Transcript
1 The Holocene 13,1 (2003) pp. 73–81 231234567891011121318192526 27 Holocene palaeoenvironmental evolution 28 in the Sa˜o Paulo State (Brazil), based on 29 anthracology and soil 13 C analysis 30 R. Scheel-Ybert, 1 * S.E.M. Gouveia, 2 L.C.R. Pessenda, 2 31 R. Aravena, 3 L.M. Coutinho 4 and R. Boulet 5 32 ( 1 Laboratoire de Pale´ oenvironnements, Anthracologie et Action de l’Homme 33 (UMR CNRS 5059), Institut de Botanique, Universite´ de Montpellier II, 163 rue 34 Auguste Broussonnet, 34090 Montpellier cedex, France; 2 Laborato´ rio de 14 C, 35 Centro de Energia Nuclear na Agricultura, Universidade de Sa˜ o Paulo, Caixa 36 Postal 96, 13400–970 Piracicaba, SP, Brazil; 3 Laboratory of Environmental 37 Isotopes, University of Waterloo, Waterloo N2L 3G1, Ontario, Canada; 38 4 Departamento de Ecologia, Universidade de Sa˜ o Paulo, Caixa Postal 11461, 39 05422–900 Sa˜ o Paulo, Brazil; 5 Instituto de Geocieˆ ncias, Universidade de Sa˜ o 40 Paulo, 01498–970 Sa˜ o Paulo, Brazil) 41 Received 25 January 2001; revised manuscript accepted 23 January 2002 42 Abstract: This paper presents a reconstruction of the Holocene palaeoenvironmental evolution in the centralSa˜o Paulo State (Brazil) based on anthracological analyses, in association with soil isotopic composition ( 13 C)and radiocarbon dating from four sites. Anatomical identification of charcoal particles allows the reconstitutionof past plant associations, and consequently of the vegetation and climate history. Rather precise interpretationsmay be achieved when associating anthracology and soil 13 C analysis. In the early Holocene, climate wasdry and an open cerrado vegetation (savanna) covered most of this area. A cerrada˜ o (forested savanna) or asemideciduous forest existed in the more humid localities. After 3500/3000 14 C yr BP the climate was morehumid, similar to the present, leading to the establishment of forested vegetation in all the studied sites. Com-parison of these results with various palaeoenvironmental studies carried out in the Brazilian phytogeographicalzone of cerrado and semideciduous forest suggests that at least its greater part presented a similar trend inthe climatic evolution during the Holocene. This phytogeographical zone presented a dry climate during theearly Holocene, then a more humid climate during the late Holocene. Climatic conditions similar to the presentappeared from 5000 to 1000 yr BP, depending on the site. 43 Key words: Palaeoenvironment, palaeoclimate, anthracology, charcoal, savanna, soil isotopic composition,oxygen isotopes, Brazil, Holocene. 596061 Introduction 62 Charcoal fragments are frequent in Brazilian soils. They are 63 present in the savannas of central Brazil, e.g., in Sa˜o Paulo 64 (Penteado, 1968; Coutinho, 1981; Scheel et al ., 1995; Melo et al ., 65 1996; Gouveia et al ., 1999) and Minas Gerais States (Vernet 66 et al ., 1994; Pessenda et al ., 1998), but also in areas presently 67 occupied by dense evergreen forests (Soubie`s, 1980; Sanford 68 et al ., 1985; Servant et al ., 1989). Many charcoal horizons dated 69 from the late Pleistocene and the Holocene have also been found 17 *Author for correspondence (e-mail: rita@s cheel. com ) 1 © Arnold 2003 2 10.1191/0959683602hl596rp 1 HOL: the holocene 2 16-10-02 07:35:00 Rev 16.04x HOL$$$596P 70 in Amazonian French Guyana (Tardy, 1998). These vestiges, 71 mostly recording natural fires, are the result of dry periods and 72 suggest the occurrence of climatic changes in the past. 73 Many palaeoenvironmental studies, mostly based on palynolog- 74 ical analyses, have shown that important climatic oscillations 75 affected the Brazilian territory during the late Quaternary (see, for 76 example, Absy et al. , 1991; Oliveira, 1992; Ledru et al ., 1995; 77 Ferraz-Vicentini and Salgado-Labouriau, 1996). The climate in 78 different Brazilian regions seems to have evolved distinctly, since 79 many of these oscillations were not synchronous, or divergent. 80 For instance, the late Holocene is characterized by rising humidity 81 in the Amazon region (Absy et al ., 1991) and in southern Brazil 1 74 The Holocene 13 (2003) 2382 (Behling, 1996), both after c. 3000 yr BP, while in the north- 83 eastern caatinga there is a marked decline in moisture levels after 84 c . 4000 yr BP (Oliveira et al ., 1999). A signi fi cant effort in new 85 studies and the use of new techniques might improve our knowl- 86 edge of the Brazilian palaeoenvironmental history and eventually 87 improve our understanding of present and future climatic changes. 88 Microscopic charcoal found in soil or in lake sediments have 89 long been used to reconstruct fi re histories and to postulate cli- 90 matic changes (Byrne et al ., 1977; Salgado-Labouriau and Ferraz- 91 Vicentini, 1994; Piperno, 1997; Behling, 2000; Haberle and 92 Ledru, 2001), but they cannot be used to reconstruct the veg- 93 etation history. 94 The occurrence of macroscopic charcoal ( 0.5 mm) in soils 95 provides solid evidence for local fi re in fl uence and can be used 96 to identify fi re-prone areas with high spatial precision (Ohlson 97 and Tryterud, 2000). Anatomical identi fi cation of the charcoal 98 particles (anthracological analysis) also allows the reconstitution 99 of past plant associations, of the vegetation history, and conse- 100 quently of palaeoclimate. 101 Identi fi cation of ancient charcoal is based in the analysis of the 102 wood anatomy, which is very well preserved after carbonization. 103 Taxonomic determination, frequently at genus level, is quite 104 accurate. In the tropics, only charcoal pieces over 4 mm are usu- 105 ally analysed. Smaller fragments are normally impossible to ident- 106 ify, because in general they do not present a suf fi ciently large set 107 of anatomical characters (Scheel-Ybert, 2001). 108 Anthracological work applied to soils is still rare, but some 109 studies have already allowed reconstruction of the vegetation his- 110 tory for the late Pleistocene and the Holocene, e.g., in sites from 111 Australia (Hopkins et al ., 1993), France (Carcaillet, 1998) and 112 French Guyana (Tardy, 1998). 113 Study of the isotopic composition of soils also allows deduction 114 of the plant cover in a site, since plants of different photosynthetic 115 pathways have distinct carbon stable-isotope values. This analysis 116 is based on the determination of 13 C values of soil organic matter 117 (SOM). 13 C values of C 3 plants (shrubs and trees in general) 118 vary between – 32 ‰ and – 20 ‰ PDB, while C 4 plants (grasses) 119 range from – 17 ‰ to – 9 ‰ (Boutton, 1991). 13 C analysis at differ- 120 ent depths in soil pro fi les can characterize transitions between C 3 - 121 and C 4 -dominated vegetation types. 14 C dating of charcoal 122 samples characterizes the events chronology. 123 This study presents results on the Holocene climatic history in 124 the Sa ˜ o Paulo State, based on anthracological analysis, soil iso- 125 topic composition analysis and charcoal radiocarbon dating from 126 four sites. It supports and completes the results already obtained 127 by Gouveia et al . (1999; 2002), who presented palaeoenvironmen- 128 tal reconstruction and inferred climate changes during the late Ple- 129 istocene and the Holocene in southeastern and central-western 130 Brazil, based principally on isotopic approaches, and Gouveia and 131 Pessenda (2000), who studied the role of biological remount of 132 soil matter and colluvium in the formation of ferrasols of Sa ˜ o 133 Paulo State. 134 Previous work 135 Palaeoenvironmental investigations are still rare in the Sa ˜ o Paulo 136 State. However, various palaeoecological studies exist for the 137 same phytogeographical zone, specially in southern and central 138 Brazil. They are summarized in Figure 1. 139 At Piracicaba (22 ° 43 S – 47 ° 38 W), the analysis of SOM car- 140 bon isotopes showed more enriched 13 C values (around – 16 ‰ ) 141 at the deeper part of the pro fi les and more depleted (around – 142 26 ‰ ) at the upper part. These results suggest a dry period and 143 the presence of grasses before c . 3500 yr BP, and an increasing 144 in humidity thereafter (Pessenda et al ., 1998). 145 In Sa ˜ o Paulo City, a dry climate has been identi fi ed around 1 HOL: the holocene 2 16-10-02 07:35:00 Rev 16.04x HOL$$$596P 146 4700 yr BP, characterized by high percentages of grass pollen 147 in peat-bog samples from the Vale do Anhangabau ´ (Takiya and 148 Ybert, 1991). 149 In Lago da Serra Negra (19 ° 00 S – 46 ° 50 W) a dry phase, with 150 cerrado vegetation, is identi fi ed from the early Holocene to 5000 151 yr BP, followed by a more humid climate and by the establish- 152 ment of the semideciduous forest (Oliveira, 1992). 153 In Salitre (19 ° 00 S – 46 ° 46 W), pollen analysis allowed the 154 identi fi cation of two phases of increasing cerrado vegetation dur- 155 ing the late Pleistocene and the Holocene: 11000 – 10000 and 156 5500 – 4500 yr BP. The later interval is characterized by the author 157 as arid. Between 10000 and 8000 yr BP climate would be cold 158 and humid and an Araucaria forest would be present on the site. 159 Between 8000 and 5500, and after 4500 yr BP, the climate was 160 warm and humid, similar to the present, with establishment of the 161 semideciduous forest (Ledru, 1993). Quanti fi cation of charcoal in 162 a soil pro fi le from this same region indicated a more humid per- 163 iod, with decreasing of the fi re frequency, only after c . 3000 yr 164 BP (Vernet et al ., 1994). SOM carbon-isotope analysis and char- 165 coal dating in three pro fi les suggested the existence of a mixture 166 of trees and grasses from the early Holocene to c . 1700 yr BP. 167 After this period until the present, the predominance of C 3 veg- 168 etation (forest) is attributed to an increasing of the humidity in 169 the region (Pessenda et al ., 1996). 170 A dry period during the early Holocene is recognized also in 171 Lagoa Santa (19 ° 30 S – 44 ° 07 W). There, humidity increased 172 after 5400 yr BP and a vegetation similar to the present (mosaic 173 of cerrado , semideciduous forest and gallery forest) was estab- 174 lished after c . 4600 yr BP (Parizzi et al ., 1998). In the same 175 region, at Lagoa dos Olhos (19 ° 38 S – 43 ° 54 W), a dry period 176 with frequent fi res between c . 13700 and 7000 yr BP, charac- 177 terized by cerrada ˜ o and cerrado , was followed by a phase of 178 increasing humidity between 7000 and 4000 yr BP. After 4000 179 yr BP, the climate became even more humid and there was estab- 180 lishment of the semideciduous forest, which still coexisted with 181 cerrado vegetation. Present climatic conditions were reached 182 around 1300 yr BP (Oliveira, 1992). 183 At Lago do Pires (17 ° 57 S – 42 ° 13 W), a dry climatic phase 184 with cerrado vegetation is recorded between 9700 and 8800 yr 185 BP; expansion of the gallery forests indicates a slightly more 186 humid climate between 8800 and 7500 yr BP; then another dry 187 climatic phase is recorded between 7500 and 5500 yr BP. After 188 this time, the vegetation evolved to a cerrada ˜ o which was main- 189 tained until 2700 yr BP, when a more humid climate allowed the 190 establishment of a denser vegetation. Increasing humidity allowed 191 the establishment of the semideciduous forest characteristic of this 192 region at around 1000 yr BP (Behling, 1995). 193 Other authors identi fi ed signi fi cant climatic changes in South 194 America during the late Quaternary (see, for example, Van der 195 Hammen, 1991; Absy et al ., 1991; Servant et al ., 1993). Studies 196 have suggested that in the north of the continent climate was col- 197 der and drier before 10000 yr BP; humid, similar to the present 198 in the interval 10000 – 8000 yr BP; colder and drier at 6000 – 4000 199 yr BP, and similar to the present since 4000 yr BP (Markgraf and 200 Bradbury, 1982). In central and southern Brazil (15 – 35 ° S) three 201 major palaeoenvironmental phases are distinguished during the 202 Holocene: dry climatic conditions between 10000 and 7000 yr 203 BP, increasing moisture levels and a more seasonal climate 204 between 7000 and 4000 yr BP, and establishment of modern con- 205 ditions after 4000 yr BP, with development of the cerrado veg- 206 etation to the north, semideciduous forest in the centre and Arau- 207 caria forest to the south of the area considered (Ledru et al ., 208 1998). In the eastern Amazonia and in central Brazil frequent for- 209 est fi res and an opening of the vegetation are noticed between 210 8000 and 4000 yr BP, indicating a dry climatic phase (Turcq 211 et al ., 1996). After 4000 yr BP the climate was more humid, with 1 R. Scheel-Ybert et al. : Holocene palaeoenvironmental evolution of Brazilian savanna soils 75 23743744745 010 000200030005000700060008000100040009000 S a l i t r e ( A ) L a g o d a S e r r a N e g r a L a g o a S a n t a L a g o a d o s O l h o s L a g o d o P i r e s t h i s s t u d y P i r a c i c a b a 1 9 º 0 0 ’ S / 4 6 º 4 6 ’ W 1 9 º 0 0 ’ S / 4 6 º 5 0 ’ W 1 9 º 3 0 ’ S / 4 4 º 0 7 ’ W 1 9 º 3 8 ’ S / 4 3 º 5 4 ’ W 1 7 º 5 7 ’ S / 4 2 º 1 3 ’ W 2 2 ° - 2 3 ° S / 4 7 ° - 4 9 ° W 2 2 ° 4 3 ’ S / 4 7 ° 3 8 ’ W S a l i t r e ( B ) 1 9 º 0 0 ’ S / 4 6 º 4 6 ’ W S a l i t r e ( C ) 1 9 º 0 0 ’ S / 4 6 º 4 6 ’ W EW presentconditions drier climatearidmore humid cCf acgf cf f f f f cccccCcgCcgcCcf fccCfC 746747 Figure 1 Summary of climatic changes documented for the cerrado and semideciduous forest phytogeographical zone in southern and central Brazil. Sa ˜ o 748 Paulo State: Piracicaba (Pessenda et al ., 1998); Minas Gerais State: Lago da Serra Negra (Oliveira, 1992); Salitre: A (Ledru, 1993); B (Vernet et al ., 749 1994); C (Pessenda et al ., 1996); Lagoa Santa (Parizzi et al ., 1998); Lagoa dos Olhos (Oliveira, 1992); Lago do Pires (Behling, 1995). c = cerrado sensu 750 amplo ; C = cerrada ˜ o; f = semideciduous forest; g = gallery forest; a = Araucaria forest. 212 forest reconstitution in many areas (Turcq et al ., 1996; Pessenda 213 et al ., 2001; Freitas et al ., 2001). 214 Regional setting 215 In this study, samples from sites Botucatu (22 ° 51 S – 48 ° 29 W), 216 Jaguariuna (22 ° 40 S – 47 ° 10 W), Anhembi (22 ° 45 S – 47 ° 58 W) 217 and Pirassununga (22 ° 02 S – 47 ° 30 W), in the Sa ˜ o Paulo State 218 (Figure 2), were analysed for anthracology, soil isotopic compo- 219 sition ( 13 C) and radiocarbon dating. 220 The study area is located in the southeastern part of the cerrado 221 phytogeographical zone, at the transition to the semideciduous 222 forest zone. Both vegetation types occur under a tropical seasonal 223 climate. The average annual precipitation in most areas of this 224 region ranges from 1000 to 1750 mm, the average annual tem- 225 perature is 20 – 26 ° C, and the dry season lasts 5 – 6 months in the 226 cerrado and 3 – 5 months in the semideciduous forest (Nimer, 227 1989). 228 Semideciduous forests contain trees up to 30 m high with an 229 important component of understorey vegetation. Hymenaea , 230 Copaifera , Peltophorum , Astronium and Aspidosperma are some 231 of the dominant genera in this formation (IBGE, 1992). 752753754 0º 23 º 27 ’ S ã o PauloState22 º 24 º 20 º 51 º 49 º 47 º 45 º + + + ++ + ++++ + 3421 755756 Figure 2 Study area in the Sa ˜ o Paulo State, Brazil. (1) Botucatu; (2) 757 Jaguariuna; (3) Anhembi; (4) Pirassununga. 1 HOL: the holocene 2 16-10-02 07:35:00 Rev 16.04x HOL$$$596P 232 The cerrado (Brazilian savanna) is a vegetation type of low 233 trees with twisted branches and stems, growing in the midst of 234 herbaceous and dwarf woody plants. In fact, the cerrado is a veg- 235 etation gradient that includes fi ve principal physiognomies 236 (Coutinho, 1990). The cerrada ˜ o is a forested formation with trees 237 up to 15 m high, without a shrubby understorey but with a her- 238 baceous stratum in tufts. Arboreal species such as Curatella amer- 239 icana , Qualea grandiflora , Q . parviflora , Caryocar brasiliense , 240 Bowdichia virgilioides and Stryphnodendron barbatiman are typi- 241 cal of this formation. Although the cerrado has typical arboreal 242 vegetation, the cerrada˜ o has a fl ora composition that is partially 243 the same as the semideciduous forest (Leita ˜ o Filho, 1992). The 244 cerrado sensu stricto is characterized by sparse trees and shrubs 245 up to 5 m high. Its fl oristic composition is similar to the cerrada˜ o , 246 but the trees are smaller and sparser, while the understorey veg- 247 etation of shrubs and grasses is more signi fi cant. Its most typical 248 species are Stryphnodendron adstringens and Dimorphandra 249 mollis . The campo cerrado and the campo sujo are intermediate 250 savanna formations consisting of grassland with scattered rachitic 251 woody plants, without an arboreal cover. Representative taxa are 252 Andira humilis , Byrsonima spp. and species of Palmae, Composi- 253 tae and Malvaceae. The campo limpo is basically a grassland. 254 At present, poor patches of secondary semideciduous forest, 255 surrounded by agro-pastoral fi elds, exist in Botucatu, Jaguariuna 256 and Anhembi. At Botucatu, the arboreal stratum is dominated by 257 Anadenanthera cf. macrocarpa (Leguminosae), and at Anhembi 258 by Peschiera fuchsiaefolia (Apocynaceae). In Jaguariuna, the very 259 open forest is dominated by Eucalyptus exogenous species. 260 Bauhinia and Piptadenia (Leguminosae) are also well rep- 261 resented. Present vegetation in Pirassununga is campo-cerrado . A 262 gallery forest exists near the sampling site. 263 Material and methods 264 Charcoal and soil samples from Botucatu (BOT and BOT II), Jag- 265 uariuna (JAG and JAG II) and Anhembi (PIN) were collected in 266 trenches 1.0 m 2.0 m large, up to 2.4 m in depth. Soils were 267 sampled by collecting up to 10 kg of material in 10 cm layers. 268 Samples deeper than 240 cm were drilled from the base of the 1 76 The Holocene 13 (2003) 23269 trenches using an auger. Botucatu samples were taken in two 270 trenches, at the top of two slopes c . 1500 m apart, at the Fazenda 271 Experimental Lageado, Faculty of Agronomy Sciences, Sa ˜ o Paulo 272 State University (UNESP). In Jaguariuna, samples were collected 273 from two trenches located at the top and middle parts of a slope 274 separated by c . 75 m, at the Centro Nacional de Pesquisa de Moni- 275 toramento e Avaliac ¸ a ˜ o de Impacto Ambiental (Embrapa). 276 The sample from Pirassununga (EMAS) was collected from a 277 single layer of c . 20 cm, between 180 and 200 cm depth from a 278 trench 5 m long (Coutinho, 1981). Sampling was undertaken in 279 an area of Ferrovia Paulista , along the road between Pirassununga 280 and Cachoeira das Emas. 281 In the laboratory, soil samples from Botucatu, Jaguariuna and 282 Anhembi were dried and sieved through a 2 mm mesh. 13 C 283 analysis was carried out at the Laboratory of Environmental Iso- 284 topes, University of Waterloo (Canada) and at the Laboratory of 285 Stable Isotopes, Center for Nuclear Energy in Agriculture 286 (Piracicaba, Brazil). Results are expressed relative to the inter- 287 national standard PDB. Precision for three determinations was 288 0.2 ‰ . 289 Charcoal fragments present in the soil were hand-collected. 290 Large charcoal samples ( 10 g) were subjected to acid-alkaline- 291 acid treatments for removal of resins, fulvic acids and lignin. After 292 drying, dating was carried out at the 14 C Laboratory of the Center 293 for Nuclear Energy in Agriculture, using the benzene synthesis 294 and liquid scintillation counting method (Pessenda and Camargo, 295 1991). Small charcoal samples ( 2 g) were entirely combusted 296 and CO 2 samples sent for AMS dating at the Isotrace Laboratory 297 (Toronto, Canada). Results are representative of the mean char- 298 coal age for each 10 cm layer. 299 The remaining fragments were examined for anthracological 300 analysis under a re fl ected light microscope. Charcoal pieces were 301 c . 1 cm of average size in all samples. Most of the charcoal was 302 partially vitri fi ed ( fi bre cells were ‘ melted ’ , while parenchyma 303 cells seemed normal). Systematic identi fi cation was carried out 304 with the help of a program for computer-aided identi fi cation, 305 coupled to a data bank of anatomical features from extant and 306 fossil charcoal (Scheel-Ybert et al ., 1998), and by comparing the 307 fossil material with a reference collection of charred wood and 308 with descriptions and photographs from the literature (Gregory, 309 1994). 310 The results are presented as presence/absence of taxa. Cumulat- 311 ive histograms indicate the number of cerrado and forest taxa. 312 Taxa that present species on both formations were classi fi ed as 313 ‘ forest/ cerrado ’ . 314 Results and interpretation 315 Most of the anthracological samples analysed at each site present 316 only a small number of fragments and few taxa. Results must 317 therefore be considered carefully. 318 In Botucatu, pro fi le BOT covers the last 8000 years (Figure 3). 319 It shows an important contribution of cerrada ˜ o taxa in the lower 320 levels – Vochysiaceae ( Qualea sp., Vochysia sp.) are typical of 321 the cerrado , but particularly frequent in the cerrada ˜ o – and an 322 increasing contribution of forest taxa ( Sebastiania sp., Cryptoca- 323 ria sp., Acacia sp.) in the upper levels (0 – 80 cm). Some forest 324 taxa are also present between 140 and 190 cm ( c . 6000 to 5500 325 yr BP). 326 Pro fi le BOT II covers the last 6700 years (Figure 4). Typical 327 forest taxa are present only in the lower levels (120 – 200 cm), 328 from c . 6500 to 5700 yr BP, in a period when these taxa are also 329 present in BOT. Open cerrado taxa are rare all along the pro fi le, 330 but cerrada ˜ o elements are abundant between 40 and 170 cm depth 331 ( c . 6000 to 3000 yr BP). ‘ Forest/ cerrado ’ taxa are present in the 332 majority of the samples. The most important, in the upper levels, 1 HOL: the holocene 2 16-10-02 07:35:00 Rev 16.04x HOL$$$596P 759760761 Forest Forest / Cerrado Cerrado T a x o n 14 C age Level P i p t o c a r p h a a f f . m a c r o p o d a S e b a s t i a n i a s p C r y p t o c a r y a s p c f . A c a c i a / A c a c i a s p L e g u m i n o s a e s p p L u e h e a s p U n i d e n t i f i e d P e l t o g y n e s p D i m o r p h a n d r a s p Q u a l e a s p p V o c h y s i a s p c f . V o c h y s i a c e a e N u m b e r o f c h a r c o a l p i e c e s 0-10 ** * 9 30-40 ** * 6 40-50 * * 12 50-60 * * 12 3040 ± 180* 70-80 * * * 12 4150 ± 110 4630 ± 80 4830 ± 70 5500 ± 70 120-130 * * 22 5560 ± 80 140-150 * * * 15 150-160 * * 14 160-170 * ** 20 170-180 * 16 6080 ± 300* 180-190 * 4 200-210 ** 10 210-220 * * 16 8000 ± 430* 220-230 * * 5 123 762763 Figure 3 Botucatu (pro fi le BOT). Results of anthracological analysis 764 (presence of taxa in each sample) and number of charcoal pieces analysed 765 per sample. Histograms indicate the number of taxa from each vegetation 766 type ( forest; a forest or cerrado ; ` cerrada ˜ o (arboreal savanna); 767 cerrado sensu amplo ). *AMS 14 C dates. 333 are Croton and Luehea , genera that have several species typical 334 of semideciduous and gallery forests. 335 Anthracological results from both sites point to the existence 336 of forested vegetation in Botucatu during all the studied period, 337 with an important contribution of semideciduous forest and cerra- 338 da ˜ o taxa. 339 Soil organic matter 13 C analysis (Figure 5) complements this 340 approach. In BOT, 13 C values vary between – 24.7 ‰ (210 – 220 341 cm) and – 26.3 ‰ (surface level), indicating arboreal vegetation 342 through the entire pro fi le. In BOT II, 13 C values fl uctuate from 343 – 22.3 ‰ (lower levels) to – 26.1 ‰ (surface). This indicates the 344 predominance of arboreal C 3 vegetation. In the lower levels, 13 C 345 enrichment may be explained by the SOM isotope fractionation 346 (Nadelhoffer and Fry, 1988; Boutton, 1991), but the existence of 347 a more open arboreal vegetation and a greater in fl uence of C 4 348 plants (grasses) is also possible. 349 In Jaguariuna, there are chronological inversions in the two 350 studied pro fi les. In JAG, these inversions are attributed to bio- 351 turbation (Gouveia et al ., 1999; Gouveia and Pessenda, 2000; 352 Gouveia, 2001). There are three series of ages relatively similar 353 (Figure 6), delineating different sedimentary blocks. In JAG II, 354 besides the biological activity, the inversions are attributed to the 355 colluvial nature of the deposit (Gouveia et al ., 1999; Gouveia and 356 Pessenda, 2000; Gouveia, 2001). Indeed, there is in JAG (situated 357 upstream) a possible sedimentary hiatus between 180 and 200 cm, 358 since the date 6240 yr BP (level 170 – 180 cm), is followed by 359 9120 yr BP (level 200 – 210 cm). This material is found in JAG 360 II (situated downstream) between 120 and 240 cm, where several 361 dates cluster around 8000 yr BP. We assume this material can 362 be used for palaeoenvironmental reconstruction, because colluvial 363 source is very near and, in consequence, charcoal pieces re fl ect 364 the local vegetation. In both cases (JAG and JAG II), 13 C curves 365 present a characteristic zigzag trend which suggests that soil 366 material was reworked (Schwartz, personal communication). 367 However, chronological inversions do not invalidate the palaeo- 368 environmental interpretations proposed, because the general 369 trends of climatic evolution may be inferred from the different 1 R. Scheel-Ybert et al. : Holocene palaeoenvironmental evolution of Brazilian savanna soils 77 23770771772 Habitat Forest Forest / Cerrado Cerrado T a x o n 14 C age Level S e b a s t i a n i a s p C r y p t o c a r y a s p c f . A c a c i a / A c a c i a s p C r o t o n s p A n a d e n a n t h e r a s p L e g u m i n o s a e c f . L e g u m i n o s a e / V o c h y s i a c e a e G u e t t a r d a s p L u e h e a s p P a l m a e U n i d e n t i f i e d P e l t o g y n e s p Q u a l e a s p V o c h y s i a s p c f . V o c h y s i a c e a e G r a m i n e a e N u m b e r o f c h a r c o a l p i e c e s 0-10 * * 8 40-50 * * * * 5 3080 ± 70* 50-60 * 2 60-70 * 6 70-80 * 5 80-90 * * * 8 4630 ± 80* 100-110 * * 10 110-120 **** 8 5660 ± 270* 120-130 * * 8 130-140 * * 5 140-150 * 6 150-160 * * * 9 160-170 * ** 18 170-180 * * * 5 190-200 * * 7 200-210 * 4 6690 70* 1234 773774 Figure 4 Botucatu (pro fi le BOT II). Results in anthracological analysis (presence of taxa in each sample) and number of charcoal pieces analysed per 775 sample. Histograms indicate the number of taxa from each vegetation type (see legend for Figure 3). *AMS 14 C dates. 777778779 BOTBOT IIPINJAGJAG II cm ‰ PDB-27 -25 -23 -21 -19 -17050100150200250300350 4320463080006690463091205580807075804290252030405500608030806330463080006690463091205580807075804290252030405500608030806330 780781 Figure 5 Soil organic matter 13 C variation (in % PDB) in Botucatu 782 (BOT and BOT II), Jaguariuna (JAG and JAG II) and Anhembi (PIN) as 783 function of soil depth. 14 C ages are given for each curve. 370 sedimentary blocks, among which the chronological sequence is 371 coherent. 372 In the JAG pro fi le, anthracological results cover the last 9000 373 years (Figure 6). Typical forest taxa are restricted to the upper 374 part of the pro fi le (0 – 30 cm), probably after c . 3000 yr BP. Taxa 375 typical of the cerrado ( Andira , Bowdichia , Qualea , etc.) are abun- 376 dant in the lower levels (80 – 260 cm). ‘ Forest/ cerrado ’ taxa (e.g., 377 Tabebuia sp., Cassia sp., Leguminosae) are present throughout 378 the pro fi le. 379 JAG II pro fi le covers the last 8000 years (Figure 7). Forest taxa 380 are also restricted to the upper levels (0 – 40 cm), probably after 381 c . 3000 yr BP. Between 50 and 120 cm ‘ forest/ cerrado ’ taxa pre- 382 dominate (various Leguminosae species), while in lower levels 383 (150 – 240 cm), around 8000 yr BP, there is practically a single 384 species (cf. Andira ), typical of the cerrado . 385 These results suggest the presence of cerrado on this site during 1 HOL: the holocene 2 16-10-02 07:35:00 Rev 16.04x HOL$$$596P 386 the late and mid-Holocene, and the establishment of a forested 387 vegetation after c . 3000 yr BP. 388 The 13 C analysis of soil samples (Figure 5) shows in both 389 pro fi les the predominance of forested formations after c . 3000 yr 390 BP, with values between – 23.0 ‰ and – 20.7 ‰ . Both curves show 391 higher 13 C values between c . 3000 and 5500 yr BP, slightly 392 decreasing values from c . 5500 to 6500 yr BP, then increasing 393 values from c . 6500 yr BP to the base. The lower levels present 394 13 C values characteristic of a vegetation type where C 3 and C 4 395 plants coexist ( – 20 ‰ to – 17.9 ‰ ), i.e., an open cerrado with a 396 great proportion of grasses. In both pro fi les, higher 13 C values 397 in levels lower than 200 cm depth suggest the predominance of C 4 398 plants (grasses), i.e., probably a campo-sujo , prior to 8000 yr BP. 399 At Anhembi, anthracological results of the PIN pro fi le cover 400 the last 8000 years (Figure 8). The presence of Securinega aff. 401 guarayuva in the surface sample indicates the existence of a for- 402 ested formation in a recent period. The Gramineae charcoal in the 403 same sample suggests it can be a cerrada ˜ o . Bauhinia sp. is fre- 404 quent in the upper part of the pro fi le. Although it can occur in 405 the open cerrado , this genus is mostly characteristic of a forested 406 vegetation. Dimorphandra sp., typical of the cerrado , and Qualea 407 sp., particularly frequent in the cerrada ˜ o , occur in the lower part 408 of the pro fi le, from 80 cm to the base. 409 These results point out to the existence of an arboreal veg- 410 etation, possibly a cerrada ˜ o , from the base of the anthracological 411 sequence until c . 3000 yr BP, when the establishment of a dense 412 cerrada ˜ o or a forest occurred. 413 SOM 13 C analysis also indicates the predominance of C 3 414 plants in the whole period (Figure 5). In lower levels (250 – 415 300cm), SOM 13 C values are more enriched ( – 22.3 ‰ up to 416 – 23.0 ‰ ). In addition to the SOM fractionation, these enriched 417 values can be related to the existence of a more open arboreal 418 vegetation and to the in fl uence of C 4 plants at the early Holocene, 419 probably due to the presence of a drier climate. From 240 cm to 420 the soil surface, and particularly in the upper 50 cm (after 3000 421 yr BP), 13 C values are more depleted (up to – 24.3 ‰ ), suggesting 422 the progressive establishment of a forested formation, probably 423 related to a more humid climate. 424 The Pirassununga sample (EMAS) shows a clear predominance
