Radiocarbon Age Offsets Between Two Surface Dwelling Planktonic Foraminifera Species During Abrupt Climate Events in the SW Iberian Margin.

This study identifies temporal biases in the radiocarbon ages of the planktonic foraminifera species Globigerina bulloides and Globigerinoides ruber (white) in a sediment core from the SW Iberian margin (so-called Shackleton site). Leaching of the outer shell and measurement of the radiocarbon content of both the leachate and leached sample enabled us to identify surface contamination of the tests and its impact on their 14C ages. Incorporation of younger radiocarbon on the outer shell affected both species and had a larger impact downcore. Interspecies comparison of the 14C ages of the leached samples reveal systematic offsets with 14C ages for G. ruber being younger than G. bulloides ages during the last deglaciation and part of the Early and mid-Holocene. The greatest offsets (up to 1,030 years) were found during Heinrich Stadial 1, the Younger Dryas, and part of the Holocene. The potential factors differentially affecting these two planktonic species were assessed by complementary 14C, oxygen and carbon isotopes, and species abundance determinations. The coupled effect of bioturbation with changes in the abundance of G. ruber is invoked to account for the large age offsets. Our results highlight that 14C ages of planktonic foraminifera might be largely compromised even in settings characterized by high sediment accumulation rates. Thus, a careful assessment of potential temporal biases must be performed prior to using 14C ages for paleoclimate investigations or radiocarbon calibrations (e.g., marine calibration curve Marine13, Reimer et al., 2013, https://doi.org/10.2458/azu_js_rc.55.16947).

Introduction

For decades, fossil planktonic foraminifera have been a valuable source of paleoceanographic information, providing proxies for variations in ice volume, sea level, salinity, temperature, and nutrients (e.g., Pearson, 2012). Since the discovery of the radiocarbon (14C) dating technique in the late 1940s (Libby et al., 1949), radiocarbon age determination of planktonic foraminifera has become a cornerstone for paleoclimate investigations spanning the last 50,000 years. Most studies rely on this method to build chronostratigraphic frameworks for marine sediment sequences and constrain changes in thermohaline circulation by estimating radiocarbon ventilation ages. However, prior works have demonstrated that planktonic foraminifera 14C ages might not always be a reliable indicator of their depositional ages due to numerous causes, as summarized by Mekik (2014). For instance, contamination trough radiocarbon addition by secondary calcite precipitation or adhesion of atmospheric carbon, which can go unnoticed during visual sample inspection under an optical microscope, can lead to large deviations in 14C ages (Wacker et al., 2014; Wycech et al., 2016). Other possible causes of temporal biases include bioturbation along with differential dissolution and fragmentation (Barker et al., 2007, and references therein), differential bioturbation coupled with species abundance gradients (e.g., Bard et al., 1987b), transport and deposition of reworked specimens (Broecker et al., 2006), and distinct calcifying habitats (Lindsay et al., 2015). All these might differentially affect foraminifera species, and their influence on foraminifera 14C ages might be largely overlooked if, as in most paleoinvestigations, only samples of one species are analyzed per sediment horizon. Thus, a more thorough assessment of the potential temporal biases between co‐occurring foraminifera species is required prior conducting investigations primarily based on climate signals derived from foraminifera tests. Given age discrepancies might exceed the duration of abrupt climate events (>1,000 years; Mekik, 2014), important questions arise in relation to the applicability of the latter approach in regions where marine sediments have a unique potential to unravel rapid climate and environmental changes.

In this regard, the so‐called Shackleton sites, MD95‐2042 and Integrated Ocean Drilling Program (IODP) Site U1385, on the SW Portuguese margin constitute benchmark cores for paleocenographic studies. For instance, Bard et al. (2004) produced a downcore sequence of 14C ages in core MD95‐2042, which was incorporated into IntCal09/Marine09 (Reimer et al., 2009) and subsequent updates (Reimer et al., 2013). This location has also emerged as one of the few regions in the world where direct correlation of marine signals with both Greenland and Antarctic ice core signals is feasible (Shackleton et al., 2000), detailed chronostratigraphies have been developed (e.g., Bard et al., 1987a; Shackleton et al., 2004), and where ventilation and reservoir ages have been studied (Skinner et al., 2014; Skinner & Shackleton, 2004), all these based on 14C ages of one species of planktonic foraminifera per sediment horizon.

Despite the importance attached to this location and prior works posing severe pitfalls to the latter approach, assessment of potential temporal biases trough 14C determinations on paired species‐specific samples has not yet been conducted. Consequently, potential temporal biases might have been disregarded in derived paleoclimate interpretations from this key study area. We aimed at identifying possible temporal biases in the 14C ages of planktonic foraminifera species, analyzed in samples from a sediment core retrieved close to the location of IODP Site U1385, and assessing the potential causes for age deviations. To accomplish this, we investigated paired 14C ages of two of the most commonly used planktonic foraminifera species: and (white) and measured complementary oxygen (δ18O) and carbon (δ13C) isotopes, and species abundance data to elucidate possible reasons why radiocarbon ages may diverge for different foraminifera species from the same sample.

Study Area

The SW Iberian margin (NE Atlantic Ocean) is a transitional region where the Portugal Current, a branch of the North Atlantic Current, flows southward year round (Figure 1a; Brambilla et al., 2008; Pérez et al., 2001). From October to March, the Iberian Poleward Current, a branch from the Azores Current, flows poleward along the W Portuguese margin (Haynes & Barton, 1990). This shift in the near‐shore surface circulation is linked to the seasonal changes in the regional atmospheric circulation, which determine two well‐differentiated oceanographic regimes. From March/April to September/October, prevailing northeasterly winds may induce Ekman transport offshore and subsequent upwelling of subsurface waters. During the rest of the year, coastal downwelling occurs under prevailing southwesterly winds (Peliz et al., 2005). Upwelled subsurface (100–500 m) waters consist in North Atlantic Central Water of either subtropical (NACWst; 100–250 m) or subpolar (NACWsp; 250–500 m) origin. The warmer and nutrient‐poor NACWst overlies the colder, nutrient‐richer NACWsp, which only upwells during strong upwelling events. Below the NACW, the denser Mediterranean Outflow Water flows poleward between 500 and 1,700 m. Below the intermediate waters, the Northeast Atlantic Deep Water flows southward (van Aken, 2000), along with varying contributions of the Upper Circumpolar Deep Water, the Upper Labrador Sea Water, and the Antarctic Bottom Water (Jenkins et al., 2015).

Figure 1

Location of core SHAK06–5K and age‐depth model. Study area and surface circulation. PC: Portugal Current. IPC: Iberian Poleward Current. Modified from Voelker and de Abreu (2011).

Materials and Methods

We analyzed downcore sediment samples from kasten core SHAK06–5K (37°34′N, 10°09′W, 2,646 m), recovered by RSS James Cook during the cruise JC089 in 2013 in the vicinity of the Shackleton Sites (Hodell et al., 2014).

Radiocarbon Determinations

The majority of the organic matter contained in the initial sediment was extracted with organic solvents following Ohkouchi et al. (2005) to use the organic fraction in a follow‐up investigation. To assess the possible influence of this procedure on the foraminifera contained in the solvent‐extracted residue, we also analyzed five samples of tests selected from nonextracted sediments. Between 15 and 30 g of dry sediment were diluted in MiliQ® water and sonicated for only 15 s for disaggregation while avoiding shell fragmentation. The solution was then wet sieved through 300‐ and 250‐μm mesh sieves and thoroughly washed using a high‐pressure stream of MiliQ® water. The resulting 250‐ to 300‐μm size fraction was immediately dried at 60 °C overnight, prior to collecting 45–100 well‐preserved shells of or from each sample. In some intervals, only 7–20 specimens of were available, limiting the amount of measured carbon (Tables S1 and S2 in the supporting information). Radiocarbon determinations (14C/12C) were performed with a gas ion source in a Mini Carbon Dating System at the Laboratory of Ion Beam Physics, ETH Zürich, with an automated method for acid digestion of carbonates whose sensitivity allows for less than10 μg of total carbon to be measured (Wacker et al., 2013). The method is outlined as follows: vials (septa sealed 4.5‐ml exetainers vials from Labco Limited, UK) containing the samples were purged for 10 min with a flow of 60 ml/min He to remove atmospheric CO2. Later, samples were briefly leached by adding 100 μl of ultrapure HCl (0.02 M) with an automated syringe to remove possible surface contaminants. The CO2 released from the leachate, referred to as leachate was transported by helium to a zeolite trap and automatically injected into the ion source to be measured for radiocarbon. The remaining sample, containing 12 μg C and referred to as leached sample, was subsequently acidified by adding 100 μl of ultrapure H3PO4 (85%) that was heated to 60 °C for at least 1 hr. The released CO2 was loaded in a second trap and injected into the ion source to be analyzed for radiocarbon (Wacker et al., 2014). Bard et al. (2015) showed that the F14C (fraction modern according to Reimer et al. (2004)) of leachates from sequential leaching of discrete samples converge toward a comparable value to that of the F14C of the leached sample (Bard et al., 2015). Thus, we propose differences <5% between the two values as an indication of near‐complete removal of surface contaminants. Five replicates of samples, referred to as untreated, were directly measured without leaching the outer shell to assess the necessity of this method. This gas ion source Accelerator Mass Spectrometry (AMS) system has a background 14C/12C value of F14C 0.0020 + −0.0010 (50000 BP), determined on marble (IAEA‐C1). Radiocarbon determinations were corrected for isotopic fractionation via 13C/12C isotopic ratios and are given in conventional radiocarbon ages. Radiocarbon ages and errors were not rounded to avoid artificial increments of age offsets and propagated errors.

Age‐Depth Model

The age‐depth model for core SHAK06–5K is a depositional model (P_Sequence type) based on 41 14C ages of monospecific samples of (Table 1) built with the calibration package Oxcal (Bronk Ramsey, 2009). Conventional radiocarbon ages were calibrated to incorporate a static marine reservoir effect using Marine13 curve (Reimer et al., 2013). The resulting age‐depth model spans the last 28,000 years.

Table 1

Age Model for Core SHAK06–5 K, Based on Monospecific Samples of the Planktonic Foraminifera

Laboratory code	Depth (cm)	Radiocarbon age (14C year BP) ± 1σ	Calendar age (year cal. BP) ± 2σ
82182.2.1	0	790 ± 150	414 ± 112
82183.2.1	4	1,010 ± 150	591 ± 92
72979.2.1	10	1,250 ± 70	815 ± 72
82185.2.1	14	1,450 ± 70	1,001 ± 73
72981.2.1	20	1,820 ± 55	1,367 ± 60
72983.2.1	30	2,300 ± 50	1,920 ± 60
72985.2.1	40	3,090 ± 65	2,879 ± 82
75040.1.1	44	3,620 ± 75	3,514 ± 86
70397.1.1	48	3,760 ± 60	3,702 ± 82
75041.1.1	54	5,300 ± 80	5,670 ± 86
72987.2.1	60	7,470 ± 60	7,923 ± 68
72989.2.1	70	8,740 ± 70	9,404 ± 70
75042.1.1	76	9,960 ± 80	10,925 ± 128
72991.2.1	82	11,050 ± 85	12,566 ± 75
72993.2.1	90	11,450 ± 90	12,913 ± 108
70400.1.1	100	12,100 ± 110	13,517 ± 112
72995.2.1	110	12,400 ± 100	13,909 ± 117
72997.2.1	120	13,250 ± 95	15,276 ± 141
70403.1.1	130	13,600 ± 110	15,875 ± 149
72999.2.1	140	14,100 ± 100	16,522 ± 158
75043.1.1	146	14,300 ± 100	16,864 ± 161
73001.2.1	152	14,900 ± 100	17,527 ± 121
73002.2.1	160	14,900 ± 110	17,742 ± 113
73003.2.1	172	15,350 ± 110	18,219 ± 133
73005.2.1	180	15,950 ± 140	18,791 ± 122
75044.1.1	196	16,650 ± 120	19,642 ± 155
75016.1.1	200	17,100 ± 120	19,989 ± 143
75018.1.1	210	17,300 ± 120	20,347 ± 130
75020.1.1	220	17,400 ± 140	20,679 ± 162
75022.1.1	230	18,600 ± 180	21,899 ± 180
75024.1.1	240	18,750 ± 140	22,241 ± 131
70406.1.1	260	20,000 ± 180	23,537 ± 200
75028.1.1	270	20,400 ± 150	24,012 ± 156
75030.1.1	280	20,700 ± 150	24,482 ± 179
75048.1.1	284	21,000 ± 160	24,781 ± 215
75032.1.1	290	21,300 ± 160	25,245 ± 186
75033.1.1	300	22,100 ± 170	25,936 ± 125
75034.1.1	310	22,600 ± 180	26,416 ± 184
75036.1.1	320	23,000 ± 180	26,974 ± 210
75038.1.1	329	24,100 ± 200	27,800 ± 163

Note. Convention radiocarbon ages and associated 1σ uncertainties have been rounded according to convention.

Scanning Electron Microscope Imagery

Representative well‐preserved specimens were selected from discrete intervals to assess surface preservation and possible early diagenetic overgrowth. Samples were graphite coated and scanning electron microscope (SEM) images were generated using a JEOL JSM‐6390LA digital SEM with a W filament.

Oxygen and Carbon Stable Isotope Analyses

Oxygen and carbon stable isotope analyses were determined every 2 cm when possible. In total, 164 samples of and 140 samples of were considered. Between 6 and 12 specimens of each species were measured with a Gas Bench II connected to a Delta V Plus isotope ratio mass spectrometer at the Stable Isotope Laboratory of Climate Geology, ETH Zurich (Breitenbach & Bernasconi, 2011). Calibration to the Vienna Pee Dee Belemnite (VPDB) scale was accomplished using two in‐house standards previously calibrated against the NBS‐18 and NBS‐19 international standards. The associated long‐term standard deviation is <0.07‰.

Species Abundance

Representative aliquots of the 250‐ to 300‐μm size fraction, containing at least 300 planktonic foraminifera shells, were obtained with a splitter. The relative and absolute abundances of and were analyzed in 33 samples spaced every 10 cm. Absolute abundances were calculated using the dry weight of the initial sieved sample.

Results

Radiocarbon ages of. samples from both extracted and nonextracted sediments show younger leachates (up to 2,000 years) compared to the corresponding leached samples (Figure 2 and Table 2). The leached samples from both types of sediments agree very well within their 1σ error.

Figure 2

Influence of the sample preparation method on radiocarbon ages. (a) 14C ages of the leachate (open circle) and the leached samples (dot) of picked from sediments extracted with organic solvents (light blue) and nonextracted sediments (dark blue). (b) Age differences between paired leachates and leached samples from extracted (light blue) and nonextracted (dark blue) sediments, and between paired leached samples (black diamonds).

Table 2

Influence of the Sample Preparation Method on Radiocarbon Ages

Depth (cm)	G. bulloides from nonextracted sediments					G. bulloides from sediments extracted with organic solvents					G. bulloides‐G. bulloides
	Leached sample		Leachate		Leached sample‐ Leachate	Leached sample		Leachate		Leached sample‐leach fraction	Leached Sample (extracted sediment)‐leached sample (nonextracted sediment)
	Lab code ETH‐	14C age (year) ± 1 σ	Lab code ETH‐	14C age (year) ± 1 σ	Age difference (year)	Lab code ETH‐	14C age (year) ± 1σ	Lab code ETH‐	14C age (year) ± 1 σ	Age difference (year)	Age difference (year)
120	90559.1.1	12,901 ± 86	90559.2.1	12,846 ± 135	55 ± 160	72997.2.1	13,228 ± 93	72997.1.1	12,328 ± 190	900 ± 211	327 ± 126
172	90557.1.1	15,262 ± 100	90557.2.1	13,377 ± 134	1,885 ± 167	73003.2.1	15,346 ± 115	73003.1.1	13,730 ± 202	1,616 ± 232	84 ± 152
210	90555.1.1	17,303 ± 109	90555.2.1	16,651 ± 167	652 ± 199	75018.1.1	17,292 ± 123	75018.2.1	15,468 ± 242	1,824 ± 271	−11 ± 164
240	90553.1.1	18,529 ± 119	90553.2.1	16,378 ± 162	2,151 ± 201	75024.1.1	18,735 ± 134	75024.2.1	16,214 ± 256	2,521 ± 288	206 ± 179
300	90552.1.1	22,171 ± 152	90552.2.1	21,509 ± 237	662 ± 281	75033.1.1	22,110 ± 172	75033.2.1	20,832 ± 342	1,278 ± 382	−61 ± 229

Note. 14C ages and associated 1σ confidence level (68.2% probability), and corresponding age discrepancies, shown in Figure 2. Age offsets that can be explained within the 1σ confidence level of the associated dates are indicated in bold.

The five untreated samples are younger than the paired leached samples and older than the leachate (Figure 3a). Age discrepancies among these three types of material measurements increase downcore.

Figure 3

Radiocarbon ages and related offsets of planktonic foraminifera. (a) Radiocarbon ages of and (b) . (c) 14C‐age discrepancies between the leached sample and the leachate of each species. (d) 14C‐age discrepancies between leached samples of both species calculated as ‐. Open diamonds and dots in (c) and (d) indicate age offsets that fall within the 1σ uncertainty envelope of the two 14C dates, respectively. Gray bars mark periods or maximum age offsets, coinciding with the Heinrich Stadials (HS) 2 and 1, the Younger Dryas (YD), and part of the Early and mid‐Holocene (E/M‐H).

Radiocarbon determinations generally reveal younger ages for the leachate in relation to the corresponding leached samples for both species (Figures 3a and 3b and Table 3). Leached samples display a systematic aging downcore with few reversals of minimal magnitude. By contrast, 14C ages of the leachate deviate from this trend, showing increasing variability downcore. While many of the age offsets between leached samples and paired leachates within the top 90‐cm fall into their associated 1σ uncertainty envelope, they show an apparent increase in magnitude downcore (up to 1,595–1,660 years for both species at 260 cm and up to 4,015 years for at the bottom of the core; Figure 3c and Table 3). Differences <5% between the F14C of leachates and corresponding leached samples indicate near‐complete removal of surface contaminants for all the samples (Tables S1 and S2). Interspecies age differences of the leached sample reveal age offsets of up to 1,030 years, and only three of them overlap within their associated 1σ uncertainty (Figure 3d and Table 3). ages are generally older than ones, a pattern that is reversed for two samples of the last glacial maximum, and within the top 20 cm of the core. The largest offsets coincide with the occurrence of three abrupt climate events: the Heinrich Stadial 1 (HS1), Younger Dryas (YD), and part of the Holocene (approximately 9–6 kyr). Limited material prevented some samples to be leached and were measured as untreated samples. Three of these samples (280 cm, 270 cm, and a replicate of the latter) strongly deviate toward younger ages.

Table 3

Radiocarbon Ages and Associated 1σ Confidence Level (68.2% Probability), and Corresponding Age Discrepancies

Depth (cm)	G. bulloides					G. ruber					G. bulloides‐G. ruber	G. bulloides‐G. bulloides
	Leached sample		Leachate		Leached sample‐leachate	Leached sample		Leachate		Leached sample‐leach fraction	Leached sample‐leached sample	Leached sample‐untreated sample
	Lab code ETH‐	14C age (year) ± 1σ	Lab code ETH‐	14C age (year) ± 1σ	Age difference (year)	Lab code ETH	14C age (year) ± 1 σ	Lab code ETH	14C age (year) ± 1 σ	Age difference (year)	Age difference (year)	Age difference (year)
0	82182.2.1a	788 ± 151
4	82183.2.1a	1,012 ± 153
10	82184.2.1	1,253 ± 71	82184.1.1	1,373 ± 77	120 ± 105	72980.2.1	1,463 ± 45	72980.1.1	1,216 ± 108	247 ± 117	−210 ± 84	−205 ± 131
	72979.1.1a	1,458 ± 110
14	82185.2.1a	1,451 ± 70
20	72981.2.1	1,820 ± 55	72981.1.1	2,078 ± 124	−258 ± 136	72982.2.1	1,884 ± 46	72982.1.1	1,930 ± 113	−46 ± 122	−64 ± 72
30	72983.2.1	2,301 ± 47	72983.1.1	2,229 ± 120	72 ± 129	72984.2.1	2,471 ± 75	72984.1.1	2,349 ± 123	122 ± 144	−170 ± 88
40	72985.2.1	3,087 ± 64	72985.1.1	2,927 ± 117	160 ± 133	72986.1.1a	2,628 ± 185
44	75040.1.1	3,619 ± 74	75040.2.1	3,823 ± 124	−204 ± 144
48	70397.1.1	3,762 ± 62	70397.2.1	3,848 ± 122	−86 ± 137	70399.1.1	3,389 ± 63	70399.2.1	3,137 ± 123	252 ± 138	373 ± 88
54	75041.1.1	5,295 ± 80	75041.2.1	5,343 ± 122	−48 ± 146
60	72987.2.1	7,470 ± 63	72987.1.1	6,556 ± 149	914 ± 162	72988.2.1	6,705 ± 60	72988.1.1	6,964 ± 207	−259 ± 215	765 ± 87	220 ± 90
	90560.1.1a	7,250 ± 64
70	72989.2.1	8,744 ± 69	72989.1.1	8,731 ± 156	13 ± 171	72990.2.1	8,482 ± 89	72990.1.1	8,261 ± 157	221 ± 180	262 ± 113
76	75042.1.1	9,957 ± 76	75042.2.1	9,338 ± 160	619 ± 177
82	72991.2.1	1,1056 ± 84	72991.1.1	10,351 ± 180	706 ± 199	72992.2.1	10,204 ± 75	72992.1.1	10130 ± 175	74 ± 190	852 ± 113
90	72993.2.1	1,1437 ± 86	72993.1.1	11,191 ± 178	246 ± 198	72994.2.1	10,806 ± 104	72994.1.1	10854 ± 174	−48 ± 203	631 ± 135
100	70400.1.1	1,2077 ± 107	70400.2.1	11,261 ± 193	816 ± 221	70402.1.1	11,900 ± 105	70402.2.1	11442 ± 201	458 ± 227	177 ± 150
110	72995.2.1	1,2385 ± 103	72995.1.1	12,413 ± 187	−28 ± 213	72996.1.1a	12,318 ± 210
120	72997.2.1	1,3228 ± 93	72997.1.1	12,328 ± 190	900 ± 211	72998.2.1	12,198 ± 91	72998.1.1	12688 ± 198	−490 ± 218	1030 ± 130
130	70403.1.1	1,3615 ± 109	70403.2.1	12,794 ± 204	821 ± 231	70405.1.1	13193 ± 109	70405.2.1	12905 ± 304	288 ± 323	422 ± 154	336 ± 140
	90558.1.1a	13,279 ± 88
140	72999.2.1	14,090 ± 104	72999.1.1	13,535 ± 199	555 ± 224	73000.2.1	13,252 ± 596	73000.1.1	11,980 ± 272	1,272 ± 655	838 ± 605
146	75043.1.1	14,290 ± 101	75043.2.1	13,079 ± 225	1,211 ± 247
152	73001.2.1	14,884 ± 105	73001.1.1	14,160 ± 216	724 ± 240
160	73002.2.1	14,924 ± 108	73002.1.1	14,334 ± 210	590 ± 236
172	73003.2.1	15,346 ± 115	73003.1.1	13,730 ± 202	1,616 ± 232	73004.1.1a	14,572 ± 328					191 ± 154
	90556.1.1a	15,155 ± 102
180	73005.2.1	15,977 ± 138	73005.1.1	14,560 ± 207	1,417 ± 249	73006.2.1	15,261 ± 230	73006.1.1	15,071 ± 339	190 ± 410	716 ± 268
190	73007.2.1	15,916 ± 206	73007.1.1	16,179 ± 247	−263 ± 322	73008.1.1a	15,513 ± 260
196	75044.1.1	16,636 ± 120	75044.2.1	15,351 ± 270	1,285 ± 295
200	75016.1.1	17,066 ± 120	75016.2.1	16,105 ± 238	961 ± 266	75017.1.1	16,786 ± 134	75017.2.1	16,599 ± 267	187 ± 299	280 ± 180
210	75018.1.1	17,292 ± 123	75018.2.1	15,468 ± 242	1,824 ± 271	75019.1.1a	17,064 ± 161
214	75045.1.1	17,242 ± 122	75045.2.1	16,159 ± 279	1,083 ± 304
220	75020.1.1	17,427 ± 142	75020.2.1	16,248 ± 270	1,179 ± 305	75021.1.1	17,511 ± 137	75021.2.1	16,493 ± 260	10,18 ± 294	−84 ± 197
230	75022.1.1	18,634 ± 176	75022.2.1	17,495 ± 259	1,139 ± 313	75023.1.1a	18,146 ± 170
234	75046.1.1	18,305 ± 130	75046.2.1	17,318 ± 278	987 ± 307
240	75024.1.1	18,735 ± 134	75024.2.1	16,214 ± 256	2,521 ± 289	75025.1.1	18,301 ± 177	75025.2.1	17,803 ± 280	498 ± 331	435 ± 222	1,154 ± 182
	90554.1.1a	17,581 ± 123
250	75026.1.1	18,726 ± 150	75026.2.1	18,314 ± 288	412 ± 325	75027.1.1	19,231 ± 141	75027.2.1	18,481 ± 289	750 ± 322	−506 ± 206
260	70406.1.1	19,979 ± 181	70406.2.1	18,387 ± 301	1,592 ± 351	70408.1.1	19,831 ± 180	70408.2.1	18,166 ± 307	1,665 ± 356	148 ± 255
264	75047.1.1	19,776 ± 143	75047.2.1	17,717 ± 276	2,059 ± 311
270	75028.1.1	20,361 ± 152	75028.2.1	17,665 ± 287	2,696 ± 325	75029.1.1a	18,348 ± 172
270 r						82186.2.1a	18,310 ± 320
280	75030.1.1	20,684 ± 155	75030.2.1	17,045 ± 257	3,639 ± 300	75031.1.1a	15,814 ± 166
284	75048.1.1	20,991 ± 159	75048.2.1	18,691 ± 319	2,300 ± 356
290	75032.1.1	21,347 ± 161	75032.2.1	20,247 ± 338	1,100 ± 374
300	75033.1.1	22,110 ± 172	75033.2.1	20,832 ± 342	1,278 ± 383
310	75034.1.1	22,573 ± 178	75034.2.1	20,153 ± 339	2,420 ± 383	75035.1.1a	21,912 ± 278
314	75049.1.1	23,133 ± 189	75049.2.1	21,020 ± 484	2,113 ± 519
320	75036.1.1	22,984 ± 185	75036.2.1	19,376 ± 305	3,608 ± 357	75037.1.1a	22,763 ± 286					1,419 ± 242
	90551.1.1a	21,565 ± 157
329	75038.1.1	24,126 ± 203	75038.2.1	20,116 ± 317	4,010 ± 376	75039.1.1a	23,166 ± 329

Note. Numbers in bold indicate age offsets that can be explained within the 1σ confidence level of the associated dates.

Untreated samples.

SEM Imagery

Overall, tests of both species exhibit good preservation with minor overgrowth (i.e., secondary calcite) on the original base of the spines (Figure S1). Such features are consistently observed in all samples, irrespective of their depth interval. Both and show variable amounts of coccoliths glued on the outer wall. Nevertheless, this feature does not affect all the samples nor all the specimens, and there is no relationship between the presence nor the amount of coccoliths and sample depth.

Isotopic Composition of and

Carbon isotopes of range between −0.4‰ and −1.8‰ and show higher values during the cold intervals associated to the HS2, HS1, and YD, and part of the Holocene (Figure 4b). The δ13C data of vary between 1.4‰ and −0.4‰ and show relatively constant values for the first half of the record (340–170 cm) and an increasing trend toward more positive values thorough the Holocene. Oxygen isotopes of range between 0.1‰ and 3.0‰ and record short‐term isotopic changes associated with HS2, HS1, and YD (Figure 4c). The δ18O data of range between −0.1‰ and 2.2‰. This record shows a smoother profile than that of and lacks samples for part of HS1. Both isotopic curves are out of phase by at least 10 cm for most of the last deglaciation (70–140 cm). The oxygen isotopic difference between both species (Δδ18Ob‐r) ranges from −0.3‰ to 1.7‰ and shows highest values during the HS2, HS1, and YD (Figure 3c).

Figure 4

Oxygen isotopic records and abundances. (a) Sedimentation rate of core SHAK06–5K based on 14C ages of leached samples of . (b) Carbon and (c) oxygen isotope record of and . (d) Oxygen isotopic difference between and . (e) Species absolute and relative abundances. Gray bars mark periods or maximum age offsets shown in Figure 3, coinciding with the Heinrich Stadials (HS) 2 and 1, the Younger Dryas (YD), and part of the Early and mid‐Holocene (E/M‐H).

Variation in Species Abundances

Average absolute and relative abundances of are 6 specimens per gram and 24%, respectively, and show large increases during the cold intervals HS2, HS1, and the YD (up to 25 specimens per gram and 72%; Figure 4e). shows average absolute and relative abundances of 1 specimen per gram and 4%. This species is almost absent during HS2, HS1, and YD, and increases to up to 8 specimens per gram and 13% during the late Holocene (top 30 cm).

Discussion

Contamination Through Secondary Radiocarbon Addition: The Need for a Leaching Step

Age discrepancies between paired leached samples and leachates highlight the secondary addition of younger carbon and subsequent contamination on the outer shell (Figures 3a and 3b and Table 3), as observed by previous authors when applying similar leaching steps (Bard et al., 2015). Such contamination was not introduced by using organic solvents for lipid extraction, as the leachates were always younger than corresponding leached samples, regardless of whether foraminifera come from solvent‐extracted or nonextracted sediments (Figure 2 and Table 2). The magnitude of such age discrepancy does not always agree for both methods, but this can be explained by the varying and small amounts of C measured from the leachate (Table S1). Moreover, comparison of 14C ages of leached samples from both types of sediments shows negligible differences (Figure 2). These results are in line with previous findings of Ohkouchi et al. (2005), who concluded that tests from solvent‐extracted sediments can be reliably used for 14C determinations. Additional influence of other sample preparation steps cannot be fully discarded. For instance, soaking of foraminifera during wet sieving can activate their reactive surface and enable adhesion of ambient carbon. However, we minimized the potential influence of this process by drying the samples in the oven right after sieving. Another possibility to consider is the influence of early diagenesis. Minor signs of secondary calcite precipitation are apparent by SEM imagery in all the tests (Figure S1), regardless of sample depth and species. Diagenetic alteration of shells through ∑CO2 exchange with pore waters with a younger 14C signature might explain the negligible impact of secondary calcite precipitation on samples from the top 60 cm and the more variable and larger effect observed downcore (Figure 3c). These results highlight the need of a leaching step to remove surface contaminants, especially for older samples, for which age biases can be greater than 1,000 years (Figure 1a and Table 3).

Regarding the untreated samples of , two large deviations towars younger‐than‐expected ages are also evident at the bottom of the core (Figure 3b). Within single depth horizons of a core retrieved from the Portuguese margin, Löwemark and Grootes (2004) found large intraspecies age discrepancies (up to 2,590 years) when comparing sediments affected and unaffected by trace fossils indicating bioturbating organisms (e.g., Zoophycos). Because ichnofossils occur throughout the sediments of IODP Site U1385 (Rodríguez‐Tovar et al., 2015; Rodríguez‐Tovar & Dorador, 2014), they most certainly also affect the sediments of core SHAK06–5K. Their influence would imply that discrete samples from the same sediment horizon would consist of a mixture in different proportions of foraminifera tests from both bioturbated and nonbioturbated material. The excellent agreement between the two replicates of samples from depth horizon 270 cm excludes bioturbation as the reason for such age deviations. Addition of younger secondary calcite might also explain these age deviations, although lack of material prevented further assessment.

Interspecies Radiocarbon Age Differences

Assuming removal of the majority of external contamination by the leaching step (Table S1), secondary radiocarbon addition does not account for the 14C age differences between the leached samples of the two species (Figure 3d), and mechanism(s) differentially affecting foraminifera species must be sought to explain the systematic younger‐than‐ 14C ages for . Ideally, such mechanism(s) should also explain changes in the magnitude of the observed age offsets with abrupt climate events. In the following, we discuss four possible mechanisms.

Contrasting Calcifying Habitats

Differences in calcifying depth and season of the two species might have also played a role in 14C age discrepancies. Mollenhauer (1999) demonstrated that interspecies differences of 540 years are possible in upwelling settings, where deep, less‐ventilated, older waters are upwelled to the surface. Currently in the study area, the average living depths of and are 58 ± 6 and 102 ± 21 m, respectively (Rebotim et al., 2017). While is characteristic of winter hydrographic conditions, is more abundant during the upwelling season (i.e., summer; Salgueiro et al., 2008). Figure 5 shows the natural radiocarbon content (Δ14C) depth profile from a station corresponding to the water column overlying the depositional area of the study site, extracted from the Global Ocean Data Analysis Project (Key et al., 2004). Corresponding natural Δ14C values for average living depths of and are −59‰ and ~−65‰, respectively, equivalent to an age discrepancy of ~50 years, which is insufficient to explain age offsets between species. As seasonality also impacts on the optimal conditions for and proliferation, we calculated the winter and summer natural Δ14C for the upper 500 m of the water column. We applied the linear relationship between natural Δ14C and dissolved silicate for North Atlantic latitudes (equation ((1))) proposed by Broecker et al. (1995), using summer and winter dissolved silicate estimates (García et al., 2014) averaged at 100‐ and 60‐m water depth, respectively, from the 2013 World Ocean Atlas.

Figure 5

Modern estimated natural Δ14C data at station ID15364 from Global Ocean Data Analysis Project (Key et al., 2004) corresponding to the overlying water column of SHAK06–5K core location. Data were plotted with Ocean Data View (ODV) (Schlitzer, 2014).

Yet the estimated seasonal difference in Δ14C is minimal (−3.2‰) and negligible in relation to the large uncertainty derived from the silicate method (±15‰; Rubin & Key, 2002).

However, it is still possible that the associated radiocarbon reservoirs (or at least one of them) varied in the past during HS1, YD, and part of the Holocene related to the large hydrographic changes that occurred during abrupt climate events in the study area (Voelker & de Abreu, 2011). This argument was put forward by Löwemark and Grootes (2004) to explain the large age discrepancy they found between and during the YD on the Portuguese margin. In this regard, the incursion of intermediate, extremely 14C‐depleted waters characterized by high nutrient content has been suggested to reach latitudes as far as 60°N in the Atlantic during the abrupt cold intervals HS1 and YD (Pahnke et al., 2008; Rickaby & Elderfield, 2005; Thornalley et al., 2011). The authors pointed to Antarctic Intermediate Water (AAIW), which would have extended northward as a consequence of Atlantic Meridional Overturning Circulation (AMOC) weakening or collapse. Indeed, such drastic reductions of AMOC during HS1 and YD prevented the formation of new North Atlantic Deep Water (McManus et al., 2004), which would have then been replaced by AAIW. However, the hypothesis of markedly different radiocarbon reservoirs affecting each of the species is not fully supported by other data. δ13C values give no clear indication of upwelling of nutrient‐rich waters occurring during HS2 or YD, and lack of during HS1 prevents further interpretation (Figure 4b). More positive δ13C values of rather suggest that upwelling had decreased at those times. Although less negative δ13C values could also be the result of upwelling and subsequent nutrient consumption by primary producers, resulting in a 13C‐enrichment of surrounding waters, this scenario disagrees with previous studies. Estimates of export production by (Salgueiro et al., 2010) and of primary productivity and upwelling occurrence by (Incarbona et al., 2010) are best explained with the arrival of freshwater during HS1 and YD resulting in water column stratification, decreased upwelling, and a large drop in productivity. Moreover, assuming that the general ecological preferences of each species remained constant during the last deglaciation, upwelling of AAIW would preferentially affect . Yet radiocarbon ages corresponding to the δ18O excursions of associated with HS2, HS1, and YD are in very good agreement with the established age ranges for these abrupt climate events (Figure S2), which underpins the notion that 14C ages are not, at least severely, biased in relation to their depositional ages. Additionally, we believe this mechanism fails to explain temporal discrepancies during the Holocene. Even though a relative increase of AAIW influence in higher northern latitudes can be recognized from neodymium isotope ratios (Pahnke et al., 2008), there is no evidence of a large reduction of AMOC at that time, which is believed to have been relatively strong during the Holocene (Gherardi et al., 2005; Thornalley et al., 2011). Although we cannot completely refute that the influence of water masses with distinct radiocarbon content (Δ14C) contributed to the observed age offsets during HS1 and YD, an additional mechanism is needed to explain the smoothed δ18O curve of in relation to that of (Figure 4c) a feature typical of bioturbated sediment (Bard et al., 1987a).

The Barker Effect

The Barker effect (first proposed by Andree et al., 1984, Broecker et al., 1984, Peng & Broecker, 1984, and Broecker et al., 2006, and coined by Broecker & Clark, 2011), refers to the differential effect of partial dissolution and subsequent fragmentation of shells along with bioturbation on the 14C ages of different species planktonic foraminifera (Barker et al., 2007; Broecker & Clark, 2011). Given that different species may dissolve at different rates, fragile and dissolution‐prone species (i.e., ) will fragment in the sediment mixed layer more easily than more robust, dissolution‐resistant species (i.e., ; Berger, 1968, 1970). This translates into shorter residence times in the sediment for relative to . Consequently, the pool of nonfragmented shells of at a given horizon will be biased toward younger specimens, because specimens that reside in the bioturbated layer for longer periods are more likely to be fragmented. As only well‐preserved whole tests were picked for 14C analyses, monospecific samples of will be, on average, younger than .

This effect was invoked to account for age discrepancies among planktonic foraminifera species of up to several thousand years especially in cores characterized by low sediment accumulation rates (<3 cm/kyr; Barker et al., 2007; Broecker et al., 2006; Broecker & Clark, 2011; Peng & Broecker, 1984). The latter is an important factor to be taken into account since the lower the sedimentation rate, the longer the exposure time to the effect of bioturbation. High sedimentation rates of core SHAK06–5K only decrease to a minimum of 6 cm/kyr for the interval from 80 to 50 cm (Figure 4a). However, the observed apparent increase in the interspecific 14C age offset is not exclusive to this horizon and visual inspection of nannofossils confirmed their excellent preservation thorough the Holocene.

Yet highly productive settings may have favored acidification of underlying waters and pore waters through CO2 release by respiration. Despite being part of a major upwelling system, total organic content in core SHAK06–5K and broader region (Baas et al., 1997; Magill et al., 2018) ranges from only 0.2% to 0.7% for the whole studied period, suggesting that substantial dissolution by organic carbon oxidation is unlikely. Similarly, changes in the depth of the calcite lysocline are also assumed to have had a negligible effect, because the water depth of the core (2,578 m) is located well above that level. Influence of more corrosive water masses could have promoted increased dissolution of . However, incursion of southern sourced water mass was mostly limited to glacial periods (Skinner & Shackleton, 2004), characterized by relatively high sedimentation rates. Therefore, we consider it is unlikely that the Barker effect had a major influence in the observed 14C age discrepancies between foraminifera species.

Lateral and Along‐Slope Transport

Introduction of reworked specimens by advection and along‐slope sedimentary processes could also contribute to radiocarbon age discrepancies, a mechanism proposed in cores from the Eastern Equatorial Pacific, the Mid‐Atlantic Ridge, and the South China Sea (Broecker et al., 2006). Addition of reworked calcareous nannofossils by lateral transport has been observed in the study area (Incarbona et al., 2010) and in core SHAK06–5K (Magill et al., 2018), especially during HS1. Simulated bottom velocities in the study area might locally exceed 10 cm/s and be able to transport dense, 250‐ to 300‐μm sized grains of foraminifera when locally reaching >40 cm/s (Hernández‐Molina et al., 2011). To explain the observed older‐than‐ ages for by any of these mechanisms, transport and deposition of large numbers of reworked (old) would be necessary, along with preferential fragmentation of during transport. This might be a feasible scenario, albeit it would imply that samples of are the ones affected by a temporal bias between biosynthesis and deposition. We thus discard this hypothesis based on (i) the good agreement of δ18O excursions during short‐term climate changes and their associated established age ranges (Figure S2) and (ii) the smoothed δ18O curve of that hardly resolves the major abrupt climate events occurred the last deglaciation (Figure 4c). Such results suggest that , rather than , accounts for the age offsets between the two species.

Differential Bioturbation Coupled With Changes in Species Abundances

The joint effect of downward mixing of foraminifera due to bioturbation and changes in their abundance might promote 14C offsets between species (Andree et al., 1984; Bard et al., 1987a; Broecker et al., 1984, 1999; Peng & Broecker, 1984). Foraminifera will always be mixed from a horizon of high abundance to low abundance. Given an increase (decrease) in the abundance of a certain species in a sediment horizon, bioturbation is expected to downmix (upmix) some of these young (old) foraminifera. As a result, the horizon underneath (above it) will be enriched in younger (older) specimens, leading to corresponding deviations in their expected 14C ages. The clear aging trend with depth gives no indication of homogenization by bioturbation >10 cm (Figures 2a and 2b). However, the δ18O record of lags that of by 10 cm during the HS1, last deglaciation, and YD (Figure 4d). This shift is more apparent when comparing samples at lower resolution (every 10 cm only; Figure S3) and suggests a mixed layer depth equivalent to ≤10 cm. Similar out‐of‐phase relationships between species‐specific isotopic records have previously been explained through this mechanism (Bard et al., 1987a, 1987b; Hutson, 1980). Löwemark and Grootes (2004) also invoked it to account for differences of 75–350 years between and in a nearby core from the SW Portuguese margin. According to these authors, and given the large changes in the abundance of relative to those of (Figure 4e), a larger impact on the 14C ages of the former species would be expected. This hypothesis is difficult to reconcile with the smoothed δ18O curve of . We would expect to be the species more affected by differential bioturbation than . Indeed, and with the exception of the sample at 60 cm, each large increase in Δδ18O is followed by a rise in absolute abundance (Figures 3c and 3d) that, despite their moderate magnitude, also follow periods of extremely low abundance or near absence. Our data are a faithful reproduction of previous mathematical simulations of Trauth (2013) and Bard et al. (1987a), who demonstrated the effects of bioturbation coupled with abundance changes in the oxygen isotopic record of a warm species (i.e., ) during deglaciation (see Figure 4 in Bard et al., 1987a). Our results do not agree well with their model for the cold species (i.e., ) because they are permanently present, and authoctonous specimens can make up for the radiocarbon addition from foraminifera belonging to adjacent sediment horizons.

Conclusions

Radiocarbon dates of paired monospecific samples of and (white) were determined in marine sediments retrieved from the SW Iberian Margin. 14C age differences of several thousands of years between paired leachates and leached samples indicate addition of younger radiocarbon in both species. This process is attributed to precipitation of younger secondary calcite by ∑CO2 exchange with 14C‐rich pore waters and/or ambient carbon adhesion during sample sieving, thus having a more variable and greater impact downcore. Leaching of the outer shell has proven to be a powerful diagnostic for external contamination, and more importantly, a tool to obtain more reliable radiocarbon dates, especially when dealing with older samples (>10 kyr). Our findings underscore the need to properly leach foraminiferal samples prior to radiocarbon dating.

Interspecies age discrepancies of the leached samples ranged between 60 and 1,030 years. yielded younger ages than paired in the same sample throughout most of the record. Larger age discrepancies were found during HS1, YD, and part of the Holocene and were attributed to the effects of bioturbation coupled with species abundance changes. This mechanism has a greater impact if the species in question has periods of absence (i.e., ) rather than greater abundance changes (i.e., ) because the population of rarer species is more affected by the addition of asynchronous foraminifera compared to a more abundant species. This process alone appears to provide a satisfactory explanation for the observed age offsets, although additional influences such as past variations in the 14C reservoirs of the respective calcifying habitats cannot be fully ruled out.

After a careful evaluation of potential 14C age anomalies in these two species, we conclude that unlike , can be reliably used to develop foraminifera‐based 14C age chronostratigraphies and to assess ocean ventilation ages in the study area.

Author Contribution

B. A. and T. I. E. planned this investigation. N. H. and L. W. assisted with radiocarbon analyses. N. L. assisted with SEM imagery. B. A. prepared the samples, analyzed the results, and wrote the manuscript with contributions by all coauthors.

Supporting Information S1

Click here for additional data file.

Table S1

Click here for additional data file.