor 0 1124197 G 10.3(107) 1.4(73) dormant season = 30.4%) and there was (MS) Aster dumosus no significant effect of bum season (t = W-3 Oll21l97 D 14.0 12.5(96) df = 10, P = 0.37). To test effects of (MS) Oll24l97 G 50.0(128) 47.4(97) time-since-bum, we collected seed of Eu- Aster linariifolius L. patorium rotundifolium and E. leucolepis WC) Oll21l97 D 5.0(80) 0.0(26) from a site that had been burned approxi- (MS) OU24l97 G 46.8(79) 48.0(25) mately 2 years previous and from sites Aster tortifolius burned earlier within the same year. Ger- VW Oll21l97 D 20.7(29) 33.3(10) mination rates were similar regardless of Aster walteri the time elapsed since burning. It appears 0-W Oll21l97 ‘D 78.7(61) 66.7(15) that for these species, quantities of viable WV 01l24l97 G 63.7(104) 74.7 seed may be obtained from sites not burned Bigelowia nua!ata D 52.0 51.0 for at least 2 years, assuming that compe- WC) 11127196 01/21/97 D 29.8(57) tition from shrubs is not excessive. W.3 WC) Oll21l97 D 0.0(60) (ABORTED SEED?) Carphephorus paniculatus Our results suggest that quality of seed col- U-W 1O/26/96 D 1.8(57) 5.0(20) lection sites for longleaf ground-layer spe- WC) 11127196 D 35.0 30.4(92) cies cannot be entirely predicted by time- 0-W Oll21l97 D 42.2(90) 42.0(88) since-bum or bum-season. To maximize (MS) 01124197 G 21.8(101) 19.0 genetic diversity, it might be advisable to l Carphephorus tomentosus collect seed following a variety of different PL) 1O/28/96 G 29.6( 142) 45.0(80) NE bum treatments or periods of time without w-1 10128196 G 44.4(36)E burning. m 11127196 G 17.0(53) 41.6(36) NE m 11127196 G 21.0 30.9(81) E PL) 11/27/96 G 35.3(68) 72.2(54) G continued

Volume 21 (l), 2001 Natural Areas Journal 91 Seed Collecting Technique Table 1, continued To maintain appropriate levels of genetic diversity in new plant populations, it is Collection Blll-ll Germ Grow important that collection methods ensure Date Season Tray Tray genetic diversity (Center for Plant Conser- vation 1991, Guerrant 1996). Random or Chaptalia tomentosa systematic sampling can avoid biasing the N-U 04114198 - 88.7(150) sample toward large, vigorous, or easily (SW 04/16/99 - 64.0(150) Chrysopsis gossypina accessible plants (Huenneke 1991). At a (C169) 01/19/99 G 32.7(150) minimum, effort should be made to collect Chlysopsis mariana seeds from all areas of a donor popula- (MS) 01124197 G 28.3(99) 40.0(90) tion’s habitat and from all plant morpho- Coreopsis oniscicarpa Fem. logical types. (MS) 01124197 G 30.1(83) 28.9(45) vernus A goal of seed collecting is to obtain ma- (C195) 06lO7l98 G lO.O(150) ture or hardened seed rather than imma- (SCTS) 05124199 D 1.2(80) ture “soft” seed. A time-honored method leucolepis of assessing maturity is to bite the seed. (MS) Oll24l97 G 43.0 27.1(96) Generally, ripe seed cannot be bitten in (SO) 01/14/99 G-l 32.0(150) two. Apfelbaum et al. (1997) list four ad- ditional cues to guide seed collection of WC) Oll21l97 D 24.0(121) 26.3(99) Oll24l97 G 15.0 tallgrass prairie species: (1) seeds are full (MS) 7.4(94) (SO) lOl14l99 G-l 46.7(150) . size, (2) seeds have changed color (i.e., Liatris squarrosa from green to a darker hue), (3) seed- (C195) 12JOll98 G 62.0(150) bearing stalks have dried, and (4) seed has Pityopsis graminifolia begun to drop. We disagree somewhat with U-0 Oll21l97 D 2.6(115) 2.0(99) cue 3. In the southeastern United States, Pterocaulon pycnostachyum (Michx.) Ell. plant stem and leaf tissue often remains w-w 06123196 D 31.0 21.7(115) green into the winter, thus plants dispers- mw 06129196 D 42.0(115) 20.0 ing seed may have green stems. Also 0-W 07lO2l99 D 44.7(150) longleaf pine savannas contain many Solidago jistulosa spring-flowering plants that stay green in (MS) Oll23l97 G 49.1(57) the summer after flowering. Solidago odora 0-W Oll21l97 D 0.0(29) 0.0(20) Vemonia angusttfolia Another good cue for detecting ripe seeds (C195) lOl24l98 G 35.3(150) in many plant families (e.g., Melastomata- ceae, Onagraceae, Saxifragaceae) is the C. FABACEAE presence of dehiscent capsules. Many cap- Desmodium tenuifolium sules act as “salt and pepper shakers” that WC) Oll21l97 D 59.6(57) 53.8(26) continue to disperse viable seeds for many Lespedeza capitata months after opening. Another cue is the WC) lOlO6l96 D 11.4(35) ease with which the seeds can be removed (MS) 01123197 G 7.3(41) from the plant: ripe seeds can typically be Rhynchosia renifomtis dislodged with minimal effort; unripe seeds mw 06l15l96 D 94.9(59) 42.4(59) are much harder to remove. Young and mw 06123196 D 32.8(61) 13.3(60) Young (1986) suggested a more quantita- (MC) 06129196 D 0.0(59) 0.0(60) (C184) 06122198 D 48.7( 150) tive indicator: monitoring seed moisture Tephrosia virginiana content. Moisture content is high in imma- (C195) 08lO3l98 G 70.0(150) ture seeds, but declines to approximately 10% as seeds mature. Philips ( 1985), Cline- D. HYPERICACEAE bell (1997), and Zettler (1997) provided Hypericum setosum numerous suggestions for collecting seeds 0-U Oll21l97 D 13.4(120) 8.3(108) of more specialized groups of plants such WC) Oll21l97 D 46.7(120) as prairie genera and Orchidaceae (a noto- NOTE: top line is after 4 months, bottom line is germination after 1.5 years. riously challenging family). continued

92 Natural Areas Journal Volume 21 (l), 2001 One method to determine the best time to Table 1, continued collect seeds of a particular species is to resample the same population over time as Collection Burn Germ Grow seeds mature (e.g., Seamon et al. 1989). Date Season Tray Tray We employed this technique for represen- tative species of four major plant families of longleaf pine ground-layer plants as E. ** Rhexia aristosu . part of the germination study mentioned above. To avoid bias, we established 0-W 1O/06/96 D 3.6(55) 0.0(S) transects at two sites (PL and HC in Table 0-W 10128196 D 1.1(90) 6.2(81) Rhexia nashii 1) and selected different random locations . 1O/06/96 D 56.7(60) 63.3(60) for each collection date. WI G-W 10128196 D 2.5(80) 60.0(80) Germination tests of these various collec- U-W 10/06/96 D 12.0 2.0 tions (see “Germination Tests” for methods) WC) 10126196 D 9.0(95) 6.3(95) indicated substantial changes in seed germi- 0-U 1 l/27/96 D 22.1(68) 20.0(115) nation rates among collection dates for most W-3 01/21/97 D 13.4(98) 2.4(85) species (Figure 2). Furthermore, the chang- (C195) 09113198 G 4.7(150) es were systematic, rather than entirely ran- dom, so that it was possible to determine an F. ONAGRACEAE optimal collection period. We recommend Ludwigia virgata collecting Rhynchosia reniformis seeds in WC) 01/21/97 D 6.1(115) 11.0 mid-June, Rhexia alifanus in late November G. PAFWASSIACEAE (despite the fact that this species in ** mid-summer and capsules appear to open W-U 12/19/94 D 34.0 (Cov) by early September), and Carphephoruspan- W-N 12119194 D 20.0 (Uric) iculatus seeds in mid-December. aromaticum, a dominant grass of wet savan- H. PLANTAGINACEAE nas in South Carolina, did not display a ** Plantago sparsijlora discrete germination maximum across the W-0 6115198 Mowed 17.6(228) range of collection dates. For this species it appears that viable seed can be collected I. over a broad period from early October Aristia’a virgata Trin. through early February. 0-W 01/21/97 D 48.4(93) 43.1(65) (MS) Oll23i97 G 58.3(103) 66.2(74) Prior results such as those presented above U-W 1O/06/96 D 27.0 25.8(89) can provide guidance as to when to collect WC) 10/28/96 D 30.0 23.1(91) seeds of particular species. Because matu- U-V 1l/27/98 D 21.0 21.0 ration dates can vary among years for a WC) 01/23/97 D 19.4(98) 29.8(94) variety of reasons, it is also important to (MS) 01/24/97 G 36.4(110) 20.9(110) evaluate seed maturity cues when decid- (C195) 12lO2l98 G 3.0( 150) ing when to collect seed. For example, Muhlenbergia expansa (Poir) Trin. seeds of Calphephorus tomentosus (a fall- W-J 10129196 G 12.0 13.0 flowering composite at the southern edge Panicum virgatum of its range in the PMNP and consequent- (MS) 01123197 G 9.9(91) 1.3(80) ly locally rare) collected from site PL on Schizachyrium scoparium two dates in the autumn of 1996 (Table 1) WJ 10124196 G 5.0 11.0 (C196) 1l/23/98 G 55.3(150) showed best germination from seed col- ** Sporobolus pinetorum Weakley & P.M. Peterson lected in late November and classified as (PL) 10129196 G 51.0 23.3(30) . “globose” (i.e., pappus fully expanded and ** Tridens ambiguus seed arranged in a 360” pattern around the (PL) 10128196 G 88.0 68.0 peduncle attachment point; Figure 3). ‘Qp- ically, globose Asteraceae inflorescences J. POLYGALACEAE contain seeds that have good germination Polygala lutea rates unless the seed is damaged in some WV 07/07/98 G 28.7(150) fashion. In another example, a striking

Volume 21 (l), 2001 Natural Areas Journal 93 greenhouse (Young and Young 1986) or drying room where relative humidity can PLANTFAMILY be controlled (Apfelbaum et al. 1997). N = 42 SPECIES D Asteraceae Some species may require special han- 8 m Poaceae Fabaceae dling. For example, orchid capsules should BEST CASE Mdastomataceae be dried thoroughly (i.e., to around 5% 8 mmmmrm onagracaae moisture content) within 24 hours of col- Polygalaceae lection using a desiccant (such as Drierite, Hypericaceae CaS04) to minimize damage by bacteria = Apiaceae and fungi (Zettler 1997). ~ Parnassiaceae m Plantaginaceee Threshing, which is the process of sepa- rating seed from the , occurs 8 10 20 30 40 50 60 70 80 90 100 naturally as capsules or pods dry and split apart. Vigorously shaking capsules in a g 6 MEAN paper bag may be all that is required (Phil- ips 1985). For recalcitrant species several techniques are available for hand thresh- ing: rubbing fruits against a coarse screen, using rolling pins, or rubbing between paddles. For the screening technique (our preference), Apfelbaum et al. (1997) and 8 -+’ 10 20 30 40 50 60 70 80 90 100 Clinebell (1997) recommend 0.25-inch mesh for “most species,” and 0.5-inch for 8 WORST CASE very large seed (e.g., Silphium spp.). Hand threshing can be tedious, and the use of a 4 mechanical hammermill is recommended for processing large amounts of seed (Young andYoung 1986, Apfelbaum et al. 1997).

0 10 20 30 40 50 60 70 80 90 loo Scalping, or coarse cleaning, can be done GERMINATION (%) CLASS with the same size screen used for thresh- ing (Young and Young 1986, Apfelbaum Figure 1. Germination results summarized according to plant family. Top graph (“best case” includes et al. 1997, Clinebell1997). Seed can then the best germination results for each species; middle graph (“mean”) includes mean of all tests; bottom be cleaned more thoroughly by using pro- graph (“worst case”) includes the single poorest result for each species. gressively finer screens (Philips 1985, Apfelbaum et al. 1997; see Clinebell 1997 for recommendations on screen sizes). Me- decline in germination rates was observed Seed Processing and Storage chanical cleaning can involve a fanning among seed collections of Rhynchosiu mill or several other types of equipment reni- Generally, there are four stages in seed formis (a species of dry sites in the Fa- (Young and Young 1986, Apfelbaum et al. processing: seed drying, seed threshing, baceae) made during June 1996 (Figure 1997). After cleaning seed, Band and Hen- seed scalping, and seed cleaning (Young 2). This probably indicated hardening of dry (1993: 9) recommended additional and Young 1986, Band and Hendry 1993, the seed coat and development of induced drying in a “dehumidifying chamber main- Apfelbaum et al. 1997). Processing is re- dormancy, typical of seed maturation in tained at a constant relative humidity of quired if seeds are to be stored or if seed this family (Fenner 1985, Pfaffand Gonter 25% and temperature of 19-23” C until the purification is desirable. 1996).To successfully germinate legume moisture content has fallen to 5-7%.” seed, it appears best to collect shortly be- Harrington (1972, 1973), cited in Apfel- Seeds can be dried by: (1) placing them on fore complete maturity and to germinate baum et al. (1997), recommended a mois- wire mesh trays and circulating air around seeds as soon as possible after collection; ture content of 5-14% for nonaquatic her- them (Apfelbaum et al. 1997), (2) placing otherwise mechanical or chemical scariti- baceous seeds. seed trays in a warm greenhouse (Apfel- cation may be required to promote germi- baum et al. 1997) or ventilated room with nation (Young andYoung 1986; R.J. Mitch- Seed moisture is an important consider- low humidity (Philips 1985), or (3) plac- ell, scientist, Joseph W. Jones Ecological ation in seed storage. Drying times and ing seeds in paper bags in a ventilated Research Center, Newton, Ga., pers. corn.). storage conditions vary with the length of

94 Natural Areas Journal Volume 21 (l), 2001 mination rate dropped to 16%, and germi- 30 1 Rhexia alifanos nants were malformed and chlorotic. Seeds of wiregrass (Aristida beyrichiana) and other longleaf pine ground-layer species are adapted to a high humidity and high temperature environment (even during the winter), and seed dispersal often takes place over several months. Seeds of these spe- Ctenium aromaticum cies may have evolved internal mecha- nisms that allow persistence of seed via- bility under less than optimal humidity 0 and temperature conditions. I I I 9l7l96 1 O/27/96 12l16l96 214197 9l7196 10127l96 12l16l96 2l4l97 Germination Tests Germination tests may be performed prior 60 - to seed collection to determine if seed is Carphephorus paniculatus worth collecting (Pfaff and Gonter 1996) i or before planting to determine how much seed to place in each cell of a seedling tray for later outplanting (Booth and Hendry 1993). Seeds (we commonly use n=lOO) can be placed in suitable containers in an appropriate environment. Seeds are count- ed as they germinate, and a germination percentage is then calculated.

Germination containers can be Petri dish- 10127196 12l16l96 2l4l97 619196 6/l 7196 6125196 713196 es lined with filter paper, although these tend to dry out. A better alternative is a Figure 2. Germination test results plotted according to date of seed collection for four “representative” small plastic germination tray with a re- species of different plant families. Curves are for best fit polynomials. Solid circle = germination tray; movable top (available from Tii-Star Plas- open circle = growing tray. tics, Henderson, KY.) lined with crepe cel- lulose paper wadding (i.e., Kimpak) covered with a single thickness of steel time that seeds must be stored. There are bags in an unheated building. Zetier (1997) blue blotter. This technique, recommend- three basic rules of thumb in seed storage reported that seeds of the orchid Platan- ed by the federal Seed Testing Laboratory (Harrington 1972,1973;Young andYoung thera clavellata remain viable for > 5 years in Dry Branch, Georgia, was used in our 1986; Apfelbaum et al. 1997): (1) each 1% if stored at 0% relative humidity and -7°C; studies. Alternative germination media reduction in seed moisture doubles seed furthermore orchid seeds should be stored include washed silica sand, compost, and life, (2) each reduction of 5” C in seed in complete and continuous darkness be- polyethylene beads (Booth et al. 1993). temperature doubles seed life, and (3) the cause embryos are light sensitive. Gordon-Reedy (1997) used a mixture of sum of the relative humidity (%) and tem- perlite, vermiculite, peat, and sand in seed perature (“F) should not exceed 100%. For We have found that seeds of most longleaf flats to successfully germinate Dodecahe- long-term storage, Band and Hendry pine ground-layer species can be stored in ma Zeptoceras (slender-homed spineflow- (1993) suggested refrigeration of sealed paper bags for several months under nor- er), an endangered plant in California. A polyester/l5+m aluminum foil/polyeth- mal indoor conditions without a serious controlled environment chamber (many are ylene packets at -18” C, an optimal tech- decline in germination. For example, Aris- available commercially at reasonable pric- nique for seeds of rare plants. Apfelbaum tida beyrichiana Trinius & Ruprecht seed es) may be required to satisfy particular et al. (1997) listed three seed storage tech- collected from South Carolina mesic sa- humidity, temperature (Thompson and niques: (1) in unsealed containers placed vannas in December 1992 had a germina- Band 1993), and photoperiod (Thanos in a controlled environment room or cham- tion rate of 52.0% when tested in April 1993) requirements for germination. Al- ber where temperature and humidity are 1993 (n=l 00). One year later (April 1994) ternatively, a germination chamber can be set to specified levels, (2) in gasketed con- the germination rate was nearly identical constructed out of an old refrigerator by tainers with silica gel dessicant refrigerat- at 52.4%. However, after 2 years of stor- installing a fluorescent light source (Young ed at 1.7-4.4” C, and (3) in paper or burlap age under these same conditions, the ger- and Young 1986).

Volume 21 (l), 2001 Natural Areas Journal 95 dry1 compounds, and ethylene (Young and Carphephorus tomentosus Young 1986, Jones and Foote 1990, Th- ompson 1993, Thompson and Booth 1993, Seeds Collected: Thanos 1993 and references therein, Gor- 8ol I- 28 Ott 96 -I I-~ 27 Nov 96-l don-Reedy 1997, Steffan 1997, Zettler 1997). Detailed discussion of these vari- ous procedures is beyond the scope of this paper, and the reader is referred to the above publications (see also Fenner 1985: Chap. 5). We note that in our experience (also B. van Perden, The Nature Conser- vancy, Charlottesville, Va., pers. corn.), prolonged stratification in moist vermicu- lite in a cold storage room does appear to substantially enhance germination of a variety of perennial pine savanna grasses including Andropogon gerardii, Schizachy- rium scoparium, Ctenium aromaticurn, and Erianthus giganteus (Walt.) P. Beauv.

Several publications provide recommen- INFLORESCENCE dations for germinating seeds of native North American plant taxa (e.g., Young and Young 1986, Wiesner 1991, Steffen 80 Carphephorus paniculatus 1997 and references therein). With few *- exceptions (e.g., generally widespread grasses and some legumes that also occur e 60 -Winter Burn-l Summer Burn in tallgrass prairie and oak woodland hab- B itats in the midwestem United States), these 5 40 publications do not list longleaf pine ground-layer species. A potential problem 3 in applying these recommendations to pine $ 20 savanna plants is that significant geograph- 8 ical variation can exist in germination cues. Thus, seeds of a particular species or ge- nus collected from a Wisconsin prairie 26 Ott 96 27 Nov 96 21 Jan 97 24 Jan 97 might require cold stratification to germi- DATE OF SEED COLLECTION nate but this treatment may be unneces- sary when the same species is collected from a South Carolina pine savanna. As an Figure 3. Germination teat results for Carphephorus tomentosus for different dates and stages of seed example, we report (Table 1) a high rate of maturity. Results for C. panicuZutus are shown for comparison. germination for Sporobolus pinetorum Weakley 8z P.M. Peterson, a rare pine sa- vanna grass, without pretreatment. In some instances seeds remain dormant ing. An apparent example of an after-rip- even when exposed to favorable germina- ening effect on seed germination in Schiza- We conducted two germination tests on tion conditions and require treatments to chyrium scoparium is discussed below in each seed collection (see “Seed Collection stimulate germination. One simple treat- “Seedling Issues: Seedling Cultivation.” Sites and Dates” and “Seed Collecting ment is after-ripening, accomplished sim- Technique,” above). For the first test, seeds ply by leaving seeds in storage at room Other treatments useful in breaking seed (n= 50 to 100, see Table 1) were placed in temperature for up to 3 months. According dormancy include cold stratification (dry plastic germination trays on moist blotting to Booth et al. (1993) winter annuals, fall- and moist), scarification, moist heat, alter- paper as described above. Trays were fruiting perennial grasses, and small-seed- nating temperatures, water-soluble inhibi- placed on a bench near a window at room ed species that seed bank are among the tors, avoidance of high irradiation, non- temperature. No attempt was made to con- species most likely to demonstrate im- water-soluble inhibitors, hydrogen trol for light or temperature regime. Seed proved germination following after-ripen- peroxide, variable light regimes, sulphy- germination was tracked for 2 3 months.

96 Natural Areas Journal Volume 21 (l), 2001 For the second test, seedling trays (hereaf- results within families were quite variable: rare by Weakley (1999). Parnassia caro- ter, “growing trays”) with 38 cells/tray some species in each family germinated liniana and Plantago sparsiflora are forbs were filled with fine sand and placed in an well and other species germinated poorly. of wet pine savannas; Rhexia aristosa, an- outdoor nursery at the Santee Experiment When germination data for the different other forb, inhabits cypress savannas and Station, FMNF, under a light shade-cloth families are plotted by species, results for other wet depressions; and Tridens ambig- covering. Seeds were pressed flat on the the two germination techniques appear uus is a grass that lives in wet pine savan- sand, with n=5 or 10 seeds/cell depending highly correlated (Figure 4). This is ex- nas and other types of depressional wet- on seed availability (see Table 1 for over- pected if seed quality is the major limita- lands. Sporobolus pinetorum is a locally all sample sizes). Germination was tracked tion to germination. When data are plotted dominant grass in pristine wet savanna for 2 months. Fresh seed (i.e., seed col- so that growing tray germination is the y- habitats over a very small area of south- lected not more than 3 weeks prior to the axis and germination tray germination is eastern . It was not known start of the test) was used in both tests. the x-axis, the slope of the regression line from South Carolina until very recently. Germination conditions for the outdoor relating the two variables is generally less We discovered it in the FMNF while col- test took advantage of natural germination than one. This indicates that germination lecting seed at PL in 1996 and have since cues typically experienced by seeds of the in the germination trays was generally found it at a few other locations as well. test species. The goal of our germination greater than that in the growing trays. The FMNF populations represent the cur- study was not to understand germination Again, this result is not surprising given rently known southern range limit for the cues, but to determine whether it was pos- the more favorable substrate and higher species. sible, using relatively unsophisticated humidity in the germination trays. More methods, to obtain an acceptable seed ger- interesting is the observation that substan- Except for Rhexia aristosa, which essen- mination rate for a variety of different tial deviations from the various regression tially failed to germinate in both indoor longleaf pine ground-layer plants, includ- lines tended to be positive (i.e., above the and outdoor tests, germination results for ing some rare species. Very little is known line) rather than negative (i.e., below the these five rare species were encouraging about seed germination rates for most of line) (see Figure 4). This suggests that (Table 1). The two grasses in particular the study species (Young and Young 1986, certain species received stimulatory ger- had high germination rates in the indoor but see Whelan 1985, Pfaff and Gonter mination cues in the outdoor treatment test (5 1% for T. ambiguus, 80% for S. 1996). that were lacking in the indoor treatment. pinetorum), suggesting that they lack a This appeared to be particularly true for cold stratification requirement. Pamassia Results of the germination studies were some fall-seeding composites that, like the caroliniana tests were done as part of an encouraging. Of the 42 species, 32 had grasses mentioned earlier, most likely re- experiment to investigate seed density ef- germination rates greater than 20% in at ceived cold stratification prior to germina- fects on germination. Fenner (1985) stated least one test (Figure 1, top). This trans- tion. Another striking example was the that positive or negative density effects are lates into what we might term a “best case much higher germination of Rhexia nashii not uncommon in popu- scenario” success ratio of 78% if a suc- seeds collected in late October 1996 in the lations. Additional factors in the experi- cessful seed collection is defined as one outdoor tray (60%) vs. in the indoor tray ment involved seed spacing (aggregated with at least 20% germination. If species (2.5%) (Table 1). This difference was par- vs. dispersed) and whether or not seeds response is based on mean germination, ticularly striking because the results of the were covered with a thin layer of soil. the success ratio is 74%, still quite good two techniques were quite similar for an- Details of the experimental design are in (Figure 1, middle). Finally, even in a worst other collection made at the same site ear- Glitzenstein et al. (1998a). Results indi- case scenario (i.e., results for a species lier in the same month. It appears that R. cated little effect of seed density or spac- summarized according to its poorest test nashii had developed some form of innate ing (Figure 5). However, covering seeds result), the success ratio is a respectable dormancy (sensu Fenner 1985) over less with soil did produce a significant enhance- 57% (24 of 42 species). This includes tests than a month, which was overcome by ment of germination (i.e., ca. 14 %; see from collections known to be suboptimal some unknown aspect of the outdoor envi- Figure 5). This may indicate a negative due to excessively early or late collecting. ronment. effect of light (see Fenner 1985) or im- proved moisture uptake by covered seeds. In practical terms our results suggest that Four of the species in our germination it should be possible to fill a seedling tray tests appear on Walker’s (1993) list of rare II: SEEDLING ISSUES by placing five seeds into each partition, longleaf pine plants: Pamassia carolini- an operation that is not inordinately diffr- ana, Plantago sparsiflora, Rhexia aristosa, Seedling Cultivation cult. In addition to practical issues, several and Sporobolus pinetorum (this last spe- other interesting observations can be made cies, recently described by Weakley and Germinating seed is the first step in seed- from the germination test results (Table 1, Peterson [1998], is listed as Sporobolus ling cultivation. The second step is to en- Figure 1). First, there was no clear differ- spp. by Walker [1993]). A fifth species, sure adequate growth and survival prior to ence in germination rates among the dif- Tridens ambiguus, did not appear on Walk- outplanting. However, cultivation condi- ferent plant families studied. Germination er’s (1993) original list but is considered tions should not be so favorable as to alter

Volume 21 (I), 2001 Natural Areas Journal 97 the natural balance between root and shoot development or to select against slow- OTHER FAMILIES growing stress-tolerant genotypes that can comprise an important component of nat- ural populations.

Choice of substrate or growth media can be critical for seedling cultivation. Nu- merous growth media are noted in the eco- logical and horticultural literature. For example, Booth et al. (1993), Jones and - LharM Foote (1990), Schramm (1997), Thomp- son (1992), and Gordon-Reedy (1997) discussed media ranging from amended available soil to commercial mixes. We employed two different strategies to culti- 0 20 40 60 80 100 vate rare plant species of longleaf pine habitats. We (JSG, DRS, DDW) have pri- marily used a conservative approach with unmodified soil taken from the A-horizon POACEAE of the restoration/reintroduction site and 80 mixed with a small quantity of soil from the seed donor site. Adding donor site soil is potentially important to provide mycor- rhizal inoculum (Seamon et al. 1989). By using unmodified soil we hoped to select genotypes capable of survival and growth under field conditions. This approach has been tried with some success for two very rare plants of wet sandy pine savannas, Pamassia caroliniana and Plantago spar- sifora (see Glitzenstein et al. 1998a and results below in “Gutplanting and Trans- 0 20 40 60 80 100 locatio n”) as well as various dry-site forbs (e.g., Liatris squat-rosa, Chrysopsis gos- 100 sypina). In contrast, JB uses an artificial ASTERACEAE soil mix that has proven effective for cul- 80 tivating orchids and a wide variety of oth- + / er rare plants of bogs, seepage slope hab-

-/+ A itats, and wet savannas (Table 2). The T 60 mixture consists of 5 parts perlite, 3 parts 1 / vermiculite, and 2 parts chopped or milled sphagnum. The mixture provides pore space and an oxygen supply to roots, and promotes the wet, highly acid conditions required by bog species. To cultivate na- tive grasses, we use a medium composed of 75% soil and 25% vermiculite (or per- lite). These latter media are sterile; in oth- er words, they do not alter soil fertility but 100 0 20 40 60 80 improve drainage and water and nutrient GERMINATION TRAY GERMINATION (%) retention.

Figure 4. Relationship between indoor (germination tray) and outdoor (growing tray) germination. See Steffen (1997) provides a helpful discus- text for descriptions of the two germination techniques. sion of propagation techniques. He advo- cates outdoor germination methods that

98 Natural Areas Journal Volume 21 (l), 2001 We conducted an experiment using Schiza- chyrium scoparium to evaluation growth media and propagation techniques. The ex- periment was a randomized block-split plot design with a seed storage treatment (i.e., room temperature vs. cool outdoors) as the main plot factor and three ratios of soil to vermiculite (lOO:O, 75:25, and 50:50) as the split-plot factor. Results indicated a signifi- cant effect (P I 0.1) of the split-plot factor . (F = 4.56, P = 0.06, df = 26) with fewest seedlings in the soil trays without vermicu- lite (mean = 13.8) and almost three times as many seedlings in the trays with the 75:25 soil:vermiculite mixture (mean = 34.3). In- termediate numbers of seedlings were count- ed in the 50:50 trays (mean = 24.8). The main plot contrast was not statistically sig- nificant (F = 3.71, P = 0.3, df = 1,l). How- ever, the apparent difference in seedling emergence (30.8 indoors vs. 17.7 outdoors) may represent an effect of after-ripening, as discussed above.

Outplanting and Translocation

Site Selection Suitable introduction sites should encom- pass habitat requirements and be within the historic range of the species (Falk et SEEDS al. 1996). However, given the sporadic distribution of many rare plants and recent Figure 5. Relationship of Parnassiu caroliniana germination to seed density. Also shown are effects of range contractions, historic range might seed spacing and germination differences between covered (i.e., with a thin soil layer) and uncovered not be easily determined. For example, Parnassia caroliniana is one of several wet savanna species with a distribution focus in southeastern North Carolina and take advantage of natural germination cues. An important aspect of plant cultivation another in north Florida-southeast Geor- Among his useful suggestions are the fol- that has received little experimental atten- gia (Peet and Allard 1993). Furthermore, lowing: (1) cover seeds lightly with a thin tion is watering regime. By varying water- it occurs sporadically through northern and layer of vermiculite; this material not only ing frequency, species characteristic of both central South Carolina, with a local max- holds moisture but also transmits light and dry and wet habitats can be maintained in imum in the vicinity of Plantersville, about therefore does not interfere with germina- the same soil. For native grasses, soils 20.3 km north of Georgetown. Recent dis- tion of light-requiring seeds; (2) cover should be kept continuously moist during covery of a new population extends this unsheltered trays with window screening seed germination and early seedling estab- distribution about 12.6 km southward to prevent seeds from washing out during lishment. After this stage, grasses charac- (Percher et al. 1999). Our introduced pop- heavy rain; and (3) place over-wintering teristic of mesic-dry sites (e.g., Schizachy- ulation in the FMNF occupies a similar flats in a cold-frame or Quonset hut to rium scoparium, Aristida beyrichiana, A. habitat to the new population. However, prevent excessive freeze-thaw cycles; al- virgata Trin.) tolerate, or even prefer, rel- the reintroduction site is another 44.4 km ternatively, flats can be covered with a tarp atively infrequent watering (e.g., refilling farther south, still outside of the currently and a layer of bark or mulch (in the south- irrigation trays every l-2 weeks) whereas documented southern limit of the northern eastern United States it is necessary to wet savanna species (e.g., Ctenium aro- section of the range of I? caroliniana. Is protect newly germinated seedlings from maticum, Erianthus giganteus) prefer con- this a suitable outplanting site? Only time frost). tinuously wet soils. will tell. We would argue, however, that the apparent advantages of this site (i.e.,

Volume 21 (I),2001 Natural Areas Journal 99 Table 2. Plants in John Brobaker’s bog garden, Awendaw, The above arguments aside, scoparium, Sorghastrum nutans ???????) SC. (E = endangered; SC= Species of Concern; R = rare.) known extirpation sites are and two rare forbs (Pamassia caroliniana, among the best candidates Plantago sparsiflora). We have also done for rare plant reintroduc- a small translocation experiment for Agri- Federal SC State tion, but only if the site can monk incisa, a rare forb. Three of the Species G-Rank List List be managed properly grasses (Aristidu beyrichiana, Ctenium (Fiedler and Laven 1996). aromaticum, Sorghastrum nutans) and the 4 R Implicit in the use of ex- three forbs were outplanted, or transplant- Calopogon pallidus 4 tinction sites is that the rea- ed, in the case of the Agrimonia, into ex- Calopogon tuberosus son for the initial loss of perimental fire research plots in the FMNF Dionaea muscipula 3 SC R the species is known and (see Table 3 for planting dates for each Eriocaulon decangulare can be remedied. In species). These plots occurred at three sites Habenaria repens longleaf pine-dominated that are typical of different longleaf pine Lachnocaulon anceps habitats, the most common dominated habitats. The “dry” site is sub- factor causing local extinc- Macbridea caroliniana 2 SC xeric longleaf pine woodland (sensu Peet tion of rare plants is tire and Allard 1993) on Chipley loamy fine Pamassia caroliniana 2 SC R exclusion or reduced fire sand (thermic coated Aquic Quartzipsam- Pinguicula caerulea frequency. For example, ments). The “mesic” site is mesic longleaf Platanthera blephariglottis 4 populations of the federal- pine woodland with occasional Quercus ly endangered longleafpine marilandica and Q. stellata. Moist savan- 5 ground-layer forb Schwul- na patches are also present but these are Platanthera flava bea americana (American now mostly closed woodlands dominated Platanthera integra 3 R chaffseed) in the FMNF by Pinus taeda, Liquidambar styractjlua, Rhexia alifanus declined catastrophically Nyssa sylvatica, and Q. phellos. The soil is Rhexia lutea with a few years of fire ex- a Craven fine sandy loam (clayey mixed Rhexia mariana clusion (Figure 6). Enhanc- thermic Aquic Hapludult). The “wet” site colorata ing these populations with is moist flatwoods intermixed with wet new seedlings would prob- savanna (see Peet and Allard 1993 for typ- jlava ably not be effective unless ical wet savanna dominants). F’latwoods 3 regular burning was re- patches are dominated by , Sarracenia minor sumed. Furthermore, me- Clethra alnifolia, and Arundinaria recta 4 chanical treatments to re- (Walt.) Muhl. Soils are Quitman loamy 3 R move loblolly pine (Pinus sands (fine loamy siliceous thermic Aeric Spiranthes cemua tuedu) saplings that have Paleaquult) (soils information is from Long Thalictrum cooleyi 1 E invaded due to lack of bum- 1980 and Soil Conservation Service, Xyris platylepis ing may also be required. Charleston, S.C., pers. corn.).

Of the six species planted into the exper- Outplanting Techniques remote location, apparently suitable habi- imental fire plots, three (Aristidu beyri- tat, and a high probability of continued A discussion of outplanting techniques is chiana, Pamassia caroliniana, and Plan- appropriate management by frequent pre- found in Schramm (1997). The author tug0 spursiforu), are true introductions, scribed burning) outweigh the minor un- advocated several site preparation meth- because they were not previously found in certainty about range boundaries (see also ods that should be done prior to outplant- the plots. Plantago sparsiflora and A. bey- . Fiedler and Laven 1996). This argument is ing. Besides fire, we feel that such meth- richiana do, however, occur elsewhere in supported by recent suggestions that sto- ods are generally unnecessary for the FMNF. In fact, extant populations of chastic and historical effects are consider- outplanting seedlings in longleafpine hab- these two species were found very recent- ably more important than has been appre- itats (see results below). Examples of suc- ly (Percher et al. 1999), confirming histor- - , ciated in determining population cessful outplanting projects using a vari- ical observations (Elliott 18 16-1824, distributions and range boundaries of rare ety of propagation methods are presented Percher 1848). Aristidu beyrichiana is a plant species (P White, unpubl. data). Fur- in Guerrant (1996). dominant grass of original pine savannas ’ thermore, Primack (1996) argued that in- over much of its range, but it is rare in troducing a species slightly outside of its Over the last 8 years we have outplanted central South Carolina where it is at the current range is justified because the act nursery grown seedlings of seven perenni- northern limit of its range (see Peet 1993). simulates the process of gradual dispersal al grasses (Andropogon gerardii, Aristida In the FMNF there is only a single popu- that has probably always been part of the beyrichiana, A. virgata, Ctenium aromat- lation (Percher et al. 1999). Plantago spar- ecology of the species. icum, Erianthus giganteus, Schizachyrium sifora presently occurs in three roadside

100 Natural Areas Jourhal Volume 21 (l), 2001 Sorghastrum nutans, the last of the outplant- ed species, is a characteristic component of the mesic woodland portions of the mesic SITE NAME site. It also occurs rarely at the wet site. 200 Presumably it also has been reduced in abun- l Ballfield dance by past episodes of fire exclusion.

The fire experiment was initiated in 1992 .Q 0 Witherbee \ and terminated in 2000. It consisted of 63 160 ,’ B-e fire treatment plots, each 1 ha in area (100 I ‘6 \ n Lethcoe m x 100 m on a side) arranged into three I \ 2 I \ blocks (i.e., the different study sites de- \ scribed above) of 21 plots apiece. Judging 3 120 ! n\ q Highway 41 from the condition of the vegetation at the start of the study, all the sites had a history of at least occasional fire with more fre- quent burning over the last decade. Each 80 of the sites burned in 1991, 1 year prior to the start of the study. Experimental burns were originally to be administered at mean intervals of 2,4, or 6 years with additional treatments incorporating variability in burning season and randomness (i.e., burns occurring at regular intervals vs. burns with random inter-fire intervals but the same mean tire return time). Plots were not burned as scheduled, particularly during 1992 1994 1996 1998 2000 the latter part of the study. Results present- ed herein are summarized according to Fipre 6. Population size changes in four FMNF populations of Sclrwalbea americana with differing number of fires that actually occurred at burn histories. All populations were burned annually from 1992 to 1994. The Witherbee Rd population particular outplanting locations during the also burned in 1998 and Lethcoe Rd was burned in 1996 and 1998. The Ballfield and Highway 41 period 1991-1997 regardless of the as- populations have not burned since 1994. signed burn treatment.

Planting locations within plots were ran- populations maintained by mowing and we attempted to start another A. incisa domly selected for the grasses. However, susceptible to severe disturbances from population in a different research plot at rare forbs were planted into microsites that road grading operations (Percher et al. the same site by moving 16 newly germi- appeared particularly favorable (i.e., can- 1999). nated seedlings and three nongerminated opy gaps, wet savanna patches for Pamas- seeds. The new site is on the edge of a sia caroliniana and Plantago sparsiflora; Three other outplanted species, Agrimo- large canopy opening in the vicinity of a oak-hickory transition for Agrimonia in- nia incisa, Ctenium aromaticum, and patch of Carya alba. cisa). Thus the grass data represent an Sorghastrum nutans, were already present unbiased estimate of bum history effects in the fire research plots prior to the start Ctenium aromaticum is a common com- on planting success within each habitat, of the study. On a standwide scale, plant- ponent of wet savannas throughout its range whereas forb data cannot be extrapolated ings of these species constitute enhance- (Peet and Allard 1993) and is the domi- beyond the selected microhabitats. Seed- ments, rather than introductions per se. nant grass of these habitats in the outer lings (plugs from growing trays) were Agrimonia incisa is a state endangered South Carolina Coastal Plain (Percher outplanted following fires using a small (federal species of concern) forb found in 1995). It is abundant in high quality sa- bulb planter. No additional site prepara- dry longleafpine forests, particularly along vannas in the PMNF, but occurs patchily tion was employed, except in one Planta- oak-hickory ecotones and in other subxer- in the wet site. In the mesic site it occurs go sparsijlora plot where the seedlings ic situations (Weakley 1999). Three known as scattered plants in the savanna patches were planted into small (20 cm x 20 cm) populations occur in the PIvINE One of now dominated by gums and pine. We soil disturbances created by shallow spad- these, a population consisting of approxi- hypothesized that the species was former- ing. Additional details of the experimental mately 250 individuals, spans two plots in ly more abundant at both sites. treatments and outplanting arrangements our “dry” study site. On 30 January 1998 can be found in Glitzenstein et al. (1998b).

Volume 21 (1),2001 Natural Areas Journal 101 Table 3. Survival percentages of seedlings (i.e., plugs from growing trays) planted out or (in the Ultimately, an introduction project can be case of Agrimonia in&a) transplanted into the fire treatment plots. Number of fires refers to considered successful only when it results experimental burns during the period 1991-1998; in parentheses are listed the number of burns in a self-maintaining population with suf- actually survived by the planted seedlings. Abbreviations are as follows. Site: D = dry study plot; ficient genetic diversity for long-term per- M = mesic study plot; W = wet study plot. Dates: W = winter; Sp. = Spring; Su = summer. sistence (Pavlik 1996). In practice, few introduced populations are monitored long enough or with sufficient rigor to deter- Planted No. No. Last Survival mine whether these goals are actually at- Species Site out Plants fires Census (%I tained (Pavlik 1996, Sutter 1996). Over A. GRASSES the short term, success criteria for peren- nial plants typically include survival and Aristida beyrichiana D w 93-94 73 4~2) Jan. 98 32.9 growth of the outplanted individuals along D W 93-94 72 3(l) Jan. 98 55.6 with information on fecundity. Based on D W 93-94 24 31) Jan. 98 54.2 these modest criteria, results of our out- D W-Sp. 97 24 5(l) Jan. 98 45.8 planting experiments are encouraging. D W-Sp. 97 72 4(O) Jan. 98 100.0 Almost all species had initially high sur- D W-Sp. 97 48 3(O) Jan. 98 95.8 vival (i.e., > 90%) during the first year D W-Sp. 97 24 m Jan. 98 100.0 after outplanting (Table 3). Furthermore, in all but a few cases, survival remained M w 93-94 49 4~2) Jan. 98 81.6 high (> 60%) for several years afterwards M w 93-94 90 3(l) Jan. 98 61.5 and most plants have now survived at least M W 93-94 32 2(l) Jan. 98 90.6 one fire (Table 3). Further, many plants are M W-Sp. 97 25 5(0”) Jan. 98 100.0 growing into larger size classes (results M w-sp. 97 73 4(O) Jan. 98 100.0 available online at http://www.talltimbers. M W-Sp. 97 48 3(O) Jan. 98 97.9 org/research.html). M W-Sp. 97 24 20 Jan. 98 100.0 Another criterion for the success of plant W W 93-94 72 4~2) Su. 98 87.5 introductions is persistence (Pavlik 1996). W W 93-94 72 3(l) Su. 98 69.4 Pavlik (1996: 139) cites a review by Bir- W W 93-94 24 2(l) Su. 98 75.0 kinshaw (1991) concerning introduction W W-Sp. 97 24 5(l) Su. 98 87.5 attempts in Britain. Of 144 attempts, 29% W w-sp. 97 88 4(O) Su. 98 96.6 did not persist, 15% were less than 5 years W W-Sp. 97 24 3(O) Su. 98 100.0 old and could not be evaluated, and 22% W W-Sp. 97 24 2(o) Su. 98 91.7 were more than 5 years old and were deemed persistent. There were also a few Ctenium aromaticum M W 93-94 48 4(2) Jan. 98 43.8 examples of introduced populations that M W 93-94 87 30) Jan. 98 55.2 had persisted over the long term (i.e., up to M W 93-94 32 2(l) Jan. 98 53.1 83 years in the case of Arabis stricta Huds., M W-Sp. 97 24 5(0”) Jan. 98 100.0 a limestone outcrop plant). Of the six spe- M W-Sp. 97 72 4(O) Jan. 98 98.6 cies outplanted into our fire plots, three M W-Sp. 97 48 3(O) Jan. 98 100.0 (Aristida beyrichiana, Ctenium aromati- M W-Sp. 97 24 2(O) Jan. 98 100.0 cum, and Parnassia caroliniana) have reached or exceeded the 5year threshold - W W 93-94 72 4(2) Su. 98 77.8 for persistence. The other three species W W 93-94 72 3(l) Su. 98 51.4 cannot yet be evaluated, but results to date W W 93-94 24 2(l) Su. 98 33.3 indicate these populations should survive W W-Sp. 97 25 5(l) Su. 98 84.0 at least 5 years. None of the introductions . , W w-sp. 97 88 4(O) Su. 98 98.9 has thus far failed. W W-Sp. 97 24 3(O) Su. 98 100.0 W W-Sp. 97 24 2(O) Su. 98 100.0 Of all the introductions, the Purnassiu . caroliniana plantings are closest to achiev- Sorghastrum nutans M W-Sp. 97 25 5(08) Jan. 98 92.0 ing true population status. Some flowering M W-Sp. 97 72 4(O) Jan. 98 95.8 has occurred for this species in each of the M w-sp. 97 47 3(O) Jan. 98 95.7 last 4 years. Seeds collected from one cap- M W-Sp. 97 24 NJ) Jan. 98 95.8 sule were planted into a “peat-plug” type continued germination tray and abundant seedlings

102 Natural Areas Journal Volume 21 (l), 2001 III: DIRECT SEEDING Table Fire Plot Introductions Planted No. No. Last Survival A second approach to beginning new pop- Species Site out Plants thS Census (%) ulations is through direct seeding (Primack and Miao 1992, Pavlik 1996, Primack W w-sp. 97 25 5(l) su. 98 44.4 1996, Pfaff and Gonter 1996, Packard W w-sp. 97 88 . 4(O) su. 98 80.0 1997, Tilman 1997). We have experiment- W W-Sp. 97 24 3(O) Su. 98 94.4 ed with this approach but not with rare W W-Sp. 97 24 269 Su. 98 100.0 plants. During October-December 1992 we collected mixed ground-layer seed from B. RARE! FORBS a variety of high quality dry, mesic and wet longleaf pine sites using a Woodward- D Jan. 98 19 Agrimonia incisa 3(O) Apr. 00 26.3 Flail Vat mechanical seed stripper (avail- able from Ag Renewal Inc., Weatherford, W Nov. 95 228 Pamassia caroliniana 40) Fall 98 66.7 Okla.; shown in use in Pfaff and Gonter W Nov. 95b 47 3(O) Fall 98 78.7 1996). Seeds from the various collection sites were mixed together and introduced W Aug. 98 21 Plantago sparsiflora 4(O) Apr. 00 66.7 into seed-addition subplots (n = 3 per main W Apr. 99 18 4(l) Apr. 00 83.3 plot) randomly located within FMNP fire treatment plots. Seeds were introduced a Fire was patchy and planted grass plots did not bum. immediately after experimental fires in b Includes a few individuals planted out the prior November. winter or spring 1993. The subplots are 1.5 m x 2.0 m and were subdivided, using plot-frames, into 48 25-cm x 25cm cells. We tallied all species present in each cell of every subplot, both before and 2 years germinated (germination percentages were and Agrimoniu incisu) were among the after the seed introductions. The same pro- not determined). If any “wild” seedlings least successful in terms of survival (Table cedure was followed in control subplots were produced, they most likely did not 3). We hypothesize that this is because (also n = 3 per main plot). Multivariate survive, because all Purnassia caroliniana moisture stress at this site provides a strong analysis of the dataset was done using plots were burned early in spring 2000 as limitation to initial establishment, a find- CANOCO (ter Braak 1987-1992). There ing supported by results of direct seeding part of a FMNF prescribed burn of the was no significant effect of seed introduc- trials (see below). Because of the addition- surrounding compartment. Seedling estab- tion at the dry site (F = 1.07; P = 0.22), but al stress associated with moisture limita- lishment is a relatively rare phenomenon significant effects were observed at the in most perennial species (Sutter 1996), tion, it may be advisable to wait longer mesic and wet sites (F’s = 1.54,1.27 re- and this may be the case in longleaf pine (i.e., up to 2 years) for plants to establish spectively; P = O.Ol).‘However, these sig- savannas where all but the most vigorous before administering the first fire. In fact, nificant results were due almost entirely to very low first-year survival rates (c 45%) seedlings are likely to be eliminated by one species, Aristidu beyrichianu. When fires within l-2 years after germination. for A. beyrichiunu in the five-bum treat- this species was excluded, the analysis was We can say that, given the precarious con- ments (i.e., plots burned five times be- no longer significant for either site (F’s = dition of Purnussia caroliniana in South tween 1991 and 1998; see Table 3) ap- 1.01, 0.92; P = 0.13, 0.70). Carolina (i.e., most populations threatened peared to be due mostly to effects of by fire exclusion, intensive forestry, drain- burning too soon after outplanting. Sec- Our results suggest that with the exception age, or some combination thereof; see ond, except for Ctenium uromuticum at of a few competition-tolerant species (see the wet site, grasses planted into infre- Percher et al. 1999), our introduced pop- also Tilman 1997 for a similar study in- quently burned plots (i.e., two fires during ulation appears as likely to survive over volving oak savanna plants), establishment 1991-1998, 6 years between fires) had the long term as most of the original pop- from seed is rare in undisturbed longleaf unexpectedly high survival (Table 3), al- ulations. pine habitats. These habitats are typically though the surviving plants had little or no dominated by perennial plants that appear growth (results online at http://www. A detailed analysis of site and fire regime to inhibit establishment of new seedlings effects on dynamics of the introduced pop- talltimbers. org/research.html) More fre- (Brewer et al. 1996, Primack 1996, Tilman ulations will be left to subsequent publica- quent fire after establishment may be nec- 1997). Thus, seed introduction may not be essary for functioning populations of these tions. However, we would like to empha- a particularly effective method for starting size two key points. First, plantings at the grasses. new plant populations. We should empha- dry study site (i.e., Aristida beyrichianu

Volume 21 (l), 2001 Natural Areas Journal 103 size, however, that conclusions from our Beds were then leveled to a uniform depth early spring when up to 3 weeks were small-scale experiment may not be rele- approximately 17 cm below the soil sur- required between censuses to tally all vant to large-scale restorations (see Pfaff face, varying by l-2 cm among beds. emerging seedlings. In addition to moni- and Gonter 1996). Seed introductions that Absolute elevation of each bed was 14 cm toring seedling emergence, we monitored involve seed drills or other machinery may above the water table on 14 April 1998, seedling growth and mortality for planted be successful in locating the occasional the day the beds were constructed. Surface species, and seedling mortality for weeds safe-site and initiating the occasional pop- soil was replaced in each bed, but with soil that died before they could be pulled. These ulation. This technique is also likely to textures randomly assigned: bed 1 was variables were monitored weekly during provide strong initial selection for appro- back-filled with the sandy loam surface 1998. During 1999 growth and mortality priate genotypes, so that, once established, soil that had been removed during the pro- were checked in May and September. In populations have a strong likelihood of cess of excavating the beds; bed 2 was the September census we also recorded ultimate success. For most rare plants the filled with Lakeland sand surface soil; and flowering and various measures of plant limit to such an approach is seed availabil- bed 3 received a surface soil taken from a and inflorescence size. ity. With at most a few hundred seeds, the site mapped as Bonneau loamy sand. Each probability of successful establishment by bed was reconstructed so that the filled Water levels in the wells were monitored direct seeding would be too low to justify soil formed a micro-elevation/drainage on a weekly basis. We also measured soil the approach (see also Reinartz 1997). An gradient with the highest end at 55 cm moisture (i.e., matric potential) using a agricultural plot technique might work, above the clay subsoil and the lowest end QuikDraw tensiometer (Soil Moisture wherein numbers of seeds are first in- at 5 cm, with a constant slope in between. Equipment Corp., Goleta, CA, USA). creased by several orders of magnitude A PVC pipe well (2-inch inner diameter, 5 These two variables were strongly related and then introduced to the restoration site. feet long, 4 feet deep) was installed in (Figure 7), suggesting that hydrological Artificial site preparation, either mechan- each bed to monitor water level fluctua- variations strongly determine droughtiness ical or chemical, might also enhance the tions. in Coastal Plain environments. probability of success by creating addi- tional space for seedling establishment (see To reduce.edge effects, a small buffer of 5 In contrast to the results of the fire plot Morgan 1997). However, such treatments cm was left on both sides of each bed. introduction study described above, nu- are inappropriate when the goal is to es- Beds were then subdivided into sections merous seedlings of a variety of species tablish new rare plant populations in exist- of 90 cm x 50 cm. Each section was fur- successfully germinated (Table 4) and be- ing quality longleaf pine ground-layer. ther subdivided into 2-cm rows, with each came established on the experimental gra- row consisting of a l-cm-wide planting dients (Table 5). Germination was strong- strip and a l-cm-wide buffer. Within the ly related to hydrology/drought, with peak Gradient Study planting strip, potential planting locations periods of germination strongly correlat- A common gardening technique is to cul- were limited to increments of 2 cm begin- ing with prolonged periods of high water tivate plants in competition-free beds ning at 0 cm. As seeds of different species tables and moist soils (Figure 7). There (Jones and Foote 1990). We have used this became available, they were planted into were interesting differences among spe- technique in an experimental context to the bed by randomly selecting one plant- cies (Figure 7). Fall-fruiting grasses and examine establishment from seed of ing location for each of the 25 rows. The composites tended to appear in late winter longleaf pine ground-layer species in the same planting arrangement was used for or early spring. In contrast, spring and absence of competition. Beds were spe- each of the 18 sections (i.e., 3 beds x 6 summer fruiting species had two germina- cially constructed so that we could also sections/bed) to avoid confounding bed or tion peaks, a minor peak in late spring explore effects of important environmen- gradient position with seed arrangement. shortly after seeds were placed, and a major tal gradients (i.e., soil texture and distance The planting technique was to press seeds peak in autumn (Figure 7). Legumes, which to B-horizon) on establishment processes flat and then cover with a thin layer of soil. in our study included the summer-fruiting and community composition of longleaf Seeds were covered with soil in part to species Rhynchosia reniforrnis and Teph- pine ground-layer communities. Three keep them from moving away from their rosia virginiana, were distinctive in that beds were constructed at the FMNF seed assigned gradient positions. they tended to appear throughout the year orchard in an area occupied by natural and were apparently able to take advan- longleaf pine ground-layer (wet-mesic sa- The experiment was initiated in early May tage of relatively brief periods of favor- vanna) but free of canopy trees. Soils in 1998 and terminated in late November able soil moisture. These differences in the area are of the Lenoir series, a sandy 1999. Sixteen species were planted onto seedling germination patterns within loam with a drainage-resistant clay sub- the beds as seeds became available. We longleaf pine ground-layer species are rem- soil at approximately 18 cm (Long 1980). selected plant species characteristic of a iniscent of similar differences within flood- range of different longleaf pine habitats, plain trees that are related to seed size and For each bed, a l-m x 3-m area was estab- from xeric to hydric. Germination was time of seed deposition (Streng et al. 1989). lished and existing vegetation and surface monitored approximately weekly through- soil were removed to expose the subsoil. out the study, except during late winter- By September 1999 many of the plants

104 Natural Areas Journal Volume 21 (l), 2001 ,

+ Matric Potential

2 100 Depth to Water Table * Legumes ti

$ 75 Spring/Summer Fruiting

e Fall Fruiting z 50 5

$j 25

P 0

Figure 7. Seedling appearance patterns on the experimental gradients as a function of hydrology and matric potential, a measure of soil moisture availability.

recruited onto the experimental gradients layer. For the present discussion, the im- Amica acuulis, a spring-fruiting compos- during summer 1998 through spring 1999 portant point is that starting populations of ite, provided a good example of this sort of 9‘ had grown to maturity and flowered (Ta- longleaf pine ground-layer species from variability. Seeds of this species, previous- ble 5). Detailed discussion of soil texture seed is feasible if gradient effects are con- ly determined to be highly viable (Table and gradient influences on vegetation de- sidered and competition is reduced or elim- l), were introduced onto the experimental velopment and compositional patterns will inated. Another point is that a successful gradients in spring 1998 and again in spring be deferred to future publications. For now introduction from seed requires a certain 1999. Due to prolonged drought during we merely state that such effects were degree of good luck in the form of favor- 1998 that lasted into autumn (Figure 7), evident and results provided experimental able climatic conditions. Even if an appro- virtually no seeds of this species germinat- evidence that the two variables included in priate microhabitat is found and competi- ed in the first year of the study. However, the study do in fact control much of the tion is reduced, the introduction may fail if as a consequence of more favorable cli- vegetation pattern in longleaf pine ground- rainfall is inadequate. Study results for mate, numerous A. acuulis seedlings were

Volume 21 (l), 2001 Natural Areas Journal 105 and habitat fragmentation inhibits natural Table 4. Seedlings germinating on the three experimental gradients from May 1998 through November 1999. Seed number refers to number of seeds added during the experiment. Seedlings immigration processes (Matlack 1994, Til- are individuals observed to originate from cotyledons or otherwise determined to be new man et al. 1997). The consequence is that germinants. Note that the few observed seedlings of and Eupatorium rotundifolium may large areas of potentially suitable but un- be explained mainly by the inadequate observation period for these species (i.e., about 1 month occupied habitat may exist for many spe- between planting and the end of the experiment). cies (Primack and Miao 1992, Matlack 1994). In longleaf pine habitats, reinitiat- ing frequent prescribed burning provides a No. Date No. Appearance necessary framework to support high lev- Species Seeds Added Seedlings . (%I els of plant diversity, but dispersal limita- tion could prevent a site from accumulat- Amica acaulis (1998) 900 5112-29198 7 0.8 ing species that would otherwise occur Amica acaulis (1999) 450 6/16-21199 65 14.4 there. The idea that large areas of suitable Chaptalia tomentosa (1998) 450 5112-15198 7 1.6 habitat may exist for rare longleaf pine Chaptalia tomentosa (1999) 450 ' 4118-29199 44 9.8 plants argues for a “Johnny Appleseed” Chrysopsis gossypina 450 l/29-2/8/99 94 20.9 approach, emphasizing initiation of multi- Ctenium aromaticum 450 l/9-11/99 30 6.7 ple new populations instead of expending scarce resources on exhaustive studies of Erigeron vemus (1998) 450 6/15-16l98 12 2.7 habitat factors. Recent findings point to Erigeron vemus (1999) 450 6116-21199 5 1.1 the importance of stochasticity in plant 450 l/29-2/8/99 171 38.0 population initiation and persistence Eupatorium rotundifolium 450 10/14-29199 7 1.6 (Menges 1992). These findings also sup- Liatris squarrosa 450 1/9-l l/99 111 24.7 port the strategy of starting many new Oxypolis filiformis 450 lOl14-29199 4 0.9 populations, since some populations will fail due to random and unpredictable phe- 450 7/13-14198 15 3.3 Polygala lutea nomena including continued anthropogen- Pterocaulon pycnostachyum ic impacts. According to Guerrant (1996: (Michx. Ell) 450 7114-15199 31 6.9 198) “the single most effective design fea- Rhexia alifanus 450 9122-23198 27 6.0 ture for reducing overall risk of failure is Rnynchosia reniformis 450 6123-25198 44 9.8 to reintroduce multiple populations as a Schizachyrium scoparium 450 1 l/30-12114198 118 26.2 standard procedure.” In the FMNF and Tephrosia virginiana 450 814198 112 24.9 other national forests this strategy has been employed successfully for red-cockaded Vemonia angustifolia 450 10124-28198 74 16.4 woodpeckers (Picoides borealis Vieillot); we suggest that a similar strategy can work equally well for many rare plant species. observed in fall 1999. The lesson of this tion and frequent fire on seedling estab- study is that multiple trials may be neces- lishment. However, starting new popula- ACKNOWLEDGMENTS sary if direct seeding is used to start pop- tions may not be particularly difficult if ulations. this “bottleneck” is bypassed by outplant- We thank Andrew Hulin, Mike Lamb, and ing seedlings or cuttings (see Gordon 1996, T. J. Savereno for help with fieldwork. Ledbetter 1997 for two more case stud- Numerous individuals associated with SYNTHESIS AND CONCLUSIONS ies). Of course, considerable effort may be Francis Marion National Forest (FMNF) Our results to date are consistent with a needed to grow and outplant sufficient helped to bum the experimental plots. We theory of longleaf pine plant community numbers of individuals to ensure appro- especially thank Bill Twomey for helping dynamics in which frequent fires reduce priate levels of genetic diversity, but this is to initiate the study and suggesting the but do not entirely eliminate competition a logistical rather than a biological chal- bum block locations. Eddie Stroman acted (Peet et al. 1983). The implication of this lenge. as bum boss for most of the fires. Joe theory for conservation is that unlimited Pinson provided information on Pamas- opportunities exist for starting new popu- Another recent hypothesis supported by sia caroliniana populations. Bates Hill lations of rare and other longleaf pine our results is that many plant species can Plantation allowed us to collect Pumassia ground-layer plants in sites managed ap- be more limited by dispersal than intrinsic seed. The Santee Experimental Forest and propriately with frequent prescribed burn- site factors (Primack and Miao 1992, FMNF Seed Orchard provided nursery ing. As we have discussed, starting popu- Matlack 1994). This is not surprising in a facilities. We thank the National Seed Test- lations from seed can often be difficult due landscape in which anthropogenic impacts ing Laboratory in Dry Branch, Georgia, to interactive effects of limited competi- have led to large declines in biodiversity, for advice on germination procedures.

106 Natural Areas Journal Volume 21 (l), 2001 Table 5. Additional results from the experimental gradients: survival, size, and reproduction in September 1999 of plants originating from seeds germinated prior to that date. Sample size (n) indicates the number of germinated seeds. Note that for some species (e.g., spring-summer fruiting species Arnica and Pterocaulon), most of the germination occurred after the September 1999 census and sample sizes shown below are consequently not as large for these species as they are in Table 4. Flowering is given as a percentage of living plants. Numbers under LItKen are mean values for plant height in cm unless followed by an L which indicates that leaf length measurements are given instead (this was for species with basal rosettes only). For species with a mixture of bolting and non-bolting stems, mean heights were calculated by assigning a height of 1 cm to individuals present only as rosettes.

Bed 1 (sandy loam) Bed #2 (sand) Bed #3 (loamy sand)

, % % I-w % % Ht/ % % I-w Species n alive flower Len n alive flower Len n alive flower Len

4 0.0 - - 2 0.0 - - 3 0.0 - - Chaptalia tomentosa I 0.0 - - 22 0.0 - - 19 2.1 0.0 0.8L Chrysopsis gossypina 28 10.7 - 1.0 33 78.8 50.0 17.2 29 58.6 0.0 1.0 Ctenium aromaticurn 12 0.0 - - 7 0.0 - - 11 33.3 0.0 6.7L 4 0.0 - - 3 0.0 - - 12 0.0 - - Eupatorium leucolepis 53 17.0 44.4 20.9 42 67.3 81.5 34.2 68 61.8 52.4 25.5 Liatris squarrosa 34 52.9 5.6 2.2 43 51.2 50.0 15.0 34 82.4 14.3 4.1 Polygala lutea 4 0.0 - - 3 33.3 100.0 15.0 4 25.0 0.0 1.9 Pterocaulon pycnostachyum (Michx. Ell) 1 0.0 ------Rhexia alifanus 9 0.0 - - 1 0.0 - - 3 0.0 - - Rhynchosia reniforrnis 11 27.3 0.0 0.7 10 10.0 0.0 0.9 22 63.6 0.0 0.9 Schizachyrium scoparium 59 49.2 31.0 25.8 25 72.0 72.2 49.0 34 55.9 31.6 23.5 Tephrosia virginiana 36 19.4 0.0 20.1 34 32.4 0.0 24.0 42 21.4 0.0 18.6 Vernonia angustifolia 25 16.0 0.0 10.2 22 50.0 18.2 12.2 26 38.5 0.0 8.3

Wiregrass seed was collected at Webb isiana for posting their rare lists on the Donna Streng, a Research Associate ut Tall Wildlife Center and Tillman Sand Ridge WWW. Finally, we thank an anonymous Timbers Research Station, is particularly Heritage Preserve operated by state of reviewer for the thoughtful comments. interested in the conservation ofplant biodi- South Carolina, Department of Natural Financial support was provided by the versity in fire-maintained communities of Resources. Richard Percher and Danny USDA Forest Service, Southern Research the southeastern United States. Her other Carlson initiated the Schwalbea study and Station (grants 29-703, 29-898, 29-1267, interests include rare and common plant collected the data through 1994. Albert SRS 33-G-97-110), the Department of demography as well us floristics of South- Meier and Margaret Bailey helped with Energy, Savannah River Site (grant 29- east coastal plain plant communities. the 1997 Schwulbea recensus. J. K. Hiers, 1171), and Francis Marion and Sumter . R. Wyatt, and R. J. Mitchell provided a National Forests (Challenge Cost Share Dale Wade is a Research Forester with the preprint of their interesting manuscript. Agreement SO-97-CCS-24). USDA Forest Service Southern Research Pat McMillan identified Sporobolus pine- Station. His interests includejre behavior torum from site PL in FMNF. Jame Amo- Jeff Glitzenstein is a Reseat-&Associate at and vegetation dynamics in southern pine 1 roso, Bill Carr, and Dorinda Scott provid- stands, with particular emphasis on link- , Tall Timbers Research Station. His inter- ed rare plant lists for North Carolina and ages between these two sets of variables. ests include effects of environmental gru- , respectively. Ms. Amoroso, Mr. He maintains long-term studies of pre- dients and natural and managed distur- . Carr, and Ms. Scott, along with Carolyn scribed burning in several southeastern bance regimes on plant communities and Kindell (Florida Natural Areas Inventory) ecosystems. populations. Recent research he has con- and Cathy Boyle (South Carolina Natural ducted with his wife, Donna Streng, in- Heritage Program), helped us to locate John Brubaker is a Medical Researcher in volves experiments with prescribed burn- recent G-ranks. We also acknowledge the the Department of Pharmacology, Medi- ing and other restoration techniques in Natural Heritage Programs of South Caro- longleaf pine savannas. cal University of South Carolina. He is lina, Georgia, Alabama, Florida, and Lou- also an avid native plant gardenel; uma-

Volume 21 (l), 2001 Natural Areas Journal 107 teur botanist, and conservationist. Aspres- and C.F. Mutel, eds., The Tallgrass Resto- plants as mitigation for environmental im- ident of the Lowcountry Chapter of the ration Handbook for Prairies, Savannas and pacts. Pp. 413-430 in T.S. Elias, ed., Con- South Carolina Native Plant Society he Woodlands. Island Press, Washington, D.C. servation and Management of Rare and has fought to maintain native longleafpine Elliott, S. 1816-1824. A Sketch of the Botany Endangered Plants: Proceedings of a Cali- fornia Conference on the Conservation and ground-layer vegetation and prescribedjre of South-Carolina and Georgia. (Reprinted 1951) Hafner Publishing, . Management of Rare and Endangered in South Carolina. Plants. California Native Plant Society, Falk, D.A., C.I. Millar, and M Olwell. 1996. Sacramento. Guidelines for developing a rare plant rein- LITERATURE CITED troduction plan. Pp. 453-490 in D.A. Falk, Harcombe, P.A., J.S. Glitzenstein, R.G. Knox, C.I. Millar and M. Olwell, eds., Restoring S.L. Orzell, and E.L. Bridges. 1993. Vege- Apfelbaum, S.I., B.J. Bader, F. Faessler, and Diversity: Strategies for Reintroduction of tation of the longleaf pine region of the D. Mahler. 1997. Obtaining and processing Endangered Plants. Island Press, Washing- West Gulf Coastal Plain. Proceedings Tall seeds. Pages 99-126 in S. Packard and CF. ton, D.C. Timbers Fire Ecology Conference 18:83- Mutel, eds., The Tallgrass Restoration Hand- 104. book for Prairies, Savannas and Woodlands. Fenner, M. 1985. Seed Ecology. Chapman and Hall, London. 151 pp. Hardin, E.D. and D.L. White. 1989. Rare vas- Island Press, Washington, D.C.. cular plant taxa associated with wiregrass Fiedler, P.L. and R.D. Laven. 1996. Selecting Band, S.R. and G.A.F. Hendry. 1993. Seed () in the Southeastern Unit- reintroduction sites. Pp. 157-169 in D.A. collecting, cleaning and long-term storage. ed States. Natural Areas Journal 9:234-245. Pp. 9-10 in G.A.F. Hendry and J.P. Grime, Falk, C.I. Millar, and M. Olwell, eds., Re- storing Diversity: Strategies for Reintro- Harrington, J.F. 1972. Seed storage and lon- eds., Methods in Comparative Plant Ecolo- gevity. Pp. 145-245 in T.T. Kozlowski, ed., gy: a Laboratory Manual. Chapman and duction of Endangered Plants. Island Press, Washington, D.C. Seed Biology III: Insects and Seed Collec- Hall, London. tion, Storage, Testing and Certification. Ac- Birkinshaw, C.R. 1991. Guidance notes for Frost, C.C. 1993. Four centuries of changing ademic Press, New York. landscape patterns in the longleaf pine eco- translocating plants as part of recovery plans. Harrington, J.F. 1973. Problems of seed stor- CSD Report i225, Nature Conservancy system. Proceedings Tall Timbers Fire Ecol- ogy Conference 18: 17-44. age. Pp. 251-264 in W. Heydecker, ed., Council, Peterborough, England, U.K. Seed Ecology. Pennsylvania State Univer- Gaston, K.J. 1997. What is rarity? Pp. 30-41 in Booth, R.E.. and G.A.F. Hendry. 1993. Seed sity Press, University Park. viability and germination. Pp. lo-13 in W.E. Kunin and K.J. Gaston, eds., The Bi- ology of Rarity: Causes and Consequences Hiers, J.K., R. Wyatt, and R.J. Mitchell. 2000. G.A.F. Hendry and J.P. Grime, eds., Meth- The effects of fire regime on legume repro- ods in Comparative Plant Ecology: a Labo- of Rare-Common Differences. Chapman and Hall, London. duction in longleaf pine savannas: is season ratory Manual. Chapman and Hall, Lon- selective? Oecologia 125:521-530. don. Glitzenstein, J.S., D.R. Streng, andD.D. Wade. Huenneke, L.F. 199 1. Ecological implications Booth,R.E., J.M.L.Mackey,I.H. Rorison,R.E. 1998a. A promising start for a new popula- tion of Parnassia caroliniana Michx. of genetic variation in plant populations. Spencer, P.L. Gupta, and R. Hunt. 1993. Pp. 3 l-44 in D.A. Falk and K.E. Holsinger, ISP germination and rooting environments: Longleaf Alliance Report 344-58 (Proceed- ings of the Resto- eds., Genetics and Conservation of Rare sand, compost and solution culture. Pp. 19- Plants. Oxford University Press, New York. 23 in G.A.F. Hendry and J.P. Grime, eds., ratation Symposium, Nov. 12-15, 1997, Ft. Methods in Comparative Plant Ecology: a Lauderdale, Fla.). Jones, S.B. and L.E. Foote. 1990. Gardening Laboratory Manual. Chapman and Hall, Glitzenstein, J.S., D.R. Streng, and D.D. Wade. with Native Wildflowers. Timber Press, London. 1998b. Maintaining and restoring species Portland, Ore. Brewer, J.S., W.J. Platt, J.S. Glitzenstein, and diversity in longleaf pine groundcover: ef- Kunin, W.E. 1997. Population biology and D.R. Streng. 1996. Effects of fire-generated fects of fire regimes and seed/seedling in- rarity: on the complexity of density depen- gaps on growth and reproduction of golden troductions. Longleaf Alliance Report 3:72- dence in insect-plant interactions. Pp. 150- aster (Pityopsis gruminifolia). Bulletin of 75 (Proceedings of the Longleaf Pine 169 in W.E. Kunin and K.J. Gaston, eds., the Torrey Botanical Club 123:295-303. Ecosystem Restoration Symposium, Nov. The Biology of Rarity: Causes and Conse- 12-15, 1997, Ft. Lauderdale, Fla.). quences of Rare-Common Differences. Bridges, E.L. and S.L. Orzell. 1989. Longleaf Chapman and Hall, London. pine communities of the West Gulf Coastal Gordon, D. 1996. Experimental translocation Plain. Natural Areas Journal 9:246-263. of the endangered shrub Apalachicola rose- Ledbetter, W. 1997. Restoration of Texas trail- mary (Conradina glubru) to the Apalachi- ing phlox in the pineywoods of east Texas. Buchele, D.E., J.M. Baskin, and C.C. Baskin. cola Bluffs and Ravines Preserve, Florida. Restoration and Management Notes 15:99- 1991. Ecology of the endangered species Biological Conservation 77: 19-26. 101. Solidago shortii. III: seed germination ecol- . ogy. Bulletin of the Torrey Botanical Club Gordon-Reedy, P. 1997. Endangered slender- Lemon, PC. 1949. Successional responses of 118:282-291. homed spineflower: germination and prop- herbs in the longleaf-slash pine forest after agation studies reported (California). Res- fire. Ecology 30:35-145. Center for Plant Conservation. 1991. Genetic toration and Management Notes 15:91-92. sampling guidelines for conservation col- Long, B.M. 1980. Soil survey ‘of Berkeley lections of endangered plants. Pp. 225-238 Guerrant, E.O. 1996. . Designing populations. County, South Carolina. USDA Soil Con- in D.A. Falk and K.E. Holsinger, eds., Ge- Pp. 171-207 in D.A. Falk, C.I. Millar, and servation Service and Forest Service in co- netics and Conservation of Rare Plants. M. Olwell, eds., Restoring Diversity: Strat- operation with South Carolina Land Re- Oxford University Press, New York. egies for Reintroduction of Endangered sources Conservation Commission and Plants. Island Press, Washington, D.C. South Carolina Agricultural Experiment Clinebell, R.R. 1997. Tips for gathering indi- Station, Moncks Comer, S.C. vidual species. Pp. 127-134 in S. Packard Hall, L.A. 1986. Transplantation of sensitive

108 Natural Areas Journal Volume 21 (l), 2001 Matlack, G.R. 1994. Plant species migration in 1991. Evaluating pyrogenicity and its ef- Monographs 59: 177-204. a mixed history forest landscape in eastern fects on vegetation in longleaf pine savan- Streng, D.R., J.S. Glitzenstein, and W.J. Platt. North America. Ecology 751491-1502. nas. Proceedings of the Tall Timbers Fire 1993. Evaluating effects of season of bum Means, D.B. 1997. Wiregrass restoration: prob- Ecology Conference 17:143-162. in longleaf pine forests: a critical literature able shading effects in a slash pine planta- Percher, F.P. 1848. Medico-botanical catalogue review and some results from an ongoing tion. Restoration and Management Notes of the plants and ferns, St. Johns Berkeley, long-term study. Proceedings Tall Timbers 15:52-55. SC. M.D. thesis (in part), Medical College Fire Ecology Conference 18:45-82. Menges, E. 1992. Stochastic modeling of ex- of South Carolina. Bucone and James, Sutter, R.D. 1996. Monitoring. Pp. 235-264 in . tinction in plant populations. Pp. 253-275 Charleston, S.C. D.A. Falk, C.I. Millar, and M. Olwell, eds., , in P.L. Fiedler and S. Jam, eds., Conserva- Percher, R.D. 1995. Wildflowers of the Caro- Restoring Diversity: Strategies for Reintro- tion Biology: the Theory and Practice of lina Lowcountry and lower Pee Dee. Uni- duction of Endangered Plants. Island Press, , Nature Conservation, Preservation and Man- versity of South Carolina Press, Columbia. Washington, D.C. agement. Chapman and Hall, New York. Percher, R.D., P. McMillan, D.R. Streng, and ter Braak, C.J.F. 1987-1992. CANOCO: a L Morgan, J.P. 1997. Plowing and seeding. Pp. J.S. Glitzenstein. 1999. A floristic and FORTRAN program for canonical commu- 193-216 in S. Packard and C.F. Mutel, eds. ecological inventory of potential high qual- nity ordination by [partial] [detrended] [ca- The Tallgrass Restoration Handbook for ity longleaf pine sites in South Carolina. nonical] correlation analysis, principal com- Prairies, Savannas and Woodlands. Island Final report to Natural Heritage Program, ponents analysis and redundancy analysis Press, Washington, D.C. South Carolina Department of Natural Re- (version 2.1). Morse, L.E. 1996. Plant rarity and endanger- sources, Columbia. Thanos, C.A. 1993. Germination and the high ment in North America. Pp. 7-22 in D.A. Primack, R.B. 1996. Lessons from ecological irradiance reaction. Pp. 187-189 in G.A.F. Falk, C.I. Millar, and M. Olwell, eds., Re- theory: dispersal, establishment and popu- Hendry and J.P. Grime, eds., Methods in storing Diversity: Strategies for Reintro- lation structure. Pp. 209-234 in D.A. Falk, Comparative Plant Ecology: a Laboratory duction of Endangered Plants. Island Press, C.I. Millar, and M. Olwell, eds., Restoring Manual. Chapman and Hall, London. Washington, DC. Diversity: Strategies for Reintroduction of Thompson, K. 1993. Germination at altemat- Packard, S. 1997. Interseeding. Pp. 127-134 in Endangered Plants. Island Press, Washing- ing temperatures. Pp. 183-185 in G.A.F. S. Packard and C.F. Mutel, eds., The ton, D.C. Hendry and J.P. Grime, eds., Methods in Tallgrass Restoration Handbook for Prai- Primack, R.B. and S.L. Miao. 1992. Dispersal Comparative Plant Ecology: a Laboratory ries, Savannas and Woodlands. Island Press, can limit local plant distribution. Conserva- Manual. Chapman and Hall, London. Washington, D.C. tion Biology 6:513-519. Thompson, K. and S.R. Band. 1993. Germina- Pavlik, B.M. 1996. Defining and measuring Ravenel, H.W. 1876. Some rare southern plants. tion at a range of constant temperatures. Pp. success. Pp. 127-156 in D.A. Falk, C.I. Bulletin of the Torrey Botanical Club 6:81- 185-187 in G.A.F. Hendry and J.P. Grime, Millar, and M. Olwell, eds., Restoring Di- 82. eds., Methods in Comparative Plant Ecolo- versity: Strategies for Reintroduction of Reinartz, J.A. 1,997. Restoring populations of gy: a Laboratory Manual. Chapman and Endangered Plants. Island Press, Washing- rare plants. Pp. 89-98 in S. Packard and Hall, London. ton, D.C. C.F. Mutel, eds., The Tallgrass Restoration Thompson, K. and R.E. Booth. 1993. Dorman- Peet, R.K. 1993. A taxonomic study of Aristi- Handbook for Prairies, Savannas and Wood- cy breaking. Pp. 181-183 in G.A.F. Hendry da stricta and Aristida beyrichiana. Rhodo- lands. Island Press, Washington, D.C. and J.P. Grime, eds., Methods in Compar- ra 95:25-37. Schramm, P. 1997. Hand-planted prairies. Pp. ative Plant Ecology: a Laboratory Manual. Peet, R.K. and D.J. Allard. 1993. Longleaf 217-222 in S. Packard and C.F. Mutel, eds., Chapman and Hall, London. pine vegetation of the Southern Atlantic The Tallgrass Restoration Handbook for Thompson, P. 1992. Creative Propagation: a and Eastern Gulf Coast regions: a prelimi- Prairies, Savannas and Woodlands. Island Growers Guide. Timber Press, Portland, Ore. nary classification. Proceedings Tall Tim- Press, Washington, D.C. Tilman, D.1997. Community invasibility, re- bers Fire Ecology Conference 18:45-82. Schultz, R.P. 1976. Environmental change after cruitment limitation, and grassland biodi- Peet, R.K., D.C. Glenn-Lewin, and J. Walker site preparation and slash pine planting on versity. Ecology 78:81-92. Wolf. 1983. Prediction of man’s impact on a flatwoods site. Research Paper SE-156, $ Tilman, D., C.L. Lehman, and C. Yin. 1997. plant species diversity: a challenge for veg- U.S. Department of Agriculture, Forest Habitat destruction, dispersal and determin- etation science. Pp. 41-54 in W. Holzner, Service, Southeast Forest Experiment Sta- istic extinction in competitive communi- M.J.A. Werger, and I. Ikusima, eds.; Man’s tion, Asheville, N.C. 20 pp. ties. American Naturalist 149:407- Impact on Vegetation. Dr. W. Junk Pub- Seamon, P.A., R.L. Myers, L.E. Robbins, and Waldrop, T.A., D.L. White, and S.M. Jones. , lishers, The Hague, The Netherlands. G.S., Seamon. 1989. Wiregrass reproduc- 1992. Fire regimes for pine-grassland com- k Pfaff, S. and M.A. Gonter. 1996. Florida na- tion and community restoration. Natural munities in the southeastern United States. tive plant collection, production and direct Areas Journal 9:264. Forest Ecology and Management 47:195- seeding techniques: interim report. U.S. Steffen, J.F. 1997. Seed treatment and propa- 210. Department of Agriculture, Natural Resourc- ? gation methods. Pp. 151-162 in S. Packard Walker, J. 1993. Rare taxa asso- es Conservation Service, Plant Materials and C.F. Mutel, eds., The Tallgrass Resto- ciated with longleaf pine ecosystems: pat- Center, Brooksville, Fla, ration Handbook for Prairies, Savannas and terns in and ecology. Proceed- Philips, H.R. 1985. Growing and Propagating Woodlands. Island Press, Washington, D.C. ings Tall Timbers Fire Ecology Conference Wildflowers. University of North Carolina Streng, D.R., J.S. Glitzenstein, and P.A. Har- 18: 105-126. Press, Chapel Hill. combe. 1989. Woody seedling dynamics in Weakley, A.S. 1999. Flora of the Carolinas Platt, W.J., J.S. Glitzenstein, and D.R. Streng. an east Texas floodplain forest. Ecological and Virginia. Unpublished working draft of

Volume 21 (l), 2001 Natural Areas Journal 109 January 1999, The Nature Conservancy, Wiesner, L. (ed., Association of Official Seed Zettler, L.W. 1997. Terrestrial orchid conser- Chapel Hill, N.C. Analysts). 1991. Rules for testing seeds. vation by symbiotic seed germination: tech- Weakley, AS. and P.M. Peterson. 1998. Tax- Journal of Seed Technology 12:1-109. niques and perspectives. Selbyana 18: 188- onomy of the jloridanus com- Young, J.A. and C.G. Young. 1986. Collect- 194. plex (Poaceae: Sporobolinae). Sida 18:247- ing, Processing and Germinating Seeds of Zettler, L.W. and T.M. McInnis, Jr. 1992. Prop- 270. Wildland Plants. Timber Press, Portland, agation of Platanthera integrilabia Correll) Whelan, R.J. 1985. Patterns of recruitment to Ore. Luet, an endangered terrestrial orchid, plant populations after fire in Western Aus- through symbiotic seed germination. Lind- tralia and Florida. Proceedings Ecological leyana 7:154-161. Society of Australia 14:169-178.

1 IO Natural Areas Journal Volume 21 (l), 2001