Scaling relationships for woody tissue respiration in two tropical rain forests

Authors

  • P. Meir,

    1. Institute of Ecology and Resource Management, University of Edinburgh, Darwin Building, Kings Buildings, Mayfield Road, Edinburgh EH9 3JU, UK
    Search for more papers by this author
  • J. Grace

    1. Institute of Ecology and Resource Management, University of Edinburgh, Darwin Building, Kings Buildings, Mayfield Road, Edinburgh EH9 3JU, UK
    Search for more papers by this author

Correspondence: Patrick Meir. Fax: + 44 (0)131 6620478; e-mail: pmeir@ed.ac.uk

Abstract

The relationship between gross primary productivity (GPP) and net primary productivity (NPP) is not fully understood. One of the uncertainties relevant to this issue is the magnitude of woody tissue respiration. Although some data exist for temperate and boreal zones, measurements of woody tissue respiration in tropical forests are sparse. We made in situ chamber measurements of woody tissue respiration in two tropical rain forests, one in the Brazilian Amazon (Reserva Jarú) and one in Central Cameroon (Mbalmayo Reserve). We made measurements on a wide range of species at each site and over a range of stem diameters from 0·02 to 1·4 m. The rate of efflux of carbon dioxide (CO2) from bark at 25 °C, Rt, varied from 0·1 to 5·2 µmol m−2 s−1 across the two sites, and the efflux was related to both volume and surface area components of the measured stem sections. The temperature response in Rt was slightly higher at Jarú than at Mbalmayo, with Q10 values of 1·8 (± 0·1 SE) and 1·6 (± 0·1 SE), respectively. A log–log regression showed that Rt was significantly related to stem diameter, D (P < 0·001; r2 = 0·58–0·62) and was significantly higher at Mbalmayo than at Jarú (P < 0·001), but that the rate of increase in Rt with stem diameter, D, was similar between sites. At the Mbalmayo site, tree growth measurements made over a 4 month period were used to make two estimates of the maintenance (Rm) and construction (Rc) components of respiration embedded in Rt. The two methods agreed closely, suggesting that Rm was approximately 80% of Rc at this site. Rm could be strongly related to D using a sigmoidal relationship that described both surface area and volume components as sources of respiratory CO2 (r2 = 0·71). This functional model was combined with inventory, growth and climate data for the Mbalmayo site to make a first estimate of annual above-ground woody tissue respiration, RA, which was 257 (± 18 SE) g C m−2 year−1. This value corresponds to approximately 10% of GPP, slightly lower than that found for another tropical rain forest, but higher than for temperate forests. When combined with data from six other sites in tropical, temperate and boreal settings, a very strong relationship was found between RA and leaf area index (LAI), and between RA/GPP and LAI (P < 0·001, r2 = 0·98). This indicates that RA exerts an appreciable constraint on NPP and that this constraint varies closely with LAI across widely differing types of woody vegetation.

Introduction

Net primary productivity (NPP) is determined by the difference between gross primary productivity (GPP) and autotrophic respiration, the respiration of plants. However, autotrophic respiration is difficult to estimate so its relationship with GPP is sometimes assumed in forest stand growth models (e.g. Battaglia & Sands 1997; Landsberg & Waring 1997). Some reports suggest that the ratio of NPP to GPP is constant and close to 0·5, although data clearly supporting this are lacking (Waring, Landsberg & Williams 1998; Medlyn & Dewar 1999), and it has been argued elsewhere that this value varies instead from 0·5 to 0·7 among stands (Amthor & Baldocchi 2001) or with age (Mäkelä & Valentine 2001). Autotrophic respiration includes below-ground and above-ground components, the latter comprising leaf and woody tissue. The last of these, the respiratory flux from woody tissue, is relatively under-studied (Sprugel & Benecke 1991) even though it varies significantly among stands (Lavigne, Franklin & Hunt 1996; Ryan, Lavigne & Gower 1997) and with temperature (Ryan 1991). Above-ground woody tissue respiration (RA) has been estimated to be 6–13% of GPP (e.g. Lavigne, Ryan & Anderson 1997; Law, Ryan & Anthoni 1999); there are very few data available for tropical forests although the upper limit of this range has been reported (Ryan et al. 1994; Malhi, Baldocchi & Jarvis 1999).

Respiration rates in tropical forests have traditionally been considered to be large because of the higher temperatures at low latitudes (Yoda 1967; Yoda 1978; Whitmore 1984), but early measurements were made on excised sections of woody tissue which could have yielded inflated respiration rates through the effects of wound respiration. More recent studies (e.g. Ryan 1990; Sprugel 1990) have emphasized the utility of in situ chamber measurements. The calculation of annual respiration totals has been facilitated by the separation of the costs resulting from the growth of new cells from those resulting from the maintenance of existing cells, a distinction embodied in the ‘functional model’ (McCree 1970; Thornley 1970). Respiration associated with the construction of cells can be calculated from the mass and composition of newly produced tissue (Penning de Vries 1975, Williams et al. 1987), whereas that associated with maintenance is related to the mass and activity of live cells. The exact distinction between the two can be difficult to define (Sprugel & Benecke 1991; Thornley & Cannell 2000) but the model continues to provide a useful analytical framework.

The choice of scalar from which stand-scale estimates of woody tissue respiration are derived varies among studies. Sapwood volume has been successfully related to respiration rate in some temperate and boreal forests where detailed information is available (Lavigne et al. 1997; Law et al. 1999). This has also been done for one humid tropical forest, although measurements were made on only two species at this site, where maintenance costs for sapwood were calculated to be 54 or 82% of total respiration (Ryan et al. 1994). However, sapwood volume has not always been identified as the best correlate for respiration; other scalars, such as nitrogen concentration, total tree size, and stem surface area, have proved equally or more successful (Lavigne et al. 1996; Yokota & Hagihara 1998; Levy & Jarvis 1998). At tree and stand scales, the requirement for respiratory support tissue increases with leaf area, but biomechanical and carbon gain trade-offs are associated with each marginal increase in leaf area, and it is not clear if this leads to stand-specific or more general expressions of respiratory cost (Givnish 1995). In species-diverse tropical forests, where sapwood volume is very poorly known because of interspecific variation, estimates of stem respiration based on relatively simple measures of stem size may be appropriate. However, information on the variability in respiration rates amongst individuals is needed before we can establish and evaluate any general relationship.

In this study we made measurements in two tropical rain forest reserves, one in Cameroon and one in Brazil, in order to identify factors governing woody tissue respiration at these sites. We asked the following questions:

  • 1What is the response of woody tissue respiration to temperature; does this differ between sites?
  • 2Using simple measures of stem size can we account for the variation in respiration rate among individuals, species and sites?
  • 3What proportions of total respiration are used for growth and maintenance processes?
  • 4Can the functional model be used to make estimates of total woody tissue respiration in the absence of detailed measurements of sapwood volume?

We then used inventory, growth and climate data at one site to make a first estimate of above-ground woody tissue respiration, and compared this with estimates from other temperate, boreal and tropical sites in order evaluate the potential for a general relationship linking physical stand properties and above-ground woody tissue respiration.

Methods

Sites

The site characteristics for each forest are given elsewhere (Meir, Grace & Miranda 2001), and summarized in Table 1. The first site was at Jarú Biological Reserve, Rondônia State, SW Brazil (10°05′ S, 61°55′ W) and is referred to here as ‘Jarú’. The forest is classed as undisturbed ‘open forest’ grading to ‘dense forest’ in places (IBGE 1993). The second site was at the Mbalmayo Reserve, Central Province of Cameroon (3°23′ N, 11°30′ W) and is referred to here as ‘Mbalmayo’. This is a secondary deciduous forest that was selectively logged in 1988 (Lawson 1995). The forests are of similar height (30–40 m) with Mbalmayo having a slightly higher leaf area index (4·4 versus 4·0 m2 m−2). The soil at Mbalmayo is a deep red or yellow oxisol (Ngeh 1989), with a sandy clay surface layer, and at Jarú it is a red or yellow orthic acrisol with a sandy surface layer (Hodnett, Oyama & Tomasella 1996). Concentrations of nitrogen and organic carbon in the surface soil layer are higher at Mbalmayo (0·16 and 1·83%, respectively) than at Jarú (0·10 and 1·18%) (Meir et al. 2001). Rainfall is slightly lower at Mbalmayo (Table 1), with a bimodal annual rainfall distribution, yielding wet season peaks in May and October, whereas only one wet season is experienced at Jarú, from December to March. Mean monthly temperatures at both sites vary little over the year, between 23 and 26 °C (Ngeh 1989; Culf et al. 1996).

Table 1.  Site characteristics for Reserva Jarú, Brazil (undisturbed forest) and Mbalmayo Reserve, Cameroon (secondary forest)
 Jarú, BrazilMbalmayo,
Cameroon
  1. Sources for data marked as superscripts: aMeir (1996) bNgeh, 1989, cCulf et, al., 1996). Error in biomass estimate is SE, obtained from a 2 ha sample.

Dominant tree familiesMoraceae,
Leguminoseae,
Palmeae
Sterculiaceae,
Ulmaceae,
Leguminoseae
Mean canopy height (m) 35a 36a
Leaf area index (m2 m−2)  4·0a  4·4a
Rainfall (mm year−1)1900c1520b
Above-ground biomass
 (kg m−2)
 22·0a  8·7 (0·47)a

Gas exchange and stem growth measurements

Gas exchange measurements were made on trees at both sites, but tree growth measurements were only made at Mbalmayo. Sampling was designed to balance the needs to account for a large number of species and range of stem sizes, and as large a number of replicates as was logistically possible. A total of 101 measurements were made on the stems of 23 species at Jarú and 15 species at Mbalmayo. The range in stem diameter, D, was from 0·02 to 1·4 m; both pioneer- and climax-stage species were included (Table 2). At Mbalmayo growth bands were attached to each of 38 trees immediately above the point where the CO2 efflux chamber was fixed. Girth measurements (resolution = 0·1 mm) were made at monthly intervals from February to May 1994, at 0830–0930 h on each occasion.

Table 2.  Species sampled at (a) Jarú, Brazil and (b) Mbalmayo, Cameroon
(a) Jarú, Brazil(b) Mbalmayo, Cameroon
SpeciesFamilyDRtTypeSpeciesFamilyDRtType
  1. Rt rate of CO2 efflux (µmol m−2 s−1), corrected to 25 °C; mean of three measurements per stem. D is diameter of measured stem sections; where more than one individual is measured, the range in D is given and number of individuals specified in parentheses (individual tree data in Fig. 2). ‘Type’ describes ecological class of each species: 1 denotes pioneer, 2 denotes shade-tolerant and/or climax, and 0 denotes unknown ecology (D. Edwards, J. Ratter & T. Pennington, personal comms)

Astronium lecointei DuckeAnacardiaceae0·040·0502Xylopia etiopicaAnnonaceae0·13–0·26 [2]2·34–4·211
Xylopia sp.Annonaceae0·05–0·11 [5]0·29–1·241Vernonia confertaAsteraceae0·050·071
Orbigynia speciosaArecaceae0·26–0·30 [3]0·55–0·791Santira trimera (Oliv.) Aubr.Burseraceae0·121·792
Licania sp.Chrysobalanaceae0·351·222Terminalia superba Engl. & DielsCombretaceae0·03–0·55 [2]0·18–1·711
Hironima sp.Euphorbiaceae0·030·170Distemonanthus benthamianus Baill.Leguminoseae; Caes0·02–0·59 [8]0·48–5·362
Ocotea cf caudata (Nees.) MezLauraceae0·140·292Ptercarpus soyauxii Taub.Leguminoseae; Pap.0·080·910
Bertolettia excelsaLecythidaceae1·443·182Desmostachys tenuifoliusIcacinaceae0·020·230
Sclerolobium sp.Leguminoseae;  Caes0·04–0·29 [3]0·12–1·821Desbordesia glaucescens (Engl.) Van  Tiegh.Irvingaceae0·08–0·22 [2]1·64–2·692
Guarea kunthii A. juss.Meliaceae0·12–0·25 [2]0·36–0·382Klainedoxa gabonensis Pierre ex Engl.  var oblongifoliaIrvingaceae0·131·802
Trichilia quadrijuga H.B.K.Meliaceae0·180·422Musanga cecropoides R.Br.Moraceae0·02–0·51 [10]0·39–3·331
Cecropia ficilifolia Snethl.Moraceae0·070·461Coelocaryon preussi WarburgMyristicaceae0·050·800
Cecropia sciadophylla Mart.Moraceae0·04–0·65 [3]0·22–1·451Lophira alata Banks ex Gaertn.f.Ochnaceae0·09–0·34 [2]0·96–4·551
Naucleopsis glabra Spr. ex Baill.Moraceae0·05–0·08 [2]0·23–0·472Panda oleosa PierreOlacaceae0·532·732
Naucleopsis krunnii (Standl.) C.C.  BergMoraceae0·02–0·10 [5]0·21–0·442Triplochiton scleroxylon K. Schum.Sterculieaceae0·05–0·55 [7]1·01–3·841
Pseudomeldia sp.Moraceae0·120·712Trema orientalis (Linn.) Bl.Ulmaceae0·04–0·19 [10]0·77–3·961
Sorocea guilleminiana Grand.Moraceae0·150·432     
Trymatococcus amazonicus Poepp  et Endl.Moraceae0·24–0·40 [2]0·32–1·150     
Virola calophylla Warb.Myristicaceae0·06–0·12 [4]0·08–0·442     
Virola michelii HackelMyristicaceae0·03–0·40 [2]0·16–1·002     
Sterculia pruriens (Aubl.) SchumSterculiaceae0·02–0·57 [6]0·28–1·160     
Theobroma microcarpum Mart.Sterculiaceae0·170·250     
Rinorea pubiflora (Benth.) Spreng.Violaceae0·150·212     

Measurements of CO2 efflux from woody tissue were made from trees in Jarú, from May to June 1993, and in Mbalmayo, from February to May 1994. Although the rate of CO2 efflux from bark can occasionally be slightly affected by the CO2 concentration in sap (Levy et al. 1999), we treated our efflux rates as measures of woody tissue respiration. Measurements were made by attaching a chamber to the stem of a tree and connecting it to an infra-red gas analyser (IRGA). Chambers were constructed out of transparent acrylic plastic, and sealed to the bark using neoprene gaskets; for narrow stems a split-cylinder design was used to envelop the stem. For these chambers adequate mixing of air was achieved by minimizing chamber volume and placing inlet and outlet nozzles on opposite sides of the measured wood section. For the larger chambers, which were sealed against the stem, a small fan was inserted into the back wall to provide gentle mixing. The smaller chambers were 10–15 cm in length and 100–250 cm3 in volume whereas the larger chambers covered a bark area 15 cm × 6 cm, and including the fan, were 400–500 cm3 in volume. At each site measurements were made under a closed canopy with unshaded chambers, as a pilot experiment indicated a change of less than 1% in efflux rate between unshaded and shaded (black cloth) conditions at the measurement points.

The principal measurement method used was a closed-path analysis, where the chamber was connected in closed circuit to an IRGA and CO2 efflux rates calculated from the increase in CO2 concentration in the chamber (Licor 6200; Licor Inc., Lincoln, NE, USA). In order to cancel possible errors resulting from leaks, chamber CO2 concentration was drawn down to a point below ambient levels (360–450 µmol mol−1) and allowed to rise an approximately equal amount above ambient (Hutchinson & Livingston 1993). Measurements were made by logging data every 5 s for 60–120 s intervals. Bark surface temperature was measured inside the chambers using a copper–constantan thermocouple. Means were taken of three consecutive measurements made on each stem at the same location, and the diameter of the woody section enclosed by the chamber recorded (measurement resolution 0·1 mm). Measurements were made on stems of 50 individuals at Jarú and 51 at Mbalmayo. Measurements were made on five or more individuals of four species at each site, and on one to three individuals of other species (Table 2). At the Mbalmayo site monthly measurements were made on a subsample of 20 trees from February to May in order to identify any seasonally related changes in CO2 efflux rates. Overnight measurements were made on six individuals at Jarú and five individuals at Mbalmayo in order to characterize temperature responses. The efflux rate of CO2, Ra, was calculated using Eqn 1.

image(1)

where Δ[CO2] is the change in concentration of CO2 in chamber (µmol mol−1 s−1), Vch and Ach are the chamber volume (m3) and the enclosed leaf area (m2), respectively, and VT is the volume (per mole) of a gas at ambient temperature and pressure.

A subsample of measurements was also made using an open-path IRGA (LCA2; ADC, Hoddesdon, UK) at the Mbalmayo site (Table 3) in order to obtain additional temperature response information on three of the five species measured using the closed chamber system. The IRGA was connected in series to a chamber and air was drawn in by a mass-flow controlled air supply unit (MASU; ADC) at 350–500 mL min−1. Before entering the cuvette, ambient air was passed through a 2 L buffer chamber; and before entry into the optical bench of the analyser it was passed through a column of silica gel to remove cross-sensitivity to water vapour in this instrument. Bark surface temperature inside the chamber was measured using a copper–constantan thermocouple , and all data were stored in a Campbell 21X datalogger (Campbell Scientific, Leicester, UK) as 15 min averages of raw data measured at 1 Hz. Efflux rates of CO2 were calculated according to Eqn 2.

Table 3.  Temperature responses for woody tissue CO2 efflux at Jarú and Mbalmayo.
SpeciesDRokQ10r2
  1. Q10 is the multiple by which the efflux rate (µmol m−2 s−1) increases in response to a 10 °C increase in temperature, obtained by fitting Eqn 3. For Mbalmayo, open-path measurements of three species (*) gave very similar Q10 values (specified in brackets). Mean (±SE) values are of the fitted values for all species; r2 values are for fits to Eqn 3 of the closed chamber measurements; D is the diameter of each measured stem section (m)

Jarú
N. krunnii0·040·030·092·380·90
Pseudomeldia sp.0·10·060·072·070·84
T. microcarpum0·020·050·061·800·89
V. michelii0·040·270·061·750·71
N. krunnii0·050·080·041·490·78
Mean  0·06 (0·01)1·90 (0·12) 
Mbalmayo
M. cecropioides0·341·170·041·54 (1·46)0·93
X. etiopica0·250·410·061·450·85
T. orientalis0·140·410·051·57 (1·53)0·68
V. conferta0·050·220·061·820·75
D. benthamianus0·230·960·051·57 (1·59)0·91
Mean  0·05 (0·01)1·65 (0·06) 
image(2)

where Fch is the molar flow rate through the chamber and Ach is the area of bark enclosed by the chamber.

Data analysis

Temperature response

An exponential model was fitted to data from both closed and open system measurements made on individual trees (Eqn 3).

image(3)

where T is temperature (°C), k (°C−1) is a temperature coefficient of R, and R0 is the rate of efflux at T= 0 °C. The rate at which a process increases for an increase in T of 10 °C is termed the Q10, and is given by k(e10k). Measured respiration rates were normalized to 25 °C (Rt) using species-specific temperature responses, or where these were not available, using the mean value for each site.

The relationship between CO2 efflux and stem size

The relative contribution to total CO2 efflux (Rt) from surface area (S) and volume (V) components of each stem section was examined using the method of Levy & Jarvis (1998). If the efflux rate is related to stem surface area, then, if re-expressed on a volumetric basis, we would expect: Rt ∝ 1/D. Alternatively, if the efflux rate was dependent on stem volume, then, expressed on an area basis, we would expect: Rt ∝ D. The empirical relationship at each site between Rt and D was analysed further using a regression of ln Rt on ln D. After estimating maintenance respiration, Rm, from measurements of total and construction respiration at the Mbalmayo site (see below), the overall relationship between Rm and D was modelled as a sigmoidal curve (Eqn 4) representing the contributions from S and V components of each stem to the total respiration rate over the measured range in D.

image(4)

where a, b and c are fitted parameters in the model.

Separating maintenance and construction respiration

Since stem growth measurements were needed to separate construction and maintenance costs, this analysis was restricted to the Mbalmayo dataset. Temperature-corrected efflux rates were assumed to represent ‘total’ respiration (Rt), the sum of maintenance respiration, Rm, and construction respiration, Rc. Calculations of Rc and Rm were then obtained from the growth and respiration data, using two methods, M1 and M2.

M1.

The increase in wood volume under the chamber was calculated from growth measurements and the specific gravity of each species (Reyes, Brown & Lugo 1992) was then used to obtain the mass of new wood. Where specific gravity data were unavailable (three species), a value of 0·5 g cm−3 was assumed. The amount of carbon per gram of dry wood was assumed to be 50% of the ash-free dry weight (Edwards et al. 1980; Matthews 1993) and ash-free dry weight was taken to be 99·3% of dry weight (Ryan et al. 1994). Penning de Vries (1975) estimated a minimum metabolic construction requirement of 0·43 g CO2 per gram of new woody tissue and experimental determinations show good agreement (0·46, Sprugel & Benecke 1991 and 0·47, Ledig, Drew & Clark 1976). Taking the mean value of these, and converting to grams of carbon expended per gram of carbon in newly constructed wood yields a value of 0·248 g g−1. This was used to calculate Rc based on stem growth measurements, expressed on an area basis for each tree, and was subtracted from Rt to give Rm, the maintenance respiration rate.

M2.

Initial stem diameters and growth measurements were used to obtain relative growth rates (RGR, m−1) of each tree. Using the proportional differences between measured and modelled Rt, the empirical regression between the natural logarithms of D and Rt was used to normalize Rt for each tree to the mean diameter of all trees for which growth rate had been measured. This efflux rate, Rtd, was then plotted against RGR, and the regression between them extended back to the ordinate (RGR = 0) in order to determine the mean Rtd at zero growth, assumed to represent the mean maintenance respiration rate, Rm0. The mean Rc across all trees, Rcm, was then calculated as the difference between the mean value for Rtd and Rm0.

Results

Variation in CO2 efflux with respect to temperature and stem diameter

A strong response in CO2 efflux to temperature was found at each site using both open and closed measurement systems (Fig. 1); measurements made on the same areas of bark of individual trees using both systems gave similar respiration rates (paired t-test, P = 0·43, n = 8). The temperature coefficient, k (Eqn 3), varied from 0·040 to 0·087 in Brazil and from 0·037 to 0·060 in Cameroon, with the means for each site yielding similar Q10 values (1·8 ± 0·1 SE and 1·6 ± 0·1 SE, respectively; Table 3). Bark surface temperatures ranged from 18 to 30 °C during the diurnal cycle, with mean daily values during measurements of 24·2 °C (Jarú) and 26·5 °C (Mblamayo), close to the reference temperature of 25 °C at which respiration rates were compared.

Figure 1.

The relationship between temperature and rate of CO2 efflux from woody tissue in four species (Db** and Mc** from Mbalmayo; Nk* and Vm* from Jarú). Species identity is given by the letters Xx which represent, respectively, the first letter of the genus and species of each individual (full names provided in Table 2); * denotes closed- and ** denotes open-path measurement system.

The value of Rt at Jarú was between 0·1 and 3·3 µmol m−2 s−1, measured on woody sections of diameter (D) = 0·02–1·4 m (Table 2). At Mbalmayo, Rt was higher, between 0·2 and 5·2 µmol m−2 s−1, measured from woody sections of D = 0·02–0·6 m. Measurements made at different points around the circumference of woody stems where D was large showed that radial variation in Rt at constant D was very slight. Tests for the source of CO2 indicated that the efflux rate was proportional to both woody tissue volume (V) and bark surface area (A). Figure 2a & b suggest that the contribution from A was more important at low D and that from V was more important at high D. This pattern was also observed for individual species (data not shown). However, logarithmic transformation of the data produced strong linear relationships between D and Rt at both sites (P < 0·001, n = 50, Jarú, P < 0·001, n = 51, Mbalmayo; Fig. 3a & b). There was little interspecies variation in this relationship, with non-significant differences found between regressions for individual species, and between individual species and the pooled data set (P > 0·1 for all tests, Fig. 3a & b). The slopes of the regressions for the overall data sets from each site (Fig. 3, Table 2) were almost identical, but the intercept was significantly higher (P < 0·01) at Mbalmayo, reflecting the higher Rt values there.

Figure 2.

Relationships at Mbalmayo and Jarú between rate of CO2 efflux from woody tissue (respiration, Rt) and stem section diameter (D). (a) variation in respiration rate (on area basis) and D; (b) variation in respiration rate (on volumetric basis) and 1/D.

Figure 3.

Log–log relationships at Mbalmayo and Jarú between CO2 efflux rate from woody tissue corrected to 25 °C (Rt), and stem section diameter (D), including SE errors. Open symbols are Mbalmayo; closed symbols are Jarú. Data for four different species at each site are shown by: M-Xx or J-Xx, where M and J denote the sites Mbalmayo and Jarú, respectively, and the letters Xx are, respectively, the first letter of the genus and species of each species (full names provided in Table 2). At each site, where the sample size for a species is less than three, the data have been pooled and denoted M-Spp and J-Spp, respectively.

Maintenance and construction respiration

Growth measurements were obtained for 38 trees at Mbalmayo. Maintenance respiration, calculated using M1, was 80% of Rt across all individuals (sample mean), with construction respiration (Rc) varying among individuals by up to 53% of Rt. The cross-species value for maintenance respiration (Rm0), calculated using M2, was very similar, 85% of Rtd (Table 4, Fig. 4). The fit of Eqn 4 to variation in Rm with D was highly significant (P < 0·001, r2 = 0·71; Fig. 5), indicating that for this site, mean Rt could be estimated well by combining estimates of Rm from Eqn 4 and calculations of Rc from growth measurements.

Figure 4.

Relationship between normalized CO2 efflux rate (Rtd) and relative growth rate (RGR) at Mbalmayo. RGR is calculated from measurements of changes in D over 4 months (February to May); Rtd is normalized by diameter, as described in Methods.

Table 4.  Estimate of the contribution (%) to total respiration (Rt) by maintenance respiration at Mbalmayo Reserve
Calculation methodMaintenance respiration
as percentage of Rt
n
  1. Maintenance respiration is calculated by two methods (M1 and M2, see Methods), using growth measurements on individual trees: M1 yields estimates for individual trees (Rm) and M2 gives an estimate of average maintenance respiration across all individuals (Rm0). n= number of trees in each estimate; uncertainties are ±SE obtained from the mean value of Rm in M1 and the regression fit in M2 (Fig. 4).

M180 (4)51
M285 (7)38
Figure 5.

Variation in maintenance respiration rate (Rm) and stem section diameter, D, at Mbalmayo. Rm is calculated from growth measurements of individual stems (see Methods, M1). Species data are shown for four species and the remaining pooled dataset; the legend follows the coding used in Figure 3.

Discussion

The temperature response of woody tissue respiration tends to have a Q10 between 1·5 and 2·5 (Ryan 1991), although results for tropical species have spanned a smaller range (Q10 = 1·6–2·2; Ryan et al. 1994; Levy & Jarvis 1998). Our data fall within this range: the Q10 for Jarú was not significantly different from 2·0, with that for Mblamayo slightly lower. Hagihara & Hozumi (1991) note that Q10 values for woody tissue respiration tend to decline slightly above 25 °C, as found here for Mbalmayo. The Q10 values of woody tissue respiration clearly do not vary widely, consistent with the assumption that the underlying biochemical processes are similar. Differences in measured efflux rates from bark may be affected by changes in the CO2 concentration in sap, especially in the early morning (Levy et al. 1999), but our open system measurements did not suggest that this process was significant in our study.

The overall range in Rt in both forest reserves (0·1–5·2 µmol m−2 s−1) was also consistent with previous studies (Ryan et al. 1994; Edwards & Hanson 1996; Lavigne et al. 1996), but Rt at Jarú was significantly lower than at Mbalmayo (Fig. 3). The same measurement system (closed) was used at both sites, and the apparent sources of respiration were also similar, including both surface area (principally cambium and phloem cells) and volume (also including xylem parenchyma cells) components (Fig. 2). This discrepancy in Rt between sites suggests a basic difference in metabolic activity. Significant differences in respiration rate were not detected in relation to the ecological characteristics of the sampled species at either site (Table 2, Figs 3 & 5). However, because the lower biomass in the secondary forest at Mbalmayo supported a similar leaf area index (LAI) to that at Jarú (Table 1), it is possible that a relatively higher respiration rate per stem was required at Mbalmayo to maintain the necessary associated phloem and xylem tissues. The nitrogen concentration of the woody tissue samples was not measured, but the significantly higher levels of leaf nitrogen reported for the Mbalmayo forest over that at Jarú (Meir et al. 2001) imply that the higher Rt at Mbalmayo also reflected a higher nitrogen concentration and hence respiratory enzyme activity in the woody tissue (Lavigne et al. 1996).

Respiration from woody stems has been found to scale with both surface area (Levy & Jarvis 1998) and sapwood volume (e.g. Ryan 1990; Law et al. 1999), and in some cases equally well with both (Lavigne et al. 1996). In our data respiration scaled with volume and surface area components, but was better represented by a log–log graph of Rt on D (Fig. 3; r2 = 0·58–0·62). Although there was a significantly larger intercept for the Mbalmayo data, the slope was very similar at each site, and the fits did not differ significantly among species or between species and the overall datasets (Fig. 3). This similarity in the relative change in Rt with D suggests that the rate of increase with stem size in the volume of physiologically active cells is quite well conserved in tropical rain forests.

Since some of the variation in the relationship between Rt and D may have resulted from unknown quantities of respiration used for growth processes, we decomposed Rt for Mbalmayo into maintenance (Rm) and construction components (Rc), using stem growth measurements. The two methods (M1 and M2) gave very similar mean estimates across all species and individuals, indicating that Rm constitutes approximately 80% of Rt. This is in close agreement with Paembonan, Hagihara & Hozumi (1992) who obtained 79% for Chamaecyparis obtusa, and with Ryan et al. (1994) who obtained 84% for a slow-growing wet-forest species in Costa Rica, Minquartia guianensis; but it is rather higher than that for a faster growing tree, Simarouba amara (54%) from the same study. The wide variation in Rm calculated using M1 (Rm = 47–100% of Rt) describes the range of growth rates (Rc) encountered at Mbalmayo. Our mean values derived from M1 and M2 are closer to that reported by Ryan et al. (1994) for a slower-growing species and indicate that overall growth patterns were representative of a closed mature forest dominated by slow-growing species, even though both slow- and fast-growing species were considered in our sample (Table 2).

After removing Rc from Rt, D explained 71% of the variation in Rm using Eqn 4. In this relationship Rm is a multiple of D where D = 0·05–0·30 m, but at D > 0·30 m Rm stops increasing, implying that sapwood thickness (i.e. the respiratory activity of sapwood per unit surface area of bark) is roughly constant for larger stems. Data for individual species also fitted this model, with the clearest pattern visible for Musanga cecropioides (Fig. 4). Consistent with this, examination of Rm using the analysis employed in Fig. 2, indicated that at D < 0·30 m, Rm was proportional to surface area, and at D > 0·30 m, Rm was proportional to volume (data not shown). Equation 4 explains more of the variation in respiration than the logarithmic model in Fig. 3, and predicts lower Rt at D > 0·6 m. Measurements on larger trees are needed to distinguish between the two models, but we hypothesize that Fig. 4 more accurately represents Rm for this forest because it is based on a functional interpretation of woody tissue respiration. Corresponding results obtained by Ryan et al. (1994) indicated that variation in sapwood thickness in a wet tropical forest in Costa Rica (5–30 mm) was one-third of that found in temperate conifers, implying that the increase in sapwood thickness with diameter tends to zero at relatively low D.

In order to make a first estimate of annual stand-scale above-ground woody tissue respiration (RA) from our data we calculated Rc and Rm separately [RA = Rc + Rm (g C m−2 ground area year−1)]. To obtain Rc we assumed that growth rates over the year remained similar to the values measured during the study period; tree growth measurements by Ngeh (1989) for a nearby site support this assumption. To calculate Rm using Eqn 4 we needed tree height, basal area and monthly temperature data for the site, for which we used annual temperature and tree-by-tree inventory data from Ngeh (1989) and Meir (1996). Using a cone form for tree boles, we estimate RA to be 214 (± 18 SE) g C m−2 year−1 (the standard error was calculated from the asymptotic SE to the fit of Eqn 4 in Fig. 4 and the uncertainty in the inventory data, Table 1). This value excludes any estimate of respiration by branches. If scaled by biomass, branch respiration is approximately 23% of bole respiration (Deans, Moran & Grace 1996), although this could represent a minimum value as cellular respiration rates in the canopy are sometimes higher than in stems, as is the proportion of physiologically active cells (Sprugel 1990). Including branch respiration, RA at Mbalmayo is 257 g C m−2 year−1, approximately 10% of GPP (Meir 1996; Grace, Meir & Malhi 2001). This value, although based on a simple up-scaling procedure, is higher than the 6% estimated by Law et al. (1999) for a temperate pine forest but lower than much earlier estimates for tropical forests (23–50%; Müller & Nielson 1965; Yoda 1967; Whitmore 1984) made using data from excised woody sections. A more recent study of a wet tropical forest using in situ measurements gives a similar value to ours (13%; Ryan et al. 1994). The slightly lower annual respiration cost (i.e. RA/GPP) found for Mbalmayo may reflect the lower biomass at this forest, but could also represent a more general trend consistent with the lower LAI at this site. Combining data from seven tropical, temperate and boreal sites, including conifer and broadleaf stands, we find an exponential relationship between LAI (indirectly measured) and RA (r2 = 0·85), and a strong linear relationship (r2 = 0·98) between LAI and RA expressed as a fraction of GPP (Fig. 6a & b). The data describe a general pattern which is consistent with the observed correlation between sapwood area and leaf area in trees (Shinozaki et al. 1964; Mencuccini & Grace 1995). The relationship shows that the total amount of respiratory activity required per unit of leaf area depends strongly on LAI across a range of forests that differ widely in ecology, growth form and latitude. The increased marginal cost in RA associated with supporting a marginal increase in LAI (Fig. 6b) also suggests that total autotrophic respiration increases with GPP. This outcome contradicts the hypothesis that the ratio of NPP to GPP is constant for all forests (Waring et al. 1998; but see Medlyn & Dewar 1998), although Mäkelä & Valentine (2001) also argue that NPP/GPP may change during stand development. Indeed for NPP/GPP to be constant, Fig. 6b implies that there should be a relative reduction in leaf and/or root respiration with increasing LAI, despite the expected allometric increases in leaf and root biomass (Vanninen et al. 1996). Such reductions are not impossible, but measurements and further modelling are needed to determine whether or not the increase in RA/GPP with LAI is also reflected in other components of the autotrophic respiration budget.

Figure 6.

(a) Variation in above-ground woody tissue respiration (RA) with leaf area index (LAI). (b) Variation in RA as a percentage of gross primary production (GPP) at seven sites in tropical (four sites), temperate (two sites) and boreal (one site) woody vegetation. The datum for Mbalmayo is shown with uncertainty (± SE) in respiration based on uncertainties in parameter estimates to Eqn 4 and in biomass from a 2 ha inventory (Deans et al. 1996; Meir 1996); GPP is obtained from eddy covariance, physiology measurements and a canopy model (Meir 1996). GPP and RA for sites 3, 5 and 6 were obtained from eddy covariance measurements, measurements of leaf and woody tissue activity and modelling of canopy physiology (Baldocchi & Harley 1995; Baldocchi 1997, Lavigne & Ryan 1997; Rayment 1998; Malhi et al. 1999; Rayment & Jarvis 2001); GPP and RA for site 2 was obtained from measurements of net primary productivity and net gas exchange for foliage and woody tissue (Law et al. 1999); GPP for site 7 was obtained from measurements of biomass increment and woody tissue respiration (Ryan et al. 1994). LAI and other site details are specified in the above references, except site 7, where LAI is estimated at 6·5 m2 m−2 (R. Chazdon, personal comm.). LAI and RA for site 1 are obtained from Levy & Jarvis (1998); GPP has not been estimated for this site. LAI was indirectly measured at all sites [using hemispherical photography or an LAI 2000 sensor (Licor Inc.)], except site 5, where the canopy was directly sampled; destructive LAI measurements made for sites 1 and 6 agree with indirect measurements (Levy & Jarvis 1999, Meir, Grace and Miranda 2001).

Conclusion

Woody tissue respiration rates in Mbalmayo and Jarú are similar to those found at other tropical and temperate sites and have mean Q10 values ranging between 1·6 and 1·9. The relative rate of increase in respiration with stem diameter is almost identical between the two sites, but the higher absolute respiration rates at Mbalmayo, perhaps determined by higher tissue nitrogen concentrations, indicate that measurements are necessary to characterize diameter–efflux relationships at different sites. The proportion of total respiration, Rt, that is used for maintenance processes, Rm, varies upwards from 47%, with higher values for slower-growing climax-stage species. The mean value for tropical rain forests is probably close to the 80% thought to be representative of slower-growing species, as most large trees in closed forests are climax-stage species. Stand-scale estimates of annual above-ground woody tissue respiration (RA) can be made by calculating maintenance and construction respiration costs, but their accuracy is inevitably limited by our knowledge of the physiologically active woody volume in the boles and branches of a canopy. Despite these sources of uncertainty, estimates from a wide range of forests show RA to represent 6–13% of gross primary productivity (GPP), and that this variability between sites in RA and RA/GPP is strongly and positively related to leaf area index.

Acknowledgments

We are grateful to ABRACOS and TIGER for financial and infrastructural support in Brazil and Cameroon. ABRACOS was a collaboration between the Agência Brazileira de Cooperação and the UK Overseas Development Administration; TIGER was the ‘Terrestrial Initiative for Global Environment Research’ programme which was funded by the UK NERC (grant no. GST/02/065). We gratefully acknowledge the support and help of local collaborating institutions, including INCRA in Brazil and ONADEF in Cameroon. We would also like to thank L. Kruuk, L. Gormley, J. Gash, M. Mencuccini, and P. Levy for providing comments on the manuscript and/or technical support. We particularly thank R. Chazdon for generously making available data on the leaf area index of the forest at La Selva, Costa Rica.

Ancillary