13^th Annual Harvard Forest Ecology Symposium - 2002

March 28^th, 2002

Long-term Suppression of Soil Respiration by Nitrogen Deposition: Contribution of Microbial Decomposition

C. Arabia*, E. Davidson, K. Savage,P. Steudler and R. Bowden

An increasing number of studies show alterations in soil respiration due to nitrogen additions to soil. At the Harvard Forest Chronic N Study, red pine and mixed deciduous forests showed immediate changes in soil respiration following nitrogen additions (50 and 150 N ha^-1 yr^-1) during the initial year (1988) of the study (Fig.1). The pine site showed an immediate decline in soil respiration rates that were also observed in the second year. The hardwood site, in contrast, experienced an initial increase in soil respiration rates after N additions in the first year. This increase was attributed to poorer soil conditions in the hardwood site compared to the pine site; N additions were hypothesized to have increased either root or microbial activity, or perhaps both. In the second year of N additions, however, both sites exhibited unchanged or decreased rates of soil respiration with continuing N fertilization. Weekly measures of soil respiration conducted on the Chronic N plots during the summer of 2001 (Fig. 2.) show that long-term continuous nitrogen fertilization continues to reduce total soil respiration. The observed reductions in total soil respiration could be due to nitrogen additions that reduced the need for root activity, reduced microbial activity, or both.

To investigate the contribution of microbial decomposition to total soil respiration, a soil incubation experiment was conducted in the laboratory to measure CO₂ fluxes from root-free soil. Soils were collected at each of the Chronic N plots during the summer of 2001 and roots were removed by passing soils through a 5.6 mm sieve. Soils were then incubated in sealed jars at field moisture content and ambient air temperature for three hours. Gas samples were removed from the jars at 0, 60, 120, and 180 min after onset of the incubation and analyzed for CO₂ concentrations.

The pattern in average CO₂ respiration for the incubated soils were similar to those observed from soil respiration sampling conducted in the field (Fig. 3.) The similarity in patterns between soil incubation rates and field respiration rates suggests that nitrogen deposition within the Chronic N Experiment has affected the rate of microbial decomposition, thus contributing to the decrease in CO₂ production measured in the treated sites in the field. Reduced root respiration may also contribute to the decline in total soil respiration measured at the plots. Declines in forest productivity measured at both sites, as well as substantial tree mortality observed at the high N sites, are likely to reduce total root activity. Declines in aboveground and belowground productivity may indirectly reduce microbial decomposition by reducing organic matter inputs to soil microbes.

Carbon Storage Over 32 Years in a Temperate Deciduous Forest

A. Barker Plotkin and D. Foster

Permanent plots provide an empirical understanding of forest change over time, and are an invaluable part of forestry and ecological research (Zeide 2001). Such studies provide concrete stand development information, and can be applied to answer numerous ecological questions, many of which could not have been anticipated when the study was initiated. In this case, 32 years of measurements are applied to current questions of carbon sequestration in temperate forests.

Walter Lyford began measurements of a 7.1-acre area of red oak-red maple forest on the Prospect Hill tract of Harvard Forest in 1969. All trees over 2 inches were mapped on very large-scale (1 inch = 5 feet) hand-drawn maps, and included live and dead trees, stumps, windthrows and other features such as stonewalls, boulders, soil moisture and a damage boundary from the 1938 hurricane. In Summer 2001, we relocated and measured all trees, marking the fourth measurement cycle in the site's 32-year history. Diameter and canopy class were recorded for live trees, and condition, decay class, diameter, bole length and orientation were measured for dead stems. New trees >2 inches were added and mapped using triangulation (Boose et al. 1998). All of Lyford's original maps were digitized into GIS format.

Overall trends for the site show typical patterns for maturing forests. Basal area rose from 108 ft²/acre in 1969 to 144 ft²/acre in 2001, while stem density declined from 517 stems/acre in 1969 to 346 stems/acre in 2001 (Table 1). Red/black oak and red maple are the major species in the stand, along with 21 other species. Red and black oak comprised more than half of the total basal area throughout the measurement period, while red maple comprised about half of all stems.

The role of temperate forest in the global carbon cycle is currently a major research issue. The Lyford site data can be used to show how carbon sequestration changes through 32 years of stand development, and can complement shorter-term data collected by eddy-flux and vegetation measurements in a similar forest type at Harvard Forest. We plan to complete a carbon accounting of this site over time, including growth of existing stems, recruitment of new stems and mortality. Tree diameter measurements will be converted to total aboveground biomass using allometric equations. The role of coarse woody debris will also be explored using repeated measurements of dead and fallen trees on the site. The forest at the Lyford site initiated mainly after clearcutting old-field white pine in the 1890s. Therefore, this analysis will quantify carbon pools and flux between approximate stand ages of 70, 75, 90 and 100 years.

Boose, E. R., E. F. Boose, and A. L. Lezberg. 1998. A practical method for mapping trees using distance measurements. Ecology 79: 819-827. Zeide, B. 2001. Thinning and Growth: A Full Turnaround. Journal of Forestry 99(1):20-25.

Table 1. Tree density and basal area in the Lyford Grid, 1969-2001. Total basal area values are ft²/acre. Other species include sugar maple, striped maple, alder, shadbush, chestnut, beech, white ash, witch-hazel, black gum, red spruce, big-tooth aspen, black cherry, white oak, hemlock and elm.

	Percent Stems/acre				Percent Basal Area
Species	1969	1975	1987-92	2001	1969	1975	1987-92	2001
Red maple	54	55	54	47	30	30	29	26
Red/Black oak	20	20	19	18	53	55	58	59
Birch species	13	12	16	18	8	8	7	7
White pine	5	5	2	3	3	3	2	2
Other	8	7	9	13	6	5	5	5
Total (absolute values)	517	466	375	346	108	117	132	144

Thirty-two Years of Stand Development After Hurricane Disturbance

A. Barker Plotkin, Kristin Wilson* and D. Foster

Permanent plots provide an empirical understanding of forest change over time, and are key to understanding stand recovery after disturbance. This permanent plot study, established by Walter Lyford, records detailed spatial features, including a damage boundary from the 1938 hurricane (Fig. 1). The objective of this analysis was to examine long-term trends of recovery after moderate-severe wind damage. Specifically, we sought to determine whether the effects of the 1938 hurricane are still evident, in terms of forest stand structure and species composition. The patterns seen in this study are then coupled with measurements from an experimental hurricane site to explore stand development following hurricane disturbance.

Walter Lyford began measurements of a 7.1-acre area on the Prospect Hill tract of Harvard Forest in 1969. All trees over 2 inches were mapped, and included live and dead trees, stumps, windthrows and other features such as stonewalls, boulders, soil moisture and a damage boundary from the 1938 hurricane. The history of the site from 1900 was complied from Harvard Forest records. In summer 2001, we relocated and measured all trees, marking the fourth measurement cycle in the site's 32-year history. Diameter and canopy class were recorded for live trees, and condition, decay class, diameter, bole length and orientation were measured for dead stems. New trees >2 inches were added and mapped using triangulation (Boose et al. 1998). All of Lyford's original maps were digitized into GIS format.

Densities of windthrown trees are much higher inside the damaged area than outside (Fig. 1), supporting Lyford's original placement of the hurricane-damage boundary. Overall stem densities are higher in the damaged area. Both areas show a decrease in stem density over time with a much more rapid decrease in the damaged area. Basal areas are higher outside of the damaged area; however, both areas show an increase in basal area with a slightly more rapid increase in the damaged area. These trends support the idea that the hurricane set back the development of the damaged area.

These stand development patterns are coupled with measurements for the first 10 years after an experimental hurricane (Cooper-Ellis et al. 1998), extending the timeline of observation. This is a very preliminary analysis, and there are some important differences between the two sites. The experimental site was initially 70-year-old mixed-hardwood forest, whereas the damaged area in the Lyford site was younger and had more pine. Also, damage in the Lyford site was more severe (90-100%) than in the experimental hurricane (67%). Caveats aside, stand development follows a logical sequence for the combined dataset. Density initially drops sharply, increases up to 30 years after the disturbance, then declines as a result of competition (Fig. 2). The initial density of the experimental hurricane stand (age 70) matches the density that the recovering Lyford site is approaching at age 63. After the initial low, basal area increases over time at a steady pace. Continuing observations in the experimental hurricane site will test whether the development patterns seen at the Lyford site apply more generally to stand development following moderate-severe wind disturbance.

Boose, E.R., E.F. Boose, and A.L. Lezberg. 1998. A practical method for mapping trees using distance measurements. Ecology 79: 819-827.
Cooper-Ellis, S. M., D.R. Foster, G. Carlton and A. Lezberg. 1999. Forest response to catastrophic wind: results from an experimental hurricane. Ecology 80: 2683-2696.

Evaluation of Greenhouse and Ozone-Depleting Gases in Rural New England

D. Barnes, J. Elkins and M. McElroy

The Montreal Protocol on Substances that Deplete the Ozone Layer in 1987 and its subsequent London (1990) and Copenhagen (1992) Amendments mandated control measures on the production and sales of ozone-depleting substances (UNEP, 1985-1997). The majority of the substances, including CFC-11, CFC-12, and CH₃CCl₃, were scheduled for 100 percent reductions in production and sales by 1 January 1996 in developed countries. The success of the Protocol in the U.S. has yet to be established for the post-1996 ban years. Here we report on the recent urban pollution histories of both the U.S. and Europe and discuss the implications for the health of the atmosphere.

This study provides an independent measure of emissions from a major U.S. emitting region of ozone-depleting species for the years 1996-2001, the first six years after the full implementation of the Montreal Protocol. The measurements were taken every 24 minutes at Harvard Forest, MA, downwind of the Northeast urban/industrial corridor, including the greater metropolitan region of New York City. Using the well-documented EPA carbon moNO_xide emissions, which are reported on a per county basis (and whose accuracy we have previously established, see Barnes (2000)), we estimate the annual and seasonal urban/industrial emissions of CFC-11 (CCl₃F), CFC-12 (CCl₂F₂), and methyl chloroform (CH₃CCl₃) all on a per capita basis. Using the same spatial statistical methodology on the hourly measurements of the AGAGE site at Mace Head, Ireland, we likewise estimated the annual European emissions of the same three gases for 1996-2000.

Our results for the U.S. and Europe are presented in Figure 1, along with all other U.S. and European emissions estimates. Those of Bakwin et al. (1997), Derwent et al. (1998a and 1998b), and Ryall et al. (2001) are based on atmospheric observations. Those of McCulloch et al. (1994) and Midgley and McCulloch. (1995 and personal communication) are based on audited sales inventories of major producers of the three gases. The data indicate that on a per capita basis the Americans consume more than their European counterparts. The announcement of the Montreal Protocol and its upcoming bans precipitated a dramatic change in emissions, as users shifted towards substitute chemicals. From 1986-1996, the U.S. emissions of CFC-11, CFC-12, and CH₃CCl₃ dropped by 86%, 69%, and 87%, respectively. It had been predicted that, following the 1 January 1996 ban, some residual emissions of CFC-11 and CFC-12 would continue, due to leakage from long-term reservoirs of these two gases (refrigerants and hard foams, in particular). Methyl chloroform emissions, on the other hand, were expected to drop to zero in 1996 or soon after, due to the difficulty in keeping it stored and the prior behavior of users who tended to consume their CH₃CCl₃ within a year of purchase. As is indicated in Figure 1, only emissions of CFC-12 are tailing off as expected. Methyl chloroform emissions are continuing at rates much higher than anticipated, with significant implications for the OH radical calculations by atmospheric modelers. Most surprising is the increase in emissions of CFC-11 in the U.S. since the Montreal Protocol ban. Such a signal suggests that long-term reservoirs are larger than previously thought, that users stockpiled the substance in large quantities in anticipation of the ban, or that a black market has developed in violation of the ban.

Bakwin, P. S., D. F. Hurst, P. P. Tans, and J. W. Elkins. 1997. Anthropogenic sources of halocarbons, sulfur hexafluoride, carbon moNO_xide, and methane in the southeastern United States. J. Geophys. Res. 102: 15915-15925
Barnes, D. H. 2000. Quantifying Urban/Industrial Emissions of Greenhouse and Ozone-Depleting Gases Based on Atmospheric Observations. Ph.D. Thesis, Harvard University.
Derwent, R. G., P. G. Simmonds, S. O'Doherty, and D. B. Ryall. 1998a. The impact of the Montreal Protocol on halocarbon concentrations in northern hemisphere baseline and European air masses at Mace Head, Ireland over a ten year period from 1987-1996. Atmospheric Environment, 32: 3689-3702.
Derwent, R. G., P. G. Simmonds, S. O'Doherty, P. Ciais, and D. B. Ryall. 1998b. European source strengths and northern hemisphere baseline concentrations of radiatively active trace gases at Mace Head, Ireland. Atmospheric Environment, 32: 3703-3715.
McCulloch, A. P. M. Midgley, and D. A. Fisher. 1994. Distribution of emissions of chlorofluorocarbons (CFCs) 11, 12, 113, 114, and 115 among reporting and non-reporting countries in 1986. Atmospheric Environment. 28: 2567-2582.
Midgley, P. M., and A. McCulloch. 1995. The Production and global distribution of emissions to the atmosphere of 1,1,1-trichloroethane (methyl chloroform). Atmospheric Environment. 29: 1601-1608.
Ryall, D. B., R. G. Derwent, A. J. Manning, P. G. Simmonds, and S. O'Doherty. 2001. Estimating source regions of European emissions of trace gases from observations at Mace Head. Atmospheric Environment, 35: 2507-2523.
United Nations Environmental Programme. 1985. Vienna Convention for the Protection of the Ozone Layer, 1987 Montreal Protocol to Reduce Substances that Deplete the Ozone Layer Report, Final Report (New York), 1990 London Amendment, 1992 Copenhagen Amendment, 1997 Production and Consumption of Ozone Depleting Substances, 1986-1995, Nairobi, Kenya.

Legacies of the Agricultural Past in the Forested Present: An Assessment of Historical Land-Use Effects on Rich Mesic Forests

J. Bellemare

Rich mesic forests are a species-rich northeastern variant of the mixed mesophytic forest tyype of eastern North America. The community is a conservation priority in the Northeast due to its limited extent high diversity and numerous associated rare taxa. To develop effective conservation strategies for rich mesic Forests (RMF), it is critical to understand factors driving the community's distribution and variation. Despite a long history of agricultural land-use in the region and indications that species typical of RMF may be sensitive to human disturbance, there has been little previous investigation of historical controls on the modern distribution, structure and species composition of RMF. In order to evaluate the relative importance of historical and environmental controls on RMF variation, this study investigated the effects of past land-use on the composition and structure of RMF in an area of western Massachusetts with a well-documented agricultural history beginning in the late 18^th century with European settlement. During the 19^th century up to 80% of the landscape was cleared for pasture and farmland. With the regional decline of agriculture in the late 19^th century, natural reforestation has resulted in ~ 75% forest cover at present. Persistent compositional differences between continuously forested stands and post agricultural forests indicate that distribution patterns for many species still strongly reflect the agricultural landscape of the 19^th century, despite the current predominance of forest. A major factor driving modern RMF vegetation patterns is the ability and rate of colonization by forest herbs. In particular, species with seeds lacking morphological adaptations for dispersal and those which produce seeds with eliasomes to encourage ant dispersal are less frequent in secondary forests, apparently as a result of dispersal limitation. Results from this study underscore the conservation importance of large RMF sites in primary forest, as well as the role of relict forest herb populations outside primary forest, in landscape-level species recovery.

Hurricane Impacts Across the Yucatan

E. Boose, D. Foster, A. Barker Plotkin and B. Hall

The ecological impacts of hurricanes across the Yucatan Peninsula, Mexico were investigated using a simple meteorological model (HURRECON) developed at the Harvard Forest and a database of historical hurricane data (HURDAT) maintained by the U.S. National Hurricane Center. A total of 105 hurricanes over the period 1851-2000 were reconstructed to produce estimates of wind damage on the Fujita scale across the region. Results showed considerable variation in hurricane activity from year to year and from decade to decade, while at the half-century scale there was an increase in hurricane intensity since the mid-19^th century. Ninety percent of the hurricanes causing F1 damage or higher (on the Fujita scale) occurred in the months of August, September, and October. A strong spatial gradient in hurricane frequency and intensity extended across the region from northeast to southwest, resulting from the greater number of hurricanes to the north, the east to west movement of most hurricanes across the area, and the tendency for most hurricanes to weaken significantly after landfall (Fig. 1). For example, during the study period, northeastern parts of the peninsula experienced a minimum of one F3 hurricane, six F2 hurricanes, and thirty F1 hurricanes, while southwestern parts experienced no F2 or F3 damage and fewer than five F1 storms. Though a significant disturbance across much of the Yucatan, hurricanes may have shorter-lived and less severe ecological impacts than fire or human land use. However the interaction of these factors (e.g., fires following hurricanes) may be very significant and deserves further study.

Effect of Prolonged Summer Drought and Snow Cover on Soil Methane Oxidation

W. Borken, E. Davidson, K. Savage, P. Steudler and E. Sundquist

Well-drained forest soils are thought to be a significant sink for atmospheric methane. Soil moisture strongly controls the oxidation of atmospheric methane by limiting the diffusion of methane into the soil, resulting in a negative correlation between soil water content and methane oxidation rates under most non-drought conditions. However, little is known about the effect of water stress on methane oxidation in temperate forests during severe drought. We hypothesized that methane oxidizing bacteria decrease their activity when soil water content falls below a value where water stress to soil microorganisms becomes important.

We established three translucent roofs, 1.30 m above the soil surface, and three control plots, each of 20 m², in a mixed deciduous forest at Harvard Forest in the spring of 2001. Throughfall water was excluded for 12 weeks from July to September. CH₄ oxidation was measured weekly from May to mid-October and monthly during winter using four closed chambers in each plot. Mean CH₄ fluxes were similar between treatment plots before the rainfall exclusion was initiated (Fig. 1).

In the control plot, CH₄ oxidation increased as a natural summer drought progressed, reaching a maximum of 329 mg CH₄ m^-2 hr^-1, and then declined as rain rewetted the soil (Fig. 1). In the throughfall exclusion plot, CH₄ oxidation reached a maximum of 317 mg CH₄ m^-2 hr^-1 two weeks sooner than in the control plot and then decreased to a lower level during the following weeks, indicating water stress of methanotrophic organisms. The roofs were removed in September, and CH₄ oxidation decreased in both treatments when subsequent heavy rainfalls increased soil moisture.

The parabolic relationship between CH₄ oxidation and soil moisture content (r²=0.75) reveals an optimum volumetric water content of 0.13 cm³ cm^-3 in the exclusion plot (Fig. 2). Because of a less pronounced soil drought, only a linear relationship was found for the control plot. In summary, the expected negative correlation between soil water content and methane oxidation was observed in the control plots during the natural summer drought, whereas the enhanced drought induced by the rainfall exclusion resulted in a brief decrease in oxidation rates relative to the control.

The incubation of sieved mineral soil from the A-horizon at different water contents also revealed a quasi-parabolic relationship between CH₄ oxidation and gravimetric soil water content with an optimum at 0.23 g g^-1 (not shown). The incubated O-horizon consumed no or only very small amounts of atmospheric CH₄.

Snow cover is a diffusion barrier for atmospheric CH₄ depending on the thickness and condition of the snow cover. CH₄ oxidation decreased from 156 mg CH₄ m^-2 hr^-1 before snowfall to 19 mg CH₄ m^-2 hr^-1 five weeks after the first snowfall. Soil temperature at 10 cm depth dropped from 5.0 to 1.0^oC during this period. Removal of a 10 cm thick snow cover increased CH₄ oxidation by a factor of 8 compared to the uptake rate measured over the snow (4 mg CH₄ m^-2 hr^-1) in February. Our results suggest that the duration of snow cover as well as the amount of snow may affect the annual rate of CH₄ oxidation.

Historical Research and Archive Additions

J. Burk

As part of the coastal project, colonial records for Long Island at the New York State Library and Archives in Albany were researched for witness tree and agricultural land-use records. Data sets of 50 or more trees were collected for six towns, including Hempstead and Oyster Bay in Nassau County. The year range of records for the towns was from the mid-1600s to the 1720s and 1730s. To help interpret the land-use history of the Hempstead Plain, agricultural resolutions in the town records from 1655 to 1845 were researched and summarized. During this process, as part of a collaboration with visiting researchers from Belgium, early Dutch records relating to the history of the Albany area including maps and witness tree data were added to the archives.

The census and land use history database for every town in Massachusetts was completed. Data from the Umass/MaConnell 1985, 1971, and 1950 studies was added to the 19th century census records, which dated back to 1801. All agricultural, woodland, and lumbering data from these records were recorded, and copies of the records retained in the Massachusetts historical records collection.

The wildlife project was updated with data from recent years. Hunting, trapping, sighting, and accident records were collected from the Fish and Wildlife office in Westborough for several species including black bear, turkey, moose, deer, fisher, and coyote.

In response to changes in record management procedures at Harvard, all microfilm and paper backups of important Harvard Forest documents were transferred from the university archives in Cambridge to the Arnold Arboretum archives in Jamaica Plain. The Arboretum also donated copies of over 70 1938 Hurricane photos to the Harvard Forest.

A detailed summary of Harvard Forest land transactions from 1907 to the present and a table documenting the history and present status of the plantations were created as part of the ongoing management plan. A large collection of updated and historic plans and property documents was added to the buildings and grounds files.

Other additions to the Archives collections included over 40 research files from various projects, a large mounted map of 1938 Hurricane mill and salvage sites, historic MDC maps of the Prescott Peninsula, 1938 aerial photos of the Cape and Islands, and large sample collections from OEB and the University of New Hampshire. As the 1830s mapping project was completed, high quality copies of some town maps were added to the collection and a file was created for all digitized maps. Physical additions to the building included bookshelves for the thesis collection and a map cabinet.

Studies of Hemlock Woolly Adelgid Infestation on Decomposition

R. Cobb and D. Orwig

Eastern hemlock (Tsuga canadensis) is experiencing rapid and widespread decline due in part to the Hemlock Woolly Adelgid (Adelges tsugae Annand - HWA), an aphid-like forest pest. HWA has spread rapidly through the mid-Atlantic and Southern New England range of hemlock. Infestation within stands is rapid, though poorly documented, and results in gradual tree dieback and mortality over the course of five to ten years. Southern Connecticut has experienced extensive hemlock loss and profound change in the structure and function of these unique ecosystems. Hemlock forests tend to be cool and dark systems that have low species diversity and are buffered against diurnal and seasonal climate cycles. These characteristics of hemlock forests are altered as the dense canopy thins, light interception at the forest floor increases, and high light species become established.

We have been studying decomposition in infested forests for the last three years. Our initial experiment compared the relative impacts of feeding by HWA with altered microclimate associated with canopy thinning (described in Cobb et al., In review). In this study canopy thinning and associated microclimate change had an overwhelming impact on decomposition. Changes due to feeding alone could not be distinguished from site to site-foliar chemistry variability.

We have ongoing studies to develop our understanding of microclimate and species-structure change on decomposition in hemlock forests. Although HWA feeding has little impact on foliar chemistry, species change will have great influence on above ground foliar chemistry (Fig. 1). Black birch foliage has two-fold lignin and nitrogen concentration compared to hemlock and is likely to have rapid short-term decomposition and N release. We have initiated a 3-year comparative study of hemlock, black birch, and hemlock-birch mixed bags in hemlock and formerly hemlock dominated sites throughout Connecticut. We are matching these measurements with study of infested and uninfested hemlock, red maple (Acer rubrum), red oak (Quercus rubra), and mixed birch (Betula sp.) foliage at the Harvard Forest Prospect Hill stand.

To determine the extent of microclimate change within the soil profile and to quantify its impacts on decomposition we have applied a common-substrate technique to all of our study stands. We deployed cellulose paper at the surface and at the mineral-organic soil interface at all 11 study sites. Below ground mass loss was rapid in the most open stands and similar between low/moderate damage stands and stands with intact birch canopies (Fig. 2). Below ground mass loss was greater compared to the soil surface.

HWA infestation, like other canopy removal disturbances, results in dramatic and long-lasting changes in ecosystem processes. Decomposition is accelerated in infested stands, reaches a peak at maximum defoliation, and recovers to a level greater than intact hemlock forests in young black birch stands. These changes are likely to interact with community and biomass structural change to exacerbate undesirable impacts of HWA infestation like increased N loss to surface waters.

Cobb, R.C., Orwig, D.A., and Currie, S.C. 2002. The effects of hemlock woolly adelgid infestation on foliar decomposition in eastern hemlock forests of southern New England. Can. J. For. Res. (In Review).

Forest-Atmosphere Exchange Processes: Report on Activities 2001-2002

D. Fitzjarrald, R.. Staebler, R.. Sakai and M. Czikowsky

New data system. In early October of 2001, we replaced our Unix-based data acquisition system with a new Linux-based system. This system can collect an arbitrary number of serial streams and merge them into a single data file in real time, thus eliminating any synchronization problems. Every half-hour, the means, moments, covariances, spectra, auto- and cross-correlations are calculated using both block and running mean Reynolds averaging. The data can then be transferred to Albany through a network connection.

Subcanopy flows. We continue our research into the characteristics of subcanopy flows and their role in forest-atmosphere exchange. Our aims are: a) to improve understanding of the factors that produce and modify the flow of air in the understory; b) to investigate whether advective processes in the subcanopy may be a significant mechanism for transporting CO₂; and c) determine whether subcanopy may explain an apparent underestimation of nocturnal respiration fluxes by the eddy covariance method on calm nights.

In the fall of 2001, the third year of the "Draino" field studies was performed. Recent improvements include an array of high-resolution 2D sonic anemometers, and a system to sample the horizontal CO₂ field in an 80 m radius around the EMS tower. These were set up in an attempt to estimate horizontal advection terms in the immediate vicinity of the flux tower. A snapshot of the measured fields is provided in Figure 1.

As expected, estimating the advective terms (u dC/dx, v dC/dy) turned out to be a difficult task. Variability of the signal can be large (Fig. 2), especially at night when large concentration gradients exist. The magnitude can approach that of the eddy covariance flux.

Table 1 summarizes the statistics on the relationship between drainage flows and missed nocturnal fluxes of CO₂ (the "CO₂ flux deficit"). This deficit was calculated as the difference between the expected forest respiration rate, based on soil temperature, and the actually measured turbulent flux rate (eddy covariance at 30m above ground). Typically about a third of all nights are associated with drainage flows (flows from Prospect Hill under stable thermal inversions), and about half the nights display flux deficits. However, only about a third of the flux deficit nights are associated with drainage flows. Apparently, it may be premature to generically blame drainage flows for any fluxes that are missed by EC.

Climatological studies. Streamflow records can be found reaching to the beginning of the 20^th century. We ask whether streamflow exhibits a detectable change at the time of leaf emergence, as enhanced evapotranspiration competes for ground water. We identified three independent runoff characteristics that change with spring onset. The first is the date of the return of the precipitation minus runoff curve, averaged over 30-plus years, to pre-snowmelt pulse values (Fig. 3). This provides an indication of the transition from the snowmelt portion to the leaf-out portion of spring. The second characteristic lies in using daily runoff values to assess changes in streamflow recession following rainfall events. We would expect the streamflow recession time constant to decrease following leaf emergence due to enhanced evapotranspiration. Analysis of data at Wappinger's Creek, NY shows a shortening of the streamflow time constant concurrent with leaf emergence and development (Fig. 4). The third characteristic involves using 15-minute runoff data to assess seasonal changes in the amplitude of the diurnal streamflow signal observed in some watersheds. During the growing season, transpiring vegetation draws on groundwater that composes the baseflow of the stream. At night, groundwater replaces some of the water that is transpired during the day. Preliminary results (Biscuit Brook in the Catskill Mountains, NY; Fig. 5) indicate that an abrupt change in the diurnal amplitude of streamflow occurs at about the date of leaf emergence.

Table 1: Summary of nightly conditions. The last 4 rows refer to subsets of nights with CO₂ flux deficits.

	1999	2000	2001
Study period	28 Jul - 19 Nov	23 Aug - 24 Nov	9 Oct - 26 Nov
Number of nights with all data	70	74	46
Total number of nights with drainage flows	21 (30%)	27 (36%)	15 (32%)
Number with CO2 flux deficit (N)	38 (54%)	47 (64%)	19 (41%)
Number of these N with u* < 0.2 m/s	19 (50%)	20 (43%)	7 (37%)
Number of these N with winds from the Northern sector	27 (71%)	35 (74%)	13 (68%)
Number of these N with good radiative cooling	26 (68%)	30 (64%)	10 (53%)
Number of these N satisfying all 3 drainage flow criteria	12 (32%)	15 (32%)	5 (26%)

Distribution of Midge (Chironomidae) Remains in New England Lakes and Implications for Paleoclimate Reconstruction

D. Francis, E. Doughty, E. Drew*

The Little Ice Age was a cold climatic period that lasted from approximately AD 1450 to 1850. The colder conditions have been well documented in New England historical records and anecdotal evidence, because European settlement occurred about halfway through this event. This climatic oscillation may be responsible for vegetation changes observed in pollen records from around the northeastern U.S., namely a decline in hemlock and beech. In order to investigate the factors that could be driving these changes, a multiproxy paleoecological study of vegetation, climate, and human impacts over the last 1000 years has been undertaken using stable oxygen isotope studies, geochemistry, pollen, diatoms, and chironomid remains.

Chironomid analysis will be used to infer past temperature variation. In order to construct a calibration or "training" set of modern samples, surficial lake sediments were collected from 37 ponds on a transect from southern Connecticut (41.3° N) to northern New Hampshire (45.1° N). Associated environmental data were also collected, including surface water temperature, pH, conductivity, secchi depth, and sediment organic matter. Mean July air temperatures for each site were determined by extrapolation from weather station data.

A total of 72 chironomid taxa (genera and species groups) were recovered from the 37 lakes. The distribution of taxa show a distinct trend with latitude. Some taxa such as Heterotrissocladius and Tanytarsus lugens type, which are associated with colder water, were only present in the northernmost samples. Some warm water types were much more abundant at the southernmost sites, including Chaoborus and Tanytarsus sp. C. In a correspondence analysis, the axis-1 sample scores are negatively correlated with latitude, with an r² of 0.465. Using canonical correspondence analysis, a direct gradient analysis that constrains the ordination axes to be linear combinations of the environmental variables, mean July air temperature is the variable with the strongest explanatory power and has a strong correlation with axis-1. Maximum water depth is strongly correlated with axis-2.

The importance of temperature has been shown in other regions, including Atlantic Canada, British Columbia, Norway, Sweden and Switzerland. It is significant that the relationship of midge distributions to temperature holds true over a relatively short latitudinal and climatic gradient. This relationship is the basis for inferring paleotemperatures based on chironomid assemblages recovered from sediment cores. A calibration model (a transfer function that uses modern data to infer past temperature from core data) using this small data set would not be statistically robust; however, these data will be combined with other sites from Atlantic Canada and Baffin Island to produce an extremely robust and reliable model. Preliminary results from North Round Pond in New Hampshire, as well as some other core sites, show that chironomid-inferred summer water temperatures were cooler by approximately 1.3 °C during the Little Ice Age. The temperature reconstructions, along with the other proxy data, will be used to help interpret the vegetation changes that are recorded in the pollen records at these sites.

Old Growth Hemlock: The Annual Pattern of Carbon Exchange

J. Hadley

CO₂ exchange has been measured in the 200-year old hemlock stand at Harvard Forest for more than 16 months, using the eddy covariance method. The initial estimate of net carbon storage for the first full year of data (November 2000 through October 2001) was 3.1 metric tons C per hectare per year (Mg/ha/yr). This is more C storage than has been measured in recent years in deciduous forest 0.5 km to the east (Barford et al. 2001). However, the initial estimate for the hemlock stand will likely be reduced ˜5-30% by adjustments for under-measurement of ecosystem respiration, which are often needed in eddy covariance studies. Net carbon storage in the hemlock stand was the difference between two much larger estimated annual carbon fluxes: annual photosynthesis (˜13 Mg/ha/yr) and ecosystem respiration (˜10 Mg/ha/yr). The annual pattern of net carbon exchange showed that in 2001 the majority of annual carbon storage occurred in the spring and fall months. About 55% of all carbon storage from November 2000 through October 2001 occurred in April, May, and October (Fig. 1c), primarily as a result of low soil respiration associated with cool soil temperatures in these months. Near-freezing soil temperatures throughout most of April did not place any very strong restrictions on carbon storage. During late summer, particularly in August, high soil temperature and high soil respiration greatly reduced net carbon storage, even though high short-term rates of carbon uptake also occurred in August (Fig. 2).

These results indicate that carbon storage in hemlock stands would probably be greater in a climate with less extreme seasonal variation than is currently typical of central Massachusetts (i.e., a climate with milder springs and autumns and relatively cool summers). Thus the effect of climate warming on carbon storage will likely depend on when it occurs. Carbon uptake by hemlocks was particularly sensitive to sub-freezing temperatures in spring (Fig. 3). This effect was seen in photosynthesis measurements in an earlier, cooler autumn (Hadley 2000). Therefore, reduced diurnal temperature variation in spring and fall could be a strong positive influence on carbon storage.

The general pattern of carbon uptake and the effects of spring and fall frosts seen in eddy covariance data support the physiological model of carbon exchange by hemlock developed in the late 1990's (J. Hadley, Harvard Forest symposium abstracts for 1999 and 2000). The model estimated however that annual C storage from October 1997 through September 1998 was only about 1 Mg C/ha/yr, much lower than the more recent estimate from eddy covariance data. New data show that leaf area index (LAI) in the model (3.5 from October through May, 4.4 in summer) was probably about 20% too low. The next step in model testing will be to run the model for the period with eddy covariance data, and with higher LAI estimates, which will increase estimated carbon storage.

Barford,C.C., S.C. Wofsy, M.L .Goulden, J.W. Munger, E.H. Pyle, S.P. Urbanski, L. Hutyra, S.R. Saleska, D. Fitzjarrald and K. Moore. 2001. Factors controlling long- and short-term sequestration of atmospheric CO₂ in a mid-latitude forest. Science 294:1688-1691.
Hadley, J.L. 2000. Effect of daily minimum temperature on photosynthesis in eastern hemlock (Tsuga canadensis L.) in autumn and winter. Arct. Antarct. Alp. Res. 32:368-374.

Historical Land Cover and Land-use Patterns in Massachusetts

B. Hall, G. Motzkin, D. Foster, M. Syfert and J. Burk

Several previous studies at Harvard Forest have documented the influence of historical land-use on forest structure and composition at local and regional scales (e.g., Prospect Hill, central Massachusetts, or Cape Cod and the Islands). Over this past year, we have gathered data that will enable us to extend these studies to evaluate historical changes in forest composition and distribution across the state of Massachusetts, a region with more varied physiography, climate, geology, and land-use. This broad-scale approach has direct conservation application and is now feasible as a range of remotely sensed data and physical and historical data layers have become more readily available.

Historical data from a wide range of sources have been incorporated into the HF GIS. For instance, boundary-trees from early-settlement survey records were gathered to assess forest composition prior to widespread alteration of the landscape by European settlers. Agricultural census data from the 1880s provides information on forest type and the age of stands that were cut. Information on forest structure in 1951 and 1971 has been tabulated from data originally developed as part of the MapDown project at the University of Massachusetts, Amherst. Data on structural and compositional variation in modern forests are derived from US Forest Service Forest Inventory and Analysis (FIA) plot data.

A series of town maps surveyed in 1830 has been digitized at the Harvard Forest and will be a critical component of our analyses. Most of these maps show forest cover, allowing investigations of the extent and geographic location of forests near the peak of forest clearance in New England. We will use 1831 tax evaluations to estimate the extent of forest cover for those towns where forests were not mapped in 1830.

Preliminary analysis of the surveyors' witness trees indicate that there were 5 general forest types at the time the Massachusetts towns were settled (Fig. 1). Cape Cod, the Islands, and the southeastern part of the state were dominated by pine-oak forests, with more wetland taxa in Plymouth and Bristol Counties than in the dryer coastal area. Hemlock and northern hardwoods such as beech, birch, and maple dominated the western part of the state, whereas oak-hickory forests with chestnut and pine were widespread in central Massachusetts. This variation in forest types is correlated with a complex environmental gradient involving concurrent changes in growing degree days (GDD), precipitation, elevation, surficial geology, and bedrock geology, with GDD explaining more of the variation in early historical forest composition than the other variables.

Primary Effects of Selective Harvest on Carbon Stocks and Stand Structure Adjacent to the Prospect Hill Tract

L. Hutyra, C. Barford, E. Pyle, D. Bryant, B. Curry,* J. Silvis,* R. Wofsy and S. Wofsy

During 2001, a selective harvest was conducted on a 43 ha private parcel immediately south of the Harvard Forest, Prospect Hill Tract. This provided an opportunity to investigate the impact of forest management on carbon (C) cycling processes. Initial baseline measurements were taken on the parcel and an adjacent control stand in order to characterize standing biomass and detrital C pools. Harvested wood was tracked from the landing to the mill to quantify the mass of C extracted in saw timber and firewood. These data provide a quantitative example of C pools and fluxes following a typical New England selective harvest. In addition, the parcel is currently being monitored to determine the effects of harvest on subsequent C cycling rates.

Thirty-three previously established circular plots (314 m² each), within the fetch of the EMS tower, were used as the control for the harvest experiment located directly to the south. Above ground woody increment (AGWI) data were collected weekly during the growing season, from the spring of 1999 through fall of 2001 on these and nine additional plots in the harvested site. Coarse woody debris (CWD) was measured on both sites in 1999 and again on the harvested site in 2001. Stand heterogeneity increased with harvest, therefore the plot size was increased to 706 m² to improve sampling precision. It was suspected that loggers may have avoided the research plots, making them no longer characteristic of the stand structure; increasing the size of the plots following the cut allowed testing for this bias.

Prior to harvest, the species composition (Fig. 1a) and standing biomass (Fig. 1b) of the harvest site differed slightly from the Prospect Hill plots, 81 vs. 97 tons C ha^-1, respectively. The biomass distribution also varied with the species composition. The harvested site is more dominated by beech (Fagus grandifolia), hemlock (Tsuga canadensis) and birch (Betula spp.). The CWD on the control and harvested sites was similar in 1999, at 7.5 and 9.3 tons C ha^-1, respectively.

During the harvest, a total of 42.8 m³ ha^-1 of wood was removed from the cut site, 15.8 m³ ha^-1 and 27.3 m³ ha^-1 of saw timber and firewood, respectively. This is consistent with typical harvest rates of 44.7 m³ ha^-1 for the Quabbin region (Kittredge pers. comm.). Based on mill estimates of wood density, this indicates approximately 26.9 tons C ha^-1 were removed from the site. At the end of 2001, 56.1 tons C ha^-1 of above ground live biomass, including the AGWI, was present on the site. The species composition did not change significantly after cutting (Fig. 1a). Further, comparison of the expanded plots with the nested interior plot showed no significant bias (Fig. 1a,1b).

Continued monitoring of these and other carbon pools and fluxes will quantify the C dynamics relative to an undisturbed stand. Further, concurrent measurement of net ecosystem exchange (via eddy covariance) at the EMS tower will allow for examination of the potential for C sequestration through forest management.

The Effect of Nitrogen Deposition on Alliaria petiolata and Indigenous Species Population Dynamics

S. Kaufman and F. Bazzaz

Alliaria petiolata is an herbaceous plant introduced from Europe that is invading forest understories across much of the United States. The rapid spread of Alliaria petiolata may be linked to increased levels of nitrogen deposition in northeastern forest ecosystems. If A. petiolata has a greater ability to utilize nitrogen or other resources, it may out-compete native plants in these ecosystems. This project experimentally investigates responses of indigenous plants and A. petiolata to increased and decreased nitrogen levels over two growing seasons. Twelve 1 m² plots are established in uninvaded forest understory and 12 plots are established in adjacent forest understory invaded by garlic mustard at Harvard Forest. Similar experiments are established at a preserve in Belmont, MA, at Bartholomew's Cobble in Sheffield, MA, and at Castle Hill in Ipswich, MA. The two treatments and controls of no amendments and no amendments + Alliaria removal are randomly assigned among the plots. Ammonium nitrate was added at the rate of 30 kg/ha/yr to increase nitrogen levels. To decrease nitrogen levels, 2000 kg/ha/yr of carbon in the form of sucrose was added. Treatments were added on eight dates from April 23 - October 11, 2001. No significant differences in soil C/N ratios (measured using the Harvard Forest Fison autoanalyzer) were detected among plots before or after amendments however. Carbon and nitrogen additions will be changed in 2002. Species abundance and percent cover of each plant species were determined August 4, 2000 for each plot prior to the addition of treatments, and were again measured on May 10 and Sept. 5, 2001. At the first two census dates, invaded plots have slightly higher total species richness, species diversity (calculated using the Shannon-Wiener index) and native species richness. The Sept. 2001 sampling date shows virtually no difference in species richness or diversity between invaded and uninvaded plots. All plots have higher native species percent cover than exotic species cover. Mean A. petiolata cover was low in most invaded plots. The only other non-native species in the plots was Euonymus alatus. Light and soil water availability were also measured for each plot. The results at Harvard Forest will be compared with those from more heavily invaded sites. The information from this experiment, and from a complimentary common garden experiment at the Concord Field Station, will help to determine whether nitrogen deposition is changing competitive interactions between A. petiolata and indigenous plant species.

	August 4, 2000		May 10, 2001		September 5, 2001
	Invaded	Uninvaded	Invaded	Uninvaded	Invaded	Uninvaded
Species richness	13	9	11	9	11	12
Species diversity	1.73	1.61	1.44	1.40	1.53	1.56
Native species richness	12	9	10	8	10	11
Native spp. Mean % cover	14.33	23.50	17.92	17.0	21.25	25.42
Exotic spp. Mean % cover	< 5	< 5	< 5	0.42	3.33	0.45

The Spectrum of Forest Preservation to Management: Conservation in a Landscape with Diverse Ownership

D. Kittredge

In a world with increasing human population and decreasing forest, the demand for products and services from woodland grows more acute. Since wood products are traded globally, and the United States is a net importer of wood, questions of sources of wood are raised. In addition, forests are important sources of habitat for a wide variety of flora and fauna, and serve as refugia for biodiversity. Questions of forest conservation in an era of less woodland and more people boil down to deciding on where forest can be managed intensively for fiber, where it can be managed less intensively for timber and other products and services, and where it can be managed for certain preservation values where human influences are minimized, such as wildland recreation and interior forest habitat. Some people refer to the latter as wilderness, or forest that is preserved. The other two management regimes vary in their intensity, and variations on the three conditions combine to form a so-called spectrum of conservation.

It is both an illusion and a luxury to expect that we can do without any one element of the spectrum. The question is no longer whether we need preservation lands at the expense of productivity, but where to locate all activities of the spectrum in landscapes with growing human populations. Answering this question of where to implement varying levels of conservation both abroad and "at home" in our own backyards is a biogeographical question, depending on forest productivity (i.e., where is there potential to grow large amounts of wood per hectare with nominal disturbance?) and biodiversity questions of habitat and rarity at both species and community levels. The question is further complicated when overlaid by the current human landscape features of development, transportation networks, political boundaries, and property ownership. Indeed, this latter factor, more than others, is of critical concern, due to the private owner rights and various public agency mandates that influence activities on any given hectare. In addition, land protection activities by both the state and ngos result in forest that is restricted from development or other use to varying degrees. Lastly, the question of where to focus various activities along the conservation spectrum is complicated by the fact that private ownership changes- for example, it is estimated that 82% of Massachusetts forest owners acquired their land between 1980 and 1994, indicating a significant turnover rate (Birch 1996). In short, there is a kaleidoscope of conditions and circumstances that shift over time and space that influences the spectrum of conservation. Changing ecological, political, sociological, and organizational influences combine to influence where and the extent to which various forms of management are practiced.

Massachusetts is the third most densely populated state in the nation, and by area, the 8^th most heavily forested. It is logical to explore in a Massachusetts context how the spectrum of conservation may be applied to meet local needs. We extend our work in the North Quabbin landscape of central Massachusetts to explore the extent to which the spectrum of conservation is being applied, and the potential for a coordinated application of it in the future, to ensure the variety of human benefits. Previous work (Golodetz and Foster 1997) showed that within this 168,312-ha area land protection activities between 1900 and 1993 resulted in 37% of this heavily forested region (i.e., 79% by area) being protected from development, either through fee simple state acquisition or the acquisition of development rights alone via easement, leaving the land in private hands. Between 1993 and 2001, an additional 2,370 ha of land were protected from development. 67% of the NQ landscape is considered either a core area or supporting natural landscape based on the BioMap analysis of the Massachusetts Natural Heritage and Endangered Species Program. 53% of the NQ forest area is currently managed for multiple-use timber production, either on state lands (three agencies) or private lands enrolled in the current use property tax program which requires timber production in return for a reduced assessed value. In addition, analysis of timber harvest regulatory data indicates that regardless of ownership management objective identified in a management plan or by agency mandate, 1.5% of the region's forest sustains a harvest annually, resulting in 26% of the forest being harvested at least once between 1984 and 2000. On the other end of the conservation spectrum, preliminary data suggest that 5,504 ha of forest are set aside as actual preserves (i.e., 4.1% of all forest). We anticipate that spatial analysis of the variety of activities occurring along the conservation spectrum in this forested region will allow for the development of alternative strategies to consider, and varied policy tools that can be used to achieve these strategies.

Timber Harvest as a Form of Disturbance in Massachusetts

D. Kittredge, J. Burk, D. Foster and G. Motzkin

Based on a pilot study of commercial timber harvesting in the 19-town North Quabbin region, we have continued gathering spatial data on harvest activity throughout Massachusetts. Specifically, we use proposed cutting plans filed with the state forestry agency in compliance with Chapter 132 - the Forest Cutting Practices Act - as a source of spatial and related data to describe the extent to which the forested landscape undergoes commercial harvest. These data are filed for every harvest activity greater than 60 m³, regardless of landowner or area. Though they are prepared for regulatory purposes, and may not have the precision or accuracy of data generated for research analysis, they nonetheless provide the best possible indication of broad scale pattern and intensity of harvest.

We have to date reviewed data on commercial harvest for the last 5 years, for 4 of 5 administrative regions of the state, representing roughly 80% of Massachusetts. All that remains to be captured is the area to the north and east of the Boston metropolitan area. Thus, we believe we have covered almost all of the predominantly forested areas of the commonwealth. Preliminary results so far indicate a total of 54,112 ha harvested (roughly 10,800 ha harvested annually) in 3,153 operations, with a mean harvest area of 17.4 ha per occurrence. Harvest operation size ranged from 0.1 ha to 364 ha. Overall estimated harvest intensity is 16 m³/ha. Recent statewide inventory estimates suggest an average of 36.2 m³/ha, thus commercial harvest operations remove approximately 44% of standing volume.

Harvest on non-industrial private ownerships represent 83% of all operations, in keeping with the predominant role they play in the landscape (i.e., non-industrial private forest represents 78% of total forest area). Interestingly, harvesting by the Metropolitan District Commission (MDC) on state watershed lands (total ownership of ca. 40,470 ha statewide) represents 7% of total operations, while Department of Environmental Management (DEM) harvests represent less than 2% of operations though they are responsible for 119,390 ha of state forest lands. Unlike other disturbance regimes, ownership seems to influence the likelihood of disturbance.

While we have not yet analyzed the data spatially, we know from previous work in the north Quabbin pilot area that harvest is essentially random with respect to physical and biological factors such as topography, soils, and forest type. A simple review of the plotted data suggests the absence of a pattern continues for the state as a whole. Virtually all of western Massachusetts (i.e., Worcester county west) experiences the same intensity of random, small harvest occurrences over time and space. This intensity of occurrence tapers off as one travels east, towards the more densely populated influence of metropolitan Boston.

We intend to use these data to confirm patterns disclosed earlier in analysis of the pilot area. We furthermore will use them to explore disturbance regimes in areas of greater population density, as well as in heavily forested areas outside the original pilot area (e.g., the Berkshire hills).

Vegetation and Ecosystem Response to Eastern Hemlock Decline and Logging: Direct and Indirect Consequences of the Hemlock Woolly Adelgid

M. Kizlinski, D. Orwig and D. Foster

Infestation by exotic pests and pathogens represents a significant agent of ecosystem perturbation that often leads to selective removal of the species and stimulates unprecedented logging of the target species. Such opportunistic logging may proceed with little knowledge of its short or long-term effects, and may generate more dramatic ecosystem disruption than the pest or pathogen itself. The recent unimpeded migration of hemlock woolly adelgid (Adelges tsugae, HWA) across the eastern U.S. is generating widespread mortality and stimulating intensive preemptive and salvage logging of eastern hemlock (Tsuga canadensis). This study compares the magnitude and trajectory of vegetation and ecosystem dynamics following HWA infestations with logged and healthy hemlock forests at ten sites located in Connecticut and Massachusetts.

Logging generated more rapid and pronounced microenvironment and vegetation changes than HWA damage. Percent open sky was 25 - 35 % in newly cut (i.e. = 3 years old) forests, closer to 10% in HWA infested forest, and < 5 % in older cuts (i.e. = 7 years old) and healthy stands. Black birch (Betula lenta) seedling densities and percent cover of brambles (Rubus spp.), sedges (Carex spp.) and hay-scented fern (Dennstaedia punctilobula) were significantly higher in cut stands versus HWA damaged and healthy stands. High black birch sapling densities (> 8000 ha -1) were common in the oldest cuts but not in adjacent, HWA-damaged portions of these stands. Residual trees had little influence on the species composition of emerging vegetation in logged sites.

Forest floor mass was significantly different among treatments, as healthy stands had 20% more mass than HWA damaged stands and 50% more mass than older cuts. Mass loss rates of cellulose paper suggest that conditions were more favorable for decomposition in the damaged and older logged stands, although laboratory incubations suggest substrate quality diminishes in these forests over time compared to healthy forests. Recently cut stands had significantly larger inorganic N pools than healthy forests, although total net nitrogen (N) mineralization rates were not significantly different among treatments. As harvest age increased, nitrification accounted for a larger proportion of total N mineralization. Nitrification rates of 0.2 kg ha^-1 day^-1 measured in the oldest cuts were almost three times greater than in HWA damaged stands and over 200 times greater than in healthy hemlock stands. However, the amount of nitrate captured on resin bags in the oldest cuts was similar to the amounts captured in healthy and damaged forests, suggesting that the excess nitrogen was being utilized in vegetative uptake. In contrast, large amounts of ammonium and nitrate captured in new cuts indicate higher N availability, less vegetative uptake, and the potential for N leaching.

Results suggest that both the decline associated with HWA infestation and the indirect effects of logging are generating profound changes in structure, composition, and ecosystem function in these forests. In addition, intense logging of healthy stands may lead to greater N losses prior to vegetation establishment than losses associated with HWA infestation or logging of deteriorated hemlock stands.

Defense Chemistry in Invasive Plant Species: What Role Does it Play in Invasiveness?

K. Lewis

We are investigating the evolutionary and phenotypic changes in defense chemistry that occur in invasive species in their new ranges in response to escape from herbivory. We have selected three species: Alliaria petiolata, Hesperis matronalis, and Cardamine impatiens, all from the family Brassicaceae and all introduced from Europe and invasive in Massachusetts. Plants in this family produce defense compounds in the form of glucosinolates, which are high in nitrogen and sulfur, two often limiting elements in plant communities.

Specifically, we are determining if invasive populations differ significantly from original populations in defense chemistry, defense allocation (inducible or constitutive defenses), and reproductive capacity. We are determining the role of resource availability in geographic patterns of chemical defense expression. Finally, we will determine the proportional contribution of genetics, abiotic environmental factors, history of herbivory, and intergenerational (maternal) effects on defenses in these plants.

Harvard Forest is not heavily invaded by these species; however, both A. petiolata and H. matronalis occur on or near the roadway that passes through the forest behind Benson house. We are performing herbivore exclusions on some of the A. petiolata plants and collecting leaves and seeds from both species behind Benson house for use in the comparisons described above. Furthermore, we are also comparing the defense chemistry of these invading crucifers with some of the native crucifers to understand how invading species avoid heavy herbivory by generalist herbivores already adapted to feeding on glucosinolate-producing plants.

Megaplots, Warming and Ecosystem Carbon Balance - The Barre Woods Project

H. Lux, F. Bowles, S. Morrisseau, J. Melillo and P. Steudler

Two of the major results of our original soil warming study were that: (1) warming stimulated the decay of a labile soil carbon pool; and (2) it also increased the availability of inorganic nitrogen to plants. Because of the small size of the original warming plots an important question we have not been able to answer is: Has the increase in available nitrogen led to an increase in carbon storage in the vegetation? And if yes, how much? In a systems context an additional question is: What is the balance between the carbon lost from the soil and the carbon stored in the vegetation in response to soil warming?

We are now addressing these important questions with a new warming experiment in the Barre Woods area of the Harvard Forest. During the summer and fall of 2001 we:

• Established two 30 x 30m plots (1 to be heated, 1 control)
• Gridded the plots and mapped all trees > 5 cm DBH (Fig. 1)
• Fitted all trees >7 cm DBH with dendrometer bands
• Cored a subset of trees to establish historical growth rates by species (Fig. 2)
• Installed 9 gas sampling collars and 6 tension lysimeters in each plot (locations indicated in Fig. 1)
• Made some initial gas measurements (Fig. 3)
• Brought line power and phone lines to the site
• Installed 3.4 miles of heating cable in the plot to be heated
• Built a "control shed", installed control equipment and completed all wiring

In April 2002, we will start the new study with an emphasis on the soil-plant coupling in response to warming. We want to determine whether or not warming results in the movement of nitrogen from the soil to the trees and how this movement affects the net carbon balance of the ecosystem.

Chronic Nitrogen Additions to Two Forest Stands

A. Magill and J. Aber

The nitrogen saturation potential of temperate forests in the Northeastern United States and in Europe is of mounting concern as nitrogen deposition from anthropogenic sources continues to increase. In order to determine the long-term impact of N inputs, key indicators of nitrogen saturation, such as forest productivity, foliar chemistry, soil net mineralization rates and soil solution chemistry, have been measured on plots receiving NH₄NO₃ additions at the rate of 0, 5 and 15 g N m^-2 yr^-1 since 1988. A nitrogen plus sulfur treatment was also included through 1998 with NH₄NO₃ additions at the same rate as the low N plots plus 7.4 g N m^-2 yr^-1 of sodium sulfate. Beginning in 1999, the N+S plots received only the NH₄NO₃ additions. Plots were established in two forest stands, a red pine plantation and a mixed hardwood stand, located in the Prospect Hill tract of Harvard Forest.

Root biomass cores were collected in 1999 to a depth of 20 cm. Data from these reveals that the largest percentage of fine roots (kg roots ha^-1 soil) were found in the 10 - 20 cm depth. However there were no significant differences in root mass between treatments. Root percent N in the high N plots was significantly lower than controls for mineral horizon samples in both stands. Small sample size made analysis of O horizon root samples impossible although we are planning to combine samples and re-run them this year. Soil water nitrate concentrations remain elevated in the high N treatment plots of both stands, as does green foliar nitrogen content. Data currently under analysis include fungal and microbial biomass, and zero tension lysimeter cation concentrations. A major field season is planned for 2002.

Soil Respiration and Response to Soil Temperature over 11 Years of Above- and Belowground Litter Manipulations

P. Micks and K. Nadelhoffer

Year 2001 began the second decade of the Harvard Forest DIRT Experiment (Detritus Input and Removal Treatments), a long-term study of linkages between forest soil organic matter dynamics and above- and belowground plant inputs to soils in a mixed hardwood forest ecosystem. The goal of the DIRT project is to assess how rates and sources of plant litter inputs control the accumulation and dynamics of organic matter and nutrients in forest soils over decadal time scales. Results demonstrate the relative importance of above- and belowground sources on soil organic matter dynamics, and also show emerging non-linear changes in carbon release and storage.

Treatments established in 1990 are: doubling aboveground litter (Double Litter), exclusion of aboveground litter (No Litter), exclusion of root inputs by trenching (No-Roots), and exclusion of aboveground litter and root inputs (No Inputs), on replicated 3m x 3m plots (n=3 for treatments, 6 for controls). The O/A-less treatment was implemented in 1991 to track the recovery of impoverished soil by replacing O and A horizon soil with B horizon material and allowing normal litter inputs to occur thereafter.

Since year 1, roots have exerted a stronger influence on soil respiration than above-ground litter. We continue to observe a decline in labile C resulting from exclusion of root inputs to the soil as soil C becomes more recalcitrant with time. Respiration from both No Roots and No Inputs decreased from 60% to 32% of controls over 11 years. Respiration from the root-excluded treatments is nearly the same whether or not above-ground inputs are allowed. CO₂ efflux from No Litter plots remains below controls, not changing appreciably relative to the controls throughout the 11 years. Doubling above-ground litter initially increased CO₂ efflux and soil C storage, but after year 8 CO₂ efflux dropped to near or below control levels even as soil C content continued to rise. This indicates that the proportion of labile C has decreased below the control and demonstrates a long-term negative effect on decomposition. Such changes in soil microbial activity could increase long-term soil C storage beyond expectations. The impoverished soil continues to recover, with respiration rates climbing from a low of 30% to 47% of controls. In year 11, CO₂ efflux from this treatment surpassed that of the root-excluded treatments.

We examined soil respiration response to soil temperature for years 4, 5, 10, and 11, finding changes among treatments over time (Fig. 1). Q₁₀ values (Table 1) for Double Litter fell steadily, indicating declining decomposition rates, and were below No Litter by year 10. No Litter Q₁₀ changed little over time. No Roots and No Inputs responses to temperature were similar in years 4 and 5, then subsequently decreased, but No Inputs Q₁₀ decreased more than No Roots despite the fact that its respiration rates are similar to or greater than No Roots'. Q₁₀ for these treatments decreased more than any treatment, dropping below 2 in year 11. -OA soil Q₁₀ showed a rising trend over time, and was higher than the root-excluded treatments' Q₁₀ in year 11, as were its respiration rates.

Our results demonstrate that soil organic matter dynamics respond non-linearly to changes in above- and belowground plant inputs. Changes in these inputs and their influence on temperature-controlled processes will be significant in determining the effects of a warmer world on the net flux of carbon from soils to the atmosphere.

Impacts of Aerosols and Clouds on CO₂ Uptake over Harvard Forest

Q. Min

A fundamental variable affecting carbon uptake by ecosystems is the amount of photosynthetically active radiation (PAR) -- not only the total amount, but also its spectral distribution, as well as its partitioning between direct and diffuse components. Taking advantage of simultaneous measurements of direct and diffuse spectral irradiance from a MFRSR, we are able to quantitatively study the issues of "diffuse use efficiency" and water use efficiency and to catalog the impacts of aerosols and clouds through photosynthesis on CO₂ uptake.

We have developed a simple and fast retrieval algorithm for synthesizing spectrum from multifilter narrowband instruments and applied it to predict the true PAR for both direct and diffuse components by convolving with a normalized action spectrum of Chlorophyll-a. Validations based on collocated the MFRSR and the RSS, demonstrated the accuracy under various sky conditions (Fig. 1). Based on the measurements during the growing season of 1998, we characterized the atmospheric conditions over Harvard Forest, and illustrated that probability distributions of optical depth for both thin and thick overcast clouds exhibit gamma distributions (Fig. 2). The occurrences of thin clouds are much higher than that of overcast evens, due to forest induced boundary layer cumulus clouds.

By combining observation and modeling, we found optical properties of aerosols and clouds have significant impacts on photosynthesis not only through the partition of direct and diffuse components but also through the changes of spectral distribution (or light quality). Figure 3 shows the statistics of direct and diffuse PAR of Chlorophyll-a against optical depths of aerosol and cloud under three categories of clear-sky, thin clouds, and thick overcast clouds for different solar airmasses or solar zenith angles. For a given optical depth, the atmosphere loaded with aerosols has higher direct PAR than that of thin clouds, but has lower diffuse PAR than that of thin clouds. Further, the incident angle of photons reaching forest canopy, associated with solar zenith angle, affects how photons penetrate through forest and possible saturation of leaf response to light, resulting in the different impacts on carbon uptake. As demonstrated, simultaneously measured direct and diffuse "true PAR" is needed for study beyond simple comparisons between cloudy and clear days and to directly address the canopy photosynthetic differences between diffuse and direct radiation. However, unlike Harvard Forest equipped with a MFRSR, most sites only measure the global PAR and reflection of the PAR using LICORs. Figure4 shows the comparisons of the true PAR from measurements of a MFRSR with the global PAR measurements of a LICOR during the growing season of 1998 at Harvard Forest site. For a given global PAR (LICOR), the direct and diffuse partition varies in a fairly large range. This means that a simple radiation partition model might result in significant errors in the partition between diffuse and direct PARs.

We also derived "diffuse use efficiency" based on maximum correlation between CO₂ uptake and effective true PAR based on . Figure 5 shows scattergrams of direct, diffuse, and effective PAR vs. measured CO₂ uptake and water use efficiency (WUE) under different sky conditions. The "diffuse use efficient" coefficients are 1.95 and 3.40 with maximum correlation coefficients of , 0.673 and 0.768, for aerosol and partial and thin clouds, respectively. This means that under partial and/or thin cloud conditions the enhanced diffuse irradiance due to clouds more efficiently penetrates into the forest, resulting in enhancement of carbon assimilation. Further, the straight lines in the Figure 5 are linear fits between the effective true PAR and carbon uptake. The slopes of these lines, representing the efficiency of carbon uptake, are 0.401, 0.409, and 0.495 for clear-sky, partial and/or thin clouds, and overcast clouds, respectively. The efficiency of carbon uptake under overcast clouds is substantially higher, over 20% more than aerosol and thin and partial cloud conditions. Combining both "diffuse use efficiency" and efficiency of carbon uptake, e.g. using total true PAR (simply direct plus diffuse true PAR) instead of effective true PAR, we may conclude that the presence of clouds enhances the efficiency of carbon uptake. As shown in the Figure 5, the WUE is also significantly enhanced as atmospheric conditions changed from clear-sky to overcast cloudy. Therefore, changes in many other factors such as temperature, moisture, and latent heating, precipitation, due to the presence of clouds have direct and/or indirect influences on carbon assimilation.

Microwave Land Surface Emissivities over Harvard Forest During the Growing Season of 1998

Q. Min and B. Lin

Understanding interactions between vegetated surfaces and the atmosphere is critical to our ability in modeling and predicting the earth's climate on local, regional, and global scales. We have exploited the possibility of using SSM/I to estimate leaf moisture or plant water content over a heavily forested area, Harvard Forest, since it is an important geophysical parameter to link the atmospheric, hydrological, and biospheric systems. We combined surface observation and coincident SSM/I radiance to retrieve the microwave land surface emissivity during the growing season of 1998, and further studied the implication on forest-atmospheric exchanges.

Figure 1 shows the time series of estimated microwave land surface emissivity, surface albedos at 415 nm, 862 nm and total shortwave albedo during the growing season of 1998. The cross represents each SSM/I sampling of the emissivities with vertical polarization at 19 GHz, and solid red line is the Lowess smooth fitting. The surface albedos at 862 nm and total shortwave, which are correlated with the LAI and the normalized difference vegetation index (NDVI) (Sakai et al. 1997), show a rapid change during spring due to leaf emergence, and then gradually decrease throughout the rest of the growing season. The microwave land surface emissivity decreases before leaf emergence occurs (Fig. 1), corresponding to the increase of soil moisture due to the rainfall and snow meltdown. Due to missing data in the longwave radiation (thus the skin temperature), we cannot know the behavior of the MLSE during the transit period of spring. After surface albedo (and LAI) reaches the maximum, over 90 % of the footprint of SSM/I pixel is covered by forest leaves. The MLSE shows the same trend of decreasing as surface albedo at 862 nm with periodic modulations. The decreasing trend of the MLSE indicates increasing moisture over the top layer of the forest, which decouples the trend of measured soil moisture underneath the forest and specific humidity in the air (Fig. 2). Thus, MLSE is likely an indicator of leaf moisture instead of soil moisture in this forest area. There was no evidence of water stress during the growing season of 1998, because of the buffering effect of deep soil moisture. The deep roots of trees uptake water to leaves through the plant water system, increasing plant water content to sustain evapotranspiration to the atmosphere. The increasing plant water content may also partially contribute to the decreasing trend of surface albedo during the growing season, along with possibilities of changes in leaf anatomy and canopy resistance (Moore et al. 1996, Sakai et al. 1997). Increasing water content conserved in the leaf will reduce the stomatal resistance and enhance the green matter production for growth. The more water, the lower the surface albedo.

Figure 3 shows the polarization difference and diurnal variation of emissivity at 19 GHz. There was no significant trend in polarization difference during the growing season, except for periodic modulations associated with synoptic system consistent with the MLSE. The periodic modulations of the MLSE and polarization difference are associated with synoptic systems and precipitation. However, diurnal variation (difference of the MLSE between the morning pass and the evening pass) shows a decreasing trend after leaf emergence. During nighttime, root uptake, dewfall, condensation, and guttation increase leaf wetness and lower the morning MLSE values. In the afternoon, evapotranspira-tion to the atmosphere depletes the interception storage and plant water storage along with ground surface and soil moisture storages, resulting in the higher MLSE. Therefore, without precipitation the diurnal variation of the MLSE is negative, and an indicator of the magnitude of the evapotranspira-tion process. Furthermore, the decreasing trend of diurnal variation indicates acceleration of evapotranspiration process due to the maturing leaves and decreasing of moisture in the atmosphere as the growing season proceeds. In order to sustain evapotranspiration to the atmosphere, the forest conserves more water in the plant water system (in drier environments), resulting in higher plant water content.

Moore, K. E., D. R. Fitzjarald, R.K. Sakai, M.L. Goulden,J.W. Munger, and S. C. Wofsy. 1996. Seasonal variation in radiative and turbulent exchange at a deciduous forest in Central Massachusetts, J. Appl. Met. 35: 122-134.
Sakai, R., D. R. Fitzjarrald, and K. E. Moore. 1997. Detecting leaf area and surface resistance during transition seasons, Agric. For. Meteor. 84: 273-284.

Eleven Years of Soil Warming on Prospect Hill - Do We Keep the Heat On?

S. Morrisseau, H. Lux, J. Melillo, P. Steudler and F. Bowles

We have now completed eleven years of our soil warming study at the Prospect Hill site in the Harvard Forest. Our research focused on a two-part question during the eleventh growing season: Did warming stimulate either CO₂ flux or net nitrogen mineralization?

In 2001, we observed relatively low CO₂ emissions and net nitrogen mineralization rates (Figs. 1a,1b). The CO₂ fluxes were the lowest they have ever been in the heated plots. We measured a flux of 507 g C m^-2 y^-1 in the control plots and 477 g C m^-2 y^-1 in the heated plots. Only once before in 1999 have emissions from the control plots exceeded emissions from the heated plots, resulting in a small percent decrease in the heated versus control plots (Fig. 2a).

The annual relative differences in CO₂ emissions between the heated and control plots are presented as three-year running means in Figure 3a. The overall pattern is suggestive of exponential decay of a labile carbon pool.

To preserve the plots, we did not measure mineralization in either 1999 or 2000. Annual net nitrogen mineralization in 2001 were the lowest we have measured (Fig. 1b; note 1991 was not a full year). Combining the current year's data with over-winter rates from the past, we project that a mean of 41 Kg N ha^-1 y^-1 was mineralized in the heated plots and a mean of 33 Kg N ha^-1 y^-1 was mineralized in the controls from April 2001 to April 2002. The seven percent increase in the heated plots relative to the controls is the smallest stimulation of net nitrogen mineralization in response to heating we have observed during the eleven-year period (Fig. 2b).

The annual relative differences in mineralization rates between heated and control plots are presented as three-year running means in Figure 3b. This way of presenting the data masks the very small relative differences in mineralization rates between heated and control plots in 2001. What is the significance of this small difference? Is it an anomaly in the time series caused by a relatively dry year? Or is it foreshadowing the exhaustion of a mineral nitrogen pool that was created by warming? To address these questions we plan to continue net nitrogen mineralization measurements at the site for a twelfth year.

Incorporating Regional-Historical Perspectives into Conservation of Uncommon Plant Communities: a Study of the Northeastern Coastal Region

G. Motzkin, D. Foster, B. Hall and D. MacDonald

Although the importance of determining the history of uncommon communities is increasingly recognized as critical for conservation efforts, most studies have been restricted to investigations of individual sites, with little attempt to place local results within a regional context. Such a context is necessary to determine the degree to which local dynamics are characteristic of the community type targeted for protection and to determine geographic and temporal variation in disturbance regimes, species distributions, and ecosystem processes. We are investigating the history and development of upland plant communities in the northeastern Coastal Region, an area that supports numerous uncommon species and assemblages and is highly threatened by development. Land-use and natural disturbance history vary considerably across the region; Nantucket and Block Island were almost completely cleared for agriculture in the 17-early 20^th C., whereas extensive areas on Long Island, Cape Cod, and Martha's Vineyard were never cleared for agriculture. Modern community patterns strongly reflect this history of land-use; several characteristic or rare lepidopteran species are largely restricted to sites that were never in agriculture, whereas numerous rare plant species are most frequent on sites that were disturbed intensively for historical agriculture or other uses. Land-use history is also a primary determinant of distribution patterns of common species. Although recent conservation efforts have emphasized the introduction of prescribed fire, vegetation patterns have largely been determined by historical land-use practices rather than fire history, suggesting that prescribed fire alone may not be adequate to ensure the long-term persistence of uncommon species and assemblages.

Reactive Nitrogen at Harvard Forest: Update for 2001

J. Munger, C. Horii, A. Bright, J. Budney, S. Wofsy, D. Nelson, J. McManus and M. Zahniser

Total nitrogen oxides (NO_y = NO + NO₂ +HNO₃ + peroxyacetylnitrate (PAN) + aerosol nitrate + other organic nitrates), as well as the components, NO, NO₂, and PAN, are measured continuously at the Harvard Forest EMS. HNO₃ was measured during the spring and fall of 2000 by a tuneable diode laser instrument (Horii, 2001). HNO₃ deposition fluxes were computed as the product of HNO₃ concentrations and a deposition velocity (V_d) derived from wind speed data. This approach is analogous to the dry deposition inferential method (DDIM) (Matt and Meyers1993; Hicks et al., 1987; Meyers et al., 1998) that is used widely. The total NO_y flux was determined using eddy covariance.

Median diel cycles of the measured concentrations of NO, NO₂, HNO₃, PAN, and NO_y are plotted in Figure 1 for June, July, and August 2000, for Northwest and Southwest wind sectors, which are the clean air and polluted sectors, respectively. The measured flux of NO_y is shown along with the inferred flux of HNO₃, the parameterized NO and NO₂ deposition fluxes, and estimated PAN flux, which is estimated by comparison of PAN and O₃ removal rates.

During northwesterly flow conditions, the sum of individual reactive nitrogen species, S[NO_y]_i = [NO] + [NO₂] + [HNO₃] + [PAN], was within the cumulative errors on the measurements to the observed concentration of NO_y. HNO₃ accounted for roughly half of NO_y during the day in summer, with the remainder of the NO_y budget consisting largely of NO_x. The summertime concentration of PAN was on the order of 15% of [NO_y]. During northwesterly flows, NO_y fluxes were dominated by HNO₃ deposition. During southwesterly flow conditions, a larger fraction of the NO_y concentration and flux is in the form of unmeasured species. HNO₃ and NO_x contributed less than 70% of the total NO_y concentration; with summertime PAN accounting for 8 to 17%. HNO3 flux was typically between 40 and 70% of FNO_y. We postulate that hydroxyalkyl nitrates may account for the remainder of the deposition. Alkenes produced by vegetation (isoprene and monoterpenes) react with NO to produce these nitrates, which are highly soluble and would deposit like HNO₃.

Hicks, B. B., D. D. Baldocchi, T. P. Meyers, R. P. Hosker, Jr., D. R. Matt. 1987. A preliminary multiple resistance routine for deriving dry deposition velocities from measured quantities, Water, Air, and Soil Pollution 36:311-330.
Matt, D. R. and T. P. Meyers. 1993. On the use of the inferential technique to estimate dry deposition of SO₂, Atmos. Environ. 27A: 493-501.
Meyers, T. P., P. Kinkelstein, J. Clarke, T. G. Ellestad, P. F. Sims. 1998. A multiplayer model for interring dry deposition using standard meteorological measurements, J. Geophys. Res. 103: 22645-22661.

Regeneration Following Clearcutting of Red Pine Overstory - Year 12

J. O'Keefe

Measurements of regeneration following removal in 1990 of a 64-year old red pine plantation on the Prospect Hill tract were continued for the twelfth year in 2001. Species, height, origin and evidence of browsing were recorded for all woody stems on 50 milacre (1.13 m radius) plots established on a five-meter grid within the clearcut. A fenced exclosure was initially erected around half of the plots. The exclosure fence has not been maintained since year 5 because no evidence of significant differences in regeneration between the exclosure and the open area was found. Extensive mixed hardwood regeneration (generally less than 7 m tall) was cut back to the ground during harvest, ensuring at least initial dominance by sprouts.

Browsing in 2001 remained at very low levels (<2% of stems). As mean tree height continues to increase both the amount of browsing and the impact of browsing on future stand characteristics should remain low. Overall, our observations show that browsing has had little long-term impact during the regeneration of this stand. White ash, the most heavily browsed species, remains the most common species in the plots. After remaining quite stable over the past five years, in 2001 the overall stem density of tree species declined to 17,883 stems/ha, compared with 19,464 stems/ha, 19,414 stems/ha in 1999, 19,958 stems/ha in 1998, 19,414 stems/ha in 1997, and 20,696 stems/ha in 1996. The relative importance of major species has remained the same over the past six years (Fig. 1). In 2001, white ash (36.5%) remained the most numerous tree species, followed by red maple (26.9%), sugar maple (14.4%) and black cherry (9.4%). These percentages changed little from 2000. After a slight decrease in 2000, red oak increased slightly to 7.5% of tree stems in 2001, the majority of which were small seedlings. Overall, the percentage of stems that originated as seedlings rather than sprouts decreased to 19.3%, down from 23.1% in 2000, 23.4% in 1999, 25.4% in 1998, and 23.7% in 1997. The majority of these seedlings (55.7%) were white ash, most less than .5 m tall.

Mean stem height rose to 3.46 m, compared to 3.20 m in 2000, 3.24 m in 1999, 3.01 m in 1998, 2.92 m in 1997, 2.87 m in 1996 and 2.67 m in 1995. The resumption in mean height growth over the past year probably reflects low seedling establishment and mortality of seedlings and young sprouts less than .5 m tall along with continued growth of the taller stems. The tallest stems were 20 white ash, 15 red maples, 15 sugar maples, 6 black cherries, 5 pin cherries, 3 paper birches, and 1 trembling aspen greater than 7 m tall. Diameter at breast height (dbh) is now being recorded for all stems taller than seven meters. Of the five most common species, sugar maple had the tallest mean height (4.75 m), followed by red maple (3.98 m), black cherry (3.69m), and white ash (2.99 m). Because of the preponderance of small seedlings, red oak mean height was only 0.67 m. It remains to be seen how many seedlings will survive to play a role in the developing stand. Our next sampling will be done in year 15.

Woody Species Phenology, Prospect Hill Tract, Harvard Forest - 2001

J. O'Keefe and T. Jones

2001 was the twelfth year in our ongoing investigation of the timing of woody vegetation development during the growing season. We observed bud break, leaf development, flowering, and fruit development on three or more individuals of 33 woody species at 3-7 day intervals from April through June. Weekly observations began again in September and continued through leaf fall. All individuals are located within 1.5 km of the Harvard Forest headquarters at elevations between 335 and 365 m, in habitats ranging from closed forest, through forest-swamp margins, to dry, open fields.

The winter of 2000-2001 began somewhat milder than normal with slightly below normal precipitation until March brought below average temperatures and heavy snow. This resulted in a deep snow pack that persisted well into April. Then the end of April and early May brought record warmth followed by a hard freeze on May 7^th that killed the newly emerged leaves of many oaks on exposed sites. Summer and fall were generally warm with near normal precipitation.

For most species initial leafout in 2001 was quite early (Table 1), despite the late snow cover in April. Interestingly, perhaps due to the late snow cover followed by record warmth, there was much less variation among species in the timing of bud break than in previous early years (1991, 1993, 1998). Leaf development then progressed steadily so that for most species 75% leaf development occurred close to the dates observed in the early years (1991, 1993, 1998), although there was more variability in leaf development timing than in bud break. Of note is the late leaf development of white oak. This delay resulted from the freeze on May 7^th killing most of the newly emerged leaves on one exposed individual, which caused a significant delay until a second flush of leaves could be produced. Some observed individuals of black oak, red oak, and red maple in exposed locations suffered similar, although somewhat less severe, damage from this freeze. Differing species and individual plant responses to this freeze emphasize the complex factors controlling these phenological processes over the development period. Each species and plant presumably integrates these factors somewhat differently as it develops.

Fall coloration and leaf fall in 2001 were somewhat later than average, with color peaking about October 12^th. This slight lateness was still within the overall pattern in which leaf senescence has shown much less variability than leaf development (Fig. 1). For most species only a week to ten days has separated the earliest and latest falls during nine years of observations.

Table 1. Estimated mean leaf development dates (month-day) for individuals of six representative species over twelve years (IBB = initial bud break, 75% = 75% leaf development).

	Quercus rubra (n=4)		Acer rubrum (n=5)		Betula allegheniensis (n=3)
	IBB	75%	IBB	75%	IBB	75%
1990	127 (5/7)	166 (6/15)	121 (5/1)	164 (6/13)	122 (5/12)	155 (6/4)
1991	128 (5/8)	143 (5/23)	123 (5/3)	142 (5/22)	117 (4/27)	141 (5/21)
1992	135 (5/14)	163 (6/11)	133 (5/12)	170 (6/18)	132 (5/11)	164 (6/12)
1993	122 (5/2)	153 (6/2)	118 (4/28)	154 (6/3)	113 (4/23)	136 (5/16)
1994	129 (5/9)	154 (6/3)	124 (5/4)	151 (5/31)	122 (5/2)	147 (5/27)
1995	132 (5/12)	150 (5/30)	129 (5/9)	148 (5/28)	126 (5/6)	144 (5/24)
1996	133 (5/12)	151 (5/30)	121 (4/30)	150 (5/29)	118 (4/27)	145 (5/24)
1997	136 (5/16)	158 (6/7)	133 (5/13)	154 (6/3)	134 (5/14)	153 (6/2)
1998	123 (5/3)	142 (5/22)	120 (4/30)	140 (5/20)	121 (5/1)	138 (5/18)
1999	126 (5/6)	148 (5/28)	126 (5/6)	146 (5/26)	127 (5/7)	141 (5/21)
2000	129 (5/8)	161 (6/9)	129 (5/8)	156 (6/4)	129 (5/8)	149 (5/28)
2001	122 (5/2)	149 (5/29)	123 (5/3)	145 (5/25)	123 (5/3)	142 (5/22)
	Quercus alba (n=3)		Hammamelis virginiana (n=3)		Acer pensylvanicum (n=4)
	IBB	75%	IBB	75%	IBB	75%
1990	132 (5/12)	163 (6/12)	124 (5/4)	163 (6/3)	119 (4/29)	155 (6/4)
1991	127 (5/7)	150 (5/30)	116 (4/26)	144 (5/24)	118 (4/28)	141 (5/21)
1992	138 (5/17)	158 (6/6)	132 (5/11)	168 (6/16)	125 (5/4)	159 (6/7)
1993	124 (5/4)	150 (5/30)	121 (5/1)	149 (5/29)	121 (5/1)	144 (5/24)
1994	134 (5/14)	160 (6/9)	126 (5/6)	148 (5/28)	128 (5/8)	151 (5/31)
1995	135 (5/15)	157 (6/6)	127 (5/7)	148 (5/28)	127 (5/7)	148 (5/28)
1996	139 (5/18)	159(6/7)	123 (5/2)	150 (5/29)	130 (5/9)	151 (5/30)
1997	146 (5/26)	164 (6/13)	130 (5/10)	154 (6/3)	133 (5/13)	153 (6/2)
1998	126 (5/6)	145 (5/25)	113 (4/23)	139 (5/19)	118 (4/29)	142 (5/22)
1999	132 (5/12)	155 (6/3)	122 (5/2)	142 (5/22)	125 (5/5)	146 (5/26)
2000	132 (5/11)	162 (6/10)	126 (5/5)	151 (5/30)	127 (5/6)	155 (6/3)
2001	124 (5/4)	162 (6/11)	122 (5/2)	143 (5/23)	125 (5/5)	149 (5/29)

Stand and Community Reorganization Dynamics Associated with Chronic Hemlock Woolly Adelgid Infestations in Southern New England

D. Orwig

Vegetation dynamics were monitored for 6 years in eight hemlock (Tsuga canadensis) stands with varying levels of hemlock woolly adelgid (HWA; Adelges tsugae) damage in south-central Connecticut to examine the patterns of overstory hemlock mortality and subsequent community reorganization associated with chronic HWA infestations. Since 1995, overstory and understory mortality has risen 5 to 15 % per year to overall values of 50 to 99% (Fig. 1). There has been no sign of tree recovery and the health and vigor of remaining trees deteriorated in all infested stands. Results suggest that trees on some sites can remain alive for over 10 years following initial infestation and remain standing for 6 to 8 years following mortality. Rapid recolonization of black birch (Betula lenta) occurred at most sites, along with low densities of red maple (Acer rubrum) and oak (Quercus) species. Seedling densities increased in moderately damaged stands and thinned but increased rapidly in height, reaching 3 to 7 m tall in heavily damaged stands. The few hemlock seedlings found contained HWA infestation. Shrub cover remained low following infestation while herbaceous cover, consisting primarily of hay-scented fern (Dennstaedtia punctilobula), has gradually increased over time. Cover of the invasive Japanese stilt grass (Microstegium vimineum) increased in one of the stands. Tree-ring analysis of hardwood and declining hemlock trees, coupled with age-structure analysis of newly established birch saplings was effective in determining the timing of initial HWA impact in stands of unknown infestation date. Over the last 10 years, radial growth of hemlock declined precipitously in most stands, while oak and maple growth exhibited concomitant, large annual increases. Results suggest dramatic stand structure and composition changes accompany heavy HWA infestations that continue long after hemlock mortality.

Vegetation and Disturbance History of Ridgetop Pitch Pine and Red Pine Communities of Southern New England

D. Orwig, G. Motzkin and D. Foster

As part of a broader investigation of ridgetop communities in the Northeast, we initiated a study examining the long-term history, development, and vegetation dynamics of Mt. Everett and nearby summits in the southern Taconic region of southwest Massachusetts. We are using a combination of dendroecological, historical, and field studies to address several questions: 1) what is the historical fire and disturbance history of the region and how has this influenced vegetation composition and structure? 2) What is the history of land-use and how has this influenced vegetation composition and structure? 3) What is the age structure of the current stand and how is this related to disturbance history? 4) How do the composition, structure, and dynamics of Mt. Everett compare with other sites in the southern Taconics and elsewhere in central New England?

We have visited several sites and intensively sampled the dwarf pitch pine community on Mt. Everett. This rare forest type, occurring in only a few locations in the northeastern U.S., is characterized by exposed bedrock, shallow soil depths of 10 to 18 cm, and abundant stunted pitch pine 0.5 to 2.0 m tall. The uneven-aged stand includes red maple, red oak, and birch species. Pitch pine exhibits continuous recruitment since the 1830s and ranges in age from 12 to 170 years old. Hardwood species established in most decades since 1860 and in higher densities in the 1950s and 1960s. Diameter is a poor predictor of age in this forest as trees with diameters of only 10 cm differ in age by more than 100 years. Many pines exhibit multiple stems, prostrate growth forms, and evidence of terminal branch damage. There is no historical or visual evidence of fire or human impacts in this forest, but on this exposed summit (795m a.s.l.), wind, snow, and ice damage are dominant disturbance factors. Pitch pine ring-width patterns displayed extremely slow growth, with many individuals averaging < 0.40 mm yr. ^-1, and a few growing only 0.08 to 0.30 mm yr. ^-1 for periods of up to 50 years. The unusual structure, extremely slow growth rates, and continuous recruitment in the absence of recent fire suggest that weather and harsh site conditions continue to maintain this unique community of disturbance-adapted species.

We will continue to examine additional ridgetop communities to place the results of this site into a broader landscape context and to make comparisons with forests that have been influenced by fire and other factors.

Ecosystem Analyses of Hemlock Woolly Adelgid Outbreaks in Southern New England

D. Orwig, R. Cobb, M. Kizlinski and D. Foster

In 1998 we began examining the response of ecosystem processes to the stress and mortality caused by the introduced hemlock woolly adelgid (HWA) in southern New England. Healthy hemlock forests typically have slow decomposition and N cycling rates due to their low foliar N content and cool microclimate. However, thinning canopies associated with HWA infestations should reverse this trend, due to dramatic increases in light levels and soil temperature. Within 8 study sites varying in HWA infestion level, we continued to investigate the magnitude and duration of N dynamics associated with HWA infestations by measuring nitrogen (N) mineralization rates using close-topped soil cores during the last four years. In addition, ion-exchange resin bags were used to estimate the spatial availability of N within sites and the extent to which NO₃ is being lost. Measurements of gravimetric moisture content and soil temperature were used with hemispherical photographs to assess microenvironmental changes. During the first four years of this study, thinning canopies from heavy HWA damage resulted in increased light, increased soil temperature, and decreased forest floor moisture content. Nitrogen cycling was impacted by damage class and microenvironmental conditions (Fig. 1). Heavily infested sites had larger extractable NH₄ and NO₃ - N pools, and had net nitrification rates 25 times greater than healthy hemlock forests (Fig. 1). In addition, resin bags captured more ammonium and nitrate in infested versus uninfested stands. Higby Mountain, the most heavily damaged stand, exhibited increasing amounts of nitrogen capture over time at levels 5 to 40 times greater than control stands. Low and undamaged stands typically showed net immobilization of nitrate. Results indicate that introduced pests and selective tree decline can rapidly and dramatically alter ecosystem processes, even prior to the onset of extensive tree mortality. In 2001, we began examining 2 additional stands that contain high overstory hemlock mortality and a dense black birch understory. We will continue to sample these stands as they deteriorate to determine the extent to which changes in overstory composition, microenvironment, and soil conditions produce fundamental changes in the cycling of nitrogen.

History of Fire in New England from Charcoal and Pollen in Lake Sediments

T. Parshall, D. Foster, S. Barry Musielewicz, E. Faison, D. MacDonald and E. Doughty

Fire plays a central role in many ecosystems by killing living vegetation (selectively), removing dead biomass, changing soil composition, and altering nutrient cycling. Like most disturbance agents, fire is heterogeneous. Ignition and spread is influenced by site and landscape-scale characteristics. Once started, macro and microclimatic variables influence fire intensity and magnitude, and some species are better able to survive fire than others. Humans have always influenced fire regimes by increasing the chance of ignition and controlling the spread of fire once it starts. The traditional view that fires have increased in abundance since European settlement in New England is undoubtedly true in many cases, but some studies have shown that fire has declined since settlement. Furthermore, Native American activity likely increased the frequency of fire before European settlement, but probably not to the same degree throughout New England. An understanding of the presettlement fire regime across a range of vegetation types is critical for developing a better understanding of ecosystem dynamics and for management plans that include prescribed fire.

The goal of this study is to evaluate the variation in fire occurrence across the range of vegetation types that occur in New England (from pitch pine-oak to spruce-fir) and to assess how these fire regimes have changed through time as climate and human land use have changed. We are using the Harvard Forest paleoecology database, which now includes a wide range of lakes in New England, in combination with existing studies of fossil charcoal. Presettlement fire and vegetation are closely related across New England. Fire occurrence was lowest in the spruce-fir and northern hardwood forests and highest along the southern coast of New England, where pitch pine and oak forests dominate. Variability in fire occurrence from site to site is high, which can also be explained primarily by vegetation. For example, fires were just as common in central New England and along the coast where pitch pine and oak forests occur. At the same time, fires are uncommon in central and northern New England where northern hardwoods and spruce-fir forests occur. The vegetation on this spatial scale is largely related to substrate characteristics and topographic position. On a more local scale, landscape features have apparently contributed to the occurrence of fire by protecting sites from its spread. Presettlement fires were most common along the coast around sites on the eastern end of a landmass that are not protected by firebreaks.

Fires have been more common following European settlement at almost every site that we studied, and the shift occurs coincidently with a change toward more open vegetation. We find no evidence for extensive grasslands anywhere in New England, including Cape Cod and Martha's Vineyard where many people suggest that open vegetation was common. Perhaps there were smaller patches of grasslands and shrublands in unique situations or along the immediate coastline. If so, this level of resolution is not available using our paleoecological sites. For this reason and because we find little correspondence between local archaeological evidence and fire occurrence, we fail to find strong support that Native American populations had an extensive effect on landscape-scale fire or vegetation patterns. Much more documentation of local archaeological sites and detailed paleoecological evidence is necessary to determine the role of Native Americans on New England ecosystems.

The 'Hydrology' of Temperate Deciduous Leaves: Linkages Between Leaf Hydraulic Conductance and Leaf Form, Structure and Function

Lawren Sack and N. Michele Holbrook

Recent studies indicate that the hydraulic conductance of the leaves, relative to that of the rest of the plant, can be surprisingly low. Leaves can thus form a major bottleneck in the overall liquid-phase path through the plant, which, in turn, can constrain maximum stomatal conductance, and ultimately the carbon exchange capacity of the leaves. Recently we refined three methods for reliably measuring leaf lamina hydraulic conductance (K_lamina) of single detached leaves. We tested the associations between K_lamina and leaf form, structure and physiology, for five deciduous woody species at Harvard Forest. How is K_lamina linked to other leaf traits already known to be related, such as leaf water content and leaf mass per area? Linkages between K_lamina and other features may reflect optimizations or trade-offs that determine leaf design during evolution and development. We found that K_lamina ranged substantially across species, and that sun leaves had significantly higher K_lamina than shade leaves (Fig 1). K_lamina was strongly correlated with leaf structural, anatomical and physiological traits, for instance maximum total stomatal pore area (Fig. 2a). This relation may drive a correlation of K_lamina with maximum stomatal conductance (Fig. 2b). Such relationships provide fundamental links between leaf structure, water-use, and carbon metabolism. Further, these relationships have potential applications in predicting whole-plant function and ecology.

Drought Effects on Soil Respiration in a Temperate Forest

K. Savage, W. Borken and E. Davidson

Temperature plays a primary role in the seasonal variation of soil respiration, and interannual variations in respiration can be explained, in part, by variations in annual precipi-tation. Long term measurements at our study site at the Harvard Forest, have shown lower annual respiration during the relatively dry years of 1995 (0.71 kg C m^-2 yr^-1), 1997 (0.64 kg C m^-2 yr^-1) and 1999 (0.48 kg C m^-2 yr^-1) compared to wet years of 1996 (0.78 kg C m^-2 yr^-1), 1998 (0.86 kg C m^-2 yr^-1) and 2000 (0.89 kg C m^-2 yr^-1). The objective of this study was to quantify the effects of drought in a replicated and controlled experiment.

We established three replicate plots (4 x 5m) for each of two treatments: throughfall exclusion and control. The exclusion plots were covered with tilted plastic roofing panels about 1 m above the ground to exclude throughfall from July 2^nd, 2001 through Sept. 24^th2001. Four collars were placed in each of the plots for measuring soil respiration, using a dynamic chamber system and portable infrared gas analyzer. Soil respiration was measured at weekly intervals during the summer. Soil temperature and moisture were measured hourly.

The year 2001 was drier (865mm) than the long-term average (1050mm). During the time the roofs were installed 168 mm of throughfall was excluded from the exclusion plots. The difference in soil respiration rates was most evident following rainfall events, when respiration rates were higher in the control compared to the exclusion (Fig.1a). Cumulative mean soil respiration was lower in the exclusion treatment (237 g C m^-2 per 84 days) compared to the control (338 g C m^-2 per 84 days). Soil respiration increased in both treatments following the first rain event after the removal of the roofs from the exclusion treatment, due to the naturally dry soil conditions present in the control plots (Fig. 1a, c, d). The roofs will be put pack in place in the spring of 2002, and respiration measurements will be repeated throughout the summer using an automated chamber system that will provide higher temporal resolution for studying responses to wetting events in the control plots.

Ecology, Evolution and Physiology of the Invasive Plant Species, Alliaria petiolata: Do Source-sink Dynamics Facilitate Habitat Expansion into Forest Understory?

K. Stinson and F. Bazzaz

This study investigates invading populations of garlic mustard, Alliaria petiolata, at Harvard Forest and other sites in Massachusetts. A. petiolata (Fig. 1) is an exot