Introduction
The Atlantic Multidecadal Oscillation (AMO), whose frequency varies
between ∼60 and 90 yr (e.g. Schlesinger and
Ramankutty, 1994; Kerr, 2000; d'Orgeville and Peltier, 2007), is
a basin-wide, sea surface temperature (SST) cycle that has been
identified in a number of localities in the North Atlantic Ocean
and the Caribbean Sea over the last 8 kyr interval (Knudsen
et al., 2011). Changes in North Atlantic Ocean SST on these time
scales are sufficiently large to imprint an AMO pattern on global
mean “unforced” temperature variability (Crowley et al.,
2014). Sedimentary evidences include the spectral analyses of the
oxygen isotope (δ18O) record of Lake Chichancanab
(Guatemala) and the Ti% record of the Cariaco Basin
(Venezuela) where 58–61 and 54–60–73 yr periodicities
appear to be dominant from 6 to 1.5 kyr, and 6.5 to
0.9 kyr, respectively (Knudsen et al., 2011). These results
are in accord with the suggestion of a southern migration of the
intertropical convergence zone (ITCZ) after 3 kyr as
evidenced in the Cariaco record (Haug et al., 2001). This ITCZ
migration trend has also been documented in coastal terraces in the
southwest Caribbean Sea (Martinez et al., 2010), and the Cauca
paleolake in the northern Andean region (Garcia et al., 2011;
Martinez et al., 2013). This region is influenced by the annual
migration of the ITCZ, and El Niño – Southern Oscillation
(ENSO), and the North Atlantic Oscillation (NAO) phenomena. Besides
these, the northern Andes, which are more than 2000 m high,
impose a significant barrier to the interaction between Atlantic
and Pacific wind regimes, resulting in climate dynamics at decadal
to centennial time scales that are still poorly understood. At
decadal timescales, the 60 yr component of the Pacific Decadal
Oscillation (PDO; e.g. Kayano and Andreoli, 2007), appears to be
time-lag correlated with the AMO (Orgeville and Peltier,
2007). Evidence suggesting that the dynamics of the South American
Monsoon (SAM) system is driven by the AMO cycle have been reported
for the Amazon basin, and are found as far south as 30∘ S
in the South Atlantic Ocean (e.g. Chiessi et al., 2009; Vuille
et al., 2012). When AMO is in a positive (warm) phase, the ITCZ
moves north leading to increased precipitation in western Europe
and the Sahel and reduced precipitation in North America and
northeastern Brazil. Conversely, when AMO is in a negative (cold)
phase the ITCZ moves south and opposite precipitation conditions
occur (e.g. Kayano and Capistrano, 2014; Garcia-Garcia and
Ummenhofer, 2015).
Here we explore the San Nicolás sedimentary succession of the
Cauca paleolake (6∘30′ N;
75∘50′ W), represented by the San Nicolás-1
core, as a unique high-resolution paleoenvironmental and
paleoclimate record for the late Holocene. The sedimentary
successions of the Cauca paleolake were recognized in three terrace
levels, deposited during the late Holocene, that were formerly
attributed to lacustrine sedimentation as the product of the
episodic damming of the Cauca river by landslides in the Liborina
region (Page and Mattson, 1981). In accompanying papers we: (1)
document the palynofacies content of the San Nicolás succession
in response to hydrological connectivity of the Cauca River with
its tributaries and climate dynamics in northern South America
(Garcia et al., 2011) and, (2) demonstrate that the formation of
the terrace deposits was not due to landslides, but to
sedimentation in a ria lake environment that changed from
igapo (black water) to varzea (white water)
conditions in the ∼3.5 to ∼1 kyr BP interval
with sediment accumulation rates in excess of
600 cmkyr-1 (Martinez et al., 2013). The hydrological
connectivity of the Cauca river with its tributaries, at seasonal
to decadal time scales, resulted in a unique, high-resolution
record of laminated sediments. In this previous contribution
(Martinez et al., 2013), La Caimana, and the Sucre II field
sections were presented, together with the San Nicolás-1
core. The latter record, which is unaffected by weathering
processes common in the field sections, is documented herein in
detail, together with the analysis of its time series. We
demonstrate that in addition to the ENSO signal there is a strong
multidecadal component, which matches the AMO frequency thus
demonstrating its influence south of the Cariaco Basin, in the
northern Andes, during the late Holocene.
Climatology and hydrology of the present Cauca Valley
Presently, climate in the Cauca Valley, as for most of the northern
Andes, is dominated by a bimodal regime, i.e. two rainy seasons
during March–May and September–November and two dry seasons
during December–March and June–August (e.g. Poveda et al.,
2006). This precipitation pattern results from the annual migration
of the ITCZ. Conversely, the northern Andes act as a major barrier
creating a very dynamic atmospheric system characterized by
meso-scale convective cells and intense precipitation, mostly on
the western and eastern flanks of the Western and Eastern
Cordilleras, respectively.
In the eastern equatorial Pacific, the westerly low-level Choco
jet, whose intensity is controlled by the sea surface temperature
gradient between the cold tongue (south of the equator) and the
Panama Basin, annually shifts with the ITCZ, and inter-annually
with the ENSO phenomenon (Poveda and Mesa, 2000; Poveda et al.,
2006, 2011). The Choco jet is forced by orographic lifting to
deliver most of its moisture on the western flank of the Western
Cordillera, thus producing a rain shadow or “dry island” effect
on the inter-montane Cauca and Magdalena Valleys. However, through
the Mistrato pass, located at 5∘ N on the Western
Cordillera, some moisture reaches the Cauca Valley particularly
between September and October when the ITCZ is in the north (Poveda
and Mesa, 2000). This has an important effect on the hydrology of
the middle Cauca Valley, and the Cauca paleolake, the subject of
the present study.
Precipitation shows decadal to multi-decadal variability, which is
smaller than yearly changes, and appears to be related to the
PDO. However, this is a relation based on few climate stations with
long time records (Garreaud et al., 2009).
Total annual precipitation in the Cauca Basin is 1887 mm,
with 243 mm during the maximum month (Restrepo et al.,
2005a). Furthermore, there is an inverse relationship between
precipitation, induced by the positive phase of ENSO (El Niño),
and water discharge and suspended sediment yield in the Cauca
River, whose annual averages are 2373 m3s-1, and
49.1 Mta-1, respectively (Restrepo et al.,
2005a). The Cauca hydrographic basin, that is
59 615 km2 large and extends from 2 to >8∘ N (Restrepo et al., 2005b), is the recipient of
precipitation from the ITCZ on its annual path (Fig. 1). This would
make it difficult to reconstruct the ITCZ mean position in the past
at a particular site in the middle Cauca Valley. Analogously, this
also appears to be the case for the reconstruction of the ENSO
phenomenon, whose influence is broad in northern South America,
spreading from the Ecuadorian Andes to the Cauca Valley
(e.g. Garreaud et al., 2009; Poveda et al., 2011).
Because the northern Andes act as a barrier to the South American
Monsoon System (SAMS), it is expected that the North American
Monsoon System (NAMS), has a major influence on the inter-montane
Cauca and Magdalena valleys that drain to the Caribbean Sea. Both
monsoon systems, however, are considered to be two extremes of the
same climate cycle (Vera et al., 2006).
The Cauca Valley runs along the Cauca – Romeral Fault system that
is a major structural suture between the Western Cordillera, of
Cretaceous volcano-sedimentary oceanic origin, and the Central
Cordillera, of Paleozoic – Mesozoic plutonic origin (e.g. Cediel
et al., 2003). In the northern Cauca Valley the Cauca–Romeral
Fault System is inverse, sinestral, and braided. This suggests that
the Santa Fé–Sopetrán depression is a pull-apart basin,
limited by the NW–SE Sopetrana, the NE–SW Cauca, and the N–S San
Jerónimo Faults (Suter and Martínez, 2009; Suter et al.,
2011). The Santa Fé–Sopetrán depression is partly filled
with late Holocene fluvio-lacustrine sediments. The middle Cauca
Valley is a steep and narrow valley, thus controlling the course of
the Cauca River, which is entrained and mostly braided. Therefore,
it has a high bed load, and is relatively unstable (cf. Schumm,
1981, 2005; Schumm et al., 2000).
Methods
The San Nicolás-1 core was retrieved, by rotary drilling, from
the San Nicolás terrace, about 100 m south of La
Caimana Creek (Fig. 1). Core sections, about 50 cm long,
were transported to the laboratory where they were longitudinally
cut and stored at 4 ∘C to prevent oxidation and
degradation of organic matter. Following Nederbragt et al.'s
(2004) recommendations, extreme care was taken during the
acquisition of digital picture images of the core sections. After
several trials with different source lights, digital pictures were
taken with natural, indirect, light. Because of the limited length
(50 cm) of each section, and the core retrieval technique
used, some noise remained due to the combination of several
variables, including light distribution, cracks, and core
voids. These were digitally corrected using the image analysis
software NIH ImageJ version 1.4, in order to obtain a composite
section and a grey scale signal. The chronology of the San
Nicolás-1 core is based on 14 radiocarbon (accelerator mass
spectrometry, AMS14C) analyses on bulk sediment samples,
i.e. carbonaceous muds. Analyses were done at the Australian
National University radiocarbon facility (Martinez et al., 2013).
Calibrated age ranges were determined from the University of
Oxford Radiocarbon Accelerator Unit calibration program
(OxCal4.1). Therefore, in the present paper all ages indicated
refer to calibrated years before present (calyrBP).
Results
The sedimentary succession at the San Nicolás-1 core is
22.14 m thick (Figs. 2 and 3). It consists of laminated
clays and silts, which are conspicuous all along the succession,
whereas fine-grained sands are restricted to its top. Color varies
from greenish dark grey (10BG4/1) to bluish grey (5B6/1) at the
base, to dark red (2.5YR3/6) and yellowish brown (2.5Y3/6) moving
upward in the section until 16.4 mbt (meters below the top) where
the grey colors are less frequent and are replaced by yellowish
light brown (2.5Y6/4), reddish brown (7.5YR4/3), grey (10YR5/1),
and white (5Y8/1), among others (Fig. 2). This change in color
reflects organic matter (OM) content and iron oxidation state
(Martínez et al., 2013), which is reflected in the dark-light
fluctuations of the grey-scale (Fig. 3).
Bioturbation along the San Nicolás succession are of animal
origin at the bottom to plant origin at the top. In both cases,
bioturbation is low and does not destroy lamination. As documented
in the field sections and the San Nicolás-1 core, vertical
burrows are <0.8 cm in diameter, internally contain
spreite structures and are conspicuous in the lower part of the
succession, whereas horizontal burrows occur sporadically all
through the succession. They were assigned to the Scoyenia
and Mermia ichnofacies (Buatois and Mangano, 2009),
respectively (Martinez et al., 2013).
Chronology of the San Nicolás succession
Except for the lowermost and the uppermost parts of the sections,
where there are no chronologic controls due to the absence of
enough OM for radiocarbon analysis, seven samples provided
reliable data for building the age model by linear interpolation
(Fig. 3). As previously explained, samples considered as the
product of reworking of OM and outliers, were excluded from the
age model (Martinez et al., 2013). This suggests that the
succession ranges in age from ∼4000 to <500 calyrBP and was deposited at an accumulation rate
that exceeded ∼600 cmkyr-1 (Fig. 3).
Image and wavelet spectrum analyses
From the composite image (Fig. 2) a gray scale record was obtained
for the San Nicolas core (Fig. 3). Then, this was separated into
its components: red, green and blue. The red channel component was
transformed to derive a time series signal (Fig. 4a). It was
de-trended by subtracting the first reconstructed component,
indicated by Singular Spectral Analysis to contain the trend, then
filtered to remove frequencies larger than 5 yr-1. Wavelet
spectrum analysis of the 3750 to 350 yr PB time interval
(Fig. 4b) was performed on the red scale data following Obrochta
et al. (2012) method, using Matlab (Torrence and Compo, 1998).
Significant, large variance (95 % confidence), 70 year peaks,
appear at 3100, 2200, 1600 and 1000 yr BP; whereas,
secondary peaks in the 8 to 32 yr band occur all over the
3750 to 350 yr PB time interval (Fig. 4b).
Discussion
Analogously to the Chichancanab Lake and the Cariaco Basin, we
interpret the 70 yr periodicity as the AMO mode of climate
variability. In particular, the reconstructed AMO pattern for the
San Nicolás-1 core is very similar to the Cariaco Basin
pattern for the 3750 to 350 yr BP time interval. Major
differences might be due to age model uncertainties, and the
potential bias introduced by the core retrieval technique used and
the handling of core sections during drilling of the San
Nicolás hole. Despite these possible sampling biases, which
potentially mask and overprint the natural cycle, the hypothesis
that AMO forcing controls the hydrological regime of the Cauca
River basin, is supported as the evidence suggest that the
dynamics of the SAM system is driven by the AMO cycle (e.g. Vuille
et al., 2012). Assuming this is true, the AMO forcing might have
controlled the discharge of the Cauca River and the episodic
flooding of its tributaries, which resulted in ria lakes.
As documented for La Caimana Creek, there appears to be
a conspicuous change in the hydrological regime of the Cauca
River, from a igapo (black-water) to a varzea
(white-water) regime ca. 3000 yr BP (Martinez
et al., 2013). This is at 16.4 mbt when color change from grey to
yellow-red (Fig. 2), and charcoal concentration abruptly drops in
tandem with the steady increase in altered lignocellulose debris
(Garcia et al., 2011; Martinez et al., 2013). All this indicates
that precipitation significantly increased in response to the
southern migration of the ITCZ, with its mean location probably
lying over the upper reaches of the Cauca River, i.e. at about
2–3∘ N.
It has been hypothesized that the warm phase of the AMO results in
a weaker ENSO variability (Dong et al., 2006) and/or, is related
to the PDO (d'Orgeville and Peltier, 2007). Both, the Cariaco
Basin and the Cauca paleolake records suggests otherwise, as ENSO
variability appears to increase between 3800 and
2800 yr BP (Haug et al., 2001) when the AMO signal is not
particularly significant as shown by Knudsen et al. (2011) and
herein. By contrast, for the Galapagos Islands it has been
suggested that precipitation and ENSO variability (more El
Niño events) increased between 2000 and 1500 yr BP
(Conroy et al., 2009), thus agreeing with a positive AMO signal in
both, the Cariaco Basin and the Cauca paleolake.
The results we have obtained are very encouraging and underline
the ria lake sedimentary successions as potential high-resolution
records for paleoclimate research in the Neotropics. Nonetheless,
we need to retrieve a new, more complete core with fewer section
breaks, in order to test our AMO hypothesis obtained from the
wavelet analysis of the San Nicolas-1 core. This will rule out any
potential bias derived from coring artifacts.