14^th Annual Harvard Forest Ecology Symposium - 2003

February 12^th, 2003

Sources and Dynamics of Dissolved Organic Carbon and Nitrogen in a Hardwood Forest Floor

J. Aitkenhead-Peterson, W. McDowell, K. Lajtha, P. Micks and K. Nadelhoffer

The ongoing DIRT experiment (Detritus Input and Removal Treatments) at Harvard Forest enabled us to quantify the importance of several sources of soil solution dissolved organic carbon (DOC) and nitrogen (DON) in a mixed hardwood forest floor. The goal of the DIRT project is to assess how rates and sources of plant inputs control the accumulation and dynamics of organic matter and nutrients in forest soils over decadal time scales. Long-term treatments begun in 1990 include double annual leaf litter input (DL), zero annual leaf litter input (NL), root exclusion (NR), root and leaf litter exclusion (NI), controls (C) and removal of the O and A horizons (OA-less), in replicated 3m x 3m plots. Similar experiments in Pennsylvania, Wisconsin, Oregon, and Hungarian forests provide insights into links between microbial processes, labile C and N pools, and plant inputs to forest soils.

The objective of this study was to quantify the contribution of DOC and DON from above- and below-ground litter, live roots, and organic soil to soil solution. We hypothesized that removal of organic matter sources would reduce DOC and DON fluxes, and that doubling leaf litter would increase DOC and DON fluxes. While our hypothesis was not fully supported by the results, we were able to apportion the sources of DOC and DON to soil solution. Soil solution was collected from zero-tension lysimeters installed at the interface between the forest floor and mineral soil in each plot. Samples were taken within 24 hours after fifteen rainfall events from May - November 1997 (yr 7 of treatments). Volume-weighted mean DOC and DON concentrations were significantly higher in DL than in controls, and significantly less in OA-less soils. Response was mixed in the litter and/or root-excluded treatments (Table 1). Sources of DOC and DON were quantified using mean annual fluxes calculated from volume-weighted mean concentrations and the PnET-II model (Aber et al. 1995) (Fig. 1). Based on differences between controls and treatments, the Oea horizon contributes 40% to soil solution DOC, leaf litter contributes 44%, and root exudate and decay contributes 16%. DON source partitioning was less straightforward because roots appeared to be a sink for DON produced in the forest floor. Oea soil appears to contribute 107% and leaf litter 39% to soil solution DON.

Mean annual DOC flux was strongly related to forest floor C:N ratio (R² = 0.95 p < 0.01) (Fig. 2) but DON flux was not. Mean annual DOC and DON fluxes were positively related to fungal biomass (R² = 0.78 and R² = 0.70 respectively, p < 0.05), suggesting that fungal biomass may be responsible for a large proportion of DOC and DON production. A seasonal effect was observed where DOC:DON ratios were significantly narrower in root-intact treatments DL, NL, and C in summer compared to autumn. This decoupling of DOC and DON with manipulated organic matter inputs and season contrasts the traditional model which suggests that DOC and DON are produced in relatively fixed ratios.

DOC fluxes in year 7 show overall patterns similar to cumulative DOC and CO₂ release from incubated Oea soils collected in year 5. They also correspond to treatment patterns of CO₂ efflux measured in the field in year 8. Continuing soil solution analysis will allow us to track DOC and DON fluxes and examine their linkages to long-term organic matter dynamics in this forest soil.

Aber, J.D., S.V.Ollinger, C.A. Federer, P.B. Reich, M.L. Goulden, D.W. Kicklighter, J.M. Melillo, and R.G. Lathrop Jr. 1995. Predicting the effects of climate change on water yield and forest production in the northeastern United States. Climate Res. 5:207-222.

Modeling the Impact of Hemlock Loss on New England Forests with the Ecosystem Demography Model

M. Albani and P. Moorcroft

The Hemlock Woolly Adelgid (adelges tsugae, HWA), an exotic aphid pathogen of eastern hemlock (tsuga canadensis), is spreading unchecked across the eastern United States. As the pathogen spreads north, New England's forests are expected to experience substantial losses of hemlock both directly from the pathogen and by associated pre-emptive and salvage logging. Since eastern hemlock is the dominant species in several forest ecosystems, its loss and replacement by birch and other hardwoods will likely lead to a dramatic cascade of biological, biogeochemical, and physical changes.

We plan to use the Ecosystem Demography (ED) model, an individual-based vegetation model describing the growth, reproduction and mortality dynamics of a plant community coupled to biogeochemical models of the associated soil fluxes of carbon, water and nitrogen, to predict the impact of hemlock losses on ecosystem structure and function.

Using data for other investigations at Harvard Forest, we will modify ED to incorporate the impact of HWA on the physiology and demography of hemlock. Concurrently, we will develop and calibrate a predictive, spatially-explicit model of pathogen infection that will be used to predict the spatio-temporal pattern spread of the adelgid as it advances northward.

Hemlock Forest Dynamics at Harvard Forest

A. Barker Plotkin, D. Orwig, J. Guzman* and D. Foster

Hemlock (tsuga canadensis) forests in New England are changing rapidly with the invasion of the hemlock woolly adelgid (HWA, adelges tsugae), an exotic insect pest that kills hemlock trees. The adelgid is just beginning to be observed at Harvard Forest, making this the right moment to begin intensive physiological, ecological and monitoring studies of our hemlock forests. We gathered detailed baseline information for several hemlock stands across Harvard Forest. We surveyed community vegetation structure and composition, and began to monitor soil nitrogen cycling. These measurements will allow us to link the long-term history of these hemlock stands to other current studies of hemlock response to adelgid infestation (Orwig 2002), and will provide the basis for intensive study of hemlock dynamics with the stands' anticipated decline. The five sites in this study include existing plots with known land-use and paleoecological histories (Foster and Zebryk 1993, McLachlan and Foster 2000). Currently, all sites are healthy with no observed HWA. Overstory trees and saplings were sampled in fixed-radius plots. Hemlock had importance values between 65-84% (Table 1), with white pine, black birch, red maple and red oak as common minor components. Nearly all saplings were hemlock; sapling densities ranged from 183 - 1645 stems/ha. Understory flora characteristics were sampled in 10- 1m2 subplots within each overstory plot. Across all sites, herb and shrub cover was very low, reflecting the dense shade and deep organic layer present in these hemlock stands (Table 1). Species richness ranged from 6 -15 understory species found in each 20x20 or 30x30m plot. Compared to a suite of hemlock study sites established in Connecticut in 1995 (Orwig and Foster 1998), these sites have more white pine and a deeper organic layer, suggesting that vegetation dynamics following hemlock decline may differ from the Connecticut sites. We initiated baseline measurements of N pools, net N mineralization and nitrification rates and N availability at the Hemlock Grid and Slab City sites. We will continue to monitor vegetation and nutrient dynamics in these stands as HWA becomes established at Harvard Forest. These sites represent a cross-section of the hemlock forests present at Harvard Forest (Fig. 1) and provide a basis for future intensive studies. HWA has been observed this winter at a few sites at Harvard Forest, so these stands are likely to experience hemlock decline and major changes in vegetation composition, forest structure and nutrient cycling within the next decade.

Foster, D.R and T.M. Zebryk. 1993. Long-term vegetation dynamics and disturbance history of a Tsuga-dominated forest in New England. Ecology 74:982-998.
McLachlan, J. and D.R. Foster and F. Menalled. 2000. Anthropogenic ties to late-successional structure and composition in four New England hemlock stands. Ecology 81:717-733.
Orwig, D. and D.R. Foster. 1998. Forest response to the introduced hemlock woolly adelgid in southern New England. Bulletin of the Torrey Botanical Society 125:59-72.
Orwig, D. 2002. Stand dynamics associated with chronic hemlock woolly adelgid infestations in southern New England. Pages 36-46 in Onken, B., R. Reardon and J. Lashomb (eds.). Symposium on the Hemlock Woolly Adelgid in Eastern North America, February 2002, Rutgers, NJ.

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

D. Barnes, S. Wofsy, J. Elkins, M. McElroy and E. Gottlieb

This study provides an independent measure of emissions of ozone-depleting species for the first years (1996-2001) following the full implementation of the Montreal Protocol1. The measurements were taken at Harvard Forest in western Massachusetts, a site that receives air from the northeastern urban/industrial corridor of the U. S. 30-40% of the time. These data exhibit well-correlated spikes in concentrations of CO, CFC-11 (CCl₃F), CFC-12 (CCl₂F₂), and methyl chloroform (CH₃CCl₃). The annual and seasonal emissions of CFC-11, CFC-12, and CH₃CCl₃ are estimated by correlating their concentrations with that of CO, and the extensive inventory data for CO emissions in the U. S. (EPA) was used to infer sources strengths for these species. For further details on the method used, see Barnes (2000).

The emissions of CFC-11, CFC-12 and CH₃CCl₃ in two other regions of the U. S. were estimated by applying the same statistical methodology used for the Harvard Forest data to measurements from NOAA/CMDL instruments located in North Carolina (WITN: data for 1995-1998) and Wisconsin (WLEF: data for 1996-2000).

Summarized in Figure 1 are the annual emissions of CFC-11, CFC-12, and CH₃CCl₃ for these three regions of the U. S.

• Methyl chloroform emissions are continuing at rates higher than anticipated. Emissions of CH₃CCl₃ were expected to drop to zero in 1996 or shortly thereafter, based on the pattern of usage for CH₃CCl₃. Prior to the 1996 ban, the solvent was utilized within a year of purchase. The continuation of the CH₃CCl₃ emissions has significant implications for OH radical calculations by atmospheric modelers.
• Emissions of CFC-12 are decreasing as expected. Residual emissions of CFC-12 and CFC-11 were expected following the 1 January 1996 ban, due to leakage from long-term reservoirs of these two gases, particularly from refrigerants.
• CFC-11 emissions have remained steady or increased slightly during the six years since the ban in 1996.
• Our CFC-11 and CH₃CCl₃ emission results suggest that long-term reservoirs are larger than previously thought, that users stockpiled the substances in large quantities in anticipation of the ban, or that a black market has developed in violation of the ban.

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, 1987-1992). A complete ban in developed countries for CFC-11, CFC-12, and methyl chloroform (CH₃CCl₃), was set for 1 January 1996.

Barnes, D. H. 2000. Quantifying Urban/Industrial Emissions of Greenhouse and Ozone-Depleting Gases Based on Atmospheric Observations. Ph.D. Thesis, Harvard University.
United Nations Environmental Programme. 1987, Montreal Protocol to Reduce Substances that deplete the Ozone Layer Report, Final Report (New York); 1990 London Amendment; 1992 Copenhagen Amendment.

The Little Ice Age Signal in New England: a Survey of Pre-settlement Climate Change

S. Barry, D. Foster and B. Hall

The "Little Ice Age", ca. 1450 to 1850 A.D., was an apparently global climatic period characterized in the Northern Hemisphere by highly variable conditions, frequent long winters and short, cool summers. In northwestern Europe environmental signals of this period are recorded in glacial activity, coastal sediments, tree rings, and historical documents, which indicate a period of cool and variable weather including shorter growing seasons and increased storminess. More recently, glacial, tree ring, lake level, and historical data indicate coincident changes in terrestrial and aquatic ecosystems in North America.

The low-resolution and poor temporal control of most pollen and stratigraphic data has restricted reconstructions of subtle paleoclimatic changes in New England. Furthermore, the end of the "Little Ice Age" coincides with the period of European settlement and broad-scale deforestation. However, some studies suggest that significant environmental and vegetation changes occurred prior to settlement. For example, our recent studies of the southern New England landscape determined that the major changes since European settlement, including the decline in beech and hemlock and the regional homogenization of vegetation composition, were actually initiated some 300-500 years before European arrival.

In this current multi-proxy study, we combine pollen data from new sites, the Global Pollen database, and the Harvard Forest archives with detailed historical land-use and land cover maps to analyze vegetation and environmental change in Massachusetts over the past 2000 years. Massachusetts is a particularly good study region because it was colonized early, is covered by unique historical records, and embraces a wide range of vegetation, environments, and cultural conditions. We analyzed records from over 20 ponds arrayed across the state's gradients in physiography, climate, geology and natural disturbance. Chronological control was provided through lead-210 and radiocarbon dating and sediments were sampled at high-resolution intervals. Small (<10 ha) and primarily spring-fed kettle lakes were selected to emphasize the local to sub-regional vegetation signal. Historical data include archaeological information and pre-settlement vegetation derived from witness trees records in early land surveys. Collaborators at Laval University have provided isotopic analyses and diatom surveys from several of our sites, increasing our understanding of regional changes.

Preliminary results suggest that the "Little Ice Age" effects varied geographically with the magnitude and nature of vegetation changes differing by vegetation type and ecoregion. A decrease in beech and hemlock is most evident in high-elevation sites, while low elevation sites of oak-hardwood forest composition show relatively little change. In many cases, changes in forest composition dynamics are accompanied by coincident changes in aquatic vegetation and % organic matter, presumably reflecting changes in water balance and lake levels or watershed clearing.

Hydrological Stations

E. Boose, E. Colburn and A. Barros

Permanent hydrological stations are planned for two small headwater streams in the Prospect Hill Tract. The first station (funded) will be located on the stream that drains the Black Gum Swamp and adjacent woodlands and flows near Shaler Hall on its way to Nelson Brook and the Millers River. The second station (funding pending) will be located on the stream that drains the steep valley to the west of Prospect Hill and flows through the wetland north of the EMS on its way to Bigelow Brook and the Swift River. Both streams have surface flow for an estimated 10-12 months of the year. The first watershed and all but the uppermost 5-10% of the second watershed are owned by Harvard University.

Each station will include a weir, piezometer array, sensors, and datalogger to measure and record surface discharge, water temperature, and subsurface flow. Stations will be provided with electricity and a communications link to networked computers in Shaler Hall and the EMS Shack. Heating cables and hoods will be used to prevent ice buildup in winter. Measurements will be posted on the Harvard Forest web page in near-real time, and contributed regularly to the LTER HydroDB database. Construction is planned for spring and summer 2003. Additional funding will be sought to deploy piezometers and soil moisture sensors at upstream locations in both watersheds.

The two watersheds, though adjacent and comparable in size, differ significantly in topography, soils, hydrology, land-use history, stream biota, and forest vegetation (including abundance of hemlock). Long-term measurements from both stations (and the Fisher Met Station) should greatly improve our understanding of the relationships among climate, soils, groundwater, surface streamflow, and stream biota. The two watersheds may also provide some control for disturbances that impact one watershed more than the other, including the anticipated loss of hemlock from the hemlock woolly adelgid.

Effects of Experimental Drought on Soil Respiration and Radiocarbon Efflux from a Temperate Forest Soil

W. Borken, K. Savage, E. Davidson and S.Trumbore

Soil respiration is clearly affected by both soil temperature and water content. Relying upon natural variation of temperature and precipitation does not permit a clear distinction between them as factors controlling seasonal and interannual variation in soil respiration. To clarify the effects of summertime drought on soil respiration processes, we conducted a drought experiment at the Harvard Forest of central Massachusetts. Both control and throughfall exclusion plots experienced the same temperature, but different moisture regimes. In addition to measuring the effects of drought on respiration, we also measured radiocarbon content of soil CO₂ to determine if heterotrophic and autotrophic respiration are affected differently by drought.

Three replicate throughfall exclusion roofs (5x5m) and 3 control plots were established in a well-drained, mixed hardwood stand. The roofs were translucent corrugated plastic panels about 1 m above the ground surface. In 2001, 168mm of throughfall were excluded over an 84 day period, and in 2002, 344 mm were excluded over 126 days. Soil respiration was measured weekly using a portable infrared gas analyzer and 4 manual chambers in each of six plots. Soil moisture and temperature probes were installed in all plots. The water content of the O-horizon was measured using DC half bridges. In the summer of 2002 we installed an automatic chamber in each of the six plots, which made hourly measurements. Radiocarbon of the soil CO₂ profiles was measured from the throughfall exclusion and control plots several times over the summer period. Mean soil respiration was lower in the exclusion plots when the panels were in place in both years (Fig. 1a) with the cumulative difference between treatments of 100 g C m^-2 per 84 days in 2001 and 210 g C m^-2 per 126 days in 2002. Soil temperature varied little between treatment plots (Fig. 1b). Litter layer water content in the control plots responded dramatically to rainfall events (Fig. 1c, d). The high resolution sampling frequency of the automated chambers allowed us to capture the rapid response of soil respiration to precipitation events (Fig. 2a-d). Many of the highest fluxes were measured shortly after a wetting event (Fig. 3a). Fluxes can remain high up to 3 days following the precipitation event (Figs. 2a and 3a). The residuals of the temperature function were positively correlated with litter layer water content (Fig. 3b). Higher Δ ¹⁴C values were observed when CO₂ concentrations were low (Fig. 4), which occurred at low water content. In the control plots this is a response to natural late summer drought, but it is more pronounced in the throughfall exclusion plots. When the soil is moist, high rates of root respiration and decomposition of young substrates result in high soil CO₂ concentrations with low Δ ¹⁴C values. Under dry conditions, root respiration produces less CO₂, while gradual decomposition of old, radiocarbon-rich substrate continues at low rates, resulting in low CO₂ concentrations with high radiocarbon content. In summary, experimentally imposed summer drought caused a reduction of soil respiration by 1 - 2 Mg C ha^-1, which is 10 - 30% of the annual rates (7-8 Mg C m^-2 ha^-1). Rainfall caused rapid increases in soil respiration, which lasted several days. A significant fraction of the variation in soil respiration not explained by soil temperature was correlated with water content of the litter layer. Radiocarbon data demonstrated that respiration of recently fixed C substrates was more strongly affected by drought than decomposition of older substrates. This experiment clearly shows that soil water content, and particularly the water content of the litter layer, is an important factor controlling daily, seasonal, and interannual variation in soil respiration of this temperate forest.

Tree Regeneration and Understory Vegetation in an Eastern Hardwood Forest 20 Years After Creation of Small Forest Openings

P. J. Burton, C. M. Burton, T. W. Sipe and D. E. Hibbs

Some suggest that forest harvesting methods mimicing the scale of natural gap dynamics can both regenerate desired tree species and maintain the conservation value of mature forests. In central New England, the process of hardwood regeneration in small openings has not been explored as thoroughly as in other regions. Fifteen roughly circular forest openings were created at Harvard Forest in the winter of 1983 to address the differential effects of opening size, and location within openings, on the establishment, survival and growth of local tree species. The experiment was established in a mixed hardwood and conifer forest dominated by red oak (Quercus rubra), red maple (Acer rubrum) and white oak (Q. alba), with lesser amounts of white ash (Fraxinus americana), black oak (Q. velutina), white pine (Pinus strobus ) and several other incidental tree species. Timber harvesting created a series of canopy gaps ranging in size from 30 m2 to 700 m2 (6 to 30 m average diameter). Mean canopy height of residual trees soon after logging was 15.5 m. The study site is located in Compartment 9 (west of Route 32) of the Prospect Hill Tract, on gently rolling bouldery terrain formerly cleared for agriculture.

In the mid-summer of 1984, the second growing season after logging, forest regeneration was sampled in intersecting transects of contiguous 2 m x 2 m quadrats running north-south and east-west across each gap and into the adjacent uncut forest. A total of 384 quadrats were used to map and count tree seedlings and sprouts, and to estimate the cover of dominant understory species. Height and basal diameter were measured for the tallest regenerating stems of each tree species. In the late summer and early fall of 2002, 20 growing seasons after gap creation and 18 years after the previous monitoring, all quadrats were relocated and resampled. Data recording for this most recent round of sampling was done on copies of the 1984 data sheets, which facilitated the relocation of mapped tree seedlings and sprouts. Stems of all tree species were measured for basal diameter and height, and the percent cover of all vascular plant species was estimated.

Forest regeneration both 2 and 20 years after logging was dominated by sprouts of red oak and red maple and by seedlings of black birch (Betula lenta), which collectively formed a closed canopy in most gaps when surveyed in 2002. Many of the stems mapped in 1984 survived to 2002, but clumps of sprouts from the same stump had thinned out considerably. Some black birch and yellow birch (B. alleghaniensis) trees had attained heights of more than 13 m after 20 years of growth, and trees of many species were found to be more than 10 m tall. Approximately as many trees had established after the 1984 survey as before it. But most late-establishing regeneration was of small stature, suppressed by the earlier cohort. Preliminary data analysis suggests that length of the north-south gap axis is a stronger determinant of regeneration dynamics than simple gap area. It appears that a north-south gap diameter of 21 m (135% of canopy height) may constitute a threshold for the processes governing successful tree establishment and growth (Fig. 1).

Ground-level vegetation in 1984 was very dense, especially in the larger gaps, and was dominated by hay-scented fern (Dennstaedtia punctilobula). By 2002, where regenerating trees formed a closed canopy, this vegetation had receded considerably, but not quite to the cover levels found in the uncut forest nearby. The 2002 vascular plant community in the experimental gaps was dominated by flowering-ferns (Osmunda spp.), wild sarsaparilla (Aralia nudicaulis), and wild blueberries (Vaccinium spp.) as well as hay-scented fern. We are currently analyzing how the abundance of individual species and the composition of the plant community are related to gap size and gap position.

The Effects of Hemlock Woolly Adelgid Infestation on Decomposition in Eastern Hemlock Forests

R. Cobb and D. Orwig

For the last four years we have been working to quantify decomposition changes that occur as eastern hemlock (tsuga canadensis) ecosystems decline. Our study has focused on three key drivers of decomposition: (i) changes in foliar quality due to HWA herbivory, (ii) changes in forest floor microclimate that occur as the canopy thins, and (iii) the effect of species composition change. All three have had an impact on decomposition. Furthermore, our data suggest that these changes are coupled with N-cycling dynamics and may helped to elucidate the mechanisms driving increased N availability in these forests.

We sampled hemlock foliar N, C, and lignin along an extensive gradient of hemlock forests. These stands range from uninfested to forests with complete hemlock mortality and currently dominated by black birch (Betula lenta) (see Orwig and Foster 1998). Foliar % C and lignin were not affected by HWA infestation; however, foliar N was higher in infested stands (S. Meyer unpublished data; see Cobb et al. this volume). Higher initial foliar N was found to increase the rate of N immobilization in decomposing foliage. Additionally, our data suggest that in many forests infested foliage may switch from a net sink to a net source of N more rapidly compared to uninfested foliage. This may be an important contribution to increased N availability in infested forests (Jenkins et al. 1999)

Altered microclimate has had an important effect on foliar decomposition. Surface litter decomposition was slowed in many infested forests due to poor conditions for microbial establishment on litter. However, our work suggests this effect may be limited only to litter at the forest floor surface. Cellulose paper buried at the forest floor mineral soil interface had significantly greater mass loss in infested stands. We also observed that as surface litter became buried, its rate of mass loss increased in infested stands.

In 2001 we began a comparative study of hemlock, black birch, and mixed litter decomposition. During 2002 we collected litter bags after six and twelve months of decomposition. Results after six months show that black birch litter decomposed more rapidly than hemlock and mixed litter (Fig. 1). The higher rates of litter decomposition that occur after the switch from hemlock to black birch litter-fall are likely a major mechanism of forest floor mass loss after hemlock mortality and logging (Kizlinski et al. 2002).

Taken as a whole, our study illustrates the dynamic effects of hemlock woolly adelgid infestation on decomposition in eastern hemlock forests. All stages of infestation studied had significant effects on decomposition by altering the chemical changes in litter, altering litter inputs (hemlock to black birch), and soil microclimate conditions.

Kizlinski M. L. D. A. Orwig, R. C. Cobb and D. R. Foster. 2002. Direct and indirect ecosystem consequences of an invasive pest on forests dominated by eastern hemlock Journal of Biogeography 29:1489-1503. Jenkins, J.C., J. D. Aber and C. D. Canham. 1999. Hemlock woolly adelgid impacts on community structure and N cycling rates in eastern hemlock forests. Can. J. For. Res. 29:630-645.
Orwig, D. A. and D. R. Foster. 1998. Forest response to the introduced hemlock woolly adelgid in southern New England, USA. J. of Torrey Bot. Soc. 125:60-73

Seasonal Dynamics of Hemlock Woolly Adelgid and Eastern Hemlock Foliar Chemistry

R. Cobb, B. Stadler and D. Orwig

We documented the seasonal dynamics of hemlock woolly adelgid (adelges tsugae - HWA) densities and eastern hemlock (tsuga canadensis) foliar chemistry (% C and N) over the 2002 growing season. We sampled 12 trees at three dates, May 31st, July 1st, and September 1st. Three branches from each tree were selected at random, cut and returned to the laboratory. We determined HWA density, length, and number of needles for each age class of hemlock foliage. HWA were recorded as egg-sack forming (individuals that matured over the 2001 winter) or non-egg sack forming (current generation). Considerable overlap of HWA generations occurred during our sampling, especially on July 1st when as many as 15 life cycle stages may be present on a single tree (McClure et al. 2001). Egg sacks were removed, foliage was dried at 45°C for 48h, ground to pass 800 um screen, and analyzed with a Fison CHN dry-combustion autoanalyzer. We found HWA densities and population distribution to be highly seasonally dependent. Our initial sampling occurred before current year foliage was elongated and no HWA were found on this age class at the beginning of the field season (Fig. 1). However, current year foliage is clearly the preferred substrate, infestation increased to 100% by July and this age class supported the greatest number of individuals. In May and July, egg-laying sistens (individuals that survived over winter) dominated the HWA population on 1-year-old foliage. The relatively high percentage of HWA on the 2-year-old age class in May suggests that the crawling progredien-sexuparae (present in early May) may be forced to establish on the low quality substrate where survivorship is probably poor. The 2-year-old age class supported the least number of individuals but was the only age class with egg sacks at the end of the growing season. This suggests that HWA grows and differentiates more slowly on this substrate and may produce a small population of mobile HWA late in the growing season.

Foliar chemistry tended to be highest during the early portion of the growing season but did not vary significantly. Percent foliar N was greatest in HWA infested trees (Table 1). HWA density and foliar % N were positively correlated (Fig. 2) suggesting that chemical induction occurs in hemlock after infestation by this insect. This conclusion is supported by the increase in volatile organic compounds emissions documented from infested trees (Broeckling and Salom 2003). However, we cannot determine if increased N is a function of improved HWA performance on higher quality foliage or an effect of HWA herbivory. A simple study of pre and post infestation foliar N levels would be useful for resolving this question.

Broeckling C.D. and Salom S.M. 2003. Volatile emissions of eastern hemlock, tsuga canadensis, and the influence of hemlock woolly adelgid. Phytochemistry 62:175-180.
McClure M.S., Salom S.M., and Shields K.S. 2001. Hemlock woolly adelgid. USDA Forest Health Technology Enterprise Team - Morgantown, West Virginia. FHTET-2001-03 p.14.

Vernal Pools in the New England Landscape

E. A. Colburn

Relationships between short- and long-term changes in physical factors, biotic interactions, and species' distributions remain a key focus of ecological investigation. Vernal pools are small, shallow, seasonally flooded, fishless forest ponds that provide breeding habitat for forest amphibians and support diverse invertebrate faunas. They are tightly coupled to surrounding uplands through transport of materials and movement of animals. Short- and long-term variations in water levels and flooding duration are dramatic and may be cyclic. We hypothesize that species' distributions in pools reflect integrated responses to hydrology and other pool-specific physical factors, current watershed conditions, and historic influences. Our long-term studies of vernal pools provide an opportunity to examine the influence of climate, hydrology, and landscape-scale factors on the distributions and abundances of aquatic organisms.

Macroinvertebrate and vertebrate species' distributions and reproductive success in vernal pools are correlated with hydroperiod, geography, vegetation, and water quality. Hydrologic differences are associated with high within-year faunal heterogeneity between pools. Between years, vegetation and fauna change with variations in hydroperiod, and effects may carry over to the next year. Relationships among long-term hydrologic patterns, land use and plant cover in the watershed, pool vegetation, and pool fauna are unclear. Existing data suggest that periodic hydrologic cycles may extend over decades for some pools. Three unanswered questions are: (1) For a given pool, is there a predictable long-term pattern of hydrologic variation? (2) Among pools, are there parallel patterns in plant and animal distributions and abundances? (3) How do short- and long-term patterns in hydrology interact with other physical and biological variables to shape species' distributions and abundances?

We are analyzing a long-term database of aquatic invertebrates and habitat conditions from a cluster of 14 pools in the Cape Cod National Seashore (CACO), 3 additional Cape pools, a cluster of 7 pools in central Massachusetts, and a subset of 18 pools elsewhere in the state. Single-date samples taken from an additional 50 pools complement the long-term data set. Processing of samples and identification of organisms is ongoing for some sites. This long-term study (6-12 years of data per site over 19 years) of macrofaunal community dynamics in southern New England vernal pools investigates within-pool biological responses to year-to-year variations in climate and pool hydroperiods, between-pool differences in habitat and aquatic community composition, and the influence of landscape-scale variables within the broader context of long-term climatic variation and watershed changes.

Patterns appearing in our preliminary analysis include chemistry, vegetation, and historic land use. A pH gradient occurs across the central Massachusetts pool cluster, and it fluctuates between years in relation to changes in maximum water depths. Distributions of molluscs and fairy shrimp within the pool cluster suggest a threshold pH tolerance for these taxa. In the CACO pools, a gradient of pH associated with the relative abundance of Sphagnum sp. is evident. The distributions of a number of invertebrates, including a northern caddisfly, Phanocelia canadensis, are correlated with pH and Sphagnum in these pools (Colburn and Garretson 2003, submitted). In individual pools across the state, the distributions of some species of aquatic insects and amphibians closely track patterns in hydroperiod from one year to the next. Between pools, differences in vegetation and canopy closure, as well as average hydroperiod, explain some of the variation in species' distributions, as does past history of forest clearing for agriculture (see abstract by Musgrove and Colburn, this volume).

Colburn, E. A. and F. M. Garretson 2003. Habitat and life history of a northern caddisfly, Phanocelia canadensis (Banks) (Trichoptera: Limnephilidae), at the southern extreme of its known range. Northeastern Naturalist (Submitted).

Establishing a Long-term Monitoring Protocol for Freshwater Macroinvertebrates at the Cape Cod National Seashore

E. A. Colburn and H. Jensen-Herrin

Freshwater lakes, ponds, and wetlands, and the plants and animals that they support, are among the most important natural resources of the Cape Cod National Seashore. The development of an inventory of the biological resources of the Seashore's fresh waters, and the establishment of a program for long-term biological monitoring to detect changes in water quality and other parameters in Seashore waters, are high priorities for the National Park Service (Godfrey et al. 1999). We are beginning a process of research designed to categorize freshwater bodies in the Cape Cod National Seashore; to carry out an inventory of a selected subset of those water bodies in order to identify indicator species, assemblages, or metrics useful for monitoring changes over time; to develop a protocol for use by Seashore staff for long-term monitoring of the ecological state of the water bodies over time; to present the National Park Service with a report detailing the results of the initial classification of water bodies, inventory, and method development; and to provide the Seashore with a step-by-step protocol for implementation of the monitoring program.

Development of a protocol for monitoring the Seashore's fresh waters will involve several steps. Preliminary work requires preparation of a list of water bodies, classification of these waters into categories, selection of freshwater bodies of interest within each category, assessment of likely threats to each water body and of potential impacts of those threats, and identification of major habitat types within each chosen water body. Once study sites have been chosen and target habitats identified, we will sample for macroinvertebrates using several different techniques at different times of year. Identification of collected animals, and analysis of the results of the sampling in relation to the water body classes and a variety of physical, chemical, and habitat data will allow us to determine sampling methods and target taxa or associations that promise to provide the best resolution for monitoring long-term environmental changes of interest.

Effects of Hemlock Decline on Forest Stream Communities

E. A. Colburn and D. A. Orwig

The invasion of an Asian insect, the hemlock woolly adelgid (adelges tsugae), is focusing attention on hemlock and its ecological role in northeastern forests. The adelgid attacks both mature trees and seedlings. Infested trees die within a few years and are usually replaced by hardwoods (Orwig and Foster 1998). Preliminary research in the Delaware Water Gap (DWG) suggests significant differences in water quality and biota of streams draining hemlock- and hardwood-dominated stands (Snyder et al. 2002). Hemlock is an important component of New England's forests, including many nature reserves and water-supply watersheds, and the effects of adelgid-induced hemlock mortality on stream water quality and biodiversity are of concern.

Soils under conifers tend to be acidic, high in organic carbon, and low in nitrogen and phosphorus (Howard and Howard 1993, Lapin 1994). Streams in hemlock-dominated watersheds tend to have higher dissolved organic carbon (DOC) and lower dissolved nutrients than streams in hardwood catchments (Snyder et al. 2002). Adelgid-induced hemlock decline may increase decomposition and nitrogen cycling in the forest soil (Yorks et al. 1999), potentially contributing to increased nutrient export in groundwater and streams. In the DWG, hemlock-dominated streams are more thermally stable, remain cooler in summer and warmer in winter, and appear to be less susceptible to seasonal drying than hardwood-dominated streams (Snyder et al. 2002). Hemlock streams also support a richer aquatic macroinvertebrate fauna and a higher proportion of predatory fish species, and they appear more likely to sustain populations of native brook trout than hardwood streams. We predict that adelgid-induced hemlock decline in New England will alter habitat, increase algal primary production, change aquatic community composition, and affect the relative abundance of different functional feeding groups of invertebrates and vertebrates in streams draining catchments currently dominated by hemlock.

In late summer, 2002, we began collecting baseline data on habitat structure, water flow, temperature, stream salamanders, and aquatic macroinvertebrates in 17 headwater streams dominated by hemlock or hardwood in central Massachusetts. In summer, 2003, we plan to expand our surveys to a number of streams located along a space-for-time transect of hemlock decline from southern Connecticut to northern Massachusetts (Orwig and Foster 2002). We will also intensify sampling in headwater streams in hemlock and hardwood watersheds at Harvard Forest. Our stream data will be tied to analysis of vegetation composition and structure, soil decomposition, and nutrient cycling in the adjacent forest.

Howard, P.J.A. and D.M. Howard. 1993. Ammonification of complexes prepared from gelatin and aqueous extracts from leaves and freshly-fallen litter of trees on different soil types. Soil Biology and Biochemistry 25, 1249-1256. Klemens, Michael W. 1993. Amphibians and Reptiles of Connecticut and Adjacent Regions. State Geological and Natural History Survey of Connecticut Bulletin 112: 1-318.
Lapin, B. 1994. The impact of hemlock woolly adelgid on resources in the lower Connecticut River valley. US Forest Service Report. FS-6200-28b(3/92), 1-45.
Orwig, D. A. and D. R. Foster. 1998. Forest response to the introduced hemlock woolly adelgid in southern New England, USA. Journal of the Torrey Botanical Society 125:59-72.
Orwig, D.A., D. R. Foster, and D. L. Mausel. 2002. Landscape patterns of hemlock decline due to the introduced hemlock woolly adelgid. Journal of Biogeography 29: 1475-88.
Snyder, C.D., J.A. Young, D.P. Lemarie, and D. R. Smith. 2002. Influence of eastern hemlock (tsuga canadensis) forests on aquatic invertebrate assemblages in headwater streams. Canadian Journal of Fisheries and Aquatic Sciences 59, 262-275.
Yorks, T. E., J. C. Jenkins, D. J. Leopold, D. J. Raynal, and D. A. Orwig. 1999. Influences of Eastern Hemlock Mortality on Nutrient Cycling. pp. 126-133 In: Symposium on sustainable management of hemlock ecosystems in eastern North America, June 22-24, 1999, Durham, New Hampshire. US Forest Service General Technical Report NE-267.

Physical and Biological Characteristics of Headwater Habitat Streams in Central Massachusetts: A Preliminary Assessment

E. A. Colburn, J. Choiniere, H. Jensen-Herrin, D. Williams and C. Dunn

Intermittent headwater streams are poorly understood and receive limited regulatory protection in Massachusetts. This research is evaluating the role of headwater streams as habitat for salamanders and aquatic invertebrates, with a long-term goal of contributing to volunteer monitoring of these systems. Starting in late summer, 2002, we mapped and sampled 17 intermittent headwaters tributary to 6 perennial streams in Westminster, Princeton, and Petersham, northern Worcester County, Massachusetts. We flagged each stream at 20-m intervals, recording GPS coordinates and describing channel characteristics, substrate, gradient, flow, riparian vegetation, and land use between flags. Based on the mapping results, we subdivided each stream into 50-m study reaches for biological sampling. In late fall we began intensive visual surveys and net sampling for stream salamanders (Gyrinophilus p. porphyriticus, Desmognathus f. fuscus, and Eurycea bislineata) and terrestrial salamanders (Plethodon cinereus, Notophthalamus viridescens) in randomly selected, area-constrained transects 1 m wide by 4 x mean channel width in each reach. Salamander sampling is complemented by dip-netting for aquatic macroinvertebrates in square-meter quadrats upstream of each salamander transect. Sampling is ongoing and will continue until after the streams have dried in summer, 2003.

Some of the study areas were cleared historically for agriculture but have regrown to forest over the past century; some were not farmed but have been harvested for forest products; others have been continuously forested. All of the streams are small, one half to one kilometer long and one-to-three meters wide, and most of them flow into impoundments created by farmers in past centuries, or by beavers in recent decades. The history of intensive land use in central Massachusetts is reflected in the presence of unpaved roads at varying distances upgradient of all of the headwater channels, and some streams receive high loads of eroded sand from these roadways.

Our preliminary analysis shows that these streams do not follow the expected longitudinal pattern of ephemeral channels leading to interrupted, then interstitial sections before reaching perennial flow. Wooded wetlands with no surface flow or evident channels are interspersed along the length of several of the headwater streams. Perennially flowing sections occur upstream of wetlands and seasonally dry channel reaches. In most of these headwater streams, reaches of high gradient with boulders are interspersed among gravel stretches and meandering reaches with seasonally deep pools. Fountain mosses (Fontinalis sp.) are common on cobbles and occur in seasonally dry stream reaches. Preliminary data on benthic invertebrates from the streams suggest strong similarities in species composition both along the length of the headwaters and between streams.

A Nitrogen Experiment in a Maine Spruce-Hemlock Forest: Results from a Wet NH₄NO³ Canopy Fertilization

B. Dail, D. Hollinger, E. Davidson, J. D. Aber, H. C. Sievering, I. Fernandez, S. Goltz, N. Scott, M. LeClerc and E. Gaige

Roughly 90% of nitrogen oxides (NO_x) in the atmosphere are emitted as a result of human activities. The Midwestern and northeastern United States, and parts of Canada, are recipients of much of this nitrogen (N), deposited as both wet and dry deposition. Northern forests are thought to be limited in growth by the availability of N. Increased N inputs have coincided with increased carbon (C) contents reported in northern forest inventories. The fate of anthropogenic N, its residence time, and the potential N-fertilization effect on forest growth remains unknown. We report throughfall findings from an ecosystem-level N addition made as wet precipitation to the canopy of a spruce-hemlock forest in central Maine. Beginning in the growing season of 2001, ammonium nitrate was sprayed on 21 hectares of forest canopy by helicopter in six applications from May through September. The addition rate was 18 kg N ha^-1yr-1; approximately 3X wet+estimated dry ambient atmospheric deposition for this site. Additional fertilizer N at a 1% ¹⁵N enrichment was aerially applied to 30x100 m (0.3 ha) plots during each fertilization. These smaller plots received N as either ¹⁵NH₄ or ¹⁵NO₃. We measured "throughfall" N reaching the forest floor the day of the fertilizations, and also in throughfall on an event basis after all precipitation events prior to winter.

Filter paper randomly placed on the forest floor the day of three fertilizations was retrieved ~2 hours post-fertilization and leached with a 2M KCl solution to remove mineral N. Approximately 9% (range: 4 to 20%) of the N applied was reaching the forest floor immediately after the foliar application. An additional 30% of the total N added to the canopy was recovered in throughfall over the course of the growing season. Finally, 5-10% of the experimentally added 18 kg N ha^-1 yr-1 was returned to the atmosphere by gaseous release of, mainly, NH3. Thus, about 50-60% of the experimentally added N was retained within the canopy, lost by way of stemflow, gaseous losses from the canopy, or translocated within the trees. We are using recovery of ¹⁵N in foliage samples taken at the end of the growing season to assess total N uptake by the canopy;. The N uptake fraction = (foliage¹⁵N - background¹⁵N) * (canopy N mass + applied N)/(applied¹⁵N*applied N mass). Using this latter approach, we estimate a very different value for canopy N retention. The ¹⁵N technique suggests uptake may be as low as ~10%. The ¹⁵N approach is likely to underestimate canopy uptake because it assumes no isotopic exchange or redistribution within the canopy. The mass balance approach may be an overestimate since gaseous N losses from the canopy might account for greater losses of the applied N. Both methods suggest that some of the applied N is being retained by the canopy, and further sampling of woody tissue should demonstrate whether significant translocation is occurring. The effects of N uptake by the canopy on C exchange and ecosystem C storage has not yet been detected after two years of eddy covariance and biometric measurements, and long-term monitoring of these processes is continuing.

QTL Analysis of Parental Effects on Seasonal Dormancy in Arabidopsis thaliana

K. Donohue, S. Takao*, A. Wheeler and L. Dorn

Seed dormancy is an important life history character that can strongly influence fitness. Appropriate germination responses to environmental cues may contribute to persistence in a variable environment or expansion into novel environments. This research investigates the genetic basis and adaptive significance of germination responses to seasonal cues in Arabidopsis thaliana. Using a combination of quantitative trait loci (QTL) mapping and manipulative selection studies in the field, we will determine which QTL contribute significantly to variation in seasonal dormancy and how variation in seasonal germination timing influences fitness.

This research program focuses on germination responses to two seasonal cues: the photoperiod under which maternal parents mature seeds, and the temperature experienced by seeds after dispersal. Plants can mature seeds in the late spring under long photoperiods or in late autumn under short photoperiods. Seeds dispersed in spring experience hot temperatures immediately after dispersal, while seeds dispersed in autumn experience cold temperatures. We therefore manipulated the photoperiod experienced by maternal plants and the temperature experienced by seeds in order to determine their effects on germination.

In one summer project, we used seeds collected from replicates of recombinant inbred lines that had been grown in either a long day or short day photoperiod under controlled conditions. After the seeds afterripened for six months, we exposed them to different incubation temperatures in the dark before putting them in the light at temperatures that promote germination (23°C). Incubation treatments included two weeks at 31° C, two weeks at 23°C, and two weeks at 31°C followed by five days at 4°C.

The incubation temperature significantly influenced the percentage of seeds that germinated. Warm incubation caused less germination, while cold broke the dormancy induced by warm incubation. Seeds incubated at 23° C displayed no dormancy. Maternal photoperiod did not influence percentages of germination in these fully afterripened seeds. Significant genetic variation was detected for the percentage of seeds that germinated in both treatments that included a warm incubation period. Genetic correlations between germination in the different incubation treatments were strong, but significant genotype-environment interactions were also detected. Therefore, different genotypes responded differently to the different incubation treatments. A very preliminary analysis identified one chromosomal region associated with variation in germination in both the warm incubation and warm/cold incubation treatments. Evidence for effects of a different chromosomal region was found for germination in only the warm incubation treatment. Therefore, one gene appears to be associated with variation in germination regardless of incubation treatment, whereas another gene may determine germination only under certain conditions.

Quantitative Genetic Architecture and Plasticity of Seed Dispersal in Arabidopsis thaliana

K. Donohue, N. Wender* and C. R. Polisetty

This project addressed the genetic basis of seed dispersal and its plasticity in Arabidopsis thaliana, a model organism for molecular genetic studies. Our goal was to determine whether dispersal ability has a genetic basis so that we can predict whether dispersal can evolve in response to natural selection. We also characterized plasticity in dispersal in response to density and seasonal conditions. By determining whether genetic variances and covariances change with density, we will be able to predict how the evolutionary potential of dispersal changes as dispersal itself evolves.

The most significant finding from this project to date is evidence for a novel genetic constraint on the evolution of dispersal that is mediated by phenotypic plasticity. First, we identified maternal plant traits that significantly influenced dispersal. Branching angle, branch number, silique length, percent fruit dehiscence, and the percentage of fruits on the primary stem all influenced seed dispersion patterns. Second, we found significant genetic variation for these maternal traits. The genetic variation for maternal traits that influence seed dispersion patterns, moreover, did not differ with density. However, these plant traits determined dispersal much more strongly when plants were grown at high density than when they were grown at low density. Therefore, genetic variation for dispersal itself would be greater at high density than at low density. As increased dispersal ability evolves and plants experience lower densities as a result, the expression of genetic variation for dispersal is expected to decline. Thus, a limit to the evolution of dispersal may result. Genetic constraint can thereby be mediated by the evolution of the trait itself.

We found substantial plasticity in dispersal in response to density. Plants that grew in high density dispersed their seeds shorter distances, and their seeds experienced less sibling competition although non-sibling density was very high. Thus plants that were themselves efficiently dispersed and consequently grew in low density also dispersed their own seeds efficiently. This dynamic can create a positive environmentally-induced correlation between generations for seed dispersal that may accelerate responses to selection on dispersal.

Effects of Nutrient Stress on a Co-evolved Food Web

A. M. Ellison, and N. J. Gotelli

Ecological communities are made up of interacting species that have evolved in response to particular types, frequencies, and intensities of stress and disturbance. Environmental cycles with lengthy return times, natural and anthropogenically-induced climate change, and unintentional or deliberate alterations of the environment may modify the types, frequencies, or intensities of stressors. The consequences of such stressors for species interactions and community structure are unknown, and may be especially severe for communities of strongly co-evolved species. In the summer of 2003, we will begin a three-year research project to determine how a novel stressor - atmospheric deposition of nutrients - is changing the dynamics of the detritus-based food web that has co-evolved with its host, the northern pitcher-plant Sarracenia purpurea. The focus of this research is on a detritus-based food web, because 70-90% of all primary productivity enters food webs as detritus. However, the development of model systems for understanding the structure and dynamics of detritus-based food webs has not paralleled similar developments in studies of grazing food webs, in which the basal trophic level is composed of phytoplankton, algae, or vascular plants. Specifically, in the course of this research, we will test three null hypotheses using a combination of greenhouse and field experiments:

1. Nitrogen derived from atmospheric sources and nitrogen from captured prey are used equivalently by the Sarracenia food web and the plant itself;
2. The trajectory of assembly of the Sarracenia food web is independent of initial starting composition, habitat type, and the type of, and rate at which, nutrients are supplied;
3. Host plant growth morphology is independent of food web structure and nutrient supply.

The results will be used to develop Markov models of community assembly and coupled differential equation models of food web dynamics. These models will be used to predict the responses of integrated communities to environmental stress. This research project will reveal the importance of anthropogenic stressors on co-evolved systems, and contribute to a deeper understanding of the assembly of food webs by contrasting effects of top-down and bottom-up forces on the dynamics of non-equilibrial assemblages. It also will lead to an appropriate model system for the experimental study of detritus-based food webs.

Biology of Loss and Replacement of Core Species Caused by an Invasive Insect

A. M. Ellison, D. R. Foster, D. Orwig and the FIBR collective

We describe a new five-year project aimed at elucidating and modeling how a single core species - Eastern Hemlock - determines the biological, biogeochemical, physical, and ecological processes of the ecosystem it creates. We propose to document and predict how the elimination of that core species, its replacement, and subsequent community re-assembly change these fundamental biological and ecological processes. This research program uses as a model system the current elimination of eastern hemlock by an invasive, exotic insect, the hemlock woolly adelgid.

The four objectives of this project are to:

1. Develop a comprehensive understanding of the way that a single core species can create and maintain an environment that supports a wide variety of other organisms.
2. Predict how a biological community is reorganized and reassembled following the loss of that core species and assess the implications of this loss at local, landscape, and regional scales.
3. Elaborate a mechanistic understanding of the biological, biogeochemical, and physical processes associated with this reorganization and reassembly.
4. Produce a predictive, general model of how coupled biological-physical systems will respond to loss and replacement of core species.

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

D. R. Fitzjarrald, R. M. Staebler, M. Czikowsky, R. K. Sakai and A. Tsoyref

1. Forest phenology and streamflow. We hypothesize that the greatly enhanced growing season evapotranspiration (ET) by 'robbing' streams of runoff provide a phenological indicator. In last year's abstract, we identified three independent runoff characteristics that accompany spring onset and leaf emergence. These are the return of the P-R curve to pre-snowmelt pulse values, the streamflow recession time constant for return to base flow after a rain event, and the amplitude of the diurnal streamflow signal. At some level, replacement of pasture with forest should alter the long-term hydrological balance in an analogous manner, but these observed effects of reforestation on watershed runoff are often subtle. However, we detected long-term changes in the streamflow recession time constant that may be attributed to successional forest growth (Czikowsky 2003).

Two competing factors are at work that could change the length of streamflow recessions due to successional forest growth. The more apparent factor is that growing-season streamflow recessions should be shortened due to increased ET. The less apparent factor comes from application of the variable source concept of hydrology (Dunne 1978). Successional forest growth not only results in increased forest cover, but in an increasingly well-developed rooting system. Greater soil infiltration at depth should occur from a better-developed rooting system. With greater soil infiltration, a larger proportion of precipitation will seep deeper into the soil column and result in delayed baseflow runoff rather than the more quickly contributing overland flow. Consequently, streamflow recessions should be lengthened. In the Wappinger Creek, NY watershed, there has been about a 40 percent increase in forest cover since streamflow records became available in 1928. Note that recession lengths here (Fig. 1) increased in the long-term during the late spring period, owing to higher soil infiltration and a greater delayed runoff contribution which lengthened recession times. During the growing season, the effect of increased ET demand by the successional forest cover dominates as recession lengths are shortened in the long-term. By contrast, we observed an increase in the recession lengths in the middle of the growing season during the 1970-2000 period for a central Massachusetts watershed of similar size but with a different land-use history, the East Branch of the Swift River (Fig. 2). Forest cover fractions on this watershed have changed less than 10 percent over the period of record. The longer recession times may indicate that the aging forest of this region is transpiring less than it was earlier in the 20^th century.

After identifying the three methods that we used to infer spring onset, we examined which watershed characteristics are related to these indicators. The diurnal streamflow signal appears in the smallest watersheds (< 200 km2) more frequently (Fig. 3) and with greater amplitude (Czikowsky 2003). The fraction of time the diurnal streamflow signal appears as well as its amplitude scale to watershed area (Aw) through the power-law . Mainstream length (Eagleson, 1970) and total stream length (Rosso and Bacchi 1991; Crave and Davy 1997) have also been found to scale to watershed area ( ). If we hypothesize that the riparian area has approximately constant width. Ar is proportional to total stream length (L), increased riparian area fraction in small watersheds contributes to the greater likelihood of the diurnal streamflow signal being observed there.

Until the late 1990's, only daily-averaged streamflow data were available for nearly all of the USGS streamflow observing sites. Reasons for not archiving high-frequency data included limited storage capacity and that daily data was believed to be sufficient for most hydrological studies. The recent practice of archiving 15-minute streamflow data at the USGS gauging sites is clearly of great utility and should be continued.

2. Canopy-atmosphere exchange processes. We continue our efforts to understand the physics of subcanopy flows and their connection to potential CO₂ flux deficits. Fluxnet carbon budgets now rely on an empirical rule of thumb to account for respiration on calm nights. To understand flows on calm nights, we developed a microkinematic approach, the DRAINO experiments performed from 1999-2002.

Canopy structure. In 2002, subcanopy measurements of the wind and CO₂ fields in a 120x120m grid around the EMS tower were conducted from mid-April to early December, covering the whole growing season. In addition, in collaboration with G. Parker of SERC, detailed canopy structure measurements were made using the methods of Parker et al. [2002], walking horizontal transects through the forest with an upward-looking laser range finder. Laser return readings were taken at 170Hz and binned in horizontal 2m sections along the path walked, equivalent to about 680 data points. Corrections were performed to correct for the bias to closer targets by assuming that leaves are Poisson randomly distributed [MacArthur and Horn, 1969]. An absolute canopy area density (CAD) is then derived using an empirical relationship between leaf area indices measured by leaf litter collection and laser return signals [Parker and Lefsky, 2003].

Six 80m transects covering the "DRAINO grid" were made, with the EMS tower roughly in the middle. The northeast transects were distinctly bottom-heavy, primarily because of the dense, low hemlock found along this transect. Transects oriented towards the southwest indicated more evenly distributed canopy elements. Here the forest is dominated by maples and oaks. The average of all CAD profiles reflects these two types of canopy encountered near the EMS tower, and displays a bimodal distribution (Fig. 4). 90% of all canopy area is found below 20m, although some canopy elements are found as high as 25m. Direct observation of understory winds. Understory flows in complex terrain are affected by four physical forces, as shown in the momentum equation, here simplified to two dimensions:

where h represents the topography, t is the stress tensor, qv is the virtual potential air temperature and qv' the local departure from it. FD represents the sum of all drag forces. The Boussinesq approximation has been applied, and the rotation of the Earth and molecular dissipation neglected.

The first force is coupling with the flows aloft, which can take the simple form of momentum transfer into the canopy through the stress, term [4] in Eq. (1). The second force [5] is gravitational, when radiative cooling near the ground creates negative buoyancy, which will cause the colder air to flow downhill. Term [6] (pressure perturbation) is also potentially important, especially in the lee and wake of hills. Upon generation, these understory flows are then modified by the immediate local topography and by obstacle effects (e.g. the vegetation). At any given place and time, different combinations of these forces and complications can be expected.

Drainage flows. Depending on which force dominates, the subcanopy flow should have different distinguishing characteristics. In Fig. 5, the effect of the buoyancy term [5] is made clear. As the fraction of this force in the sum of all forces increases, the north and northwest wind directions starts to dominate. This direction agrees with the longest continuous slopes at Harvard Forest.

Consequences of understory flows on the carbon budget. We describe the CO₂ budget in a conceptual box of height h=29m and horizontal size of the Draino grid in terms of the equation of conservation of c:

Term [1] is the storage term, [2]+[3] the horizontal advection, [4] the vertical advection term advocated by Lee [1998], [5] the eddy covariance flux at z=h, and [6] the sum of all sinks and sources between z=0 and z=h, including everything crossing the lower boundary at z=0.

Table 1 summarizes the statistics of the relationship between nights when the eddy covariance fails to properly measure the respiration flux of CO₂ ("deficit nights") with subcanopy flow conditions for all four years of "Draino" observations. Nights with drainage flows do indeed correlate well with nights with missing eddy fluxes. In fact, high buoyant fractions describe these nights significantly better than the commonly used u* (friction velocity) criterion.

Fig. 6 summarizes the mean diurnal variation of all measured components of the CO₂ budget. Both the measured eddy covariance (EC) and the u*-corrected signals are shown. The advective terms are significant at night and disappear during the day. The sum of EC, storage and advective terms agree within limits of error. Table 2 separates nights with clear deficits from those with apparently correct EC fluxes. Horizontal advective terms increase during deficit nights. However, adding the advection term does bring deficit and non-deficit nights into closer agreement, within the resulting error limits.

Crave, A. and P. Davy. 1997. Scaling relationships of channel networks at large scales: Examples from two large-magnitude watersheds in Brittany, France. Tectonophysics, 269: 91-111.
Czikowsky, M. 2003. Seasonal and successional effects of streamflow and evapotranspiration. M. S. Thesis, The University at Albany, SUNY.
Dunne, T. 1978. Field studies of hillslope flow processes. Hillslope Hydrology, M. J. Kirkby, Ed., John Wiley and Sons, 227-293.
Eagleson, P. S. 1970. Dynamic Hydrology. McGraw-Hill, 462 pp.
Lee, X. 1998. On micrometeorological observations of surface-air exchange over tall vegetation. Agric. For. Meteorol., 91, 39-49.
MacArthur, R.H. and H.S. Horn. 1969. Foliage profiles by vertical measurements. Ecology, 50, 802-804.
Parker, G. G., M. E. Harmon, M. A. Lefsky, J. Chen, R. van Pelt, S. B. Weiss, S. C. Thomas, W. E. Winner, D. C. Shaw and J. F. Franklin. 2003. Three dimensional structure of an old-growth Pseudotsuga-Tsuga canopy and its implications for radiation balance, microclimate, and atmospheric gas exchange. Ecosystems (Submitted).
Parker, G. G. and M. A. Lefsky. 2003. Leaf overlap in some deciduous broadleaf forests and its implications for canopy structure and radiation environment.(In preparation).
Rosso, R. and B. Bacchi. 1991. Fractal relation of mainstream length to catchment area in river networks. Water Resources Res. 27: 381-387.

The Regulation of Transpiration Rate in Plants

P. Franks, M. Zwieniecki and N. M. Holbrook

The survival of all land plants depends primarily on their ability to maintain hydration while performing photosynthesis. Central to this task is the regulatory role of stomata. Increasingly dry air is a threat because the hydraulic gradients required to sustain higher transpiration rates, particularly in trees, may constitute an intolerable state of cellular desiccation or induce air embolisms in xylem. To protect against this, stomata appear to have evolved functionally to down-regulate transpirational water loss as air becomes drier. The intensity of this regulation is known to vary considerably between species (Fig 1), but little is known about both the mechanism of this regulatory process, or the basis of its variation across species.

Recent data on the hydromechanical properties of stomatal guard cells and hydraulic conductivities of leaf fine veins suggests that control of transpiration rate must depend heavily on the nature of the hydraulic resistances of the non-vascular tissue between leaf xylem elements and sites of evaporation on the guard cell walls and/or sub-stomatal chamber walls. Our hypothesis is that variations in leaf tissue hydraulic resistances (Fig. 2) influence the mode of stomatal control of transpiration rate. It is possible that these differences in hydraulic properties are determined by different leaf anatomical characteristics. Our objective in this study is to use micro-probe techniques to manipulate and measure the degree of hydraulic coupling between leaf veins and different points along the hydraulic flow path between veins and evaporative sites near stomata, and to use this quantitative information to develop a mechanistic model of transpiration control.

This study will yield insights into the role played by leaf hydraulic resistances in stomatal control of transpirational water loss. Information gained from this will greatly improve our understanding of the linkages between leaf anatomy, leaf hydraulic properties and regulation of transpiration rate in plants. Application of this information to both living and fossilized leaf material will aid our interpretation of plant evolution and provide an improved mechanistic basis for understanding forest responses to changing climates.

Chronic Nitrogen Enrichment Affects the Structure and Function of the Soil Microbial Community

S. D. Frey and M. Knorr

Inputs to the terrestrial N cycle have doubled in the past century due to anthropogenic activities, particularly fertilizer use and fossil fuel combustion. The biota in many terrestrial ecosystems evolved under low N conditions and thus, one anticipated consequence of elevated N inputs is the change in species composition and/or diversity. There are still major uncertainties in our understanding of how N enrichment will impact the soil community and ultimately ecosystem function. We have begun a study to examine the effects of N enrichment on the relative abundance, diversity, and activity of bacteria and fungi in forest soils exposed to chronic N additions. During fall 2002 we collected and analyzed soil samples from the Chronic Nitrogen Addition Plots at Harvard Forest. One soil core (5.6 cm diam; O horizon plus 10 cm of mineral soil) was collected from four of the 5 x 5 m subplots within the control, low N (5 g N m^-2 y^-1), and high N (15 g N m^-2 y^-1) plots in both the hardwood and pine stands. The samples were analyzed for active bacterial and fungal biomass, microbial functional diversity, and the activities of celluloytic and ligninolytic enzymes. To date, we have found that long-term N additions have decreased total microbial biomass, relative fungal biomass, functional diversity, and lignin-degrading activity (Fig. 1). Total microbial biomass was 35 and 38% lower in the high N compared to control plots in the hardwood and pine stands, respectively. The proportion of the microbial community comprised of fungal biomass declined from 70-94% in the control plots to 45-53% in the high N treatments. Catabolic evenness, assessed by measuring the respiratory response of the microbial community to the addition of 25 different C sources, indicates that the functional or metabolic diversity of the microbial community is reduced under long-term N additions. Finally, the activity of phenol oxidase, a lignin-degrading enzyme, was significantly reduced by N enrichment, with a particularly strong effect observed in both the low and high N plots in the hardwood stand. We expect that these changes in the microbial community will be reflected in the quantity and quality of the active soil organic matter pool. During the coming year we will be measuring the concentrations of the amino sugars muramic acid and glucosamine in order to quantify the relative contributions of bacteria and fungi, respectively, to the production of microbial-derived organic matter. We will also conduct an experiment to examine the relative contributions of mycorrhizal and saprotrophic fungi to N immobilization in these soils.

Quantifying the Importance of Belowground Plant Allocation for Sequestration of Carbon In Soils

J. Gaudinski, M. S. Torn, T. Dawson, J. Bird and S. Mambelli

How much carbon is pumped belowground by root growth in temperate forests? How long do roots live and how does the fate of this belowground productivity compare to the microbial and humic transformations of leaf and needle productivity? These are pressing questions for global change and carbon sequestration research, as well as for ecological theory about plant allocation strategies and their nutrient feedbacks.

Our work aims to better characterize dynamics of carbon cycling belowground for roots versus aboveground plant allocation. We are (1) Quantifying the stocks and lifetime of live fine and coarse roots; (2) Determining the lower bound of NPP "pumped" into soil carbon through these roots; (3) Comparing leaf and root decomposition including rates, microbial communities and humification products; (4) Characterizing the turnover times of soil organic matter pools, and (5) Tracking the partitioning of recent plant photosynthate to rapidly lost root respiration and exudate mineralization, and more slowly lost root tissues and soil organic matter (SOM). Our approach will take advantage of several new methods (radiocarbon analysis of roots and SOM, ¹³C tracking of decomposition products, and isotope-label PLFA analysis). The radiocarbon method in particular allows direct determinations of root age, a measure not currently possible with any other technique. At four northern latitude forest research stations, including Harvard Forest, MA, we will make comparisons of belowground allocation sequestration potential based on species and forest type, including deciduous vs. conifer and re-growing vs. mature managed forests. At Harvard Forest we are focusing on items 1-3 discussed above, namely fine root dynamics and the decomposition dynamics of roots vs. leaves.

Estimates of high belowground net primary productivity (50% or more) in forest ecosystems are often based on assumptions that all fine roots (< 2 mm in diameter) live and die within one year. Recent radiocarbon (¹⁴C) measurements of fine root cellulose in three eastern temperate forests of the United States show that at least a portion of fine roots are living for more than 8 years (Gaudinski et al. 2001) and that fine root lifespans likely vary as a function of both diameter and position on the root branch system. New data from Harvard Forest, MA, further support the idea of large variations in fine root lifespans. We have sampled soil cores in the O, A and B horizons and sorted all roots (removing fern roots) into five size classes (< 0.3, 0.3-<0.5, 0.5-<0.8, 0.8-<1.1 and 1.1 - 2 mm in diameter) and live versus dead root categories. Samples have also been analyzed for age as a function of position on the root branching system for red maple. Additionally, we have compared ¹⁴C-derived- and mini-rhizotron-derived lifespans for sugar maple from Michigan (in collaboration with Kurt Pregitzer). Our preliminary results are (1) position on the root branching system shows a correlation with age, with ages up to 7 years for 4th order roots of red maple (Fig. 1). (2) The method by which roots are sampled also affects the (¹⁴C)-estimated age, with roots sampled via soil cores showing longer lifespans and less variation in lifespan relative to roots sampled by position on the root branch system (when similar diameter classes are compared). (3) Mini-rhizotron and radiocarbon lifespan estimates agree well when a correction is made to account for the age of carbon making up new root tissues (~ 1 year). This age correction is right within the resolution range of our bomb-¹⁴C technique and thus this value comes from parallel work being done at Oak Ridge National Research Park (where there is an enriched ¹⁴C signature ecosystem wide). In the near future we hope to compare our ¹⁴C-derived data with minirhizotron data from Tom Gower's research group working in the same area at Harvard Forest.

To predict the sequestration potential of plant C inputs, it is essential to better understand the biological and chemical regulation of the conversion to stable soil organic matter (SOM). Much of the recent research in forest ecosystems has focused on the initial decomposition of above-ground litter. We are examining both above-ground (needles and leaves) and below-ground inputs (fine roots), and determining decomposition rate as well as the conversion of C inputs into stable SOM fractions. Our objectives are to assess the effects of soil depth (O vs. A horizon) and litter type (roots vs. needles or leaves) on C and N stabilization rates and humification pathways in temperate forest soils. As part of this goal, we are conducting a litter bag study at Harvard Forest, MA (deciduous forest) and Blodgett Forest, CA (conifer forest). In fall 2001, litter bags were placed in the organic and mineral (O and A) horizons at Blodgett Forest and Harvard Forest. The litter types are: three sizes classes of roots, needles (Blodgett); leaves (Harvard Forest), and a common wood substrate placed at all sites. By placing each kind of litter at both soil depths and having a substrate common to the forests, the results will enable us to evaluate the influence of soil environment, litter type, and litter chemistry on litter decomposition rates.

Gaudinski J. B., S. E. Trumbore, E. A. Davidson, A. Cook and D. Richter. 2001. The age of fine-root carbon in three forests of the eastern United States measured by radiocarbon. Oecologia 129: 420-429.

Carbon Exchange Measured in a 55-Year-Old Oak-Maple Forest on a Dry Upland Site in 2002

J. Hadley

Carbon exchange was measured using the eddy covariance technique on an upland site near the center of a forest fire that destroyed about 20 ha of forest in 1957. The forest within 200 m to the SW and NW of the carbon exchange site is dominated by red oak and red maple, which established after the fire, along with smaller numbers of black oak, white birch, black birch, eastern hemlock and white pine. Beyond 200 m to the SW is a ˜70-year old red pine plantation and older forest outside the burn, while to the NW are older oaks, maples, black birch and some white pine.

Eddy covariance measurements were begun in May 2002, along with measurements of soil respiration using chambers and a portable gas analyzer. Daily net carbon storage by the forest was first observed on May 21, but cold and cloudy weather prevented consistent daily net carbon storage until after May 29. The forest stored 1.1 to 1.4 Mg/ha of carbon per month in June, July and August, and slightly less than 1 Mg/ha in September. Data from the dormant period beginning in October are still being evaluated.

One of the challenges for estimating carbon exchange of the relatively young forest in the area burned in 1957 is the relatively small size of the burned area. A mathematical model to estimate the source area for measured carbon fluxes (footprint model) based on the work of Schmid (1994) was created using a detailed wind profile above the canopy. The model shows that nighttime footprints are much larger than the 1957 burn, but in daytime the footprint is sometimes primarily in this area. A detailed analysis of measured carbon exchange rates as a function of footprint area is currently underway.

Soil respiration was measured at 30 locations to the SW and NW of the tower where eddy covariance measurements of carbon exchange were conducted. Soil respiration was primarily affected by temperature, which explained over 90% of the variation in the average rate of the 30 sites. However, during July and August low soil moisture also limited soil respiration, as indicated by a large increase in respiration rate soon after a precipitation event. The highest average soil respiration rate for all 30 sites was about 5 µmol m^-2 Ssup>-1 (215 mg m^-2 Ssup>-1), similar to that measured at other sites at Harvard Forest, except during very wet years (Savage and Davidson 2001). There was no clear difference in soil respiration between the forest established after the 1957 fire and older forest outside the burn, but total basal area of trees within 6 m of a particular site was positively correlated with the soil respiration rate, as was the basal area of oak trees in particular.

Savage, K.E. and E.A. Davidson. 2001. Interannual variation of soil respiration in two New England forests. Global Biogeochem. Cycles 15: 337-350.
Schmid, H.P. 1994. Source areas for scalars and scalar fluxes. Boundary-Layer Meteorol. 67: 293-318.

300 Years of Forest Change in Massachusetts: Forest Cover in 1830

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

In 1830 the Commonwealth of Massachusetts mandated that all towns make maps at a scale of 100 rods to an inch (19,800:1). All towns depicted cultural features such as roads and important industrial, educational and governmental buildings; most maps also show environmental features such as woodland cover, meadows, and water bodies. Since 1830 was near the period of maximum agricultural land clearance in Massachusetts, the maps serve as a reasonable estimate of the maximum extent of forest clearance. This map series has proven to be an invaluable source for historical woodland cover data. While other sources, such as tax evaluations, censuses, and State Foresters' Reports list only total area of forest, the town maps are spatially explicit and allow spatial analysis and description of forest cover and subsequent change at the local level.

The Harvard Forest has recently finished digitizing the town maps for the whole state into a Geographic Information System (GIS) to facilitate future analysis. Such analysis will be valuable in archeological, historical, and natural resource assessment, planning, and management. Recent and current studies have used the woodland data to identify potential primary forest sites and areas of historic agriculture, to describe long-term trends in forest cover, and to compare with other sources of long-term data.

Ecosystem-Atmosphere CO₂ Exchange in a Maine Spruce-Hemlock Forest: Updated Results

D. Hollinger, B. Dail, E. Davidson, S. Goltz, J. Lee, H. Hughes, C. Rodrigues, J. Walsh, N. Scott, J. Aber, H. Sievering, I. Fernandez and M. LeClerc

Measurements of carbon dioxide exchange between a spruce-hemlock forest and the atmosphere have been carried out since 1996. The complete flux data set was recently re-evaluated with corrections applied for low frequency flux losses. CO₂ fluxes were examined as a function of the friction velocity, u*, and a minimum threshold of 0.25 (rather than 0.20 as used previously) used for accepting nocturnal flux values. These and other minor changes result in the following integrated NEE values for the Howland Forest for 1996-2001; 158, 153, 133, 178, 271, and 175 g C m^-2 y^-1. On average, the revised values are about 5% lower than previous estimates. Interannual variability of NEE is due about equally to variation in respiration and GPP. A variety of factors including total incident quantum flux, temperature, and precipitation affect either or both GPP (photosynthesis) and respiration. The 1996-2002 period of flux measurements at Howland (especially since 1998) represents a time of unusually warm summer temperatures, based on a 50-year climate record from the Millinocket National Weather Service Station.

Carbon Cycling at the Harvard Forest, a Bottom-up Approach

L. Hutyra, V. Y. Chow, C. C. Barford, E. H. Pyle, C. M. Jones, D. M. Bryant, G. W. Santoni*, S. Pears*, J. Wofsy and S. C. Wofsy

Agricultural land use and active management have changed the vegetation dynamics of the New England landscape. Temperate deciduous forests have been identified as a potentially significant carbon sink (Curtis et al. 2002). Our study uses a bottom-up approach to quantify carbon pools and fluxes thereby allowing us to independently test the findings of the EMS tower (see Urbanski et al. abstract) and understand the linkage between the forest and the atmosphere. We are specifically looking at variables driving the seasonal and annual dynamics of carbon within Harvard Forest. A second experiment was initiated in 2001 to directly measure the long and short-term changes in the carbon cycle resulting from active management.

Forest carbon cycling is not singularly driven by accrual in live wood, but encompasses a multitude of other variables. The forest's mortality and recruitment demographics characterize the future composition of the live biomass and create the material for decomposition (See Liu et al. abstract). Litter fall and chemistry provide a metric for assessing inter-annual variation in photosynthetic capacity and nutrient cycling.

In 1993, forty vegetation plots with a radius of 10-m, were established within the fetch of the EMS tower. In 1998, the vegetative plots were resurveyed and a suite of high frequency biometric measurements was added to the study. All trees greater than 10-cm in diameter were outfitted with spring mounted stainless steel dendrometers. Using fine resolution dendrometry measurements we could examine variations in tree growth and responses to long and short-term climatic forcing. In 1999, an active forest management study was created by combining three of the original study plots with six additional plots to monitor ecosystem responses to a typical New England harvest. From February through November 2001, the harvest site was selectively logged. Litter inputs have been monitored at both of these sites since the commencement of the biometric measurements.

Total carbon uptake by live trees showed significant inter-annual variation (Table 1). Late season growth appears to differentiate the total strength of the carbon sink (Fig. 1a). Severe summer drought followed by late season rain in 1999 likely dampened the total rate of carbon uptake. A sunny and warm 2002 growing season may have helped propel the growth rate into the fall. Prior to logging, the harvest site contained approximately 81-MgC ha^-1, 42.8-m³ ha^-1 of wood was removed leaving 56.1-MgC ha^-1 in post-logging live biomass. The total stock of coarse woody debris more than doubled following harvest (Table 2). The harvest site showed a dramatic increase in carbon uptake in 2002, rebounding to near pre-harvest levels in one-year (Fig. 1b). Continued monitoring of these two sites will provide valuable insight towards understanding the effects of natural variations and anthropogenic activities on the carbon cycle.

Curtis, P.S., Hanson, P.J., Bolstad, P., Barford, C., Randolph, J.C., Schmid, H.P., and Wilson, K.B. 2002. Biometric and eddy-covariance based estimates of annual carbon storage in five eastern North American deciduous forests, Agric. For. Meteor. 113: 3-19.

Using History for Land Management in New England

A. E. Ingerson

I am investigating whether, how, and why landowners and managers in New England are (or are not) using long-term or historical methods and data to make and evaluate practical decisions, through qualitative interviews with the staffs of nonprofit organizations and public agencies, as well as some private landowners and consultants.

Much science policy focuses on improving access to research, by broadening distribution or simplifying the language used to report results. However, my interviews to date have not identified accessibility as a critical issue. Instead, they have drawn attention to issues of institutional design and practice, for both management and research:

1. turnover State and federal resource management agencies are now shifting from a generation of employees trained to value institutional memory and long-term record keeping to a generation trained to respond to changing public values and implement "state of the art" research and policies. In the private sector, data and plans developed for short-term transactions (timber, land, or easement sales and donations) are seldom archived for future use.
2. continuity Some data collection has been motivated by the threat of potential litigation, or by a need for economic valuation. Paradoxically, data may neither be collected nor archived for land in uncontested, stable ownerships that are not expected to produce periodic income.
3. history as a static "past" History is often used to identify "reference" states to which ecosystems should be returned, or in which they should be maintained. Although many managers are skeptical of strict preservation as an approach to both ecological systems and human institutions, there is a dearth of practical examples using historical methods and analyses to understand, manage, or promote rather than prevent ecological and institutional change.
4. history as retrospective explanation By explaining the past too thoroughly, history can make past actions appear "overdetermined" or inevitable, and thus irrelevant to current decisions-which usually seem anything but inevitable. Historical ecology may compound this problem by documenting mostly the actual consequences of human actions, while managers face mostly choices between competing routes to the same intended consequences, or between competing intentions.

These tentative findings suggest several new or nonstandard approaches to promoting the practical application of historical ecology:

• support place-based partnerships between management and research, to encourage the integration of long-term knowledge from different specialties (including history and ecology, or the social and natural sciences), to help research and management "co-evolve" over time, and to provide visible intermediate benefits from long-term research and data archiving
• promote a stronger flow of ideas about change management between the managers of cultural and natural resources (for example, between practitioners of ecologically "adaptive management" and practitioners of "adaptive reuse" in historic preservation)
• create new models of historical research and teaching, centered on critical empathy, which could be used to help policymakers and the public understand the intentions of past and present managers, while still judging management choices by their long-term ecological consequences; such models would make historical methods and data more directly useful for adaptive management.

Landscapes Dominated by Non-Industrial Private Forest (NIPF) Ownership: Sustainability at Woodlot and Regional Scales

D. B. Kittredge and J. Burk

Sustainability of the forest at a regional scale in landscapes dominated by non-industrial private forest (NIPF) ownership depends on the often-independent actions and behaviors of thousands of private families and individuals.

These NIPF lands comprise the dominant forest ownership in many parts of the United States, and represent an important part of the greater forest landscape matrix even in parts of the region where industrial and/or public lands dominate. In northeastern states, NIPF lands can represent 75% or more of total forest area. While forest landowner attitudinal survey work in the past several decades has explored reasons for ownership, motivations, and perspectives on traditional management (e.g., do you intend to harvest in the next 10 years?), little if any study has focused on attitudes and, importantly, documented behaviors related to sustainability on their lands. Some landowner attitudes pertaining to the notion of sustainability can be inferred from earlier work (e.g., documented interest in wildlife habitat and nature, aesthetics, and privacy), but these do not directly link to sustainability or timber productivity on their own lands.

As the urban-rural interface expands from metropolitan centers, though this wooded landscape may appear to be forested from the air, it no longer sustains a number of benefits upon which society has grown to depend. In particular, timber harvesting declines as a viable and sustainable land use activity. We will use the assembled Massachusetts timber harvest spatial database to: 1. Study the decision-making process, priorities, and behaviors of different types of NIPF owners, in terms of sustainable harvesting, and sale/ development; and 2. Use landscape-scale spatial data and associated community-level demographic data to assess the extent to which such landscapes can remain sustainable producers of wood products in the face of expanding urban/ suburban influence. In so doing, we will identify characteristics of a NIPF landscape in which harvesting or the production of timber is no longer sustainable.

Logging as Disturbance in Massachusetts

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

In a pilot study in the 19-town North Quabbin region, we captured the spatial location and associated information describing all commercial timber harvesting from 1984 to 2001 (Kittredge et al. in press). Analysis indicates that harvest occurs randomly within this heavily forested region. There is no pattern of spatial occurrence with respect to terrain, forest cover type, soils or other natural features. There is likewise no pattern of occurrence with respect to roads. The only pattern that emerges is with respect to ownership. Some state agencies actively harvest, and others are more passive and their level of harvest is not proportional to the amount of the landscape they control. Interestingly, non-industrial private landowners collectively own roughly 60% of the forest, and were responsible for 64% of the harvested area, suggesting a relative interest in harvest that is not matched by attitudinal surveys which portray owners as disinterested in this form of disturbance on their land.

Based on the results of our pilot study, we are expanding data collection to cover the entire state of Massachusetts. The process is ongoing, and to date we have captured commercial harvest information for approximately 300 of the state's 350 communities dating back to 1984. Data has been gathered for 4 of the 5 DEM regional districts, covering all of eastern MA, central MA west to the Connecticut River Valley, and Berkshire County. This represents 9,212 harvest operations over this period, and a total of over 141,400 hectares. The average harvest operation is 15 hectares. More than 3/4ths of the harvest occurrences are on non-industrial private land, mirroring the forest ownership pattern in the state. 4.4 percent of the harvests have occurred on DEM state lands, and 8.8 percent have been conducted by the MDC. Detailed species data has been recorded for each harvest, and added to all files in the pilot study. With the exception of portions of Middlesex and Essex Counties, harvest records are available across the state for the 18-year period.

Data collection for the remaining western MA towns and verification of all spatial information should be completed by Summer 2003. We will use this information to study the influence of widespread harvesting disturbance on successional trajectories and future forest composition. Vegetation sampling in harvested versus unharvested areas will also enable us to evaluate the impact of harvesting on native and invasive species distribution and abundance. Together with spatial data on harvesting patterns, our results will provide information that is critical for guiding conservation planning across the region.

Kittredge, D.B., A.O. Finley, and D.R. Foster. 2003. Timber harvesting as ongoing disturbance in a landscape of diverse ownership. Forest Ecology and Management (In Press).

Paleolimnological Assessment of Human-Induced Impacts on the Nutrient Balance of Walden Pond (Massachusetts, USA) During the Last Three Centuries Using Diatoms and Stable Isotopes

D. KÖster, R. Pienitz, B. Wolfe, S. Barry, D. Foster and S. Dixit

The evaluation of human disturbance on ecosystems and the development of effective and appropriate restoration and conservation management strategies require knowledge of natural pre-disturbance ecosystem structure. The catchment and lake of Walden Pond, Massachusetts, have experienced multiple historical anthropogenic impacts, a site which has attracted much attention since the publication of the American classic "Walden" by the natural philosopher Henry David Thoreau. Despite the public interest in the protection of the lake as a natural and historical monument, no comprehensive study of the pre- and post-disturbance limnology of Walden Pond has been conducted.

As part of a multi-proxy analysis of a sediment core spanning 1600 years from Walden Pond, fossil diatoms reveal substantial changes in the nutrient status over the past 250 years, resulting from anthropogenic impacts on the lake and watershed. Following a period of environmental stability from about 430 AD to 1750 AD, the abundance of the diatom Cyclotella stelligera increased (Fig. 1), organic content declined, and bulk organic ¹³C decreased and bulk organic ¹⁵N increased. These changes coincided with logging in the watershed, and are mainly attributed to an increase in detrital input of inorganic sediment from the watershed. With the beginning of intensive recreational development of Walden Pond in the mid- 20^th century, oligotrophic diatom species were largely replaced by disturbance indicators and the diatom-inferred lake pH increased by 0.5 units, while the bulk organic carbon and nitrogen stable isotope composition markedly shifted to lower and higher values, respectively. These changes reflect higher nutrient availability and primary production due to inorganic inputs from erosion related to trails, beaches, and construction as well as increased nutrient inputs by waste water seepage into groundwater and massive swimmer use. During the last 25 years, diatom assemblages stabilized, suggesting that initial management measures have been effective in reducing the rate of eutrophication.

From about 600 AD to the period of European Settlement, a gradual increase in the abundance of Tabellaria flocculosa, which has been recently observed during spring overturn and late summer in Walden Pond, might indicate longer full circulation periods due to lower summer temperatures or earlier ice-breakup than before 600 AD.

Notably, the human-induced changes observed over the past 250 years are well beyond the range of natural variability of the past 1600 years, yet the pre-disturbance record provides a useful target for developing additional restoration and conservation measures to ensure future environmental protection of this historical site.

KÖster, D., Pienitz, R., Wolfe, B.B., Barry, S., Foster, D.R., Dixit, S.S. (2003). Paleolimnological assessment of human-induced impacts on the nutrient balance of Walden Pond (Massachusetts, USA) during the last three centuries using diatoms and stable isotopes. Canadian Journal of Fisheries and Aquatic Sciences (Submitted).

Variation in Ecosystem Respired CO₂ in a Deciduous Forest Is Related to Climatic Drivers

C. Lai, A. Schauer and J. Ehleringer

The carbon isotope ratios of carbon dioxide fluxes from terrestrial ecosystems are key measurements needed to constrain interpretations of sinks in carbon cycle analyses. The current research began in January 2001 and is a multi-faceted effort addressing photosynthetic and respiratory isotope exchanges across the biosphere-atmosphere boundary at four AmeriFlux sites (Harvard Forest, Howland Forest, Konza Prairie, and Wind River). We have developed, tested, and established automated CO₂ sampling devices at these sites, capable of continuously sampling CO₂ concentration and filling air flasks for subsequent isotope ratio analyses. We focus on understanding the magnitude of changes in the carbon isotope ratio of ecosystem respiration (δ¹³CR) through the growing season and a mechanistic basis for observed isotope changes. Observations at the Harvard Forest in 2001 indicate that δ¹³CR varied ~ 2.4 per mil during the growing season and were significantly correlated with the changes in environmental variables. These observations appear to relate directly to water stress and stomatal closure dynamics within this temperate deciduous ecosystem. After leaf fall, δ¹³CR varied by nearly 10 ‰, from -26.4 to -35.9 ‰, far exceeding the natural range of this ecosystem. It is suggested that fossil fuel combustion might have contributed significantly during winter.

Focusing at AmeriFlux sites provides a direct link to NEE measurements. Over the next 2 years, we will continue to measure seasonal observations of the isotope ratio of CO₂ fluxes. In addition, we will be combining eddy covariance and isotope observations to partition NEE into assimilation and respiratory components. This will provide an independent estimate of flux components over a growing season in response to seasonal stresses.

Both short-term and long-term monitoring of the seasonal and interannual variations in ecosystem ¹³C values feed directly into regional and global CO₂ inversion models as a constraint on land-ocean partitioning. Our data sets complement long-term observations of tropospheric CO₂, which are the best available long-term data sets constraining interannual estimates of global productivity. Our data sets are being linked with global CO₂ inversion modeling efforts (e.g., TRANSCOM) to better understand the seasonal dynamics of tropospheric CO₂, particularly with respect to how much land/ocean partitioning varies on an interannual basis.

Rapid and Transient Response of Soil Respiration to Rain

X. Lee, H. Wu, J. Sigler and C. Oishi

The need for a precise understanding of the terrestrial carbon sink and climate feedback has motivated extensive studies on environmental influences on soil respiration. However, current knowledge of soil respiration is based in large part on field research in fair weather conditions, and the process during rainstorms and its role in the ecosystem carbon balance are not well understood. Here we combine the data from eddy covariance (EC), rain simulation and incubation experiments to show that, for a mixed temperate forest at Great Mountain, Connecticut, soil respiration increases rapidly and instantaneously in response to the onset of rain and returns to the pre-rain rate shortly after the rain stops. The pulse-like flux, most likely caused by the release of active carbon compounds from the litter layer, amounts to an annual loss on the order of 0.2 t C ha^-1 yr-1 to the atmosphere, or as much as 20% of the net ecosystem production (NEP) of mid-latitude forests. If precipitation becomes more variable in a future warmer world, the rain-induced flux should play an important part in the transient response of the ecosystem carbon balance to climate, particularly for ecosystems on ridge-tops with rapid water drainage.

Evolution of Defense Chemistry in Alliaria petiolata Since Introduction into North America

K. C. Lewis and F. A. Bazzaz

Invasive species threaten the integrity and function of ecosystems across the globe. However, prediction of and prevention of invasion have thus far proved elusive because of the dearth of commonalities among invasive species that might predict invasiveness. One of the characteristics common among invasive plant species is escape from herbivory in their new range. It has been predicted that plants released from herbivory will adjust their resource allocation to reduce investment in defense, particularly when defenses are costly. This could result in increased availability of resources for other activities such as growth and reproduction. In invasive plant species, this may contribute to strong competitive ability and rapid spread. Alternatively, release from specialists and continued selection pressure from generalists in the new range could increase investment in defenses if they are mainly effective against generalists. By using a cruciferous invasive species, Alliaria petiolata, which has costly defenses, I am investigating whether and how this species might shift resource allocation between chemical defenses and other fitness-related traits. If this is an important contributing factor to invasiveness, then potential for release from herbivory and response flexibility in resource allocation could become important screening tools to predict which species may become invasive when introduced outside their home range.

Using field populations of these plants in New England (including Harvard Forest) and in Hungary, I have been measuring growth, herbivory, defense chemistry, and reproductive capacity to determine whether invasive US populations differ significantly from native-range European populations. Using herbivore exclusions, I will also be able to distinguish between constitutive (inflexible) defense expression and defenses induced by herbivore damage. These data will provide insight into the changes that have occurred since introduction in these plants and the role of defense chemistry in competitive ability and invasiveness.

Respiration of Coarse Woody Debris in Adjacent Shelterwood-Cut and Mature Central Hardwood Stands

W. Liu, D. M. Bryant, L. Hutyra and S. C. Wofsy

Carbon cycle investigations have recently noted the potential importance of dead woody biomass as a storage pool of organic C (Barford et al. 2002). In addition, dead wood provides food and shelter for multiple trophic levels from microbes to macrovertebrates. The decay of CWD transfers nutrients from to the forest floor thus influencing forest productivity (Harmon et al. 1986).

Coarse woody debris (CWD) is generally defined as the deadwood portion of litter fall measuring > 10cm diameter and includes snags (dead trees), logs and stumps. CWD stocks, mass loss rates and density changes are well characterized for forests of the Pacific Northwest Region but data for the Northern Hardwood forests remains sparse.

Logs of the dominant hardwood genera (Betula, Acer, Quercus) and conifers within the footprint of the EMS tower and the adjacent cut site (south of Pierce Farm Rd) were tagged and numbered, measured for diameter at the midpoint. The current state of decay of each log was categorized to five decay classes. Locations were logged with a handheld GPS unit.

Cross-section disks were removed from the midpoint of logs that were randomly chosen from this tagged population. Respiration of decay organisms within the wood is trapped in the pore spaces resulting in accumulation of CO₂ that often reaches > 10%. Consistent with Chambers et al. (2002) we exposed CWD samples to ambient air, under shade, for 3 hours prior to measurement. Three replicates of each factorial class (genus X decay class X diameter class) were measured in each of the two stands.

Flux of CO₂ from cross-sections was measured in the headspace of one of 2 static chambers depending on the size of the sample. Concentration of CO₂ in the sample stream was measured at 5 sec intervals with an Infra-red Gas Analyzer (IRGA). The rate of concentration change was determined by the slope of the time vs. concentration regression.

Disks were weighed prior to flux measurement. Samples were measured for volume by water displacement and dried in a 105° oven to constant mass. Moisture content of the disks was calculated as a function of oven-dry vs. fresh cut mass. Density is the oven-dry mass per unit volume.

Individual regressions of temperature, moisture and sample density against flux rate (flux = mg CO₂, g sample^-1, sec^-1). Correlations were weak for all variables in the following order: density (r2 = 0.31, p < 0.01), > moisture (r2 = 0.23, p < 0.01) > Temperature (r2 = 0.029, p < 0.02). Combining temperature and moisture as independent variables vs. flux in a multiple regression improved the fit slightly (r2 = 0.26, p < 0.01). Including density in the model further improved the multiple r2 but only slightly above the fit of density alone (r2 = 0.36, p < 0.01.

A three-factor nested model ANCOVA was performed with genera, decay class and diameter as categorical factors; temperature and moisture were included as continuous variables. Flux among genera was found to be significant, but only Betula and Quercus varied significantly in pair-wise comparisons (p = 0.0247, see Table 1). No significant pattern was found for flux among decay class. A single factor ANOVA of fluxes across genus and decay class showed a significant difference of flux rates between the cut site and the adjacent undisturbed stand (p = 0.027). Including moisture and temperature in the ANCOVA model of stand flux showed no significant effect of stand. Therefore, the microclimate difference between these stands appears responsible for the variation in flux rate. The effect is particularly apparent when scaling fluxes to inventoried CWD in both stands. The total estimated growing season flux rate (genus mean x biomass x 6 mos) of the cut stand was 784 kg/ha/yr almost 294 % greater than the adjacent mature forest (268 kg/ha/yr). The greater volume of CWD in the cut stand (see Hutyra et al., this vol. Table 1) produced a much greater areal flux rate. However, turnover rate of the total stand biomass was higher in the mature stand than the cut site (5.5 % yr-1 vs 4.1 % yr-1 respectively). These differences will likely diminish as the canopy closes and residual stocks of CWD decline during regrowth of the cut site.

Barford, C.C., S.C. Wofsy, M. L. Goulden, J. W. Munger, E. Hammond-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.
Chambers, J. Q., J. P. Schimel and A. D. Nobre. 2001. Respiration from coarse woody litter in central Amazon forests.
Harmon, M.E. J.f. Franklin, F.J. Swanson, P. Sollins, S. V. Gregory, J.D. Lattin, N.H. Anderson, S.P. Cline, N.G. Aumen, J.R. Sedell, G.W. Lienkaemper, K. Cromack Jr., K.W. Cummins. 1986. Ecology of coarse woody debris in temperate ecosystems. Advances in Ecological Research 15:133-302.

Table 1. Mean (s.d.) Flux rates of CWD in adjacent shelterwood cut and mature stand. Units are nanogram CO₂, gm CWD -1, sec -1. (nanogram = mg x 10-3)

Decay Class:

Genus	1	2	3	4
Shelterwood cut
Acer	1.1 (0.6)	1.2 (0.7)	5.8 (n/a)	11 (0.6)
Betula	2.3 (1.5)	3.4 (2.4)	5.1 (4.3)	5.9 (4.3)
Quercus	1.9 (0.9)	1.6 (2.0)	2.5 (3.4)	2.9 (2.4)
Conifer	1.2 (0.6)	2.5 (0.7)	3.3 (2.6)	1.6 (1.4)
Mature Stand
Acer	3.2 (2.9)	4.3 (2.5)	3.9 (2.9)	9.4 (6.8)
Betula	3.2 (4.4)	1.5 (1.3)	5.3 (4.0)	8.3 (5.0)
Quercus	3.5 (2.6)	1.7 (1.6)	3.1 (2.6)	2.9 (1.9)
Conifer	2.5 (2.1)	3.9 (5.9)	1.7 (1.4)	4.8 (2.4)

Vulnerability of Xylem Vessels to Cavitation in Acer saccharum (Marsh.): Scaling from Individual Vessels to Whole Branches

P. J. Melcher, M. A. Zwieniecki and N. M. Holbrook

The relation between xylem vessel age and vulnerability to cavitation of Acer saccharum Marsh. was quantified by measuring the pressure required to force air across bordered pit membranes separating individual xylem vessels. We found that the bordered pit membranes of vessels located in current year xylem could withstand greater applied gas pressures (3.8 MPa) compared to bordered pit membranes in vessels located in older annular rings (2.0 MPa). A longitudinal transect along six-year-old branches indicated that the pressure required to push gas across bordered pit membranes of current year xylem did not vary with distance from the growing tip. To understand the contribution of age-related changes in vulnerability to the overall resistance to cavitation, we combined data on the pressure thresholds of individual xylem vessels with measurements of the relative flow rate through each annual ring. The current year's annual ring contributed only 16 % of the total flow measured on 10-cm-long segments cut from six-year-old branches, but more than 70 % of the total flow when measured through six-year-old branches to the point of leaf attachment. The vulnerability curve calculated using relative flow rates measured on branch segments were similar to vulnerability curves measured on six-year-old branches (P₅₀ = 1.6 to 2.4 MPa), whereas the vulnerability curve calculated using relative flow rates measured on six-year-old branches were similar to ones measured on the current year's extension growth (P50 = 3.8 MPa). These data suggest that, in A saccharum, the current year's xylem can withstand larger xylem tensions than older wood and dominates water delivery to leaves.

The Barre Woods Megaplot Soil Warming Experiment

J. M Melillo, P. A. Steudler, H. Lux, F. P. Bowles, S. Morrisseau and E. Burrows

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 conclusively 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:

1. 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. The Barre Woods site was chosen because of its similarities to the Prospect Hill site - stand age, tree species composition, and soil type. There is no evidence of a plow horizon at the Barre Woods site, though surrounding stone walls indicate past use as pastureland. Historical records indicate the site was destroyed during the 1938 hurricane and then regrew naturally.

During the summer and fall of 2001 we buried 3.4 miles of heating cable in a 30 x 30m plot. Cables were buried at a 10cm depth, spaced 20cm apart. A second 30 x 30m area was delineated to serve as the control plot. Results from the original soil warming experiment confirmed that the soil disturbance associated with the installation of heating cables has had no effect on soil temperatures and only minor and variable impacts on soil moisture.

In April 2002, we began a one-year period of baseline measurements before turning on the heat in the new megaplot. These measurements included nitrogen mineralization, trace gas fluxes (CO₂, CH₄, N₂O), tree growth (dendrometer bands), an under story species survey, canopy foliage analysis (C:N, NIR), and lysimetry. Thermistors were installed in both plots to begin tracking soil temperatures. The baseline measurements have confirmed that there are no significant differences between the two plots for key ecosystem processes prior to the initiation of the warming manipulation (Figs. 1-4). We will turn on the power in April 2003 to begin the "manipulation phase" of this study. From this study, we expect to determine whether or not warming results in the movement of nitrogen from the soil to the trees and to learn how this movement affects the net carbon balance of the ecosystem.

Soil Warming, the First Decade and Beyond

J. M Melillo, P. A. Steudler, H. Lux, F. P. Bowles, S. Morrisseau and E. Burrows

In the first decade of the soil warming experiment in the Prospect Hill site at the Harvard Forest, we have documented changes in soil carbon and nitrogen cycling and here we consider some of the consequences of these changes for the climate system. For the first time we have shown that while soil warming accelerates soil organic matter decay and CO₂ fluxes to the atmosphere, this response is small and short-lived for a mid-latitude forest because of the limited size of the labile soil carbon pool (Fig. 1a, b). We have also shown that warming increases the availability of mineral nitrogen to plants (Fig. 2a, b). Since plant growth in many mid-latitude forests is nitrogen limited, warming has the potential to indirectly stimulate enough carbon storage in plants to at least compensate for the carbon losses from soils.

By the twelfth year of the soil warming study, the effect of warming on soil nitrogen availability has declined dramatically. The once large increase in available nitrogen created in response to warming has almost disappeared by year twelve (Fig. 3a, b). We do not know where the "excess" available nitrogen has gone. We speculate that over time most has become tied up in nitrogen pools with long storage times such as refractory soil organic matter and the woody tissue of the trees.

A Regional Climatology of Cloud and Aerosol for Forest-Atmosphere Exchange

Q. Min

Carbon uptake by forest is enhanced on partly cloudy days due to the enhancement of diffuse component scattered by cloud droplets (Min 2002, and references herein). Furthermore, ecosystems affect both cloud formation through the process of evapotranspiration, and cloud properties, by influencing aerosol and water vapor concentrations. These studies point to the needs for regional climatological measurements of clouds and aerosol, and for establishing the link between the microphysics of cloud and aerosol, through the PAR irradiance, for CO₂, water vapor and sensible heat exchange. To investigate climatology of cloud for forest-atmospheric exchange, we developed a fast retrieval algorithm to derive cloud optical depth from broadband global shortwave measurements that have been taken for past ten years at the Harvard Forest site. First we compared cloud optical depths derived from this retrieval algorithm against well-validated results at the ARM SGP site where various passive and active instruments for detecting clouds are operated (Min et al., 2003). As shown in Figure 1, this simple retrieval with assumption of standard mid-latitude summer atmospheric profiles agrees within 10% of accurate narrowband retrieval. Various sensitivity studies indicate that the overall uncertainty of this simple retrieval is about 15%.

We applied this retrieval to the hourly averaged shortwave measurements at Harvard Forest from 1997 to 2000. For four years, data sets are more continuous and have better quality. Figure 2 shows the histogram of cloud optical depth over Harvard Forest. The averaged optical depths of