1. Measurements are presented of CO2 flux from woody stems of two shrub species in the Sahelian zone of West Africa (Guiera senegalensis and Combretum micranthum). Measurements were made on excised stems and in situ.
2. An experiment suggested that the respiration rate was closely related to the stem surface area but showed little relationship with stem volume, and was therefore associated with the cambium and phloem rather than the sapwood. This contrasts with several studies in which sapwood volume appeared to be the dominant component and the difference is attributed to the comparatively small diameter of stems in the shrubs studied here.
3. Measurements were made of the response of stem CO2 flux to photosynthetic photon flux density (Q) and gave evidence of corticular photosynthesis in the stems.
4. The response of stem respiration to temperature was determined in situ. The data were analysed using a variant on the standard exponential relationship with temperature, R = (Rom + Rog) exp(k T), where R is respiration rate, Rom is maintenance respiration rate at 0 °C, Rog is growth respiration rate at 0 °C, k is a temperature coefficient and T is temperature. Data from the dry season were assumed to represent maintenance respiration and used to define Rom and k. The Rog term thus quantifies the increase in respiration during the wet season beyond this baseline level, presumably attributable to growth. Values of Rom were 0·054 and 0·074μmol m–2 s–1 in G.senegalensis and C. micranthum, respectively, whilst Rog ranged between 0·061 and 0·96μmol m–2 s–1 during the wet season.
5. At ecosystem scale on an annual basis, stem respiration represented 17% of leaf photosynthesis, whilst corticular photosynthesis was at most 11·1% of stem respiration.
In woody stems, CO2 is produced as a by-product of respiration in the living cells of the xylem parenchyma, cambium and phloem, and is reassimilated to some extent where photosynthetic chlorenchyma is present in the cortex or periderm. Compared with leaves, little work has been done on the gas exchange of woody tissues. As a consequence of this, even some of the most basic questions remain unanswered (Sprugel & Benecke 1991). However, when considering the carbon balance of an ecosystem over a year or more, stem CO2 efflux may be an important component, as total respiration (including the foliage and soil components) would be expected to equal photosynthesis in a steady-state ecosystem, such as a mature forest or other climax community. In tropical ecosystems, above-ground woody-tissue respiration has been estimated to be around 10–13% of gross photosynthesis (Ryan et al. 1994; Meir 1996), although the estimates are very few.
The aim of the present study was to quantify CO2 fluxes from woody stems in two Sahelian shrub species, Guiera senegalensis L. and Combretum micranthum G. Don., and to relate these fluxes to environmental variables and seasonal physiological activity. To our knowledge, these are the first studies of wood respiration in the semi-arid tropics and only two other studies appear to have been made elsewhere in the tropics (Ryan et al., 1994; Meir 1996). The study forms part of a larger project concerned with the carbon balance of Sahelian vegetation (Levy et al., in press; Moncrieff et al., in press) which was carried out within the framework of the HAPEX-Sahel experiment in Niger, West Africa (Goutorbe et al. 1994). The study described here has five foci: (1) to determine the source of respiratory CO2 within the wood; (2) to derive the response of respiration to temperature; (3) to derive the response of stem CO2 flux to photosynthetic photon flux density (Q); (4) to analyse the seasonal change in respiration in terms of growth and maintenance respiration; (5) to quantify the contribution of stem CO2 fluxes to the ecosystem carbon balance.
1. SOURCES OF RESPIRATORY CO2
Possible sites of respiration include the resting and dividing cambial cells, phloem cells and xylem parenchyma in the sapwood. If the cambial and phloem cells are the main source of CO2, then the measured flux should be proportional to the surface area (S) of the stem. Alternatively, if sapwood xylem parenchyma cells are a more important source, the volume (V) of wood will be the most important factor. There was no heartwood formation in the stems in this study, as most of the shrubs were less than 10 years old, so that V was assumed to be equivalent to sapwood volume. The simplest approach to identifying the main source of respiratory CO2 would be to correlate the flux from stem sections of different sizes with S and with V. However, this does not give unambiguous results, as S and V change together as stem diameter, d, varies (e.g. Lavigne, Franklin & Hunt 1996). Yoda et al. (1965) and Yoda (1967) developed an empirical equation describing the relationship between respiration rate and d, but did not attempt to separate S and V components of respiration explicitly. Here, a simple graphical analysis of this relationship is used in order to distinguish between these components.
If the respiration rate, R, is related to the surface area, then there should be a positive linear relationship between the reciprocal of stem diameter, d–1 and CO2 flux per unit volume (equation 1, Fig. 1a) but no relationship with CO2 flux per unit area.
If R∝S, then R/V∝S/V
Thus R/V ∝ d–1.
Alternatively, if the flux is volume dependent, there should be a positive linear relationship between diameter and the area averaged flux (equation 2, Fig. 1b) but no relationship with the volumetrically averaged flux.
If R∝V, then R/S∝V/S
A distinction can be made between these two possibilities if the respiration rate of stem sections of different diameters is measured and expressed both on a surface area basis and on a volume basis and related to the diameter of the stem.
2. RESPONSE TO TEMPERATURE
The response of respiration to temperature classically takes the form of an exponential increase, described by the equation
where Ro is the respiration rate at 0 °C, k is a temperature coefficient and T is temperature in °C. This relationship is commonly described by the Q10 parameter, where Q10 is equal to exp(10k) and is the increase factor for a 10 °C rise in temperature.
3. CORTICULAR PHOTOSYNTHESIS
Photosynthesis has been reported in woody stems of several species (e.g. Ludlow & Jarvis 1971; Perry 1971; Bossard & Rejmanek 1992). The cortex of many G. senegalensis stems was clearly green beneath the bark, particularly in fissures where the bark was thin, indicating the presence of chloroplasts and the capacity for photosynthesis. Furthermore, the shrubs are virtually leafless for 6 months of the year, so corticular photosynthesis may be particularly advantageous in this instance.
4. SEASONAL CHANGE IN RESPIRATION
The products of respiration are used in two rather different processes: maintenance and growth (Lambers, Szaniawski & de Visser 1983; Farrar 1985). The former includes all the processes involved in maintaining the cells, such as protein turnover, repair of membranes and maintaining ion gradients. Maintenance (or basal or dormant) respiration, Rm, is a strong function of temperature and is proportional to the mass of active cells. Growth (or construction) respiration, Rg, results from the metabolic processes involved in producing new tissue. As the biochemical pathway involved is the same, it too is a strong function of temperature, although it is also a function of the amount of growth and the chemical composition of the tissue formed. Although the distinction may not be clear in some cases [e.g. ‘growth-related maintenance respiration’ may exist (Sprugel & Benecke 1991)], the distinction still provides a useful conceptual framework for interpreting results, even if the partitioning of respiration between maintenance and growth is only an approximation.
Relatively few attempts have been made to separate growth and maintenance respiration in woody plants. Two methods have generally been used in these studies [although various other methods have been used in herbaceous plants (Amthor 1984, 1989) which have occasionally been extended to woody perennials (e.g. Paembonan, Hagihara & Hozumi 1992)]. One method involves using measurements of wood volume increment to infer the mass of woody tissue produced over a period of time (e.g. Ryan et al. 1994). This can be converted to a growth respiration rate by assuming that 0·43 g of CO2 are evolved for each gram of wood produced (Penning de Vries 1975). This amount can then be subtracted from the measured CO2 efflux to give the maintenance respiration rate. Measurements must be made over a sufficiently long time period so that volume growth can be detected, and there may be a lag between volume increase and growth respiration (Yokota, Ogawa & Hagihara 1994; Edwards & Hanson 1996). There is also a degree of uncertainty associated with the construction cost of wood, as this will vary with the chemical composition and density of the tissue.
A second method is to use dormant season measurements to establish the temperature response of maintenance respiration and to calculate growth respiration from measurements in the growing season by subtraction (e.g. Edwards & Hanson 1996). This assumes that the temperature response of maintenance respiration derived in the dormant season remains the same during the growing season, an assumption which appears to be reasonable (Sprugel 1990; Sprugel & Benecke 1991). Previous studies have generally presented growth respiration as a constant term added to the temperature-dependent maintenance respiration term. However, as both involve the same biochemical pathways, both terms are equally dependent on temperature, and this should be taken into account if growth respiration is to be described in a meaningful way. Here, we quantify growth respiration by redefining equation 3 to incorporate a growth term, Rog:
where Rom and Rog are the maintenance and growth respiration rates at 0 °C respectively. The parameters Rom and k were derived during the dry season (when Rog = 0), and assumed to remain constant over the year. With these parameters fixed, Rog can be determined by fitting data collected in the growing season to equation 4.
5. CONTRIBUTION TO ECOSYSTEM CARBON BALANCE
Very few studies have attempted to extrapolate results from chamber measurements to ecosystem scale. One of the main problems is the difficulty of accurately measuring S or V per unit ground area in woody vegetation, as well as confusion over whether S or V is needed at ecosystem scale. However, because of the relatively small stature of the vegetation in this study, direct measurements of stem surface area index (SSAI), leaf area index (LAI) and biomass were feasible. These data were combined with measurements of leaf and canopy gas exchange and micrometeorological variables (described by Levy et al., in press) to estimate the magnitude of ecosystem stem CO2 fluxes and leaf photosynthesis over the course of a year.
The work described here was carried out at the fallow and tiger bush sub-sites within the Southern super-site of the HAPEX-Sahel experimental area in Niger, in the Sudano-Sahelian zone of West Africa. The major annual climatic feature of the region is the wet season, approximately from June to September, during which more than 90% of the annual precipitation falls. This is followed by a warm dry season from October to mid-March and a hot dry season from mid-March to May. Average maximum daily temperatures range from 33°C in January to 41°C in April. Rainfall at Niamey (45 km to the north) averages 562 mm per year (1905 to 1986 mean). This contrasts with a potential evaporation of 2057 mm per year [1953 to 1962 mean (Sivakumar 1987)]. Consequently, the water balance is in deficit for much of the year.
The fallow site (13°14·63’N, 2°14·65’E) comprised an area which had previously been cultivated for millet but which had been left fallow for 7 years (see Wallace et al. 1994 for more details). Fallow land throughout the region is characterized by shrub vegetation dominated by G. senegalensis, with a ground layer of annual grasses and herbs. Bush density was 327 ha–1 and the average bush height was 2·2 m. The bushes are mostly leafless over much of the dry season. Leaf growth in G. senegalensis tends to precede the start of the rains, beginning in May, and senescence starts in late September. Leaf area reaches a maximum in late August – early September, before senescence and leaf fall begin in late September – October. The bushes are mostly leafless again by November.
TIGER BUSH SITE
Tiger bush is characterized by stripes or arcs of vegetation, tens of metres in width and up to several hundred metres in length. These are separated by areas of completely bare, indurated soil. The cause of this pattern is not fully understood, but most work suggests that such ‘vegetation arcs’ are caused by a positive feedback process between the vegetation and soil moisture in areas where sheet wash is an important source of water for plants (White 1970, 1971). The tiger bush site was located at the centre of an irregularly-shaped area of tiger bush which was approximately 3 km in diameter (13°11·89’N, 2°14·37’E) (see Culf et al. 1993 for more details). The vegetated strips covered approximately 33% of the surface area, as estimated from aerial photographs. The dominant species were Combretum nigricans Lepr. ex Guill. et Perrott (a tree up to 10 m in height) and the shrubs C. micranthum and G. senegalensis. A herb layer was present, similar to that at the fallow site. Leaf phenology is similar to that in the fallow vegetation, although leaf growth tends to begin a few weeks later.
Materials and methods
Measurements of CO2 flux from stems were made on G. senegalensis at the fallow site and C. micranthum at the tiger bush site in February and March 1992 and between June and October 1992, using an infra-red gas analyser (LCA3, ADC Ltd, Hoddeson, UK) and purpose-built chambers. Two techniques were used: (1) measurements on excised sections of stems in chambers in the laboratory and (2) in situ measurements, in which chambers were sealed on to stems in the field. In both cases, an open system was used with flow rates between 150 and 500 cm3 min–1.
Excised sections from stems of G. senegalensis of different diameters were measured in February 1992. The measurements were repeated on C. micranthum in July 1992. Eight stems of each species were cut at ground level and taken to the laboratory. Here, they were cut into smaller sections which fitted into cuboidal ‘Perspex’ chambers, with dimensions of 200mm × 60mm × 60 mm or 150mm × 25mm × 25mm, fitted with fans to promote mixing of the air. Diameters of the stem sections ranged from 0·1 to 4·8 cm. During measurements, the chamber was completely shaded using black cloth. Temperature was kept approximately constant by the air-conditioning in the laboratory, ranging from 23 to 27 °C. Stem temperature was measured using a copper-constantan (Cu-Con) thermojunction on the stem surface, referenced to the panel of a data logger (21X, Campbell Scientific Ltd, Shepshed, Leicester, UK).
Corticular photosynthesis was investigated in G. senegalensis, using the above chambers, placed outdoors in natural light on a clear day. Lower values of Q were obtained by shading with cloth. Q incident on the chamber was measured using a quantum sensor. Temperature inevitably increased over the measurement periods (typically by around 4·0°C) which were kept as short as possible for this reason.
Potential problems with measurements on excised tissue include wound respiration and the release of CO2 stored in the gas and liquid phases of the wood. Consequently, the ends of the stems were coated with petroleum jelly to minimize the flux across the cut surface. Measurements were always made within 6h of first excising the stem and stems were only cut into smaller sections immediately before being measured. Five G. senegalensis stem sections were measured both immediately after sectioning and 4h later to check for increases in traumatic respiration over this period. No difference was found. Furthermore, rates measured in the laboratory and in situ were similar at equivalent temperatures.
In situ measurements were made using cylindrical ‘Perspex’ chambers (210mm×40 mm diameter), which could accommodate stems with diameters between 20 and 30 mm. The chamber was placed at a height of 0·2 to 1 m. Reference air was pumped in from a 2·75 m mast. Reference and differential CO2 concentrations were recorded every 10s and averaged every 10min on the data logger. Stem temperatures inside and outside the chamber were measured using Cu-Con thermojunctions placed in cracks in the bark and referenced to the panel temperature in the data logger. During the daytime, the temperature of the stem inside the chamber was typically 2 to 3°C higher than the stem temperature measured outside the chamber. Q was measured by a quantum sensor placed adjacent to the chamber. The chambers could be covered with a black cloth, effectively reducing incident Q to zero. The system was left to run continuously for one to 10days. The approach was to measure a relatively small number of stems (two to four of each species) each month and to leave the chamber on for several days so as to get complete diurnal curves. Feedback between CO2 efflux and the chamber conditions was assumed to be negligible, although there is some evidence that there is a reduction in respiration in leaves in elevated CO2 (e.g. Bunce 1990).
CANOPY STRUCTURE MEASUREMENTS
In February 1992, a survey was carried out of all G. senegalensis bushes within an area of 6750 m2 at the fallow site. For each of the 221 bushes, the x-y co-ordinates with respect to a micrometeorological tower were recorded, along with bush height, maximum and minimum radii, height of maximum diameter, number of stems less than 10 mm diameter and diameter of all larger stems. All stem diameters were measured at a point 20 cm above the base.
Harvests of up to 26 G. senegalensis stems were made at approximately monthly intervals between June and October 1992, and the following were recorded: stem diameter, mass of leaves, leaf area and mass of wood. Linear regression was used to derive relationships between stem cross-sectional area at 20cm and both leaf area and biomass. These relationships were then used with the frequency distribution of stem diameters in the surveyed area to estimate LAI and biomass per square metre for G. senegalensis. Wood surface area was estimated on a small sample of three individual stems. Each stem was cut into small sections which were then sorted into diameter classes, incrementing by 5 mm. The volume of wood in each diameter class was calculated from mass and density measurements. These volumes were converted to surface areas by assuming a surface area to volume ratio of a cylinder with diameter equal to the mid-point of the diameter class. Direct measurements of LAI and biomass of the ground flora were made by harvesting within quadrats (Levy et al., in press). Only a limited number of direct measurements of canopy structure were made at the tiger bush site (Levy 1995) and are not presented here.
Results and discussion
SOURCE OF RESPIRATION
The theoretical relationships between R and d for the cases where R is proportional to either the stem surface area or volume are shown in Fig. 1a and b. In C. micranthum, there is a clear increase in the flux per unit volume with d–1 (Fig. 1c) but no relationship with the flux per unit area (Fig. 1d). The data thus indicate that the major respiratory source is area related. The situation is slightly less clear in G. senegalensis but results appear very similar, i.e. there is a linear increase in the flux per unit volume with d–1 (Fig. 1e) but no relationship between d and flux per unit area (Fig. 1f), except in the smallest stem sections (d<10mm). A weak positive relationship between diameter and area averaged flux is found in these data (Fig. 1f, r2 = 0·693, n = 9), suggesting that the volume component is more important in these tissues. This may be because (1) a higher proportion of the xylem cells are alive in the young thin stems or (2) the cambium and phloem make up a significant part of the stem volume in the thinnest stems. However, with the exception of very small stems, stem respiration in both species can be satisfactorily expressed on a surface area basis but not on a volume basis, suggesting that the bulk of the respiring tissue is associated with the cambium and phloem. It may be significant that the relationship with surface area is closer in the measurements on C. micranthum than in the measurements on G. senegalensis (Fig. 1), as the former were made in the growing season and the latter were made in the dry season, and growth respiration will mostly occur in the cambium and so is likely to be related to the surface area.
The view that the surface area component was generally more important, attributable to activity of cambial and phloem cells, was the accepted wisdom until recently (e.g. Landsberg 1986). However, several recent studies (Ryan 1990; Sprugel 1990; Ryan et al. 1994, 1995) have found that sapwood volume was the major component in stems of several coniferous species. Based on these and other studies, Sprugel et al. (1995) state that ‘when construction respiration is absent … phloem and cambium respiration are insignificant compared with sapwood maintenance respiration’ and conclude that area-averaged rates will not be generally useful. The difference in conclusions drawn here and by Sprugel et al. (1995) may be related to the size of the stems considered. Sprugel et al. (1995) were reviewing results from trees with considerably larger stem diameters than the shrubs considered here (approximately 8 to 41 cm cf. 0·1 to 4·8 cm here). Thin stems have a high S/V ratio, so the volume component of respiration will be relatively small, but at some point as d increases, the volume of living cells in the sapwood will surpass that of the cambium and phloem.
This idea was investigated using a simple model in which fixed values were assigned to the respiration rate per unit surface area and to the respiration rate per unit volume. For hypothetical stems, as d increases (and S/V decreases), the change in the relative magnitude of the two components can be calculated (Fig. 2). The respiration rates chosen were 0·19μmol m–2 s–1 (the mean of area based rates for G. senegalensis and C. micranthum in February, Table 2) and 15μmol m–3 s–1 for the volume based rate (Ryan et al. 1995, Table 2), both normalized to 10 °C. Using these values, the surface area component of respiration was larger than the volume component in stems where d was less than 5 cm. The values in Fig. 2 are somewhat arbitrary, as the true respiration rate per unit volume is not known, but the curves illustrate the general pattern. The pattern may break down where d < 10 mm as discussed above.
RESPONSE TO TEMPERATURE
Field respiration rates were relatively high in both species because of the high temperatures attained by stem surfaces. The response of respiration to temperature is approximately exponential as expected (Fig. 3). There is a suggestion that the response flattens off slightly above 40°C, as noted in other studies (Larcher 1980) but the effect is not large. It has been suggested that this effect is related to a decrease in enzyme activation energy at high temperatures, and should be modelled using an Arrhenius-type equation (Ryan 1991; Lloyd & Taylor 1994). However, this approach did not improve the fit to the data, and was not used. Values of Q10 derived from the dry season data were 2·17 for G. senegalensis and 1·64 for C. micranthum. Q10 values for woody plant respiration typically range from 1·5 to 2·3 (Ryan 1991). There is a clear increase in stem CO2 efflux when the chamber was shaded, suggesting that corticular photosynthesis occurred, as discussed below.
There was a demonstrable photosynthetic response of stem sections to Q of up to 0·85μmol m–2 s–1 when placed horizontally in full Q (Table 1, Fig. 4). Refixation did not exceed respiration and represented up to 75% of stem respiration in the dark. By comparison, refixation reached up to 90% of respiration in the data reviewed by Sprugel & Benecke (1991). The photosynthetic response was linear in four out of five cases, suggesting that photosynthesis was limited by Q. This is likely for three reasons: (1) one side of the stem was always in shade, (2) only a small fraction of Q incident on the bark is transmitted to the chloroplasts in the cortex and (3) the supply of CO2 to these cells is likely to be plentiful as CO2 evolved in respiration of surrounding tissue can be refixed. The quantum efficiency, α, was calculated as the slope of a regression line through the data in Fig. 4, giving a value of 260 μmol CO2 per mol quanta.
Table 1. . Refixation in excised Guiera senegalensis stem sections and in situ. Refixation is calculated as the difference between the CO2 efflux in the dark and in full light (Q = 1500 to 1800μmol m–2 s–1). In the dark, temperature ranged between 26·4 and 35·5 °C, and on average, was increased by 4°C in full light. The in situ measurements are the same as those presented in Fig. 3, where temperatures ranged from 15 to 50°C
The rates of refixation from excised sections are likely to be overestimates of values occurring in situ, as the stem was positioned horizontally during the experiment and so would absorb more radiation than a stem that was upright within a bush, shaded by a canopy of leaves. However, the value of α may apply in situ, as it is determined from the change in efflux with change in Q, and so is not dependent on the absolute amount of radiation absorbed. Using α to predict refixation in situ gives a maximum value of 0·52 μmol m–2 s–1 at Q = 2000 μmol m–2 s–1. This agrees well with the in situ rates of refixation shown in Table 1, estimated from the data in Fig. 3. In situ refixation ranged from 0 μmol m–2 s–1 at 15°C to around 0·4μmol m–2 s–1 at 45°C, and represented up to 33% of dark respiration. However, it is harder to relate these data to Q as the irradiance received by a vertical stem may differ considerably from that measured by the adjacent horizontal sensor, depending on solar angle and shading.
Work on leaves suggests that ‘dark’ respiration per se can be reduced by Q (e.g. Jackson & Volk 1970; Brooks & Farquhar 1985), although the effect is difficult to study because of confounding increases in photosynthesis and photorespiration (Cornic & Jarvis 1972). It is possible that some component of the apparent photosynthetic response in Figs 3 and 4 is actually photoinhibition of respiration rather than refixation of respired CO2, but it is not possible to distinguish between the two with the method used. Photorespiration is likely to be very restricted in the chlorenchyma as CO2 concentrations are high there, together with low O2 concentrations and Q.
It seems likely that refixed carbon will be used locally in respiration once again, as corticular photosynthetic rates never exceeded the respiration rate and so an assimilate pool will not accumulate. There is probably little or no water loss associated with corticular photosynthesis, as internal CO2 is refixed and the chlorenchyma is not exposed to the air via stomata or lenticels. It may therefore be interpreted as a xerophytic adaptation for carbon economy, which will be particularly effective during the 6 months of the dry season, when the plants are leafless and the stems receive a higher Q than in the wet season.
SEASONAL CHANGE IN RESPIRATION
Figure 5 shows that at 40°C, respiration rates in C. micranthum and G. senegalensis were generally less than 1μmol m–2 s–1 in the dry season, and between 2 and 6μmol m–2 s–1 in the wet season (only measurements made in the dark are compared so as to remove Q as a complicating variable). There are also differences within the growing season, as the rates were highest in August and September, the peak of the season, and were lower in July and October. Seasonal differences were confounded to some extent by variation amongst stems, especially in the C. micranthum data for July, although the broad seasonal pattern is the same in both species. To quantify the change in respiration attributable to growth, the growing season data were fitted to equation 4. The derived Rog parameters are shown in Table 2.
Table 2. . Seasonal change in parameters in the equation R = (Rom + Rog) exp(k T). The parameters were derived from the data shown in Fig. 5. The units of R are μmol m–2 s–1 in all cases. Rm and Rg are the maintenance and growth respiration rates at 25°C calculated from equation 4
The values of growth respiration, Rg, fall within the range of 1·3 to 5·3μmol m–2 s–1 at 20°C given in the review by Jarvis & Leverenz (1983) for a wide range of species during times of high meristematic activity. When extrapolated to this temperature, the highest value obtained in this study was 2·6μmol m–2 s–1 for C. micranthum in September. Only two data sets from tropical species are available for comparison. These gave Rg values of 0·21 and 0·57μmol m–2 s–1 (Ryan et al. 1994) and 0·30–0·44μmol m–2 s–1 (Meir 1996) at around 25°C. At this temperature, Rg is 2·2μmol m–2 s–1 in G. senegalensis and 3·3μmol m–2 s–1 in C. micranthum. However, the other studies were made in mature rain forest and higher rates might be expected in fast-growing shrubs during the short growing season of the Sahel.
There is a strong theoretical basis for the assumption that the value of k derived in the dry season will remain constant throughout the year, as the temperature response is an intrinsic property of respiration related to enzyme kinetics. Figure 5 shows that the major change during the growing season was in the offset (Rom + Rog), and variation in the slope, k, was slight. Similarly constant values of k have been observed in apple trees over 8 months and in different phenological states (Butler & Landsberg 1981), in stems and roots of Scots Pine throughout the year (Linder & Troeng 1981) and in a wide range of conifers (see Jarvis & Leverenz 1983). Only in the C. micranthum data from September was the value of k noticeably different, when the slope was slightly less. In a few cases where k has been found to vary seasonally (e.g. Paembonan, Hagihara & Hozumi 1991), the variation has been related to temperature. In this study, effects of temperature and seasonal changes in k can be easily separated, as temperatures were not markedly different during the growing season (as is usually the case in temperate climates) but there is no obvious explanation for the anomalous value of k in September.
Changes in the Ro parameter have also been found in other studies and attributed to changes in growth respiration (e.g. Linder & Troeng 1981) on the presumption that Rom was constant over the year. This assumption is less easily justifiable because Rom may vary in relation to seasonal changes in the amount, composition and activity of living tissue, for example, as a result of an increase in the number and mass of living cells per unit area by augmentation of the cambial layer with partially differentiated xylem cells (Havranek 1981) or changes in phloem activity. However, there are few relevant data available, as more sophisticated methodology is required to distinguish amongst these possibilities. Some of these phenomena may fall into the category of growth-related maintenance and for practical purposes are best classed as growth respiration. An attempt by Sprugel (1990) to test whether the rate of maintenance respiration was constant over the year suggested that this assumption may be valid, although the data were somewhat noisy. Given this assumption, equation 4 provides a simple means of quantifying the growth and maintenance components of respiration, taking into account the temperature dependence of both components.
CONTRIBUTION TO ECOSYSTEM CARBON BALANCE
The data shown in Fig. 6 were used to estimate SSAI for the fallow site. This gave a value of 0·12 which was used to convert fluxes from a stem surface area basis to a ground area basis, so as to represent ecosystem fluxes for the fallow site. These were extrapolated to the annual scale using equation 4. The Rom and k parameters used were those in Table 2. A polynomial was fitted to the Rog data in Table 2 so that Rog could be expressed as a function of day of year and updated daily. Hourly air temperature data for the whole year were available from automatic weather stations and were assumed to be representative of stem temperature. Using these data, hourly ecosystem stem respiration was calculated and summed over the whole year to give a value of 3·97 mol CO2 m–2.
In order to relate this to the total ecosystem carbon balance, annual leaf photosynthesis was estimated using a procedure which has been described elsewhere (Levy 1995; Levy et al., in press). Measurements of leaf and ecosystem gas exchange were made at the fallow and tiger bush sites, as well as at a millet field crop site. The leaf- and stem-scale data, together with canopy structure measurements, were used to paramaterize a process-based canopy model, MAESTRO (Wang & Jarvis 1990), which predicts ecosystem CO2 fluxes. This was tested against independent eddy covariance measurements of net ecosystem CO2 flux. Agreement was generally good and the validated model was used to predict carbon sequestration at the millet site over the growing season, using a year's weather data. This prediction agreed closely with harvest measurements (Levy et al., in press). The same procedure was used here, with parameters that had been measured for the fallow vegetation. This predicted a value of 23·93mol CO2 m–2 for net leaf photosynthesis (gross photosynthesis minus leaf dark respiration) at the fallow site accumulated over the growing season.
The results show that stem respiration was 17% of net leaf photosynthesis on an annual basis, and so was not a negligible term in the ecosystem carbon balance. This is close to the values of 10 and 13% for tropical rain forest from Meir (1996) and Ryan et al. (1994), respectively. The higher respiratory fraction in the fallow vegetation may be related to the high prevailing temperatures and the fact that the bushes are leafless for approximately half the year. Future studies of carbon balance in other ecosystems may reveal whether the close agreement among these values is merely coincidental or has any functional significance.
As the Q response of corticular photosynthesis was linear, annual ecosystem corticular photosynthesis, Pc, could be estimated as:
where Qannual is the total incident Q over the year (14·54kmol m–2 for 1992) and SSAI is SSAI. This gave a value of 0·44mol CO2 m–2, which represents 1·8% of net leaf photosynthesis and 11·1% of stem respiration. The true value of Pc will be somewhat less because of the erectophile angle distribution of the stems and shading by leaves during the growing season and so is a rather small fraction of the total ecosystem budget. However, corticular photosynthesis may result in a significant saving in the annual carbon budget of the bushes and be particularly important in offsetting the respiratory costs of surviving over the long dry season.
We would like to thank Joost Brouwer and the staff at ICRISAT Sahelian Centre for help during field work, and Patrick Meir for helpful discussion.