2010 NORTHEASTERN NATURALIST 17(3):437–454 Soil Respiration and Ecosystem Carbon Stocks in New England Forests with Varying Soil Drainage Aletta A. Davis1,2, Jana E. Compton1,3,*, and Mark H. Stolt1 Abstract - Northern temperate forests play an important role in the global carbon (C) cycle. Individual stands can differ in C content and storage, based on characteristics such as vegetation type, site history, and soil properties. These site differences may cause stands to vary in their response to extreme weather events such as droughts. We examined ecosystem C pools, soil respiration, and litterfall in four hardwood stands with widely varying soil drainage in Rhode Island. Total ecosystem C increased as soils became more poorly drained, ranging from 181 Mg C ha-1 in the excessively drained Entisol to 547 Mg C ha-1 in the very poorly drained Histosol. The proportion of ecosystem C contained in the soil was much higher in the poorly drained soils, and ranged from 57% in the excessively drained Entisol to 91% in the poorly drained Histosol. While total ecosystem C stocks varied by a factor of three, rates of litterfall and soil respiration were similar among sites. Soil carbon content was highest in the very poorly drained site, and respiration was lowest from this site. During the summer drought of 1999, all soils except the Histosol had lower respiration rates than predicted from temperature alone. Rain events that ended the drought produced a pulse of soil respiration in all mineral soils, stimulating soil C flux more than expected from temperature alone. The effect of drought and rewetting on soil respiration varied by site, suggesting that the response to climate variability will depend upon soil drainage to some extent. Soil respiration rates were most variable in dry conditions, and current and antecedent soil moisture conditions played an important role during those times. In general, soil respiration was much more variable over time than across sites, even among these sites with very different total soil C content, indicating that climate—mainly temperature—is the main determinant of soil CO2 release even across soils with widely varying drainage. Introduction Northern temperate forests are a net sink for atmospheric carbon dioxide (CO2), largely as a consequence of forest regrowth on abandoned farmland, fire suppression, and possible fertilization by increasing atmospheric CO2 and nitrogen (N) deposition (Goodale et al. 2002, Magnani et al. 2007, Pacala et al. 2001). From 1900 to 2000, forest cover in Rhode Island increased from 20% to 70% of the landscape (Hooker and Compton 2003). While large-scale modeling indicates that regrowing forests are net carbon (C) sinks when aggregated over large geographic areas, within these regions 1Department of Natural Resources Science, University of Rhode Island, Kingston, RI. 2Current address - Department of Forestry and Environmental Sciences, North Carolina State University, Campus Box 8002, Raleigh, NC 27695. 3Current address - US Environmental Protection Agency, National Health and Environmental Effects Research Laboratory, Western Ecology Division, 200 SW 35th Street, Corvallis, OR 97333. *Corresponding author - compton.jana@epa.gov. 438 Northeastern Naturalist Vol. 17, No. 3 variation exists in ecosystem types, soil drainage (Davidson and Lefebvre 1993), rates of N deposition (Ollinger et al. 1993), and current and historical management land-use history (Foster et al. 1998). All of these factors can influence ecosystem C storage. Soil drainage is a potentially important site factor influencing ecosystem C balances under a changing climate. Soil respiration and net ecosystem C exchange are strongly related to temperature (Lloyd and Taylor 1994), but soil moisture can influence this relationship (Davidson et al. 1998, Nikolova et al. 2009, Xu and Qi 2001). Wetter sites may have lower soil respiration on an annual basis, but during dry periods, poorly drained soils could serve as substantial sources of available C to be respired and added to the atmospheric pool (Savage and Davidson 2001). Soil moisture can vary dramatically among forests spatially, and within a forest over seasonal and inter-annual time scales. Our objective was to determine how temporal and spatial variation in soil water availability influences forest ecosystem C stocks and soil respiration in southern New England. We examined sites presented as part of the soil C inventory conducted by Davis et al. (2004). The current paper presents ecosystem C pools including vegetation (above and belowground biomass and soil C to 1 m depth), plus litterfall and soil respiration of four hardwood stands covering the extremes of soil drainage from excessively drained to very poorly drained. The occurrence of a severe drought during the summer of 1999 also provided the opportunity to examine the differential response of this range of soil drainage classes to prolonged drying. We hypothesized that respiration would increase dramatically in the poorly drained soils during drought, since they would dry out and decomposition would no longer be limited by low-oxygen conditions (Scanlon and Moore 2000). We also hypothesized that well-drained soils would have little change or even a decrease in soil respiration during drought as a result of increased water limitation of microbial activity. Methods Site description This study was conducted in the Pawcatuck Watershed in Rhode Island. We further examined a subset of sites where we previously measured soil C (Davis et al. 2004). The climate is temperate, with a mean annual temperature of 9.1 °C, mean monthly maximum of 21.1 °C, and mean monthly minimum of -3.7 °C. Annual precipitation averages 1270 mm, distributed relatively evenly throughout the year (data from Kingston NOAA weather station site; Carl Sawyer, NOAA, Kingston, RI, pers. comm.). We chose four soil types for study, which represent a range of drainage classes, parent materials, and textures (Rector 1981): the loamy sand phase of the Windsor series (Entisol) formed in glacial outwash (excessively drained; ED), the silt loam phase of the Enfield series (Inceptisol) formed in loess-over-outwash (well drained; WD), silt loam phase of the Raypol series (Inceptisol) formed 2010 A.A. Davis, J.E. Compton, and M.H. Stolt 439 in loess-over-outwash (poorly drained; PD), and the Carlisle series (Histosol) that forms in organic deposits (very poorly drained; VPD). Forests ranged in age from 64–105 years and were dominated by hardwood trees (Table 1). Mixed Quercus (oak) species and Acer rubrum L. (Red Maple) dominated the ED and WD sites, with Red Maple increasing in dominance in the PD site. The presence of a plow layer (Ap horizon) indicated that all three mineral soil series were historically cultivated (ED Entisol, WD Inceptisol, and PD Inceptisol). The VPD Carlisle site is classified as a Red Maple swamp and was probably not cleared for agriculture. There is evidence that swamps in this region remained as isolated woodlands within an agricultural landscape (Foster et al. 1992). These swamps were not completely undisturbed, and during the last century may have been selectively harvested in winter months when the frozen peat made for easier timber removal (F. Golet, University of Rhode Island, Kingston, RI, pers. comm.). Soil sampling and analysis Soil C for a range of sites was determined by Davis et al. (2004); the sites examined in this paper are a subset of that larger sampling effort. We build upon the Davis et al. (2004) soil work and determined above- and belowground biomass, litterfall, and soil respiration for this subset of sites. One map unit per soil series was chosen for estimating total ecosystem C storage (Table 1). Both plot (20-m x 20-m) and random transect (100-m length) sampling approaches were used in this study. Plot-scale measurements provide detailed information on ecosystem components and are a manageable scale at which to study ecosystem processes. To better estimate the variability of the soil C pool across larger stands, random transects were used to distribute the soil sampling locations across the map unit. Soil morphology and classification within the plots were representative of transects. Two 100-m transects were identified within each map unit, and sampled at 25-m intervals along each transect. Soil samples from the three soil series developed in mineral soil (Windsor, Enfield, and Raypol) were collected by horizon to a depth of 1 m using a 47-mm diameter split-core sampler and slide hammer. A total of 10 cores were collected from each site. The O horizon was collected intact from a 15-cm x 15-cm area and later separated into Oi, Oe, and Oa materials. In the organic soil (Carlisle), fibric and hemic horizons were collected using the 15- x 15-cm area, while sapric horizons were collected by horizon using a Macauley peat sampler. Bulk density calculations were corrected for >2-mm coarse fragment mass and volume. Loss-on-ignition was determined for all soils by combustion at 550 °C for 5 hours (Nelson and Sommers 1996). Samples were analyzed for total C concentration using an automated combustion NA1500 Carlo-Erba CN analyzer (Carlo-Erba Instruments, Milan, Italy). The C concentration was multiplied by the <2-mm soil mass for each horizon to obtain an areal estimate of total soil C. 440 Northeastern Naturalist Vol. 17, No. 3 Plot-level biomass and litterfall In order to estimate stand establishment age, the largest tree (based on diameter at 1.4 m) within 10–15 m of each sampling point along the 100-m transects was cored at the base, and we mounted the cores and counted tree rings using a dissecting microscope. Stand age was estimated from the mean age of the four oldest trees cored at each site (Table 1). Basal area at each sampling point along the 100-m transect was estimated by prism. Within each mapping unit, one 20-m x 20-m plot was established for the detailed study of ecosystem C pools and fluxes. Plots were located in an area considered to be representative of the mapping unit. Soil auger borings were made within and around the edges of each plot to confirm that soil morphology was representative of the entire unit; if a mapping unit was very heterogeneous, a new mapping unit was selected. Basal area estimates from transect sampling were compared to the basal area within each plot to identify areas where the vegetation present within the plot was similar in basal area to the transects (Table 1). Table 1. Soil and vegetation characteristics by site. Stand age, transect basal area, and plant composition are from the 100-m transect sampling. Plot basal area and trees per hectare are from the 20-m x 20-m plots. Site (by soil type) Characteristic Windsor Enfield Raypol Carlisle Soil order Entisol Inceptisol Inceptisol Histosol Great group Psamment Dystrudept Endoaquept Medsaprist Drainage class Excessively Well drained Poorly drained Very poorly drained drained Stand age (yr)A 64 81 105 105 Transect basal area 23.0 26.9 25.7 18.6 (m2 ha-1) Plot basal area 20.5 28.2 32.4 18.9 (m2 ha-1) Trees (ha-1)B 600 800 550 400 Composition Overstory (O)C Q. r., Q. a., A. r., P. r., Q. r., A. r., Q. a. A. r., C. t. Q. c., Q. v. Q. c., Q. a N. s. Understory (U)D G. b., V. p. Lycopodium sp. S. r., T. r., V. d., C. a., S. f. Lycopodium sp. AStand age was determined by taking the mean age of the four oldest trees cored at each site. BLive trees, >5 cm diameter at 1.4 m (DBH). CQ. r. = Quercus rubra L. (Northern Red Oak), Q. a. = Q. alba L. (White Oak), Q. c. = Q. coccinea Muenchh. (Scarlet Oak), Q. v. = Q. velutina Lam. (Black Oak), A. r. = Acer rubrum L. (Red Maple), P. r. = Pinus rigida P. Mill (Pitch Pine), N. s. = Nyssa Sylvatica L. (Blackgum), C. t. = Chamaecyparis thyoides (L.) B.S.P. (Atlantic White Cedar). DG. b. = Gaylussacia bacata (Wangenh.) K. Koch. (Black Huckleberry), V. p. = Vaccinium pallidum Aiton (Lowbush Blueberry), S. r. = Smilax rotundifolia L. (Roundleaf Greenbrier), T. r. = Toxicodendron radicans (L.) Kuntze (Poison Ivy), V. d. = Viburnum dentatum L. (Southern Arrowwood), C. a. = Clethra alnifolia L. (Sweet Pepperbush), and S. f. = Symplocarpus foetidus (L.) Salisb. ex Nutt. (Skunk Cabbage). 2010 A.A. Davis, J.E. Compton, and M.H. Stolt 441 Tree biomass was estimated using published allometric equations based on diameter at 1.4 m (Tritton and Hornbeck 1982). Carbon storage in aboveand belowground overstory biomass was calculated based on average C concentrations in hardwoods and softwoods of the Northeast (0.521 and 0.498 g C g-1, respectively; Birdsey 1992). Regional estimates of the proportion of root biomass to total aboveground biomass for hardwoods (0.1834) and softwoods (0.2048) were used to estimate root biomass (Birdsey 1992). Understory aboveground biomass was determined by destructively sampling three 2-m x 2-m areas during September and early October 1999 before vegetation began to senesce. Understory biomass was collected from three randomly selected locations outside the 20-m x 20-m plots, no more than 10 m away from any edge of the larger study plots in order to minimize disturbance of the respiration plots. Although we did not measure coarse woody debris in the present study, Hooker and Compton (2003) found that this pool contained <3% of total ecosystem C in Pinus strobus L. (White Pine) stands on well-drained soils in Rhode Island. Litterfall was measured within each plot at regular intervals for one full year (June 1999–May 2000). Three plastic litter baskets (38 cm x 52.5 cm) were located randomly in each plot to capture litterfall. All litter was collected biweekly during the fall and monthly throughout the rest of the year. The litter was dried at 105 °C to constant weight. Carbon in the understory and litter were estimated assuming a C concentration of 0.50 g C g-1 (Houghton 1996). Soil respiration Point-in-time CO2 fluxes from the soil surface were measured from June 1999 to May 2000 with a Li-Cor 6262 infrared gas analyzer (Li-Cor, Lincoln, NE) using the dynamic, closed-chamber method (Rolston 1986) as described by Davidson et al. (1998). Beveled-edge 25-cm diameter PVC rings were installed at approximately 2–3 cm into the forest floor to form a seal at the soil surface with minimal impact on fine roots (Raich and Nadelhoffer 1989). Rings were permanently placed at four randomly selected points within each 20-m x 20-m plot. Ambient CO2 was measured for 60 seconds by placing the chamber on its side near the ring. After 60 seconds, the PVC chamber was fitted snugly over the ring for 5 minutes. Carbon dioxide concentrations within the chamber were recorded at 3-second intervals. Chamber concentrations were plotted as a function of time, and a linear regression between the chamber concentrations and time was fitted to a 60-second portion of the data to calculate the rate of CO2 flux from the soil. The slope of the regression (CO2 rate) was converted to gaseous C losses on a g C m-2 d-1 basis. The airspace volume was determined using five measurements of ring height. Biweekly flux measurements began in June 1999 and ended in November 1999, except for the Carlisle site, where measurements began in late July. Soil respiration was measured monthly from May 1999 to May 2000. Soil temperature beside each chamber was determined (10-cm soil depth) from two instantaneous measurements each time soil respiration was 442 Northeastern Naturalist Vol. 17, No. 3 measured. In addition, hourly temperatures were measured throughout the entire year for one random location within the plot, using a temperature logger placed in the forest floor and the mineral soil at 10 cm (HOBO H8, Onset Corporation, Woods Hole, MA). When temperature data were missing due to equipment problems (<5% of the time), we used a linear regression between that site’s temperature and the nearest site’s temperature for the 10 days prior to and following the data gap to fill the gap. At each time we measured respiration, we also collected randomly located soil samples (n = 3) for soil moisture (using a 6-cm diameter steel sampler for organic soils and a 4.5- cm diameter steel corer for the mineral soils). These samples were collected away from the rings to avoid disturbance, but should represent the relative soil moisture conditions during the period respiration was measured. Soil moisture was determined by drying for 24 hours at 105 °C. Average bulk density for these soil depths was determined from the transect data by series, and then soil moisture (g H2O g-1 soil) was converted to volumetric water content (cm3 cm-3). Daily precipitation data were obtained from a NOAA weather station in Kingston, RI (C. Sawyer, University of Rhode Island, Kingston, RI, pers. comm.). Total annual CO2-C flux (Mg C ha-1 yr-1) from the soil surface was estimated by developing CO2 vs. temperature regressions for each ring in each of the four sites. This approach allowed us to calculate standard errors and make comparisons across sites. Temperature data used in these regressions were the average of the two probes inserted beside each chamber when flux was measured. Daily average 10-cm soil temperature for each sampling date was calculated from the hourly measurements recorded by the data logger installed at each site. Soil temperatures (10 cm) from the probes were within 1–2 °C of the hourly temperature measured by the temperature datalogger at each site and strongly positively correlated. Once the CO2-temperature regression equations for each ring were calculated, daily average soil temperature for the site was used to extrapolate annual CO2 emissions from each ring. Mean comparisons for the point-in-time values were conducted using repeated measures analysis of variance with SPSS (SPSS Science, Chicago, IL). Nonlinear regressions were also conducted with SPSS. Results Carbon in vegetation, soils, and litterfall Total aboveground biomass C increased with decreasing drainage in the mineral soil types from 66 Mg C ha-1 in the ED Windsor site to 120 Mg C ha-1 in the PD Raypol site (Table 2). The VPD Carlisle site had the lowest biomass C of all sites. Understory biomass was a small proportion of total ecosystem C (<1%; Table 2), and increased with drainage. The VPD Carlisle site had the greatest understory C, and also highly variable accumulation of C in the understory (CV = 125%), probably a result of the pronounced hummocky or pit-and-mound nature of the soil surface. 2010 A.A. Davis, J.E. Compton, and M.H. Stolt 443 Soil C was the largest pool of C stored within all four ecosystems. The soil C pools (including O horizons) in the three mineral soils accounted for 57–62% of the total C in these forests, while in the VPD Carlisle site the soil accounted for 91% of total ecosystem C (Table 2). Root biomass accounted for only 6% of total ecosystem C in the mineral soils and 1% in the organic soil. Soil C content was lowest in the ED Windsor (64 Mg C ha-1) site. The VPD Carlisle site had over seven times more soil C than the ED and WD sites, and more than two times more C than even the PD Raypol site (190 Mg C ha-1). Aboveground litterfall was similar across these sites (2.1–2.9 Mg C ha-1 yr-1), with lowest rates in the VPD site (Table 2). The ratio of litterfall to total aboveground biomass varied from 2–5%. Litter inputs from mid-September through the beginning of November accounted for the majority of total annual aboveground litter inputs to the soil (68–85%, data not shown). Soil respiration There was substantial seasonal variation in soil respiration, with the highest rates of respiration occurring from April to November (Fig. 1). Soil respiration increased by a factor of 2–4 from June through August. The highest point-in-time respiration was measured at the end of July in the Histosol and in mid-August after rains in the Entisol and Inceptisols (Fig. 1). Differences across sites were much more pronounced during these peak times of soil respiration. Little variability among sites was observed during the winter. Total annual soil respiration in these sites, based on the average of the annual sum of daily individual ring estimates, ranged from 5.95–9.54 Mg C ha-1 yr-1, with the lowest fluxes occurring from the VPD organic soil (Table 2). Table 2. Carbon pools, distribution, and fluxes. Values are means, ± standard errors. Soil type and drainage class Windsor ED Enfield WD Raypol PD Carlisle VPD Carbon pools (Mg C ha-1) Plant Overstory aboveground 65.0 94.5 118.5 37.8 Understory aboveground 1.0 ± 0.3 0.3 ± 0.1 1.7 ± 0.1 4.8 ± 3.5 Roots 11.6 18.1 21.7 7.2 Soil Oi plus Oe horizon 40.3 ± 4.3 40.7 ± 4.3 37.7 ± 5.7 16.1 ± 6.1 Oa plus 0–100 cm depth 63.5 ± 3.3 128 ± 14.1 190 ± 13.2 486 ± 45.4 Total ecosystem C content 181 281 370 552 Carbon distribution (% of total) Plant biomass (total) 43% 40% 38% 8% Forest floor (Oi + Oe horizon) 22% 14% 10% 3% Total soil (O horizon + 0–100 cm soil) 57% 60% 62% 92% Carbon fluxes (Mg C ha-1 yr-1) Aboveground litterfall 2.78 ± 0.30 2.87 ± 0.08 2.32 ± 2.20 2.12 ± 0.37 Soil respiration 9.08 ± 1.12 7.13 ± 0.73 9.54 ± 1.17 5.95 ± 0.92 444 Northeastern Naturalist Vol. 17, No. 3 We found that soil type had a marginally significant influence on soil respiration (P = 0.088; Table 3). Sampling date and the interaction between sampling date and soil type were highly significant factors influencing soil respiration (P < 0.001), suggesting that moisture and temperature changes over time were important and affected the sites in different ways. The ED Windsor soil had slightly higher CO2 fluxes than the other soils throughout the growing season, except for the early growing season and the peak of the drought in late July, when the PD Raypol soil had the highest rates of soil respiration. The highest total fluxes were observed in the PD Raypol site. This finding could be most likely due to higher fluxes in the early growing season and high rates during the dry, warm period when this site still had abundant soil moisture (Fig. 2). The lowest respiration rates were observed in the VPD Carlisle soil (Table 2). Possible indications of drought effects can be observed by plotting measured CO2 fluxes from these soils against predicted values based on daily average soil temperature (Figs. 2, 3). Several time-periods stand out and may indicate drought effects. Substantial drought occurred during June and July 1999 (Fig. 4). At the end of July 1999, soil respiration was much lower than Figure 1. Soil respiration measured from the four study sites from May 1999 until May 2000. Bars represent standard error (n = 4 rings per site). Table 3. Repeated measures ANOVA effects on soil respiration. Source DF MSE F P value Soil 3 16.611 2.702 0.088 Ring (soil) 12 7.419 5.745 <0.001 Date 23 32.091 24,849.000 <0.001 Date*Soil 32 2.715 2.102 0.001 2010 A.A. Davis, J.E. Compton, and M.H. Stolt 445 fluxes predicted solely from soil temperature in the Windsor, Enfield, and Raypol soils (Fig. 2). Until this time, moisture does not appear to limit soil respiration in these soils. In the second week of August it rained frequently, and soil respiration measured just after this period was much higher than predicted in the ED Windsor soil (Fig. 2) and indicate that CO2 fluxes were moisture-limited in this soil near the end of summer. While total ecosystem C differed by a factor of three (181 vs. 552 Mg C ha-1) among these forests, soil respiration varied only by a factor of 1.5 (range = 6.0–9.1 Mg C ha-1 yr-1) and litterfall was uniform across sites (range = 2.1–2.9 Mg C ha-1 yr-1). The similarity in soil respiration across sites may have been the result of a severe drought during the summer of 1999. Low soil moisture appears to have negatively affected soil respiration Figure 2. Soil respiration (predicted and measured), temperature, soil moisture (0–10 cm mineral soil), and precipitation for the four sites. Predicted respiration was determined using the site-specific exponential relationship shown in Figure 3. Precipitation data were obtained from the NOAA weather station in Kingston, RI. 446 Northeastern Naturalist Vol. 17, No. 3 in the ED and WD soils for at least a portion of the drought. Soil respiration declined significantly with increasing soil moisture for all sites (Table 4). We conducted linear regressions using both soil moisture and temperature, but only temperature was a significant predictor in these relationships. The relationship between temperature and moisture is complex and generally inverse over the course of the year (Table 4). Soil moisture is often highest during the coldest periods, but can covary with temperature during certain periods. Here, we examined temperature-moisture interactions by plotting the residuals of the regression between respiration and temperature against soil moisture (Fig. 5). Using this approach, we observe greater deviation from the relationship at low soil moisture. This result suggests that soil Figure 3. Relationship between soil temperature (10 cm) and CO2 efflux. Equations shown are derived from the average of the relationships used for the calculation of annual fluxes. Table 4. Linear regressions of the influence of volumetric water content on CO2 fluxes. ns = not significant (P > 0.05). ED = extremely well drained, WD = well drained, PD = poorly drained, and VPD = very poorly drained. Soil Depth Slope Intercept r2 P value Windsor (ED) Forest floor -24.90 6.43 0.35 0.013 0–10 cm -21.19 6.32 0.18 ns Enfield (WD) Forest floor -6.53 3.96 0.40 0.007 0–10 cm -9.28 4.83 0.43 0.004 Raypol (PD) Forest floor -5.31 4.82 0.27 0.038 0–10 cm -6.37 5.51 0.34 0.017 Carlisle (VPD) Forest floor -23.49 8.72 0.41 ns 0–10 cm -11.87 7.14 0.21 ns 2010 A.A. Davis, J.E. Compton, and M.H. Stolt 447 temperature effects are less predictable during drought periods, and that moisture availability influences respiration during those periods. Discussion Ecosystem C totals and distribution Our study only examines one site per drainage class, thus making it difficult to draw inferences about the specific drivers of differences between Figure 4. Daily precipitation and departure from a 30-year average from June 1999 through May 2000. Data obtained from the NOAA weather station in Kingston, RI. Asterisks indicate periods of moderate or greater drought according to the Palmer Drought Severity Index (National Climatic Data Center 1999–2000). 448 Northeastern Naturalist Vol. 17, No. 3 sites. We report these data and place them in the context of other studies to assist in the interpretation. Plant biomass varied among sites, with the maximum biomass occurring in the intermediate drainage classes. Our Figure 5. Soil moisture vs. the residuals of soil respiration from the soil temperature model for each site (nonlinear regression equations shown in Table 5). Table 5. Regression equations (CO2 vs. 10-cm soil temperature) for individual rings of the four sites. Soil Ring Equation r2 Windsor 1 y = 0.2040e0.1589x 0.84 2 y = 0.2744e0.1726x 0.86 3 y = 0.3378e0.1611x 0.85 4 y = 0.3698e0.1348x 0.72 Enfield 1 y = 0.3538e0.1119x 0.83 2 y = 0.337e0.1438x 0.86 3 y = 0.4720e0.1056x 0.81 4 y = 0.5902e0.1084x 0.80 Raypol 1 y = 0.1607e0.1929x 0.83 2 y = 0.3117e0.1585x 0.78 3 y = 0.6417e0.1280x 0.86 4 y = 0.5137e0.1034x 0.66 Carlisle 1 y = 0.1203e0.2030x 0.88 2 y = 0.1930e0.1407x 0.85 3 y = 0.1704e0.1895x 0.82 4 y = 0.1202e0.1767x 0.82 2010 A.A. Davis, J.E. Compton, and M.H. Stolt 449 estimates for overstory aboveground biomass C (38 to 118.5 Mg C ha-1) are similar to those found in other studies. Botkin et al. (1993) estimated that the aboveground biomass in deciduous forests of New England contains 40.2 Mg C ha-1 (89.4 Mg biomass ha-1 times 45% C from their study), which is lower than our mineral soil sites but similar to the VPD site. Trettin et al. (1999) found a similar range in biomass C (82.2–117.6 Mg C ha-1) for oak-, hickory-, and maple-dominated forests in Tennessee. The site with VPD Carlisle soil contained much less aboveground biomass C than the sites with mineral soils (Table 2), even though the Carlisle stand was older. Soil moisture status may limit C storage in aboveground biomass by limiting growth at the extremes of drainage classification (ED and VPD soils). Soil C stocks varied nearly five-fold with drainage class, reinforcing the idea that soil drainage is a very important variable for incorporation into spatial assessments of soil C (Tan et al. 2004). Soil C content was lowest in the ED Windsor (104 Mg C ha-1) site. The VPD Carlisle site held 502 Mg soil C ha-1, which is over seven times greater soil C than the ED site, and more than twice greater than even the PD Raypol site. The VPD Carlisle site is a peat soil, with bulk density of less than 0.2 g cm-3 and C concentrations were approximately 50% throughout the profile to 100 cm (Davis et al. 2004). Peat soils have very high C content because anoxic conditions lead to very slow decomposition and preservation of organic matter. Rapalee et al. (1998) found that the largest C stocks occurred in poorly drained sites across the boreal regions. The distribution of C throughout the three well-drained forests was slightly different than the average US forest described by Birdsey (1992), who estimated that above- and belowground vegetation accounted for ≈32% of total ecosystem C storage. In our study, above- and belowground vegetation on the three mineral soil sites accounted for a slightly higher proportion of ecosystem C (38–42%), while vegetation in the Carlisle site accounted for only 8% of total ecosystem C storage (Table 2). Estimates of regional C storage often are based on broad categories of ecosystem or soil types. This approach does not take into account the complexity of these systems and the various factors that contribute to gains and losses in C. Approximately 5-fold differences in soil C and 3-fold differences in ecosystem C content were found among our study forests between the excessively drained and very poorly drained sites. The most apparent effect of soil type on ecosystem C storage was seen in the soil C pool, which was greater in the VPD and PD soils. The differences in soil C pools were not strongly related to present-day biomass, respiration, or litter inputs, but are most likely a consequence of two potentially interacting factors: poor organic matter quality, and the effects of poor soil drainage reducing turnover and decomposition of organic matter. Our study cannot separate these factors. However, we can gain some information on the importance of soil moisture by comparing the short-term responses of soil respiration to changing soil moisture status during wet and dry periods by drainage class. 450 Northeastern Naturalist Vol. 17, No. 3 Comparison of respiration and litterfall fluxes across sites Our estimates of annual respiration rates (5.95–9.08 Mg C ha-1 yr-1) are similar to other rates observed in hardwood forests of New England. Davidson et al. (1998) reported an annual respiration rate of 7.2 Mg C ha-1 yr-1 for drought-affected forest soils at Harvard Forest, MA. In a five-year study at Harvard Forest using eddy covariance techniques, Goulden et al. (1996) found a range in soil CO2 effluxes from 8.1–11.4 Mg C ha –1 yr-1. Similar soil respiration rates have been reported for oak forests and cedar swamps in Minnesota (7.9 and 7.4 Mg C ha-1 yr-1, respectively [Reiners 1968], and boreal aspen forests, 8.9 Mg C ha-1 yr-1 [Russell and Voroney 1998]). Despite differences in aboveground biomass across the sites we studied, litterfall inputs were similar across these sites. Litterfall for these sites (2.12–2.78 Mg C ha -1 yr-1) were just below the range of 2.9–5.3 Mg C ha -1 yr-1 for hardwood forests reported by Scott and Binkley (1997). Similar litterfall C fluxes (2.3–2.6 Mg C ha-1 yr-1) have been reported for hardwood forests in the Appalachians (Garten et al. 1999), while lower litterfall (1.5 Mg C ha-1 yr-1) was reported for a well-drained site in central Massachusetts (Gaudinski et al. 2000). While soil and ecosystem carbon differed across sites by a factor of 5, soil respiration varied only approximately 2-fold. We found a marginally significant difference between sites in respiration (P = 0.088), and the very poorly drained site had the lowest rates. Davidson et al. (1998) also found that variability in soil respiration across the landscape was related to soil drainage class, with better drained soils having higher rates. However, we found that among mineral soils, the highest rates were observed in the poorly drained site. This site tended to have consistently the highest rates during the early growing season and high rates during the warmest periods. Differences in respiration between sites were relatively small, indicating that even with substantial differences in soil C stocks, respiration is largely driven by temporal variations in temperature and moisture conditions. Soil respiration: Interactions between soil temperature and moisture Soil temperature was the main factor influencing soil respiration, as found by others (Kicklighter et al. 1994), but soil moisture also influenced soil respiration in our study (Fig. 2, Table 4). Low soil moisture can constrain soil CO2 efflux (Davidson et al. 1998, Linn and Doran 1984, Xu and Qi 2001). Volumetric water content accounted for 18–43% of the variation in CO2 in the soils we studied (Table 4), compared to 72–88% for soil temperature (Table 5). We did not combine respiration and temperature into one model, since the relationship is complex and may require information about the matric potential in upland soils (Savage and Davidson 2001), which was not available for the study sites. By plotting the residuals of the regression between respiration and temperature against soil moisture (Fig. 5), we observed greater deviation from the relationship at low soil moisture. This result indicates that during dry periods, the relationship between soil temperature and respiration is weaker, possibly due to the influence of moisture limitations during dry periods and release of this limitation after rain events. 2010 A.A. Davis, J.E. Compton, and M.H. Stolt 451 In the summer of 1999, New England suffered a moderate to severe drought as reflected by the Palmer Drought Severity Index (http://www. cpc.ncep.noaa.gov/index.html; Fig. 4). Monthly precipitation was far lower than the 30-year average for April, June, and July 1999 (Fig. 4). Drought periods can have substantial effects on soil moisture and in turn respiration. Davidson et al. (1998) reported a significant effect of a 1995 drought on soil respiration at volumetric water contents less than 12%. The ED Windsor and WD Enfield sites had volumetric water contents of 12% or less on a few sampling dates (Fig. 2). The ED Windsor site had low volumetric water contents beginning in June and continuing through the middle of September. The Enfield site experienced low water contents during the last two weeks of June. In the ED and WD soils, soil respiration was lower than predicted from temperature alone during periods of low volumetric water contents, which suggests that soil respiration in the drier soils may have been moisturelimited throughout the drought. During the periods when soil moisture was the lowest (during the end of the 1999 summer drought), the VPD soil had the highest soil respiration rates. Our data suggest that there are key periods during which the soil respiration is strongly regulated by soil moisture. Rainfall events at the end of the drought period stimulated soil moisture flux more than predicted from temperature alone. At the end of October, after an extended period of little rainfall (Fig. 2), soil moisture increased and actual CO2 fluxes in the ED Windsor and PD Raypol sites were also higher than predicted (Fig. 2). In addition, at the beginning of the growing season in May 2000, actual CO2 fluxes were higher than predicted from temperature alone in the Windsor and Enfield series (Fig. 2). Increased fluxes immediately following rainfall events may be the response of roots or microbes to increased soil moisture in the surface horizons of the soil (Borken et al. 2003, 2006; Law et al. 2001). Our study, while limited because of a lack of replication, shows that ecosystem carbon content can vary dramatically across sites, with poorly drained soils having very high carbon contents. Climate variation, however, seems to be the strongest regulator of the release of soil carbon from these mature forests. Soil temperature was the main driver of soil respiration, with soil moisture being a secondary factor, which was negatively related to respiration during most of the year. Soil respiration rates were most variable during droughts, and current and antecedent soil moisture conditions played an important role during those times. Future efforts to understand the exchange of C between soils and the atmosphere should continue to examine the importance of climate variability and soil moisture on the net storage of carbon. Acknowledgments We thank José Amador and Keith Killingbeck for advice in the design and interpretation of this study. Fred Pollnac assisted with the soil-respiration measurements. This research was supported by the Rhode Island Agricultural Experiment Station. 452 Northeastern Naturalist Vol. 17, No. 3 Literature Cited Birdsey, R.A. 1992. Carbon storage and accumulation in United States forest ecosystems. USDA Forest Service, Washington, DC. General Technical Report 59. Borken, W., E.A. Davidson, K. Savage, J. Gaudinski, and S.E. Trumbore. 2003. Drying and wetting effects on carbon dioxide release from organic horizons. Soil Science Society of America Journal 67:1888–1896. Borken, W., E.A. Davidson, K. Savage, E.T. Sundquist, and P. Steudler. 2006. Effect of summer throughfall exclusion, summer drought, and winter snow cover on methane fluxes in a temperate forest soil. Soil Biology and Biochemistry 38:1388–1395. Botkin, D. B., L.G. Simpson, and R.A. Nisbet. 1993. Biomass and carbon storage of the North American deciduous forest. Biogeochemistry 20:1–17. Davidson E.A., and P.A. Lefebvre. 1993. Estimating regional carbon stocks and spatially covarying edaphic factors using soil maps at three scales. Biogeochemistry 22:107–131. Davidson, E.A., E. Belk, and R.D. Boone. 1998. Soil water content and temperature as independent or confounded factors controlling soil respiration in a temperate mixed hardwood forest. Global Change Biology 4:217–227. Davis, A.A., M.H. Stolt, and J. E. Compton. 2004. Spatial distribution of soil carbon in southern New England hardwood forest landscapes. Soil Science Society of America Journal 68:895–903. Foster, D.R., T. Zebryk, P. Schoonmaker, and A. Lezberg. 1992. Post-settlement history of human land-use and vegetation dynamics of a Tsuga canadensis (Hemlock) woodlot in central New England. Journal of Ecology 80:773–786. Foster D.R., G. Motzkin, and B. Slater. 1998. Land-use history as long-term broadscale disturbance: Regional forest dynamics in central New England. Ecosystems 1:96–119. Garten, C.T., Jr., W.M. Post III, P.J. Hanson, and L.W. Cooper. 1999. Forest soil carbon inventories and dynamics along an elevation gradient in the southern Appalachian Mountains. Biogeochemistry 45:115–145. Gaudinski, J.B., S.E. Trumbore, E.A. Davidson, and S. Zheng. 2000. Soil carbon cycling in a temperate forest soil: Radiocarbon-based estimates of residence times, sequestration rates, and partitioning of fluxes. Biogeochemistry 51:33–69. Goodale, C.L., M.J. Apps, R.A. Birdsey, C.B. Field, L.S. Heath, R.A. Houghton, J.C. Jenkins, G. H. Kohlmaier, W. Kurz, S. Liu, G.-J. Nabuurs, S. Nilsson, and A.Z. Shvidenko. 2002. Forest carbon sinks in the northern hemisphere. Ecological Applications 12:891–899. Goulden, M.L., J.W. Munger, S. Fan, B.C. Daube, and S.C. Wofsy. 1996. Exchange of carbon dioxide by a deciduous forest: Response to interannual climate variability. Science 271:1576–1578. Hooker T.D., and J.E. Compton. 2003. Forest ecosystem C and N accumulation during the first century after agricultural abandonment. Ecological Applications 13:299–313. Houghton, R.A. 1996. Converting terrestrial ecosystems from sources to sinks of carbon. Ambio 25:267–272. Kicklighter, D.W., J.M. Melillo, W.T. Peterjohn, E.B. Rastetter, A.D. McGuire, and P.A. Steudler. 1994. Aspects of spatial and temporal aggregation in estimating regional carbon dioxide fluxes from temperate forest soils. Journal of Geophysical Research 99:1303–1315. 2010 A.A. Davis, J.E. Compton, and M.H. Stolt 453 Law, B.E., A.H. Goldstein, P.M. Anthoni, J.A. Panek, M.H. Unsworth, M.R. Bauer, J.M. Fracheboud, and N. Hultman. 2001. CO2 and water vapor exchange by young and old Ponderosa Pine ecosystems during a dry summer. Tree Physiology 21:299–308. Linn, D.M., and J.W. Doran. 1984. Effect of water-filled pore space on carbon dioxide and nitrous oxide production in tilled and non-tilled soils. Soil Science Society of America Journal 48:1267–1271. Lloyd, J., and Taylor J.A., 1994. On the temperature dependence of soil respiration. Functional Ecology 8:315–323. Magnani, F., M. Mencuccini, M. Borghetti, P. Berbigier, F. Berninger, S. Delzon, A. Grelle, P. Hari, P. G. Jarvis, P. Kolari, A. S. Kowalski, H. Lankreijer, B. E. Law, A. Lindroth, D. Loustau, G. Manca, J. B. Moncrieff, M. Rayment, V. Tedeschi, R. Valentini, and J. Grace. 2007. The human footprint in the carbon cycle of temperate and boreal forests. Nature 447:849–851. National Climatic Data Center. 1999–2000. Palmer drought severity indices. National Oceanic and Atmospheric Administration. Rhode Island historical data obtained in December 2009. Available online at http://www.ncdc.noaa.gov/oa/ climate/research/drought/palmer-maps/. Accessed 30 December 2009. Nelson, D.W., and L.E. Sommers. 1996. Total carbon, organic carbon, and organic matter. Pp. 961–1010, In D.L. Sparks, A.L. Page, P.A. Helmke, R.H. Loeppert, P.N. Soltanpour, M.A. Tabatabai, C.T. Johnston, and M.E. Sumner (Eds.). Methods of Soil Analysis. Part 3: Chemical Methods. Soil Science Society of America, Madison, WI. Nikolova, P., S. Raspe, C. Andersen, R. Mainiero, H. Blaschke, R. Matyssek, and K.-H. Häberle. 2009. Effects of the extreme drought in 2003 on soil respiration in a mixed forest. European Journal of Forest Research 128:87–98. Ollinger, S.V., J.D. Aber, G.M. Lovett, S.E. Milham, R.G. Lathrop, and G.M. Ellis. 1993. A spatial model of atmospheric deposition for the northeastern United States. Ecological Applications 3:459–472. Pacala, S.W., G.C. Hurtt, D. Baker, P. Peylin, R.A. Houghton, R.A. Birdsey, L. Heath, E.T. Sundquist, R.F. Stallard, P. Ciais, P. Moorcroft, J.P. Caspersen, E. Shevliakova, B. Moore, G. Kohlmaier, E. Holland, M. Gloor, M.E. Harmon, S.-M. Fan, J.L. Sarmiento, C.L. Goodale, D. Schimel, and C.B. Field. 2001. Consistent landand atmosphere-based US carbon sink estimates. Science 292:2316–2320. Raich, J.W., and K.J. Nadelhoffer. 1989. Belowground carbon allocation in forest ecosystems: Global trends. Ecology 70:1346–1354. Rapalee, G., S.E. Trumbore, E.A. Davidson, J.W. Harden, and H. Veldhuis. 1998. Soil carbon stocks and their rates of accumulation and loss in a boreal forest landscape. Global Biogeochemical Cycles 12:687–701. Rector, D.D. 1981. Soil survey of Rhode Island. United State Department of Agriculture, Soil Conservation Service, Washington, DC. Reiners, W.A. 1968. Carbon dioxide evolution from the floor of three Minnesota forests. Ecology 49:471–483. Rolston, D.E. 1986. Gas flux. Pp. 1103–1120, In A. Klute (Ed.). Methods of Soil Analysis. Part 1. 2nd Edition. Agronomy Monograph 9. American Society of Agronomy and Soil Science Society of America, Madison, WI. Russel, C.A., and R.P. Voroney. 1998. Carbon dioxide efflux from the floor of a boreal aspen forest. 1. Relationship to environmental variables and estimates of C respired. Canadian Journal of Soil Science 78:301–310. 454 Northeastern Naturalist Vol. 17, No. 3 Savage, K.E., and E.A. Davidson. 2001. Interannual variation of soil respiration in two New England forests. Global Biogeochemical Cycles 15:337–351. Scanlon, D., and T. Moore. 2000. Carbon dioxide production from peatland soil profiles: The influence of temperature, oxic/anoxic conditions, and substrate. Soil Science 165:153–160. Scott, N.A., and D. Binkley. 1997. Foliage litter quality and annual net N mineralization: Comparison across North American forest sites. Oecologia 111:151– 159. Tan, Z.X., R. Lal, N.E. Smeck, and F.G. Calhoun. 2004. Relationships between surface soil organic carbon pool and site variables. Geoderma 121:187–195. Trettin, C.C., D.W. Johnson, and D.E. Todd, Jr. 1999. Forest nutrient and carbon pools at Walker Branch watershed: Changes during a 21-year period. Soil Science Society of America Journal 63:1436–1448. Tritton, L.M., and W.J. Hornbeck. 1982. Biomass equations for major tree species of the Northeast. USDA Forest Service. General Technical Report NE-69. Northeastern Forest Experiment Station, Radnor, PA. Xu, M., and Y. Qi. 2001. Soil-surface CO2 efflux and its spatial and temporal variations in a young Ponderosa Pine plantation in northern California. Global Change Biology 7:667–677.