Click Here
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 113, G02030, doi:10.1029/2007JG000583, 2008
for
Full Article
Diurnal, seasonal, and interannual variation in carbon dioxide and energy exchange in shrub savanna in Burkina Faso (West Africa) Christian Bru¨mmer,1 Ulrike Falk,2 Hans Papen,1 Jo¨rg Szarzynski,2,3 Reiner Wassmann,1 and Nicolas Bru¨ggemann1 Received 1 September 2007; revised 1 April 2008; accepted 14 April 2008; published 18 June 2008.
[1] Savannas cover large areas of the Earth’s surface and play an important role in global
carbon cycling. West Africa is dominated by a variety of savanna ecosystems; however, they have been poorly studied up to now. In the present study, energy and carbon dioxide exchange was measured over a 2-year period using the eddy covariance technique at a southern Sudanian savanna site in Burkina Faso that was not subject to human disturbances except for annual burning. Turbulent energy exchange was dominated by sensible heat flux in the dry season (November–May) and by latent heat flux in the wet season (June–September), with peak values higher than 300 W m2 and lower than 100 W m2 for the dominating and the minor component, respectively. The ecosystem was a marginal CO2 source in the dry season, whereas significant CO2 uptake was found in the rainy season. CO2 fluxes showed a clear diurnal pattern, with mean release rates of 0.2 mg CO2 m2 s1 during nighttime and mean maximum uptake rates of 1.0 mg CO2 m2 s1 in July and August around midday. Diurnal courses of CO2 flux were in phase with photosynthetically active radiation (PAR). Highest CO2 uptake rates of more than 0.8 mg CO2 m2 s1 occurred at PAR levels in excess of 1000 mmol m2 s1. Total net ecosystem CO2 uptake was 179 ± 98 g C m2 in the first year and 429 ± 100 g C m2 in the second year of investigation, including an estimate of annual fire C loss of 149 g C m2. The remarkable difference in net CO2 uptake between the two years could be explained by higher rainfall in 2006, revealing the sensitivity of the ecosystem to water availability and rainfall distribution that could lead to a significant change in C sequestration patterns under a changing climate. Citation: Bru¨mmer, C., U. Falk, H. Papen, J. Szarzynski, R. Wassmann, and N. Bru¨ggemann (2008), Diurnal, seasonal, and interannual variation in carbon dioxide and energy exchange in shrub savanna in Burkina Faso (West Africa), J. Geophys. Res., 113, G02030, doi:10.1029/2007JG000583.
1. Introduction [2] Temporal variations in magnitude and location of terrestrial carbon sources and sinks as well as their response to environmental parameters are key determinants of seasonal and interannual variations in global carbon fluxes. For a better estimation of cumulative carbon budgets at broader scales, investigations of net ecosystem exchange (NEE) of carbon dioxide in various types of ecosystems are needed. Diurnal as well as seasonal changes in NEE reflect the amplitude of photosynthetic carbon uptake and respiration losses [Falge et al., 2002]. Hence, determination of diurnal
1 Atmospheric Environmental Research Division, Institute of Meteorology and Climate Research, Forschungszentrum Karlsruhe, GarmischPartenkirchen, Germany. 2 Department of Ecology and Resource Management, Center for Development Research, University of Bonn, Bonn, Germany. 3 Now at Office for Outer Space Affairs, German Aerospace Center, Bonn, Germany.
Copyright 2008 by the American Geophysical Union. 0148-0227/08/2007JG000583$09.00
and seasonal variation in NEE provides a better insight into carbon sequestration potentials of any ecosystem. [3] In the last few years, numerous studies focused on the analysis of carbon, water, and energy budgets of tropical, temperate, and boreal forest ecosystems [e.g., Goulden et al., 1996; Wilson and Meyers, 2001; van Wijk and Bouten, 2002; Chen et al., 2004; Paw U et al., 2004]. In contrast, much less attention has been paid to carbon cycling in savanna ecosystems, covering about 17 million km2 of the Earth’s surface between the equatorial rain forests and midlatitude deserts, i.e., 65% of Africa, 60% of Australia, 45% of South America, and 10% of India and Southeast Asia [Cole, 1986]. As savannas potentially play a significant role in the global carbon budget, a better quantification of their carbon balance is needed. However, only a few studies on carbon dioxide exchange in West African savannas have been reported up to now, e.g., by Verhoef et al. [1996] and Hanan et al. [1998] in Sahelo-Sudanian sites in Niger, receiving an annual rainfall of 300 – 600 mm. In both studies, a strong decrease in CO2 uptake was found in the weeks following the last rainfall of the wet season, and decreasing soil water availability was identified as the main controlling parameter of CO2 exchange. However, to our
G02030
1 of 11
G02030
¨ MMER ET AL.: CO2 AND ENERGY EXCHANGE IN SAVANNA BRU
knowledge all carbon flux studies in West Africa available up to now cover only periods of some weeks, making a reliable calculation of annual NEE, which is crucial for understanding the role of savannas in global carbon fluxes, impossible. [4] To overcome the lack of long-term measurements in savanna ecosystems covering periods of more than a year, the objective of the present work was to quantify the NEE as well as latent and sensible heat flux in a shrub savanna and its daily, seasonal, and interannual variability related to rainfall distribution over the time span of two consecutive years. The specific questions this work asks are the following: (1) Are the magnitude and amplitudes of daily and seasonal fluxes of CO2 and latent and sensible heat comparable to other savanna sites as well as to other ecosystems? (2) To what extent does the amount and temporal distribution of precipitation drive the annual carbon balance and turbulent flux partitioning? (3) Which other factors control the magnitude and the diurnal and seasonal courses of CO2 fluxes?
2. Materials and Methods 2.1. Site Characteristics [5] The research site (10°5105600N, 3°402200W, 293 m above sea level) was located in a nature reserve in the southwest of Burkina Faso (Ioba province, 280 km from the capital Ouagadougou), bearing the name of a bordering small village (Bontioli). The study area corresponded to the southern Sudanian savanna type (800– 1000 mm annual rainfall); the long-term average annual rainfall was 926 mm in the period 1960– 1999 (FAOClim-NET, Climpag: Climate Impact on Agriculture, Environment, Climate Change and Bioenergy Division, Food and Agriculture Organization, 2007, available at http://tecproda01. fao.org/climpag), with over 75% of the precipitation occurring from June to September. The nature reserve consisted mainly of shrub-dominated savanna, with some treedominated areas. The measurements were conducted in an area mainly dominated by grasses and shrubs. The maximum north-south and west-east extension of the reserve was 15 and 18 km, respectively. The position of the instrumentation was chosen to allow for the required fetch and, at the same time, to be as representative of the whole nature reserve as possible. During the rainy season Andropogon gayanus and Loudetiopsis kerstingii were the dominating grass species in the study area. In late October of every year the dry grass layer was removed by controlled fires arranged by the local environmental authority. After burning, the vegetation consisted of nearly equal portions of the shrub and tree species Vitellaria paradoxa, Detarium microcarpum, Entada africana, Terminalia laxiflora, Combretum glutinosum, Acacia dudgeoni, Combretum collinum, and Lannea microcarpa throughout the dry season. All trees were found only at distances greater than 100 m from the tower and not in the two main wind directions. During the peak of the rainy season, the grass layer reached 0.7 m in height, whereas the height of the shrubs (1 –2 m) and trees (maximum 12 m) remained nearly constant during the observation period. Leaf area index (LAI) values were measured along transects in the immediate vicinity of the eddy covariance (EC) tower location in late September 2005 and were highly variable (1.1 –
G02030
5.7), with a mean value of 3.2. Mean LAI for the grassdominated spots was 4.5. [6] The soil was classified as a sandy loam. It consisted of one homogeneous layer of approximately 40 cm thickness above laterite rocks with no development of soil horizons and with a stone fraction >2 mm of 35%. The pH value (in 0.01 M CaCl2) was 4.9 ± 0.2. Soil organic carbon content was 0.56 ± 0.09%, and the C:N ratio was 11.08 ± 0.15. 2.2. Meteorological Measurements [7] Air temperature and relative humidity were measured with a HMP45C probe (Vaisala, Helsinki, Finland). Net radiation (Rn) was determined using a NR-Lite net radiometer as well as a CNR1 net radiometer (Kipp & Zonen, Delft, Netherlands) for the determination of incoming and outgoing solar and far-infrared radiation. Photosynthetically active radiation (PAR) was measured with a PAR-Lite sensor (Kipp & Zonen). Rainfall was recorded with an ARG 100 rain gauge (Campbell Scientific, Lougborough, United Kingdom). All parameters were determined at a height of 2.0 m, except for rainfall that was recorded 1.0 m above ground. [8] The instrumental setup for soil measurements consisted of a heat flux plate (HFT-3, Campbell Scientific, Logan, Utah, United States) in 0.08 m depth, three soil temperature probes (107 Thermistor, Campbell Scientific, Lougborough, United Kingdom) in 0.02, 0.1, and 0.3 m depth, respectively, and a theta probe soil moisture sensor (Delta-T, Cambridge, United Kingdom) in 0.2 m depth. Hourly values of volumetric water content (VWC) (percent) were converted into water-filled pore space (WFPS) (percent) using the following equation: WFPS ¼
VWC ; BD 1 2:65
ð1Þ
where BD (g cm3) is the bulk density of the soil, and 2.65 (g cm3) is the density of quartz. 2.3. Measurement, Data Processing, and Postfield Correction of Carbon Dioxide and Water Vapor Fluxes [9] The EC technique was used to determine fluxes of carbon dioxide (FC), latent heat (lE), and sensible heat (H) between biosphere and atmosphere. Principles and characteristics of this measurement technique were described and discussed in detail, e.g., by Baldocchi et al. [1996], Aubinet et al. [2000], Massmann and Lee [2002], Finnigan et al. [2003], and Foken [2003]. [10] The data required for the calculation of sensible heat, latent heat, and CO2 fluxes (i.e., CO2 and H2O mixing ratios and air temperature and pressure, as well as the threedimensional wind speeds) were measured with a threedimensional sonic anemometer (CSAT3, Campbell Scientific Instruments, Logan, Utah, United States) in conjunction with an open-path infrared CO2/H2O gas analyzer (Li-7500, Li-Cor, Inc., Lincoln, Nebraska, United States) 2.65 m above ground. The data were recorded with a data logger (CR5000, Campbell Scientific Instruments, Logan, Utah, United States) at a frequency of 20 Hz. In addition, the logger was programmed to calculate and store the turbulent fluxes as half-hourly averages, which were used for com-
2 of 11
G02030
¨ MMER ET AL.: CO2 AND ENERGY EXCHANGE IN SAVANNA BRU
parison with the final flux values obtained after application of the postfield data processing and correction. Measurements in the Bontioli nature reserve were conducted from 1 November 2004, at the onset of the West African dry season, until 8 December 2006. Zero and span calibration of the infrared analyzer was performed every 2 months. The instrument drift was always below 3% within this period. [11] In order to assure an accurate as possible calculation of turbulent fluxes using the EC technique, an application of correction procedures, considering site characteristics and instrument-specific configurations as well as atmospheric conditions, is necessary [Webb et al., 1980; Foken and Wichura, 1996; Aubinet et al., 2000; Baldocchi et al., 2001; Falge et al., 2001a, 2001b]. First, the data set was tested for stationarity of both mean values and variance of 5-min periods of each data channel. Then, spikes were removed by applying a user-defined filter for each data channel. Running mean vertical and horizontal velocity components were transformed to zero by performing a two-dimensional coordinate system rotation. Flux calculations included the friction velocity, the Monin-Obukhov stability parameter, a frequency response correction, which accounts for the inability of the measurement system to capture very high or low frequency fluctuations in the signals, as well as Webb-Pearman-Leuning (WPL) density corrections [Webb et al., 1980]. After application of the correction procedures, half-hourly values of sensible heat, latent heat, and CO2 fluxes were calculated, which were compared afterward with the quality flag records of the data logger. Unusable values because of rainfall events or dust deposition on the glass aperture of the open-path sensor or insufficient nighttime turbulence (u* < 0.1) were eliminated. Smaller data gaps were filled using the mean diurnal variation (MDV) (14 days) method [Falge et al., 2001a]. A large data gap from January to April 2005 was filled by linear interpolation between the half-hourly values of the same time of the days closest to the gap. This was possible without large error, as no significant photosynthetic and respiratory activity was prevalent in the period. As a last step, an additional heat flux term was included in the WPL correction of the open-path gas analyzer data because of measurement errors due to self-heating of the sensor [Burba et al., 2006; Grelle and Burba, 2007]. The average percentage of gaps in the flux data for the two experimental years was 37.4%, which was mainly caused by power failures and dust deposition on the sensor from January to April 2005. [12] NEE was estimated by cumulating all CO2 fluxes and adding the storage term by using hourly changes in CO2 concentration assuming a well-mixed nighttime boundary layer with a height of 100 m, which corresponds to the estimation of Octavio et al. [1987]. Uncertainty estimates for the total annual NEE values equal the standard errors (n = 17,520). A specific leaf area (SLA) value of 13.8 m2 kg1 for the annually burned Andropogon grass layer was taken from Wray and Strain [1987]. Given the LAI of 4.5 m2 m2 at the grass-dominated spots and a C content of A. gayanus of 45.7% [Diallo et al., 2006], the C loss with the annual burn was estimated to be 149 g C m2. This value was included in the monthly (for October) and annual NEE estimates. Components of NEE, i.e., gross ecosystem productivity (GEP) and ecosystem respiration (R), were deter-
G02030
mined by applying the following Q10 model [e.g., Stoy et al., 2006]: ðTa Tref Þ=10 ; R ¼ q1 q2
ð2Þ
where q1 and q2 are model parameters describing the base R at a reference temperature (Tref), here 20°C, and the exponential temperature response of R, respectively. Here q1 and q2 were determined for the whole data set; air temperature (Ta) was used as the driving variable. GEP was calculated by subtracting R from NEE. CO2 flux values were expressed as mg CO2 m2 s1, all daily, monthly, seasonal, and annual values of NEE as well as GEP and R were expressed as g C m2 (time unit)1. Data processing, statistics, linear and polynomial regression analyses for measured fluxes, and meteorological parameters were performed with Origin 7.0 (OriginLab Corporation, Northampton, Massachusetts, United States), SigmaPlot 2000, and SPSS 8.0 statistical analysis software (SPSS Inc., Chicago, Illinois, United States).
3. Results 3.1. Meteorological Conditions [13] Monthly average air temperatures ranged from minima of 24.9°C and 24.7°C in August 2005 and September 2006, respectively, to maximum values of 32.1°C in May 2005 and 31.9°C in May 2006 (Figure 1a). Relative air humidity was lowest in January and February 2005 and 2006 (values between 14.2 and 18.6%), whereas highest values of over 80% were reached in the months of July – September in both years. [14] The precipitation pattern at the study site showed the distinct wet and dry seasons typical for the tropics. From April to September 2005, 93 rainy days were recorded with a maximum monthly sum of 245 mm in September 2005, while only 60 mm of rainfall fell from November 2004 to March 2005 on only 4 days (Figure 1b). A similar rainfall distribution was observed in the second year. Compared to the long-term precipitation record for this region, November 2004 was an uncommonly wet month with 59.6 mm of rain in total, of which 58.2 mm were measured during a severe thunderstorm on 2 November 2004. While in April 2005 a higher amount of precipitation was recorded than in May 2005, a delayed onset of the rainy season was observed in 2006. Total annual rainfall in 2005 and 2006 was 785 and 919 mm, respectively. [15] In April of both years, soil moisture increased over a short period of time from an entirely dry soil with 0– 76% WFPS after the first rainfall at the end of the dry season, and the soil dried up again until the beginning of May (Figure 1c). With the onset of the rainy season in the middle of May, soil moisture reached values constantly above 33% WFPS, with periods of nearly water-saturated conditions (92% WFPS) from July to the beginning of October. Drying of soil started in October of both years, reaching a WFPS close to zero in the middle of December. [16] Average daily Rn during the whole investigated period was 101 W m2 (Figure 1c). Minimum daily average Rn of 46 W m2 was observed in December 2004, while
3 of 11
G02030
¨ MMER ET AL.: CO2 AND ENERGY EXCHANGE IN SAVANNA BRU
Figure 1. (a) Air temperature (n = 632– 744, plus/minus SE) and relative humidity (n = 632– 744, plus/minus SE), (b) monthly total measured and average long-term (1961 – 1990) precipitation as well as number of rainy days, and (c) hourly values of soil moisture and average monthly net radiation (n = 28– 31, plus/minus SE) from 1 November 2004 to 31 October 2006 in the Bontioli reserve, Burkina Faso. maximum daily average values based on hourly resolution were measured in the rainy season with a maximum of 152 W m2 in September 2005. 3.2. Diurnal and Seasonal Variation in CO2 Exchange [17] In the dry season (November – April) the CO2 exchange was close to zero most of the time (Figure 2b). With more frequent rainfall events in 2005 and 2006, maximum uptake rates continuously increased to values around 1.5 mg CO2 m2 s1 from the beginning of June to the beginning of July and remained nearly constant until the beginning of September. In this period of the highest CO2 exchange, nighttime CO2 release rates of about 0.8 mg CO2 m2 s1 were observed. Daytime CO2 uptake began to decrease at the beginning of September in both years, while nighttime CO2 release remained approximately at the same level for 2 months longer (Figure 2b). In 2006, the period of net CO2 uptake lasted until the end of October and, thus, approximately 3 weeks longer than in 2005. [18] The mean diurnal CO2 fluxes and cumulative daily NEE are shown in Figure 3 for each single month of the rainy season and lumped together for the months of the dry season (November – May). The data of the dry period from November to May were combined because no significant
G02030
differences in CO2 exchange between the single months were observed within this period, when CO2 flux rates were generally very low and did not exceed maximum uptake rates at 1200 LT of 0.05 mg CO2 m2 s1 and maximum release rates of 0.05 mg CO2 m2 s1 during nighttime. In contrast, clear diurnal patterns of CO2 fluxes were found during the rainy season. Highest amplitudes in both investigated years were observed in July and August. While mean uptake rates reached values up to 0.95 mg CO2 m2 s1 at 1200 LT, CO2 release of 0.2 mg CO2 m2 s1 occurred during nighttime. June and September were the transition months before and after the peak of the rainy season. Flux rates in October declined nearly to the level of the dry season months (Figure 3). Differences in mean diurnal flux rates between the two years were small except for June and October (Figure 3). Midday uptake rates were approximately 0.45 mg CO2 m2 s1 in June 2005 compared to 0.6 mg CO2 m2 s1 in June 2006. In October 2005 midday uptake rates were around 0.1 mg CO2 m2 s1 and on average 0.1 mg CO2 m2 s1 lower than in October 2006. [19] Mean daily NEE was approx. 0.4 g C m2 d1 (November 2004 to May 2005) and 0.2 g C m2 d1 (November 2005 to May 2006) (Figure 3). From June to September, however, total daily mean uptake was always higher in the second year. While only minor differences between 2005 and 2006 of less than 1.4 g C m2 d1 were determined in July and August, when ecosystem C uptake was always between 4.5 and 5.9 g C m2 d1, a 2.0 g C m2 d1 higher C uptake was observed in June 2006 (3.2 g C m2 d1) compared to June 2005 (1.2 g C m2 d1), and C uptake was 2.2 g C m2 d1 higher in September 2006 (5.7 g C m2 d1) than in September 2005 (3.5 g C m2 d1). 3.3. Monthly and Annual Carbon Balance [20] Average monthly release of CO2 to the atmosphere was 10 g C m2 in the first dry season (November 2004 to May 2005, Figure 4a). In 2005, net C uptake on a monthly basis was observed only during June – September, with a maximum value of 156 g C m2 in August 2005. The total carbon uptake of the ecosystem during the first investigated year (1 November 2004 to 31 October 2005) was 179 g C m2 (Figure 4b). This value included the estimated carbon loss due to the controlled burning in the last week of October 2005 (149 g C m2). [21] In the second year, however, a considerably higher ecosystem CO2 uptake of 429 g C m2, again including the C loss during the annual fire, was determined (Figure 4b). While only minor variations were found in the dry period (total exchange of 7 g C m2), highly significant differences were found for the rainy season. CO2 uptake in June 2006 was 2.8-fold higher than in June 2005 (Figure 4a) and was on average 46% higher from June to September 2006 than from June to September 2005. [22] Partitioning of NEE into GEP and R is shown in Figure 4c. The C loss of 149 g C m2 due to the annual fire was included in the October values. Ecosystem respiration in both years was highest in June (116 g C m2 in 2005 and 128 g C m2 in 2006) and decreased to values between 70 and 80 g C m2 in the course of the growing season from July to September, when soils were nearly water saturated. Dry season respiration was always between 1 g C m2
4 of 11
G02030
¨ MMER ET AL.: CO2 AND ENERGY EXCHANGE IN SAVANNA BRU
Figure 2. (a) Hourly means of air temperature and daily sums of rainfall, (b) half-hourly values of CO2 flux, (c) latent heat flux, and (d) sensible heat flux in the Bontioli nature reserve.
Figure 3. Mean diurnal CO2 fluxes and cumulative daily NEE in the dry seasons 2004/2005 and 2005/ 2006 (November– May) and for each single month in the rainy seasons 2005 and 2006 (June – October). Error bars indicate SE of the mean of 30, 31 or 212 half-hourly flux values. Negative values indicate CO2 uptake from the atmosphere, and positive values indicate CO2 loss to the atmosphere. 5 of 11
G02030
G02030
¨ MMER ET AL.: CO2 AND ENERGY EXCHANGE IN SAVANNA BRU
G02030
of 1000 mmol m2 s1. The parameter y0 (Table 1) represents the CO2 uptake saturation threshold, which was lowest for July (0.91) and August (1.17). In contrast, no significant relationship between the independent variable PAR and the dependent variable CO2 flux could be found for the dry season months November – May as well as for October.
Figure 4. Carbon dioxide biosphere-atmosphere exchange in the Bontioli nature reserve from November 2004 to October 2006. (a) Monthly mean net ecosystem exchange (NEE) (plus/minus SE), (b) annual course of cumulative NEE for the two investigated years, and (c) monthly NEE, gross ecosystem productivity (GEP), and ecosystem respiration (R). (February and March 2005) and 34 g C m2 (February 2006) except for April and May (85 to 102 g C m2), when the first occasional heavy rain events were observed. While only minor differences in R between 2005 and 2006 were observed in the period June– September, GEP was 27% higher in 2006 than in 2005. 3.4. Effect of Photosynthetically Active Radiation on CO2 Exchange [23] Figure 5 displays the fitting of the dependency of CO2 fluxes on photosynthetically active radiation (mmol photons m2 s1) with a first-order exponential decay model in the form of CO2 flux (mg CO2 m2 s1) = A1exp(PAR/t1) + y0 (for parameters see Table 1). Data sets of both investigated years were included in the regression analysis. A comparatively good match between measured data and model output was obtained for the months with the highest net CO2 exchange, i.e., July (R2 = 0.80) and August (R2 = 0.78). Highest ecosystem CO2 uptake of more than 0.8 mg CO2 m2 s1 occurred at PAR values in excess
3.5. Latent and Sensible Heat Fluxes and Energy Balance Closure [24] Monthly mean diurnal courses of latent and sensible heat flux as well as net radiation from November 2004 to October 2006 are shown in Figure 6a. In the dry season (November– May) turbulent energy exchange was dominated by sensible heat flux, with peak daytime values in excess of 300 W m2 at 1200 LT and nighttime values down to 50 W m2. Peak latent heat flux values in the dry season were as low as 25 W m2 at 1200 LT and were close to zero during nighttime. In the rainy season (June – September), the relation of sensible to latent heat flux was reversed. Although net radiation was higher in the rainy season than in the dry season, only turbulent energy exchange was dominated by latent heat flux in the rainy season not total energy exchange. Peak values of latent heat flux were between 300 and 450 W m2. From June to August of both years, the sensible heat flux followed a similar pattern like the average latent heat flux in the dry season, with mean peak values of approximately 100 W m2 at 1200 LT. In October of both years, latent and sensible heat flux reached a similar level of approximately 250– 300 W m2. [25] Monthly average midday Bowen ratios and the monthly average evaporative fraction lE/(lE + H) are displayed in Figure 6b. Bowen ratios in the wet season months were always below 0.7 and were around 1 in October 2005 and 2006. In contrast, Bowen ratios were always higher than 1.5 in the dry season months and peaked in March 2006 (27.4). An opposite pattern was observed for the evaporative fraction, with values never exceeding 0.45 in the dry season months and always being higher than 0.6 in the rainy season. [26] Energy balance closure (EBC) for the entire investigation period, analyzed with linear regression statistics based on the ideal one-to-one relationship between H + lE and Rn G, where Rn is the net radiation and G is the soil heat flux, is shown in Figure 7. From November 2004 to December 2005 as well as from July to November 2006, month-specific slope values equaled or were higher than 0.75. From January to May 2006, however, turbulent fluxes were approximately 30 – 35% lower than the available energy. No seasonal trend was found in the EBC, while a diurnal variation based on the entire data set could be demonstrated (Figure 7b). The energy balance was closed within a range of 100 and +100 W m2 during nighttime in the majority of the cases but varied between 500 W m2 and +500 W m2 during daytime, with highest values between 1200 and 1500 LT. A histogram of classified EBC data (Figure 7c) revealed that high positive values in excess of 100 W m2 appeared more frequently around 1200 LT, meaning that the flux of available energy was higher than the turbulent energy flux. In the late afternoon and early evening hours, however, turbulent energy flux
6 of 11
¨ MMER ET AL.: CO2 AND ENERGY EXCHANGE IN SAVANNA BRU
G02030
G02030
Figure 5. Dependency of half-hourly mean carbon dioxide fluxes on photosynthetically active radiation (PAR) and vapor pressure deficit during the dry season (November –May), in the rainy season (June – September), and in October. The data set includes all data from November 2004 to October 2006. The curve fittings were done with a first-order exponential decay model in the form of CO2 flux (mg CO2 m2 s1) = A1exp(PAR/t1) + y0 (for parameters see Table 1).
was frequently higher than the available energy flux by more than 100 W m2.
4. Discussion 4.1. Daily and Seasonal NEE [27] The savanna site investigated in this study showed clear diurnal and seasonal patterns of net CO2 exchange between the biosphere and the atmosphere. During the dry months November – May, low net ecosystem CO2 release rates (Figure 3 and 4) were found. With the first more frequent rain events in June, the net ecosystem carbon source turned into a carbon sink. C uptake increased until August to mean peak values of 0.95 mg CO2 m2 s1 at midday, associated with a rapid plant growth in this period. A similar seasonal variability in CO2 exchange as well as the occurrence of maximum C uptake rates 2 – 3 months after the onset of regular rainfall in regions with a pronounced dry season were reported by several authors [e.g., Hanan et al., 1996; Veenendaal et al., 2004; Hastings et al., 2005]. The peak values of CO2 uptake at other seasonally dry sites, however, were in most cases lower than those in the present study. Verhoef et al. [1996] found an ecosystem C sink of 0.44 mg CO2 m2 s1 at a fallow savanna site in Niger in the Sudano-Sahelian savanna zone, consisting of scattered shrubs separated by herbaceous understory vegetation in the wet season, while C uptake rates of 0.22 mg CO2 m2 s1 were observed during the transition period
from wet to dry season. The same peak C uptake rates of 0.44 mg CO2 m2 s1 were reported by Veenendaal et al. [2004] for a Mopane woodland in South Africa. Hanan et al. [1998] quantified a maximum sink strength of 0.66 mg CO2 m2 s1 in another Sudano-Sahelian savanna in Niger. At all sites the annual rainfall was significantly lower (300– 600 mm) than at our site. Therefore, the significantly higher net C uptake in the Bontioli nature reserve compared to the other savanna sites could be explained by the greater water availability, reflected in the very dense grass layer with much higher LAI values leading to higher C uptake rates on an area basis. LAI values in Bontioli varied between 1.1 and 5.7 with a mean value of 3.2 in September 2005, whereas LAI values reported by Veenendaal et al. [2004] in the Mopane woodland reached only values up to 1.3 during maximum canopy development. Verhoef et al. [1996] reported values from 0.4 to 1.35 for their Sudano-Sahelian Table 1. Parameters of the Exponential Decay Model in Figure 5a
R2 y0 A1 t1
November – May
June
July
August
September
October
0.06 0.07 0.11 3305.02
0.66 0.50 0.65 524.41
0.80 0.91 1.07 543.06
0.78 1.17 1.25 781.26
0.54 0.80 0.83 1090.12
0.10 0.08 0.17 1177.58
a Following the equation CO2 flux (mg CO2 m2 s1) = A1exp(PAR/t1) + y0, where PAR is photosynthetically active radiation (mmol photons m2 s1).
7 of 11
G02030
¨ MMER ET AL.: CO2 AND ENERGY EXCHANGE IN SAVANNA BRU
G02030
GEP, was on a similar level in June and September of each year, although NEE was 3 times lower in June 2005 and 2 times lower in June 2006 than in September of each respective year. NEE values in June of both years were comparatively low because of the relatively high soil respiration values in this period (beginning of the rainy season), whereas high soil moisture inhibited R from July to September favoring higher C sequestration in this period. [30] Verhoef et al. [1996] showed that plant stress induced by unfavorable atmospheric and soil conditions (vapor pressure deficit above 30 hPa and low soil water content) caused NEE to peak during the early morning hours. In our study, NEE peaked at 1200 LT and was therefore fairly in phase with PAR, indicating that the majority of the photosynthetically active vegetation at our site was not drought stressed during our measurements. A light saturation of CO2 flux rates at PAR values in excess of 1000 mmol m2 s1
Figure 6. (a) Monthly means of diurnal courses of latent (lE) and sensible (H) heat flux as well as net radiation (Rn) from November 2004 to October 2006 and (b) monthly average midday Bowen ratios and monthly average evaporative fractions lE/(lE + H). savanna sites in Niger, whereas Scanlon and Albertson [2004] measured an LAI value of 1.67 at their wettest site in Mongu, Zambia. At the latter site, just slightly lower maximum CO2 uptake rates of 0.88 mg CO2 m2 s1 were found during a Kalahari transect study. This site received an annual rainfall of 879 mm and is, therefore, more comparable to our site in regard to water availability than to the other sites. [28] The mean peak C uptake of 0.95 mg CO2 m2 s1 of the savanna ecosystem investigated in the present study has a medium position in the wide range of C uptake by ecosystems worldwide, with values, for example, from 0.1 mg CO2 m2 s1 for a boreal spruce forest [Fan et al., 1995], 0.7 mg CO2 m2 s1 in a moist tropical rain forest in Brazil [Lloyd et al., 1995], 0.9 mg CO2 m2 s1 in an old-growth Douglas fir forest [Paw U et al., 2004], 1.1 mg CO2 m2 s1 in a temperate grassland [Verma et al., 1992], 1.4 mg CO2 m2 s1 in a boreal aspen forest [Black et al., 1996], and 0.3 to 1.7 mg CO2 m2 s1 in deciduous temperate forests [Wofsy et al., 1993] up to 2.0 –3.0 mg CO2 m2 s1 for closed wheat crops under favorable conditions [Baldocchi, 1994]. [29] An immediate response of soil respiration to rainfall and soil moisture changes during the transition period between dry and rainy season was revealed by partitioning NEE into GEP and R using the Q10 model. The partitioning demonstrated that photosynthetic activity, expressed as
Figure 7. (a) Total energy balance closure (EBC) as determined by 30-min means of turbulent energy (latent heat plus sensible heat) and available energy (net radiation minus soil heat flux) for the period November 2004 to October 2006, (b) diurnal variation of the total EBC, and (c) histogram of classified EBC data.
8 of 11
G02030
¨ MMER ET AL.: CO2 AND ENERGY EXCHANGE IN SAVANNA BRU
could be demonstrated in this study for the rainy season (Figure 5). While a similar threshold was observed by Paw U et al. [2004] for an old-growth Douglas fir forest, Baldocchi [1994] found a linear relationship between canopy CO2 exchange and PAR for the entire range of values between 0 and 1750 mmol m2 s1 for well-watered crops with relatively closed canopies. A CO2 flux compensation point, e.g., at 50 mmol m2 s1 as reported by Ruimy et al. [1995] for several closed canopy forests, could not be found in this study. A net CO2 flux close to zero appeared during nighttime as well as during daytime when the vegetation cover was only sparsely developed (May and beginning of June) or already drying out (October), associated with relatively high respiration rates. In the transition months May/June and October, when soil moisture had not fully reached the values of the rainy season or the soil was already drying up, the response of CO2 uptake to radiation was less pronounced. A similar relationship was found in a grassland study by Kim and Verma [1990], who suggested that water limitation could have caused low leaf water potentials, which may have limited CO2 exchange by reducing the turgor of the stomatal guard cells and by affecting enzyme reactions associated with photosynthesis. 4.2. Interannual Variability in NEE and Annual Carbon Sequestration [31] The Bontioli nature reserve was a net carbon sink of 179 ± 98 g C m2 in the first year of our investigations (November 2004 to October 2005) and 429 ± 100 g C m2 in the second year (November 2005 to October 2006). These values included estimates of the C loss during the annual fire (149 g C m2), based on SLA data in the literature, which could not be reliably quantified with our instrumental setup. [32] The ecosystem was more or less carbon neutral in the dry season months from November to March. In this period no ground vegetation was present; therefore, only the scattered shrubs in close vicinity to the tower were capable of photosynthesis. On the other hand, soil respiration was also low because of the very low soil moisture, leading to a very low net CO2 exchange. In April and May, however, markedly higher CO2 release rates were found, which could be explained by a few single rain events leading to an increase in soil respiration triggered by higher soil water content. This increase in soil respiration exceeded the CO2 uptake by the sparsely developed grass layer at that time. From June to September of both years, the period with the highest net C uptake was observed. However, CO2 net uptake rates during this period were much higher in the second year than in the first year. Especially in June 2006 a higher sink activity than in June 2005 was found (96 g C m2 compared to 34 g C m2). A possible explanation could be that the sum of precipitation in May 2006 (78 mm) was higher than in May 2005 (49 mm); hence, in June 2006 the development of the grass layer was in a more advanced state with higher photosynthetic capacity than in June 2005. GEP results derived from the Q10 model confirm this assumption by revealing much higher photosynthetic activity in June 2006 (225 g C m2) than in June 2005 (150 g C m2). NEE in September 2006 (171 g C m2) was also substantially higher than in September 2005 (104 g C m2). It can be assumed that the extremely dry August in 2005 (around
G02030
100 mm less rain than the long-term average) led to an early dry up of the vegetation, which was reflected in both lower NEE and GEP values in September 2005 than in September 2006. [33] A further significant difference in NEE was found in October of both years. While in October 2005, 30 g C m2 were naturally released to the atmosphere, i.e., by ecosystem respiration only without C loss due to the annual fire, the ecosystem was almost carbon neutral (3 g C m2) in October 2006. A reason for these differences in CO2 flux could be the difference in rainfall. In October 2005, only 25 mm were measured on 4 rainy days causing an early dry up of the vegetation followed by a cessation of photosynthetic activity. Although at the same time the soil water content started to decrease, it was still high enough to sustain soil respiration exceeding the C uptake by the plants. In contrast, 75 mm of rain precipitated on 15 rainy days in October 2006, leading to a longer photosynthetically active period than in October 2005. [34] Only a few long-term CO2 exchange measurements in seasonally dry ecosystems have been conducted up to date. There is a lack of C flux studies especially in the West African Guinean and Sudanian savannas, making an interpretation and comparison of our data with other annual C budget estimates for this region impossible. Therefore, our findings could only be compared with diurnal and seasonal flux values reported for this ecosystem type. Hastings et al. [2005] also found a clear rainfall dependency in their long-term NEE measurements in a desert shrub community in Mexico, but those values ranged on a much lower level. In two consecutive years, 39 and 52 g C m2 yr1 were fixed by the ecosystem, while the recorded rainfall sums were 147 and 197 mm, respectively. Although their measurements were carried out at a desert site, the response of CO2 exchange to water availability was similar to that observed in our savanna study. Veenendaal et al. [2004], however, estimated a net uptake of 44 g C m2 over the course of 12 months for their semiarid savanna (Mopane woodland), a site receiving on average 464 mm of rain a year. 4.3. Energy Exchange [35] Half-hourly values of latent heat transfer were positively correlated with NEE, soil moisture, and relative humidity throughout the entire investigation period and were negatively correlated with sensible heat flux. Verhoef et al. [1996] reported that the increase and/or decrease in latent heat was mainly caused by the variation in stomatal conductance. In our study, the higher amount of rainfall in April 2005 than in May 2005 was also reflected in latent heat. This suggested that the ecosystem, i.e., the biomass production and therefore the CO2 uptake potential, reacted quite sensitively to water availability especially in the transition period from the dry to the rainy season. Highest daily mean latent heat flux values of 133 W m2 in 2005 and 131 W m2 in 2006 as well as Bowen ratios 0.8 (Figure 6) were associated with the period of the highest biomass production in June with high gross CO2 flux rates (highest soil respiration and relatively high peak CO2 uptake rates at 1200 LT, see section 4.1) in contrast to July and August when the highest CO2 net uptake rates were found.
9 of 11
G02030
¨ MMER ET AL.: CO2 AND ENERGY EXCHANGE IN SAVANNA BRU
[36] The energy balance closure observed in this study was in a typical range for micrometeorological measurements of this type in most of the investigated months. Gaps in the EBC can be induced by differences in the spatial scales of the footprint of the eddy covariance sensors, the net radiometer, and the soil heat flux plates as well as by the short (half-hourly) averaging period. Large gaps in the EBC are normally associated with an underestimation of the CO2 uptake by the ecosystem [Aubinet et al., 2000; Wilson et al., 2002]. Here, the highest number of gaps occurred between January and May 2006, when the CO2 uptake potential was very low, and therefore, the impact of a poor EBC on the associated uncertainties in CO2 flux calculations was low. [37] The diurnal course of the EBC (Figures 7b and 7c) showed that the available energy flux exceeded the turbulent energy flux at midday, whereas values below 100 W m2 were found in the late afternoon hours. During the entire investigation period, only 5% of the EBC values were below 100 W m2, whereas 6% were above +100 W m2, indicating a partial imbalance and an underestimation/overestimation of the turbulent energy flux, which could have been caused by heat storage in vegetation, with vegetation storing energy during morning and midday hours and releasing it in the afternoon.
5. Conclusions [38] To our knowledge, the present study provides the first long-term data set of EC measurements of carbon dioxide and energy exchange in a savanna in sub-Saharan West Africa (southern Sudanian savanna). Although each year considerable quantities of C were released again during burning of the senescent vegetation at the beginning of the dry season, the ecosystem acted as a remarkable C sink, however, with significant year-to-year variability. An important finding was that especially during the transition months between dry and wet season (April – June) as well as between wet and dry season (October) the ecosystematmosphere CO2 flux responded immediately to changes in water availability. The first rains at the beginning of the wet season stimulated soil respiration more quickly and intensely than plant growth, leading to a burst in soil CO2 efflux, whereas the drying out after the last rain at the beginning of the dry season led to a quicker cessation of photosynthetic activity than soil respiration. These findings stress the sensitivity of this ecosystem to interannual climate variability and changes in rainfall patterns, not only as a result of natural year-to-year variability but also as a consequence of global climate change. However, to get a better insight into the effect of climate change on C sink/source strengths of the savanna ecosystems, more long-term and additional manipulation studies, also with different savanna and land use types, are required. [39] Acknowledgments. Funding for this research was provided by the Helmholtz Association of German Research Centers (Virtual Institute, VH-VI-001). The authors thank Konrad Vielhauer of the Center for Development Research (ZEF) and Dominik Schmengler of the Dreyer Foundation for infrastructural support during the field campaigns and Gildas Houe´nagnon Boko for technical assistance and for collecting data. We also thank Michael Dannenmann, Christian Werner, Boris Matejek, Ralf Kiese, and Bettina Ketzer for scientific advice as well as Lazare Tia for providing LAI values.
G02030
References Aubinet, M., et al. (2000), Estimates of the annual net carbon and water exchange of forests: The EUROFLUX methodology, Adv. Ecol. Res., 30, 114 – 177. Baldocchi, D. D. (1994), A comparative study of mass and energy exchange over a closed C3 (wheat) and an open C4 (corn) crop: II. CO2 exchange and water use efficiency, Agric. For. Meteorol., 67, 291 – 321, doi:10.1016/0168-1923(94)90008-6. Baldocchi, D. D., R. Valentini, S. R. Running, W. Oechel, and R. Dahlman (1996), Strategies for measuring and modelling CO2 and water vapour fluxes over terrestrial ecosystems, Global Change Biol., 2, 159 – 168, doi:10.1111/j.1365-2486.1996.tb00069.x. Baldocchi, D. D., et al. (2001), FLUXNET: A new tool to study the temporal and spatial variability of ecosystem – scale carbon dioxide, water vapor, and energy flux densities, Bull. Am. Meteorol. Soc., 82, 2415 – 2434, doi:10.1175/1520-0477(2001)0822.3.CO;2. Black, T. A., et al. (1996), Annual cycles of water vapour and carbon dioxide fluxes in and above a boreal aspen forest, Global Change Biol., 2, 219 – 229, doi:10.1111/j.1365-2486.1996.tb00074.x. Burba, G. G., D. J. Anderson, L. Xu, and D. K. McDermitt (2006), Additional term in the Webb-Pearman-Leuning correction due to surface heating from an open-path gas analyzer, Eos Trans. AGU, 87(52), Fall Meet. Suppl., Abstract C12A-03. Chen, J., K. T. Paw U, S. L. Ustin, T. H. Suchanek, B. J. Bond, K. D. Brosofske, and M. Falk (2004), Net ecosystem exchanges of carbon, water, and energy in young and old-growth Douglas-fir forests, Ecosystems, 7, 534 – 544, doi:10.1007/s10021-004-0143-6. Cole, M. M. (1986), The Savannas: Biogeography and Geobotany, 438 pp., Academic, London. Diallo, M. D., R. Duponnois, A. Guisse, S. Sall, J.-L. Chotte, and J. Thioulouse (2006), Biological effects of native and exotic plant residues on plant growth, microbial biomass and N availability under controlled conditions, Eur. J. Soil Biol., 42, 238 – 246, doi:10.1016/j.ejsobi.2005.01.003. Falge, E., et al. (2001a), Gap filling strategies for defensible annual sums of net ecosystem exchange, Agric. For. Meteorol., 107, 43 – 69, doi:10.1016/ S0168-1923(00)00225-2. Falge, E., et al. (2001b), Gap filling strategies for long term energy flux data sets, Agric. For. Meteorol., 107, 71–77, doi:10.1016/S0168-1923(00)00235-5. Falge, E., et al. (2002), Phase and amplitude of ecosystem carbon release and uptake potentials as derived from FLUXNET measurements, Agric. For. Meteorol., 113, 75 – 95, doi:10.1016/S0168-1923(02)00103-X. Fan, S.-M., M. L. Goulden, J. W. Munger, B. C. Daube, P. S. Bakwin, S. C. Wofsy, J. S. Amthor, D. R. Fitzjarrald, K. E. Moore, and T. R. Moore (1995), Environmental controls on the photosynthesis and respiration of a boreal lichen woodland: A growing season of whole-ecosystem exchange measurements by eddy correlation, Oecologia, 102, 443 – 452, doi:10.1007/BF00341356. Finnigan, J. J., R. Clement, Y. Malhi, R. Leuning, and H. A. Cleugh (2003), A re-evaluation of long-term flux measurement techniques part I: Averaging and coordinate rotation, Boundary Layer Meteorol., 107, 1 – 48, doi:10.1023/A:1021554900225. Foken, T. (2003), Angewandte Meteorologie: Mikrometeorologische Methoden, 1st ed., 289 pp., Springer, New York. Foken, T., and B. Wichura (1996), Tools for quality assessment of surfacebased flux measurements, Agric. For. Meteorol., 78, 83 – 105, doi:10.1016/0168-1923(95)02248-1. Goulden, M. L., J. W. Munger, S.-M. 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, doi:10.1126/ science.271.5255.1576. Grelle, A., and G. Burba (2007), Fine-wire thermometer to correct CO2 fluxes by open-path analyzers for artificial density fluctuations, Agric. For. Meteorol., 147, 48 – 57, doi:10.1016/j.agrformet.2007.06.007. Hanan, N. P., J. A. Elbers, P. Kabat, A. J. Dolman, and H. A. R. de Bruin (1996), CO2 flux and photosynthesis of a Sahelian savanna during HAPEX-Sahel, Phys. Chem. Earth, 21(3), 135 – 141, doi:10.1016/S00791946(97)85574-7. Hanan, N. P., P. Kabat, A. J. Dolman, and J. A. Elbers (1998), Photosynthesis and carbon balance of a Sahelian fallow savanna, Global Change Biol., 4, 523 – 538, doi:10.1046/j.1365-2486.1998.t01-1-00126.x. Hastings, S. J., W. C. Oechel, and A. Muhlia-Melo (2005), Diurnal, seasonal and annual variation in the net ecosystem CO2 exchange of a desert shrub community (Sarcocaulescent) in Baja California, Mexico, Global Change Biol., 11, 927 – 939, doi:10.1111/j.1365-2486.2005.00951.x. Kim, J., and S. B. Verma (1990), Carbon dioxide exchange in a temperate grassland ecosystem, Boundary Layer Meteorol., 52, 139 – 149. Lloyd, J., J. Grace, A. C. Miranda, P. Meir, S. C. Wong, H. S. Miranda, I. R. Wright, J. H. C. Gash, and J. McIntyre (1995), A simple calibrated model of Amazon rainforest productivity based on leaf biochemical properties, Plant Cell Environ., 18, 1129 – 1145, doi:10.1111/j.1365-3040.1995.tb00624.x.
10 of 11
G02030
¨ MMER ET AL.: CO2 AND ENERGY EXCHANGE IN SAVANNA BRU
Massman, W. J., and X. Lee (2002), Eddy covariance flux corrections and uncertainties in long-term studies of carbon and energy exchanges, Agric. For. Meteorol., 113, 121 – 144, doi:10.1016/S0168-1923(02)00105-3. Octavio, K. H., A. Arrocha, and E. Sanhueza (1987), Low nocturnal atmospheric dispersion capacity of the Venezuela savannah, Tellus, Ser. B, 39, 286 – 292. Paw U, K. T., et al. (2004), Carbon dioxide exchange between an oldgrowth forest and the atmosphere, Ecosystems, 7, 513 – 524. Ruimy, A., P. G. Jarvis, D. D. Baldocchi, and B. Saugier (1995), CO2 fluxes over plant canopies and solar radiation: A review, Adv. Ecol. Res., 26, 1 – 68. Scanlon, T. M., and J. D. Albertson (2004), Canopy scale measurements of CO2 and water vapor exchange along a precipitation gradient in southern Africa, Global Change Biol., 10, 329 – 341, doi:10.1046/j.1365-2486. 2003.00700.x. Stoy, P. C., G. G. Katul, M. B. S. Siqueira, J.-Y. Juang, K. A. Novick, J. M. Uebelherr, and R. Oren (2006), An evaluation of models for partitioning eddy covariance-measured net ecosystem exchange into photosynthesis and respiration, Agric. For. Meteorol., 141, 2 – 18, doi:10.1016/ j.agrformet.2006.09.001. van Wijk, M. T., and W. Bouten (2002), Simulating daily and half-hourly fluxes of forest carbon dioxide and water vapor exchange with a simple model of light and water use, Ecosystems, 5, 597 – 610, doi:10.1007/ s10021-002-0155-z. Veenendaal, E. M., O. Kolle, and J. Lloyd (2004), Seasonal variation in energy fluxes and carbon dioxide exchange for a broad-leaved semi-arid savanna (Mopane woodland) in southern Africa, Global Change Biol., 10, 318 – 328, doi:10.1111/j.1365-2486.2003.00699.x. Verhoef, A., S. J. Allen, H. A. R. De Bruin, C. M. J. Jacobs, and B. G. Heusinkveld (1996), Fluxes of carbon dioxide and water vapour from a
G02030
Sahelian savanna, Agric. For. Meteorol., 80, 231 – 248, doi:10.1016/ 0168-1923(95)02294-5. Verma, S. B., J. Kim, and R. J. Clement (1992), Momentum, water vapor, and carbon dioxide exchange at a centrally located prairie site during FIFE, J. Geophys. Res., 97, 18,629 – 18,639. Webb, E. K., G. I. Pearman, and R. Leuning (1980), Correction of flux measurements for density effects due to heat and water vapour transfer, Q. J. R. Meteorol. Soc., 106, 85 – 100, doi:10.1002/qj.49710644707. Wilson, K. B., and T. P. Meyers (2001), The spatial variability of energy and carbon dioxide fluxes at the floor of a deciduous forest, Boundary Layer Meteorol., 98, 443 – 473, doi:10.1023/A:1018779506973. Wilson, K. B., et al. (2002), Energy balance closure at FLUXNET sites, Agric. For. Meteorol., 113, 223 – 243, doi:10.1016/S0168-1923(02)00109-0. Wofsy, S. C., M. L. Goulden, J. W. Munger, S.-M. Fan, P. S. Bakwin, B. C. Daube, S. L. Bassow, and F. A. Bazzaz (1993), Net exchange of CO2 in a mid-latitude forest, Science, 260, 1314 – 1317, doi:10.1126/science.260. 5112.1314. Wray, S. M., and B. R. Strain (1987), Interaction of age and competition under CO2 enrichment, Funct. Ecol., 1, 145 – 149, doi:10.2307/2389718.
N. Bru¨ggemann, C. Bru¨mmer, H. Papen, and R. Wassmann, Atmospheric Environmental Research Division, Institute of Meteorology and Climate Research, Forschungszentrum Karlsruhe, Kreuzeckbahnstraße 19, D-82467 Garmisch-Partenkirchen, Germany. (
[email protected]) U. Falk, Department of Ecology and Resource Management, Center for Development Research, University of Bonn, Walter-Flex-Straße 3, D-53113 Bonn, Germany. J. Szarzynski, Office for Outer Space Affairs, German Aerospace Center, UN Campus, Hermann-Ehlers-Straße 10, D-53113 Bonn, Germany.
11 of 11
