Atmospheric forcing over Holocene productivity in the Alboran Sea (W Mediterranean) from coccolithophores and Mg/Ca-estimated SST records (стр. 2 )

3.2.  Coccolithophore analysis and taxonomy

A total of 85 samples were taken systematically every 2-3 cm. Slides for micropaleontological analyses were prepared following the settling techniques proposed by Flores and Sierro (1997). Qualitative and quantitative analyses were made using a Nikon Eclipse 80-i petrographic microscope with a phase contrast device at 1000x magnification. Nannofossils census counts are based on at least 500 identified specimens in a first count, which is representative for studying the main species (Fatela and Taborda, 2002). In a second count, 25 fields of view were observed in order to avoid underestimation and/or overestimation of those species whose abundance is less than 1% in the first count. Nannofossil Accumulation Rate (NAR) is given in number of coccoliths*cm-2*ka-1, and calculated in every sample considering dry-sediment density, sediment rate and absolute abundance of every species (number of coccoliths*g-1). Relative abundance (%) was also calculated. All the taxa identified in this study have been previously reported in the sampling location (Weaver and Pujol, 1988; Бlvarez et al., 2010). The “small Gephyrocapsa” group is integrated by all Gephyrocapsa specimens smaller than 3 µm (Flores et al., 1999). Along with Emiliania huxleyi (<4 µm), they have been lumped together as “small placoliths” Other taxa identified in this study are Gephyrocapsa caribeanicca, Gephyrocapsa oceanica (specimens larger than 5.5 µm) and “small G. oceanica” (<5.5 µm), Helicosphaera carteri, Syracosphaera spp. and Florisphaera profunda (as majority species). Rare taxa identified were Braarudosphaera bigelowii, Calcidiscus leptoporus, Calciosolenia murrayi, Coccolithus pelagicus subsp. braarudii, Coccolithus pelagicus subsp. pelagicus, Oolithotus fragilis, Pontosphaera spp., Rhabdospahera clavigera, Umbilicosphaera sibogae, Umbellosphaera spp. and Discosphaera tubifera. The last four mentioned have been lumped together as the warm-water group (WWG) due to their common and relative high record in warm waters (McIntyre and Bй, 1967; Okada and Honjo, 1973). Reworked nannofossils are nannoliths pertaining to an older period than the studied. They are usually related to terrigenous material input belonging to the continental margin (Flores et al., 1997; Colmenero-Hidalgo et al., 2004) that once was part of the seabed where these nannoliths deposited.

Preservation of the coccolithophore assemblages is good (little or no evidence of dissolution; diagnostic features fully preserved) (Flores and Marino, 2002). Besides, NAR’s have been transformed into coccoliths*m-2*day-1. The results are comparable to the flux found by Bбrcena (2004) in the same area for sediment traps samples deployed between 1997-1998, proving that dissolution and taphonomic effects are negligible in our core.

3.3.  N ratio

N ratio portrays the relationship between small placoliths inhabiting the upper photic zone (UPZ) and F. profunda, which lives in the LPZ (Flores et al., 2000). Abundance of small placoliths is linked to upwelling intensity and medium to high productivity (Beaufort et al., 1997). F. profunda, on the contrary, proliferates when nutrients are available in the LPZ, and thus, it has been used as an indicator of the depth of the nutricline (Molfino and McIntyre, 1990).

3.4.  Stable isotope

Up to 20 well-preserved tests of planktic foraminifer Globigerina bulloides were picked from the >150 µm size fraction in 66 samples. The individuals were crushed, ultrasonicated and cleaned with methanol before the isotopic analyses within a SIRA mass spectrometer equipped with a VG isocarb common acid bath system at the University of Barcelona. Calibration to the Vienna Pee Dee Belemnite (VPDB) standard scale (Coplen, 1996) was made through the NBS-19 standard, and the analytical precision was better than 0.06‰ for δ18O.

3.5.  Mg/Ca ratio

A total of 60 well-preserved tests of Globigerina bulloides (>150 µm) were picked in 43 samples for Mg/Ca ratio, performed according to Pena et al., (2005) technique. The reductive bath step was halved in order to minimize sample dissolution due to their small size. Routinely along with Mg/Ca ratio, Mn/Ca and Al/Ca ratios were also measured. Every four samples a standard solution was measured in order to correct any derived deviation due to instrumental instability. All samples were corrected according to this. Besides, four blanks were prepared and analysed in order to assure that samples were not contaminated during the analytical process. All solution ICP-MS measurements were carried out at the analytical services unit of the University of Barcelona using a Perkin Elmer Elan6000, quadrupole ICP-MS. Final Mg/Ca values were converted into SST values according to Elderfield and Ganssen (2000) equation.

3.6.  Molecular biomarkers

A total of 86 samples were selected for the analysis of fossil organic compounds (long chain alkenones, alcohols and hydrocarbons), taken every 2-3 cm, and coincident with those used in Mg/Ca analysis when sediment was available. Experimental procedures are described in Villanueva et al., (1997). Samples were analyzed with a Varian Gas Chromatograph (GC) model 450, an autoSampler 8400, Cold On-Column (COC) Injector 1093 and a Flame Ionization Detector (FID). Hydrogen was the carrier gas (2.5 mL/min). C37 unsaturated alkenones (di-unsaturated and tri-unsaturated) are synthesized by coccolith flora. Their identification and quantification allow the calculation of the Uk’37 index, which has been calibrated according to the Mьller et al., (1998) equation in order to measure SST. N-hexacosan-ol and n-nonacosane are allochthonous vascular plant debris. Their relative content is examined through the alcohol index: n-hexacosanol/ (n-nonacosane+n-hexacosanol).

4.  Results

4.1.  Calcareous nannoflora distribution

Small placoliths constitute up to 80% of the nannofossil assemblage on average (Figure 3). G. muellerae (Figure 3) starts a drastic reduction at 10.5 ka cal. BP until 7.7 ka cal BP. where it records a large peak. From 5.7 ka cal BP. across the late Holocene low NAR and scant variability are observed. NAR of the WWG is very low for the whole studied period, except the large peak observed at 7.7 ka cal. BP (Figure 3). F. profunda and small G. oceanica exhibit similar variability (Figure 3): low NAR at 12 ka cal. BP and increasing trend from 10.7 ka cal. BP until 7.7 ka cal BP, when a large peak is observed. From that time up core, they mimic both NAR values and variability. Small placoliths and G. muellerae also exhibit parallel variability during this period. NAR of reworked nannoliths show high variability, especially from 7.7 ka cal. BP onward. Percentage values of this group show large peaks up to 2% of the nannofossil assemblage across the whole period. Total NAR (Figure 3) shows increasing trend from 12.5 to 7.7 ka cal. BP. From that time until 6.5 ka cal. BP, it records its higher values. From 6.5 ka cal. BP to present total NAR records a decreasing trend. N ratio variability (Figure 3) shows a general decreasing trend for the whole studied record. Its values ranges from 1 (maximum value) to 0.9, which is a variation of 10%, showing high variability inside this interval.

4.2.  Isotope Record

From 12 to 7.8 kyr (Figure 4) oxygen isotope record shows a decreasing trend varying a total of 1.48 (‰). From 7.8 ka cal. BP to 3.8 ka cal. BP an increasing trend is observed interrupted by several slight depletions. The total variation during that time is an increase of 0.98 (‰). From 3.8 ka cal. BP to the top, δ18O shows a slight decreasing trend with values ranging a total of 0.43 (‰).

4.3.  Sea Surface Temperature records

Mg/Ca-estimated SST (Figure 4) shows a maximum of 28.7°C at 11.5 ka cal. BP and a minimum of 20.8 °C at 3.2 ka cal BP (Figure 3). A general cooling trend is observed from 9.5 ka cal. BP onward, punctuated by six short warming events, with increases between 2 and 5.4 °C and peaking at 11.5, 9.6, 7.4, 6, 3.8, and 2.5 ka cal. BP.

- SST (Figure 4) ranges between 14.6 at 12.0 ka cal. BP and 20.1°C at 8.9 ka cal. BP. From 12 to 8.9 kyr, it shows a warming trend, which turns into cooling trend from that time to the top with no remarkable short-term changes. Mg/Ca-estimated and -estimated SST records show similar general trend. However, they are inconsistent regarding to their absolute values and internal variability.

4.4.  Alcohol index

From 12 to 7.6 ka cal. BP alcohol index shows decreasing trend. From 7.6 ka cal. BP up core, a general increasing trend is recorded, punctuated by several fast depletions centred at 7.6, 6.5, 6, 5, 4.1, 3.3 and 2.4 ka cal. BP.

5.  Discussion

5.1.  SST reconstruction

Both SST records show a general cooling trend (Figure 4). More in detail, alkenone-estimated SST shows a smooth and monotonous internal variability, while a remarkable point-to-point scatter is seen in Mg/Ca-estimated SST. This later record has been compared with that of core MD95-2043 from the Alboran Sea (Cacho, personal communication, 2013), since the use of the same species discards ecological implications. Both records show fast short-term oscillations, not observed in our the Uk’37-estimated SST record (Figure 4). This discrepancy between quantitative approaches for reconstruction of SST is a normal feature in studies comparing results of these two different biological proxies (Leduc et al., 2010). Emiliania huxleyi, the most prominent alkenone-synthesizing coccolithophorid (Winter and Siesser, 1994) is assumed to represents annual-mean mixed-layer (0-10 m) temperatures (Mьller et al., 1998). However, foraminifera G. bulloides is more abundant in spring (Schiebel and Hemleben, 2000). Acording to Leduc et al., (2010), differences in SST estimates between both methods would be related to paleoenvironmental conditions that affect seasonality and water column depth of the planktic foraminiferal calcification and alkenone production. Nevertheless, previous works have shown that the general cooling trend across the Holocene was affected by several oscillations at a minor time scale (Wanner et al., 2011), as shownby previous Uk’37-estimated SST records from the same study area Cacho et al., (2001). With regard to absolute values, an overestimation of 5°C is recognised in our the Mg/Ca-estimated SST record compared with our the Uk’37-estimated SST record and Mg/Ca-estimated SST from core MD95-2043. This effect could be due to both ontogenetic and analytical implications. Friedrich et al. (2012) analysed Mg/Ca ratio in nine modern planktic foraminiferal species across fourteen mostly 50 µm window sieve fractions. They found that the smaller the picked size, the higher SST was obtained. In core MD95-2043 sieved fraction was restricted to 315-355 µm, while in our core a non-specific fraction >150 µm was sieved. According to analyses developed by Friedrich et al., (2012), these differences in size of specimens and wide window fraction would become a variation of 2.5°C in SST, which could explain up to 50% of the overestimation in our Mg/Ca-estimated SST values. Besides, the reductive bath was halved during the cleaning protocol applied to our samples. Pena et al. (2005) found that one of the Mn phases that is related to high Mg/Ca values is only removed during this reductive bath, implying temperature decreases of 1 to 6.2°C in samples of Neogloboquadrina dutertrei, and of 0.9 to 2.1°C in samples of Globigerinoides ruber. Due to similarities between G. bulloides and G. ruber, we could assume that shortening of the reductive bath could account for uncertain part of SST overestimation in our record. According to this, we consider internal variability of Mg/Ca-estimated SST is reliable. However, we are aware that SST overestimation must be taken with caution and so, we will not interpret Mg/Ca-estimated SST in terms of absolute values. Several peaks shape the internal variability of Mg/Ca-estimated SST record. They consist of successions of warming and cooling episodes that last few hundreds of years, something that is interpreted in terms of SST oscillations at centennial time-scale.