ARTICLE IN PRESS ELSEVIER Forest Ecology and Management 5863 (2002) 1-13 Forest Ecology and Management www.elsevier.com/locate/foreco Carbon storage of harvest-age teak (Tectona grandis) plantations, Panama Margaret Kraenzel^'*, Alvaro Castillo'', Tim Moore'^, Catherine Potvin^ "^Department of Biology, McGill University, 1205 Dr. Penfield Ave., Montreal, Que., Canada H3A IB I Autoridad Nacional del Ambiente, Apartado 2016, Paraiso-Anc?n, Panama "Department of Geography, McGill University, 805 Sherbrooke St. W., Montreal, Que., Canada H3A 2K6 Received 28 April 2001; accepted 26 December 2001 10 11 12 13 14 15 16 17 18 19 20 21 22 23 Abstract Reforestation is being considered as a mitigation option to reduce the increase in atmospheric carbon dioxide and predicted climate change. Forestry-based carbon storage projects are being introduced in many tropical countries, and assessment of carbon storage potentials is made difficult by a lack of species-level information. We measured above- and belowground biomass and tissue carbon content of 20-year-old teak {Tectona grandis) trees in four Panamanian plantations to estimate carbon storage potential. A regression relating diameter at breast height (DBH) to total tree carbon storage was constructed and used to estimate plantation-level tree carbon storage, which averaged 120t/ha. Litter, undergrowth and soil compartments were estimated to contain 3.4, 2.6 and 225 t C/ha, respectively. The soil carbon was a one-time measurement, not an estimate of soil C accumulation. We estimate carbon storage in Panamanian harvest-age teak plantations to be 351 t C/ha. Various methods of calculation of carbon storage in short-rotation plantations are discussed. ? 2002 Published by Elsevier Science B.V. R?sum? 24 25 26 27 28 29 30 31 32 33 34 35 36 37 Hoy en d?a, la reforestaci?n est? siendo considerada como una opci?n para mitigar los cambios clim?ticos predichos como resultado de la contaminaci?n atmosf?rica por di?xido de carbono. En muchos pa?ses tropicales se est?n introduciendo proyectos forestales de almecenaje de carbono. Este estudio se enfoca en la teca {Tectona grandis) para medir varias caracter?sticas que afectan el potencial de almacenaje de carbono tanto de los ?rboles como de las plantaciones donde se encuentran. Se midieron la proporci?n ra?z-v?stago, la biomasa total y el contenido de carbono en los tejidos en ?rboles de teca de veinte a?os de edad en plantaciones paname?as. Se desaroll? una regresi?n que relaciona el di?metro a la altura del pecho con la cantidad total de carbono en el ?rbol que fue utilizada para estimar la cantidad de carbono almacenada en los ?rboles de cuatro plantaciones. Encontramos un promedio de 120 t C/ha en los arboles. Se estudiaron la hojarasca, el sotobosque, y los perfiles de los suelos, y encontramos promedios de 3.4, 2.6 y 225 t C/ha en esos compartimentos, respectivamente. Estimamos un almacenaje de carbono de 351 t/ha por estas plantaciones. Se discuten varios m?todos de c?lculo del almacenaje de carbono en plantaciones de rotaci?n corta. ? 2002 Published by Elsevier Science B.V. Keywords: Forestry; Carbon dioxide mitigation; Root biomass; Allometric equations; Soil carbon Corresponding author. Present address: PO Box 4, Douglastown, Que., Canada G4X 2Z1. Tel.: E-mail address: mkraenzel@hotmail.com (M. Kraenzel). + 1-418-368-6260. 1 0378-1127/02/$ - see front matter ? 2002 Published by Elsevier Science B.V. 2 PII: 30378-1127(02)00002-6 ARTICLE IN PRESS M. Kraenzel et al. /Forest Ecology and Management 5863 (2002) 1-13 1 38 1. Introduction 39 Of the 130 million ha of forest plantations in the 40 world (Allan and Lanly, 1991), just over half are 41 located in the tropics (FAO, 1995). The total carbon 42 storage that can be credited to global forest plantations 43 today is an estimated 11.8 Pg C (Winjum and Schroe- 44 der, 1997), about 10% of the carbon lost through land 45 conversion since industrialization. Forestry activity 46 designed to store carbon is often proposed for the 47 tropics, as tropical climates support rapid vegetation 48 growth rates (Schroeder and Ladd, 1991). Marland 49 (1998) estimated that based on higher potential growth 50 rates, the area required to capture annual carbon 51 emissions could be reduced by 25% if afforestation 52 efforts were centred in the tropics. Grainger (1988) 53 calculated that the tropics contain 758 million ha of 54 depleted or degraded lands which were once forested. 55 Reforestation of these areas would capture significant 56 amounts of atmospheric carbon, and would be 57 expected to contribute to soil quality and conservation 58 (Schroeder, 1992). Although there are several esti- 59 mates of carbon storage in various forest types 60 (Brown, 1993; Lugo and Brown, 1992; Vogt, 1991), 61 few estimates of individual species' carbon storage 62 potential have been published. To allow informed 63 choices between species when establishing carbon 64 storage projects, it is important to characterize various 65 traits which influence carbon storage on a per species 66 basis. Such information would also be useful for 67 inclusion in global carbon storage/cycling models. 68 For most species used for reforestation, only above- 69 ground biomass potentials are known. To have a whole 70 picture of species' carbon storage potential, one must 71 know aboveground-to-belowground biomass alloca- 72 tion patterns. Belowground allocation of biomass in 73 forests ranges widely, e.g., in tropical dry forests the 74 contribution of roots to total biomass has been esti- 75 mated to range from 18 to 46% (Sanford and Cuevas, 76 1996). 77 This study was conducted in Panama, where for- 78 estry plantation is increasing rapidly in popularity. 79 From 1992 to 1998, the area of abandoned land that 80 had been reforested rose from 11 000 to 34 600 ha. 81 Just over half of these reforestation projects have been 82 conducted using teak (ANAM, 1999a). Today, teak 83 ranks third among tropical hardwood species in terms 84 of plantation area established world-wide, covering 2.25 million ha (Krishnapillay, 2000). It is planted 85 extensively in the world's tropics for high-quality 86 timber. Because of teak's increasing popularity as a 87 plantation species, we choose to study its carbon 88 storage potential. Schroeder and Ladd (1991) point 89 out the importance of considering a species' cumula- 90 tive carbon storage potential rather than its potential 91 maximum growth rate at some point during its life- 92 cycle when estimating its carbon storage potential. For 93 this reason, this work was conducted in plantations of 94 harvest-age, which for teak in Central America is 20 95 years. 96 The goals of this work were: (1) to measure teak 97 root-to-shoot ratio, total biomass and tissue carbon 98 concentrations, as well as litter production, under- 99 growth biomass and carbon storage, and soil carbon 100 storage in teak plantations, (2) to develop two non- 101 destructive predictors of teak tree carbon storage and 102 biomass (one for whole trees, the other for the root 103 compartment), and (3) to produce an estimate of the 104 carbon storage potential of Panamanian teak planta- 105 lions at harvest age. The tree carbon measured in this 106 work represents the carbon sequestered by a planta- 107 tion over its lifetime. To translate this to carbon 108 storage potential, it is necessary to include informa- 109 tion about the harvest and replanting of such a planta- 110 tion. A discussion of the possible methods of ill calculation of carbon storage of these plantations 112 follows. 113 2. Materials and methods 114 2.1. Study site 115 This study was conducted in four 20-year-old teak 116 plantations in Panama's Canal Zone (9?20'N, 117 79?50'W), established by Panama's Ministry of Envir- 118 onment (ANAM) in 1978-1979. Three of the planta- 119 lions are on Lago Alajuela in Chagres National Park 120 (Boquer?n, Pe?as Blancas and Tranquilla), the other is 121 in Soberan?a National Park (Aguas Claras), all within 122 25 km of each other, inside the watershed of the 123 Panama Canal. These are small-scale plantations of 124 about 5 ha each, and have received very little manage- 125 ment, with only natural thinning and no undergrowth 126 removal. Basic characteristics of the trees of these 127 plantations are listed in Table 1. Common under- 128 ARTICLE IN PRESS M. Kraenzel et al. /Forest Ecology and Management 5863 (2002) 1-13 Table 1 Basic characteristics of the study plantations (trees, n = 48 per plantation)"' Plantation Average tree Average DB H Average tree Tree species composition name density (per ha) (cm) height (m) (teak:palm:other) Boquer?n 586 23.7 (7.6) ab 20.7 (4.1) 98:0:2 Pe?as Blancas 566 26.6 (8.6) a 19.6 (4.4) 96:1:3 ^ Tranquilla 621 25.3 (6.7) ab 20. 6 (4.3) 90:8:3 Aguas Claras 723 21.9 (5.0) b 20.6 (4.2) 93:1:6 Average 624 24.4 20.4 94:3:3 '' Letters denote groups of significantly similar DBH, based on ANOVA analysis (a = 0.05). Standard deviations in parentheses. 129 growth species are Gustavia superba, Heliconia lat?s- imo patha, Andira inermis and Bactris sp. 131 Average daily temperatures in this zone range 132 between 23 and 30 ?C, and annual precipitation varies 133 between 2300 and 3000 mm, with a 4-month-long dry 134 season from December to April (ANAM, 1999b). The 135 soils of these plantations were derived from sedimen- 136 tary rocks of tertiary age (Weyl, 1980), and soil 137 textures tend to be loamy throughout the profile 138 (Table 2). 139 2.2. Scales of study 140 To investigate the carbon storage in these planta- 141 tions, we worked on two different scales: the tree level 142 and the plantation level. We measured tree tissue 143 biomass and carbon concentration to describe the 144 relationship between DBH and carbon storage of 145 individual trees. At the plantation level, the tree-based 146 work was scaled up to estimate the amount of carbon 147 stored in the trees of the plantations, using average DBH and tree density for each plantation. This was 148 supplemented by litter, undergrowth and soil carbon 149 mass estimates. 150 Average and range of tree size were estimated using 151 the 48 trees closest to two 100 m transects established 152 at right angles to each other in each plantation. DBH 153 and height were measured using diameter tape and a 154 clinometer (Haga). From these 192 (4 x 48) trees, 155 nine trees covering the range of size present in the four 156 plantations were subsampled to be harvested for 157 above- and belowground measurement of biomass 158 and tissue carbon concentrations. At each plantation 159 except Tranquilla (where the lack of water supply 160 precluded root harvest), the 48 trees were separated 161 into three groups of 16 based on size, and from each 162 size class one tree was randomly selected for harvest. 163 Felling areas were cleared of litter and undergrowth 164 and the trees were directionally felled. Aboveground 165 biomass was separated into different tissue types 166 (large, medium, small leaves, flowers, twigs, and 167 branches), and the trunk cut up into metre-long pieces. 168 Table 2 Basic characteristics of the study plantations (soil, with pH and bulk density of surface samples (0-10 cm depth, ? = 15 per plantation), and colour of dry soil according to Munsell soil colour charts; surface layer = 0?10 cm depth, bottom layers = 10 cm to bottom of pit) Plantation name Soil texture Soil colour Average profile Bulk density pH depth (cm) (g/cm=') Boquer?n Surface layer: silty loam Bottom layer: loam Surface layer: loam Light grey 2.5 years (7/2) 180 0.63 (0.07) 6.6 (0.7) Pe?as Blancas Reddish-yellow >200 0.74(0.10) 6.2 (0.2) Bottom layer: clayey loam 5 years (6/6) Tranquilla Surface layer: loam Bottom layer: loam Brownish-yellow 10 years (6/6) 160 0.75 (0.13) 5.9 (0.3) Aguas Claras Surface layer: slightly clayey loam Bottom layer: slightly clayey loam Dark yellowish-brown 10 years (4/4) 190 0.66 (0.20) 6.1 (0.4) ARTICLE IN PRESS M. Kraenzel et al. /Forest Ecology and Management 5863 (2002) 1-13 1 169 To excavate the coarse roots (>5 mm in diameter), 170 we started at the stump and followed the roots to their 171 ends. For the most part their growth was shallow and 172 lateral, without a taproot. As the tree density was high, 173 it was difficult to distinguish between fine root sys- 174 tems of different trees sharing the space. To deal with 175 this problem, pits were established around each tree as 176 the coarse roots were excavated, from which all soil 177 was removed to isolate the fine roots. The soil was 178 manually washed using a low-pressure water source 179 over a 1 cm mesh. The perimeters of these pits were 180 set halfway between the focal trees and their neigh- 181 hours (an average of 1.5 m from the focal tree). Out- 182 side of these pits no fine roots were collected, to 183 balance for the foreign fine roots which were collected 184 from within the pit. In this study, fine roots were 185 considered to be <5 mm in diameter. The technique 186 of washing the soil did not allow us to collect all fine 187 roots present. To estimate the amount of fine roots of 188 diameter smaller than 5 mm (not collected), 12 trials 189 were performed at each tree. Five litres of soil from 190 random areas in the pit were processed as normal, then 191 the washed soil was collected and all fine roots it 192 contained possible to collect by hand were isolated 193 from it. To calculate the proportion of roots left behind 194 by our >5 mm technique, we compared the <5 mm 195 root masses collected in the trials to the fine root 196 masses collected as usual. This average proportion 197 was added to each tree's fine root mass. We believe 198 this accounted for most of the roots not measured by 199 our >5 mm collection method. No attempt was made 200 to separate dead and live roots in either size class. 201 Wet masses of all materials were measured using a 202 Viking 300 lb capacity spring scale (Viking). Samples 203 were immediately taken from each tissue type to 204 obtain wet-to-dry mass conversions and for later 205 carbon content analysis. The tree-specific wet-to-dry 206 mass conversion factors for different tissues were used 207 to convert total wet mass per tissue to total dry mass 208 per tissue for each tree. These dry masses were then 209 converted to tissue carbon storage by multiplying 210 them by tree- and tissue-specific carbon concentra- 211 tions. 212 Plantation-level work was performed in all four 213 study plantations. Tree density in these plantations 214 was estimated by counting all trees in a random area of 215 25 X 25 m^. The litter layer (any dead plant material 216 on the plantation floor) was collected at the end of the dry season (1999). The accumulated mass of litter was 217 used to approximate the annual litter fall. On average, 218 the woody portion made up 17% of the litter. We do 219 not know what part of this portion of the litter came 220 from the current year or from previous years. Teak and 221 non-teak litter were separately collected from 12 222 randomly located 1 x 1 m plots. Aboveground bio- 223 mass of non-teak undergrowth was collected from five 224 3 X 3 m^ plots in each plantation at the end of the wet 225 season (1999). Because we were only able to sample 226 aboveground undergrowth, total undergrowth biomass 227 was estimated from measured aboveground biomass 228 by multiplication by 1.34, based on the root-to-shoot 229 ratio for tropical deciduous forest plants reported by 230 Jackson et al. (1996). 231 Fifteen random soil samples were taken from the 232 soil surface (0-10 cm) of each plantation. As well, 233 samples were taken at each 10 cm of depth from two 234 or three 2 m deep pits in each plantation. Soil profile 235 depth was measured as the average depth at which 236 each plantation's pits became rocky and resistant to 237 sampling. Bulk density, pH, soil texture and organic 238 matter content were measured for both surface and pit 239 samples. 240 2.3. Sample treatment and chemical analysis 241 The tree tissue samples and collected litter and 242 undergrowth were weighed wet within 3 days of being 243 collected, using a Salter-AND-EK scale with 12 kg 244 capacity (Salter). They were dried at 70 ?C for 1 week, 245 and reweighed to produce tissue-specific wet-to-dry 246 mass conversion factors. 247 To prepare for organic carbon determination, the 248 vegetation samples were ground with mortar and 249 pestle using liquid nitrogen. For each of the nine study 250 trees, all samples per tissue type were pooled into one 251 100 g sample. Subsamples of 100 g in size were taken 252 from the material from eight randomly chosen litter 253 samples per plantation. Within each subsample, teak 254 and non-teak litter were recombined in their original 255 mass proportion. Dry material from each of the five 256 undergrowth plots was chopped into fine pieces, sub- 257 sampled, ground, and for each plot a subsample of 258 100 g in size was taken for carbon determination. 259 These subsamples were analysed for carbon concen- 260 tration using gas chromatography on a CHN Elemen- 261 tal Analyser, EA 1108 (Fisons Instruments). The 262 ARTICLE IN PRESS M. Kraenzel et al. /Forest Ecology and Management 5863 (2002) 1-13 263 analyser was monitored for accuracy of readings every 264 10 samples with a sulphanilamide standard. 265 The soil samples were dried for 1 week at 70 ?C, and 266 sieved using 2 mm mesh to remove any vegetation or 267 gravel present. Soil texture was estimated manually, as 268 described by Schlichting et al. (1995). Acidity (pH) 269 was measured in 0.01 M calcium chloride in a ratio of 270 1:3, using an Orion Research Digital lonalyzer, Model 271 601 (Orion Research). Organic matter content of all 272 soil samples was estimated through loss on ignition 273 (LOI), by combustion in a muffle fumace at 350 ?C for 274 16 h (Hesse, 1971). CHN analysis (as done on the 275 vegetation samples) was performed on 30 of these 276 samples to provide organic carbon content. This data 277 were used to build a regression between organic carbon 278 content and LOI. The relationship was statistically 279 significant {p < 0.0001), had a coefficient of determi- 280 nation of 0.715, and the standard error of estimate was 281 1.044. This regression was applied to the other soil 282 samples to estimate their organic carbon content. 283 Soil data were grouped into various layers of depth 284 in all profiles. Average bulk density, organic carbon 285 concentration and organic carbon storage were calcu- 286 lated for these profile layers (Fig. 2). 287 2.4. Statistical analysis 288 Various linear regressions were constructed using 289 DBH as the independent variable, and total tree bio- 290 mass, total tree carbon storage, root biomass and 291 carbon storage as dependent variables, using data from 292 all nine trees. All these data were transformed using 293 log to the base 10, as is commonly done to linearize 294 data of this type. One-way analysis of variance was 295 used to test the differences between carbon contents of 296 the various tree tissues. As well, tissues were grouped 297 as woody (trunk, branches, coarse roots and twigs) and 298 soft (leaves, flowers and fine roots), and the difference 299 in carbon content between these groups was tested 300 using one-way analysis of variance. One-way analyses 301 of variance were also used to test whether pH, root-to- 302 shoot ratios, mass and carbon concentrations of litter 303 and undergrowth, undergrowth-to-teak litter ratios, 304 tree height and DBH varied among plantations. 305 Two-way analysis of variance was used to test whether 306 bulk density and % soil carbon varied among planta- 307 tions and depths. All statistical analyses were con- 308 ducted using Systat 9.0 for Windows. "^^ 3. Results 309 Average tree heights range between 19.6 and 310 20.7 m, and average DBH ranges from 21.9 to 311 26.6 cm (Table 1). Analysis of variance showed that 312 the trees at Aguas Claras had a smaller average 313 DBH than the trees of Pe?as Blancas (F = 3.84, 314 /? = 0.011). 315 T 3.1. Biomass and carbon concentration of teak 316 tissues 317 T While values of DBH of the nine excavated trees 318 ranged between 16.9 and 43.8 cm, total tree dry 319 biomass varied from 122 to 1365 kg. On average, 320 woody tissues (trunk, branches, twigs and coarse 321 roots) made up 95% of a tree's mass (Table 3). These 322 woody tissues have significantly higher carbon con- 323 centrations than the soft tissues: leaves, flowers and 324 fine roots (49.2 and 46.4%, respectively, F = 120, 325 p < 0.0001). By weighting the carbon concentrations 326 of the different tissue types by the proportion of the 327 total tree biomass they represent, we obtain an average 328 of teak tree carbon concentration (49.5%) which can 329 be used to obtain tree carbon storage estimates using 330 total tree biomass. The carbon storage of the nine 331 harvested trees ranges from 60 to 674 kg. 332 Simple linear regressions of log DBH versus log - 333 dry biomass, and log DBH versus log carbon storage 334 show that these relationships are strong, yielding 335 coefficients of determination (r^) of 0.978 for both 336 regressions (Fig. 1). The linear regression of DBH 337 versus root system biomass and carbon storage (Fig. 1) 338 shows that 87% of the variation in root biomass and 339 carbon in a teak plantation can be explained by DBH 340 of the trees. 341 3.2. Root-to-shoot ratio 342 Root-to-shoot ratios (R:S) ranged from 0.11 to 0.23 343 in the nine excavated trees, with a mean of 0.16. When 344 carbon concentrations of these tissues are taken into 345 account, on average 13.1% of the trees' carbon was 346 stored in their roots, and 86.9% in their shoots. 347 Variability in root-to-shoot ratio was not strongly 348 related to tree size. Linear regression was not used 349 to analyse these data due to a violation of standard 350 assumptions which could not be remedied by trans- 351 ARTICLE IN PRESS 6 M. Kraenzel et al. /Forest Ecology and Management 5863 (2002) 1-13 Table 3 Proportion of tissue types in terms of biomass and tissue-specific carbon concentrations'" Tissue type Proportion of total tree biomass (%) Tissue carbon concentration (%) Small leaves (<25 cm long x 15 cm wide) Medium leaves ((35 x 20)-(25 x 15) cm^) Large leaves (>35 cm x 25 cm) Flowers (from six trees) Twigs Branches Upper trunk (upper third) Mid-trunk (middle third) Lower trunk (lower third) Coarse roots (>5 mm diameter) Fine roots (<5 mm diameter) 0.28 0.83 1.90 0.26 1.28 16.76 14.43 19.43 31.42 11.65 1.76 46.4 (1.1) abg 46.5 (0.9) abg 47.0 (0.8) ab 47.2 (0.4) ab 47.2 (0.4) ab 48.7 (0.6) cdf 49.6 (0.9) cdef 50.2 (0.4) de 50.4 (0.8) de 48.8 (0.6) cdf 45.2 (1.1) ag " In ail tissue categories 10 samples per tree were taken, except for the trunk categories, where five samples per tree were taken. Biomass proportion values are averages over nine trees. Carbon concentration values are averages of pooled samples from nine trees. Letters denote groups of significantly similar tissue carbon concentrations, based on ANOVA analysis (a = 0.05). Standard deviations are in parentheses. 352 formation. Instead, the Pearson correlation coefficient 353 was computed to measure the strength of association 354 between the two variables. Its value was ?0.292, 355 revealing a weak negative association between DBH and root-to-shoot ratio which was statistically insig- 356 nificant. One-way ANOVA showed that plantation 357 identity did not affect tree root-to-shoot ratio signifi- 358 cantly {F = 0.62, p = 0.571). 359 3.5 3 - 2.5 2 - 1.5 - Log Tree B = 2.575(Log DBH) - 1.042, R^ = 0.978, SEE = 0.056 Log Tree C = 2.574(Log DBH) - 1.345, R^ = 0.978, SEE = 0.056 ? TreeB o TreeC ? RootB D RootC 0.5 Log Root B = 2.399 (Log DBH) - 1.671, R^ = 0.867, SEE = 0.136 Log Root C = 2.387(Log DBH) - 1.968, R^ = 0.864, SEE = 0.137 1.2 1.3 1.4 1.5 LogDBH(cm) 1.6 1.7 Fig. 1. Linear regressions of DBH versus total tree dry biomass (#), total tree carbon storage (O). root system dry biomass (?) and root system carbon storage (D), for the nine study trees (all data log-transformed). ARTICLE IN PRESS M. Kraenzel et al. /Forest Ecology and Management 5863 (2002) 1-13 Table 4 Vegetation carbon storage values at the plantation level (tree carbon storage) Carbon storage Underground tree Aboveground tree Total tree carbon per tree (kg) carbon storage (t/ha) carbon storage : (t/ha) storage (t/ha) Boquer?n 180 13.8 91.8 105.6 Pe?as Blancas 248 18.4 122.2 140.6 Tranquilla 217 17.6 117.1 134.8 Aguas Claras 138 13.1 86.8 99.8 Average 196 15.7 104.5 120.2 360 3.3. Plantation-level carbon storage 361 The largest tree carbon storage at the plantation 362 level was found at Pe?as Blancas (141 t/ha), while the 363 smallest was found at the Aguas Claras plantation 364 (100 t/ha) (Table 4). The mean carbon storage in tree 365 roots of the plantations is 15.7 t/ha, while the mean 366 shoot carbon storage is 104.5 t/ha. The mean total tree 367 carbon storage at the plantation level is 120.2 t/ha 368 (Table 4, Fig. 3). 369 There was no significant difference between the 370 biomass and carbon concentrations of undergrowth 371 collected in the four different plantations {F = 0.56, 372 p = 0.684). The average carbon concentration of the 373 undergrowth is 44.4%, about 2% smaller than the 374 carbon concentration of yearly cycling teak tissues, 375 46.4% (F = 27.92, p < 0.0001), both inputs to the 376 plantations' litter. Average undergrowth biomass was 377 calculated to be 5.8 t/ha, containing 2.6 t carbon/ha 378 (Table 5, Fig. 3). 379 No significant difference was found between the 380 mean amounts of litter collected in the four different 381 plantations (F = 0.56, p = 0.642, Table 5). Average 382 dry mass of litter which accumulated over the dry season in these plantations was 7.9 t/ha, containing 383 3.4 t C/ha (Table 5, Fig. 3). On average, litter collected 384 was made up of 7% undergrowth tissue, and 93% teak 385 tissue. Averages of the undergrowth-to-teak ratio of 386 litter mass were found to be significantly different 387 between plantations {F = 3.52, p = 0.030). The 388 mean carbon concentration of the litter was 43.3%, 389 and did not vary significantly between plantations 390 (F = 1.48, p = 0.242; Table 5). 391 The textures and colours of the soils differed 392 between plantations, reflecting differences in parent 393 material (Table 2). The surface soil at Boquer?n was 394 found to be significantly less acidic than the surface 395 soil of the other plantations {F = 7.0, p < 0.0001). 396 No difference was found when comparing the average 397 surface soil (0-10 cm) bulk densities of the four 398 plantations, which ranged between 0.63 and 0.75 g/ 399 cm . There were insignificant differences between 400 plantations in terms of average profile bulk density. 401 A significant difference was found in soil organic 402 carbon concentration among plantations {F = 7.98, 403 p < 0.001). Both carbon concentration and bulk den- 404 sity changed significantly with depth (F = 12.78, 405 p < 0.001 and F = 6.37, p < 0.001, respectively), 406 Table 5 Vegetation carbon storage values at the plantation level (litter and undergrowth carbon storage)"' Mass (t/ha) Carbon concentration (%) Litter Undergrowth Carbon storage (t/ha) Litter Undergrowth Litter Undergrowth Boquer?n Pe?as Blancas Tranquilla Aguas claras Average 8.4 a (3.2) 7.7 a (1.5) 7.3 a (3.8) 7.9 a (3.1) 7.9 4.9 a (4.7) 6.6 a (4.0) 4.19 a (2.9) 7.5 a (6.1) 5.8 42.3 a (1.4) 43.1 a (2.6) 43.9 a (1.3) 43.8 a (1.2) 43.3 45.7 a (1.3) 43.9 a (2.7) 43.8 a (0.8) 44.1 a (1.6) 44.4 3.6 3.3 3.2 3.5 3.4 2.2 2.9 1.8 3.3 2.6 '' Undergrowth plots per plantation: n = 5, litter plots per plantation; biomass: n = 24; carbon concentration: M = 8. Letters denote groups of significantly similar mass or carbon concentration values, based on ANOVA analysis (a = 0.05). Standard deviations in parentheses. ARTICLE IN PRESS M. Kraenzel et al. /Forest Ecology and Management 5863 (2002) 1-13 Q 100 200 Bulk Density (g/cm) Organic Carbon Concentration (%) 20 40 60 80 100 Carbon Storage (t/ha) 120 Fig. 2. General patterns of bulk density and organic carbon concentration as affected by depth. Bars denote carbon storage per depth increment. Values are averages over the four study plantations. 407 and the interaction between plantation and depth had 408 significant effect in the case of carbon concentration 409 (F = 1.70, p = 0.044). The bulk density and carbon 410 concentrations of the various soil samples combined 411 across plantations give a general picture of carbon 412 storage at different depths (Fig. 2). Carbon concentra- 413 tion decreased with depth in a general pattern of 414 exponential decay. ^ 415 4. Discussion \ 416 Fig. 3 summarizes the knowledge we have about the 417 carbon storage in this system. The largest new carbon 418 store, after the establishment of the plantations, is the 419 trees themselves. Average carbon storage in the trees 420 of these mature plantations is 120 t/ha. As much of the trees' carbon is located aboveground, the longevity of 421 this carbon store depends on the fate of this wood once 422 it has been harvested. The litter and undergrowth of 423 this system contain a moderate amount of carbon 424 when compared to the other compartments (Fig. 3). 425 Adding carbon stored in undergrowth and litter (2.6 426 and 3.4 t C/ha, respectively) to the plantation estimate, 427 the carbon storage figure rises to 126 t/ha. The figure 428 shows that most of the carbon in the system is in the 429 soil, averaging 225 t/ha, bringing the total carbon in 430 each hectare of these plantations to 351 t. 431 The strength of the regression relating DBH to tree 432 carbon storage allows confident use of the equation for 433 estimation of carbon stores in trees of harvest-age teak 434 plantations. This tool may prove useful both for 435 application in existing plantations, as well as for 436 prediction of potential carbon storage when combined 437 > J c o ?t? E? (O ^ 1/J o ARTICLE IN PRESS M. Kraenzel et al./Forest Ecology and Management 5863 (2002) 1-13 150 100 50 - Flowers -Leaves "?Twigs Soil Surface 50 100 150 200 0-70 cm Depth r > 70-130 cm Depth 130 cm to pit bottom Soil Fig. 3. Carbon storage in various compartments at the plantation level. Storage values below the soil surface line represent belowground carbon stores. Values are averages over the four study plantations. 438 with site-index curves which predict productivity of 439 various sites in terms of tree size. The regression 440 which predicts biomass and carbon storage of tree 441 roots allows accounting of a carbon store until now 442 unknown in size. Since the plantations studied in this 443 work were not thinned, the equations presented here 444 would have decreased accuracy in managed planta- 445 tions if R:S were affected by management treatments. 446 The amount of carbon stored in a tree's roots is 447 often substantial, but is unknown for many species. 448 Despite teak's increasing popularity as a tropical 449 reforestation species, little work had yet been done 450 investigating the species' complete biomass (Karma- 451 charya and Singh, 1992). We found only one article 452 which addressed teak's belowground biomass alloca- tion (Hase and Foelster, 1983), a study performed in 453 Venezuela in an age series of teak plantations up to 9 454 years. Comparing our root-to-shoot results with those 455 of Hase and Foelster, there is a progressive decrease in 456 the values of this ratio with increasing plantation age, 457 from 0.42 at 4 years to 0.20 at 9 years, to our result, 458 0.16 at 20 years of age. The fact that we found no 459 relationship between root-to-shoot ratio and tree size 460 (DBH) in this study suggests that this trend may be 461 linked more directly to development with age than tree 462 size. 463 The mean root-to-shoot ratio found in these teak 464 plantations is small as compared to the more general 465 ratio that Cairns et al. (1997) produced from a review 466 of tropical forest biomass studies. They found the 467 ARTICLE IN PRESS 10 M. Kraenzel et al. /Forest Ecology and Management 5863 (2002) 1-13 1 468 average R:S for primary and secondary tropical forests 469 was 0.24. The amount of root carbon storage and 470 transmission of carbon to the soil through the roots 471 may be lower in forest plantations as compared to 472 natural forests. Cuevas et al. (1991) studied a Pinus 473 caribaea plantation and secondary forest of the same 474 age, growing in the same climate and on the same soils 475 in Puerto Rico. Total biomass was similar in the two 476 systems, but the pine plantation allocated only 6% of 477 total production belowground to roots, whereas the 478 secondary forest allocated 44% of its production 479 belowground. 480 In breaking up the tissues and determining separate 481 carbon concentrations for each tissue type, a pattern of 482 decreasing carbon concentration toward the trees' 483 extremities was revealed. The biomass-weighted 484 mean carbon concentration was 49.5%, very close 485 to the 50% value often used for estimation of carbon 486 storage from dry biomass information. The biomass 487 and carbon which turned over yearly in the trees of the 488 study plantations was small relative to their total 489 biomass. These biomass compartments made up 5% 490 of the trees' total biomass at 20 years of age, while 491 long-lived, woody tissues made up 95% of the bio- 492 mass. Karmacharya and Singh (1992) investigated 493 primary production allocation in the trees of an age 494 series of teak plantations in Kerala, India, and found 495 that in later stages of development, though the more 496 ephemeral tissues make up a small part of the trees' 497 total standing biomass, the trees have shifted much of 498 their production toward these tissues. At 30 years of 499 age, 50% of the trees' production went into woody 500 parts, and 50% into softer-tissue parts which turnover 501 rapidly. In the Panamanian study trees, when consid- 502 ering total production over a tree's lifespan, the 503 ephemeral tissues take on much greater importance. 504 Though not storing carbon within the tree itself for 505 long, they fall as litter, which can channel the portion 506 of carbon not decayed directly to the atmosphere 507 toward the soil carbon pool. 508 The litter accumulated on the floors of these planta- 509 tions was comparable in quantity to the annual litter- 510 fall of surrounding forest (Table 5). Leigh and 511 Windsor (1982) found that in the forest of BCI, less 512 than 50 km away from the furthest of the study 513 plantations, litterfall was 6.1 t/(ha per year), and state 514 that litterfall in most lowland tropical forests ranges 515 between 6 and 8 t/(ha per year). Measures in Sardi- nilla, a point central to the four plantations studied 516 here, show that the litter quantity on their study 517 pasture is 2.5 t/ha (Moore et al., submitted). The 518 increase in litter from pasture to plantation is appreci- 519 able, but the gain in carbon storage in this compart- 520 ment is small compared to the gain in the tree 521 compartment. 522 4.1. Carbon storage of Panamanian teak plantations 523 The 120 t of carbon stored in the trees of 1 ha of 524 these Panamanian teak plantations is similar to the 525 ?nal stocks of Australian radiata pine and Brazilian 526 slash pine on medium site classes (171 t C/ha over 45- 527 year rotation and 112 t C/ha over 30-year rotation, 528 respectively), as estimated by Nabuurs and Mohren 529 (1995). Cuevas and Medina (1986) pubhshed biomass 530 ?gures for three types of Amazonian forest, estimated 531 equivalent to 152 t C/ha in Terra Firme forest, 178 t C/ 532 ha in Tall Caatinga forest and 155 t C/ha in Tall Bana 533 forest. The six Central American lowland tropical 534 forest sites reported by Sanford and Cuevas (1996) 535 contained an average of 146 t C/ha. Using this figure, 536 we estimate that at the end of their rotation the teak 537 plantations store about 85% the amount of carbon of 538 the surrounding forest when unperturbed. 539 The carbon stored in these plantations may also be 540 compared to carbon storage in the vegetation of 541 pasture in Sardinilla, to quantify the increase in carbon 542 storage which may accompany reforestation with teak. 543 The grazed pasture of Sardinilla supported 2 kg C in a 544 hectare of vegetation (Moore et al., submitted). This 545 figure is expected to be higher on abandoned land. 546 4.2. Carbon storage calculations 547 The IPCC's default carbon storage calculation is 548 based on the amount of carbon stored in the trees of a 549 plantation at the end of their growth cycle (UNEP 550 et al., 1995). This is not a serious source of error if the 551 trees are not harvested until some long time after they 552 reach maturity (Christie and Scholes, 1995). Teak, 553 however, is grown for valuable hardwood, and in 554 commercial plantations is cut upon reaching the 555 desired size. As short-rotation plantations have high 556 capacity for carbon sequestration but short-term capa- 557 city for carbon storage, their carbon storage potentials 558 should be examined as mean storage over time, 559 ARTICLE IN PRESS M. Kraenzel et al. /Forest Ecology and Management 5863 (2002) 1-13 11 560 including harvest and regrowth, rather than as peak 561 carbon contents just prior to harvest (Schroeder, 562 1992). Nabuurs and Mohren (1995) also underline 563 the short-term nature of the short-rotation plantation 564 carbon sink. They focus on long-term results by 565 calculating carbon storage over many rotations. 566 Schroeder proposed a revised method for estimation 567 of carbon storage by short-rotation plantations, repre- 568 senting the average tree carbon storage over many 569 rotations. We used our data for teak to calculate long- 570 term storage using this mean carbon storage method. 571 To estimate standing crop for each year of the planta- 572 tion, we used a growth curve of teak grown in Costa 573 Rica in a GTZ project (COSEFORMA, 1998) to 574 calculate what proportion of final yield had been 575 reached at each year of growth. Our calculations with 576 teak data resulted in a mean carbon storage estimate of 577 76 t C/ha. 578 Winjum and Schroeder (1997) used the mean carbon 579 storage calculation to estimate the carbon storage capa- 580 city of various forest plantations, and concluded that 581 storage in the phytomass of plantations generally 582 increases from high to low latitudes, ranging from 47 583 to 81 t C/ha. Our mean storage estimate for Panamanian 584 teak plantations falls into the upper part of this range. 585 Tree plantations also store carbon in products made 586 from harvested wood, and this makes up an important 587 part of their carbon storage potential. From our bio- 588 mass data, we estimated that the study trees contained 589 60% of their biomass in usable trunk wood. This 590 represents an average of 72 t C/ha in harvestable wood 591 per rotation. The loss of teak biomass while sawing a 592 trunk into lumber is 58% (Van den Ende, pers. comm.) 593 leaving 30 t C/ha in sawed logs. Further losses would 594 be sustained in transforming saw logs into finished 595 products, depending on the product made. Winjum 596 and Schroeder (1997) estimate that over a 50-year 597 period, harvests from plantations in low latitudes store 598 15-37 t C/ha in wood products. Our above calcula- 599 tions show that over 50 years one would obtain 60 t C/ 600 ha in saw logs. By transformation into finished pro- 601 ducts, this may be reduced to an average in the range 602 of Winjum and Schroeder's estimate, though decom- 603 position of these products would have yet to be 604 factored in to get an equilibrium storage value. 605 To recompare the carbon storage of the teak planta- 606 tions to surrounding forest, taking a longer-term view, 607 one can see that mean storage in the vegetation of the plantations is about one-half of the storage of the 608 surrounding undisturbed forest (146 t C/ha, Sanford 609 and Cuevas, 1996). Storage in wood products could 610 make this gap considerably more narrow. 611 It is important to keep in mind that mean carbon 612 storage values for plantations are only valid while the 613 plantations exist and are replaced after each harvest. 614 After the plantation is discontinued, the vegetation 615 carbon storage on the land is much lower, akin to 616 pasture values, though plantation sites may be left 617 storing more carbon than before planting in cases 618 where tree presence and management engendered soil 619 rehabilitation and soil carbon storage. In contrast, 620 forests store carbon for much longer time scales with- 621 out need for human intervention. The plantation of trees 622 whose ephemeral tissues (as opposed to their wood) are 623 used as products may approach forest carbon seques- 624 tration capacity, as their mean carbon storage is not 625 continually cut back by harvests of wood. As well, these 626 plantations support locals, and in doing so may help to 627 slow surrounding deforestation. 628 The carbon stored in the first metre of the soil of 629 these plantations is comparable to the expected 630 amount of carbon in the first metre of tropical soils, 631 130-160 t/ha (Jobb?gy and Jackson, 2000). Measure- 632 ments taken in Sardinilla have shown that the estab- 633 lishment and growth of teak plantations to the age of 634 7-8 years provokes a very slight increase in soil 635 carbon storage, amounting to less than 20 t/ha (Moore 636 et al., submitted). From this observation, we assume 637 that much of the carbon of the soils of our study 638 plantations was present before the establishment of the 639 plantations. Moore's data suggest that the plantation of 640 abandoned land with teak does not promote significant 641 increases in carbon storage in the soil as the plantation 642 grows. An important question about the soil carbon 643 storage potential of plantations is the size of the 644 contribution of decomposing stumps and roots to soil 645 carbon over many rotations. Greater addition of car- 646 bon to the soil compartment may be achieved by 647 planting more deeply rooted tree species (Jobb?gy 648 and Jackson, 2000). 649 5. Conclusion 650 From our calculations, we conclude that teak plan- 651 tations have appreciable mean carbon storage capa- 652 ARTICLE IN PRESS 12 M. Kraenzel et al. /Forest Ecology and Management 5863 (2002) 1-13 1 653 city, much greater than that of the abandoned pasture 654 they were planted on. The compartment of the planta- 655 tion with the greatest potential for carbon sequestra- 656 tion and storage is the wood biomass (120 t C/ha). The 657 litter and undergrowth together contribute only about 658 6 t C/(ha per year). The total potential storage of teak 659 plantations is considerable, but not as large and long- 660 lasting as those of surrounding natural forest or of 661 plantations established for the collection of ephemeral 662 tissues. 663 Acknowledgements 664 The authors gratefully acknowledge the support of 665 the Natural Sciences and Engineering Council of 666 Canada in the form of an operating grant to C. Potvin, 667 as well as the support of the McGill University Centre 668 for Chmate and Global Change Research (C^GCR) at 669 McGill University. In Panama, invaluable support was 670 provided by the Smithsonian Tropical Research Insti- 671 tute as well as ANAM (Autoridad Nacional del 672 Ambiente). The authors thank Dr. Jim Fyles and Dr. 673 Nigel Roulet for their advice and input and also to 674 Anthony Wall for his help throughout this project. References 676 Allan, T., Lanly, J.P., 1991. Overview of status and trends of world 677 forests. In: Howlett, D., Sargent, C. (Eds.), Technical Workshop 678 to Explore Options for Global Forestry Management. Interna- 679 tional Institute for Environment and Development, London, 680 pp. 17-39. 681 ANAM, 1999a. Plan de manejo de los Parques Nacionales 682 Soberan?a y Chagres. Autoridad Nacional del Ambiente, 683 Panama, 130 pp. 684 ANAM, 1999b. Superficie reforestada por especie e a?o a nivel 685 nacional. Departamento Nacional de Plantaciones Forestales de 686 Panam?, Panam?, 2 pp. 687 Brown, S., 1993. Tropical forests and the global carbon cycle: the 688 need for sustainable land use patterns. Agrie. Ecosyst. Environ. 689 46, 31^4. 690 Cairns, M.A., Brown, S., Helmer, E.H., Baumgardner, G.A., 1997. 691 Root biomass allocation in the world's upland forests. 692 Oecologia 111, 1-11. 693 Christie, S.I., Scholes, R.J., 1995. Carbon storage in eucalyptus and 694 pine plantations in South Africa. Environ. Monit. Assess. 38, 695 231-241. 696 COSEFORMA (Cooperaci?n en los Sectores Forestal y Maderero 697 Convenio Costarricense-Alem?n), 1998. Manual del Curso Practico: Manejo de Plantaciones Forestales, Zona norte de 698 Costa Rica, octubre 12-16, 1998. 699 Cuevas, E., Medina, E., 1986. Nutrient dynamics within Amazo- 700 nian forests. I. Nutrient flux in fine litterfall and efficiency of 701 nutrient utilization. Oecologia 68, A6(y-A12. 702 Cuevas, E., Brown, S., Lugo, A.E., 1991. Above- and belowground 703 organic matter storage and production in a tropical pine 704 plantation and a paired broadleaf secondary forest. Plant Soil 705 135, 257-268. 706 FAO, 1995. Forest resources assessment, 1990?global synthesis. 707 FAO Forestry Paper 124. Food and Agriculture Organization of 708 the United Nations, Rome. 709 Grainger, A., 1988. Estimating areas of degraded tropical lands 710 requiring replenishment of forest cover Int. Tree Crops J. 5, 711 31-61. 712 Hase, H., Foelster, H., 1983. Impact of plantation forestry with teak 713 (Tectona grandis) on the nutrient status of young alluvial soils 714 of West Venezuela. For Ecol. Mgmt. 6, 33-57. 715 Hesse, PR., 1971. A Textbook of Soil Chemical Analysis. John 716 Murray Publishers Ltd., London, 520 pp. 717 Jackson, R.B., Canadell, J., Ehrlinger, J.R., Mooney, H.A., Sala, 718 O.E., Schulze, E.D., 1996. A global analysis of root distribu- 719 tions for terrestrial biomes. Oecologia 108, 389-411. 720 Jobb?gy, E.G., Jackson, R.B., 2000. The vertical distribution of soil 721 organic carbon and its relation to climate and vegetation. Ecol. 722 Appl. 10, 423-436. 723 Karmacharya, S.B., Singh, K.P., 1992. Biomass and net production 724 of teak plantations in a dry tropical region in India. For Ecol. 725 Mgmt. 55, 233-247. 726 Krishnapillay, B., 2000. Silviculture and management of teak 727 plantations. Unasylva No. 201. www.fao.org/docrep/x4565e/ 728 x4565e04.htm. 729 Leigh Jr, E.G., Windsor, D.M., 1982. Forest production and 730 regulation of primary consumers on Barro Colorada Island. In: 731 Leigh Jr, E.G., Rand, A.S., Windsor, D.M. (Eds.), The Ecology 732 of a Tropical Forest. Smithsonian Institute Press, Washington, 733 DC. 734 Lugo, A.E., Brown, S., 1992. Tropical forests as sinks of 735 atmospheric carbon. For Ecol. Mgmt. 54, 239-256. 736 Marland, G., 1998. The prospect for solving the CO2 problem 737 through reforestation. DOE/NBB-0082. US Department of 738 Energy, Office of Energy Research, 66 pp. 739 Moore, T., English, N., Lewis, J., Potvin, C, submitted for 740 publication. Soil carbon in a humid tropical pasture undergoing 741 land use change. 742 Nabuurs, G.J., Mohren, J.M.G., 1995. Modelling analysis of 743 potential carbon sequestration in selected forest types. Can. J. 744 For Res. 25, 1157-1172. 745 Sanford, R.L., Cuevas, E., 1996. Root growth and rhizosphere 746 interactions in tropical forests. In: Mulkey, S., Chazdon, R.L., 747 Smith, A.P. (Eds.), Tropical Forest Plant Ecophysiology. 748 Chapman & Hall, New York, pp. 268-300. 749 Schlichting, H., Blume, R, Stahr, K., 1995. Bodonkudliches 750 Prakticum, 2nd Edition (p. 41). Blackwell Science, Berlin, 751 295 pp. 752 Schroeder, P., 1992. Carbon storage potential of short rotation 753 tropical tree plantations. For Ecol. Mgmt. 50, 31^1. 754 ARTICLE IN PRESS M. Kraenzel et al./Forest Ecology and Management 5863 (2002) 1-13 13 755 Schroeder, P., Ladd, L., 1991. Slowing the increase of atmospheric 756 carbon dioxide: a biological approach. Clim. Change 19,283-290. 757 UNEP, OECD, lEA, IPCC (United Nations Environment Pro- 758 gramme, Organization for Economic Cooperation and Devel- 759 opment. International Energy Agency, Intergovernmental Panel 760 on Climate Change), 1995. IPCC Guidelines for National 761 Greenhouse Gas Inventories, 3 Vols. IPCC, Bracknell. 769 Vogt, K., 1991. Carbon budgets of temperate forest ecosystems. 762 Tree Physiol. 9, 69-86. 763 Weyl, R., 1980. Geology of Central America, 2nd Edition. Gebr. 764 Bomtraeger, Berlin, 371 pp. 765 Winjum, J.K., Schroeder, P.E., 1997. Forest plantations of the 766 world: their extent, ecological attributes, and carbon storage. 767 Agrie. For. Meteorol. 84, 153-167. 768 r.i ) I 1 I \ , ^^