Temperate forests, such as those that dominate the Eastern US, are critically important ecosystems at regional to global scales. These forests harbor a high diversity of species and habitats, provide essential resources such as water, air, and wood products, offer important amenities to some of the most densely populated parts of the globe, and comprise key elements in global carbon budgets and atmospheric trace gas concentrations. Understanding the structure, function, and patterns of temperate forests and anticipating their future responses to natural disturbance, environmental variation, and human-imposed disturbance and stress are therefore critical societal goals that are dependent on a sound understanding of fundamental ecological processes. Such insight can only be gained through long-term study that integrates inter-disciplinary approaches through field studies, measurements, experiments, and modeling over a range of temporal and spatial scales.

The Harvard Forest LTER program seeks to interpret ecological pattern and process in New England forests that have been and will continue to be highly dynamic as a consequence of disturbance and environmental change and to apply this knowledge to regional and global issues in forest management, conservation, land restoration and protection, public health and policy, and the environment. Over the past decade, HF LTER has matured into a program that applies unique approaches in historical and community ecology, ecophysiology, atmospheric chemistry, and ecosystem studies to the interpretation of a suite of long-term, large-scale experiments and measurements, mechanistic studies, and retrospective research at site, regional, and global scales. In LTER I, we used primarily site-based annual and static measurements to interpret current conditions and to evaluate the comparative response of forest ecosystems to natural disturbance (hurricanes) versus anthropogenic stress (N deposition and soil warming; Fig. 13). Long-term measurements and insights from these initial results enabled LTER II to associate inter-annual variation in processes and environment, to examine direct land-use impacts at site to sub-regional (Central Massachusetts) scales, and to incorporate an understanding of historical landscape transformations into interpretations of modern pattern and process.

In LTER III we are poised to make significant advances in our understanding of the interactions among disturbance, environmental change, and ecosystem pattern and process at local to regional and even global scales. In particular, we propose to (1) extend many measurements, modeling activities, and historical studies to regional scales, (2) interpret landscape development, vegetation and wildlife dynamics, and current patterns in relationship to millennial-scale climate change, land-use history and natural disturbance, (3) evaluate ecosystem response to critical current disturbances and stresses (e.g. forest cutting, forest conversion, ozone and N deposition, and biological invasions by insect pests and exotic plants), (4) interpret long-term measurements and responses to experimental treatments mechanistically and in relationship to inter-annual and inter-decadal environmental variation and landscape history, and (5) apply this information to understanding the current and projected role of this vast forest region in global carbon budget storage. In these efforts we will (1) continue our long-term experiments and measurements, (2) strengthen our inter-disciplinary connections and cross-site and cross-regional comparisons, (3) add new mechanistic studies and disciplines (e.g. wildlife biology, plant morphological response to disturbance, social dimensions of land-cover change), and (4) address new processes while adding additional co-investigators to our long-term science team and leveraging major additional support from NSF, DOE, NASA, EPA, USDA, The Nature Conservancy, and private foundations.

The proposed work comprises two integrated initiatives that are explored below: assessing the role of disturbance and environmental change in controlling vegetation and landscape dynamics (Part B) and extending these results through additional studies and experiments to the interpretation and projection of carbon and nitrogen dynamics in forest ecosystems (Part C).

B. FOREST AND LANDSCAPE DYNAMICS IN RESPONSE TO DISTURBANCE AND ENVIRONMENTAL CHANGE

Over the past 350 years, the New England landscape has been transformed by human activities interacting with environmental change and natural disturbance (Foster & O'Keefe 2000). The forested landscape was extensively deforested and cut over, then farmed intensively through the mid 19^th C, and subsequently allowed to reforest naturally over the past 150 years as agriculture shifted to the Midwestern US and Eastern populations concentrated in urban and suburban areas (Fig. 14; Foster 2000). Today, the region is 60-95% forested; in structure, wildlife, and many ecosystem, landscape, and regional processes, it is more natural than at any time since the American Revolution. HF LTER research highlights the major conclusion that interpretations of modern ecosystem pattern and process are dependent on integrating detailed knowledge of this history of landscape changes with an understanding of changing disturbance processes and environmental conditions. Such an integrated perspective is essential if we are to meet our major objectives of interpreting modern conditions and anticipating future ones while contributing to policy discussions in conservation, forest and land management, and the environment (cf. Fig. 15).

The challenges confronting an interpretation of modern landscape pattern and process in the context of this history are relatively large, despite past research activities. Central questions that follow directly on results from LTER I and II studies include:

1. How have broad patterns of forest composition been driven by interactions among long-term climate variation, land-use, and natural disturbance?
2. What are the relative contributions of history versus site factors in controlling community composition and landscape variation?
3. How have wildlife populations responded to these changes in land cover and human activity?
4. What role do ongoing disturbances by invasive organisms, notably pathogens that selectively eliminate dominant trees species and exotic flora, exert on forest ecosystem structure, composition and function?
5. What are the important long-term mechanisms for plant survival, recovery, and regeneration following disturbance, including hurricanes, logging, and insect outbreak?

Our past studies provide us with the information and skills necessary to address and answer these questions and to relate the results to our understanding of pattern and process in New England forests. Question 1 emerges from our research in Central Massachusetts and is appropriately addressed at a regional scale across New England using paleoecological, historical, and field techniques. Question 2 has previously been addressed at only two sites and in the C Massachusetts sub-region; in order to evaluate variability across New England, detailed studies are necessary in other physiographic regions that vary considerably in environment, land-use history, and conservation imperatives. Question 3 has never been comprehensively addressed for New England and yet is essential background for detailed wildlife studies and for addressing current ecological, social, and conservation issues. We will focus on a statewide (Massachusetts) assessment where excellent historical records complement our studies of forest dynamics. Question 4 addresses one of the central processes that has shaped New England forests historically and is of immediate concern due to the arrival of the hemlock woolly adelgid and many exotic plant species that are expected to alter forest composition and function. It will be studied using landscape analyses and field and controlled-environment studies. Question 5 emerges from our hurricane experiment and other studies that identify vegetative reproduction as a key forest process that requires morphological, comparative, and field investigations.

This research plan develops logically from LTER studies and represents a major expansion to derive broader generalizations. It addresses fundamental ecological issues that will enable us to contribute meaningfully to important societal concerns.

1. Regional Analyses of Vegetation and Wildlife Patterns and Dynamics

Our LTER studies indicate that the vegetation of northeastern North America was changing in pronounced ways 300-500 years before European arrival (ca. 1620-1700) and that dynamics after European settlement were driven by interactions among climate (Little Ice Age; LIA) and human land use (Fig. 16; Foster et al. 1998; Fuller et al. 1998; Francis & Foster 2000). Building on our substantial LTER database for C. Massachusetts, we will extend these studies across New England to interpret pre- and post-European vegetation change in light of independent data on climate history and Indian and European activity. With LTER and other NSF support, we will use multi-proxy approaches from paleoecology, paleolimnology, archaeology, and history to reconstruct climate, vegetation, and cultural dynamics over the past 1500 years at 8 sites arrayed across the climatic and forest gradients of New England (Fig.17; methods follow Fuller et al. 1998, Foster et al. 1998). Analysis of records from the North American Pollen Database (NOAA) will place these results in a truly regional framework for the entire northeastern US (Fig. 18). Specific methods will include: high resolution pollen records to reconstruct vegetation; chironomids, stable isotopes, geochemistry, diatoms, tree-ring records and historical reconstructions to interpret climate history; charcoal and land-use data to document fire and human impacts; and ²¹⁰Pb and ¹⁴C for temporal control. Forest composition at the time of European settlement will also be determined independently from early Proprietor's records (cf., Foster et al. 1998), which contain tree species data for most of New England. We will develop a comprehensive Proprietor's database in collaboration with Hubbard Brook (HBR) scientists (C. Cogbill), which will add substantially to our knowledge of the early European landscape. Results from these investigations will provide: (1) an objective characterization of the LIA and climate history in New England over the past 1,500 years; (2) comparison of pre- and post-European forest dynamics in relationship to independent environmental and land-use histories; (3) re-examination of historical forest dynamics in light of prior climate and vegetation change; (4) a broad spatial and temporal context for site to regional studies and modeling in the LTER; and (5) interpretations of direct value to two other LTER sites: HBR and Plum Island.

A major focus of HF LTER research is to evaluate the relative importance of modern conditions versus historical factors in controlling stand and landscape-level vegetation composition and dynamics. LTER II showed that the relative influence of history versus site conditions on current vegetation patterns varies considerably among sites and physiographic regions. On edaphically homogeneous sand plains throughout the Connecticut Valley, land-use history is the primary determinant of modern vegetation (Motzkin et al. 1996, 1999a; Donohue et al. 2000). In contrast, across the heterogeneous Central Uplands, vegetation varies with historical factors and complex environmental gradients, especially soil drainage and C:N ratios (Foster et al. 1998; Motzkin et al. 1999b). As a result of such variation, a comprehensive understanding of controls on species distributions, vegetation patterns, and community dynamics must incorporate the full range of physiographic and historical conditions that occur across New England. Thus, we propose to augment past studies of two major physiographic areas (Uplands and Valley) with an integrated study of the history and dynamics of the Coastal region that extends from Cape Cod to Long Island, including Nantucket, Martha's Vineyard, and Block Island. This region is particularly appropriate for such investigations because: (1) the physiography and disturbance history contrast strongly with the Upland and Valley; (2) the coastal region is an international conservation priority that supports numerous uncommon species and communities and yet has never been studied comprehensively from a broad spatial and temporal perspective; and (3) previous coastal studies have failed to evaluate the effects of historical disturbance rigorously, despite the long and well-documented history of land-use and fire (Fig. 19; Foster & Motzkin 1999).

We propose integrated paleoecological, historical, and field studies to determine factors contributing to the development of modern vegetation patterns and dynamics. We hypothesize that prior to European settlement, vegetation was largely controlled by: (1) variation in soil fertility and drainage, which is strongly related to surficial landform; (2) fire history, which is correlated with surficial geology and human distribution; and (3) geographic location and exposure. We further hypothesize that the modern occurrence, composition, and dynamics of major coastal assemblages have primarily resulted from disturbance, especially land-use history. Although wind, fire, and pathogens have been important, we anticipate that they have not obscured land-use effects. In particular, we hypothesize that the distribution of rare species and communities is tightly linked with patterns of prior land-use.

To address these hypotheses, several approaches and data sources established in LTER II will be utilized to allow for direct comparison with our results from other sub-regions:

(1) Fine-resolution pollen, charcoal and sediment analyses of 15 sites, stratified by geography, landform, and distance from centers of Indian and European activity, will provide decadal information on forest composition and dynamics at regional and landscape scales over the last 1,500 years (cf., Fuller et al. 1998).
(2) Comprehensive historical data and maps on cultural features, vegetation, and disturbance (1650 to present) will be added to our large New England GIS database.
(3) Sampling of vegetation and soils in random plots will relate regional vegetation variation to site conditions and historical disturbance (Motzkin et al. 1996, 1999a, 1999b; Donohue et al. 2000).

This research addresses the major HF LTER theme of evaluating the relative importance of modern conditions versus history in controlling forest composition and dynamics. It will also make significant contributions to biodiversity conservation and management (Foster & Motzkin 1998, 1999), expand our regional analyses, and add substantially to our growing GIS, paleoecological, historical, vegetation and soils data for New England.

HF LTER studies have previously concentrated on vegetation dynamics. However, 400 years of land-use and land-cover change have also produced profound changes in wildlife populations (Foster 2000, Foster & O'Keefe 2000). Currently, many mammal and bird species (e.g., bear, beaver, fisher, moose, eagles and herons) that were uncommon during the past 200 years are expanding dramatically, whereas major declines are occurring in species characteristic of New England's agrarian past (e.g., bobolinks, meadowlarks, woodcock and open-land sparrows; Fig. 20). These changes have been driven by many ecological and cultural factors, including habitat availability, competitive interactions, climatic conditions, cultural perceptions, and management. However, few studies have attempted to detail the history of wildlife dynamics and relate these to shifts in ecological and cultural parameters. Such a perspective is critical in order to understand current wildlife dynamics and to address policy conflicts and ethical dilemmas that result from wildlife-human interactions in an increasingly suburbanized landscape (Deblinger et al. 1999).

The regional databases and historical perspective of the HF LTER present an unusual opportunity to address basic ecological questions about the factors controlling wildlife populations, the feedbacks between plant and animal species, and contrasts between the organization and dynamics of plant and animal assemblages. Consequently, we propose a historical analysis of New England wildlife to: (1) document major trends in populations since European settlement, (2) identify the loss or arrival of major species and analyze the long-term composition of wildlife assemblages, (3) identify the physical, biological, and cultural factors driving these changes, (4) integrate this information with vegetation data to increase our understanding of forest processes through time, and (5) provide a context for understanding recent changes and for guiding conservation policy. We will focus on the entire state (Massachusetts) where excellent historical records enable us to develop a semi-quantitative understanding of these dynamics and relationships (Fig. 21; Cardoza pers. comm., Bernardos, Foster & Motzkin in prep.). Results will assist species-specific investigations on the effects of suburbanization and land-use change on wildlife populations (Deblinger et al. 1999).

2. Biological Invasions: Population, Community, and Ecosystem Response

Invasive plants and animals cause significant changes in forests at population to ecosystem levels through competition with and predation on native species, and by influencing resource availability and forest structure and process. Exotic insect pests have been particularly important historically, resulting in substantial change in forest composition and function across the Eastern US and the near elimination of several dominant trees in the past century. Non-native plant species are increasingly altering forest composition in ways that are largely unknown. Due to the importance of such events in forests worldwide and their historical and current significance to New England, we propose investigations of recent invasions by a major forest pest and by exotic plants in order to determine effects on species interactions, community composition and structure, and nutrient dynamics.

Although the history of forest pathogen introduction in New England is well-documented, no studies address the factors controlling these infestations or forest ecosystem response to selective mortality of dominant species. Consequently, the introduction of HWA, a small, aphid-like insect from Japan, presents a unique opportunity and imperative to study forest damage and response to a major pathogen. HWA has caused extensive decline and mortality in parts of hemlock's range and looms as a major threat to eastern forests (Fig. 22). HWA reached southern New England in 1985, produced widespread mortality in Connecticut by 1988, and is now in >70 Massachusetts towns. Because hemlock is an important and abundant late successional species that strongly controls stand microclimate and soil conditions, many HF LTER studies have investigated the long-term dynamics and function of hemlock forests. Consequently, we are in a strong position to study HWA (Foster et al. 1992; Foster & Zebryk 1993; Compton & Boone 2000; Hadley in press, Orwig & Foster 1998).

We propose to expand studies initiated with USDA Forest Service support to undertake two related efforts: (1) stand to regional evaluation of HWA infestation and hemlock mortality (1985 onwards) and assessment of the physical, biological, and historical factors influencing damage and response; and (2) analyses of forest composition, structure, micro-environment and ecosystem responses to hemlock stress and mortality.

Regional Spread and Landscape Controls. In preliminary studies of a 5900 km² area of Connecticut, HWA occurred in nearly 90% of stands and was responsible for nearly complete mortality in many large hemlock forests (Fig. 23; Orwig & Foster 1998). Mortality decreased with latitude and duration of infestation and appears to be unrelated to site or stand factors; therefore, we project heavy to complete mortality of hemlock across the region. We will extend our study to Massachusetts to utilize our extensive data on healthy hemlock forests and to document changes that relate to ongoing HF LTER studies.

Forest Ecosystem Response. As infested stands may suffer complete mortality in 4-8 years, the potential exists for major ecosystem impacts including nutrient loss, N export, and erosion. Our initial results indicate large increases in N mineralization and nitrification rates with mortality (Jenkins et al. 1999). However, many questions remain concerning the magnitude and duration of N changes and their relationship to soil conditions and biotic uptake. Consequently, we will initiate studies of N availability, mineralization, and nitrification rates in sites recently infested with HWA and will continue these long-term measurements to investigate temporal dynamics and driving mechanisms as mortality commences. We will then compare ecosystem response to pathogens with ongoing studies of hurricane damage, N saturation, soil warming, and logging (cf. Section C).

Although the spread of aggressive invasive plants in temperate forests is generating increasing concern, few studies have evaluated the influence of historical factors on the dynamics of these invasions. Because land-use frequently creates opportunities for plant establishment and exerts long-term impacts on site conditions and patterns of succession (Foster 1992), invasion and spread of exotics is likely to be influenced by past and current activity. In turn, biological invasions may alter community dynamics, species diversity, and ecosystem function. In LTER III, we propose studies to address the following questions: (1) Does variation in land-use history influence forest susceptibility to invasion and the distribution of exotics? (2) How do invasives affect the composition and dynamics of forests and influence ecosystem processes? And, (3) What relationships exist between physical disturbances, anthropogenic stresses, and biological invasions?

We will study the distribution and dynamics of the 5 most widespread invasive woody species in C. Massachusetts (Berberis thunbergii, Rhamnus cathartica, R. frangula, Lonicera sp. and Celastrus orbiculatus), each of which has the potential to cause major forest ecosystem changes. Berberis is an understory shrub that forms dense thickets and can alter soil characteristics (Kourtev et al. 1998), Rhamnus and Lonicera are shrubs that can displace existing vegetation (Weatherbee et al. 1998), and Celastrus is a vine that can grow densely on trees and the ground (Dreyer 1994). Species distributions will be determined through extensive mapping and existing data for the Harvard Forest and adjacent Quabbin Reservation. Detailed site histories will be determined from field and archival records following Motzkin et al. (1996, 1999b) to allow direct comparison with data from the Connecticut Valley, Central Massachusetts, and the Coastal region in order to determine broad patterns of variation in the distribution and abundance of invasive species. Edaphic and other resource variables will be sampled in adjacent sites with and without invasives to identify factors that may influence species distributions. Long-term plots will be established in areas with different land-use histories, in which establishment, abundance, and demography of invasives will be monitored. In addition, we plan to measure changes in resource availability, dynamics of indigenous species, and community structure.

To complement these and past studies on species interactions and responses to resource conditions (e.g., George & Bazzaz 1999b; Wayne & Bazzaz 1997; Catovsky & Bazzaz 1999), we also propose a set of controlled studies and introductions of common invasive species in the field. The goal of these experiments is to characterize competitive interactions between invasives and native plants as well as to measure the influence of altered resources on establishment of invasives and indigenous species. We will integrate these studies with controlled environmental (glasshouse) studies in which factorial experiments will further quantify competitive interactions among invasive and native species and determine mechanistic responses to resources, including resource-use efficiency and allocation patterns. Results will increase our understanding of the differential susceptibility of plant communities to invasion and thereby improve our ability to predict short and long-term population dynamics and community structure. Our results will also have direct bearing on public land management, including the Quabbin Reservation, where resource managers have identified the control of invasive species as a major objective (T. Kyker-Snowman pers. comm.).

3. The Mechanisms of Plant Recovery from Disturbance: Vegetative Reproduction

Our prior investigations of regeneration dynamics after disturbance have focussed on resource and microsite influences on seedling establishment or on demographic limitations to colonization after disturbance (e.g., Carlton and Bazzaz 1999a,b). It is increasingly clear, however, that vegetative regeneration is often the primary mechanism by which woody species respond to physical disturbance (Del Tredici 1998, Donohue et al. 2000). For instance, results of the experimental hurricane as well as our prior studies on logged areas and areas salvaged following HWA infestation demonstrate the critical importance of rapid vegetative regeneration in controlling ecosystem response to physical disturbance (Cooper-Ellis et al. 1999; Foster et al. 1998a). To date, few studies have explicitly addressed the differing morphological responses of individual species to the range of physical disturbances that are characteristic of temperate forests (Del Tredici, in review). As a result, the degree to which ecosystem response depends on pre-disturbance stand composition and structure versus seedling regeneration is largely unknown. Such an understanding is critical if we are to compare species responses to different disturbance types and if we are to understand the mechanisms by which individuals and species respond to dynamic landscapes and management regimes.

In LTER III, we propose to document the range of vegetative morphological responses displayed by the dominant tree species in New England to characteristic disturbances including windthrow, fire, logging, and pathogens. We are particularly interested in evaluating the change in regenerative capabilities of individuals of different life stages. For instance, although vegetative regeneration is well-documented for mature stems, morphological adaptations that allow even recently germinated seedlings to regenerate vegetatively are widespread and may be a primary mechanism for successful establishment in many species. Using long-term data from the experimental hurricane, our new logging study (see Section C, below), and ongoing HF studies of sites with known disturbance histories, our specific objectives are to: (1) describe the vegetative morphological response of common tree species to dominant disturbances; (2) evaluate the regenerative capability of individual species relative to stem age; and (3) determine the importance of different regenerative strategies with increasing time since disturbance. We will develop a functional classification of damage types related to detailed morphological descriptions of mechanisms of vegetative regeneration. Results will contribute significantly to our understanding of biological and morphological constraints on species response to disturbance and our interpretations of forest ecosystem dynamics.

The hurricane manipulation, modeled after damage patterns from the 1938 hurricane (Rowlands 1941, Foster 1988a, Cooper-Ellis et al. 1999) and initiated in 1990, allows us to integrate our new studies of vegetative reproduction (LTER III) with our data on seedling regeneration (LTER II; Carlton & Bazzaz 1999a) in understanding mechanisms of ecosystem recovery. To accomplish this we will continue long-term measurements in the 0.8-ha manipulation and adjacent 0.6-ha control. All trees are mapped; data on damage, survival, and sprouting, recorded annually for each tree from 1991-1997, will continue every 3 years, as will assessment of regeneration via sprouting, new seedlings and advance regeneration. Herb and shrub cover, measured before the manipulation and in 1991, 1992, 1995, will continue every 5 years. Emphasis in LTER III will shift from initial forest recovery (Bowden et al. 1993a, Foster et al. 1997, Cooper-Ellis et al. 1999) to evaluation of long-term stand development in comparison with prior HF studies following the 1938 storm (e.g., Spurr 1956, Oliver & Stephens 1977, Henry & Swan 1974). The importance of sprouts, new saplings, and advance regeneration in forming the next forest canopy is becoming apparent as the new cohort of trees experiences severe competition and successful stems move into larger size classes. Consequently, we will focus on documenting modes of long-term regeneration and evaluating the contribution of tree population dynamics to community recovery.

The long-term continuation of this experiment is critical to understanding the mechanisms of forest vegetation recovery from a disturbance type that affects nearly every generation of New England forests (Boose et al. 2000a). Data on long-term forest recovery in the experimental hurricane will be directly comparable with our long-term data from plots established after the 1938 hurricane and with our new investigation of logging (below).

C. CARBON AND NITROGEN INTERACTIONS WITH LAND USE AND ENVIRONMENTAL CHANGE

As a consequence of the past three centuries of land-use and natural disturbance, the New England landscape, like much of the Eastern US, currently supports new and predominantly young forest that is growing rapidly and storing significant amounts of carbon (Foster & O'Keefe 2000, Wofsy et al. 1993, Aber et al. 1999, Irland 1999). The vast extent of this aggrading temperate forest (200-300x10⁶ ha in the US and Canada south of 51^oN) makes these lands critical elements of global carbon dynamics with significant influence on future atmospheric CO₂ concentrations and landscape response to climate change (Wofsy et al. 1993).

The long-term trajectory and magnitude of forest growth and carbon storage in New England and the Eastern US are uncertain. To assess the current and future role of these forests in the global carbon cycle information is required defining: (1) the successional status of forests across the landscape, (2) current rates of C sequestration and evolution for the principal vegetation assemblages, (3) the potential effects of climate variation and human-imposed disturbance and stress (e.g., N deposition, logging, forest conversion), and (4) the mechanisms controlling C and N dynamics. Past LTER studies place us in an ideal position to address these issues through intensive site measurements, experiments, and regional analyses. Process-level and cross-site studies will, in turn, allow us to use of our ecosystem models to develop regional projections of forest-atmosphere carbon exchanges.

The challenges facing an assessment of current and projected carbon storage in New England are great. Traditional measurements of forest growth and carbon uptake such as forest inventories cannot define rates for such basic processes as soil organic matter accumulation or storage of carbon in woody detritus. This compromises our ability to link carbon storage to past land-use activity and ongoing environmental change. Although the New England states currently support greater forest area than at any time in the past 200 years (Irland 1999), increasingly these forests are being logged for wood products or converted to suburban uses that greatly alter local and regional carbon balances. Thus, one objective of our work in LTER III will be to assess the impact of forest management and land use change on carbon uptake. We plan to determine the regional extent of logging and forest conversion, and to study in detail the effects of a typical harvest on carbon stocks. The goal is to be able to compare carbon budgets of disturbed lands with patterns of carbon accumulation in forests growing on undisturbed lands.

Fossil fuel combustion and fertilizer production have greatly increased the deposition of fixed nitrogen (NO_y and NH_x) on temperate landscapes (Galloway et al.1994, Melillo 1996). Increased deposition of these biologically reactive N forms can lead to eutrophication of terrestrial and downstream aquatic ecosystems (Aber et al. 1989, Stoddard 1994). Some studies have suggested that these N inputs stimulate significant C storage in mid-latitude forests (Townsend et al. 1996, Holland et al. 1997). However, more recent results from the N Saturation study (Magill et al. in press) and our cross-site comparison of ¹⁵N tracer movements in temperate forests (Nadelhoffer et al. 1999c) suggest that if there is an effect of N deposition on forest C balance, increased soil C retention rather than N-induced increased tree growth is important.

Increasing temperatures may both increase the length of the growing season and fertilize forests as accelerating decay of soil organic matter increases N availability for uptake and storage in plants. These are probably more important influences on forest C accumulation than N fertilization effects (above). Since the C:N ratio in plants is substantially greater than in soil organic matter, warming may increase net C storage by transferring N from soils to vegetation, especially in forests (McKane et al. 1995, 1997; Melillo 1996. Townsend & Rastetter 1996). The magnitudes of such increases depends on how plant C balance is affected by other factors, including the potentially deleterious effects that increased N deposition may exert on tree growth and ecosystem processes (i.e., N saturation; cf. Aber et al. 1989,1998) and other aspects of climate change (e.g., water availability, temperature effects on photosynthesis and respiration).

Results from our long-term experiments underscore the fact that many fundamental mechanisms controlling N and C dynamics in temperate forest ecosystems remain poorly understood (Berntson and Aber 2000, Nadelhoffer et al.1999, Aber et al. 2000). Consequently, in LTER III we propose to expand our long-term measurements, experiments, regional comparative studies, and modeling activities to address central aspects of nitrogen and carbon cycles in New England forests, especially as they pertain to current and prospective environmental concerns and to the global C cycle. We will focus on the response of C and N dynamics to vegetation change and succession, climate variability, forest management, and stress, especially nitrogen deposition. At the heart of this work will be intensive site-based studies extended to pertinent sub-regional, regional, and even global scales through extensive measurement and modeling activity.

The proposed work expands HF LTER I and II activities in which we established a series of unique field measurements, large-scale experiments, cross-site studies, and modeling activities designed to increase our understanding of C and N dynamics in temperate forest ecosystems. Our permanent plot measurements are coupled with long-term eddy flux measurements of forest-atmosphere exchanges of CO₂, gaseous N compounds, other trace gases, water vapor, and energy made at the Environmental Measurement Station (EMS; Wofsy et al.1993, Goulden et al. 1996, Hollinger et al. 2000). Large-scale and long-term experiments include: DIRT (Detritus Inputs and Removal Treatments), in which above- and below-ground litter inputs are manipulated to evaluate processes and mechanisms of organic matter formation; Soil Warming, in which soil temperature is raised 5 ºC above ambient to assess the effect of warming on soil carbon and nitrogen dynamics (Peterjohn et al. 1994,Melillo et al. 1995, in press); and N Saturation, in which chronic N additions to forest plots are used to assess the effects of anthropogenic enhancement of N inputs on forest structure and function (Aber et al. 1993, Magill et al. 1997, in press, Nadelhoffer et al. 1999a). These intensive site-based activities are broadened through comparative studies at Hubbard Brook, Bear Brook and Howland Forest, ME, and international networks of ecosystem studies (e.g. AmeriFlux, Hollinger et al. 2000, NITREX, Wright et al. 1994).

In LTER III we will expand these efforts to address four primary questions:

1. What are the current rates of C storage in temperate forests and how are they influenced by climate variability, succession, and forest logging and conversion?
2. How would direct (chronic N inputs) and indirect (within ecosystem N redistribution due to climate warming) influences on the Northeast's N cycle affect the C storage in the region's forests?
3. What above- and below-ground mechanisms mediate the ability of temperate forests to store carbon and to process fixed nitrogen?
4. What is the regional pattern of forest productivity in New England and how is it affected by past and current land-use, forest stresses, and N deposition?

Question 1 will be addressed using the eddy-covariance measurements that have been taken continuously in HF LTER since 1990, combined with intensive biometry and measurements of soil flux on permanent plots in the footprint of the tower. This work will be complemented by assessments of logging and forest conversion at site, sub-regional, and regional scales. Question 2 will be addressed through analyses related to the N Saturation and Soil Warming experiments. Question 3 will be explored through ongoing and new studies on the DIRT, N Saturation, and Soil Warming experiments. Question 4 will be examined through comparative studies across New England and through regional ecosystem modeling based on intensive and extensive measurements at the Harvard Forest and regionally.

1. Long-term Measurement of Carbon Storage in Relation to Climate Variability and Forest Harvesting and Conversion

We plan to enhance long-term measurements of Net Ecosystem Exchange (NEE) using the EMS eddy-covariance system, extending the longest continuous time series for NEE in the world. The flux data, in combination with comprehensive ecological measurements, provide the foundation for interdisciplinary investigations of carbon balance, legacies of prior land use, effects of logging, and biosphere-atmosphere exchanges of pollutants and greenhouse gases by addressing the following questions (Figs. 24, 25):

1) What are the important biological processes controlling NEE in temperate forests?
2) How do these processes vary with climate over annual to decadal time scales?
3) What are the quantitative relationships between NEE, forest structure (e.g. canopy nitrogen content, LAI, coarse woody debris, stand age), species composition, and climatic variations over longer time scales (inter-annual, inter-decadal)?
4) Do seasonal anomalies in climate affect photosynthetic capacity or pools of short-lived organic matter, thus affecting annual carbon sequestration in subsequent years?

Measurements in permanent plots of C cycle processes, including aboveground wood increment, soil respiration, and production of litter and coarse woody debris, will complement ongoing eddy-covariance measurements of fluxes of CO₂, H₂O, sensible heat, O₃, and NO_y and observations of environmental parameters (e.g., net radiation, PAR, soil and air temperatures, wind profiles; Wofsy et al. 1993, Goulden et al. 1996a, Munger et al. 1996). Individual biological C fluxes will be summed and reconciled with the forest NEE. Our 10-year data record provides a very accurate measure of average seasonal fluxes and defines mean ecosystem response to environmental forcing by incident light, temperature, length of the growing season, etc. Seasonal and interannual anomalies in NEE and its component fluxes will be compared in detail to climate anomalies (e.g., drought) and disturbance (e.g. defoliators, ice or wind damage) to test concepts of causal relationships.

The proposed work will deliver basic biometric data, detailed carbon budgets, observations of the sensitivity of CO₂ exchange to environmental forcing and to past land use, and integrated synthesis of ecosystem control over temperate forest NEE, including:

i. Process measurements and compartment inventories (see table below).
ii. Integration of biological/ecological measurements with climate and NEE data.
iii. Determination of control over long-term patterns of CO2 exchange from i and ii.

PROCESS MEASUREMENTS AND COMPARTMENT C INVENTORIES.

Process	Observation
NEP	Annual sum of NEE
GEE + Tree Respiration	Species-specific wood increment data
GEE + Tree Respiration	Leaf litter production data
Soil Respiration	Direct soil collar measurements
Tree mortality and decay	Annual tree and coarse woody debris surveys

Our focus on defining quantitatively the implications of past and ongoing human and natural disturbance (prior land use and hurricanes) for C dynamics is a new direction for the EMS that addresses major questions in HF LTER. The measurements of leaf and wood production, respiration, and net C storage in wood track seasonal and inter-annual feedbacks between climate and C allocation in major forest compartments and define the variable contributions of different tree species. These observations will provide a unique opportunity for direct, quantitative assessment of the influence on the carbon cycle of prior land use, and interactions between land cover change and climate in the temperate region.

Forest management plays a key role in sequestration or release of atmospheric CO₂ by forests. Although there are many models of logging effects on biomass and carbon balances, accurate observations of carbon fluxes associated with harvests are lacking for both short and long time scales. In central New England, logging has increased as a consequence of forest growth and market conditions, to become the pre-dominant forest disturbance. Determination of harvest impact on C storage and the pattern of forest cutting in the landscape is therefore a critical imperative for research. Thus we will expand on past efforts to include a detailed assessment of the fate of carbon following logging in a typical forest. We will also analyze the intensity, distribution, and impacts of logging and forest conversion in central Massachusetts, in the entire state, and across New England. In addition to providing direct information on forest disturbance and C storage, these analyses will provide essential input for regional modeling activities and vegetation studies.

Dynamics and fate of carbon following commercial logging. Logging impacts will be assessed in a commercial harvest in a mature hardwood forest, adjacent to and structurally very similar to those at the EMS and other LTER experiments. The proposed logged and control areas have been sampled since 1993 in 46 300 m² plots (9 harvest and 37 control), including litter by species, LAI, soil respiration, soil moisture, and high-frequency dendrometry on ~800 trees. In 1999 we surveyed coarse woody debris. In order to maintain an accurate carbon budget, harvest-related measurements will carefully account for removals by weighing all product and inventorying woody residue and slash. Long-term post-harvest measurements will pay particular attention to soil moisture, temperature, respiration, and organic matter profiles (total carbon and C/N ratio). We will also measure species-specific growth increments, including new growth into the >5 cm size-class, tree mortality, and decay.

The proposed work on logging will deliver the following products:

1. Detailed accounting of the short-term C balance on the timber harvest compared with adjacent undisturbed sites (for which NEE is measured by eddy correlation).
2. Accounting of the initial fate of below-ground soil carbon following harvest.
3. Integrated understanding of how both ecological and climate responses to forest harvest control subsequent evolution of forest carbon balance.
4. Quantification of the relative magnitudes of the effects of human and long-term (10-year) natural disturbance within the ecosystem.

Regional Assessment of Forest Management and Conversion. To place our site-based measurements in a broad context and to ascertain the impact of forest harvesting and conversion on C storage regionally, we will: (1) develop a spatially explicit, long-term (1984 onwards) assessment of logging activity in central Massachusetts and (2) analyze Massachusetts and New England-wide patterns of forest cutting, forest conversion to other land uses, and forest growth versus removals.

Massachusetts is unique in requiring Forest Cutting Plans for all commercial timber harvests. These reports enable development of databases and GIS overlays characterizing the spatial characteristics, intensity, motivation, landowner, and volume of all logging activities in our central Massachusetts sub-region (Fig. 26). In LTER III we will assemble this database and undertake field sampling of areas harvested in the past 15 years in order to confirm the accuracy of the data and to evaluate the compositional and successional response of forests to logging. These data will be analyzed to:

1. Determine the extent and geographic pattern of harvesting in relationship to cultural, natural, ownership, and land-use factors.
2. Assess temporal trends in wood volumes and species and quality removed in relationship to regional economic factors.
3. Evaluate land-use patterns and behavior in a landscape predominantly owned and controlled by non-industrial, private forest owners (Kittredge 1999).
4. Make current our regional analysis of forest disturbance developed in LTER I and II and include natural and human factors by including dominant land-use practices.
5. Generate sub-regional and regional estimates of forest growth and impacts of forest harvesting conversion on carbon storage.

USDA Forest Service Forest Inventory Analysis (FIA)will be used to compare these detailed and spatially explicit data with state- and New England-wide assessments of forest cutting, conversion, and forest growth versus removal. FIA provides state-level average estimates of growth, removals, and total forest area. Our detailed data enable us to estimate harvest rates for extensive non-industrial private forest lands, to compare this to FIA removal rates, and to develop factors to apply to FIA removal rates elsewhere. This will be combined with land change data to better characterize changes in timber volume and C storage due to harvesting versus change in land use. Other New England states have varying degrees of timber harvesting regulation and data that can provide indirect data on harvest and conversion. In addition, FIA crews recently measured the EMS footprint, providing a baseline for comparison of data from the Eastern US with eddy flux data and forest composition and stand characteristics at HF LTER.

2. Mechanisms of Carbon and Nitrogen Dynamics in Relation to Climate Change, Land Use and Nitrogen Deposition - Long-term Experiments

Long-term experiments initiated in LTER I to follow the course of carbon and nitrogen dynamics in relation to disturbance, climate change, and nitrogen deposition become increasingly important as we extend our research to investigate the effects of land use and the mechanisms of C and N cycling, and as we pursue a broader geographical scope. As these studies continue well into their second decade, they enable us both to evaluate the sensitivity of C and N dynamics to climate variability and to investigate above- and below-ground processes that control N and C cycling in temperate forests. Equally important, these experiments have critical cross-linkages with our land-use studies, hurricane experiment, permanent plots, and measurements made at the EMS and in the proposed logging assessment. They also provide key input to the regional and global comparative studies that we undertake and to the synthetic modeling approaches described later in this proposal. Thus, we will continue and expand these experiments in LTER III as described below.

We will continue chronic manipulations of above- and below-ground litter inputs to forest soils by maintaining and expanding the DIRT (Detritus Input and Removal Treatments) experiment, which is developing an understanding of the long-term (years to centuries) influences of plant inputs on soil organic matter development and dynamics. Treatments are conducted on replicate (n = 3) 3m ´ 3m plots and include doubling of leaf litter inputs, preventing leaf litter inputs, preventing root inputs (with root-ingrowth barriers), and preventing all (leaf and root) plant inputs. A fifth treatment involves experimental "impoverishment" of soil by replacing the top 20 cm of O and A horizons with B horizon soil.

We will conduct our third post-treatment sampling in the early summer of 2001 after 10 years of manipulations. Sampling in years 0 (pre-treatment), 1, and 5 provided information on C and N contents, basic soil properties (CEC, base saturation, pH, texture), and potentially mineralizable C and N pools (methods of Stanford & Smith 1972, as modified by Nadelhoffer 1990). Year 5 sampling and analyses (Nadelhoffer et al. submitted) were expanded to include soil microfauna, bacteria and fungal counts and measurements of dissolved organic N and C dynamics (Aitkenhead et al. in prep), which will be repeated in 2001. In addition, we are collaborating with Dr. Maura Meade, a molecular ecologist at Allegheny College, to explore the use of molecular probes for assessing how variations in plant inputs to soil influences microbial functional groups.

Field measurements of soil respiration (Bowden et al. 1993, Boone et al. 1998) and soil solution chemistry (Aitkenhead et al. in prep) will be continued during alternate growing seasons, or more often if resources allow. Our accumulated field C flux and soil solution data will complement our existing data on temporal changes in soil properties and mineralizable organic matter pools. Our combined field and laboratory studies will provide information needed to test simulation models of plant-soil interactions (e.g., GEM; Rastetter et al. 1991) and soil organic matter dynamics (e.g., DOCMOD; Currie & Aber 1997).

In addition, we plan two new, related activities. First, we will install 'impoverished' soil plots to document soil development (see above) in three additional forest types dominated by red pine, white pine, and sugar maple. Once installed, these O-A Less plots require no maintenance as natural rates of root ingrowth and litter deposition are allowed. Periodic sampling (Years 0, 1, 5, 10, 20, 30…) and assaying of soil properties and mineralizable C and N pools will allow us to characterize patterns of soil process and property recovery under different forest types.

Second, we will expand linkages to related experiments in other regions. Currently we collaborate actively with the H.J. Andrews LTER (temperate conifer forest; Drs. Kate Lajtha, Phil Sollins, etc.) and Allegheny College, PA (base-rich sugar maple forest; Dr. Richard Bowden), which have modified DIRT experiments. We have begun discussions with collaborators working at the Coweeta (GA) and Luquillo (PR) LTERs (January 2000), Hungarian investigators (lead by Kate Lajtha at the June 1999 East European ILTER symposium in Budapest), and Korean and other East Asian ILTER investigators (Nadelhoffer, October 1999 East Asia ILTER meeting in Seoul), who are considering similar experiments. Ultimately, we anticipate an array of DIRT-type experiments in a variety of climatic and vegetation zones to provide information needed to build a process-level understanding of soil organic matter genesis and function at local and global scales.

We will continue treatments and core measurements on N Saturation with the primary goal of assessing the complete, long-term sequence of responses. Of particular interest is documenting the timing and extent of increased nitrate loss to determine whether this ecosystem response is driven by the concentration or the cumulative dose of additions. If cumulative dose is the key driver, then the low N plots should take three times longer to show elevated nitrate leaching than the high N plots. We will also continue the N+S plots without the S addition, making them a replicate for low N treatment. Such replication is rare in long-term, large-plot experiments.

LTER support allows continuation of the experimental manipulation and core measurements. However, we are pursuing support for two additional process-level initiatives (G. Berntson, PI).

Effects of N Deposition on Kinetics and Magnitude of Soil N Immobilization. We have developed new short-term ¹⁵N tracer methods to quantify the kinetics and magnitude of N immobilization and to quantify more clearly the contributions of "fast" and "slow" processes to gross N immobilization. In hardwood and conifer stands, rapid immobilization accounted for a significant fraction of total N immobilized (1/3-2/3 of the total), and long-term N deposition led to a reduction in N immobilization. For the conifer forest, this reduction was due to a loss in the amount of N immobilized during the fast phase, whereas in the hardwood stand it was due to slow phase immobilization. A strong relationship exists between NO₃^- leaching losses and measured N immobilization rates but only when both rapid and slow phases of N immobilization are included. We plan to quantify fast vs. slow N immobilization in a larger number of sites, over longer periods of time, with greater replication, and with coupled field and laboratory experiments. We plan to characterize the biotic and abiotic components of these different immobilization phases through manipulation (soil sterilization) and analysis of the linkage between C and N dynamics during immobilization (soil CO₂ flux, microbial uptake).

Effects of Long-Term N Deposition on Canopy Structure and Photosynthesis. Preliminary measurements from access towers in the high N pine plots suggest that treatments have changed canopy production, turnover rate, and a critical foliar N:A_max relationship. Collaborating with the USDA Forest Service, we found that while foliar N has increased by more than 100% and soluble protein has increased by 85% in the pine high N treatment canopy, maximum photosynthesis did not change. We also found that needle retention times were reduced by 40%, and LAI reduced by 16%, while litterfall rates increased by approximately 20%. These data suggest that increased N deposition has severely reduced the pine forest's gross photosynthetic capacity through reductions in standing foliage with no physiological compensation, providing an explanation for observed reductions in tree diameter increments and soil respiration rates. We will extend these measurements to low N pine plots and all hardwood plots.

We will continue to study the effects of soil warming on soil and plant processes by maintaining our extant soil-warming experiment and by adding a new one during the next six-year LTER period. Our current study, established in 18 six by six meter plots in 1991, was designed to explore the consequences of temperature increases on soil processes including soil respiration, organic matter decomposition, methane production and consumption, net nitrogen mineralization and net nitrification, and nitrous oxide production (e.g., Peterjohn et al. 1994, Melillo et al. 1995). The plots are grouped into six blocks, each with three plots assigned to one of three treatments. The treatments are: (1) heated plots in which the average soil temperature at 5cm is elevated 5^oC above ambient using buried heating cables; (2) disturbance-control plots that are identical to heated plots except that they receive no electric power; and (3) control plots that have been left in their natural state. For at least the first three years, 2001 through 2003, we will continue to measure soil respiration, trace gas fluxes, nitrogen mineralization and nitrification, and nitrogen leaching to extend our documentation of the effects of warming on soil carbon and nitrogen stocks and cycling rates.

In addition, we are planning a new warming experiment to address a central question: Can soil warming change ecosystem N distribution and thereby change the capacity of forests to store C? We will directly test the idea that warming can result in the redistribution of N from the soil to plants and thereby enhance net ecosystem carbon storage (Melillo et al.1995). Using our warming protocol in Sweden, Professor Sune Linder (pers. comm.) has observed a 30% increase in wood increment in response to a 5^oC soil warming over five years. In our new, larger, long-term soil-warming experiment, we will make a series of plant and soil measurements that will quantify changes in C storage to provide insight as to how these changes may be related to N redistribution from soil to plants. The larger plots are required to include enough trees in a treatment to capture the plant-soil interactions. Beginning in 2001, we will establish three 30x30 m plots adjacent to the current soil warming study and will make 3 years of baseline plant and soil measurements on all plots. In 2004, we will initiate a single treatment on each plot - one heated, one disturbance-control, and one control, following the protocols in our initial warming experiment. We will use baseline time-series data in place of plot replication to determine treatment effects in the same way that whole-watershed studies do.

Plant measurements will include: (1) annual woody increment, (2) C:N ratio in the woody increment, roots and leaves, and (3) natural abundance ¹⁵N content in the various plant parts. Measurements of woody increment will provide an estimate of the degree to which soil warming and the acceleration of the N cycle stimulate C storage in trees. The measurements of C:N ratio in woody increment, roots and leaves will give insight into how "plastic" the stochiometry of various plant tissues is in response to acceleration of the N cycle through warming. We will also measure ¹⁵N in plant tissues (wood, roots, leaves) on all plots to determine whether plants on the warmed plots are acquiring N that has been mineralized from more refractory (¹⁵N-enriched) soil organic matter pools.

Soil measurements in the new study will include: (1) C and N stocks in the bulk soil and in the light density (LD) and heavy density (HD) soil fractions; (2) the ¹⁴C and ¹⁵N content of the bulk soil, the soil fractions and the fine roots; (3) CO₂ and ¹⁴CO₂ efflux from the soil surface; and (4) lignase enzyme activity in the various soil horizons; and (5) net nitrogen mineralization and nitrification rates. Measurements of C and N stocks in the soil and respired CO₂ are essential background information for developing budgets using ¹⁴C and ¹⁵N. We propose to use ¹⁴C (Trumbore et al. 1996, Paul et al. 1997, Trumbore et al. 1997) and lignase enzyme activities (Sinsabaugh et al. 1991, 1992) to determine whether soil warming has increased or decreased the decomposability of meta-stable soil organic matter. Measurements of N mineralization and nitrification will provide insights into the effects of warming on the N cycle and essential background information for interpreting our ¹⁵N natural abundance study.

3. Regional Modeling of C and N Dynamics in Relation to Land Use, Climate Change and Forest Stress

We have applied PnET models to Harvard Forest sites and across the New York/New England region and have validated these against measurements for gross ecosystem C flux, net primary production, and water yield at HF and Hubbard Brook LTER and regional data for NPP and water yield (Figs. 27, 28). The individual effects of several components of global change (CO₂, temperature, precipitation, tropospheric ozone, N deposition and land use) have been analyzed separately, and predicted responses have been published in conjunction with US Forest Service Northern and Southern Global Change Programs.

In LTER III we will extend this analysis spatially and integrate the separate effects of the different stressors listed above into a single model to provide an integrated analysis of the response of Eastern US forest ecosystems to the full suite of major environmental change factors. This work will continue in cooperation with the US Forest Service through the following steps.

First, we will improve PnET's photosynthesis routine to integrate the combined effects of CO2 and ozone based on studies of the relationship between internal leaf CO2 concentration and maximum photosynthetic rate. This relationship can be described by a Michaelis-Menten-type equation that can be used to predict the maximum rate of photosynthesis achievable at a given foliar nitrogen concentration. This approach is compatible with the realization of ozone effects on photosynthesis currently in the model and offers the potential for alternative descriptions of stomatal response to both ozone and CO₂ concentrations.

Second, we will extend the spatial extent of our study to include all forests in the Eastern US using the data from the VEMAP program, which provides interpolated climate time series data from 1895 at 0.5x0.5^o resolution. Due to discrepancies between the VEMAP 1 radiation data set and a spatial interpolation based on direct measurements within the northeastern US (Jenkins et al. 2000), we have developed and will use a summary relationship based on the limited number of actual observations made over time.

We will also develop data planes for N deposition and tropospheric ozone concentration. N deposition will be estimated using the NADP data base, with records for over 100 stations in the eastern US, along with the detailed measurements of speciated dry deposition at the EMS tower to confirm the assumed relationship between dry deposition and total (wet + dry) deposition. We will use three USEPA sources to derive a spatially explicit, monthly time-step representation of ozone dose over the Eastern US for the period of measurement: documents on mapping and forecasting ozone concentrations (EPA/454/R-99/009 and EPA/625/R-99/007), daily ozone forecasts from the Hysplit_4 model, and the AIRS data set (set of surface ozone measurements from 1982 to the present). The Northeast States for Coordinated Air Use Management (NESCAUM) also compiles and maps ozone concentrations that we will use. These regional estimates will be linked to the more detailed understanding of chemical, physical, and biological processes that are inferred from the detailed measurements at the EMS tower.

Observations and modeling of reactive nitrogen deposition, and monitoring and analysis of concentrations of selected trace gases that influence the energy balance or oxidant capacity of the atmosphere will complement the nitrogen addition, ozone impact, and carbon cycling studies.

D. SYNTHESIS AND INTEGRATION IN THE HF LTER PROGRAM

Over the history of the HF LTER program, synthesis of results and integration of interpretations have been achieved through: a system of regular science team meetings, the Annual HF Ecology Symposium and combined LTER and NIGEC (National Institutes of Environmental Change - DOE) workshop, joint lectures and articles (e.g., AAAS 1999; Foster et al. 1998), and regular collaboration among LTER co-investigators on other projects and agency proposals. Most importantly, synthesis and collaboration have been motivated by the results and conclusions that emerged from LTER I: information from apparently unrelated subjects (e.g., historical ecology and atmospheric chemistry) proved to be indispensable in interpreting modern ecosystem pattern and process. For example, the initial design of the N Saturation experiment and Environmental Measurement Station paid no attention to the details of history. Nonetheless, it became apparent that the results from each became interpretable only with a knowledge of 19^th C land-use activity and 20^th C impacts by hurricane and fire (Aber et al. 1998, Wofsy et al. 1993). Thus, exchange of information and integration of results have become essential parts of the design, methodological approach, and interpretation of all aspects of the HF LTER program. Recognition of this fact can be seen in the synthesis volume that has emerged (Foster and Aber 2000) and in the design of LTER III.

In LTER III, synthesis of activities is imbedded in all aspects of the project. Forest cutting, forest land conversion, insect outbreaks, hurricane impacts and past land-use are essential variables for studies of biological invasions, modern vegetation patterns, and conservation biology as well as the dynamics of nitrogen and carbon. Vegetation dynamics and ecosystem function will also be related to environmental variation on annual to century scales. Ultimately, the complete integration of these factors will occur through the development of models that predict ecosystem responses to complex patterns of land use and environmental change. In developing the basic models, generating input variables, and verifying output, we use our complete array of results from long-term experiments and measurements, land-use, disturbance and environmental histories, and mechanistic understanding of ecosystem processes. Thus, modeling activities not only enable us to extend our results in time and space but also represent an essential tool in the integration of our long-term research.