Long-term studies of forest dynamics on the North Carolina Piedmont

 

 

Introduction

 

Disturbance, natural or human-caused, is nearly ubiquitous in both natural and anthro­pogenic landscapes with the consequence that succession is a dominant process responsible for much of the variation in ecosystem structure and function. For this reason, the mechanisms inherent in successional change have long been and still remain of central interest to plant community ecologists.  The forests of the North Carolina Piedmont have been a focus of research on secondary forest succession for many years as documented in numerous classic papers (Oosting 1942, Billings 1938, Keever 1950, Bormann 1953, among others).  As a result, more is known about secondary forest succession on the North Carolina Piedmont than for perhaps any other system, and these forests have attained the position of a model system for succession studies (Christensen & Peet 1981).

 

The present research program began with the observation that community-level properties and processes are often intimately linked to and can be best understood as consequences of population processes (Peet and Christensen 1980, 1987; reviewed in Peet 1992).  Elucidation of the nature and mechanisms of population change during succession requires information on the fates of individual plants followed over significant time periods.  Ultimately, our understanding of community dynamics and our ability to test related theory will depend on the availability of long-term data that span a broad range of developmental stages, site conditions, and species.  Unfortunately,  the long-term records currently available are extremely limited and often insufficient for addressing the questions of broadest ecological interest.

 

In the late 1970's, acutely aware of the need for long-term observations, one of us (RKP, together with NL Christensen)  initiated a research program to study secondary forest succession on the North Carolina Piedmont centered around exploitation of a large set of permanent sample plots with individually marked trees established in the Duke and Hill Experimental Forests during the period 1933-47.  With the support of NSF funding, these sample plots were maintained and subsequently  supplemented with (1) a large set of survey plots that document compositional variation of both herbs and trees with respect to site conditions, (2) extensive areas of mapped forest for study of spatial processes, and (3) extensive monitoring of seedling and sapling demography to document the critical establishment phase of tree growth.  We know of no other database that approaches ours in the range of sites studied, the number of individuals observed, and the number of years over which the observations have been maintained.  The results of our program are partially documented in the section of this proposal on Results of prior NSF-supported research and are reviewed in Peet & Christensen (1987) and Peet (1992).

 

In September 1996, Hurricane Fran crossed central North Carolina and caused substantial tree mortality in parts of the Duke Forest.  Nearly all of our research plots lost at least a few trees and some lost in excess of 50% of the canopy. Our research program to date has of necessity focussed primarily on plots that have experienced only occasional, single-tree gaps, though several of our plots were significantly damaged by Hurricane Hazel in 1954.  The occurrence of Fran provides a unique opportunity to examine the impact of a major wind event on a series of forest sites for which there are many years of baseline data on tree, seedling, and herb dynamics. We can also compare, to at least a limited extent, the recovery patterns following Hurricanes Hazel and Fran.

 

 

 

The LTREB Program

This proposal is being submitted to the "Long-term Research in Environmental Biology" (LTREB) Program, which was established by NSF in 1982 to facilitate research requiring collection of field data over long periods of time.  Toward this end LTREB grants are intended to provide modest levels of funding over five-year periods, while usually not providing the main line of research support for an investigator.  Consequently, these grants generally do not include salary for the PIs or support for data analysis or publication.  The expectation is that analysis and publication will be supported by normal research proposals.  Accordingly, this proposal focuses more on data collection and potential application of those data than on the specific hypotheses to be tested or the actual tests to be employed.  These will provide the focus for subsequent proposals. Nonetheless, we do outline some potential applications of these data in a concluding section on Applications and Analysis.

 

Long‑term goals

The overall objective of our research program is to understand the dynamics of forest communities using North Carolina Piedmont forests as a model system.  The overall objective of this research proposal is to make it possible for us to maintain and, where necessary, expand long‑term observations that will help us  and other workers to achieve this better understanding of forest dynamics.  A basic premise of our work is that much of forest dynamics and succession can best  be understood as a consequence of the population dynamics of the dominant tree species, an approach first articulated in Peet and Christensen (1980) and more recently fully elaborated and documented in Peet (1992).  The slow growth of forest trees greatly limits opportunities to document tree dynamics over the full period of stand development, and thus greatly limits our ability to investigate the population processes that underlie succession and community dynamics.  The present proposal is designed to continue and expand efforts needed to build a database adequate for such population-based studies of forest dynamics.

 

Specific Objectives

The primary purpose of this proposal is to support maintenance and recensus of a broad array of  permanent plots located in or near the Duke Forest.  A secondary objective is to organize and archive data from Duke Forest permanent plots so as to provide better access for other researchers.  Funding for this proposal is particularly urgent because the most rapid changes resulting from the impact of Hurricane Fran can be expected to take place in the first year or two following the storm.  If we are to make the most of the unique opportunity provided by Fran, intensive sampling needs to be initiated in the summer of 1997.

 

In addition to examining the impact of Hurricane Fran and continuing our investigations of tree demography and successional convergence, we foresee major efforts to examine (1) the dynamics of the transition period between the self-thinning and steady-state phases of forest development when many ecosystem properties change dramatically (see Christensen and Peet 1981, 1984; Peet 1992), and (2) similarities and differences in forest dynamics and tree demography on contrasting sites within the region, as well as between regions. A more complete elaboration of anticipated use of the resultant data is presented in the section on Applications and Analysis.

 

Background

 

Results of Prior NSF Support:  Robert K. Peet

 

BSR‑8905926:  Long‑term studies of forest dynamics (LTREB); awarded to Robert K.  Peet (PI) and Norman L. Christensen (Co‑PI); $174,930; 1989‑1994. (Plus an REU supplement for $4321.)

 

BSR‑9107357:  Long‑term studies of forest dynamics: analysis of  seedling and sapling populations;  awarded to Robert K. Peet (PI, $146,729) and Norman L. Christensen 1991‑93.

 

During the past seven years my  research on forest succession and dynamics has been supported by two NSF grants.  The first NSF grant was an LTREB grant for maintenance and development of long-term permanent plots in and near the Duke Forest.  I will first describe the database that has been developed using this and earlier LTREB funding as it will form the basis for the proposal that follows.  I will then describe some of the applications that have been made of these data, including  those associated with the second grant specifically awarded to look at tree regeneration.

 

The Duke Forest LTREB Database ‑ a Research Resource

One of our primary objectives, as stated above, was to build a database that would allow us to examine the extent to which we can understand succession as a population process.  Although we have collected many  types of data as part of this project, we focus here on those datasets that will be a major focus of the LTREB renewal.

 

Original Permanent Sample Plots.  Between 1931 and 1947 permanent sample plots (PSPs) with individually numbered trees were established in the Duke Forest; in addition­ plots were established in the Hill Experimental Forest.  These plots have been recensused at roughly 5-year intervals since establishment.  During the period of our LTREB grant we resurveyed all 37 extant PSPs.  During remeasurements we mapped and recorded the diameter and height of all trees over 1 cm dbh, including new ingrowth.  These plots cover approximately 4 ha of forest and currently contain nearly 11,000 mapped and measured trees (Table 1).  All data have been systematically checked for errors and inconsistencies.

 

Mapped Forest Stands.  The original PSPs ranged in area from 405 to 4047 m2, though the majority were less than or equal to 1012 m2.  This modest size precludes meaningful studies of spatial patterns, gap dynamics, and demographic investigations of other than the few dominant species, all features of forests critical to understanding succession.  For this reason we mapped large forest blocks (including those with intensive seedling demography plots; see below), including 10 mapped or remapped during the most recent funding period (50% mapped twice). These 10 largest plots together cover approximately 24 ha of forest and include 38,000 living trees (Table 2).

 

Seedling Plots.  Despite their foresight, the foresters who established our permanent plots were not interested in seedlings or saplings.  Consequently, we inherited extensive long-term records for trees, but not for seedlings.  This is unfortunate as establishment is often the critical step in determining whether a species will succeed in a particular habitat (Grubb 1977, Harper 1975), and if so when.  To compensate, we initiated observations of seedling demography in 1978.  Seedlings were mapped along permanently marked transects 1-m wide and 10 to 50-m long.  These transects cross the normal range of spatial variation in each stand.  Annually, seedlings of all dominant tree species were mapped by Cartesian coordinates along each transect.  Using data sheets prepared from the previous year's census, field crews have found it relatively easy to relocate survivors and to identify new ingrowth.  The identity of each seedling, whether it was alive, and its height, and number of leaves (up to 20) were recorded.  These data were checked systematically for a broad range of possible errors.

 

We initially established 10 intensive seedling study plots (250-300 m of transect) and 11 auxiliary seedling plots (ca 50m).  The transects cross the normal range of spatial variation in a forest and thus allow us to document aver­ages for stands as well as heterogeneity within stands.  Starting in 1989 we shifted to maintaining 5 intensive seedling plots representing three types of mature hardwood forest and two stands near the stage of transition from loblolly pine to hardwoods. Annual measurements continued through 1994.  For these five sites we also monitored sapling (> 1m tall and < 1 cm dbh) growth annually in a set of 4x50 m transects that overlap the seedling transects (Table 2).

 

Compositional Survey Plots.  In 1977, under separate funding, circa 230 0.1 ha plots were distributed across the Duke Forest to document variation in species composition and tree population structure.  Unlike the great majority of permanent plots in North America, these plots have detailed information on herbaceous species and soil chemistry. In addition, during the  early 1990's 62 0.1 ha plots with detailed tree and herb data were established across a range of natural areas within the NC Botanical Garden.

 

Applications made of the Duke Forest LTREB Database

Tree seedling demography.  Tree seedling population processes are little known and of necessity typically treated as a black box in studies of forest dynamics.  In Philippi and Peet (1994, 1996)  we report age- and size-specific growth and survivorship rates over a 17-yr period for 95076 seedlings representing 14 tree taxa in 2 forest types (upland pine and upland mixed hardwoods). Our objectives were to determine the potential importance of including seedling processes in studies of forest dynamics, and to assess which aspects of tree seedling dynamics are most important for projecting changes in forest composition.  We reported several findings contrary to expected results: rates of seedling influx were not tightly coupled to basal area of adults;  relative growth rate (RGR) was essentially independent of seedling age and size (height); growth rates of individual seedlings in sequential years were not correlated; there was no evidence of seedling release from competition  as evidenced by the occurrence of high growth rates for 3 or more consecutive years.  For all species, the probability of surviving to the following year increased with both age and size.  For most species, RGR was not predictive of survivorship to the following year.  Slow growth rates preclude estimation of seedling growth and survival to sapling size in studies less than several decades in duration.  We estimated probabilities for seedling survival to 1m with simulations based on probability distributions of size-specific survivorships and growth rates.  The probability of a seedling surviving to reach 1m in height (p1m) varied among species groups from 15% to less than 1 in 1000.  The large-seeded species groups (Carya spp., Quercus spp. ) had high survivorships to 1m.  Quercus alba had a much lower p1m in hardwood stands than in pine stands, which is consistent with the commonly reported  successional trends from Pinus to Quercus and from Quercus to Acer.  Median times to reach 1m in height ranged from 17 to 38 yrs. Resprouting, both following dieback of an individual and vegetative production of additional shoots, was common, and demographic parameters of sprouts are strikingly different from those of new seedlings.  Our results suggest that seedlings in closed-canopy forests, rather than acting as a bank of advance regeneration waiting for release of a few individuals, can better be thought of as existing in a leaky pipeline, on the way to a sapling bank.

 

Within-season and small-scale spatial variation in seedling growth and survival were examined in three of  our forest stands by Lopez-Mata (1994, Lopez-Mata & Peet 1996a, b, c) in an effort to identify mechanisms of coexistence of the various species of Quercus and Carya that dominate the canopy.  Seedlings and saplings were found to be less abundant near conspecific adults than near adults of other species.  Differences were also found in plasticity of growth in response to canopy openings.

 

Community composition and dynamics.   We examined long-term trends in growth and mortality rates in even-aged stands of loblolly pine using permanent sample plots.  This work has led us to view forest stand development in terms of  the now classic  four-stage process described by Bormann and Likens (1979) and others (e.g., Oliver 1981, Peet 1981) .  During the first or establishment stage, mortality is low and growth rates are high.  The second stage is characterized by intense competition, high mortality of smaller individuals, and a roughly constant overall mortality rate.  Little establishment occurs during this stage and mortality rough­ly corresponds to the -3/2 thinning rule.  The third or transition stage is one of relaxed competition resulting from death of canopy trees.  The fourth stage is the steady-state or climax forest where gap processes result in a spatially heterogeneous forest.  The mortality patterns of  these various stages are summarized in Peet & Christensen (1987).  Among other community attributes that appear to track this developmental sequence and which can be interpreted in terms of intensity of competition (reviewed in Peet 1992)  are species diversity (Peet & Christensen 1988) and compositional predictability (Christensen & Peet 1984).

 

Tree growth and mortality.  Size and growth inequality have been proposed to corre­spond to the asymmetry of competition (see Weiner and Thomas 1986).  As part of an investigation of this topic we examined changes in size inequality during the initial phases of stand development (Knox, Peet, and Christensen 1989, Duncan 1995).  As predicted, we found inequality to increase early in stand devel­opment corresponding to asymmetric growth (where light is limiting).  Also as predicted, inequality declined with the onset of self-thinning. Subsequently, consideration of our four-stage model of forest development led us to predict that competition should be symmetric during establishment, then asymmetric during self-thinning, then less-asymmetric during the tran­sition phase, and finally, strongly asymmetric again during the steady-state phase; most of our predictions are borne out (Peet 1988).  In addition, we have found that our permanent plots show evidence of synchronous deviations from the expected patterns of asymmetry that likely correspond to climatic variation.  Such devia­tions might provide a valuable index of the exogenous stresses a forest is experiencing at a given time.  Of particular interest is an apparent dip in the asymmetry of competition that matches up with the canopy opening associated with Hurricane Hazel in 1954.

 

The importance of scale in observations of community pattern.  Variation in species composition of forest samples is often related to environment, but the dependence of the patterns observed on the scale of observation and the spatial extent of the observations is typically ignored.  Reed et al (1993) examined this question by mapping all the trees in a 6.55 ha plot and sampling soil and other environmental attributes at each grid corner.  Herb data were collected by Palmer at a series of different scales (Palmer & White 1994).  The tree map was used to calculate overall and species-specific canopy influence.  Predictability of species composition was found to increase with the grain size of the observations, and the factors most strongly correlated with composition changed with grain size.

 

Additional applications of the data.  Numerous additional projects have used the LTREB dataset.  Baker-Brosh (1996) determined the genotype of all loblolly pines in a series of 13 stands of varied initial loblolly density as part of a study of change in population heterozygosity with successional stage and intensity of competition.   DeCoster (1996) extended the LTREB tree mapping to include an area of forest damaged by a tornado and then used logistic regression to identify factors associated with tree damage. In this extreme wind event, many trees were broken and pines were disproportionately damaged. He found  similar results in US Forest Service records of damage from Hurricane Hugo in the South Carolina Piedmont. James Graves (1995, 1996) used herb data from the Duke Forest compositional survey plots to demonstrate the influence of herbaceous vegetation on tree canopy structure.  Several investigators have used our LTREB data to examine spatial pattern in forest trees and seedlings. Ribbens (1995) linked seedling establishment patterns with the distributions of canopy trees.  Robert Wolpert and associates in the Duke University Statistics Department are using our forest maps to develop new statistical tools for detecting spatial pattern (Ickstadt & Wolpert 1996).  Canopy architecture has been examined by a number of workers who have linked remote sensing data with our forest plots (e.g., Kasischke et al. 1994).

 

Contributions to development of human resources

The data collection and management activities associated with these projects were and are labor intensive.  Consequently, these projects provided numerous opportunities for undergraduates to gain hands-on experience with scientific research, plus many opportunities for graduate students to obtain broader experience with ecological research and to develop further their research ideas. 24 undergraduate were employed as summer field assistants under these grants. In addition, 4 graduate students were employed as field assistants or supervisors.  Two undergraduates and two graduate students were employed as data managers during the academic year.  Three undergraduate honors thesis projects were conducted as part of this. Four graduate students and two Postdoctoral associates worked extensively with data collected during these grants as part of their research projects.

 

Publications resulting from BSR-8905926 & BSR-9107357 (full citations in References).

Publications: Baker-Brosh 1997; Duncan 1995; Glenn-Lewin, Peet & Veblen 1992; Ickstadt &           Wolpert 1997; Kasischke, Christensen and Haney 1994; Knox, Peet & Christensen 1989;    Palmer & White 1994; Peet 1992; Peroni 1994; Reed, Peet, Palmer & White 1993.

Manuscripts (not yet accepted for publication): Graves, Peet & White 1996; Lopez-Mata & Peet       1996a, 1996b, 1996c; Muth & Peet 1996; Philippi, Peet & Christensen 1996.

Dissertations: Baker-Brosh 1996; DeCoster 1996; Graves 1995; Lopez-Mata 1994; Ribbens 1995

Honors Theses: McAdams 1992; Muth 1996, Reed 1991

Manuscripts in active preparation: (6, not fully itemized, but see Philippi & Peet 1994, Philippi,          Peet & Christensen 1992, 1993)

Abstracts: (approximately 15, not itemized)

 

 Results of Prior NSF Support:  Dean L. Urban

Urban was co-principal investigator with W. Lauenroth, H. Shugart, and others, on Coupling ecosystem processes and vegetation structure across environmental gradients (NSF grant BSR-9013888, $1.2M, 1991-1995). This project used simulation models as a framework for comparisons among grassland and forested LTER sites, focusing on the interface between community ecology (species diversity) and ecosystems (nutrient cycling, water relations). In addition to articles by other PI's, Urban produced eight articles under this grant (Urban et al. 1991, 1993; Urban and Shugart 1992; Weishampel et al. 1992; Coffin and Urban 1993; Hansen et al. 1993; Lauenroth et al. 1993; van Voris et al. 1993); five others are in prep. or in revision. This work is currently being extended under two new NSF grants. Environmental variability and forest pattern: a comparison of eastern and western landscapes (IBN-9652656, $399,684, 1996-1999) extends the cross-site comparison work to focus on the demographic mechanisms of gradient response, and thus complements the work proposed here for Duke Forest. A second new grant, Scaling forest ecosystem dynamics from trees to landscapes (BIR-9630606, $110,895, 1996-1999) uses a suite of simulation models to rigorously extrapolate forest dynamics from the small scales of tree-based information, to the larger scales at which resource management and impact assessment are addressed. The emphasis on long-term system dynamics in both new projects meshes nicely with the collection and analysis of long-term datasets as proposed here.

 

Proposed Research

 

Our proposed work consists of two related components: recensus of long-term permanent plots and improving the quality, documentation and accessibility of the associated data archive. The richness of our collection of plot data and the budgetary constraints imposed by the LTREB program will preclude resurvey of some of the plots that it would be desirable to include. In addition, the time required to locate and recensus plots that contain significant amounts of downed timber is extremely difficult to assess in advance.  Below we list the plots that we plan to recensus, and assign priorities to these varied tasks. Our intention is to begin with the highest-priority tasks and add as much additional work as logistics allow.

 

Past experience has shown us that the most time-efficient method for censusing mapped trees, saplings and seedings on moderately level terrain is by laying tapes along previously surveyed and staked grid lines.  This method has the added advantage that it minimizes the number of small, inconspicuous stems missed in the resurvey.  For trees that appear to have been damaged or killed by Hurricane Fran, we will assign codes for damage type and degree modeled after those developed by William Platt (pers. comm; see Slater et al. 1995; also see Everham& Brokaw 1996) for Gulf Coast hardwood forests and by our own group (DeCoster 1996) for tornado damage in Piedmont forests.  This will allow comparison of factors that influence the probability of tree damage in different storm types.

 

Vegetation Resurvey

 

Original Permanent Sample Plots (Priority 1)

We propose to maintain and resurvey the remaining 37 permanent sample plots dating from the period 1933-1947.  In these plots we will remap all stems >1 cm dbh to the nearest tenth meter and record the dbh and height of each stem.  Some of the more significant attributes of these plots are summarized in Table 1 (e.g. 10637 trees alive at last census).  We anticipate sampling these plots twice during the LTREB funding period (1997 and 2001), which will keep the plots on their current ~5-year recensus cycle.

 

Mapped forest stands (Priority 1)

The limited size of the original permanent plots precludes their use for intensive study of spatial pattern, gap processes, or the demography of any save the most common species.  To facilitate study of these critical aspects of forest dynamics we selected five forest areas in 1988 for intensive study.  These are all sites on which we mapped all trees >1 cm at some time during 1982-87, and again at least once during the 1989-94 funding cycle.  We also constructed a series of additional forest maps to provide a broader range of forest types and site conditions.  In total, we will recensus the 10 maps shown in Table 2.

 

The five intensive study plots include two old pine stands (70 & 80 years old) and three uneven-aged hardwood forests on contrasting soils.  The pine stands appear to be entering the transition phase of forest development which we found to be so critical in our studies of convergence (Christensen & Peet 1984).  Hurricane Fran caused moderate damage, removing 10-20% of the canopy.  One hardwood stand is on a moist, semi-alluvial site on which gap-formation is relatively frequent (Scoville 1981), but damage from Fran was limited.  A second hardwood forest occupies a typical upland mixed oak forest, was first mapped by Dr. F.H. Bormann in 1950, and is described in detail by Christensen (1977).  The final hardwood site occurs on a dry upland site with soils derived from rock rich in calcium and magnesium.  Both upland hardwood plots were significantly impacted by Hurricane Fran (canopy loss of ca 75% and 35%, respectively, which is representative of damage within large areas of the Duke Forest region).

 

All trees in each intensive study site will be resurveyed twice, once at the beginning and once near the end of the funding period.  During each resurvey, all stems >1 cm dbh will be mapped and recorded by species and diameter, and deaths (or missing trees) will be recorded. We propose to recensus the five additional mapped forests once  (Table 2) so as to facilitate comparisons across a broad range of forest types and damage levels.  The 10 mapped forests cover 233786 m2 and contained 37702 trees at last census.

    

Seeding and Sapling Demography Plots (Priority 1)

We maintained intensive, annually censused seedling (individuals of potentially arborescent species <1m tall) demography plots in four of the five intensive mapped-forest areas from 1978-1994, and in the final one from 1989-1994 (Table 1).  In addition, we maintained intensive, annually censused sapling (1m tall to 1.0 cm dbh) plots superimposed on the seedling plots.  We propose to resample these plots annually for the next five years to assess the impact of canopy damage on seedling dynamics. At last census in 1994, these 5 plots contained a total of 7848 extant mapped seedlings and 4996 extant mapped saplings (Table 2).

.

Compositional Survey Plots (Priority 2)

In 1977 Peet and Christensen established a series of approximately 230 0.1 ha plots distributed across all the major forest types and ages within the Duke Forest.  These plots included detailed data on both trees and herbaceous species as well as soil chemistry.  Data from these plots have subsequently been used for many purposes including studies of forest composition (1980), species diversity (1987), and compositional convergence during succession (1984).  Permanent plots with detailed herb and soil data are extremely uncommon in North America, yet herbs are well known to be especially sensitive to the changing environment and competitive status of forest stands.  We propose to relocate and georeference (GPS) as many of these plots as possible.  Our preliminary assessment suggests that ~25% of the plots have been destroyed and another 20% may prove impossible to relocate.  Nonetheless, relocation of roughly 130 plots with detailed herb and soil data would allow assessment of forest change at a resolution nearly unprecedented in North America.  Our ultimate goal is to resurvey those plots remaining from 1977, and to augment these with new plots stratified over Duke Forest as a whole.  Time and budget are unlikely to allow completion of this task, but we will resurvey as many as possible, deferring the remainder for a subsequent proposal.

 

North Carolina Botanical Garden Plots (Priority 2)

During the early 1990s, a series of 62 0.1 ha plots were distributed in a stratified-random fashion across the lands of the North Carolina Botanical Garden, roughly 8 miles from the Duke Forest.  The plots include both trees and herbs and were sampled using a modified version of the North Carolina Vegetation Survey protocol (Peet et al 1996).  We propose to resurvey these plots, manyof which were damaged by Hurricane Fran.  Since these plots were originally located in a stratified random fashion, they will be ideally suited for assessing the amount of damage and average herbaceous plant response.

 

Data Management and Archiving

 

To be of maximal value to the ecological community, long-term data should be documented and made available for public use. Considerable data on forest succession has accumulated as a consequence of past activities of the Duke Forest and previous LTREB grants. Our data are sufficiently unusual and valuable that even without such public access, we receive (and honor) regular requests for use of our long-term data. We propose to develop a publicly accessible web site that documents the available data and provides access to all data within a few years of collection. This activity can be expected to consume roughly 1/3 of the time of the graduate student assistant, plus much of the time contributions of Urban and Halpin.

We propose to archive databases as plain text (ASCII) files with accompanying descriptive abstracts (metadata), both of which will be accessible via standard network vehicles (anonymous ftp, web browsers). The databases will reside on the server maintained in the Landscape Ecology Laboratory at Duke (http://www.env.duke.edu/lel-acl/lel.html); these archives will be mirrored at UNC-Chapel Hill in the Biology Department under Peet's supervision.

 

Spatial databases will be maintained in GIS format as Arc/Info export files for ease of transfer, and documented according to FGDC standards. As spatial data, some of these files will comprise little more than the location of the plots, yet in this format the permanent plots can be overlaid readily with other geographic data maintained for the Forest and the surrounding area. Currently these data include digital terrain data, hydrography, roads, built structures, and soils (SCS Soil Survey data). In addition, we are currently digitizing stand maps (dominant species by age class) for all management compartments in Duke Forest (revised every 10 years since 1934). Finally, Duke Forest is a NASA SuperSite and we maintain a wealth of remotely-sensed data ranging from digital orthophotos (1-m resolution) to Landsat TM (30 m) and  MSS (80 m), to AVHRR imagery (1 km). The georeferencing of all Duke Forest data allows us to reconcile these different data within a common spatial framework so we can address questions such as whether Hurricane Fran affected stands differentially with respect to soil type or topographic position, or to pinpoint permanent sample plots that have been damaged as evidenced in airphotos acquired within weeks after the storm (these are currently being scanned). The georeferencing of permanent sample plots also allows us to use the plots as ground truth for various satellite imagery.

 

Analysis and Applications

 

Long-term datasets are scarce, yet they are essential for answering many ecological questions.  Analysis of our Duke Forest permanent plot data has provided numerous insights into the patterns and mechanisms of forest succession and regeneration.  Inevitably, as many new questions have arisen from our work as old questions have been answered.  Below we discuss some of our results and some of the additional research questions these results have led us to ask.  This discussion is not intended to be a complete or exhaustive survey of applications of the data we propose to collect.  Moreover, It would be presumptuous to suggest that we can anticipate the major applications that our data may find 20 or more years from now.  The foresters who in the 1930s established many of the permanent plots we maintain certainly did not anticipate our interest in asymmetry of competition, the processes that characterize the transition from even-aged to all-aged forests, tests of the -3/2 thinning law, or the impact of increased nitrogen in rainfall on tree growth.

 

Tree population processes & trajectories of forest change

 

Variation in tree growth and mortality.   Information on how tree growth and mortality rates vary with species, size, site, and competitive environment are critical for developing better forest simulation models and for simple projection of changes in stand composition.  Our preliminary analysis of Duke Forest data shows that loss rates for hardwoods in uneven-aged stands are generally constant through time with the slope being dependent on the species and the site; canopy species generally have lower slopes than obligate understory species. We will obtain for the major hardwood species estimates of growth and mortality rates as they vary with site conditions and tree size.  In particular, we will examine whether mortality rates are relatively variable across sites, species and sizes, but relatively constant within species, size classes, and sites.

 

Trajectories of compositional change.  The supposedly mature, all-aged deciduous forests of the North Carolina Piedmont are changing in composition.  For example, red maple and beech have increased in importance over the last 20 years far more than anyone predicted.  Oldfield pine stands are now succeeding  to hardwood stands, but with less oak and hickory than the classic and widely accepted literature would lead one to believe.  This basic information on successional trajectories is critical to future studies of the successional process in this model system. We will use the Duke Forest permanent plot data to reassess compositional trajectories of both oldfield pine forests and older hardwood forests of the North Carolina Piedmont.

 

Spatial pattern changes.   We, like other investigators, have shown that mortality rates are high for small trees and for trees located near other trees (Daniels 1976, Lorimer 1983, Peet and Christensen 1987, Knox et al. 1989).  Our work as well as other studies suggests that seedling/sapling mortality is maximal during the thinning-phase of stand development and low during the establishment and transition phases (Lorimer 1983, Cooper 1961, Christensen 1977, Peet & Christensen 1987, Whipple 1980), which leads to an expectation that spatial pattern should proceed from contagious through random to more uniform as trees increase in size. However, we have not yet demonstrated how these patterns change with succession or how mortality varies spatially in a mixed-species forest.  At present we have little permanent plot data available to assess how mortality patterns change during the transition phase, but within a decade the pine stands should have aged sufficiently to provide the critical information.  Reasonable predictions include a generally lower rate of small tree mortality during transition owing to reduced competition, or increased spatial heterogeneity in mortality of small trees during transition with high survival associated with gaps.  We see both predictions as realistic and expect that they represent extremes associated with extremes in initial density (and hence synchrony of death) of the initial trees. We will test whether spatial patterns of trees become progressively more regular during the thinning phase of stand development, and then become more aggregated during the transition phase.

 

Cross-site Comparisons.  In forest ecology we are getting to the point where we know a lot about a few study sites, but without much knowledge of the generality of the results obtained.  On the one hand, we need to know if the results are general across a range of sites conditions within a region, a question for which our diverse permanent plots are well suited. On the other hand, more cross-site comparisons are needed if the broad geographic generality of  the results is to be ascertained. We have already conducted preliminary comparisons of mortality patterns between sites and anticipate expanding this effort (Peet, Harcombe and Parker 1991).  We will compare overall and species specific growth and mortality results between sites in a region and with the results of other workers working within the eastern deciduous forest region.

 

Implementation of forest simulation models for ancient igneous soils.  Although forest gap models (Botkin et al. 1972) have been implemented for a broad range of forest conditions in eastern North America and elsewhere, they have not been parameterized for the North Carolina Piedmont.  This offers a particularly interesting challenge in that forest composition in Piedmont and Southern Appalachian Mountain sites is strongly linked with soil chemistry and mineralogy (Peet & Christensen 1980, Peet & Newell 1995; Newell et al 1996).  In addition, seedlings have generally been treated as a black box in such models owing to the lack of detailed information of the type we have been collecting.  These sites offer a valuable opportunity to implement a gap model with the advantage of extremely rich databases for parameterization and testing. Such a model will be especially useful for exploring hypotheses about the role of soils and demographic mechanisms, and for testing these hypotheses against long-term datasets.  We will use Duke Forest permanent plot data to parameterize forest simulation models with particular emphasis on tree seedling dynamics and tree response to soil cation availability.  This work will be a natural extension of the new NSF grant to one of us (DLU), which is concerned with cross-site comparisons among forests.

 

Impacts of large, infrequent disturbances

 

Hotshots and hotspots.  When we initiated study of seedling demography, we anticipated that there would be microsites characterized by more rapid seedling growth and greater survival.  We also anticipated that we would be able to identify individuals that were consistently fast growers and thus more likely to achieve sapling status.  One of the major surprises of our seedling demography work was that the individuals that grew most rapidly in one year were no more likely to grow rapidly the next year than the average seedling.  In fact, there was generally a negative correlation between growth in successive years.  These results, while based on 17 years of data and 95,000 individuals, were obtained in relatively constant and homogeneous forests.  Hurricane Fran has changed all this.  We anticipate that the large gaps created by Fran will result in release of established seedlings and saplings.  Although the loss of canopy trees may well create a microclimate that is detrimental to many of the smaller seedlings, older seedlings and saplings have larger root systems and thus can be expected to have an ample supply of soil resources and thrive in the open conditions.  We will test the hypothesis that large, established seedlings experience a sustained increase in growth rate in those portions of stands with significant hurricane damage.

 

Trajectories of stand density.  Over the history of the Duke permanent plot studies, sapling and small tree density has increased substantially.  This could represent a change induced by a change in stand structure as once disturbed forests have become less even-aged and have more heterogeneous canopies with a gap structure favorable for seedling and sapling growth.  Alternatively, the increase in density could be a consequence of  loss of  low-intensity fire and woodland grazing that may have kept seedling and sapling populations low during the previous century. Continued monitoring of the Duke Forest permanent plots as they recover from mortality caused by Fran will allow assessment of whether sapling and small tree density decrease again as canopy structure becomes more even-aged similar to that of early in the century, or stays high owing to continued absence of low-intensity fire and woodland grazing.

 

Oak and hickory regeneration.  Although various oaks (ca 15 species) and hickories (ca 5 species) dominate the canopy of much of the Piedmont including several of our mapped plots, they are underrepresented in the seedling and sapling layers relative to their canopy dominance (a situation true throughout much of eastern North America).  Our demographic work has shown that while oak and hickory seedlings and saplings do become established and can eventually achieve canopy status, there has been a steady decline in the dominance of these genera over the past 60 years, along with a simultaneous increase in abundance of red maple and beech.  One popular hypothesis is that these genera are well adapted to chronic, low-intensity fires that ceased during the 1800s (see Abrams 1992).  However, another hypothesis that has received consistent support is that  these relatively shade-intolerant species are adapted to rapid growth following major canopy disturbances such as those associated with hurricanes and tornados (see Glitzenstein et al 1986). Continued monitoring will allow us to observe whether oaks and hickories regain lost canopy dominance following hurricane damage, or whether removal of canopy oaks will simply serve to hasten the rate of replacement of these traditionally dominant genera with the more shade-tolerant maples and beeches that currently dominate the seedling and sapling layers.

 

Predictability of composition.  In 1984, Christensen and Peet used the Duke Forest survey plot data to show that the extent to which forest composition (primarily herbaceous composition) can be predicted from site variables increases dramatically during the self-thinning phase of forest development, crashes during the transition phase, and increases again (but to a lesser peak) in the mature, uneven-aged hardwood stage.  They attributed this pattern to changing intensity of competition, but admitted that the transitional phase shift might reflect sorting relative to light rather than soil attributes.  If Christensen and Peet are correct that the predictability of species distributions tracks the intensity of competition from canopy trees, then there should be a major relaxation in the predictability of species distributions for several years following the major canopy disturbances from Hurricane Fran. We will test the hypothesis that species distributions will become less predictable for several years following major canopy disturbances associated with Hurricane Fran, while remaining relatively constant in the forests with undisturbed canopies.

 

Diversity of seedlings & saplings following disturbance.  The tree mortality in the Duke Forest  attributable to Hurricane Hazel in 1954 was, like that attributable to Fran, very patchy.  In one case a 0.1 ha permanent plot lost approximately half its basal area, whereas an equivalent and adjacent plot had very little damage.  Although these plots were very similar prior to Hazel, significant differences were apparent in the following years.  Perhaps most striking was that the damaged plot had a flush of seedling and sapling establishment, and by 1977, 23 years after the storm, there were twice as many sapling species established as in the control.  This observation is consistent not only with our hypothesized relaxation of competition, but also the hypothesis that blowdowns contribute significantly to tree species diversity in Piedmont forests. We will observe changes in seedling and sapling composition of permanent plots to test the hypothesis that seedling and sapling diversity increases following significant canopy damage.  We will also determine whether this anticipated increase in sapling diversity is attributable to differences in seedling establishment (and first year survival), or to reduced mortality and/or increased growth of established individuals.

 

 

Changes in the asymmetry of competition following canopy disturbance.  Numerous permanent plots with even-aged loblolly pines showed a shift toward greater symmetry of competition immediately following Hurricane Hazel (Peet 1988), presumably because of a reduction in competition for light.  However, this change could also be a consequence of climatic variation (smaller but synchronous changes in symmetry were observed in other years). We hypothesize that competitive symmetry of canopy trees will increase within damaged plots in the years immediately following Hurricane Fran.

 

Biomass and production.  We used permanent plot records in conjunction with dimension analysis equations we developed for the trees in Duke Forest (Peet and Council 1981) to determine changes in biomass and production over the course of stand development (Peet 1981).  With the exception of Jurik et al. (1988), virtually all other attempts to study changes in biomass and production during succession have relied on single samples of different-aged stands.  A particularly interesting aspect of our biomass-production studies was that net aboveground production was remarkably constant, presumably reflecting constant resource supply rates, whereas biomass steadily increased, presumably reflecting recovery from selective cutting or low-intensity ground fires of the past century.  Although biomass decreased following Hurricane Hazel in 1954, aboveground production did not change substantially.  This suggests that nutrient supply rates are sufficiently limiting in these ancient soils that even with substantial tree mortality, available nutrients are absorbed by the remaining individuals.  We will use our permanent plots to assess changes in biomass and aboveground production during the recovery from Hurricane Fran.  We hypothesize that this storm, which caused substantially greater canopy damage than Hazel, will cause a drop in aboveground production.  We further hypothesize that the drop will be followed by a brief peak in production as the excess resources are consumed by the recovering trees, much as predicted by Peet (1981) and Sprugel (1985).

 

Consistency between storms in mortality risk factors.  Research by our group on mortality and damage caused by Hurricane Hugo on the South Carolina Piedmont in 1989 and a 1988 Tornado near the Duke Forest showed breakage to be a major form of damage, and tree species and height to strongly influence probability of death (DeCoster 1996).  Preliminary evaluation of damage by Fran to the Duke Forest suggests fewer pines and more oaks to have been damaged, and most damage to have been caused up uprooting rather than breakage. We will compare factors that influenced mortality that resulted from Fran with the factors we have previously found to be important for mortality during Hugo and the 1988 Tornado.  We will also, for the first time, be able to include information on the growth and competitive history of individuals trees within a stand in evaluating tree mortality factors.

 

Field Work Schedule

 

1997                Resurvey 37 original permanent sample plots

Resurvey 5 intensive mapped stands for trees, seedlings and saplings

 

1998                Recensus 5 remaining mapped plots

Resurvey 5 intensive mapped stands for seedlings and saplings

Resurvey Compositional Survey and NCBG plots (as time allows)

 

1999                Resurvey 5 intensive mapped stands for seedlings and saplings

Resurvey Compositional Survey and NCBG plots (as time allows)

 

2000                Resurvey 5 intensive mapped stands for seedlings and saplings

 

2001                Resurvey 37 original permanent sample plots

Resurvey 5 intensive mapped stands for trees, seedlings and saplings

 

Responsibilities

 

Peet will be responsible for overall coordination and supervision of the plot data collection and data management.  The graduate student will be responsible for the day-to-day data collection and management activities.  This will include participation in and direct supervision of the undergraduate field crew.  White will supervise data collection for the North Carolina Botanical Garden plots.  Urban and Halpin will be responsible for georeferencing the field sites, compiling metadata and archiving the associated data in the Duke Forest files.  They will develop appropriate ftp and www sites for data access, and for providing long-term facilities for data archiving in the Nicholas School of the Environment at Duke.  Undergraduates and/or graduate student interns supported by both the UNC and Duke budgets will participate for 12 weeks on the summer field crew and assist in data archiving activities.

 

Table 1.  Long-term Permanent Sample Plots to be recensused

 

Plot      Census     Census    Initial      Size           Initial      Remain     Current 

#             dates      number     age         m2             trees         trees         trees

 

A. Successional Pine Plots

 

D4         1933-92           11         9          1012            323              35            258           

D5                     A                                                         368              40            239           

D6                     A                                                         207              46            330           

D7                     A                                                         217              39            249           

D12       1933-92           11            8          405              25              16              98           

D13                   A                                                           50              14            171           

D14                   A                                                           51              17            145           

D15                   A                                                           79              19            161           

D16                   A                                                           81              17            152           

D17                   A                                                         149              19            168

D18                   A                                                         186              15            147           

D19                   A                                                         236              16            132           

D20                   A                                                         261              17            237           

D21                   A                                                         431              27            158           

D22                   A                                                         511              15            136           

D23                   A                                                       1172              10            242           

D24       1934-92           11        19          1012            324              66            319           

D25                   A                                                         125              38            328           

D26                   A                                                         331              77            264           

D28       1934-92           11        15            810            464              35            363           

D29                   A                                                         227              26            271           

D39       1934-92             9        15            810            197              29            226           

D40                   A                                                         475              43            196           

D41                   A                                                         431              25            260           


D42                   A                                                         272              44            287           

D49       1936-92           10        30          1012            292              56            191           

D50                   A                                                         311              56            185           

D51                   A                                                         127              38            288           

 

Uneven-aged Hardwood Plots

 

D10       1933-92           10       mix          1012            319              62            229           

D35       1933-92             8       mix          1012            186              68            388           

D36                   A                                                         145              41            376           

D37                   A                                                           96              22            470           

D43       1935-92             9       mix          1012            230              48            171           

D44                   A                                                         237              44            181           

H23        1947-92           10       mix          4047            419            140          1037           

H24                    A                                                         448            160          1248

H25                    A                                                         336            136                 .           

 

 In many cases we have increased the size of the plot since 1977.  In the table we show the original plot size, how many trees were present in the original census of each plot, how many of the original trees were still alive at last census (ca 1992), and how many total trees (ingrowth included) were alive at the most recent census.

 

Table 2.   Mapped forest stands

 

Plot     Forest Type                    Years      Area  Obs Trees  Trees  Seed   Seed   Sapls  Sapls

#                                                                    (m2)     #    Obs    Alive    Obs    Alive    Obs    Alive

 

Intensively sampled plots with seedling and sampling data

M13     Upland Hdwds  1950-93       19600          5 7859  3883    3863    1030      948      621

M12     Upland Hdwds 1978-90       20400          4 6089  5103    8140    1293    1312      674

M06     90 yr Loblolly                    1978-93    9750    4    3796    2620  29125    1865    2640    1524

M04     70 yr Loblolly                    1978-92  13000    4    5529    3483  27215    3254    1317      835

M94     Alluvial Hdwds  1986-93       23750          3 7662  6725      598      406    2361    1342

 

Mapped forests without recent seedling and sampling data.

M91     Mesic Hdwds/Pine             1984-91    5250    3      911      809           

M92     Chestnut Oak 1984-92         5000          3 769      670           

M93     Upland Hdwds/Pine           1986-91  15000    2    4664    3909           

M95     Longleaf Pine                   1985-93  56500    3    2012    1670           

M96     Dry-mesic Hdws                     1991  65536    1    8830    8830

 

Mapped plots have all tree stems $ 1 cm dbh located to the nearest 10 cm. The duration of the mapping (Years) varies between plots, as does the number of times the trees in the plots have been remeasured (obs #).  The area mapped refers to the most recent remapping; some of the plots have been increased in size since monitoring was initiated.  The number of tree-size individuals observed at time during the monitoring (Trees obs), as well as the number alive at the most recent observation (Trees alive) are indicated.  For the five intensively sampled plots seedlings (0-1 m tall) and sapliungs  (>1m tall and < 1 cm dbh) have been monitored through 1994.

 

 

References

 

Abrams, M.D. 1992 Fire and development of oak forests.  BioScience 42:346-353.

 

Baker-Brosh, K.F. 1996.  The genetic consequences of self-thinning in two populations of loblolly pine, Pinus taeda L.  Ph.D. Dissertation, University of North Carolina at Chapel Hill.

 

Baker-Brosh, K.F. & R.K. Peet. 1997.  The ecological significance of lobed and toothed leaves in temperate forest trees.  Ecology (in press).

 

Billings, W.D. 1938.  The structure and development of old field shortleaf pine stands and certain associated physical properties of the soil. Ecol. Monogr. 8:437-499.

 

Bormann, F.H. 1950. The statistical efficiency of plot size and shape in forest ecology. Masters Thesis, Duke University.

 

Bormann, F.H. 1953.  Factors determining the role of loblolly pine and sweetgum in early old-field succession in the Piedmont of North Carolina. Ecol. Monogr. 23:339-358.

 

Bormann, F.H. and G.E. Likens. 1979.  Pattern and process in a forested ecosystem.  Springer-Verlag.

 

Botkin, D.B., J.F. Janak, and J.R. Wallis. 1972. Some ecological consequences of a computer model of forest growth. J. Ecology 60:849‑873.

 

Christensen, N.L.  1977 Changes in structure, pattern and diversity associated with climax forest maturation in Piedmont, North Carolina.  Am. Midl. Nat. 97:178-188.

 

Christensen, N.L. and R.K. Peet 1981. Secondary forest succession on the North Carolina Piedmont.  In D.C. West, H.H. Shugart & D.B. Botkin (eds.) Forest Succession: concepts and application.  Springer-Verlag. Pages 230-245.

 

Christensen, N.L. and R.K. Peet 1984.  Convergence during secondary forest succession. J. Ecol. 72:25-36.

 

Coffin, D.P., and D.L. Urban. 1993. Implications of natural‑history traits for ecosystem dynamics: comparison of a grassland and forest. Ecol. Model. 67:147‑178.

 

Cooper, 1961.  Patterns in ponderosa pine forests. Ecology 42:493-499.

 

Daniels, R.F. 1976. Simple competition indices and their correlation with

annual loblolly pine tree growth.  Forest Science 22:454-456.

 

DeCoster, J.K. 1996. Impacts of tornados and hurricanes on the community structure and dynamics of North and South Carolina forests. Ph.D. Dissertation, University of North Carolina at Chapel Hill.

 

Duncan, R.P. 1995.  A correction for including competitive asymmetry in measures of local interference in plant populations.  Oecologia 103:393-396.

 

Everham, E.M. & N.V.L. Brokaw. 1996.  Forest damage and recovery from catastrophic wind.  Bot. Rev. (In press)

 

Glenn-Lewinm D.C., R.K. Peet and T.T. Veblen (eds.). 1992.  Plant succession: Theory and prediction.  Chapman and Hall.

 

Glitzenstein, J.F., Harcombe, P.A. & D.R. Streng. 1986.  Disturbance, succession, and maintenance of species diversity in an East Texas forest. Ecol. Monogr. 56:243-258.

 

Graves, J.H. 1995.  Resource availability and the importance of herbs in forest dynamics.  Ph.D. Dissertation, University of North Carolina at Chapel Hill.

 

Graves, J.H., R.K. Peet and P.S. White.  1996.  Resource availability and the trade-off in abundance of herbs and woody plants in temperate deciduous forests.  Manuscript.

 

Hansen, A.J., S.L. Garman, B. Marks, and D.L. Urban. 1993. An approach for managing vertebrate diversity across multiple‑use landscapes. Ecol. Applic. 3:481‑496.

 

Ickstadt, K. & R.L. Wolpert. 1997.  Multiresolution assessment of forest inhomogeneity.  In C. Gatsonis, J.S. Hodges, R.E. Kass & N.D. Singpurwalla (eds.) Case Studies in Baysian Statistics, Vol 3.  Lecture Notes in Statistics. Springer-Verlag (in press).

 

Jurik, T.W.  1988.  Aboveground biomass and production from 1938 to 1984 for

four aspen plots in northern Lower Michigan.  Michigan Botanist  27:43-54.

 

Kasischke, E.S., N.L. Christensen and E.M. Haney. 1994.  Modeling of geometric properties of loblolly pine tree and stand characteristics for use in radar backscatter studies. IEEE Trans. Geo. and Remote Sens. 32:800-822.

 

Keever, C. 1950.  Causes of succession on old fields of the Piedmont, North Carolina. Ecol. Monogr. 20:229-250.

 

Knox, R.G., R.K. Peet & N.L. Christensen 1989. Population dynamics in loblolly pine stands: changes in skewness and size inequality. Ecology 70:1153-1166.

 

Lauenroth, W.K., D.L. Urban, D.P. Coffin, W.J. Parton, H.H. Shugart, T.B. Kirchner, and T.M. Smith. 1993. Modeling vegetation structure‑ecosystem process interactions across sites and biomes. Ecol. Model. 67:49‑80.

 

Lopez-Mata, L. 1994.  Coexistence of Quercus and Carya in natural upland hardwood forests of the North Carolina Piedmont.  Ph.D. Dissertation. University of North Carolina at Chapel Hill.

 

Lopez-Mata, L. & R.K. Peet. 1996a.  Spatial dispersion of sapling oaks and hickories in upland hardwood stands of the North Carolina Piedmont, USA.  Submitted.

 

Lopez-Mata, L. & R.K. Peet. 1996b.  Regeneration niche and coexistence of tree species in Quercus-Carya forests of the North Carolina Piedmont, USA.  Manuscript.

 

Lopez-Mata, L. & R.K. Peet. 1996c.  Sapling growth strategies in response to canopy openness in a North Carolina Quercus-Carya forest.  Manuscript.

 

Lorimer, C.G. 1983.  Tests of age-independent competition indices for individual trees in natural hardwood stands.  For. Ecol Mgt. 6:343-360.

 

McAdams, A.G. 1992.  Characterization of the light environment of Piedmont forests and the effect of light on seedling growth.  B.S. Thesis. University of North Carolina at Chapel Hill.

 

Muth, C.C. 1996.  A comparison of understory tolerance measures for woody plants.  B.S. Thesis, University of North Carolina at Chapel Hill.

 

Muth, C.C. and R.K. Peet. 1996.  A comparison of understory tolerance measures for woody plants.  Manuscript.

 

Newell, C.L. And R.K. Peet 1995.  Vegetation of Linville Gorge Wilderness, North Carolina.  USFS, National Forests in North Carolina. 211 pp.

 

Newell, C.L. R.K. Peet, C.J. Ulrey, T.R. Wentworth, K.D. Patterson and D.E. McLeod. 1997.  Geographic variation in forest distribution across five landscapes in the Southern Appalachian Mountains of North and South Carolina. In Biogeography of the Southern Appalachian Mountains. In press.

 

Oliver, C.D. 1981.  Forest development in North America following major disturbances. For. Ecol. And Management 3:153-168.

 

Oosting, H.J. 1942. An ecological analysis of the plant communities of Piedmont, North Carolina.  Am. Midl. Nat. 28:1-126.

 

Palmer, M.W. & P.S. White. 1994.  Scale dependence and the species-area relationship. Amer. Nat. 144:717-740.

 

Peet, R.K. 1981. Changes in biomass and production during secondary forest succession.  In D.C. West, H.H. Shugart & D.B. Botkin (eds.) Forest Succession: concepts and application.  Springer-Verlag. Pages 324-338.

 

Peet, R.K. 1992.  Community structure and ecosystem properties. In D.C. Glenn-Lewin, R.K. Peet and T.T. Veblen (eds.). Plant succession: Theory and prediction. Chapman and Hall, London.  Pages 102-151.

 

Peet, R.K. and N.L. Christensen. 1980a.  Hardwood forest vegetation of the North Carolina Piedmont. Veröff. Geobot. Inst. ETH, Stiftung Rübel, Zürich 69:14-39

 

Peet, R.K. and N.L. Christensen 1980b.  Succession: a population process. Vegetatio 43:131-140.

 

Peet, R.K. and N.L. Christensen. 1987. Competition and tree death. BioScience 37:586-595.

Peet, R.K. and N.L. Christensen. 1988a.  Changes in species diversity during secondary forest succession on the North Carolina Piedmont.  In H.J. During, M.J.A. Werger and J. Willems (eds.). Diversity and pattern in plant communities. SPB Publishers pp 233-245.

 

Peet, R.K. and N.L. Christensen. 1988b.  Changes in symmetry of competition during development of even-aged Pinus taeda stands.  Bull. Ecol. Soc. Amer. 69:257-258

 

Peet, R.K. and O.P. Council. 1981.  Rates of biomass accumulation in North Carolina Piedmont forests. North Carolina Energy Institute, Research Triangle Park, NC.

 

Peet, R.K. , P.A. Harcombe and G.R. Parker. 1991.  Rates and patterns of mortality in eastern deciduous forests: a comparative study.  Bull. Ecol. Soc. Amer.  72:217.

 

Peet, R.K., T.R. Wentworth, R.P. Duncan and P.S. White. 1996. The North Carolina Vegetation Survey Protocol: A flexible, multipurpose method for recording vegetation composition and structure. Department of Biology, University of North Carolina at Chapel Hill. 40 pp.

 

Peroni, P.A. 1994.  Invasion of red maple (Acer rubrum L.) During old field succession in the North Carolina Piedmont: age structure of red maple in young pine stands.  Bull. Torrey Bot. Club 121:357-359.

 

Philippi, T.E. & R.K. Peet 1994.  A model of optimal life-histories for tree seedlings: allocation to growth vs belowground reserves. Bull.  Ecol. Soc. Amer. 75(Suppl):180.

 

Philippi, T.E., R.K. Peet & N.L. Christensen. 1992.  Survivorship and growth of Acer rubrum seedlings in stands representing different successional stages from old-field Pinus taeda to mature mixed hardwoods. Bull. Ecol. Soc. Amer. 73(Suppl.): 304-305.

 

Philippi, T.E., R.K. Peet & N.L. Christensen. 1993.  Tree seedling demography in old-field Pinus taeda and mature mixed hardwoods stands in a Piedmont forest.  Bull. Ecol. Soc. Amer. 74(Suppl.):393.

 

Philippi, T.E., R.K. Peet & N.L. Christensen. 1996.  Demography of tree seedlings in a North Carolina Piedmont forest.  Manuscript.

 

Reed, R.A. 1991.  Scale dependence of vegetation-environment correlations.  B.S. Thesis. University of North Carolina at Chapel Hill.

 

Reed, R.A., R.K. Peet, M.W. Palmer & P.S. White. 1993.  Scale dependence of vegetation-environment correlations in a Piedmont woodland, North Carolina, USA.  J. Veg. Sci. 4:329-340.

 

Ribbens, E. 1995.  Seedling recruitment in forests.  Ph.D. Dissertation, University of Connecticut.

 

Scoville, D.A. 1981.  Herb and vine vegetation of single tree windthrow gaps in lowland forests of North Carolina.  Masters Thesis, University of North Carolina at Chapel Hill.

 

Slater, .H., W.J. Platt, D.B. Baker & H.A. Johnson. 1995.  Effects of Hurricane Andrew on damage and mortality of trees in subtropical hardwood hammocks on Long Pine Key, Everglades National Park, Florida, USA. J. Coastal Research, Special Issue 21:197-207.

 

Sprugel, D.G.  1985.  Natural disturbance and ecosystem energetics.  In S.T.A. Pickett and P.S. White (eds.).  The ecology of natural disturbance and patch dynamics. Academic Press. Pp. 335-352.

 

Urban, D.L. Using gap models to assess forest response to climatic change: a stocktaking. Ecol. Applic. (in revision).

 

Urban, D.L., G.B. Bonan, T.M. Smith, and H.H. Shugart. 1991. Spatial applications of gap models. For. Ecol. and Manage. 42:95‑110.

 

Urban, D.L., M.E. Harmon, and C.B. Halpern. 1993. Potential response of Pacific Northwestern forests to climatic change: effects of stand age and initial composition. Climatic Change 23:247‑266.

 

Urban, D.L., and H.H. Shugart. 1992. Individual‑based models of forest succession. Pages 249‑292 in D.C. Glenn‑Lewin, R.K. Peet, and T.T. Veblen (eds.), Plant succession: theory and prediction. Chapman and Hall, London.

 

van Voris, P., W.D. Millard, J. Thomas, and D. Urban. 1993. TERRA‑Vision‑‑the integration of scientific analysis into the decision‑making process. Int. J. Geographic Information Systems 7:143‑164.

 

Weiner, J. & S.C. Thomas. 1986.  Size variability and competition in plant monocultures.  Oikos 47:211-222.

 

Weishampel, J.F., D.L. Urban, H.H. Shugart, and J.B. Smith, Jr. 1992. Semivariograms from a forest transect gap model compared with remotely sensed data. J. Veg. Science 3:521‑526.

 

Whipple, S.A. 1980.  Population dispersion patterns of trees in a southern Louisiana hardwood forest.  Bull. Torrey Bot. Club 107:71-76.