W. S. F. Schuster 34

Changes in tree biomass and carbon content over seven decades (1930 - 2000) in an aggrading deciduous forest

Schuster, W.S.F1,2, K.L. Griffin2, K.J. Brown3, M.H. Turnbull4, D. Whitehead5, and D.T. Tissue6

1 Black Rock Forest Consortium, 129 Continental Road, Cornwall, New York 12518, USA,

845-534-4517, fax 845-534-6975,

2 Lamont-Doherty Earth Observatory of Columbia University, Palisades, New York, USA.

3 Ohio University, Athens, Ohio, USA

3 University of Canterbury, Christchurch, New Zealand

4 Landcare Research, New Zealand

5 Texas Tech University, Lubbock, TX, USA.

Abstract

We sought to quantify forest biomass and net carbon (C) uptake rates over 70 years in an aggrading deciduous forest in southeastern New York, USA. Annual increment in live, aboveground tree biomass (AGB), estimated using species-specific allometric equations, was as high as 6.9 Mg ha-1 y-1 on undisturbed long-term plots in the 15-km2 Black Rock Forest. Between the early 1930s and 2000, average annual AGB increment on these plots after all losses was 2.17 Mg ha-1 y-1, equivalent to an annual sequestration of 1.08 Mg C ha-1 y-1. Forest-wide AGB accumulation, estimated from inventory measurements, averaged 1.25 Mg ha-1 y-1 (0.62 Mg C ha-1 y-1) between 1930 and 1985 despite periodic logging. A subset of 51 plots averaged further AGB accumulation of 1.85 Mg ha-1 y-1 between 1985 and 2000. We found some indication of decreasing carbon sequestration with stand maturation, but not on all plots. AGB increment was most rapid (2.91 Mg ha-1 y-1) between the early 1930s and the early-mid 1960s, and least rapid (1.28 Mg ha-1 y-1) from the early 1960s through the mid 1980s. AGB increment was positively correlated with soil depth and pH, and negatively correlated with slope steepness, initial stand age, and biomass. Red oak (Quercus rubra L.) canopy trees stored C at twice the rate of similar-sized canopy trees of other species.


Introduction

Forests cycle and store most of the earth’s terrestrial biomass and thus play a dominant role in the global carbon (C) cycle (Dixon et al. 1994). Temperate deciduous forests in the northern hemisphere comprise some of the world’s most substantial C sinks (Ciais et al. 1995, Myneni et al. 2001), thereby acting to counter anthropogenic increases of atmospheric CO2 and the associated consequences. However, estimates of average annual C sequestration for North American temperate forests range widely (i.e. 1.7 Pg y-1, Fan et al.1998; 0.04 Pg y-1, Schimel et al. 2000), and some suggest they may be approaching saturation (Field and Fung 1999). While some forests in eastern North America have been sequestering C at substantial rates in recent years (Goulden et al. 1996), other forests in the region have experienced reductions in biomass accumulation and C sequestration (Likens et al. 1994, 1996). Clearly, empirical studies are needed to provide better spatial and temporal resolution of forest C fluxes to enable further elucidation of controlling factors (Scurlock et al. 1999).

Eddy-covariance studies provide an important means of tracking seasonal and annual C flux (e.g. Valentini et al. 2000; Janssens et al. 2001). Estimates of net annual ecosystem production (NEP, equal to net primary production (NPP) minus heterotrophic respiration) generally range upward from 2 Mg C ha-1 y-1 for temperate forests. However, short-term NEP studies in forests are not sufficient to determine net C flux over longer periods and larger regions, termed net biome production (NBP, equal to NEP minus losses to human activities and all other causes over long time periods; IGBP 1998). Furthermore, factors controlling long-term C flux may differ from those controlling annual C flux (Barford et al. 2001). For example, successional age and changes in community composition can interact with other biotic and abiotic factors to regulate long-term forest C storage (Gower et al. 1996, Ryan et al. 1997). Ecosystem theory states that net C uptake will decrease to zero as ecosystems approach full maturity (Odum 1969, 1971). However, Carey et al. (2001) have argued that, while forests older than 100 years have been considered to be insignificant C sinks, net primary production (NPP) continues to increase in some forests over 200 years old, and other studies have also emphasized the C-sink potential of older forests (e.g. Schulze et al. 2000).

Here we examine long-term biomass changes in a North American temperate deciduous forest by analyzing long-term plot and inventory records compiled over a 70-year period of forest maturation. We analyze records from the Black Rock Forest, a 1530-ha oak-dominated forest in a region where forests have been aggrading following recovery from repeated clearcutting and conversion to agricultural use during the 19th and early 20th centuries. We use forest inventory records and repeated measurements from undisturbed plots between 1930 and 2000 in conjunction with species-specific allometric equations to study how aboveground biomass has changed over this period of regrowth. We address two questions: (i) how have forest biomass and net C uptake rates varied over this time period, and (ii) have these been decreasing with increasing stand age? We also analyze temporal patterns in growth rates to determine how factors such as environmental conditions and community composition have impacted long-term forest C flux.

Methods

Site Description

The Black Rock Forest (BRF) is a 1530-ha oak-dominated forest in southeastern New York State, located within the Highlands Physiographic Province, an approximately 8000 km2 forested uplands underlain by the Reading Prong geologic formation (Fig. 1; Braun 1967). Average annual precipitation is 1.2 m and air temperature is strongly seasonal, with monthly averages ranging from –2.7 °C in January to 23.4 °C in July (Ross 1958, Turnbull et al. 2001). The topography is rocky with steep slopes and elevations ranging from 110 to 450 m above sea level. The soils are predominantly brown forest soils, with bedrock or glacial till parent material at depths ranging from 0.25 to 1 m (Olsson 1981). Soil reaction is acidic, availability of nutrients is low, and site index ranges from poor to average (Table1; Lorimer 1981).

European settlement in the area began around 1700 and the land was repeatedly logged, with a small proportion of the forest completely cleared for conversion to agriculture and livestock pasture and then abandoned before 1900 (Raup 1938). Frequent clearcuts and fires resulted in a preponderance of hardwood sprout regeneration (Tryon 1939). BRF became established as a research forest in 1928 and was a unit of the Harvard University Forest system from 1949-1989. Since that time, BRF has been operated as a field station and nature preserve by the Black Rock Forest Consortium, a group of academic institutions from the surrounding region (Tryon 1930, Mahar 2000).

Inventory Methods and Experimental Plots

In 1930, an inventory of trees of “cordwood size”, roughly greater than 0.1 m diameter at breast height (dbh), was made of the whole forest, which at the time comprised an area of 1260 ha (Tryon 1930). The forest was subdivided into 150 stands based on species composition and average tree density. Stand area, stand age (based on historical records and tree ring counts), density by species, and average wood volume (1 cord = roughly 2.26 m3) were determined for each stand.

Between 1931 and 1936 a series of long-term plots ranging from 0.04-ha to 0.1-ha were established in which forest growth was monitored (Tryon 1939). Eight of these plots have remained undisturbed and all trees greater than 25.4 mm dbh on these plots have been measured approximately every five years for dbh and crown class (dominant: trees receiving full light from above and partial sidelight; codominant: trees receiving full light from above only; intermediate: trees receiving only partial toplight; suppressed: trees receiving no direct light). Since 1994, annual measurements of dbh and crown class have been performed. The height of all trees was measured in 1998. Most measurements were made in July or later months, when the majority of diameter growth had already occurred (Karnig and Stout 1969), and therefore represent conditions at the end of the growing season for that year. In four of the years, measurements were made in April through early May, and therefore represent conditions at the end of the previous year’s growing season.

The eight remaining long-term plots are located at intermediate forest elevations and exhibit typical forest characteristics (plots 1 – 8 in Fig. 1). Table 1 lists stand age (determined from increment cores; Lorimer 1981, D’Arrigo et al. 2002), average height of the canopy trees, slope and soil characteristics, and approximate site index (calculated from the heights of dominant and codominant oaks in 1998; Schnur 1937) for these plots. Based on these data, half of the plots are on sites of “good” quality, with the other half ranging from “fair” to “poor”, similar to the range in site quality across the forest (Lorimer 1981). The plots were originally established in pairs, one within an area that was experimentally thinned and the other in a nearby area left undisturbed as a control. Thinning operations removed dead, diseased, and dying trees, as well as those species considered to be less economically desirable (e.g., gray birch (Betula papyrifera Marsh.), bigtooth aspen (Populus grandidentata Michx.). However, among the thinned plots (i.e., plots 2, 4, 6, and 8) only plots 2 and 6 were significantly reduced in density and biomass compared to their paired control plots.

A second forest-wide inventory was completed in the summer of 1985 (Friday and Friday 1985), by which time the forest had increased in size to 1416 ha. A total of 56 experimental cuts and thinnings were accomplished between 1930 and 1985, impacting approximately one-third of the forest (e.g. Tryon and Finn 1949, Harrington and Karnig 1975). These areas were included in this inventory along with 10 ha of conifer trees that were planted in the forest in the 1930s and 1940s. The inventory subdivided the forest into 71 stands based on species composition, canopy height, and canopy cover. Three sample points (or occasionally more) were located at regular intervals along the long axis of each stand, producing a total of 218 plots (star symbols in Fig. 1). All trees greater than 50.8 mm dbh on these plots were tallied using a 10-factor basal area prism, totaling 2078 trees of 37 species. Data recorded for each tree included species, dbh, and crown class. In 2000 (or in a few cases 1999 or 2001) a total of 51 of the plots were remeasured for various purposes (circled stars in Fig. 1). Measurement methods were the same as in the 1985 inventory, enabling quantification of tree growth, mortality and regeneration between 1985 and 2000 for these plots.

Biomass estimation

We used previously derived regression equations to estimate live aboveground tree biomass (AGB) from tree dbh measurements. For most trees, we used species-specific equations developed at sites that were nearby and/or edaphically similar, and from studies that had a similar dbh range and a relatively large number of samples. For red oak (Quercus rubra L.), sugar maple (Acer saccharum Marsh.), red maple (A. rubrum L.), yellow birch (Betula alleghaniensis Britt. (B. lutea Michx.f.)), bigtooth aspen, quaking aspen (Populus tremuloides Michx.), and beech (Fagus grandifolia Ehrh.), we used equations from Monteith (1979). For white oak (Quercus alba L.) and scarlet oak (Q. coccinea Muenchh.) we used the equations of Whittaker and Woodwell (1968). For chestnut oak (Quercus prinus L.), black birch (Betula lenta L), black cherry (Prunus serotina Ehrh.), hickories (Carya spp.), white ash (Fraxinus americana L.), yellow poplar (Liriodendron tulipifera L.), eastern hemlock (Tsuga canadensis (L.) Carr.), and basswood (Tilia americana L.), we used the equations of Brenneman et al. (1978). For the remaining, uncommon species we used the general equations of Monteith (1979) for either hardwoods or softwoods. Biomass in tree stumps and roots, standing dead trees and woody debris, understory vegetation, forest floor and soil are not included in these formulae.

Forest-wide biomass analyses

We estimated AGB for the 150 stands inventoried in 1930 by using diameter measurements and volume estimates made in the 1930s on 12 control plots (numbers 1, 3, 5, 7, and others indicated by squares in Fig. 1). For each tree on each plot, we estimated aboveground biomass using the allometric equations and then estimated total aboveground biomass density (AGB) by summing for all trees on each plot and dividing by plot area. For plots established between 1931 and 1936, we estimated AGB in 1930 by subtracting the average annual AGB increment for that plot between the first two measurements multiplied by the number of years after 1930. We then regressed the estimated 1930 AGB for each of these plots on the average wood volume in “cords” (equivalent to 2.27 m3) per hectare for the stand in which each plot was located. The resulting regression relationship, Ba = 2.173 * V, where Ba is aboveground biomass in Mg ha-1 and V is wood volume in cords per hectare had an r2 of 0.61 (p< 0.002). AGB in 1930 was calculated for each of the 150 stands using this regression equation, and forest-wide AGB was estimated as the area-weighted average of these values. 1985 forest-wide AGB was estimated as the area-weighted average of the AGB calculated for each of the 71 stands in the 1985 inventory, and the difference was taken to represent the overall change in forest AGB over 55 years.

AGB change between 1985 and 2000 was calculated in similar fashion for the 51 plots inventoried in both of these years. However, since these plots were not selected to represent the entire forest, simple means were calculated to quantify AGB change over 15 years on this subset of plots.