Last Millennium Environmental Changes and Climate Inferences in the Southeastern Atlantic Forest, Brazil

Non-profit academic project, developed under the open access initiative " main " — 2010/8/9 — 20:09 — page 717 — #1 ABSTRACT This study presents paleoenvironmental data based on pollen, elemental and isotopic compositions of organic matt (TOC, N, δ 13 C and δ 15 N) and 14 C dating of 170 cm lake sediment record. Samplings have been made in Lagoa Grand at Parque Estadual Turístico do Alto Ribeira – PETAR, Southern São Paulo State, Southeastern Brazil. The variatio in relative frequencies (in percentage) of arboreal pollen along the core range between 40 and 80%. The δ 13 C valu ranged from –23 to –30 and C/N of ∼10 to 15, indicating the contribution of terrestrial C 3 plants and algae in t sediment organic matter. The δ 15 N results presented values from 3 to 4.5, also suggesting a mixture of algae an terrestrial C 3 plants. The 14 C dating indicates modern age for the shallow horizons to ∼1030 BP at the base of the cor A probable wetter climate in the period of ∼370 BP to ∼340 BP was inferred from the data set, which corresponds a part of the period covered by the Little Ice Age (LIA).


INTRODUCTION
Lake systems are diverse, and the sources and alterations of organic matter are geographically and temporally variable.Nonetheless, useful generalizations can be made about the different kinds of elemental, isotopic and palynologycal proxies that provide evidence of the origins and depositional histories of sedimentary organic matter and hence paleoenvironmental conditions (Meyers 2003).
Pollen records obtained from lacustrine sediments have been used for palaeoenvironmental reconstruction studies during the late Quaternary and the Holocene in several regions of South America and Brazil (Absy and van der Hammen 1976, Absy et al. 1991, Roth and Lorscheitter 1993, van der Hammen and Absy 1994, Colinvaux et al. 1996, Ledru et al. 1996, 2001, 2002, Behling 1995a, b, 1997a, b, 1998, 2002, Salgado-Labouriau et al. 1998, Sifeddine et al. 2001, 2003, Saia 2006).Considering the last 3000 BP, most of the records indicate increased moisture conditions, although differences in climatic signal are recorded according to the site location.However, no data have been produced in high resolution emphasizing the potential use of an interdisciplinary study, involving different proxies and techniques for the last millennium.
Interdisciplinarity is fundamental for improving interpretations.In most cases, analysis of the bulk composition of sedimentary organic matter will provide only general information.In fact, according to Meyers (2003), Sifeddine et al. 2004 andOrtiz et al. 2004, interpretation is not simple, in most cases, organic matter is a mixture of components from many sources and with variable degrees of preservation.Despite diagenesis, partial reworking and alteration of the original organic matter when sinking to the lake bottom, sedimentary organic matter retains important information on its origin, transportation and deposition, which helps paleoenvironmental reconstructions (Ortiz et al. 2004).
The δ 13 C values of lacustrine sedimentary organic matter may potentially reflect changing proportions of different types of vegetal matter (e.g.C 3 vs.C 4 plant abundance), as well as the presence of aquatic organisms, e.g.phytoplankton, which have similar isotopic signatures (Meyers 1997).During photosynthesis, phytoplankton preferentially utilizes the 12 C isotope, leading to typical δ 13 C values of -20 in its organic matter (Meyers 2003, Schidlowski et al. 1994).Thus, when the phytoplankton utilizes the dissolved CO 2 in equilibrium with the atmosphere, it is usually an isotope indistinguishable from organic matter produced by terrestrial C 3 plants.The correlation with C/N (total organic carbon and total organic nitrogen ratio) values is useful.C/N ratio of lacustrine sediments can be used to distinguish two main types of organic matter: (i) organic matter without cellulosic structure, originated from algae and phytoplankton, characterized by ratios between 4 and 10; and (ii) organic matter with cellulosic structure, produced by terrestrial plants, characterized by ratios ≥ 20 (Meyers andIshiwatari 1993, Meyers 1994).This correlation should thus lead to discrimination between organic matter from freshwater algae and organic matter from C 3 or C 4 type plant sources.Analysis of stable carbon isotopes and the C/N ratio of organic matter preserved within lake sediments can provide important information about paleoenvironments, which can serve to complement pollen-based paleovegetation studies.
Although not widely used as a paleolimnological proxy as δ 13 C values, nitrogen isotopic compositions (δ 15 N) can similarly help to identify sources of organic matter of lakes and to reconstruct past productivity rates (Lent et al. 1995, Herczeg et al. 2001).The isotopic difference among sources of nitrogen is roughly preserved in the δ 15 N values of organic matter from algae (+8.5 ) and from C 3 terrestrial plants (+0.5 ) (Peterson and Howarth 1987).
This study shows the first lake sediment record regarding the last millennium in Brazil.The main objective is the reconstruction of the vegetation changes in high resolution, with climate inferences that occurred in Southeastern Brazil during the last millennium.The data were obtained from a sediment record using pollen, geochemistry (total organic carbon -TOC and total nitrogen -TN) and isotope (δ 13 C and δ 15 N) analysis, in addition to a detailed chronological frame based on radiocarbon dates.We did inferences with the period of the Little Ice Age (LIA) that occurred among 14 th and 19 th centuries and the Maunder Minimum (1645-1715 AD).

STUDY AREA
Lagoa Grande is a small lake located at PETAR (Parque Estadual Turístico do Alto Ribeira; (24 • 31 59.1 S, 48 • 39 45.0 W, 364 m elevation), Vale do Ribeira, South of São Paulo State (Fig. 1).This park is a conservation unit on the left bank of the Ribeira River, approximately 350 km from the city of São Paulo, and 100 km from the Atlantic coastline.The lake developed within a centripetal drainage basin on karst topography.The catchment area is about 10 000 m 2 and the lake itself occupies an area of about 150 m 2 , with a present maximum water depth of 1 m.The lake, which is situated in a remote region, contains largely untouched natural sediments.Bedrock geology consists of a sequence of low-grade metasedimentary rocks of the Late Proterozoic Açungui Group, which is formed by fine-graded limestones interbedded mainly with metasiltites and phyllites (Campanha and Sadowski 1999).The major minerals in these rocks are calcite, quartz, mica and chlorite.The clayey soils are classified as cambisols.
The mean annual temperature is 20 • C, with minimum and maxima of 14 and 27 • C, respectively.The average annual precipitation is around 1600 mm (Karmann 1994), with low rainfall from June to August.
The natural vegetation around Lagoa Grande is the Atlantic Dense Ombrophilous Forest (forest hillside) and lies within the Atlantic Forest domain of vegetation.The modern vegetation in the lake margins is occupied mainly by trees of the family Cecropiaceae, Euphorbiaceae, Flacourtiaceae, Lauraceae, Leguminosae,

SAMPLING AND ANALYTICAL PROCEDURES
A sediment core was collected using a vibracorer (Martin et al. 1995) and a 3-inch diameter aluminum tube with 3 m length.The 170 cm long core was collected close to the lake margin and sub-sampled in 2 to 10 cm intervals for analysis.

C AND N ISOTOPES
The sediment samples separated of leaves, roots, etc., were treated with HCl 4% to eliminate carbonates, washed with distilled water up to pH 6, dried at 50 • C and homogenized.These samples were used for total organic carbon and nitrogen analyses, carried out at the Stable Isotope Laboratory of Center for Nuclear Energy in Agriculture (CENA/USP).The results were expressed in percentage of dry weight, with analytical precision of 0.09 and 0.07%, respectively.The 13 C and 15 N results were expressed as δ 13 C and δ 15 N with respect to the PDB standard and atmospheric air, respectively, using the conventional notations: where R S1 and R S2 are, respectively, the 13 C/ 12 C and 15 N/ 14 N ratios in the sample, R PDB the 13 C/ 12 C ratio for the international standard (PDB) and R Air the 15 N/ 14 N ratio for the atmospheric air.The results were expressed in delta per mil notation, with analytical precision better than 0.2 (Pessenda et al. 2004(Pessenda et al. , 2005)).
For radiocarbon analyses, wood fragments and bulk sediment samples were subjected to physical treatment (removal of vegetable remains, etc.) and treated with HCl 4% for 4 h at 60 • C to remove carbonates, fulvic acids and resins.After successive washes (up to pH 5) with deionised water and drying to 60 • C for 48 h, the samples were combusted and the purified CO 2 sent to the Isotrace Laboratory, University of Toronto, Canada, for Accelerator Mass Spectrometry (AMS).Radiocarbon ages (Table II) are expressed as 14 C BP (years Before Present) and in cal BP (calibrated years Before Present) (Stuiver et al. 1998) and with precision of ±2σ .

PALINOLOGY
The pollen analyses were performed each at 10 cm interval along the 160 cm long core.Chemical treatment of approximately 5 g of sediment was performed for pollen extraction based on Faegri and Iversen (1989).Sample treatment followed the Lüber technique using 70% HF for 18 h, followed by 50% HCl in hot water bath and by 10% KOH solution.Palynomorphs were separated using ZnCl 2 and the residue fraction of 50 μL mounted on microscope slides in silicon oil; pollen counting was performed under 1000× magnification.Approximately 300 pollen grains of tree and herb taxa were counted per sample.The reference collection of Dr. Marie-Pierre Ledru and pollen atlas were used for pollen identification (Absy 1975, Markgraf and D'Antoni 1978, Colinvaux et al. 1999).Frequencies are expressed as percentages of the total sum of arboreal pollen (AP), nonarboreal pollen (NAP) and undetermined types, with aquatic taxa, fern spores and Cyperaceae excluded from the total sum.Relative frequencies of spores and aquatic taxa were calculated in relation to the total AP and NAP sums.The determination of pollen concentration was based on the method of Cour (1974), following the mathematical equation: where: v = volume of the residue used in the feature of pollen slides; V = total volume of the residue after a chemical treatment; I = total length observed in the slides during the counting of grains; L = medium length used in each slide, defined for the longitudinal lines between the residue and the glue line that was traced in the slide.
The residue percentage in each sample is obtained from this calculation.From the fraction observed in each residue (f%), it is possible to calculate the grain concentration per gram of sediment.This calculation is obtained from the next equation:

RESULTS
The core sediment consists of grey dark clay with massive structure at the bottom, changing upward to planeparallel lamination of brown and grey clay mud, to light brown mud with massive structure and red spots (iron/ hydroxide/oxide), to plane-parallel lamination of brown and grey clay mud, to light brown mud with massive structure and red spots, and to light brown and massive structure without bioturbation at the surface (Fig. 2, Table I).
The ages of terrestrial macrofossils (wood fragments) and sediment samples ranged from 1030 ± 60 BP at 171-168 cm to modern ages (8-6 cm sediment depth), with increasing ages downward (Fig. 2, Table II).At 126-116 cm a wood fragment presented an age of 680 ± 80 BP, and at 124-122 cm the organic sediment sample an age of 550 ± 50 BP.Similar results were obtained for the wood fragment (250 ± 60 BP) at 65-63 cm, and for the organic sediment sample (320 ± 60 BP) at 64-62 cm.In order to provide age estimation relative to sediment depth, a continuous age scale based on interpolation of 14 C ages versus sediment depth is presented in Figure 3.
The results of the pollen analysis are presented in Figures 2 and 4. Figure 2 shows changes in arboreal pollen and spores frequencies and the concentration values of number of pollen grains per gram of wet sediment.The main pollen types found in the sediment of Lagoa Grande are associated to Cyperaceae and Poaceae, which record the vegetation at the lake margin, and Alchornea, An Acad Bras Cienc (2010) 82 (3)   Cecropia, Myrtaceae and Moraceae, and Podocarpus and Weinmannia, which are elements of Atlantic Ombrophilous forest and of forest of altitude, respectively.The predominance of arboreal taxa is observed along all the sediment core (Fig. 2).From 170-140 cm depth (∼1000 BP to ∼750 BP), arboreal pollen frequencies are high, with ∼70% and a peak of fern spores of ∼20% is being observed.Cecropia, Moraceae, Myrtaceae and Poaceae are also recorded.The frequency of Weinmannia and Alchornea is high (Fig. 4).Weinmannia shows the highest frequency of this interval, with 7%, and Podocarpus appears around 150 cm.In the interval 155 cm to 145 cm (∼950 to 800 BP), the TOC content increased from 1.8 to 5.0%, the C/N ranged from 12.6 to 14.9, the δ 13 C values stayed around -28.5 and δ 15 N values ranged from 3.3 to 3.8 (Fig. 2).The results are indicative of the predominance of organic matter from terrestrial C 3 plants.From 145 cm to 135 cm (∼800 BP to 720 BP), the TOC values decreased from 5% to 0.5%, the C/N from 14.9 to 10, the δ 13 C values enriched from -28.5 to -23.7 , and δ 15 N values in-creased from 3.5 to 4.5 , which is indicative of more significant presence of phytoplankton (algae) (Meyers 1994).This could be associated with the increase of the lake level due to more humid conditions than the previous period.From 135 cm to 70 cm depth (∼720 BP to ∼320 BP), the elemental and isotopic analyses (TOC ranged from 0.8 to 5.1%, the C/N ratio from 10.1 to 15.3, δ 13 C values from -23.7 to -29 , and δ 15 N from 4.25 to 3 ) indicated a mixture of organic matter (C 3 plants and phytoplankton), with higher influence of terrestrial C 3 plants.From 135 cm to 80 cm (∼720 BP to ∼350 BP), a decrease of Alchornea (∼2%) and Cecropia (∼0.5%) was observed, and the Podocarpus disappeared till ∼110 cm (∼420 BP).From 80 to 70 cm (∼350 BP to ∼320 BP), the spore frequencies presented the highest value (up to ∼50%) (Fig. 4), which is coincident with the significant increase of Weinmannia (4%), Podocarpus (0.8%), Cyperaceae (∼4 to 5%) and Myrtaceae (28%) at 90 cm (∼360 BP).From 70 to 50 cm depth (∼320 BP to ∼280 BP), the TOC increased up to 8%, the C/N values to 15 being associated with more An Acad Bras Cienc (2010) 82 (3) depleted δ 13 C (-29 ), and δ 15 N ranged between ∼3 and 3.5 , indicating a mixture of organic matter (C 3 plant and algae).A decrease of arboreal pollen frequency from ∼60% to 55% was observed at ∼60 cm (∼300 BP), simultaneously with Alchornea (4%), Myrtaceae (10%), Podocarpus (0.25%) and Weinmannia (2%).The spore frequency was around 10% (Fig. 4).
The distribution of C/N versus δ 13 C values (Fig. 5) indicates the presence of two sources of organic matter.Very few samples indicate a more significant presence of algae and most of the samples represented a mixture of algae and terrestrial C 3 plants, probably associated with the variation of the lake level.

DISCUSSION
The pollen record of Lagoa Grande indicates no drastic changes in the arboreal vegetation, suggesting that the climate of the area, at least in the last millennium, remained humid in most of the period, probably without longer dry seasons.The pollen types found from ∼1000 BP to ∼400 BP are characteristic of dense and humid forest represented by Alchorneae, Cecropia, Myrtaceae and Moraceae, of local plants on the lake margins such as Cyperaceae and Poaceae, of spores with frequency of up to ∼15-20%, and of elements of forest of altitude indicated by the low presence (<0.5%) of Podocarpus at ∼820 BP and 380 BP, and more significant frequency (∼4% and ∼6%, respectively) of Weinmannia.
An increase in the frequency of spores up to 50% is observed at ∼350 BP.One hypothesis could be linked with the oxidation and degradation of more fragile pollen types, resulting in a relative enrichment of more resistant pollens, such as spores (Colinvaux et al. 1999).However, oxidation signs or bad preservation were not observed in the pollen content.The second hypothesis is associated with the decrease of the water level, probably related with a warmer and/or drier climate period, and with consequent wider exposure of the lake margins.Fern spores and Cyperaceae, usually found in lake margins, simultaneously increased their frequencies, which were recorded between ∼400 and 320 BP (Fig. 4).A third hypothesis is also possible: an increase in both humidity and erosion process, with the consequent input of fern spores and Cyperaceae, usually found in humid environments, into the lake basin.Geochemical interpretation presented by Oliveira et al. 2009 supports this last hypothesis.According to these authors, the composition of the Lagoa Grande sediments can be described in terms of a dominant geogenic component, an association of mica, kaolinite and goethite with elements of low mobility during weathering (Ba, Cs, Rb, Sr), and with elements commonly assumed to be immobile (Ti, Cr, Sc, Zr, Hf and Ta).This component may have been brought by erosion and removal of deeper soil layers in the catchment area, where these elements are found at the highest abundance and largely derived from the B-horizon of the surrounding soil profiles.In addition, there is a minor biogenic component and related to the O-horizon (shallow layers) of these soils, which reflects the association of chlorite with the transition metals (Co, Zn and Cu).This probably represents the input into the lake of material eroded from the uppermost organic horizons of the catchment soils.In the O-horizon of soils, many metals are found forming stable organo-metallic compounds.This is particularly the case for Zn and Cu, which are among the most essential elements for plant growth and is accumulated in the humus layer (Blaser et al. 2000).This component is relatively more concentrated in the uppermost (360 BP to present) and lowermost zones (1030 to 730 BP) of the sedimentary column than in the intermediate zone (730 to 360 BP).A possi-  ble cause for this is the variation in the erosion regime in the catchment area.At the onset of the sedimentation (1030 to 730 BP), the input of deeper layers of soil was reduced as a result of the relative catchment stability.Under such conditions, a greater proportion of biogenic elements was accumulated in the sediments.With the increase in the erosion intensity between 730 to 360 BP, deeper layers of the soil profile were successively eroded.The inflow of those materials into the lake led to a relative depletion of the biogenic component of the sediments.This trend was enhanced between 490 to 360 BP.From 360 BP onwards the concentration of the biogenic component in the sediments increased again, suggesting the resumption of more stable conditions in the catch-An Acad Bras Cienc (2010) 82 (3) The changes in the erosion regime can be assumed to be controlled by climate fluctuations.A shift toward increased rainfall, particularly if it occurs in the summer months increasing the seasonality and enhancing erosion in the catchment area.This would increase the supply of geogenic material into the lake, diluting the material from the more organic topsoil.
A link can thus be established between the geochemical signature of the sediments and the climatic conditions at the time of their deposition.During the last millennium, in the Lagoa Grande region, the period inferred from the geochemical data from 730 to 360 BP (wetter), with the wettest conditions between 490 and 360 BP, overlaps approximately that one suggested by pollen and isotopic data (see the following discussion) as a warmer (inference from pollen) and wetter (inferences from both indicators) period of ∼370 to 340 BP.Therefore, the hypothesis of an increase in both humidity and erosion process, with the consequent input of fern spores and Cyperaceae, will be considered.Concerning the elemental and isotope data, it was observed in the interval ∼370 to 340 BP an increase in the concentration of grains/g of sediment, TOC concentration, C/N values, and a depletion of δ 13 C, which is indica-tive of significant input of organic matter of C 3 plant, probably associated with the increase of spores of Cyperaceae and arboreal species as Myrtaceae and Weinmannia, during more humid and erosive period (Fig. 2).In the interval ∼340 to 320 BP, it was observed the lowest values of TOC in the whole sediment profile, as well as the lowest grain per gram of sediment, which could be connected with a small input of the organic matter in the sedimentary basin, possibly associated with a less humid and erosive period.At the same time, a significant increase in Weinmannia frequencies is observed (Fig. 4).A pollen record in Cambará do Sul (29 • 03 09 , 50 • 06 04 Southern Brazil) shows that Weinmannia became a common taxa in the Araucaria forest between 1520 and 1770 AD (∼430 BP and 180 BP) (Behling et al. 2004), suggesting a shift to warmer climatic conditions.These authors associated this interval with a cooler phase (∼550 BP to 200 BP) called Little Ice Age (LIA), one of the most important variations of the last millennium (from the 14 th to the 19 th centuries), which is known from North Hemisphere records in the USA, described by Pederson et al. 2005, and in the Europe by Fontana, 1976, Font 1988, Ramil-Rego et al. 1994, and Luque and Julià 2002.Considering these associations, the record of Lagoa Grande suggests the presence of a warmer and probably wetter climate period (linked to the highest spore frequency and TOC) for the interval 370-340 BP.This can be connected with the LIA period, which has been poorly recorded in South America.Lichenometry dating of glaciers in the Cordilleras in Bolivia, (16 • 17 S, 68 • 06 W; Rabatel et al. 2005) established a chronology for the LIA in that area, with a maximum occurrence in the second half of the 17 th century (∼300 BP) and finishing between 1870 and 1910 AD (∼80 to 40 BP).These authors concluded that the second part of the LIA could have been dry in the Andes, because tropical glaciers retreat significantly when precipitation decreases.
In Southern Peru, a dry climate seems to have prevailed in the 1720-1880 AD (∼230 to 70 BP) period (Thompson et al. 1986, Rabatel et al. 2005).In Argentina, close to Buenos Aires, a cool episode was observed (Iriondo and García 1993) and associated with the LIA period, in which an aridity episode was reported.In the same country, Laguna Mar Chiquita also shows dry conditions during the LIA period (Piovano et al. 2002).
These drier events could be correlated with the Maunder Minimum (1645-1715 AD; ∼305 to 235 BP), described as a climax of the LIA, when the earth gradually warms until the warming is further accelerated by anthropogenic greenhouse gases (Mann et al. 1999, Shindell et al. 2001, Youshimori et al. 2005), mainly considering an error (precision) of ±50 years for each radiocarbon dating in the 400-320 BP, inferred warmer and wetter climate period in this work.

CONCLUSIONS
Environmental conditions were relatively stable with the maintenance of the forest during the last 1000 years and characterized by a combination of pollen (arboreal) and isotope signals (phytoplankton and C 3 land plants), which indicated humid climatic conditions for most of the period.However, the bulk and isotopic analyses of sediment core indicate significant variations, probably associated with changes in the Lagoa Grande water column, eventually linked with a wet and warm period (from ∼370 BP to ∼340 BP), that was associated with the Little Ice Age (LIA) and Maunder Minimum.
sum of grains in one slide; f = fraction observed in the residue; m = original weight used in the chemical treatment of the sample.

Fig. 5 -
Fig. 5 -C/N × δ 13 C. ment area, except for brief episodes of enhanced erosion around 340 BP, 280 BP and 200 to 170 BP.The changes in the erosion regime can be assumed to be controlled by climate fluctuations.A shift toward increased rainfall, particularly if it occurs in the summer months increasing the seasonality and enhancing erosion in the catchment area.This would increase the supply of geogenic material into the lake, diluting the material from the more organic topsoil.A link can thus be established between the geochemical signature of the sediments and the climatic conditions at the time of their deposition.During the last millennium, in the Lagoa Grande region, the period inferred from the geochemical data from 730 to 360 BP (wetter), with the wettest conditions between 490 and 360 BP, overlaps approximately that one suggested by pollen and isotopic data (see the following discussion) as a warmer (inference from pollen) and wetter (inferences from both indicators) period of ∼370 to 340 BP.Therefore, the hypothesis of an increase in both humidity and erosion process, with the consequent input of fern spores and Cyperaceae, will be considered.Concerning the elemental and isotope data, it was observed in the interval ∼370 to 340 BP an increase in the concentration of grains/g of sediment, TOC concentration, C/N values, and a depletion of δ 13 C, which is indica-

TABLE II 14 C ages of organic matter obtained in the Lagoa Grande core.
TO -Isotrace Laboratory, Toronto, Canada; (*) -cal AD.