Spatial and seasonal contrasts of sedimentary organic matter in floodplain lakes of the central Amazon basin

In this study, we investigated the seasonal and spatial pattern of organic matter (SOM) in ﬁve ﬂoodplain lakes of the central Amazon basin (Cabaliana, Janauaca, Canaçari, Miratuba, and Curuai) which have di ﬀ erent morphologies, hydrodynamics and vegetation coverages. Surface sediments were collected in four hydro- 5 logical seasons: low water (LW), rising water (RW), high water (HW) and falling water (FW) in 2009 and 2010. We investigated commonly used bulk geochemical tracers such as the C : N ratio and the stable isotopic composition of organic carbon ( δ 13 C org ). These results were compared with lignin-phenol parameters as an indicator of vascular plant detritus and branched glycerol dialkyl glycerol tetraethers (brGDGTs) to trace the 10 input of soil organic matter (OM) from land to the aquatic settings. We also applied the isoprenoid GDGT (iGDGT) crenarchaeol as an indicator of riverine suspended particulate organic matter (SPOM). Our data showed that during the RW and FW seasons, the surface sediments were enriched in lignin and brGDGTs in comparison to other seasons. Our study also indicated that ﬂoodplain lake sediments primarily consisted 15 of allochthonous, C 3 plant-derived OM. However, a downstream increase in C 4 macrophyte derived OM contribution was observed along the gradient of increasing open waters, i.e. from upstream to downstream. Accordingly, we attribute temporal and spatial di ﬀ erence in SOM composition to the hydrological dynamics between the ﬂoodplain lakes and the surrounding ﬂooded forests.

logical seasons: low water (LW), rising water (RW), high water (HW) and falling water (FW) in 2009 and 2010. We investigated commonly used bulk geochemical tracers such as the C : N ratio and the stable isotopic composition of organic carbon (δ 13 C org ). These results were compared with lignin-phenol parameters as an indicator of vascular plant detritus and branched glycerol dialkyl glycerol tetraethers (brGDGTs) to trace the input of soil organic matter (OM) from land to the aquatic settings. We also applied the isoprenoid GDGT (iGDGT) crenarchaeol as an indicator of riverine suspended particulate organic matter (SPOM). Our data showed that during the RW and FW seasons, the surface sediments were enriched in lignin and brGDGTs in comparison to other seasons. Our study also indicated that floodplain lake sediments primarily consisted 15 of allochthonous, C 3 plant-derived OM. However, a downstream increase in C 4 macrophyte derived OM contribution was observed along the gradient of increasing open waters, i.e. from upstream to downstream. Accordingly, we attribute temporal and spatial difference in SOM composition to the hydrological dynamics between the floodplain lakes and the surrounding flooded forests.

Introduction
Inland waters play a significant role in the global carbon budget. Lakes and rivers are active systems where the transport, transformation and storage of organic carbon (OC) affect the carbon cycle on a landscape and global scale (e.g., Cole et al., 2007;Tranvik et al., 2009;Raymond et al., 2013). In this context, the wetlands, are dynamic the water column reaches the sediment and is finally buried (Devol et al., 1984), the sediments in these lakes are important sinks of carbon (Moreira-Turcq et al., 2004). Most of the sedimentary organic matter (SOM) in freshwater systems is derived from terrestrial vascular plants (Goñi and Hedges, 1992;Moreira-Turcq et al., 2004;Mortillaro et al., 2012). In the Amazon basin, many studies have characterized the OM in the suspended particulate organic matter (SPOM) in the rivers system and in the floodplain lakes and concluded that the main source of OC to the aquatic system is forests and upstream Andean soils (e.g., Quay et al., 1992;Victoria et al., 1992;Hedges et al., 1994;Moreira-Turcq et al., 2004;Aufdenkampe et al., 2007;Mortillaro et al., 2011;Moreira-Turcq et al., 2013;Zell et al., 2013b). However, little is 20 known about the molecular composition of the SOM in the floodplains in general, and in floodplain lakes in particular, and the contribution of the multiple sources of OM (upland soils, flooded forest, aquatic macrophytes, and phytoplankton) remain uncertain (Mortillaro et al., 2011;Zocatelli et al., 2011;Moreira et al., 2014).
The spatiality and the seasonality of the hydrology in the Amazon basin strongly in-Introduction vertical flux of OC was observed in lake Curuai during the falling water season, which is interpreted as the result of a process of concentration of periodically resuspended sediments as lakes are becoming smaller and shallower (Moreira-Turcq et al., 2004). In downstream lakes, higher values of bulk δ 13 C were found in the sediments, when compared to upstream lakes (Victoria et al., 1992). This variability may be explained by 5 the differences in the interfaces between the river and the lakes along the upstreamdownstream transect or in aquatic primary production (mainly aquatic plants), which is more widespread in the open water lakes downstream. A previous study of bulk parameters and fatty acids in the central Amazon basin (Mortillaro et al., 2011) was not conclusive about the sources of SOM in floodplain lakes. Hence, the present work 10 applies multiple biomarkers, namely lignin phenols, branched glycerol dyalkyl glycerol tetraethers (brGDGTs) and crenarchaeol, in addition to the bulk parameters, to disentangle the sources of SOM in floodplain lakes of the central Amazon basin and the role of the spatiality and seasonality in determining the composition of the SOM. Lignin is a recalcitrant organic macromolecule composed of phenolic molecules and 15 produced by vascular plants. The products of CuO degradation of lignin (Hedges and Ertel, 1982) have been widely applied as biomarkers to trace plant material to aquatic systems Bernardes et al., 2004;Aufdenkampe et al., 2007;Kuzyk et al., 2008). It can be an important component of fossil OC in floodplain lakes (Zocatelli et al., 2013) but also a relevant source for the outgassing of CO 2 in the Amazon River Introduction the bulk parameters, analyzed in superficial sediments collected in five floodplain lakes of the central Amazon basin in four hydrological seasons, provides new insights into the link between the hydrology of the Amazon basin to the sources of SOM in floodplain lakes.

Study area 5
The Amazon River is the world's largest river with a drainage basin area of 6.1 × 10 6 km 2 covering about 40 % of South America (Goulding et al., 2003). The mean annual discharge is 200 × 10 3 m 3 s −1 at Óbidos, the most downstream gauging station in the Amazon River (Callede et al., 2010). Rivers within the Amazon drainage basin are traditionally classified according to water color, as well as physical and chemical pa-10 rameters (Sioli, 1950): white water (e.g. Solimões, Madeira and Amazon), black water (e.g. Negro), and clear water (e.g. Tapajós). The total area of wetland is 350 × 10 3 km 2 (Melack and Hess, 2011). Approximately 17 % of the central Amazon basin is subjected to periodical floods. This creates large temporary wetlands, i.e. seasonally flooded forests, woodlands and shrubs, which corresponds to 58 % of the total flooded area 15 during the high water season. Aquatic macrophytes, floating meadow and marsh cover 5 to 8 % of the wetlands and open waters correspond to 12 and 14 % in low and high water seasons, respectively (Hess et al., 2003). Five floodplain lakes were investigated in this study: Cabaliana, Janauaca, Mirituba, Canaçari, and Curuai (Fig. 1a, Table 1). The lakes are located along the Solimões- 20 Amazon river shoreline in a biogeographic gradient of upstream flooded forests to downstream flooded woodlands and open water lakes (Bourgoin et al., 2007;Abril et al., 2014). Cabaliana is a round shape lake surrounded by flooded forests and two sub-regions (Fig. 1b). In the northern region, the Manacapuru River discharges clear water while in the southern region, the white water brought by the Solimões River, Introduction north, and clear water comes through the stream system in the south. Mirituba has a round shape and receives white water from the Madeira River and the Amazon River through a complex drainage system (Fig. 1c). It can be considered as a white water lake surrounded by flooded forests and woodlands, with no significant contribution of black water streams. Canaçari has two well-defined sub-regions (Fig. 1c). In the north-5 ern region, the Urubu River discharges black water and in the southern region, the Amazon River discharges white water. It is surrounded by flooded forests and woodlands. Curuai is the largest lake in the central Amazon basin, mainly surrounded by woodlands and open waters (Fig. 1d). It receives white water from the Amazon River through small channels, apart from the main channel in the eastern side. There are 10 small contributions of black water streams in its most Southeastern part.

Sample collection
Surface (0-2 cm) sediment samples (n = 57) were collected using a grab sampler of 100 cm 3 in lakes Cabaliana, Janauaca, Mirituba, Canaçari and Curuai in the central 15 Amazon basin between Manaus and Santarém ( Fig. 1 20 In each floodplain lake, sediment samples were collected at three stations in each season. However, sometimes only two samples were collected when stations were not accessible during a specific season. Four riverbank sediments and three soils from well above the inundations known as "terra firme" were also collected during the LW season. In addition, four samples of Introduction phytes) were sampled during the HW season in the lakes Janauaca and Curuai. All samples were kept frozen (−20 • C) on the ship and transported frozen to the Universidade Federal Fluminense laboratory (Brazil), where they were freeze-dried.

Bulk geochemical parameters
Total carbon (TC), total nitrogen (TN), and δ 13 C for the samples obtained during the analyses were determined in duplicate with a precision of 0.1 mg C g −1 . TC (wt. %) and correlated very well with TOC (wt. %) with a +0.16 intercept (R 2 = 0.96; p < 0.001; n = 16). This indicates that TC in floodplain lakes sediments investigated was mostly TOC. Therefore, we considered TC as TOC in this study. In order to assess contribution of inorganic nitrogen (NH + 4 + NO − 2 + NO − 3 ) to TN, TN (wt. %) and TOC (wt. %) 20 were correlated (R 2 = 0.89; p < 0.001; n = 57). It showed that a contribution of mineral nitrogen present in fine-grained sediments accounted for ca. 0.06 wt. %. We thus subtracted 0.06 wt. % from the TN content and used this for calculation of the C : N ratio.

Lignin phenol analysis
Approximately 500 mg of freeze-dried sediments and macrophytes were analyzed for lignin monomers using the alkaline CuO oxidation method (Hedges and Ertel, 1982;Goni and Hedges, 1992) at the Universidade Federal Fluminense laboratory (Brazil). In brief, sediments or macrophytes were transferred to stainless steel reaction vials 5 and digested with 300 mg CuO in 2N NaOH under N 2 in an oxygen-free atmosphere at 150 • C for 150 min. The samples were acidified to pH 1-3 and subsequently 6 mL of ethyl acetate was added. After centrifuging at 2500 rpm for 5 min the supernatant was collected, dried over sodium sulfate (Na 2 SO 4 ), evaporated under a stream of N 2 , reconstituted in pyridine, and converted to trimethylsilyl derivatives using bis-(trimethylsilyl) 10 trifluoroacetamide (BSTFA) at 60 • C for 20 min. Oxidation products were analyzed using an HP Agilent 6890N Series gas chromatography. The recovery factor was calculated using the internal standard ethyl vanillin added prior to analysis (values above 60 % were considered). The response factor was performed using a mixture of commercial standards in four different concentrations, which 15 were periodically injected for calibration. To confirm the identification of each lignin phenol, eight selected samples were analyzed with an Agilent 7890A gas chromatography coupled to an Agilent 5975C VL MSD mass spectrometer using a selective ion monitoring (SIM) at NIOZ (The Netherlands).
Phenol concentrations were reported as the carbon-normalized sum of eight lignin-20 derived reaction products (λ8 mg g −1 oc ), including vanillyl (V -series) phenols (vanillin, acetovanillone, and vanillic acid), syringyl (S-series) phenols (syringealdehyde, acetosyringone, and syringic acid), and cinnamyl (C-series) phenols (p-coumaric and ferulic acid). Ratios S : V and C : V were calculated to identify angiosperm tissue sources. The ratio of acidic to aldehyde vanillyl phenols ((Ad : Al)v) was used as an indicator of the lignin degradation state, since acidic phenols are produced from aldehyde functional groups during the lignin degradation (Hedges and Ertel, 1982

GDGT analysis
All samples for the lipid analysis were processed at NIOZ (The Netherlands). The freeze-dried samples were extracted with a modified Bligh and Dyer technique (Bligh and Dyer, 1959;Pitcher et al., 2009). In brief, the samples were extracted three times with a mixture of methanol (MeOH):dichloromethane (DCM):phosphate buffer (8.7 g of 5 K 2 HPO 4 in 1 L bidistilled water) 10 : 5 : 4 (v : v : v) in an ultrasonic bath (10 min). Extracts and residues were separated each time by centrifugation at 2500 rpm for 2 min. DCM and phosphate buffer were added to the extracts to give a new volume ratio 1 : 1 : 0.9 (v : v : v). This mixture was centrifuged at 2500 rpm for 2 min. to obtain a good phase separation. The DCM phase was then collected in a round bottom flask.

10
The MeOH-phosphate phase was washed twice with DCM and then discarded. The collected DCM fractions were reduced under rotary vacuum. The total lipids extracts were fractioned into core lipids and intact polar lipids (IPLs). The separation was carried out on activated silica with n-hexane:ethylacetate 1 : 1 (v : v) for core lipids and MeOH for IPLs (Pitcher et al., 2009). To each fraction, 0.1 µg C 46 15 GDGT internal standard was added (Huguet et al., 2006). Two third of the IPL fraction was hydrolyzed to cleave off polar head groups. The hydrolysis was carried out by refluxing (3 h) in 2 N HCl:MeOH 1 : 1 (v : v). The solution was adjusted to pH 5 with 2 N KOH-MeOH. This mixture was washed three times with DCM. The DCM fractions were collected, reduced by rotary evaporation, and dried over a Na 2 SO 4 column. Core lipids 20 fractions were separated into polar (DCM:MeOH 1 : 1, v : v) and apolar (DCM) fraction over an activated Al 2 O 3 column.
The core lipids and IPL-derived GDGTs were analyzed using high performance liquid chromatography-atmospheric pressure positive ion chemical ionization-mass spectrometry (HPLC-APCI-MS) in selected ion monitoring (SIM) mode according to Introduction

Long-Chain n-Alkanes carbon isotopes
Two sediment samples collected in the LW season, one from lake Janauaca and another from lake Curuai, were used to compare the differences in the δ 13 C values of plant-wax derived long-chain n-alkanes in the upstream and in the downstream lakes. The extraction of n-alkanes was performed with an Accelerated Solvent Extraction 5 method (ASE). The extracts were fractionated in apolar and polar fractions using an activated aluminum oxide (Al 2 O 3 ) column with hexane and MeOH:DCM (1 : 1, v : v), respectively, as the eluents. The n-alkanes in the apolar fraction were identified by a Thermo Finnigan Trace DSQ gas chromatography (GC-MS) and quantified with an HP 6890 GC system. To quantify the concentration of the n-alkanes, an internal standard 10 was added to the apolar extracts. To further clean up the apolar fraction, the extracts were passed over a silver nitrate (AgNO 3 ) column using hexane as the eluent. The δ 13 C values of higher n-alkanes were determined using an isotope-ratio-monitoring mass spectrometer (IRM-GC-MS) Thermo Delta V Advantage and the results were obtained using the software Isodat 3.0. Four injections were performed for each sample 15 to calculate the analytical error.

Statistical analysis
To evaluate the differences in mean values between different groups, the nonparametric Mann-Whitney U-test was used, which does not need the normality assumption of the one-way analysis variance (ANOVA). Groups that showed significant 20 differences (p < 0.05) were assigned with different letters. The statistical test was performed with the software package SIGMAPLOT 11.0.

Bulk parameters
The TOC content showed lower mean value (Table 2) in the downstream lake Curuai (2.0 ± 0.6 wt. %) and the highest mean value was found in Cabaliana (3.3 ± 0.8 wt. %; Fig. 3a). No significant seasonal variation was observed (Fig. 4a). The C : N ratio did 5 not reveal significant spatial and seasonal variations (Figs. 3b and 4b). The lowest mean value was found in Curuai (10 ± 1) and the highest one in Mirituba (11 ± 2). The δ 13 C org values were significantly less negative in the downstream lakes (Fig. 3c). In Curuai the mean value was −27 ± 1 ‰ and in Cabaliana −33 ± 2 ‰. No significant seasonal variation was observed for the δ 13 C org values (Fig. 4c).

Lignin phenols
No significant changes were observed along the upstream-downstream transect for the mean values of λ8 (i.e. a proxy for the amount of lignin); the mean value of λ8 for the SOC was 44 ± 29 mg g −1 oc . However, λ8 values revealed significant seasonal changes. 20 The higher values were observed in the RW (56 ± 30 mg g −1 oc ) and in the FW seasons (62 ± 34 mg g −1 oc ) compared to the HW (23 ± 9 mg g −1 oc ) and LW (29 ± 12 mg g −1 oc ) seasons (Fig. 3g). The C : V ratio showed no significant seasonal and spatial variation, and the mean value for all sediment samples was 0.7 ± 0.4 (Figs. 3d and 4d). The values of the S : V ratio did not show significant spatial differences either but higher mean 25 values in the RW season (1.1 ± 0.1) and in the FW season (1.2 ± 0.2) were observed in 8758 Introduction comparison to that in the LW season (0.9 ± 0.1; Fig. 4e). The mean value of (Ad : Al)v ratio for the different lakes did not show spatial variation (Fig. 3f), however, it was higher in the LW (1.5 ± 0.4) and HW (1.7 ± 0.5) seasons for most lakes (Fig. 4f). For the C 3 macrophytes, λ8 values varied between 50-60 mg g −1 oc and between 70-160 mg g −1 oc for the C 4 macrophyte samples. The S : V ratio varied between 0.4 and 0.6 5 for C 3 macrophytes and between 0.4 and 0.8 for the C 4 macrophyte. The range of C : V ratio was 1.0 to 3.1 for the C 3 macrophytes and 1.4 to 2.7 for the C 4 macrophytes. The (Ad : Al)v ratio varied between 0.2 and 0.8 for all macrophyte samples (Table 3). For the riverbank and wetland soil samples, the λ8 values varied between 8 and 88 mg g −1 oc . The S : V ratio varied between 0.5 and 1, the C : V ratio varied between 0.2 and 0.5, 10 and the (Ad : Al)v ratio varied between 0.6 and 1.5.

BrGDGTs and crenarchaeol
Along the upstream-downstream transect, no significant changes were observed for the mean values of brGDGTs concentrations (Fig. 3h). The lowest value was found in Curuai (31 ± 14 µg g −1 oc ) and the highest one in Canaçari (44 ± 22 µg g −1 oc ). The mean 15 concentrations of crenarchaeol were higher in Canaçari (115 ± 57 µg g −1 oc ) when compared to Janauaca (34 ± 33 µg g −1 oc ). However, no significant difference was observed between the upstream (Cabaliana and Janauaca) lakes and the downstream lake (Curuai; Fig. 3h and i). On the other hand, brGDGTs concentrations showed significant seasonal changes. The highest mean value for brGDGTs concentrations was found 20 in the FW season (45 ± 23 µg g −1 oc ), and the lowest mean concentration was found in the HW season (24 ± 16 µg g −1 oc ). The RW and LW seasons showed intermediate mean concentrations (35 ± 12 and 38 ± 16 µg g −1 oc , respectively) and no significant difference was observed if compared to the FW and HW seasons (Fig. 4h). The concentrations of crenarchaeol did not reveal significant changes over the hydrological seasons (Fig. 4i). 25 The mean values varied between 5 ± 4 and 10 ± 6 µg g −1 oc in the HW and LW seasons, respectively. The percentage of IPL brGDGTs and IPL crenarchaeol was significantly higher in the LW season (19 ± 7 and 23 ± 9 %, respectively). In the other three seasons, it showed values around 10 ± 2 % of IPL brGDGTs and IPL crenarchaeol with no significant variability (Table 3).

Long-chain n-alkanes
The results of n-alkane analyses are summarized in Table 6. The carbon preference 5 indices (CPI), calculated according to Bray and Evans (1961), were high, confirming a plant wax origin of the higher n-alkanes. A somewhat lower CPI was found in the downstream lake (3.5) compared to those in the upstream lake (5.5). A significant increase in the δ 13 C values of the long-chain n-alkanes (C 27 , C 29 , C 31 ) was observed downstream. In the upstream lake, the mean δ 13 C for the long-chain n-alkanes was 10 −34.1 ± 0.5 ‰ and in the downstream lake the mean value was −31.6 ± 0.6 ‰ (Table 6). This represents a difference of 2.5 ‰ from upstream to downstream.

Sources of sedimentary organic matter in the floodplain lakes
To determine the origin of the SOM in the floodplain lakes, we considered five po-15 tential sources: (1) the terrestrial Andean clay-bounded and refractory SPOM, which may be transferred to the floodplain lakes via the Solimões-Amazonas and Madeira rivers (Hess et al., 2003), (2) "terra firme" soil and litter of the Amazonian lowland forest , which will be transferred to the floodplain lakes via streams, (3) Mortillaro et al., 2011). The biomarkers analyzed, lignin phenols and GDGTs, enabled us to identify most of these sources of OM, except for planktonic sources. However, in this case, some information can be obtained using bulk parameters, i.e. the δ 13 C org and C : N ratio. Our results were compared with data reported previously Martinelli et al., 1994;Meyers, 1994;Martinelli et al., 5 2003; Aufdenkampe et al., 2007;Zell et al., 2013b) and with specific OM sources sampled and analyzed in this work, such as macrophytes, wetland soil and "terra firme" soil (Table 3), in order to identify the main sources of SOM in the floodplain lakes.
The average values of the various parameters of the river SPOM (Ertel et al., 1986;, wetland soils, "terra firme" soils and the potential biological OM 10 sources (phytoplankton, macrophytes, grass, leaves and wood) are compared with those of the SOM of the floodplain lakes in Fig. 5 and Table 4. Data for the riverine SPOM is subdivided into fine particulate organic matter (FPOM) and coarse particulate organic matter (CPOM). For the interpretation of these data, it is important to note that the amount of CPOM in the Amazon river has been reported to be approximately 15 eight times lower than that of the FPOM (Richey et al., 1990). The averages of important lignin parameters (λ8, S : V ratio) but also the C : N ratio of the wood samples are significantly different from those for the sediments, which clearly indicates only a minor contribution of woody material to the SOM. Furthermore, the λ8 of riverine FPOM is substantially lower than that of the SOM of the floodplain lakes, indicating that riverine 20 SPOM is not an important source of lignin for the SOM of the floodplain lakes either. In terms of lignin parameters, the SOM is distinguished by two clear characteristics. Firstly, the (Ad : Al)v ratio is high with an average value of 1.25 (Fig. 5). Such a high value is only noted in the wetland and "terra firme" soils. However, this ratio is affected by the oxidation state of the lignin and thus, cannot be used as a source characteris-25 tic of the lignin. Secondly, the SOM is characterized by a substantially elevated C : V ratio ( Fig. 6; cf. Hedges et al., 1982). Since all of the potential lignin sources, except macrophytes, have a much lower value, this indicates that macrophyte lignin and, thus accordingly, macrophyte OM (since average λ8 values of macrophyte OM and the SOM do not substantially differ) largely contribute to the SOM. Since the S : V ratio of macrophyte OM is relatively lower than that of the lignin component of the SOM (Fig. 5), some contributions of lignin derived from other fresh plant OM (i.e. grasses/leaves) or wetland soils might explain the elevated S : V ratio of the SOM. Further information with respect to sources of SOM can be obtained from the GDGT  We have argued that the C : V ratio and the crenarchaeol concentration are the only two parameters that clearly point out one specific source of SOM (i.e., macrophytes and aquatic production in the rivers or floodplain lakes, respectively). Consequently, these parameters can be applied to a two end-member model to estimate the fractions of each of these two sources in the SOM. According to this approach (Martinelli el 5 al., 2003), the C : V values of macrophytes and the average values of soil and riverine SPOM samplescan be used to estimate the contribution of macrophyte OM to the SOM. Similarly, the concentration of crenarchaeol in the riverine SPOM and its concentration in soil samples can be used to estimate the contribution of aquatically produced OM to the SOM (Eq. 1-3, Table 5).
In Eqs. (1) and (2) (Table 5). These calculations indicate that 20-30 % of the SOM is derived from macrophytes and 20-30 % from the aquatic production either in the river or in the floodplain lake itself. Consequently, the 20 remaining 40-60 % of the SOM might be derived from other sources of OM such as the flooded forests (Eq. 3). The periodical floods link the floodplain lakes and the wetland vegetation and soil. Thus, the seasonal and spatial contrasts in the SOM should be investigated in order to better understand the connectivity between these compartments.

Spatial differences in the composition of sedimentary organic matter
Along the longitudinal transect, from upstream to downstream, most bulk geochemical parameters (i.e. TOC content and δ 13 C org ) show significant differences between the upstream and downstream lakes (Fig. 3a, c), while most of the measured biomarker parameters (λ8, S : V , (Ad : Al)v and brGDGTs) do not show such a pattern (Fig. 4e,  of organic molecules (Harvey, 2006). It is important to note that the results must be interpreted taking in consideration the high sedimentation rates in the floodplain lakes, typically 1-2 cm yr −1 (Moreira- Turcq et al., 2004), and the fact that re-suspension is induced by storms during the LW and RW seasons or by currents during the receding waters (FW). These events may have a substantial effect on the material comprising the first 2 cm of sediments of floodplain lakes, which are mixed with newly arrived SOM from the water column, and are re-oxygenated favoring the degradation. The percentage of TOC in the sediment samples shows a decrease from 3.3 (wt. %) upstream (Cabaliana) to 2.1 (wt. %) downstream (Curuai; Fig. 3a). Furthermore, over the transect of lakes the average δ 13 C org values increase by ca. 5 ‰ (Fig. 3c). How-20 ever, the average C : N ratio does not show any significant changes over the transect (Fig. 3b). These results are in good agreement with previous studies in the central Amazon Basin (Victoria et al., 1992;Martinelli et al., 2003). The increasing trend in δ 13 C org from upstream to downstream lakes may be caused by an increased contribution of C 4 macrophytes to the SOM, whose abundance increases in open water lakes 25 and floodplains. Alternatively, since the δ 13 C org values in the downstream lakes come closer to the δ 13 C org of the Solimões-Amazon SPOM (∼ −26 to −30 ‰; Moreira-Turcq et al., 2013;Mayorga et al., 2005), an increased input of riverine organic matter may also explain this. To disentangle whether this trend in the δ 13 C org values is caused by the contribution of C 4 plants or of riverine SPOM, the isotopic composition (δ 13 C) of long-chain n-alkanes was analysed. Sediments from the upstream lake Janauaca and the downstream lake Curuai, both collected during the LW season, were compared. The results (Table 6) show that the long-chain n-alkanes δ 13 C sig-5 nature is more like those of C 3 higher plants (Castañeda et al., 2009) for both lakes although for Curuai the values are slightly less negative. If one considers the values of δ 13 C in the n-alkane C 29 in the leaf waxes of C 3 and C 4 plants, one can calculate the contribution of C 4 plants sedimentary n-alkanes according to the following equation: Contribution of C 4 plants = δ 13 C org C 29 (C 3 plants) − δ 13 C org C 29 (sediment) δ 13 C org C 29 (C 3 plants) − δ 13 C org C 29 , (C 4 plants) · 100 (4) 10 where the end member value for δ 13 C org C 29 (C 3 plants) is −34.7 ‰ and for δ 13 C org C 29 (C 4 plants) is −21.7 ‰ (Castañeda et al., 2009). The measured values for δ 13 C org of the C 29 n-alkane in the sediments of Janauaca and Curuai are listed in Table 6. Accordingly, the percentage of C 4 plants in the upstream lake is only 3 %, but for the downstream lake 22 %. The difference in δ 13 C org for C 4 and C 3 higher plants is ca. 15 20 ‰. A switch from almost 100 % C 3 macrophytes to a 78 % contribution would result in a change in the isotopic composition of the macrophyte "pool" of the SOM of 4-4.5 ‰.
Since this pool is estimated to represent 20-30 % of the SOM, this cannot fully explain the observed 5 ‰ shift (Fig. 3c). However it should be considered that the increasing fraction of C 4 higher plants for the SOM in the downstream lake may not solely be the 20 consequence of changes in the contributing aquatic macrophytes. Land vegetation, mainly shrubs and grass in downstream lakes, may also affect the observed shift in δ 13 C org of SOM

Seasonal changes in the composition of sedimentary organic matter
The λ8 values and the S : V ratio show significantly higher values in the RW and FW seasons (Figs. 4e, g, and 6a) in all lakes. The mean concentrations of brGDGTs also show higher values in the FW season (Figs. 4h and 6b). The co-occurrence of these two types of molecules indicate that litter, traced by lignin phenols, and superficial soil, 5 traced by brGDGTs, are preferentially deposited during rising and receding waters, which increases the wetland soil runoff. Besides, the seasonal mean values of (Ad : Al)v show remarkably lower values in the RW and FW seasons (Fig. 4f), an inverse pattern if compared to the S : V , λ8 and brGDGTs. This means that less degraded lignin is present in the surface sediments in the RW and FW seasons. Thus, the increase in the 10 concentrations of the organic compounds is not a consequence of the re-suspension of the sediments, but to the arrival of fresher OM. In the HW and LW seasons, more degraded lignin phenols (higher values of (Ad : Al)v) are present in the sediments concomitant with lower amounts of λ8 and S : V ratio. Since the concentration of crenarchaeol (a marker for aquatic production) and the C : V ratio (manly affected by aquatic 15 macrophytes; see above) do not reveal significant seasonal changes, we conclude that such increase in the concentration of the lignin phenols in the RW and FW seasons and the brGDGTs in the FW season is not derived from the water column, riverine SPOM or in situ production but from the soil and leaf runoff. Previous works postulated that Andean and low land soil material is mainly trans-20 ferred to the lakes via river main stream, in particular, during the RW and HW seasons and that would be the main source of SOM of the floodplain lakes (e.g., Victoria et al., 1992;Moreira-Turcq et al., 2004;Mortillaro et al., 2011). However, according to our results, the lignin phenols increase their concentration in the RW and the FW seasons. Thus, based on the hydrodynamics of floodplain lakes and the concentration of the 25 biomarkers applied in this study, in the RW and FW seasons, these organic molecules are mainly derived from the drainage of local wetlands soils. This is more evident for the upstream lakes, which are surrounded by flooded forests and by larger flooded area,

Tables Figures
Back Close

Full Screen / Esc
Printer-friendly Version

Interactive Discussion
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | than for the downstream lakes, which are surrounded mainly by grass vegetation and shrubs. However, even in lake Curuai, where the primary production and the riverine SPOM is admittedly an important source of SOM (Moreira-Turcq et al., 2004;Zocatelli et al., 2013), the interface between the floodplain lake and the flooded soil drives the sedimentation of the organic compounds. Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | de Pessoal de Nível Superior (CAPES). The research leading to these results has also received funding from the European Research Council (ERC) under the European Union's Seventh Framework Program (FP7/2007 Table 4. Average values of biomarkers and bulk parameters in the possible sources of SOM and in sediment samples. The data was obtained in the present work and in the literature Hedges and Mann, 1979;Aufdenkampe et al., 2007).  Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | C a b a li a n a J a n a u a c a M ir it u b a C a n a ç a r i C u r u a i TOC (wt.%) 0 1 2 3 4 5 6 C a b a li a n a J a n a u a c a M ir it u b a C a n a ç a r i C u r u a i 16 C a b a li a n a J a n a u a c a M ir it u b a C a n a ç a r i C u r u a i C a b a li a n a J a n a u a c a M ir it u b a C a n a ç a r i C u r u a i 2.0 C a b a li a n a J a n a u a c a M ir it u b a C a n a ç a r i C u r u a i  Aufdenkampe et al., 2007;Zell et al., 2013) and the present work (Table 3). Letters over the boxes indicate significant differences (p < 0.05) between the means. Introduction