Lowland Amazon River system
Spatial distribution and isotopic composition of OCterr
The TOC contents of our riverbed sediments in individual tributaries are
larger than values of bedload sediments reported by Bouchez et al. (2014)
(mean = 0.23 ± 0.42 % in the Solimões and Madeira rivers
and the Amazon mainstream), but lower than the values of suspended materials
(e.g. mean = 1.14 ± 0.33 % in the Solimões and Madeira
rivers and the Amazon mainstream) in the Amazon lowland basin (Hedges et al.,
1986; Moreira-Turcq et al., 2003; Bouchez et al., 2014). The distribution of
TOC contents basically reflects the characteristics of the tributaries, which
are mainly influenced by the content of dissolved organic matter and
suspended sediment (Fig. 2a). The relatively high TOC contents in the Negro
River are due to the low suspended sediment content and high content of humic
substances (Ertel et al., 1986). The Xingu river is characterized by low
suspended sediment content and high phytoplankton production, which lead to
the high TOC contents in riverbed sediments (Moreira-Turcq et al., 2003). In
contrast, the Solimões River and the Madeira River, being the primary
contributors of the suspended sediment to the Amazon River mainstream, have
large suspended sediment load (Moreira-Turcq et al., 2003). Consequently, the
low TOC contents in the riverbed sediments in these tributaries are due to
dilution by lithogenic material. The lower-intermediate TOC contents for the
Amazon River mainstream results from the mixing of different signals from
these tributaries with a greater influence from the Solimões and the
Madeira rivers.
The δ13CTOC values of our riverbed sediments (i.e.
from -26.1 to -29.9 ‰) (Fig. 2b) are similar to the values
reported for riverbed sediments (e.g. from -27.6
to -28.8 ‰) and suspended particulate matter
(e.g. -28.3 ± 1.1 ‰) in the Amazon River system in previous
studies (Hedges et al., 1986; Cai et al., 1988; Kim et al., 2012; Bouchez et
al., 2014). Hedges et al. (1986) studied δ13CTOC values
of different organic carbon sources in samples from the Amazon River and
found the respective average δ13CTOC values of C3 tree
leaves, woods, macrophyte tissues, and C4 grasses to be -30.1 ± 0.9,
-27.6 ± 1.0, -21.4 ± 8.4 and -12.2 ‰. The total
average δ13CTOC value of riverbed sediments in this
study (-28.5 ± 0.9 ‰) confirms the dominant contribution
from terrestrial C3 plants. There is no significant difference in the
distribution of δ13CTOC values among the sampled
tributaries.
Characteristics of lignin phenols
With the exception of the samples from the Xingu River (r2=0.02,
p=0.71), all riverbed sediments exhibit a good relation between Λ8
values and Σ8 values (average r2=0.76, p<0.05,
n=39). In the Xingu River, the high level of in situ primary production and
low-turbidity conditions favour the settling of phytoplankton-derived organic
matter from the water column (Moreira-Turcq et al., 2003). The deposited
phytoplankton-derived organic matter dilutes the abundance of lignin in TOC
(Λ8) but has only a small influence on sediment mass and, as a
result Σ8, resulting in divergence of Λ8 and Σ8
(Rezende et al., 2010). Based on previous studies, the Λ8 values of
different organic matter fractions range from 0.45 to 2.40 mg/100 mg OC for
fine particulate organic matter (FPOM; silt and clay fraction,
< 63 µm), and from 1.21 to 10.46 mg/100 mg OC for coarse
particulate organic matter (CPOM; sand-size fraction,
> 63 µm) (Aufdenkampe et al., 2007; Hedges et al.,
1986, 2000). In this study, riverbed sediments (Fig. 2c),
except for three samples with lignin contents lower than 2.0 mg/100 mg OC,
had Λ8 values (2.42–9.27 mg/100 mg OC) similar to those of CPOM.
Most of the Λ8 values of our samples are smaller than the average
Λ8 values of tree wood tissues (19.3 mg/100 mg OC) and C4 grasses
(9.1 mg/100 mg OC), and closer to the range of tree leaves and macrophytes
(3.7 mg/100 mg OC and 6.4 mg/100 mg OC, respectively) (Hedges et al.,
1986). This finding is also supported by the distribution of C / V and
S / V ratios. The plot of S / V vs. C / V (Fig. 3) indicates that
angiosperm leaves are the major origin of lignin in the lower Amazon basin.
It is noteworthy that the range of typical C / V values for angiosperm
leaf material in the Amazon basin is larger (i.e. including C / V values
as low as 0.07) than in other regions (with lowest C / V values around
0.20) (Bianchi et al., 2011; Cathalot et al., 2013; Tesi et al., 2014). The
resulting small difference between C / V ratios of non-woody and woody
tissues of angiosperms in the Amazon region results in a larger uncertainty
in inferring the plant sources of lignin. C / V values around 0.1 could
be interpreted either as signals exclusively from leaves or as signals from a
mixture of woody tissues and leaves. To circumvent this uncertainty, the
P / V values are also used to identify the lignin sources. P phenols in
our samples are derived from lignin, as supported from the significant
correlation of the content of P phenols and lignin content (r2=0.50,
p<0.05, n=47). All P / V values of our samples (0.17–0.64)
are higher than the average P / V ratio of woods (0.05) and similar to
the range observed for leaves (0.16–6.9) (Hedges et al., 1986). Considering
all parameters, non-woody angiosperms are the most likely major source of
lignin in the lowland Amazon basin. The slightly higher C / V ratios in
the Solimões River (0.20) and the Madeira River (0.24) suggest a small
contribution of grass-derived material (C / V > 1) probably
from the Andean highlands (Aufdenkampe et al., 2007; Hedges et al., 2000).
The degradation extent of OCterr can be assessed by
(Ad / Al)V and (Ad / Al)S ratios as more
degraded lignin yields elevated (Ad / Al)V and
(Ad / Al)S values (Hedges et al., 1988; Opsahl and Benner,
1995). In the case of the Amazon basin, the (Ad / Al)V and
(Ad / Al)S ratios of typical fresh woods and tree leaves both
range from 0.11 to 0.24 (Hedges et al., 1986). All of our samples exhibiting
values between 0.26 and 0.71 for (Ad / Al)V (Fig. 2d) and
between 0.15 and 0.57 for (Ad / Al)S are outside of the range of
fresh plant materials, suggesting degraded OCterr in all samples.
Instead, the (Ad / Al)V,S ratios observed in our samples are
within the ranges of suspended particulate solids obtained in the lower
Amazon basin and Bolivian headwaters ((Ad / Al)V,S of
0.21–0.39 and 0.13–0.22 for CPOM and (Ad / Al)V,S of
0.38–0.79 and 0.22–0.41 for FPOM; Hedges et al., 1986,
2000). The Negro River displayed the highest average
(Ad / Al)V ratio (0.55), reflecting a greater degree of
degradation. This might be indicative of more efficient degradation in the
podzols of the lateritic landscapes in the Negro River watershed (Bardy et
al., 2011). The (Ad / Al)V ratios in the Solimões and the
Madeira rivers increase with increasing C / V values (r2=0.50,
p<0.05, n=13), which implies that the plant tissues with higher
C / V values (higher grass contributions) are more degraded. This further
supports the inference that the Solimões and the Madeira rivers receive
POC from grass sources from high-altitude watersheds, where deeper soil
erosion of more degraded OCterr could occur. The degradation
status of lignin in riverbed sediments does not display a downstream
increasing trend and is similar to previous studies on suspended POC of
different size fractions. This leads to the conclusion that
OCterr processing during transport through the Amazon River
system probably has limited influence on the composition of lignin recorded
in the riverbed sediments and the degradation information likely reflects
source characteristics of OCterr prior to fluvial transport
(Hedges et al., 1986, 1994). This finding contradicts the conclusion of Ward
et al. (2013) that lignin is rapidly degraded within the Amazon River. Ward
et al. (2013) studied particulate and dissolved lignin in the water column,
which is exposed to degradative environments. In contrast, our study focuses
on the lignin associated with mineral particles, which are deposited and
protected from degradation. This discrepancy highlights the important role of
matrix association effects in the preservation of organic matter.
Sedimentological control on OCterr characteristics
As grain-size data could not be obtained directly on the riverbed sediment
samples, we inferred grain-size information based on the relationship between
the Al / Si ratio and grain size of riverbed sediments observed by
Bouchez et al. (2011) in samples from the Amazon basin. High Al / Si
indicates aluminium-rich fine-grained sediment, whereas low Al / Si
suggests silicon-rich particles of larger grain size (Bouchez et al., 2011;
Galy et al., 2008).
As expected, the TOC contents increase with Al / Si (Fig. 4a), indicating
that fine particles, associated with larger specific surface areas and likely
rich in clay, carry more TOC than coarser particles. The Negro and the Xingu
rivers have larger Al / Si and TOC variations, and for a given
Al / Si ratio, the Negro and the Xingu rivers show higher TOC contents
compared to the other tributaries. As these rivers have distinct chemical
characteristics and clay mineral composition (e.g. lower pH in the waters of
the Negro River and higher kaolinite content in the sediments of the Negro
and the Xingu rivers than in other tributaries; Guyot et al., 2007), the
adsorption affinities of OCterr on different clay minerals or
under different chemical conditions may be distinct.
Different grain-size classes may have not only different TOC content but also
the composition of their OCterr might vary. For example, previous
studies on POM in the Amazon basin found that CPOM has a higher content of
lignin phenols than FPOM and that CPOM is composed of fresher lignin with
lower C / V ratios (Hedges et al., 1986). Nevertheless, contradictory
results were observed in our riverbed sediments. The Λ8 values in
the Madeira River, the Solimões River, and the mainstream Amazon River
show a remarkable increase with decreasing grain size (indicated by
increasing Al / Si ratios) (Fig. 4b). This rise in lignin content in
organic matter associated with finer minerals implies preferential
preservation of lignin on finer particles compared with other components. A
similar trend has been observed in other environments, such as in surficial
sediments of the East China Sea (Wu et al., 2013) and the Iberian margin
(Schmidt et al., 2010). However, opposite distribution patterns with lignin
enriched in coarser size classes and low density fractions have been found in
sediments from the Washington margin (Keil et al., 1998) and the Laptev Sea
(Tesi et al., 2016). This is likely because co-deposited plant debris
preferentially accumulates within coarser and low density fractions. Thus, the
apparent discrepancy between these studies may derive from the different
methods employed, i.e. partitioning sediments into discrete size and density
fractions (Keil et al., 1998; Tesi et al., 2016) vs. characterizing the
mean grain size of bulk sediments (Wu et al., 2013; Schmidt et
al., 2010, and our study). On the other hand, the distinct environmental
settings in which these studies were conducted, may influence the grain-size
distribution of lignin. The Amazon River drains tropical lowlands where
degradation on land is efficient, and the turbid waters of the Amazon River
might limit the settling of plant fragments. Consequently, in our riverbed
samples sedimentary organic matter likely contains only a small contribution
of plant debris, whereas less efficient degradation on land in the colder
climates of the higher latitudes will result in deposition of more coarse
plant debris. In the Negro River, there is only a slight increase in Λ8 values as mineral particles become finer, probably as a result of the
large amount of sediment-associated chemically stable humic substances
(Hedges et al., 1986), in which the lignin content is relatively constant.
However, Xingu River sediments exhibit decreasing Λ8 values with
decreasing grain size probably because the lignin content in finer particles
from the Xingu River is diluted by other non-lignin organic components. With
respect to the indicator of plant sources (Fig. 4c), the C / V ratios for
samples from the Madeira River, the Solimões River, and the Negro River
decrease with decreasing grain size, which implies that lignin with higher
C / V ratios is typically enriched in coarser particles. This suggests
that the non-woody tissues with higher proportions of cinnamyl phenols are
enriched in coarse-grained sediments. Xingu River and Amazon River mainstream
sediments present no pronounced trend between C / V ratios and
Al / Si values.
(Ad / Al)V values for all riverbed sediments do not show any
obvious relationship with Al / Si (Fig. 4d). Only Madeira River and
Solimões River sediments exhibit decreasing (Ad / Al)V
values with increasing Al / Si, which suggests that lignin associated
with
larger mineral particles is more degraded. This observation implies
preferential preservation of lignin in finer-grained sediments due to better
protection against oxidative degradation (Killops and Killops, 2005). Like
the correlation between Λ8 values and Al / Si, our observation
on (Ad / Al)V values is different to the trends found by Keil
et al. (1998) and Tesi et al. (2016), where lower (Ad / Al)V
values have been found in coarser fractions. Again, this is likely due to the
fact that these studies investigated individual grain size and density
fractions, and that they were conducted in higher latitudes with less
efficient processing of plant remains prior to deposition. As a result, fresh
plant tissue would be found in the coarse fractions leading to low
(Ad / Al)V values (see discussion above). Because our
sediments likely contain limited amounts of plant debris, the
(Ad / Al)V are lower in finer sediments implying that lignin
is better preserved in finer-grained particles. In previous studies on
suspended sediments, lignin in the coarse fractions is more abundant and less
degraded compared to the counterpart in fine fractions (e.g. Hedges et al.,
1986). In contrast, our results for riverbed sediments suggest that lignin is
preferentially preserved and better protected against degradation on
fine-grained material. The different grain-size effects on OCterr
composition between suspended and riverbed sediments suggest that there are
other processes working on OCterr in suspended sediments and
riverbed sediments, which cause post-depositional changes in the
OCterr characteristics.
In summary, our data indicate that lignin derives mainly from non-woody
tissues of angiosperms in the lowland Amazon basin, and there is little
evidence for contribution from C4 plants to riverbed sediments. Grain size
plays an important role in OCterr preservation and lignin composition
in the Amazon River. Fine inorganic particles have high adsorption affinity
for OCterr, especially for lignin compared to other OCterr
components and efficiently protect lignin from degradation.
Amazon shelf and fan
Spatial distribution and characteristics of OCterr and
lignin phenols
Because of the depleted average δ13CTOC values of the
riverbed sediments (-28.5 ± 0.9 ‰), contribution of C4
plants is not expected in the offshore sediments affected by the Amazon
outflow. Therefore, enriched 13CTOC values in the SE sector
(-18.6
to -21.6 ‰) likely indicate organic matter
predominantly of marine origin. δ13CTOC values in the
Amazon fan sector ranging from -21.4 to -24.5 ‰ (Fig. 5b) also
reflect dominantly marine organic matter. These values are within the range
of published values for high sea level periods (Schlünz et al., 1999),
when most terrestrial POM discharged from the Amazon River is transported to
the north-western shelf. In the NW sector, increasing δ13CTOC values with distance from the Amazon River mouth
indicate that the terrestrial POM input from the Amazon River transported to
the NW by the North Brazil Current is increasingly diluted by marine organic
matter. OCterr is dominant on the continental shelf,
corroborating previous results (e.g. Schlünz et al., 1999).
Sediments in the SE sector exhibit much lower Λ8 values than
observed in the Fan and NW sectors. Here, slightly increasing Λ8
values with distance from the Amazon River mouth suggest that this
terrestrial material is predominantly supplied by rivers to the south-east of
the Amazon mouth and not by the Amazon River itself. The Λ8 values
in sediments near the Amazon River mouth are highly variable and decrease
with distance from the river mouth to the Fan and NW sectors, reaching very
low Λ8 values in the slope of the NW sector. Lower lignin contents
(0.05–0.32 mg mg-1 OC) have been observed in the deep sea fan
sediments measured in the study of Feng et al. (2016), which means there is
increasing loss of lignin during the transport seawards to the Fan and NW
sectors. Λ8 and δ13CTOC values show similar
spatial distribution and are positively correlated (r2=0.53, p<0.05, n=27) (Fig. 6). The agreement in the spatial patterns of lignin
content and isotope composition of organic matter suggest that lignin is a
reliable tracer of OCterr in the Amazon shelf and fan, and that
the SE sector receives little OCterr contribution from the Amazon
River. The intercept of the correlation between Λ8 and δ13CTOC of NW and Fan sediments (r2=0.58, p<0.05, n=21) is at -20.8 ‰, which corresponds to conditions with
minimal OCterr input to the marine sediments. It should be noted
that the samples from the Amazon fan have the same low lignin contents as in
the SE sector, which indicates low contribution of OCterr from
terrestrial vascular plants under modern conditions. However, the Fan
sediments show more depleted δ13CTOC than sediments
from the SE sector, which implies that a small terrestrial fraction is
contained in the organic matter of the modern Fan sediments. Potentially, the
OCterr from vascular plants deposited in the Amazon fan is
readily degraded as indicated, e.g. by the high (Ad / Al)V,S
ratios, while the relict OCterr is predominantly rock derived
(with estimated δ13CTOC between -24.3 and
-25.7 ‰) (Bouchez et al., 2014) and responsible for the depleted
δ13CTOC. Petrogenic organic matter is thus likely a
significant component in the offshore sediments because of its refractory
nature and resulting high preservation potential.
Stable carbon isotopic composition of total organic carbon
(δ13CTOC) vs. carbon-normalized lignin content (Λ8) for marine surface sediment samples from the Amazon continental margin.
Panels (a), (b), (c), and (d) indicate total organic carbon (TOC) content,
carbon-normalized lignin content (Λ8), cinnamyl / vanillyl (C / V)
ratio and lignin degradation index ((Ad / Al)V) for marine surface
sediment samples from the NW sector vs. mean grain size, respectively. Empty
triangles represent the three deepest sites (> 2000 m) far from
the coast, and filled triangles represent the other sites in the NW sector
with water depth shallower than 100 m. See Table 2 and Fig. 5 for the
location of the NW sector.
C / V and S / V ratios (0.08–0.47 and 0.70–1.57, respectively;
Fig. 3) in the entire Amazon shelf and fan are comparable to those in the
riverbed sediments of the lowland Amazon basin, which indicates the same
predominant source of non-woody angiosperm tissues. This also implies no
further alteration of Amazon-derived lignin after it is discharged into the
ocean and deposited in marine sediments. Lignin in offshore marine sediments
thus can provide reliable evidence for the reconstruction of the vegetation
cover in the Amazon basin.
The distribution of the degradation state of lignin based on
(Ad / Al)V is shown in Fig. 5d. The strikingly elevated
(Ad / Al)V values in the Amazon fan are probably caused by
longer exposure to oxygen (Blair and Aller, 2012) at the sediment–water
interface under low sedimentation rates, corroborating our previous
interpretation of the low Λ8 values but intermediate
δ13CTOC values in the Fan sediments (Fig. 6). In the NW
sector, there is no obvious decreasing trend of the
(Ad / Al)V values with the distance from the river mouth. Low
(Ad / Al)V values found at shallow nearshore sites far from
the Amazon River mouth (e.g. GeoB16218-3 and GeoB16225-2) may be due to
rapid transport and deposition of the material discharged from the Amazon
River (Nittrouer et al., 1995) or contributions from local small rivers, which
may carry an indistinguishable lignin signature from the Amazon River. In the
study of Feng et al. (2016), the sampling locations in the Amazon deep sea
fan are in deeper water depths than our samples (> 4000 m).
Nevertheless, very low (Ad / Al)V,S (0.04–1.3 and 0.03–0.8 for V and S
phenols, respectively) were observed. Neither S / V nor C / V ratios
decrease with (Ad / Al)V,S in the marine sediments, which would be
expected because cinammyl and syringyl pheonls are preferentially degraded
compared to vanillyl phenols (Benner et al., 1990; Opsahl and Benner, 1995).
Despite the fact that the degradation of lignin preserved in marine sediments
is slightly higher than that preserved in riverbed sediments, degradation has
no major impact on the lignin composition.
Influence of grain size on OCterr deposition in marine
sediments
The grain size and Al / Si in the Amazon fan and SE sectors vary within a
rather small range. The grain size and Al / Si relationship in the NW
sector is in accordance with the results obtained by Bouchez et al. (2011).
The sediments in the NW sector have similar Al / Si ratios as our
riverbed sediments (Table 2), which are correlated with grain size. We use
grain-size data for the following discussion of the sedimentological control
on the distribution pattern of OCterr in the NW sector and refer
to the relationship between TOC and Al / Si as observed in the riverbed
sediments.
Fine sand sediments were observed at the position closest to the Amazon River
mouth (GeoB16209-2) and at site GeoB16225-2, which is far from the Amazon
River mouth (about 700 km) but near the coast in a water depth of 34 m. The
sample GeoB16225-2 probably receives additional input from local smaller
rivers, from which coarser sediments are discharged. Considering that the
characteristics of organic matter in the lowland Amazon basin are almost
uniform, the local contributions deposited in GeoB16225-2 cannot reliably be
distinguished from the material from the Amazon River system. The trend of
increasing TOC contents with decreasing grain size (Fig. 7a) parallels the
one demonstrated for the riverbed sediments (Fig. 4a) and in other marine
sediments (Keil et al., 1997; Mayer 1994). The lignin content in organic
matter (Λ8) and grain size are not significantly related likely
because the OCterr is mixed with marine-derived organic matter in
marine environments (Fig. 7b). For example, according to the enriched
δ13CTOC values
(mean = -20.4 ± 0.1 ‰), sites GeoB16216-2, GeoB16217-1,
and GeoB16223-1, which are the sites located farthest offshore, contain the
largest fractions of marine organic matter, which reduces their Λ8
values to about 0.18 despite their small mean grain size
(< 14 µm). Except for these three locations, the Λ8 values are higher in finer-grained sediments than in sandy sediments. This
suggests that in marine sediments, as in the riverbed sediments, sorption of
lignin on finer sediment is the dominant control on its distribution.
C / V and (Ad / Al)V values are not related to the grain
size in the NW sector (Fig. 7c, d), which suggests that the influence of
grain size on lignin composition and degradation is not as important as in
the riverbed sediments. The control on the degradation of lignin on the inner
Brazil–French Guiana shelf and slope is probably complex and influenced by
many factors, including oxygen exposure time, contribution of material by
coastal rivers, and sedimentation rate. Compared with riverbed sediments,
offshore sediments also exhibit better preservation of organic matter and
selective preservation of lignin in finer grain-size particles, but grain
size has limited impact on the lignin composition and degradation status.
In summary, the spatial patterns of lignin content and isotope compositions
of organic matter corroborates earlier findings (Geyer et al., 1996;
Nittrouer and DeMaster, 1996; Schlünz et al., 1999) that material
discharged by the Amazon River is transported north-westward by the North
Brazil Current. The modern Amazon fan area receives more marine organic
matter, and petrogenic organic matter is a significant component of
OCterr in the Amazon fan sediments. The similarity of lignin
composition (C / V and S / V) of marine and riverbed sediments
suggests that lignin is a reliable tracer reflecting the plant source of
terrestrial organic matter in the Amazon basin and can be applied to
reconstruct vegetation changes and palaeoclimate conditions. Organic matter
and lignin content furthermore vary with sediment grain size in the Amazon
shelf and slope area and show the same preservation trend, better
preservation in finer-grained sediments, as in riverbed sediments. However,
lignin composition is rather uniform in sediments of different grain sizes.