Biogenic hydrocarbon emissions from southern African savannas

. Biogenic nonmethane hydrocarbon (NMHC) emissions were investigated at two field sites in the Republic of South Africa that include five important southern African savanna landscapes. Tropical savannas are a globally important biome with a high potential for biogenic emissions but no NMHC emission measurements in these regions or in any part of Africa have been reported. Landscape average hydrocarbon emissions were estimated by characterizing plant species composition and foliar density at each site, identifying and characterizing NMHC emissions of the most abundant plant species, and identifying and characterizing NMHC emissions of plant species with the highest NMHC emission rates. A hand-held portable analyzer proved to be a useful tool for identifying plants with high emission rates. A branch enclosure system, with gas chromatography and flame ionization detector, was used to quantify isoprene and monoterpene emission rates. Emission rates were species-specific and several genera had both high and low emitters. At least some species with high emission rates were identified in most savanna types. High and low emitters were found on both nutrient-rich and nutrient-poor soils. Landscape average emission capacities for the five savanna types range from 0.6 to 9 mg C m '2 h '(cid:127) for isoprene and about 0.05 to 3 mg C m -2 h -(cid:127) for monoterpenes. The savanna emission rates predicted by existing global models are within the range estimated for these five savanna types.


Introduction
Nonmethane hydrocarbons (NMHCs) are released into the atmosphere from surface sources and influence regional photochemical oxidant formation and acid deposition. Guenther et al. [ 1995] estimate that vegetation is the source of over 90% of all NMHCs in the global atmosphere. Other sources include organic matter decomposition [Zimmerman, 1979], oceans [Plass-Duelmer et al., 1993], biomass burning [Greenberg et al., 1984], fossil fuel burning, solvent use, and chemical manufacturing [Middleton et al., 1990]. Improved estimates of NMHC emissions from vegetation are needed to accurately initialize global chemistry and transport models. Current estimates rely on a very limited database of NMHC emission rate measurements. Existing data demonstrate that there are large variations in emission rates for different plant species and thus for various landscapes. Most field investigations of natural NMHC emissions have been in North America and Europe [Guenther et al., 1995]. Measurements have also been reported for Asia [Yokouchi and Ambe, 1984], South America [Zimmerman et al., 1988], and Australia [Ayers and Gillett, 1988]  Guenther et al. [1995] estimated global NMHC emissions using the global ecosystem database developed by Olson [1992]. They found NMHC emissions data reported for 26 of the 57 global ecosystems in the Olson database. Tropical savannas are one of the ecosystems for which no NMHC emissions data were available. The high temperatures and radiation fluxes associated with tropical savannas make these regions a potentially large biogenic NMHC emissions source. Tropical savannas cover 65% of the land surface of Africa, 60% of Australia and 45% of South America [Huntley and Walker, 1982]. It is an important biome because of its potential in terms of grazing areas for livestock and crop production, and also because of the role of savannas in the global carbon cycle. The grazing and agricultural potential makes savannas susceptible to land use change. Climatic change, especially precipitation patterns, can also have a significant impact on these regions. In southern Africa the savanna biome covers 46% of all landscapes [Rutherford and Westfall, 1994]. The biome can be divided into two main types, namely the arid savannas with nutrient-rich soils and the moist, nutrient-poor savannas [Huntley, 1982]. The nutrient-rich savannas are dominated by trees of the genus Acacia, whereas the nutrientpoor sites are dominated by broad leaf species primarily from the families Corsalpinaceae and Combretaceae [Scholes, 1988].
The primary objective of this study was to develop a methodology for characterizing landscape average NMHC emissions and apply the method to a landscape where biogenic NMHC emissions have not been studied. Many of the landscapes where biogenic NMHC emission measurements are lacking are in regions where it is difficult and expensive to conduct field experiments. Thus, it is necessary to have emission characterization methods that require a minimum of 25,859 field time and equipment. The four majoi' components of the emission characterization method described in this paper include (1) characterizing plant species composition and foliar density at each site, (2) identifying and characterizing NMHC emissions of the most abundant plant species, (3) identifying and characterizing NMHC emissions of plant species with the highest NMHC emission rates, and (4) estimating landscape average NMHC emission rates. The emission characterization method is applied to two field sites in southern Africa that comprise five different tropical savanna landscapes.

Field Site Description
Biogenic hydrocarbon emissions were investigated at two locations in the Republic of South Africa (RSA). Field experiments were conducted in December 1992 at Ntoma in eastern RSA (latitude 24.1 ø S, longitude 31.3 ø E) and Nylsvley (latitude 24o39 ' S, longitude 28o42 ' E).

The Ntoma site is located within the Klaserie Game
Reserve that covers an area of about 500 km 2. The total land surface in this region consists of about 3 0% Colophosperrnurn rnopane savanna, 50% Combreturn apiculaturn savanna, and 20% Acacia nigrescens savanna. Each of these landscapes has been described in detail by Scholes [1988]. Colophosperrnurn rnopane savanna has a tree foliar density of about 80 g rn '2 consisting of about 72% Colophosperrnurn rnopane, 16% C apiculaturn, and 6 % Grewia bicolor. Average tree foliar densities of about 76 g m '2 in C. apiculaturn savannas are 53% C apiculaturn, 17% Sclerocarya birrea, and 13% G. bicolor. Acacia nigrescens savanna has a tree foliar density of about 66 g rff 2 consisting of about 64% A. nigrescens, 11% S. birrea, and 10% G. bicolor. In each of these landscapes, three to four tree species contribute over 80% of the total tree foliar density. The foliar density of grasses in these savanna regions ranges from 30 to 200 g m -2 with values of 100 to 150 g m '2 typically observed.

Methods
Isoprene and monoterpene emission rates were estimated using two field-portable enclosure measurement systems. The first system, referred to here as the hand-held emissions screening (HES)system, was used to identify plant species with high total NMHC emission rates. The HES system provides an immediate qualitative measure of total NMHC emission rate. The second system, referred to here as the I•ranch emission measurement (BEM) system, was used to quantify isoprene and monoterpene emission rates for plant species that either were dominant in a landscape or were identified as a high emitter by the HES system.
The HES system consists of a small (approximately 0.3 L) polyethylene bag enclosure connected by Teflon tubing to a total NMHC analyzer with a photoionization detector (Thermo Environmental Instruments, model 580B). The entire HES system weighs less than 5 kg, is battery-operated, and can easily be operated by one person. The lower detection limit of the analyzer is NMHC species-specific but is typically around 400 gg rff 3 for terpenoid compounds. A lower detection limit of about 0.5 gg g-• h -• can be expected with about 10 g of foliage in the enclosure and a 200-mL min -• flow rate. Water vapor in the sample air is often very high due to foliar transpiration and can be a significant interference for the HES system if it condenses in the detector. This problem can be minimized by conditioning the sample flow using a water vapor trap or by reducing the pressure or heating the detector.
The BEM system uses an approximately 14-L enclosure that consists of a rigid aluminum frame covered by a flexible Teflon bag. The enclosure is suspended from a tripod to minimize contact with the vegetation. Ambient air is pushed into the enclosure through Teflon tubing from a location selected to minimize background concentrations. Samples were stored in electropolished, stainless steel canisters and transported to Boulder, Colorado, for analysis. Foliar area and dry weight were determined for the enclosed branch so that emissions could be expressed on both a foliar area and dry foliar mass basis. Whole air samples were also collected in stainless steel canisters to characterize background NMHC concentrations in ambient air. [Greenberg et al., 1995]. This system has a lower detection limit of about 5 ppt C.

All ambient air samples and BEM enclosure air samples were analyzed by gas chromatography (GC) with cryogenic preconcentration using a flame ionization detector (HP5890) to quantify concentrations and by comparisons with mass spectrometry to identify compounds
Leaf temperature, enclosure temperature, relative humidity, photosynthetically active radiation (PAR), and general sampling conditions were recorded for each BEM enclosure measurement. A LICOR 6200 with a 4-L chamber was used to measure photosynthesis, transpiration, and stomatal conductance from representative leaves of each tree. These measurements were recorded between 800 LST and 1200 LST.
Xylem pressure potentials were measured just after sunrise at 730 LST with a Scholander pressure chamber. The average midday temperatures of around 36øC are typical of southern African savannas in summer (December). Guenther et al. [1995] have estimated that the global annual NMHC flux from vegetation is about 1 Gg (10 •5 g) of carbon and consists primarily of isoprene (44%) and monoterpenes (11%). Investigations of NMHC emission rates described in this section include NMHC screening of a large variety of plant species and quantitative estimates of isoprene and monoterpene emission rates for important plant species.

Vegetation Screening
Fifty plant species were screened with the HES system. Plants were considered to be high NMHC emitters if qualitative results with the HES system indicated that NMHC emissions exceeded about 0.5 gg C g-• h -•. Fourteen (28%) of the species shown in Table 1 were classified as high emitters.
Trees have been the focus of most NMHC emission studies but shrubs and grasses contribute significantly to the total LAI in most savanna landscapes and should be included in future investigations of biogenic NMHC. The percentage of the total The status of the soil in which they are found in the field is also indicated. NMHC is nonmethane hydrocarbon and HVS is hand-held NMHC screening system.
•hmb and tree. bFertile and infertile. Table 1. Only nine species contribute more than 5% of the LAI of woody species in either of the two areas. Seven of these dominant species were identified as high emitters by the HES system. Negligible NMHC emissions were observed for all of the Combreturn spp., Lannea spp., and Euclea spp. tested in the study areas, but the genera Acacia and Rhus each included both high and low emitters. Acacia caffra, A. karroo, A. nilotica, and Rhus pyroides were identified as low emitters, whereas A. mellifera, A. nigrescens, A. tortilis, and Rhus leptodictya were high emitters. These results indicate that emissions are species-specific and all important species in a genus should be investigated.

LAI contributed by tree and shrub species at each of the two field sites is shown in
High NMHC emissions were observed for 23% of the species from nutrient-rich and 25% from nutrient-poor soils. All of the Acacia species are found on nutrient-rich soils yet three species emit hydrocarbons and three species do not. A. mellifera, A. nigrescens, and A. tortilis on nutrient-rich soils were classified as high emitters, as were Burkea africana and Ochna pulchra which grow on nutrient-poor soils.

Colophospermum toopane, Rhus leptodictya, and Grewia fiavescens are species which can be found on both nutrientrich and nutrient-poor soils and they all emit hydrocarbons. NMHC emissions can thus occur from a species, regardless of the soil nutrient availability. The qualitative HES measurements and the quantitative BEM measurements are compared in Table 2. The HES results indicated that 11 of the 14 species have significant total NMHC emission rates. This result is validated by the BEM measurements. This finding demonstrates that the HES system is a valuable tool for identifying plant species with high
NMHC emission rates.

Emission Rate Measurements
Emission rates for the 14 plant species sampled with the BEM system are shown in Table 2. C. molle, C. apiculatum, and Sclerocarya birrea emitted negligible (< 0.5 gg g-1 hr-1) amounts of isoprene and monoterpenes. Only isoprene was emitted from B. africana and O. pulchra. Five species emitted only monoterpenes including A. tortilis, C. rnopane, G. fiavescens, T. prunoides, and T. sericea. The remaining four species emitted both isoprene and monoterpenes: A. nigrescens, Phragmites mauritianum, R. leptodictya, and Securinega virosa. In order to use these measurements to initialize emission models, the emission capacity at a leaf temperature of 30øC and PAR of 1000 gmol m '2 s -1 was estimated using the temperature and PAR dependent emission algorithms of Guenther et al. [1993]. Guenther et al. [1994] reviewed field measurements of NMHC emissions from North American plant species and considered plants to be high emitters of isoprene if branch-level isoprene emissions were around 40+20 gg g-1 hr-1. Six of the species listed in Table 2 have isoprene emission rates ranging from 32 to 110 gg g-1 hr-1 including four that fall within the range of 40+20 gg g-1 hr-1 and two species with higher rates (81 and 110 gg g-1 hr-1).
Eight different monoterpenes were observed in enclosure air samples including c•-pinene, I•-pinene, camphene, myrcene, carene, cymene, d-limonene, and terpinolene. Only c•-pinene, I•-pinene, carene, myrcene and terpinolene were a major portion of the emission rate from the eight plants with significant monoterpene emission rates.

Guenther et al. [1994] considered North American plant species to be high total monoterpene emitters if emissions
were about 3+1.5 gg C g-1 hr-1. The 14 African savanna plant species included five with emission rates < 0.5 gg C g-1 hr-1, five with rates ranging from 0.5 to 1.3 gg C g-1 hr-1, and three

Landscape Average Emission Rates
Surface emission models are essential for incorporating the impact of emissions into regional and global atmospheric chemistry models. The estimated emission rates can be used to assess potential changes in atmospheric chemical constituents. Early estimates of global biogenic NMHC emission rates used a single emission factor to represent all terrestrial landscapes [Zimmerman, 1979]. Recent efforts have resulted in global models, with high resolution (0.5 ø x 0.5 ø latitude/longitude) that can generate hourly average emission estimates [Guenther et al., 1995]. The accuracy of these estimates in most regions is severely limited by the emissions data used to develop these models. In this section NMHC emissions for southern African savannas are estimated based on the data described above and compared to the estimates of Guenther et al. [1995].
The two field sites comprise a total of fiv•e distinct landscapes: C. mopane savanna (CM), C. apiculatum savanna (CA), A. nigrescens savanna (AN), Burkea africana savanna (BA), and A. tortilis savanna (AT). Landscape average emission capacities were determined for each of the five landscapes using the species composition data in Table 1 and the emission capacities for species listed in Table 2. The emission capacities for grasses (5 gg C g'• hr 4 for isoprene and 0.2 gg C g-• hr 4 for monoterpenes) and for woody plants (16 gg C g'• hr -• for isoprene and 0.8 gg C g-• hr 4 for monoterpenes) not listed in Table 2 are based on values reported by Guenther et al. [1995]. The resulting isoprene emission capacities shown in Table 3    aThis is an overestimate if disturbances to C. mopane secretory cells caused the very high emission rate observed for this species. average foliar density of southern African savannas is about 75% of the average estimate for all savannas. The higher foliar density estimates and lower monoterpene emission capacity (per gram foliage) results in good agreement between the emission capacities (per square meter)estimated for the two field sites and by Guenther et al. [1995]. The good agreement between the per gram foliage isoprene emission capacities results in a factor of 2 to 4 higher per square meter isoprene emission capacity estimated by Guenther et al. [1995]. These results demonstrate the importance of accurate estimates of both foliar density and per gram foliage emission capacity.

Summary and Conclusions
Tropical savannas are one of many globally important biomes for which biogenic NMHC emissions data are lacking. The study described in this paper provides an initial characterization of NMHC emissions from southern African savannas and demonstrates a methodology that can be used to characterize emissions from other regions. Species with high isoprene or monoterpene emissions were identified in most savanna types. Emission rates were speciesspecific and several genera were found to have both high and low emitters. High and low emitters were also found to occur on both nutrient-rich and nutrient-poor soils.
Landscape average emission capacities for the five savanna types range from 0.6 to 9 mg C rff 2 h -• for isoprene and about 0.05 to 3 mg C rn '2 h '• for monoterpenes. Guenther et al. [1995] assumed that southern African savannas have an isoprene emission capacity of almost 10 mg C rn -2 h -1 and a monoterpene emission capacity of about 0.5 mg C m '2 h -1. On a unit foliar mass basis, the isoprene emission capacity assumed by the global model agrees with field measurements while the monoterpene emission capacity is lower. The factor of 2 to 3 higher foliar •lensity calculated by the global model, however, results in higher area average isoprene emissions estimated by the global model. Tropical and subtropical savannas clearly have the potential to make a significant impact on global NMHC budgets. Additional studies are required to accurately estimate foliar densities, species composition, and emission capacities for these globally important landscapes.