3. Análisis de gastos
3.2. Estructura del gasto
because direct radiometric age constraints exist only for the most recent (early to middle Holocene) sapropel S1 (e.g., Lourens et al., 1996; Mercone et al., 2000; Casford et al., 2007;
De Lange et al., 2008). On that limited basis, an astronomically-tuned timescale for Mediterranean sediments has been proposed that assumes a 3-kyr lag between precession minima and sapropel mid-points (Hilgen et al., 1993; Lourens et al., 1996, 2004). However, the assumed extrapolation of the specific phase relationship of S1 to older sapropels remains to be tested (e.g., Ziegler et al., 2010a). We have recently presented a radiometrically constrained chronology that spans four sapropels (Grant et al., 2012), which allows such a test to be performed.
Core LC21 in the eastern Mediterranean (Fig. 4.1) contains four sapropels (S1, S3, S4, S5), and – in addition to a radiometrically constrained chronology – has been synchronised with the Red Sea relative sea-level (RSL) record (Grant et al., 2012). Hence, it is an ideal platform for investigating phase relationships between monsoon variability, insolation forcing and global ice-volume fluctuations on orbital timescales. We use it to test the following hypotheses: 1) there is a consistent 3-kyr offset between precession minima and sapropel mid-points, 2) there is a systematic phasing between northward penetration of the EAfSM and changes in global ice volume, 3) the concept of a global monsoon is valid.
Figure 4.1 Locations of marine cores and speleothem caves discussed in this study, in the context of the East Asian monsoon (EAM), Indian monsoon (IM), East African summer monsoon (EAfSM) and West African monsoon (WAM).
76
4.2 MethodsScanning x-ray fluorescence (XRF) elemental analyses of the archive halves of sediment core LC21 (southern Aegean Sea, 35° 40' N, 26° 35' E; Fig. 4.1), were performed at the British Ocean Sediment Core Research Facility (BOSCORF) at the National Oceanography Centre, Southampton, using an Itrax XRF core scanner from Cox Analytical Systems (Gothenburg, Sweden). XRF data were collected every 0.5mm down-core using a
molybdenum tube set at 30 kV and 30mA, and a sampling time of 40 seconds directly at the core surface. The exposed core surface was covered with a 4 micron thin SPEX Certi Prep Ultralene1 foil to avoid contamination of the XRF measurement unit and desiccation of the sediment. Subsequent sub-sampling and stable isotope analyses of the same core halves have been described in Grant et al. (2012). The foraminiferal stable oxygen and carbon isotope records discussed below are for the surface-dwelling species Globigerinoides ruber (white) (δ18Oruber, δ13Cruber) and the sub-surface dwelling species Neogloboquadrina pachyderma (dextral) (δ18Opac, δ13Cpac). Tests of these species were selected from >300 m and 150-300
m sieved sediment fractions (see Methods in Grant et al., 2012).
4.3 Results
Barium and stable isotope profiles are used to accurately delineate the depths of sapropel boundaries (Fig. 4.2), which are here used to define past intervals when the East African summer monsoon (EAfSM) penetrated north of ~21°N (see Rohling et al., 2002a, 2004;
Larrasoaña et al., 2003; Osborne et al., 2008). Redox reactions at the sediment-seawater interface affect the preservation of organic-rich deposits. As a consequence, elements that are enriched in sapropels and which exhibit ‘conservative’ behaviour in sediments are most suitable for reconstructing down-core sapropel depths. Barium is ideal for this purpose because it is well-preserved in sediments (Dymond et al., 1992), and enriched in sapropels (Thomson et al., 1995; De Lange et al., 2008) due to its association with the decomposition of organic matter in the water column, which in turn is coupled to primary productivity (Dymond et al., 1992).
In core LC21, pronounced increases in Ba at sapropel horizons are accompanied by elevated Vanadium (V) and depleted δ18O and δ13C (Fig. 4.2).Vanadium is a redox-sensitive element and precipitates under reducing conditions, so although it should not be used alone to define sapropel boundaries, the good agreement between increases in V and Ba in core LC21 implies that – in this case – elevated V reliably indicates sapropel boundaries (see Thomson
77
et al., 1995;Nijenhuis et al., 1999
). Strongly depleted δ18Oruber and δ18Opac values in LC21 sapropels far exceed inferred δ18O changes associated with global ice-volume reduction at these times (see Fig. 1 in Grant et al., 2012), and are synchronous with negative excursions in δ13Cruber and δ13Cpac (Fig. 4.2). The δ18Oruber and δ18Opac depletions are unlikely to primarily reflect warming of surface and sub-surface waters or an increase in net precipitation over evaporation because neither of these processes would result in a synchronous depletion in both surface- and subsurface-water δ13C. The observed trend is easily explained, however, by monsoon-driven flooding of the North African margin, which would result in asubstantial input of isotopically light freshwater and terrestrial carbon into the eastern Mediterranean. Together, therefore, the LC21 Ba, V, δ18O and δ13C records effectively delineate the intervals of sapropel formation and preservation in core LC21, and thus the intervals of northward expansion of the EAfSM.
Figure 4.2 Photograph of core LC21 (in metres below sea floor, mbsf) with scanning XRF profiles (21-point moving average) of barium and vanadium, and planktonic foraminierfal δ18O and δ13C records from G. ruber (white) and N. pachyderma (d). Sapropels are indicated (grey rectangles).
78
4.4 Discussion4.4.1 Sapropel deposition and insolation maxima/precession minima
Here we consider the timing of sapropel deposition relative to the summer inter-tropical insolation gradient (“SITIG”, Fig. 4.3) and the precessional component of insolation. The advantage of using the SITIG rather than a specific latitudinal insolation curve is that it accounts for changes in tropical insolation in both hemispheres, and therefore best captures variability in the insolation forcing of the intertropical convergence zone (ITCZ), which in turn drives the intensity and spatial distribution of monsoon precipitation. We find that the onset of sapropels S3-S5 in core LC21 occurred ~1.5-3 kyr before insolation
maxima/precession minima (Fig. 4.3b-d), whereas sapropel S1 deposition began 0.8 kyr after the Holocene insolation maximum/precession minimum (Fig. 4.3a). Previous studies inferred a lag of ~3 kyr between the mid-point of S1 and the nearest insolation
maximum/precession minimum (Lourens et al., 1996; De Lange et al., 2008; Ziegler et al., 2010a). We find a comparable phase offset if we considering the S1 midpoint (Fig. 4.3a). It is clear that, regardless of whether sapropel bases or mid-points are used, the concept of a consistent 3-kyr phase offset between precession minima/insolation maxima is valid only within broad tolerances of a few thousand years.
The differences between phase relationships determined using either sapropel bases or mid-points are noteworthy. Sapropel mid-mid-points give the average timing of sapropel
formation/preservation and are therefore an appropriate approximation for cross-spectral phase analyses, which typically rely on peak-to-peak signal comparison at a specific frequency. However, the timing of a mid-point depends on the duration of a sapropel, and this clearly varies – both in absolute terms (number of years) and relative to insolation changes. For example, according to the radiometrically constrained chronology of LC21, sapropels S3 and S4 coincide with the interval of maximum insolation values (>520 W/m2) (Fig. 4.3c,d), but sapropels S1 and S5 extend until insolation drops to ~500 W/m2 (Fig.
4.3a,b).
The above strongly suggests that, during full interglacials (i.e., considering S1 and S5), EAfSM precipitation and/or northward expansion is sufficiently strong to prolong sapropel deposition under waning insolation, which is consistent with positive vegetation-albedo feedbacks on the meteorological conditions (Nicholson, 2009). Such changes in the duration of sapropels due to indirect (feedback) processes are unlikely to be the same for all
sapropels, and this compromises the validity of the concept of a ‘systematic’ phase
79
relationship between precession and sapropel mid-points. Because vegetation-albedofeedbacks have not yet had the time to develop at around the time of onset of sapropel deposition, it seems more promising to evaluate phase relationships between precession and the onset/base of sapropels. By that criterion, the lagged onset of sapropel S1 (relative to peak insolation) is clearly unique relative to other sapropels within the last glacial cycle.
Figure 4.3 Insolation and ice-volume changes during sapropel deposition (= EAfSM maxima) under full interglacial conditions (a, b) and periods of glacial inceptions (c, d). A maximum probability sea-level curve (dark turquoise) with 95% confidence limits (pale turquoise) (Grant et al., 2012) is based on the Red Sea sea-level reconstruction method (Siddall et al., 2003, 2004; Rohling et al., 2009). The subtropical insolation gradient (SITIG, dashed orange) is calculated as the difference between mid-summer insolation at 23˚N and 23˚S, and has been normalised by unit variance, and then negated, in order to plot on the same axis as the normalised precession curve (orange).
Previous work has related the apparently ‘delayed’ onset of S1 to millennial-scale cooling in the North Atlantic affecting atmospheric circulation over the Mediterranean (Rohling, 1994;
Ziegler et al., 2010a). Deep-water convection in the eastern Mediterranean is highly sensitive to cold northerly airflows (see Rohling et al., 2002b; Casford et al., 2003, and