The Response of Soil Processes to Climate Change: Results ... - CREAF

ECOSYSTEMS

Ecosystems (2004) 7: 625– 637 DOI: 10.1007/s10021-004-0220-x

© 2004 Springer-Verlag

The Response of Soil Processes to Climate Change: Results from Manipulation Studies of Shrublands Across an Environmental Gradient Bridget A. Emmett,1* Claus Beier,2 Marc Estiarte,3 Albert Tietema,4 Hanne. L. Kristensen,5 Dylan Williams,6 Josep Pen˜uelas,3 Inger Schmidt,7 and Alwyn Sowerby1 1

Centre for Ecology and Hydrology–Bangor, Orton Building, Deiniol Rd., Bangor, Gwynedd LL572UP, United Kingdom; 2RISA National Laboratory, P.O. Box 49, DK-4000 Roskilde, Denmark; 3Plant Ecophysiological Unit, CSIC–CEAB–CREAF, Edifici C, Universitat Auto`noma de Barcelona, 08193 Bellaterra, Barcelona, Spain; 4Center for Geo-ecological Research (ICG), Institute for Biodiversity and Ecosystem Dynamics (IBED)–Physical Geography, University of Amsterdam, Nieuwe Achtergracht 166, 1018, WV Amsterdam, The Netherlands; 5Danish Institute of Agricultural Sciences, P.O. Box 102, DK-5792 Aarslev, Denmark; 6Countryside Council for Wales, Penrhos Road, Bangor, Gwynedd, United Kingdom; 7Danish Forest and Landscape Research Institute, Hørshom Kongevej 11, DK-2970 Hørsholm, Denmark

ABSTRACT sites. Highest Q10 values were observed in Spain and the UK and were therefore not correlated with soil temperature. A trend of increased accumulated surface litter mass loss was observed with experimental warming (2%– 22%) but there was no consistent response to experimental drought. In contrast to soil respiration and decomposition, variability in net N mineralization was best explained by soil moisture rather than temperature. When water was neither limiting or in excess, a Q10 of 1.5 was observed for net N mineralization rates. These data suggest that key soil processes will be differentially affected by predicted changes in rainfall pattern and temperature and the net effect on ecosystem functioning will be difficult to predict without a greater understanding of the controls underlying the sensitivity of soils to climate variables.

Predicted changes in climate may affect key soil processes such as respiration and net nitrogen (N) mineralization and thus key ecosystem functions such as carbon (C) storage and nutrient availability. To identify the sensitivity of shrubland soils to predicted climate changes, we have carried out experimental manipulations involving ecosystem warming and prolonged summer drought in ericaceous shrublands across a European climate gradient. We used retractable covers to create artificial nighttime warming and prolonged summer drought to 20-m2 experimental plots. Combining the data from across the environmental gradient with the results from the manipulation experiments provides evidence for strong climate controls on soil respiration, net N mineralization and nitrification, and litter decomposition. Trends of 0%– 19% increases of soil respiration in response to warming and decreases of 3%–29% in response to drought were observed. Across the environmental gradient and below soil temperatures of 20°C at a depth of 5–10 cm, a mean Q10 of 4.1 in respiration rates was observed although this varied from 2.4 to 7.0 between

Key words: shrubland; drought; warming; climate change; heathland; N mineralization; decomposition; respiration.

INTRODUCTION Climate change associated with increased greenhouse gas emissions is expected to cause an increase in global mean temperature and a more vigorous hydrological cycle resulting in more severe

Received 19 July 2002; accepted 16 May 2003; published online 23 June 2004. *Corresponding author; e-mail: [email protected]

625

626

B. A. Emmett and others

droughts and floods (Houghton and others 2001). Historical records already show an increase in atmospheric CO2 and in global annual surface temperatures of 0.6°C since the late 19th century although there is little evidence for changes in rainfall pattern (Houghton and others 2001). In attempting to predict the influence of these climate changes on terrestrial ecosystems, scientists have used a range of experimental techniques including field greenhouses, open-top chambers, soil warming, infrared lamps, transplantation studies, and environmental transects (Shaver and others 2000). All have their advantages and disadvantages, but many studies have reported significant changes in soil processes such as net nitrogen (N) mineralization and soil respiration which would have profound implications for ecosystem functioning by affecting changes in nutrient availability and carbon (C) storage (Kirschbaum 1995). In a recent synthesis of results from these studies carried out across a range of ecosystem types, warming was found to increase soil respiration by a mean of 20%, whereas net N mineralization rates increased by an average of 46% (Rustad and others 2001). However, there was a large degree of variability which, surprisingly, could not be related to climatic and environmental factors at individual sites, making prediction of responses for individual sites problematic. There have been fewer ecosystem studies assessing the direct effect of changing rainfall patterns and results are not always as expected. For example, a decline in net N mineralization rates was reported in response to increased rainfall and an increase was reported following experimental drought in a calcareous grassland (Jamieson and others 1998). Further problems in predicting the long-term effects of climate change on ecosystem processes arise from the transient effects that are typically not measured in short-term studies. These include acclimation of soil respiration rates and plant responses after one or more years of experimental treatment (for example, Oechel and others 2000; Luo and others 2001). These declines have been attributed to an interactive effect between warming and a decline in soil moisture, limitations in soil nutrient supply for plants, and a decline in the availability of labile C sources for soil microbes (for example, Peterjohn and others 1994; Arft and others 1999). In this study we used a combination of an experimental and gradient approach to quantify the effect of warming and prolonged summer drought on key soil processes as part of the EU CLIMOOR (climate-driven changes in the functioning of heath and moorland ecosystems) project. Ericaceous

shrublands are a widespread vegetation type that often have high conservation value and are used for a range of activities including grazing and hunting (Gimingham and others 1979). They are generally characterized by low nutrient availability and are currently threatened by a variety of environmental pressures including over- and undergrazing, lack of management, susceptibility to uncontrolled burns, and atmospheric deposition (Gimingham and others 1979; Aerts and Heil 1993). Climate change is likely to be an additional pressure and has the potential to affect both the structure and function of shrublands. To forecast these changes, dynamic models need to be developed in combination with data acquisition to both parameterize the models and test model outputs (Shaver and others 2000). Our study contributes to this need by providing data from ecosystem-level experiments across an environmental gradient that includes the often underrepresented southern region of Europe. We have employed an experimental approach that may elucidate the direction of change and some of the transient responses to climate change. Replicating these studies across an environmental gradient may help in elucidating longer-term responses, such as changes in plant species composition, which shift at decadal time scales. Our manipulation experiments altered both temperature and summer rainfall and were carried out using a common protocol across a European climate gradient spanning latitudes of 41– 56°N, mean annual air temperatures of 8 –15°C and mean annual rainfall of 455–1741 mm. Here we report the response of key soil processes, namely litter decomposition, soil N transformations, and soil respiration. Original hypotheses included a more positive response to warming in the colder, northern sites and a more negative response to prolonged summer drought in the southern site in Spain. Companion articles are in this issue on the experimental details involving our use of retractable roofs to create a drought or nighttime warming (Beier and others 2004), the response of plants (Pen˜ uelas and others, 2004), water quality (Schmidt and others 2004), 14 C allocation (Gorissen and others 2004), and the vulnerability of the heathlands (Wessel and others 2004).

METHODS Site Description The CLIMOOR sites were located in Clocaenog in North Wales (UK), Mols in Denmark (DK), Oldebroek in The Netherlands (NL), and Garraf in

Effects of Climate Change on Shrubland Soils

627

Table 1. Site Locations and Characteristics Site name Country

Mols DK

Clocaenog UK

Oldebroek NL

Garraf ES

Location

56° 23'N 10° 57'E 58 9.4 758 25–30 Calluna vulgaris Deschampsia flexuosa

52°24'N 5° 55'E 25 10.1 1042 30 – 40 Calluna vulgaris

41°19'N 1° 49'E 210 15.1 455 10 –15 Erica multiflora Globularia alypum

500 31 34 117 Sandy podzol

53° 03'N 3° 28'W 490 8.2 1741 20 –25 Calluna vulgaris Vaccinium myrtilus Empetrum nigrum 1790 30 33 NA Peaty podzol

584 52 27 NA Sandy podzol

275 11 170 NA Petrocalcic calcixerepts

0 –3 3.7 18.5 41 0.21

0–6 3.9 37.4 89 0.09

0–4 3.7 22.5 65 0.11

0 –12 8.1 10.4 7.8 0.48

3–20 4.1 1.8 1.39

6 –17 4.0 37 0.41

4 –16 3.8 3.3 1.41

12–37 8.3 3.5 NAa

Altitude Mean air temp (°C) Precipitation (mm) N deposition (kg N/ha/year) Vegetation

Aboveground C (g C/m2) Aboveground C/N Shrub litter C/N Grass litter C/N Soil Upper soil horizon Depth (cm) pH C/N Organic matter (%) Bulk density (g cm⫺3) Lower soil horizon Depth (cm) pH Organic matter (%) Bulk density (g cm⫺3) a

NA—not applicable.

Spain (ES). All sites were dominated by shrubs with perennial woody biomass. In DK, UK, and NL, the vegetation was dominated by the ericaceous shrub Calluna vulgaris. In 1999 –2000 an outbreak of the heather beetle (Lochmaea suturalis) occurred at the DK site resulting in significant defoliation of the Calluna plants. In ES, the vegetation was dominated by Globularia alypu and Erica multiflora, the latter of the ericaceous family. Further details of the sites are presented in Table 1, 2.

Treatments Nine experimental plots (5 m ⫻ 4 m) were established in relatively homogeneous areas within each site. Three treatments were allocated randomly (UK) or in blocks (DK, NL, ES): control (C), warming (W), and prolonged drought during the growing season (D). Around each plot, a light scaffolding structure was built of galvinized steel tubes covered by thin plastic sleeves to prevent contaminants from leaching into the plot. In the warming plots, this frame supports a retractable, reflective curtain made of strips of infrared-reflective material bound

into a high-density polyethylene mesh. A small motor activated by a light sensor extends this curtain over the vegetation at night thus preventing heat loss. A tipping bucket rain sensor retracts the curtain at night to enable rain to enter the plot. To prevent damage to the cover, a wind sensor activates the retraction of the curtain at night if wind speeds exceed 10 m/s. Over the drought plots, the retractable curtain is made of transparent polyethlene plastic. Rain sensors activate the motor to extend this cover over the plots when rain is detected and to retract the cover when the rain stops. When the curtain is extended in these drought plots to exclude the rain, wind sensors again retract the curtain to prevent damage during periods of high wind. The effect of the warming treatments on mean air and soil temperature was between 0.5 and 2°C (Beier and others 2004). In terms of growing season-days this was equivalent to a 5%– 10% increase in the three northern sites but no change in ES. The summer drought treatment reduced precipitation during the growing season by 65%–94%

628


Table 2. Summary of Net Nitrification Rates and Mineralization Rates Expressed on a per Gram Organic Matter and Area Basis

Site

Species

Horizona

DK

C. vulgaris

Upper

D. flexuosa

Upper

C. vulgaris

Upper

UK

Lower

NL

C. vulgaris

Upper

ES

E. multiflora

Upper

a

Min Max Mean Min Max Mean Min Max Mean Min Max Mean Min Max Mean Min Max Mean

Nitrification (␮g N/g OM/day)

Mineralization

⫺0.02 0.52 0.23 0.08 6.65 1.82 0 0 0 ⫺0.11 0 ⫺0.03 0.52 3.13 1.23 ⫺0.12 0.99 0.19

1.65 8.36 5.4 6.73 16.00 10.76 ⫺0.17 0 ⫺0.08 ⫺0.16 0.19 ⫺0.02 3.42 7.97 4.82 ⫺2.14 1.93 0.05

Nitrification

Mineralization [g N/m2/(SE)]

0.2 (0.1)

3.8 (0.9)

0.9 (0.7)

6.6 (0.5)

0 (0)

⫺0.1 (0.1)

⫺0.1 (0.1)

⫺0.2 (0.1)

1.4 (0.5)

5.6 (2.1)

1.0 (0.1)

0 (0.1)

Depth sampled in northern sites was varied to match organic layer.

and annual rainfall by 9%– 82% (Beier and others 2004). This occurred in the main growing seasons at all sites: between May and September for 55–74 days in UK, DK, and NL and from March to December for 229 –244 days in ES. The influence of the retractable curtains on humidity (DK), radiation (DK), and wind were tested at the DK site and were found to be minimal (Beier and others 2004). All measurements of climate variables and soil processes were recorded in an inner 4.5-m ⫻ 3.5-m plot leaving an outer 0.5-m border. Edge effects were studied in the DK site and were found to be relatively small (0.1– 0.2°C).

(Erica). Three replicate bags for each time period were placed either randomly (UK), stratified beneath Calluna plants and open areas (NL), beneath Calluna plants only (DK), or beneath and on the north side of Erica bushes (ES). Bags were retrieved over a period of 1–2 years, sorted to remove material that had entered the bags after installation, and oven-dried. Results are expressed as mass remaining as a percentage of initial mass. C and N content of litter quality was determined using combustion methods (UK, DK, and NL) and acid digestion and ICAP–AES (inductively coupled argon plasma emission spectrometer) (ES) (Table 1).

Decomposition

Soil Respiration

Ericaceous litter was collected from all sites either by shaking bushes (DK, NL, ES) or by picking standing dead material from plants (UK) in September– November 1999. Material was air-dried and sorted to achieve a homogeneous sample comprising short and long shoots of Calluna (DK, UK), Calluna shoots plus flowers (NL), and 2–3-year-old Erica leaves (ES). Standing dead stems and leaves of Deschampsia flexuosa were also collected in the fall of 1999 (DK only). Material (3 g) was placed in bags with a mesh size of 1 mm (Calluna and Deschampsia) and 0.71 mm

Permanent collars were installed in all plots to record in situ rates of soil respiration in the three northern sites (1 chamber/plot in DK and NL and 3 chambers/plot in the UK). These were installed to an approximate depth of 5–10 cm to minimize disruption to roots although some damage was inevitable. In ES, measurements were made in five random locations in each plot at each sampling time. Infrared gas analyzers were used at DK, ES, and UK sites (PP Systems; Models EGM-2, EGM-1, and CIRAS-2, respectively, all with chamber SRC-1). In NL, gas vials were used to collect samples for later

Effects of Climate Change on Shrubland Soils analysis using gas chromatography. Measurements were carried out either fortnightly (DK), monthly (NL, UK), or seasonally (ES) between 09 00 and 15 00 h. In the UK, temperature and moisture measurements of the upper 7 cm were taken using a Delta Thetaprobe handheld temperature probe. At all other sites, logged data from installed TDR probes and temperature sensors at 5–10-cm depth provided soil moisture and temperature data.

Net N Mineralization In the three sites dominated by Calluna (DK, UK, and NL), an intact soil core method was used for each plot. Three replicate paired cores approximately 5 cm in diameter were taken pretreatment in 1998 (NL) and during the first two years of treatment within different seasons in 1999 – 2001 (all sites). The top soil horizon only was sampled in DK, NL, and ES, whereas a lower mineral soil horizon was also sampled in the UK. The depth of the upper horizon sampled varied in NL and DK to reflect the variable organic layer: 1– 4 cm in the NL and 0.5–5.0 cm in DK. In the UK, the upper organic layer varied between 5.9 and 6.7 cm and the lower from 5.0 to 5.7 cm. One of each pair was returned to the laboratory for analysis for initial ammonium–N and nitrate–N content. The core to be incubated in the field was placed in a polythene bag and replaced in the ground for a period of 1–2 months depending on the season. A different method was employed in ES because no organic layer was present and the soil was too hard to obtain an intact soil core. Instead, the soil was sampled to 13-cm depth (no organic layer is present). Coarse stones were removed and soil was repacked in a plastic bag and installed in the soil beneath plants protected by a polythene cover. After incubation, cores were removed and the change in ammonium and nitrate content determined. The soil material was lightly mixed and large roots removed prior to extraction in 1 M KCl using a soil-to-extract ratio of 1:5 (ES), 1:10 (DK, UK), and 1:30 (NL). Soil moisture was determined after drying at 70 – 80°C (NL, DK, UK) and 105°C (ES) and organic matter content was determined by combustion (UK, DK, NL) and wet oxidation (ES). The change in inorganic N (net mineralization) and nitrate N only (net nitrification) during the incubation period was calculated and results were expressed both as the change in extractable N content per gram organic matter (␮g N/g OM) and the change in extractable N content on an area basis (mg N/m2).

629

DATA ANALYSIS Decomposition A mean weight loss was calculated for each plot because we considered plots to be the smallest independent unit. Data were log transformed and rates of decomposition calculated using the following equation: ln共W t/W 0兲 ⫽ ln W 0 ⫺ kt

(1)

where Wt is the weight of decomposed litter at time t, W0 is the initial weight of litter, t is time, and k is the fractional weight loss. We performed one-way ANOVAs for each site on annual fractional weight loss (k) at each sampling period to determine the effect of treatments in UK, NL, and ES. For DK, an additional two-way ANOVA was performed for an interspecies comparison. Data across the sites were combined for ericaceous litter at the final sampling period to determine the effect of site and treatment on fractional weight loss. Between-treatment comparisons were carried out following ANOVA using a Tukey test if significant differences were observed. Differences were considered significant at p ⬍ 0.05 level. To compare the relative responsiveness of the treatments across the sites, we calculated the difference in mass loss in the means (M) as a percentage of the control: Warming Change: W ⫽ M W ⫺ M C/M C

(2)

Drought Change: D ⫽ M D ⫺ M C/M C

(3)

A positive value indicates an increase in the rate of mass loss, a negative value indicates a decrease. A multiple regression approach was performed on annual fractional weight loss of ericaceous litter with annual rainfall and mean annual air temperature combining data across all sites and treatments.

Soil Respiration Treatment effects for each site were compared using one-way ANOVA. Between-treatment comparisons were carried out on mean treatment data (n ⫽ 3/treat/site) for each sampling time if significant differences were found using a Tukey test. Mean treatment data over the whole sampling period were combined at a site level for each plot and an ANOVA carried out to determine overall site and treatment effects. The relative effect of the two treatments was calculated as for litter decomposition and expressed as a percentage of the control. A multiple regression approach was carried out to

630


determine the relationships between soil respiration and soil moisture and temperature for control, drought, and heated treatments for all sites.

Net N Mineralization Mean results were calculated for each plot for each time period prior to statistical analysis. One-way ANOVAs were performed to determine treatment differences within each site. Because of the large variability in rates between sites, a standardized mean difference was calculated using the following equation: d⫽

Xt ⫺ Xc sd

(4)

where Xt is the mean of the experimental treatment, Xc is the mean of the control treatment and sd is the pooled standard deviation. Mean treatment data for each site were calculated and combined for a two-way ANOVA with site and treatment as fixed variables. A multiple regression approach was used to determine the relationships between net N mineralization and nitrification rates and soil moisture and temperature using data for control, drought, and warming for all sites.

RESULTS Decomposition Accumulated mass loss at the final sampling time was between 27% and 38% for Calluna (DK, UK, NL), 38% for Deschampsia (DK), and 20% for Erica litter (ES), 350 – 656 days (Figure 1a,b). No significant effect of treatment on the cumulative mass loss of Calluna, Erica, and Deschampsia at the final sampling time was recorded within any site (Figure 1b). However, the drought treatment significantly reduced mass loss during earlier samplings by 4%– 10% in three of the four sites (DK, NL, and ES) at the p ⬍ 0.01 level. This effect disappeared after the sampling on day 72 for Deschampsia in DK, day 192 for Calluna in DK, day 84 for Calluna in NL, and day 141 for Erica in ES. No significant species effect was observed between Calluna and Deschamspia litter at the DK site. When annual fractional weight loss for the ericaceous shrubs was combined across all sites for the final sampling period, a significant site (p ⬍ 0.01) and treatment (p ⬍ 0.05) effect was observed. Annual fractional weight loss in the drought treatments (0.26) was significantly lower relative to the heated (0.31) treatments (p ⬍ 0.05). Neither significantly different from the control (0.28). Annual

Figure 1. Mass remaining in litter bags (a) in the control plots in all four CLIMOOR sites for Calluna, Deschampsia, and Erica litter, and (b) of ericaeous litter installed in the three treatments. C ⫽ control, W ⫽ warming, and D ⫽ drought. Site codes are as in Table 1 Standard errors bars are shown.

mass loss was significantly lower in the ES litter bags (0.16) compared with all other sites. UK mass loss rate (0.27) was also significantly lower compared with the NL and DK rates (0.36 and 0.34, respectively) (p ⬍ 0.01). This resulted in an overall trend of increasing annual mass loss in the following order: ES is less than the UK which is less than DK which is equal to NL. When treatment results were expressed as a percentage of control, a general trend of a positive effect of warming in the northern site (11%–22%) can be seen (Figure 2a). No consistent trend in the effect of drought was observed. Multiple regression identified temperature as accounting for 51% of the variability observed between sites and treatments (r2 ⫽ 0.51, p ⬍ 0.01). If ES data were excluded assuming moisture limitation, this resulted in temperature accounting for 81% of the variability and a Q10 based on air temperature of 6.2. Inclusion of moisture in the model did not significantly improve the relationship in either case.


631

Figure 3. (a) Seasonal soil respiration rates and soil temperature in control treatments, and (b) the relationship between soil respiration measurements and soil temperature measurements (5–10 cm) for each site excluding data below 20°C. Site codes are as in Table 1.

Soil Respiration

Figure 2. The graphs show the change due to drought and warming treatments expressed as a percentage of control for (a) annual litter mass loss and (b) soil respiration, and the standardized mean change for (c) net N mineralization rates and (d) net nitrification rates across the different sites. Standard error bars are shown. Site codes are as in Table 1.

Strong seasonal trends in soil respiration rates were observed in the DK, UK, and NL sites (Figure 3a). This trend closely followed seasonal patterns of soil temperature. There were insufficient data to determine seasonal patterns in ES. Significant differences between treatments were observed at all sites at individual sampling times (7 occasions in DK, 3 in the UK, 2 in NL, and 4 in ES). Most significant effects were between control and drought treatment plots and occurred in both summer and winter, so included periods outside the experimental drought period. Significant warming effects were observed only at the DK site and on only three occasions. When all data were combined, a significant site (p ⬍ 0.01) and treatment (p ⬍ 0.05) effect was observed. Respiration rates across sites increased in the order: NL is less than DK which is less than ES which is equal to UK. There was a trend for reduced soil respiration rates of 3%–29% in response to drought relative to the control equivalent to 0.004 – 0.046 g C m⫺2 h⫺1, and a 0%–19% increase in response to warming equivalent to

632


0 – 0.006 g C m⫺2 h⫺1 (Figure 2b). Rates in the drought and heated plots were significantly different at the p ⬍ 0.05 level, but neither were significantly different from the control. A single exponential response curve was fitted to data within the 0 –20°C range which explained 52% of the variation r2 ⫽ 0.52 with a Q10 ⫽ 4.1 at a soil depth of 5–10 cm: 5 Soil respiration ⫽ 0.0111e 0.1407T

(5)

However, a range of relationships to temperature was observed at the individual sites in this temperature range (Figure 3b): DK soil respiration ⫽ 0.0256e 0.0904T

(6)

UK soil respiration ⫽ 0.0158e 0.1951T

(7)

NL soil respiration ⫽ 0.078e 0.1434T

(8)

ES soil respiration ⫽ 0.0163e 0.1666T

(9)

All were significant at the p ⬍ 0.05 level. Q10 values ranged from 2.4 DK, to 4.2 in NL, 5.3 in ES, and 7.0 in the UK at a soil depth of 5–10 cm. When all data were combined, including the data above 20°C, a linear relationship to temperature was observed that explained 52% of the variation (not shown). Inclusion of soil moisture in the model significantly improved the variation explained to 73%: Soil respiration ⫽ ⫺0.0898 ⫹ 0.00674T ⫹ 0.367M 共 p ⬍ 0.01, p 2 ⫽ 0.73兲

(10)

Net N Mineralization and Nitrification Rates of net nitrification and N mineralization were in the range of ⫺0.1 to 6.6 ␮g N/g OM/day and ⫺2.1 to 16.0 ␮g N/g OM/day, respectively (Table 2). Rates of an area basis were calculated by combining rates per gram dry weight of soil with core bulk density values. Annual rates were calculated by multiplying mean daily rates on an area basis by 365 to provide an annual estimate. Net nitrification rates ranged rom ⫺0.1 g N/M2/year in the UK to 1.4 g N/m2/year in the NL. Net mineralization rates were equivalent to ⫺0.1 g N/m2/year in the UK to 6.6 g N/m2/year in Deschampsia-dominated areas in DK. No significant treatment effects on net N mineralization or nitrification were observed in the UK and DK at any sampling time or when data were combined across all sampling periods. In the NL a significant reduction in net nitrification rates in re-

sponse to the drought in June 2000 was observed and for net N mineralization when data were combined across all sampling periods. In ES, no significant differences relative to the control were observed. However, net nitrification rates in the drought treatments were significantly lower than the warming treatments in May 2000 and again in May 2001. A significant difference between net N mineralization rates in areas dominated by Calluna relative to Deschampsia in DK was observed during one sampling time. In the UK, net N mineralization and nitrification rates in the upper organic layer and lower mineral soil were compared. No significant differences were observed. When data for the upper soil layers beneath the ericaceous shrubs were combined across the CLIMOOR sites, a significant site (p ⬍ 0.001) and treatment effect (p ⬍ 0.05) was observed for both net N mineralization and nitrification. Treatment comparisons indicated that net N mineralization rates were significantly different between drought (1.32 ␮g N/g OM/day) and warming (3.33 ␮g N/g OM/ day) plots (p ⬍ 0.01) but neither were significantly different from the control (2.64 ␮g N/g OM/day). For nitrification rates, drought significantly reduced rates relative to the control (⫺0.07 ␮g N/g OM/day and 0.51 ␮g N/g OM/day) and was also significantly different to the control (0.82 ␮g N/g OM/day). There was no obvious relationship between effect size and the environmental gradient across the sites (Figure 2c,d). The lowest rates of both net nitrification and N mineralization were recorded at the two extremes of the CLIMOOR climate gradient (Figure 4a, 5b): the wet, cold UK site and the hot, dry site in ES. Net N mineralization rates were significantly related to soil moisture content across the sites, with positive rates generally recorded in soils with moisture content between 20% and 60% (Figure 4a,b). A regression equation containing quadratic terms was found to best describe the relationship to initial soil moisture content (M) (r2 ⫽ 0.49, p ⬍ 0.001): Net N mineralization ⫽ ⫺52.33M 2 ⫹ 44.73M ⫺ 4.42

(11)

Within the soil moisture range (20%– 60%), a significant linear response to temperature (T) was observed (Figure 5) (r2 ⫽ 0.27, p ⫽ 0.06): Net mineralization ⫽ 1.66 ⫹ 0.345T

(12)

No significant relationship with moisture was observed within this range. A regression equation containing a quadratic


Figure 4. The relationship between seasonal net N mineralization and initial soil moisture content (g/g wet soil) at the start of incubations (a) across the four sites and (b) for the three treatments (C ⫽ control, D ⫽ drought and W ⫽ warming). Only data from the upper soil horizons beneath shrubs are included. Site codes are as in Table 1.

633

Figure 6. The relationship between (a) seasonal net nitrification and initial soil moisture content (g/g wet soil) at the start of incubations and (b) net nitrification and initial soil moisture content for the three treatments (C ⫽ control, d ⫽ drought, and W ⫽ warming) across the four CLIMOOR sites. Only data from the upper soil horizon beneath shrubs in included. Arrows indicate one single low value for the DK site. Site codes are as in Table 1.

Net nitrification ⫽ ⫺7.59M 2 ⫹ 6.65M ⫺ 0.64 (13) When soil moisture was 20%– 60% (DK and NL), a linear relationship, including both temperature and moisture, was found to provide the best relationship (r2 ⫽ 0.47, p ⬍ 0.001): Figure 5. The relationship between net N mineralization and mean upper soil temperature during the incubation period for the three experimental treatments in shrubdominated soils. Data points are limited to incubations with an initial moisture content between 20% and 60%. Regression line shown (p ⬍ 0.05) is for all treatments combined.

Net nitrification ⫽ ⫺4.13 ⫹ 0.1533T ⫹ 7.39M (14)

DISCUSSION Litter Decomposition

term also described the relationship of net nitrification to initial soil moisture content, although this described only 15% of the variation (Figure 6a,b) (r2 ⫽ 0.15, p ⫽ ⬍ 0.05):

Decomposition can be affected by two main factors, litter quality and climatic variables (Fog 1988; Berg and others 1996). For example, litter with high N content has been observed to have lower rates of litter decomposition in the latter stages possibly as a

634


result of the formation of recalcitrant condensation products that are highly resistant to biological degradation. Low manganese content of litters may also have an important effect in limiting formation of lignolytic enzymes (Berg and others 1996). In our study, the affect of litter quality was compared at only the DK site, although differences in litter quality may also contribute to intersite variability across the environmental gradient. No significant differences between decomposition rates of Calluna and Deschampsia were observed, suggesting litter quality may have a limited influence on decomposition rates of plant material in this heathland system. The effect of climate can be observed both from the impacts of the experimental treatments and by comparing results across the environmental gradient. Across the environmental gradient, mass loss rates were positively related to air temperature, which explained 81% of the variance in the three northern sites (Q10 ⫽ 6.2). Rates were significantly lower in our southern site probably because of water limitation. The generally positive effect of warming treatments in the northern sites (Figure 2a) also indicates an important role for temperature in the more northern sites which is absent in ES.

Respiration Respiration rates recorded here were within those reported across a range of sites (Rustad and others 2001) and were greatest at the southern most site, ES, and the western site, UK. This is in contrast to the decomposition results in which rates were lowest at these two sites. As with many other studies, this study indicates soil respiration is largely temperature dependent. Within sites, soil respiration is largely temperature dependent with soil temperature at 5 -10cm depth explaining 52% of the variation at temperatures below 20°C, a Q10 ⫽ 4.0 was observed which is toward the high end of those suggested by Kirschbaum (2000) following a synthesis of data from seasonal and soil warming studies but similar to those suggested from 14C isotope studies (Q10 ⫽ 3.8). Our data from this gradient study would predict an increase of 0.003 g C m⫺2 h⫺1 in response to the 1°C rise in temperature in the experimental treatments. A trend of increased respiration rates was observed in the warming treatments of a similar magnitude with a mean of 0.005 g C m2 h⫺1 when all sites were combined although this was not statistical by significant because of the large variability in measurements. However, this general relationship ignores a wide range of Q10 values across the different sites (Figure 3b). These ranged from 2.4 in DK to 4.2 in NL, 5.3

in ES, and 7.0 in the UK. Past studies generally indicate a greater positive response to temperature (that is a higher Q10) in soils at lower temperatures (Kirschbaum 1995). No such relationship is observed here with Q10 values in ES greater relative to the colder DK and NL sites. Unfortunately, no relationship to any measured environmental variable was identified. The relative contribution of plant roots, their mycorrhizal fungi, and free-living heterotrophs to soil respiration and its response to temperature is not known in this or many other studies. Roots have been found to contribute between 33% and 62% of soil respiration in forest ecosystems (Raich and Tufekcioglu 2000) and are thought to be more responsive to temperature than bulk soil (Boone and others 1998). It has been proposed that soil microbes and other soil biota may actually be responding to the supply of substrate from current photosynthate rather than temperature per se (Hogberg and others 2001). Our data however, show a coincidence of maximum soil temperatures and soil respiration at the northern sites suggesting that the dominant pathway in these heathland soils may be driven by temperature rather than the immediate supply of photosynthate. Moisture limitations may sometimes mask the temperature response, particularly at higher temperatures (Davidson and others 1998). This is likely the reason for the absence of a relationship to temperature above 20°C in ES. In addition, the importance of moisture was indicated by the significant reduction in respiration rates reported at individual sampling times in drought treatments at all sites, and the 22% of variance explained by soil moisture. Overall, the experimental treatments reduced soil moisture by 6%. Using the relationship derived from across the environmental gradient, a decrease of 0.007g C m2 h⫺1 would be predicted in response to this 6% decline. Results from the experimental drought treatment indicate reductions of a similar magnitude with a mean reduction of 0.011 g C m2 h⫺1 relative to the control treatment when all data were combined. No universal function equation is available to describe the relationship between soil respiration and soil moisture (Davidson and others 2000). One reason may again be the influence of changes in plant-derived substrate on soil microbial activity. Plant productivity is generally thought to be more dependent on water availability than on soil respiration (A˚ gren and others 1996). Thus, the effect of changes in primary productivity due to changes in water availability, as observed in our study (Pen˜ uelas and others), or any other environmental

Effects of Climate Change on Shrubland Soils variable may have a significant effect on above- and below ground substrate supply for the soil microbes, as well as a direct effect on root respiration (Jamieson and others 1998).

Net N Mineralization and Nitrification A wide range of net N mineralization and nitrification rates were recorded across the sites and treatments equivalent to ⫺0.2 to 6.6 g N m⫺2 year⫺1 and ⫺0.1 to 1.4 g N m⫺2 year⫺1, respectively. The lower end of this range is similar to that previously reported for tundra and heath-type vegetation in northern latitudes (for example, Kristensen and Henriksen 1998; Schmidt and others 1999; Jonasson and others 1999). The upper end of the range is similar to that reported in drier heathland or forest sites (for example, Berendse 1990; Van Vuuren and van Breemen 2000; Verburg and others 2000; Gundersen and others 1998). In the UK site, net immobilization was often recorded (negative values), possibly as a result as a combination of wetter and colder conditions and the low availability of other nutrients (Kristensen 2001). The plants may outcompete the microbes for inorganic N when not isolated in an incubation core or they may be utilizing organic-N sources (Na¨ sholm and others 1998). Rates in ES were also low possibly a result of the hot, dry conditions. Positive net N mineralization rates were consistently reported for the NL and DK sites with intermediate soil moisture contents of 20%– 60%. High N deposition at the NL site may contribute to the relatively high rates observed there. At the DK site, the heather beetle outbreak in 1999 –2000 may have affected rates of net N mineralization as found previously (Kristensen 2001). Across the environmental gradient, the effect of soil moisture explained 46% of variance in net N mineralization rates in the upper soil layer beneath the ericaceous shrubs. Soil temperature at 5–10-cm depth was significantly related to net N mineralization rates only within a moisture range of 20%– 60% when a temperature quotient Q10 of 1.5 was observed. There were few significant effects between the experimental drought treatments and the control because of the large variability associated with measurements. No significant effect of experimental warming on net N mineralization and nitrification rates relative to the controls was observed beneath the ericaceous shrubs. The absence of any significant effect of experimental warming, even within this moisture range, is not surprising considering the modest temperature increase achieved (approximately 1°C) and the variability associated with the incubation method. Within the 20%– 60% moisture range, the

635

regression relationship derived from the environmental gradient suggests that this temperature increase would result in a rise of only 2 ␮g N/g OM/day, which is significantly lower than the standard error of measurement at all sites. The lower sensitivity of net N mineralization to soil temperature (Q10 ⫽ 1.5) relative to soil respiration (Q10 ⫽ 2.0) has previously been reported but the reasons for this are not well understood (Kirschbaum 1995). Mineralization is a more complex process, being the net effect of gross mineralization and immobilization. In addition, plant roots do not contribute directly to mineralization rates although they are a major component of soil respiration. One possibility is that the contribution of plant roots to soil respiration underlies this greater sensitivity to temperature. It is also interesting to note that soil respiration appears to be less dependent on water availability than net N mineralization, which depends on soil microbial metabolism alone. Kirschbaum (2000) argues that as both processes ultimately depend on the supply of available C, the controls of decomposition are more relevant in the long term. The problem of identifying the direct and indirect plant contributions to soil respiration in the field and thereby gaining a fuller understanding of these controls remains to be solved. Finally, the importance of the abiotic process of both fixation of inorganic N (for example, Dail and others 2001) and stabilization of organic-C pools (Thornley and Cannell 2001) should not be ignored. For example, Thornley and Cannell (2001) suggest that current dynamic models underestimate the temperature dependence of abiotic reactions of C in the soil, such as chemical bonding onto minerals, and thus overestimate the loss of soil C due to increasing temperature. These processes may also contribute to the different sensitivity to soil moisture and temperature of C and N turnover in soils.

CONCLUSIONS Results presented here suggest that key soil processes are differentially affected by the gradients of rainfall and temperature within the latitudinal range of the study. We observed soil moisture as the primary variable explaining variability in net N mineralization and nitrification rates. In contrast, soil temperature was the best predictor of soil respiration and litter decomposition, possibly because of the effect of temperature on plant root activity, which also contributes to soil respiration measurements. Our experimental warming and drought treatments provided evidence that soil processes respond to climatic change in line with the trends

636


observed across the environmental gradient at a gross scale. However, there are significant differences in the temperature sensitivity of respiration that we were not able to relate to any environmental variable. Overall, our results suggest that key soil processes respond differently to climate change and these may be difficult to predict without a greater understanding of the factors that determine sensitivity of the individual processes to the different climate variables.

ACKNOWLEDGEMENTS

Fog K. 1988. The effect of added nitrogen on the rate of decomposition of organic matter. Biol Rev 63:433– 62. Gimingham CH, Chapman SB, Webb NR. 1979. European heathlands In: Specht RLEcosystems of the World 9A: Heathlands and Related Shrublands Amsterdam: Elsevier. p 365– 413. Gorissen A, Tietema A, Joosten NN, Estiarte M, Pen˜ uelas J, Sowerby A, Emmett BA, Beier C. 2004. Climate change affects carbon allocation to the soil in shrublands. Ecosystems 7:650 – 61. Gundersen P, Emmett BA, Kjonaas OJ, Koopmans CJ, Tietema A. 1998. Impact of nitrogen deposition on nitrogen cycling in forests: a synthesis of NITREX data. For Ecol Manage 101:37– 55.

The project was funded by EU under the projects CLIMOOR (contract No. ENV4-CT97-0694) and VULCAN (contract No. EVK2-CT-2000-00094) and the participating research institutes.

Ho¨ gberg P, Nordgren A, Buchmann N, Taylor AFS, Ekbld A, Ho¨ gberg MN, Nyberg G, Ottosson–Lo¨ fvenius M, Read DJ. 2001. Large-scale forest girdling shows that current photosynthesis drives soil respiration. Nature 411:789 –92.

REFERENCES

Houghton JT, Ding Y, Griggs DJ, Noguer M, van der Lindern PJ, Xiaosu D. 2001, Climate Change 2001: The Scientific Basis. Cambridge: Cambridge University Press.

Aerts R, Heil GW. 1993, Heathlands: Patterns and Processes in a Changing Environment. Dordrecht: Kluwer Academic Publishers. A˚ gren GI, Kirschbaum MUF, Johnson DW, Bosatta E. 1996. Ecosystem physiology—Soil organic matter In: Breymeyer AI, Hall DO, Melillo JM, A˚ gren GIGlobal Change: Effects on Coniferous Forests and Grasslands. SCOPE New York: Wiley. p 207–28. Arft AM, Walker MD, Gurevitch J, Altalo JM, Bret–Harte MS, Dale M, Diemer M, Gugerli F, Henfy GHR, Jones MH, Hollister R, Jo´ nsdo´ ttir IS, Laine K, Le´ vesque E, Marion GM, Molau U, MAlgaard P, Nordenha¨ ll U, Raszhivin V, Robinson CH, Starr G, Stenstro¨ m A, Stenstro¨ m M, Totland A, Turner L, Walker L, Webber P, Welker JM, Wookey PA. 1999. Response patterns of tundra plant species to experimental warming: a metaanalysis of the International Tundra Experiment. Ecol Mongr 69:491–511. Beier C, Emmett BA, Gundersen P, Tietema A, Pen˜ uelas J, Estiarte M, Gordon C, Gorissen A, Llorens L, Roda F, Williams D. 2004. Novel approaches to study climate change on terrestrial ecosystems at the field scale: drought and passive nighttime warming. Ecosystems 7:583–97. Berendse F. 1990. Organic matter accumulation and nitrogen mineralization during secondary succession in heathland ecosystems. J Ecol 78:413–27. Berg B, Ekbohm G, Johansson M, McClaugherty C, Rutigliano F, de Santo AV. 1996. Maximum decomposition limits of forest litter types: a synthesis. Can J Bot 74:659 –72. Boone RD, Nadelhoffer KJ, Canary JD, Kaye JP. 1998. Roots exert a strong influence on the temperature sensitivity of soil respiration. Nature 396:570 –2. Dail BD, Davidson EA, Chorover J. 2001. Rapid abiotic transformation of nitrate in an acid forest soil. Biogeochemistry 54: 131– 46. Davidson EA, Belk E, Boone RD. 1998. Soil water content and temperature as independent or confounded factors controlling soil respiration in a temperate hardwood forest. Global Change Biol 4:217–27. Davidson EA, Verchot LV, Cattanio JH, Ackerman IL, Carvalho JEM. 2000. Effects of soil water content on soil respiration in forest and cattle pastures of eastern Amazonia. Biogeochemistry 48:53– 69.

Jamieson N, Barraclough D, Unkovich M, Monaghan R. 1998. Soil N dynamics in a natural calcareous grassland under a changing climate. Biol Fertil Soils 27:267–73. Jonasson S, Michelsen A, Schmidt IK. 1999. Coupling of nutrient cycling and carbon dynamics in the Arctic, integration of soil microbial and plant processes. Appl Soil Ecol 11:135– 46. Kirschbaum MUF. 1995. The temperature dependence of soil organic matter decomposition and the effect of global warming on soil organic carbon storage. Soil Biol Biochem 27:753– 60. Kirschbaum MUF. 2000. Will changes in soil organic carbon act as a positive or negative feedback on global warming. Biogeochemistry 48:21–51. Kristensen HL. 2001. High immobilization of NH4⫹ in Danish heath soil related to succession, soil and nutrients: Implications for critical loads of N. Water Air Soil Pollut Focus 1:211– 30. Kristensen HI, Henriksen K. 1998. Soil nitrogen transformations along a successional gradient from Calluna heathland to Quercus forest at intermediate atmospheric nitrogen deposition. Appl Soil Ecol 8:95–109. Luo Y, Wan S, Hui D, Wallace LL. 2001. Acclimatization of soil respiration to warming in a tall grass prairie. Nature 413: 622– 4. Na¨ sholm T, Ekbald A, Nordin A, Giesler R, Ho¨ gberg M, Ho¨ gberg P. 1998. Boreal forest plants take up organic nitrogen. Nature 392:914 – 6. Oechel WC, Vourlitis GL, Hastings SJ, Zulueta RC, Hinzman L, Kane D. 2000. Acclimation of ecosystem CO2 exchange in the Alaskan Arctic in response to decadal climate warming. Nature 406:978 –981. Pen˜ uelas J, Gordon C, Llorens L, Nielsen T, Tietema A, Beier C, Bruna P, Emmett B, Estiarte M, Gorissen T. 2004. Non-intrusive field experiments show different plant responses to warming and drought among sites, seasons and species in a north– south European gradient. Ecosystems 7:598 – 613. Peterjohn WT, Melillo JM, Steudler PA, Newkirk KM, Bowles ST, Aber JD. 1994. Responses of trace gas fluxes and N availability to experimentally elevated soil temperatures. Ecol Applic 4:617–25. Raich JW, Tufekcioglu A. 2000. Vegetation and soil respiration: Correlations and controls. Biogeochemistry 48:71–90.

Effects of Climate Change on Shrubland Soils Rustad LE, Campbell JL, Marion GM, Norby RJ, Mitchell MJ, Hartley AE, Cornelissen JHC, Gurevitch J. 2001. GCTE-NEWS. A meta-analysis of the response of soil respiration, net N mineralisation, and above-ground plant growth to experimental ecosystem warming. Oecologia 126:543– 62. Schmidt IK, Williams DL, Tietema A, Gundersen P, Beier C, Emmett BA, Estiarte M. 2004. Soil solution chemistry and element fluxes in three European heathlands and their responses to warming and drought. Ecosystems 7:638 – 49. Schmidt IK, Jonasson S, Michelsen A. 1999. Mineralisation and microbial immobilization of N and P in arctic soils in relation to season, temperature and nutrient amendment. Appl Soil Ecol 11:147– 60. Shaver GR, Canadell J, Chapin FS III, Gurevitch J, Harte J, Henry G, Ineson P, Jonasson S, Melillo J, Pitelka L, Rustad L. 2000.

637

Global warming and terrestrial ecosystems: a conceptual framework for analysis. Bioscience 50:871– 82. Thornley JHM, Cannell MGR. 2001. Soil carbon storage response to temperature: an hypothesis. Ann Bot 87:591– 8. Van Vuuren MMI, Aerts R, Berendse F, De Visser W. 1992. Nitrogen mineralization in heathland ecosystems dominated by different plant species. Biogeochemistry 16:151– 66. Verburg PSJ, van Breemen N. 2000. Nitrogen transformations in a forested catchment in southern Norway subjected to elevated temperature and CO2. For Ecol Manage 129:31–9. Wessel, WW, Tietema, A, Beier, C, Emett, BA, Penuelas, J, Riis-Nielsen, T (2004) ⬙A qualitative ecosystem assessment for different shrublands in Western Europe under impact of climate change using the results of the CLIMOOR research project⬙ Ecosystems 7:662–71.