Restoration of Longleaf Pine Ecosystems

Seed bank viability in disturbed longleaf pine sites

Cohen, Susan1, Richard Braham2, and Felipe Sanchez 1
1USDA Forest Service, Southern Research Station, 3041 Cornwallis Rd.
Research Triangle Park, NC 27709 and 2North Carolina State University, Department of Forestry, Biltmore Hall, Raleigh, NC 27695


Some of the most species-rich areas and highest concentrations of threatened and endangered species in the Southeast are found in wet savanna and flatwood longleaf (Pinus palustris Mill.) communities. Where intensive forestry practices have eliminated much of the natural understory vegetation of longleaf ecosystem, the potential for re-establishment through a seed bank presents a valuable restoration opportunity. Longleaf pine sites converted to loblolly pine plantations and non-disturbed longleaf sites on the Coastal Plain of North Carolina were examined for seedbank viability and diversity. Conducting vegetation surveys and examining the seed bank using the seedling emergence technique, allowed for verification of the seed bank presence, as well as evaluation of the quality of the seed bank in disturbed longleaf pine sites.

Over 35 species and 1000 individuals germinated, and the seed banks of both stand types contained species not noted during the vegetation survey. While many of these species were weedy and typical of disturbance, numerous taxa were indicative of stable longleaf communities. This study confirms both the presence and quality of seed banks in highly disturbed longleaf pine sites.


Due to human activities, the area occupied by longleaf pine ecosystems has been reduced from an estimated 80 million acres in presettlement times to less than 3 million acres (Frost 1990). Generally categorized according to moisture conditions, longleaf pine ecosystems range from xeric sand hills to wet flatwoods and savannas, and associated with many different soils are unique understory conditions. Land conversion to agriculture and loblolly pine (Pinus taeda L.) plantations account for much of reduction on the more moderate to poorly drained sites.

Roughly 95 percent of today’s remnant longleaf stands are found on xeric to dry-mesic sites that are too infertile to support agriculture or intensive plantation culture (Frost 1990). The wetter savanna and flatwood communities are more interesting, however, because they contain higher understory diversity (Christensen 1981). Some of the most species-rich areas and highest concentrations of endangered and threatened species in the Southeast are found on these wet sites (Walker and Peet 1983, Rome 1988, Frost 1990).

Because the wetter longleaf communities are rare, restoration efforts on wet sites will promote conservation of an endangered ecosystem. Where the natural understory vegetation has been eliminated by severe disturbance, the potential for re-establishment through a seed bank presents a valuable opportunity (Simpson et al 1989). Sowing and transplanting of native species is costly, time consuming, and may not be possible if seed is not commercially available. In addition, the increasing isolation of intact, rare community types makes natural seed dispersal of desired understory species unreliable (van der Valk and Pederson 1989).

Determining the presence of a seed bank and its composition is an essential first step before a management or restoration plan that relies on the seed bank is attempted (van der Valk and Pederson 1989). If the seed bank of disturbed sites proves to be viable with species typical of intact longleaf ecosystems, more sites will be available for restoration. This is particularly important for the rare, wet longleaf ecosystems. The objectives of this study were to verify the existence of a persistent seed bank in disturbed longleaf pine sites, and if present, evaluate the quality of the seed bank.

Study Sites

Study sites are on the Croatan National Forest, located in Jones, Craven and Carteret counties. Two treatments--disturbed and non-disturbed--were established with 4 replicate study sites each. All sites are predominantly on Leon soils--a poorly drained soil of coarse sediment, with a weakly cemented spodic horizon (Bh) (Goodwin 1987, Goodwin 1989, and LeBlond 1999).

The four sites of the disturbed treatment have all undergone severe understory disturbance associated with forestry activities, and are currently in loblolly pine plantations established between 1971 and 1975. These sites have undergone mechanical site preparation on a minimum of two occasions and are sporadically burned for fuel reduction and wildlife purposes (Cherry, pers com 1999). The four sites of the non-disturbed treatment are fire maintained natural longleaf stands that served as vegetative standards in the study. These sites primarily contain second growth longleaf established between 1897 and 1900, with occasional scattered old growth (Croatan National Forest 1997). The nondisturbed sites have never experienced understory disturbance, and are considered to be the best approximation of pristine longleaf pine flatwood or savannas sites on Leon soils within the Croatan National Forest (LeBlond 1999, Braham 1999, and Schafale 1994).


The methodology was designed to allow for sampling and evaluation of the seed bank contained in the soil from each study site and to allow for comparison of the species that germinate from each treatment with the standing vegetation currently present on each site.

A modified version of the North Carolina Vegetation Survey (NCVS) was employed to examine the standing vegetation of each site (Peet et al. (1998). Percent foliar cover of shrubs and ground line species was determined, as was density and basal area of tree species. Evaluation of the seed bank was completed using the seedling emergence technique. Soil was collected and seeds germinated in a greenhouse using the most favorable germination conditions with regards to light, temperature, and moisture. Soil sample collections were made in August 1999 and May 2000, and each collection was monitored for 9 months. Plants germinated from seed were collected and identified, with the date of germination and study site recorded.

Summary of Results

Vegetation survey

The nondisturbed sites were dominated by longleaf pine and the only other tree species was Pinus serotina (pond pine). Many species typically occurring in these types of sites have not reached tree size probably due to the regular prescribed burn pattern used to manage these sites. As expected, loblolly pine accounted for over 75 percent of the basal area in the disturbed sites. Hardwood species typically found on wet mineral soils of the Coastal Plain, such as Acer rubrum, Nyssa sylvatica , Magnolia virginiana and Persea borbonia were also present. The basal are a in the disturbed sites is approximately twice that found in the nondisturbed site and there are almost 9 times as many stems/hectare.

With regards to the shrub and groundline vegetation there is a shift from shrub to groundline vegetation dominance going from disturbed to nondisturbed sites. There is probably a higher occurrence of shrubs on the nondisturbed sites due to the dormant season growing regime. Although the sites share many of the same species -particularly with reference to shrubs, the abundance of these species varies widely among site type.

The primary shift being that from herbaceous to woody species when going from the nondisturbed to disturbed sites, in addition to the typical pocosin shrubs gaining dominance in the disturbed sites. The disturbed sites do exhibit high resiliency to disturbance and in addition to many of the ruderal type species, many species typical of intact longleaf communities can be found. Complete species list, as well as percent cover by species, will be present in the manuscript currently in preparation.

Seed bank evaluation

The disturbed sites had greater numbers of individuals and greater numbers of species than the nondisturbed sites. This is to be expected as dominant plants in natural ecosystems tend not to accumulate large seed banks and equal disturbance on a predictable schedule--like low intensity, high frequency fire--tends not to select for the accumulation of a large seedbank.

The table presented below provides a brief summary of the numbers of species detected in the seed banks from both the August 1999 and May 2000 collections. Species were considered indicative if cited as consistently occurring in intact longleaf pine sites by sources such as plant community inventories and vegetation studies. Weedy plants were those typical of highly disturbed areas. These species are not likely to perpetuate themselves in fire maintained ecosystems.

Almost half of the indicative species, germinated in the greenhouse from the disturbed sites in the 1999 collection, were not found in the standing vegetation. That number increases to almost 75% from the disturbed sites from the 2000 collection. Lack of correspondence between standing vegetation and seed bank is a very common finding in seed bank research from other community types. In addition, weedy species are not well represented, again following the literature.

There were a greater number of species in the May 2000 collection and greater numbers of individuals from the August 1999 collection. These differences may be due to seasonal variation in dormancy breaking requirements based on collection time, uneven distribution of seeds in the soil, or some other factor. Detailed results will be present in the manuscript currently in preparation.