Effects of deforestation on ecosystem carbon densities in central Saskatchewan, Canada

Abstract

Ecosystem carbon densities (aboveground vegetation plus soil organic carbon (SOC) to 45 cm depth) were compared for forested and deforested sites on hummocky glacial till landforms in central Saskatchewan. Six forest sites in Prince Albert National Park, six pasture sites in or near the Cookson Community Pasture, and six privately owned cultivated fields were sampled across a 30 000 ha study area (53°34′N, 106°20′W). The median ecosystem C density for forests (158 Mg C ha⁻¹) was significantly greater (P<0.05) than the median for pastures (63 Mg C ha⁻¹) and the median for cultivated fields (81 Mg C ha⁻¹). For soil organic carbon alone, significant differences were not detected between forested and deforested sites. The range for SOC density across the study area was 51 Mg C ha⁻¹ for natural forest sites, 54 Mg C ha⁻¹ for pastures and 92 Mg C ha⁻¹ for cultivated fields. Spatial variation in SOC densities within each treatment group precluded detection of land-use effects on SOC densities (expected to be ≤25 Mg C ha⁻¹) with this research design. SOC may have been elevated at sites with shallower groundwater, but further research is required to confirm this and to identify the responsible processes. Differences in ecosystem C densities between the forested and deforested sites were primarily the result of differences in vegetation biomass. Median live plus dead vegetation C density for the forest group was 60 Mg C ha⁻¹ greater than the pasture group (P<0.05) and 56 Mg C ha⁻¹ greater than the cultivated group (P<0.15). Including estimated C in coarse roots and lesser vegetation at the forest sites would increase these differences by approximately 15 Mg C ha⁻¹. Accounting for disturbance processes such as forest fires, differences in vegetation carbon between forested and deforested sites in this study area are estimated to vary between 30 to 75 Mg C ha⁻¹ over time. Continued agricultural expansion within the study area would result in losses of at least 30 Mg C ha⁻¹. Recent ratification of the Kyoto protocol by Canada provides an incentive to curtail deforestation and initiate reforestation actions. Future reforestation with trembling aspen dominated vegetation in central Saskatchewan has the potential to sequester 30–75 Mg C ha⁻¹ over the next 50–100 years. However, legislative protection or strong financial incentives will be required to secure long-term carbon gains through reforestation.

Introduction

After fossil fuel burning, land-use change is the next largest source of anthropogenic carbon emissions. Houghton (1999) estimated that 87% of global emissions from land-use change over the past 150 years were from agricultural expansion in forested regions. Past deforestation and associated agricultural activities reduced both vegetation biomass and soil organic matter (Houghton et al., 1983, Buringh, 1984, Schlesinger, 1984, Mann, 1986, Houghton and Skole, 1990, Davidson and Ackerman, 1993).

Peak agriculture-related deforestation in eastern North America occurred a century ago. More recently, rural depopulation, farmland abandonment and expansion of tree plantations has led to increases in the extent of forest cover in this and other temperate regions (Williams, 1989, Rudel, 1998, FAO, 2001). Land-use change in western Canada runs counter to this pattern. In central Saskatchewan, agricultural settlement was barely underway by 1900. Development of croplands and pastures persisted for the next 90 years (Ramankutty and Foley, 1999) and deforestation may not yet have subsided (Fitzsimmons, 2002a, Hobson et al., 2002).

Houghton (1999) estimates losses of 85 Mg C ha⁻¹ for vegetation, and 51 Mg C ha⁻¹ for the top 1 m of soil, for North American boreal regions converted to agriculture. Beyond this continental estimate, more specific data on the effects of deforestation on carbon densities for western Canada are not available. Numerous estimates are available for vegetation biomass (Sulistiyowati, 1998), soil carbon storage (Xiao, 1987, Ellert and Bettany, 1995; Pennock and van Kessel, 1997a, Pennock and van Kessel, 1997b) or both (Halliwell and Apps, 1997a, Halliwell and Apps, 1997b, Halliwell and Apps, 1997c; Gower et al., 1997) at sites within central Saskatchewan. However, these data do not facilitate direct comparisons of forest and agricultural lands because of differences in methodology and site conditions (e.g. soil type). Comparable estimates of C densities at forested and agricultural sites are necessary to assess the impact of past deforestation and future land-use change on C stocks.

We investigated ecosystem C densities for forested and deforested sites in the Boreal Plain Ecozone (Acton et al., 1998) of Saskatchewan. Deforestation is defined as “the removal of a forest stand where the land is put to a non-forest land-use” (Helms, 1998, p. 44). Deforested sites included both pastures and cultivated fields. The objectives of the study were: (i) to estimate the magnitude of C losses associated with deforestation at sites; and (ii) to partition these losses among specific ecosystem components such as above-ground vegetation and soils to a fixed depth.

Our research design was a comparative mensurative experiment (Hurlbert, 1984) using space as an analogue for time. Three treatment groups were compared with the expectation that the effects of land-use on C density would be exhibited as differences in C densities among the forest group, pasture group and cultivated group. Three hypotheses were investigated. For soil organic carbon (SOC), we hypothesized:

$$\text{C}{}_{\mathrm{<mtext>forests</mtext>}}\text{=}\text{C}{}_{\mathrm{<mtext>pasture</mtext>}}\text{>}\text{C}{}_{\mathrm{<mtext>cultivated</mtext>}}\text{.}$$

Cultivated sites were expected to have the lowest SOC due to tillage-induced losses (see reviews by Mann (1986), and Davidson and Ackerman (1993)). For C in aboveground vegetation, we hypothesized:

$$\text{C}{}_{\mathrm{<mtext>forests</mtext>}}\text{⪢}\text{C}{}_{\mathrm{<mtext>cultivated</mtext>}}\text{>}\text{C}{}_{\mathrm{<mtext>pasture</mtext>}}\text{.}$$

Forests were expected to have the greatest aboveground C because of their perennial woody tissue and taller stature. Because of fertilization and lack of grazing at cultivated sites, biomass C values were expected to exceed those at pastures. For estimated ecosystem C (aboveground vegetation plus SOC), we hypothesized:

$$\text{C}{}_{\mathrm{<mtext>forests</mtext>}}\text{>}\text{C}{}_{\mathrm{<mtext>pasture</mtext>}}\text{>}\text{C}{}_{\mathrm{<mtext>cultivated</mtext>}}\text{.}$$

Forests were expected to have the greatest C densities because of greater aboveground vegetation C. Cultivated sites were expected to have lowest C densities due to tillage-induced SOC losses.