dating investigation of the upper Jaramillo transition from a volcanic section at Tenerife (Canary Islands)

Earth and Planetary Science Letters 406 (2014) 59–71

Contents lists available at ScienceDirect

Earth and Planetary Science Letters www.elsevier.com/locate/epsl

A combined paleomagnetic/dating investigation of the upper Jaramillo transition from a volcanic section at Tenerife (Canary Islands) C. Kissel a,∗ , H. Guillou a , C. Laj a , J.C. Carracedo b , F. Perez-Torrado b , C. Wandres a , A. Rodriguez-Gonzalez b,c , S. Nomade a a b c

Laboratoire des Sciences du Climat et de l’Environnement/IPSL, CEA-CNRS-UVSQ, Avenue de la Terrasse, 91198 Gif-sur-Yvette Cédex, France Departamento de Física-Geología, Universidad de Las Palmas de Gran Canaria, 35017 Las Palmas de Gran Canaria, Spain Institute of Earth Sciences Jaume Almera, ICTJA-CSIC, Sole i Sabaris s/n, 08028 Barcelona, Spain

a r t i c l e

i n f o

Article history: Received 8 February 2014 Received in revised form 29 August 2014 Accepted 3 September 2014 Available online xxxx Editor: J. Lynch-Stieglitz Keywords: geomagnetic reversals unspiked K–Ar 40 Ar/39 Ar upper Jaramillo volcanic sequence

a b s t r a c t A coupled paleomagnetic/dating investigation has been conducted on a sequence of 25 successive lava flows, emplaced during the upper transition of the Jaramillo subchron in Tenerife, Canary Islands. This sequence is located along the western wall of the Güímar collapse scar, in the south central part of the island. Nine flows distributed throughout this sequence were dated using unspiked K/Ar and 40 Ar/39 Ar methods. They bracket the section between 1009 ± 22 ka and 971 ± 21 ka (2σ ). A first group of 8 flows at the bottom of the sequence is characterized by normal polarity with paleointensity values of the order of present-day field intensity in the Canary Islands. The virtual geomagnetic poles (VGP) of these 8 flows describe a short loop at high latitudes. Seven overlying flows are transitional in directions and dated between 991 ± 14 ka and 1002 ± 11 ka consistently with published ages of the upper Jaramillo reversal. This second group of flows is characterized by low paleointensity values (around 8–12 μT) that are less than 30% of the present dipole value in Tenerife. The VGPs of the first two transitional flows lie over northeastern Pacific whereas the five following transitional flows have all negative inclinations and their VGPs lie initially over East Antarctica, then describe a northward loop almost reaching New Zealand. The final group of ten flows yield intensities varying between 20 and 35 μT and VGPs close to the southern pole with two of them describing a small amplitude second loop to southeastern Pacific. Assuming a constant extrusion rate as a very first approximation, the distribution of the obtained ages suggests a duration of 7.6 ± 5.6 ka for the transitional interval. The obtained transitional positions of VGPs are consistent with the path reported for the same reversal from North Atlantic sediments but are different from the only other volcanic record from Tahiti. The intensity low characterizing the transitional interval remains the best tie point, centered at 996 ± 7 ka (2σ ) relative to 28.02 Ma FC sanidine. © 2014 Elsevier B.V. All rights reserved.

1. Introduction The Jaramillo normal polarity subchron (JNS) has been first evidenced within the reverse Matuyama Chron through the study of rhyolitic domes in New Mexico (Doell and Dalrymple, 1966). Since then, the JNS has been identified in widely distributed eolian, marine, and lacustrine sedimentary sequences. The two reversals bracketing the JNS are frequently used as age markers for Pleistocene records, yet their ages have been progressively adjusted with time using improved 40 Ar/39 Ar ages or astronomical tuning (Berggren et al., 1985; Shackleton et al., 1990; Tauxe et al., 1992; Spell and McDougall, 1992; Spell and Harrison, 1993; Izett and Obradovich, 1994; Singer et al., 1999; Channell et al.,

*

Corresponding author. E-mail address: [email protected] (C. Kissel).

http://dx.doi.org/10.1016/j.epsl.2014.09.003 0012-821X/© 2014 Elsevier B.V. All rights reserved.

2009). Relative changes in paleointensity obtained from sedimentary records across the entire JNS have also been investigated with reference to long-term dynamics of the earth’s magnetic field intensity changes (Valet and Meynadier, 1993; Leonhardt et al., 1999; Laj et al., 1996; Verosub et al., 1996; Dinares-Turell et al., 2002; Channell et al., 2009). The details of the vectorial changes of the earth magnetic field during the reversals bounding the JNS and in particular the upper normal to reverse (N–R) Jaramillo/Matuyama reversal (UJR) have, on the other hand, been seldom documented, despite their relevance to decipher the behavior of the geomagnetic field during polarity transitions. A description of the UJR has been obtained from four sedimentary sequences, all of them retrieved from ODP/IODP drill cores at latitudes ranging between 53◦ N and 60◦ N in North Atlantic (Channell and Lehman, 1997; Mazaud et al., 2009). They all show the same rather simple Virtual Geomagnetic Pole (VGP)

60

C. Kissel et al. / Earth and Planetary Science Letters 406 (2014) 59–71

path, described by a first loop over the Americas down to about 30◦ –45◦ S, followed by a full N–R path over the Indian ocean sector, from along the eastern African coastline (about 40◦ E longitude) to over India (about 75◦ E longitude) depending on the core. The associated relative paleointensity profiles all show a very low value at the time of the directional changes. Lava flows are an important archive in paleomagnetism because they provide the only way to retrieve absolute intensity values of the geomagnetic field and they can be dated using both the K–Ar and 40 Ar/39 Ar techniques. However, it is a challenge to locate volcanic sequences erupted during the short geological periods corresponding to geomagnetic reversals or excursions, given the sporadic nature of volcanic eruptions. Previously to this study, only one reliable volcanic record of the UJR has been obtained from a sequence of 23 flows at Tahiti, at about 18◦ S (Chauvin et al., 1990). Data from Tahiti are mainly directional data, only one transitional flow allowed a single paleointensity determination. On the basis of four ages from normal and transitional flows of the Punaruu valley (Singer et al., 1999), the best age estimate calculated for the termination of the JNS is 1001 ± 10 ka (Singer et al., 2004). This age is reported at 2σ uncertainty and relative to 1.194 Ma Alder Creek Rhyolite sanidine standard (ACs-2; Nomade et al., 2005) corresponding to 28.02 Ma FC (Renne et al., 1998). Here, we report on new dating and paleomagnetic results (both directional and intensity data) from a volcanic sequence located in the NE Rift Zone of Tenerife (Canary Islands) comprising 25 successive flows, 7 of which describe the transitional field during the upper Jaramillo reversal. 2. Geological setting and sampling The NE Rift Zone of Tenerife (NERZT) evolved very rapidly with high eruptive rates during the last cycle of eruptive growth between 1 Ma and 830 ka (Carracedo et al., 2011). This cycle was followed by major landslide collapses occurring at ca. 830 ka (Micheque and Güímar landslides on the northern and southern flank of the rift respectively) and between 690 and 566 ka (La Orotava landslide adjoining the Micheque landslide) (Carracedo et al., 2011). In the valley of Güímar, a sequence of lava flows were emplaced on a slightly southward tilting slope and are now exposed on the southern wall of the Güímar landslide (Guillou et al., 2013). The sampled section is located along this “Pared de Güímar” (28◦ 18 N; 16◦ 28 W) (Fig. 1). Preliminary K/Ar dating indicated that this wall, formed by a 500 m thick sequence of basaltic flows, erupted between 1008 ± 22 and 963 ± 21 ka (Carracedo et al., 2011). This high eruptive rate during the activity of the NERZT therefore allows to retrieve a detailed record of changes in the geomagnetic field. In fact, the K/Ar ages indicate that this section covers the upper part of the normal polarity Jaramillo subchron, including the entire upper Jaramillo transition to the reversed polarity of the Matuyama Chron. Preliminary fluxgate measurements in the field confirmed the presence of a reversal in this sequence (Guillou et al., 2013). Sampling of 25 successive flows was conducted over a two years period along the track of the valley of the Güímar, starting with flow TT16 in the normal polarity zone of the Jaramillo Chron and ending with flow TT55 in the reverse polarity zone of the Matuyama Chron. All the flows are accessible along the track except for the youngest flow, TT55, which is located just above it (Fig. 1). Two flows have double numbers (TT33-34 and TT31-32) because they were sampled twice along the path. Samples were collected using a gasoline powered drill with a 25 millimeter diamond barrel. Four to 14 cores were sampled in each flow and spread out as much as possible over the outcrop, both laterally and across the flows depending on the thickness and the brittle nature of the flows, for a total of 173 cores. A Schoenst-

edt magnetic locator proved very useful in selecting precise drilling location devoid of very local magnetic disturbances such as lightning strikes. 80% of the 173 cores have both magnetic and sun orientation and the difference between the two ranges between 0 and 12◦ with an average value around 7◦ . The bottom chip (thin slice cut at the bottom of the core to make it flat before cutting it into specimens) was used for rock magnetic experiments. Then, between 3 and 8 specimens half the standard size of paleomagnetic samples (11 mm thick) were obtained from each core and the deepest ones were used for stepwise demagnetization and for paleointensity experiments because they did not suffer from any surface weathering. For K–Ar and 40 Ar/39 Ar dating, large diameter cores (5 cm in diameter and up to 20–25 cm long) were collected in order to attain the freshest part of the most massive portion of the rock. 3. Dating methods and results K/Ar dating was performed on nine samples. The isotopic composition and abundance of Ar were determined using an unspiked technique described in Charbit et al. (1998). The analytical procedure is described in the supplementary material. For each sample, three independent determinations of concentrations in K were carried out by atomic absorption (flame photometry) at the Centre de Recherche Pétrographique et Géochimique (CRPG, Nancy, France). These repeated measurements have relative precisions of 1% and were combined to yield a mean value for each sample. Age determinations of each sample were made using this mean K concentration value and the weighted mean of the two independent measurements of radiogenic argon (40 Ar∗ ). Uncertainties for the Ar data are of an analytical nature only, consisting of propagated and quadratically averaged experimental uncertainties arising from the 40 Ar (total) and 40 Ar∗ determinations. Uncertainty on each age is given at 2σ and the details of the method are given in the supplementary material. K/Ar ages (Table 1) range from 1009 ± 22 ka to 971 ± 21 ka, consistently with the local stratigraphy except for TT26 (see below). This time interval is consistent with the upper limit of the JNS. On the basis of both the unspiked K–Ar ages and first paleomagnetic results, aliquotes from the same groundmass separate than for K–Ar experiments were selected from five of the flows to be analyzed at the 40 Ar/39 Ar facility of the Laboratoire des Sciences du Climat et de l’Environnement (LSCE). The analytical procedure is also described in the supplementary material. Two step-heating experiments were conducted on two splits for each of the 5 samples (the details are reported in the supplementary table). Plateau ages, isochron regressions and probability of fit estimates were calculated using ArArCALC (Koppers, 2002). Uncertainty of individual plateau age as well as inverse isochron is given at 2σ (full external error). In the following all ages are given with uncertainties at 2σ and relative to 28.02 FC sanidine. Two step-heating experiments yielded concordant spectra with 100% of the gas defining the age plateaus. The eight other plateau ages comprise between 72% and 94% of the gas released (Table 2). The 40 Ar∗ contents range from 10% to 60%, with typical values of 30% to 40% for the plateau steps. The 40 Ar/36 Ar intercept values defined for the associated isochrons are atmospheric and the total fusion ages are similar within errors to plateau or isochron ages, with the exception of the second experiment on sample TT-22 (Table 2 and Fig. 2). This indicates that for most samples the effect of argon loss or excess argon is almost negligible. This is also a powerful check of the assumptions required to validate the K–Ar ages as reliable crystallization ages, because the isochron approach makes no assumption regarding the trapped component and combines estimates of analytical precision and internal disturbance of the sample (scatter around the isochron). Therefore the isochron

C. Kissel et al. / Earth and Planetary Science Letters 406 (2014) 59–71

61

Fig. 1. Sampling along the “Pared de Güímar” (see the two inserts for the location of the section in the Tenerife island). The dotted line on the topographic map enhances the path along which the sampling was carried. The red dots are for the flows for which pooled K/Ar and 40 Ar/39 Ar dating are available. The reported numbers are for the preliminary K/Ar ages (Carracedo et al., 2011) which originally guided the paleomagnetic sampling. The latter has been made in two successive field trips. Given the seaward slight tilt of the flows emplaced on a slope, the flows are in stratigraphic order from the bottom of the section on the left hand side (TT16) up to TT31-32 and the TT55 above the trail. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

ages (Table 2) are preferred over the weighted mean plateau ages. They yield ages spanning the interval between 1002 ± 11 and 979 ± 16 ka. For one of the flows (TT26) the isochron age as well as the total fusion age are older by 50 ka than the unspiked K–Ar age of the same groundmass material. This might reflect incomplete melting of TT-26 during the K–Ar experiments and/or insufficient gas clean-up. 40 Ar/39 Ar ages are calculated from relative abundances of each isotope measured individually, and therefore a complete melting and degassing of the sample is not necessary. Step-heating experiments involved much smaller quantities of extracted gas. Therefore a better gas cleanup is achieved. For these two reasons the 40 Ar/39 Ar age at 1002 ± 11 ka is preferred. For the other four samples, the 40 Ar/39 Ar isochron ages and the unspiked K–Ar ages are equivalent at the 95% confidence level. Because the K–Ar and 40 Ar/39 Ar ages are based on inter-calibrations of a common primary standard mineral, GA-1550 biotite (Spell and

McDougall, 2003), they are directly comparable to one another (see supplementary material for details). These data are then pooled to produce a best age estimate for each flow. These four pooled ages are reported in Table 3 together with the unspiked K/Ar ages of the other flows and the 40 Ar/39 Ar isochron age of TT26. 4. Rock magnetic, paleomagnetic and paleointensity results 4.1. Rock magnetic properties Rock magnetic properties were studied using high field thermomagnetic curves, k–T curves, hysteresis cycles and FORC diagrams. k–T curves were performed using a CS-2 furnace from AGICO and thermomagnetic curves were performed using a horizontal Curie balance (Petersen instrument) up to 620 ◦ C in presence of Ar gas flow to minimize oxidation during the heating process. These

62

C. Kissel et al. / Earth and Planetary Science Letters 406 (2014) 59–71

Table 1 Unspiked K–Ar ages of the lavas from Pared de Güímar (Tenerife). Sample ID

K (±1σ ) (%)

Lab n◦

M (g)

40

Ar∗ (%)

40

40

TT-55∗

1.428 ± 0.014 “–”

7723 7741

1.25272 2.59277

23.888 31.536

2.439 ± 0.014 2.390 ± 0.012

2.411 ± 0.009

974 ± 21

971 ± 21

2.075 ± 0.021 “–”

7756 7860

1.62239 1.71420

24.666 26.984

3.523 ± 0.019 3.606 ± 0.019

3.566 ± 0.013

991 ± 21

988 ± 21

TT-27

1.868 ± 0.019 “–”

7866 8118

1.23920 1.50972

18.802 17.782

3.200 ± 0.016 3.176 ± 0.019

3.190 ± 0.012

984 ± 21

981 ± 21

TT-26

1.602 ± 0.016 “–”

7854 8009

1.33014 1.73215

14.134 22.510

2.639 ± 0.018 2.640 ± 0.014

2.645 ± 0.009

950 ± 22

947 ± 22

1.635 ± 0.016 “–”

7996 8093

2.32822 2.00961

14.718 33.641

2.803 ± 0.015 2.762 ± 0.014

2.781 ± 0.010

981 ± 21

978 ± 21

1.785 ± 0.018 “–”

7830 7862

1.52310 1.48870

26.925 12.597

3.078 ± 0.017 3.072 ± 0.019

3.075 ± 0.013

993 ± 21

990 ± 21

TT-48

1.395 ± 0.014 “–“

7940 7997

1.19588 1.55087

13.732 23.420

2.399 ± 0.010 2.480 ± 0.014

2.441 ± 0.010

1008 ± 22

1005 ± 22

TT-47

1.295 ± 0.013 “–”

7878 8094

1.14853 1.00760

13.892 15.686

2.217 ± 0.016 2.269 ± 0.013

2.249 ± 0.010

999 ± 22

996 ± 22

TT16∗

1.254 ± 0.013 “–”

7733 7738

2.22261 2.2602

21.677 29.112

2.215 ± 0.012 2.187 ± 0.012

2.201 ± 0.084

1012 ± 22

1009 ± 22

TT-30

TT-25

TT-22

Ar∗ (±1σ ) (10−12 mol g−1 )

Ar∗ weighted mean (±1σ ) (10−12 mol g−1 )

Age(1) (±2σ ) (ka)

Age(2) (±2σ ) (ka)

Ages are calculated using the decay constants of Steiger and Jäger (1977). ∗: Samples TT-55 and TT-16, equivalent to samples JCD-550 and KAR-41, in Carracedo et al. (2011), have been recalculated according to updated calibration of the K–Ar mass-spectrometer. K refers to potassium content measured from the groundmass split, Lab n◦ to the number of the K–Ar experiment, M to the mass of molten groundmass, 40 Ar∗ (%) to the percentage of radiogenic 40 Ar. The 40 Ar∗ weighted means are calculated on the basis of FC sanidine at 28.1 Ma so ages are reported in (1) relative to this age and in (2) recalculated from (1) relative to 28.02 Ma FC sanidine.

Table 2 Summary of

40

Ar/39 Ar data from incremental heating experiments (Pared de Güímar, Tenerife, Spain).

Sample site Experiment no.

wt. (mg)

K/Ca (total)

Total fusion age (ka)

Age spectrum Increments used (◦ C)

39

Ar (%)

Age ± 2σ (ka)

Isochron analysis

TT-30, groundmass FG-144 to FG-154 FG-155 to FG-163

90 86

0.59 0.42

994 ± 14 982 ± 21

650–1380 650–860

100.0 80.9

994 ± 15 990 ± 21

126 137

0.65 0.61

975 ± 13 971 ± 17

650–980 603–1059

93.7 90.8

81 80

0.39 0.51

1010 ± 12 999 ± 11

650–1120 650–1200

88.9 100.0

94 137

0.19 0.38

1006 ± 14 996 ± 7

700–1060 602–1094

71.5 89.1

119 135

0.23 0.34

994 ± 21 945 ± 16

weighted mean plateau and isochron ages from two experiments: simple mean plateau and isochron ages from two experiments:

1.41 1.60

11 of 11 5 of 8

1.55 0.91

296.3 ± 4.5 297.7 ± 7.8

978 ± 13 975 ± 17

1003 ± 11 999 ± 11

988 ± 13 1000 ± 7

700–950 700–980

85.8 76.9

994 ± 21 972 ± 17

Age ± 2σ (ka)

991 ± 20 984 ± 30 989 ± 17 988 ± 25

0.82 0.27

7 of 8 9 of 10

0.68 0.23

293.9 ± 2.7 296.3 ± 2.1

990 ± 23 970 ± 21 979 ± 16 980 ± 22

0.84 1.69

8 of 9 9 of 9

0.96 1.88

296.0 ± 3.1 294.8 ± 3.2

1002 ± 15 1001 ± 15 1002 ± 11 1002 ± 15

0.66 0.14

7 of 10 9 of 10

0.58 0.16

293.6 ± 3.7 295.4 ± 3.1

997 ± 6 994 ± 10

weighted mean plateau and isochron ages from two experiments: simple mean plateau and isochron ages from two experiments: TT-22, groundmass FG-030 to FG-040 FG-071 to FG-079

40

1001 ± 8 1001 ± 11

weighted mean plateau and isochron ages from two experiments: simple mean plateau and isochron ages from two experiments: TT-25, groundmass FG-104 to FG-113 FG-399 to FG-408

MSWD

977 ± 10 976 ± 15

weighted mean plateau and isochron ages from two experiments: simple mean plateau and isochron ages from two experiments: TT-26, groundmass FG-114 to FG-122 FG-123 to FG-131

N

993 ± 12 992 ± 18

weighted mean plateau and isochron ages from two experiments: simple mean plateau and isochron ages from two experiments: TT-27, groundmass FG-310 to FG-317 FG-409 to FG-419

Ar/36 Ar ± 2σ intercept

MSWD

1000 ± 27 1001 ± 16 1001 ± 14 1000 ± 22

1.77 2.17

7 of 11 6 of 9

2.00 0.82

297.0 ± 5.0 290.5 ± 3.5

981 ± 13 983 ± 19

Ages calculated relative to 1.194 Ma Alder Creek Rhyolite sanidine standard. wt refers the mass of molten groundmass, Increment used corresponds to the range of temperature of the retained steps in the age calculation, cumulative 39 Ar released at this temperature interval, N is the number of steps used in the age calculation.

988 ± 30 995 ± 23 992 ± 18 992 ± 26

39

Ar% to the corresponding

C. Kissel et al. / Earth and Planetary Science Letters 406 (2014) 59–71

Fig. 2. Age spectra and isochrons depicting Uncertainties ±2σ .

40

63

Ar/39 Ar experimental results from transitional flows. Grey filled boxes were rejected for the plateau age and isochron calculations.

64

C. Kissel et al. / Earth and Planetary Science Letters 406 (2014) 59–71

Fig. 2. (continued)

experiments were performed on two to eight cores per sites. About 40% of the heating curves show a slow decrease in high field magnetization (or slight increase of k) at low temperature and a final drop between 500 ◦ C and 580–600 ◦ C where complete removal of the magnetization occurs (Fig. 3a). Cooling curves are reversible (Fig. 3a). In another 40% of the samples, complete removal of magnetization is observed around 600 ◦ C, after a continuous decrease from room temperature with a rather constant slope (Fig. 3b). Cooling curves are reversible also in these cases. Finally, in 20% of the samples, magnetization decreases rapidly, reaching negligible values in a temperature range of 250–400 ◦ C. The behavior is not reversible when the heating is made in one continuous run up to 600 ◦ C i.e. well above the temperature at which the magnetization is removed. This might be due to high temperature changes (Fig. 3c). Given the Curie temperatures defined above and the low oxydation state of the samples, the main magnetic carrier in the samples is therefore magnetite with various degrees of Ti content. The magnetic mineralogy is usually rather uniform within each flow but it may happen that mixture of low- and high-Ti content magnetites is observed in different samples from the same flow. Hysteresis parameters were obtained using an AGM 2900 from Princeton Measurement Corporation after applying a field of ±1 T and after slope correction. The magnetization and coercive force ratios, when reported on the diagram proposed by Dunlop (2002), fall into the pseudosingle (PSD) domain range along the single domain – multidomain (SD-MD) low Ti content magnetite mixing line with a percentage of SD ranging between 80% and 30% (Fig. 3d). Only the points corresponding to the flow TT33-34 of ankaramitic composition are outliers with respect to this mixing line and are close to the curve corresponding to titanomagnetites with Titanium substitution of x = 0.6 (Dunlop, 2002) (Fig. 3d).

FORC diagrams constructed for 48 of the samples show that the hysteresis parameters illustrate a uniform size of the magnetic particles within the samples rather than a mixture of SD-MD endmembers. Interestingly, at least for the 90 cm thick TT27 flow, no real systematic relationship is observed between the magnetic properties and the position within the flow. On the contrary, heterogeneities in the magnetic grain size are present in the same rather central position with respect to the flow top and bottom (Fig. 3d). 4.2. Stable directions A total of 142 specimens were subjected to either alternating field (AF) or thermal stepwise demagnetization in the μmetal shielded room at LSCE. AF demagnetization was performed using a LDA-3 degaussing coil from AGICO with a tumbling sample holder. Progressively increasing peak fields from 4 to 80 mT were applied in 11 to 14 steps. Thermal demagnetizations were conducted in air using two different PYROX furnaces in which the heating space is controlled by three separate heating coils with independently controlled thermocouples keeping the thermal gradient lower than 2 ◦ C at 500 ◦ C over the entire heating zone. Cooling is obtained with an air circulation over the samples in a field less than 1 nT. After each AF or thermal step, the samples were measured using a 755R-2G magnetometer for discrete samples (high homogeneity coils). Possible mineralogical changes upon heating were monitored by measuring the low field susceptibility value after each thermal demagnetization step. Apart from a viscous component removed after the very first demagnetization step (100 ◦ C or 4 mT), the direction of the characteristic component magnetization (ChRM) is stable (Fig. 4). Only ten directions, belonging to different sites were rejected from the

C. Kissel et al. / Earth and Planetary Science Letters 406 (2014) 59–71

65

Fig. 3. a), b) and c) are examples of thermomagnetic (or k–T ) curves illustrating the different types of magnetic behavior described in the text. d) Hysteresis ratios obtained by conducting magnetic hysteresis loops with an AGM2900. They are reported on the diagram proposed by Dunlop (2002) for magnetites. The FORC diagrams obtained from flow TT27 are taken as an example of the different grain size mixtures within the same flow. These FORC diagrams are based on a minimum of 100 measured individual FORC curves and constructed using the FORCinel software (Harrison and Feinberg, 2008). SF is for smoothing factor. The photo of the flow is reported aside to show the distribution of the samples.

Fig. 4. Examples of stepwise demagnetization diagrams obtained using either stepwise thermal or AF stepwise demagnetization as described in the text. Full (open) circles are for projections onto the horizontal (vertical) planes.

66

C. Kissel et al. / Earth and Planetary Science Letters 406 (2014) 59–71

Table 3 Dating and paleomagnetic results from the Pared de Güímar (Tenerife, Spain). Flow number

TT55 TT31-32 TT33-34 TT30 TT29 TT28 TT27 TT37 TT36 TT35 TT26 TT25 TT24 TT51 TT23 TT50 TT22 TT49 TT48 TT47 TT20 TT19 TT18 TT17 TT16

Ages ±2σ (ka)

971 ± 21 989 ± 13∗ 980 ± 13∗

1002 ± 11 994 ± 12∗

991 ± 14∗ 1005 ± 22 996 ± 22

1009 ± 22

n/ N

7/7 9/9 12/12 4/6 4/4 6/6 6/6 5/5 4/4 6/6 6/6 5/6 2/3 2/3 4/4 8/8 7/7 8/8 6/8 4/6 1/1 1/1 3/3 5/5 7/8

Directions

Paleointensity

Decl. (◦ )

Incl. (◦ )

k

α95 (◦ )

Latitude VGP (◦ )

Longitude VGP (◦ )

n/ N

160 184.1 169.1 155.5 223.7 213.3 182.4 172.7 188.5 176.8 123.7 116.0 114.4 147.6 163.5 −29.6 −39.5 −1.2 − 0.7 0.51 43.9 31.8 33.3 34.5 14.4

−46.3 −31.3 −24.6 −38.0 −40.5 −46.8 −18.6 −15.0 −16.9 −20.0 −17.7 −11.8 −7.9 −45.2 −52.0 18.7 9.8 41.3 40.4 41.9 42.0 60.0 52.3 52.8 42.8

69 53 142 189 278 101 70 158 36 91 135 54 – – 184 74 96 70 151 55 – – 294 372 445

7.3 7.1 3.7 6.7 5.5 6.7 8.0 6.1 15.6 9.9 5.8 10.5 – – 6.8 6.5 6.2 6.6 5.5 12.4 – – 7.2 3.3 2.9

−72.3 −78.0 −71.5 −66.7 −50.5 −60.7 −71.1 −68.2 −68.8 −71.7 −33.9 −25.7 −23.3 −61.3 −75.2 56.5 45.9 85.3 84.7 85.8 50.6 61.2 61.1 60.1 76.7

76.1 −35.6 19.2 61.7 −109.2 −113.9 −23.7 3 .4 −40.2 −6.3 65.2 66.3 65.2 78.2 94.5 −134.5 −130.9 177.1 170.4 157.3 69.1 39.2 56.1 55.1 85.3

0/5 4/7 0/6 0/4 2/4 2/4 3/3 0/4 2/3 3/4 0/4 2/3 0/3 2/3 1/4 0/4 0/8 2/5 1/4

F ± 1σ (μT)

VADM ±1σ (1022 A m2 )

VDM ±1σ (1022 A m2 )

29.8 ± 6.4

6.0 ± 1.3

7.0 ± 1.5

34.7 ± – 22.4 ± – 19.5 ± 4.6

7.0 ± – 4.5 ± – 3.9 ± 0.9

7.0 ± – 5.6 ± – 5.0 ± 1.2

23.1 ± – 8.2 ± 1.6

4.7 ± – 1.7 ± 0.3

5.8 ± – 2.1 ± 0.4

9.6 ± –

1.9 ± –

2.4 ± –

11.7 ± – 14.9 ± –

2.4 ± – 3.0 ± –

2.2 ± – 3.8 ± –

33.1 ± – 31.8 ± –

6.7 ± – 6.5 ± –

7.2 ± – 6.8 ± –

The flows are reported in stratigraphic order. The ages (2σ uncertainty) are obtained from unspiked K/Ar technique (Table 1) except for TT26 (see text) and those with ∗ are for pooled K/Ar and 40 Ar/39 Ar ages (Tables 1 and 2). n/ N are the numbers of data taken into account in the statistical calculation/total number of studied samples. Declinations and inclinations are calculated using Fisher statistic (k: precision parameter, α95 : confidence angle). The latitude and longitude of the virtual geomagnetic poles (VGP) are deduced from these directional data. F is for field, VADM for virtual axial dipole moment and VDM for virtual dipole moment.

calculation of the mean site statistical directions because they were outliers (Table 3). Directions calculated using PCA analysis (Kirschvink, 1980) are characterized for 80% of the studied samples by MAD values lower than 4◦ , the other 20% ranging between 4 and 10◦ (Fig. 4). Mean-site directions calculated using Fisher’s statistics based on 3 to 12 independent ChRM directions have a precision angle α95 ranging between 3.3◦ and 15.6◦ with an average of 7◦ . Two flows had only two reliable samples and another two have only one. The mean-flow directions are reported with their statistical parameters in Table 3. The first eight flows (TT16 to TT49) are of normal polarity while the following seven flows have clear transitional directions and uppermost last 10 flows are reversely magnetized. 4.3. Paleointensity Paleointensity (PI) experiments were conducted on 82 samples from 19 flows. These 82 samples were selected on the basis of the experiments described above: samples in which the magnetization is carried by high proportion of SD or PSD grains (as an example, samples TT27-03 and 05 were not analyzed), and those characterized by the thermomagnetic curves showed in Fig. 3a and b. Priority was also given to the transitional flows and to the normal and reverse flows close to the reversal. We used the original Thellier and Thellier method (Thellier and Thellier, 1959) with the field applied continuously during the direct and reverse heating in a furnace in which the field gradient is negligible and the temperature gradient is less than 2 ◦ C over the ∼50 cm long heating zone. Experiments were made in a gentle flow of argon, following the procedure of Kissel and Laj (2004). The number of heating steps, defined on the basis of the thermomagnetic curves and on the thermal demagnetization spectra, was variable (always exceeding 10 double heating steps) (Fig. 5). Numerous pTRM checks were performed either at the preced-

ing step or at much lower temperatures, when arriving at the end of the procedure. The original Thellier and Thellier procedure is powerful to detect formation of new minerals because their direction of magnetization align with that of the field applied in the furnace. As seen on the stereoplots reported in Fig. 5, this is not the case for the samples selected as reliable ones. The maximum acceptable value for the angular deviation (α ) between directions of magnetization obtained using the half sum of the data at each Thellier and Thellier step (Fig. 5) and from zero field demagnetizations was fixed at 15◦ . This is one of the selection set of criteria (PICRIT-03 criteria; Kissel and Laj, 2004) that we used to assess the reliability of intensity estimates. The other PICRIT-03 requirements are that the fraction of NRM taken into account in the determination ( f ) should exceed 0.35, the DRAT (difference ratio) should be less than 7% and the Cumulative DRAT (CDRAT) should be less than 10%. Only 24 specimens yielded reliable PI values, representing a relatively low rate of success of 29%. Reasons for failure are variable and are related to the different PICRIT-03 parameters and/or to unstabilities during the experiment. In only two cases, we have accepted PI values determined with f = 0.31 and 0.33, respectively, because they were consistent with the other PI values determined with higher f values within the same flows. The same within-flow consistency test lead us to keep a few PI values for which the maximum DRAT value ranges between 7 and 10% (associated with CDRAT values <10%). Mean values obtained from each flow yielding reliable paleointensity estimates are reported in Table 3. The uncertainties were calculated only for the three flows yielding more than two reliable determinations. Depending on their stratigraphic position, these mean-flow values are very variable, ranging between 8.2 ± 1.6 μT and about 33 μT, the latter being close to the present-day field intensity in Tenerife which is 38.4 μT (IGRF source).

C. Kissel et al. / Earth and Planetary Science Letters 406 (2014) 59–71

67

Fig. 5. Examples of sets of diagrams obtained from the Thellier and Thellier experiments. On the Arai plots (NRM/NRMo versus TRM/NRMo), circles are for double-heating at the temperature indicated close to the points; pTRM checks are reported with triangles. Solid symbols correspond to the part of the diagram used for determination of paleointensity, open symbols are for the rejected points. F is the intensity value estimated from the calculated slope (grey line). f , q, DRAT and CDRAT are the selection criteria (see Kissel and Laj, 2004). The corresponding double-orthogonal projections for NRM reconstructed from Thellier and Thellier experiments are also shown with open (full) dots for projection onto vertical (horizontal) plane. The declination, inclination and mean angular deviation (MAD) values calculated using PCA analysis are reported together with α , the angular deviation between this direction and the one defined after stepwise demagnetization. On the corresponding stereoplots (full/open symbols, lower/upper hemisphere), the projection of the field direction inside the oven, i.e. of the z axis of the samples is indicated by the cross (lower hemisphere). θ is the angle between this direction and the characteristic remanent magnetization. In both the demagnetization diagrams and the stereoplots, the data corresponding to the steps not taken into account for paleointensity determinations are reported in grey.

5. Discussion The results obtained from the “Pared de Güímar” section are summarized in Fig. 6 as (a) declinations, (b) inclinations, (c) VGP latitudes, (d) intensities, (e) Virtual Axial Dipole Moment (VADM) and Virtual Dipole Moment (VDM), (f) reversal angle and (g) ages when available, all versus the flow number. 5.1. Directional and paleointensity pattern throughout the reversal The bottom part of the sequence is constituted of 8 normal polarity flows corresponding to the top part of the Jaramillo subchron. The reversal angle, i.e. the angular deviation between the mean-flow direction and that expected for a geocentric axial dipole (GAD), is smaller than 25◦ . The observed scatter is therefore within the range of usual secular variation at this latitude (McElhinny and McFadden, 1997). These normal directions can be divided in three groups: TT16, dated at 1009 ± 22 ka, yields an easterly deviated declination and an inclination value very close to the GAD one. The four successive thin flows (TT17-18-19-20) most likely belong to the same volcanic unit with directions slightly easterly deviated with respect to that of TT16. They are followed by 3 flows (TT47-48-49) with similar directions close to the GAD direction. Among the 8 normal flows, paleointensity was studied for

9 specimens from flows TT47 and TT48 and 3 reliable determinations could be obtained. They yield together a mean PI value of 32.7 ± 6.7 μT, consistent with the present-day 38.5 μT value (IGRF source). These two flows are dated at 996 ± 22 ka and 1005 ± 22 ka respectively, consistent with the top part of the Jaramillo normal polarity subchron (Singer, 2014). Above TT49, 7 flows (see Fig. 6) are characterized by anomalous directions for which the reversal angle exceeds 50◦ . The first two transitional flows (TT22, TT50) have positive inclinations but significantly lower than the GAD value (47◦ ) giving reversal angles of 50◦ and 37◦ respectively. Only one sample from flow TT50 yielded a reliable PI value of about 15 μT i.e. less than 40% of the nontransitional value. The two overlying flows (TT23 and TT51) have steep negative inclinations and they are close to the reverse axial dipole, but they are characterized by a low PI value (11.7 μT). They are followed by three flows also characterized by negative but much shallower inclinations (reversal angle ∼100◦ ) with two mean-flow PI values of 9.6 μT and 8.2 ± 1.6 μT, i.e. about 20% of the non-transitional field. All these flows have recorded transitional paleomagnetic directions and intensities. Four overlying flows (TT35 to 37 and TT27) yield declinations close to 180◦ but inclinations shallower than expected on the basis of the GAD value. The associated PI values, around 20–23 μT are

68

C. Kissel et al. / Earth and Planetary Science Letters 406 (2014) 59–71

Fig. 6. Paleomagnetic and dating results obtained from the Pared de Güímar section. The results are reported in stratigraphic order from the Upper Jaramillo (TT16 at the bottom) to the Matuyama (TT55 at the top). a) and b) are mean-flow inclination and declination respectively obtained from the Fisher statistics with the associated uncertainties (α95 for inclination and α95 / cos I for declination). The dashed lines are for the expected GAD values at this site; c) paleointensity values obtained from Thellier and Thellier (1959) experiment. The flow means are reported by the black dots and the red dots correspond to the individual values obtained in each flow. The dashed line is for the present-day value at this site (IGRF source); d) latitudes of the virtual geomagnetic poles (VGPs) across the section; e) VADM (black) and VDM (blue) values calculated from the mean flow paleointensity values; f) reversal angle; g) ages reported with their 2σ uncertainty. The red dashed line is the linear regression taking into account the entire dataset. The yellow envelope is for a 2 sigma propagating gaussian error also given by the ISOPLOT 3.75 software. The grey zone highlights the transitional interval. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

two to three times larger than those of the underlying transitional lavas. In the following flows (TT28 and TT29), the inclination is similar to the reverse GAD one but the declinations are slightly westerly deviated. The PI value obtained from TT28 is about 35 μT, similar to the present-day one. Finally, the sequence ends with the last 4 flows (TT30; TT33-34; TT31-32; TT55) characterized by reverse directions associated with a high paleointensity values (∼30 μT) and the youngest age at 971 ± 21 ka (TT-55). 5.2. Timing and duration of the upper Jaramillo reversal The comparison of the Canarian UJR record with the few other existing records of the same reversal, necessitates to calculate its central age and duration. The centered age of the reversal, obtained by pooling (weighted mean) the ages of the three dated transitional flows (pooled 40 Ar/39 Ar isochron and unspiked K–Ar ages of TT22 and TT25 and isochron age of TT26) is 996 ± 7 ka. This age is consistent with the 1001 ± 10 ka age obtained for the Punaruu sequence at Tahiti (Singer, 2014) and with the age of 990 ka attributed to the UJR in marine stratigraphy (Lisiecki and Raymo, 2005; Channell et al., 2009) on the basis of astronomical tuning. All the records of this reversal are therefore consistent in age. Another age of 28.201 Ma has been proposed for FC sanidine (Kuiper et al., 2008) and the ages of the reversal at Tenerife and at Tahiti re-calculated to this value are 1002 ± 8 ka and 1008 ± 10 ka respectively. They remain consistent with one another. It is interesting to note that the calibration to 28.02 Ma is more consistent with astronomical tuning than the calibration to 28.201 Ma. This was also the case for the Brunhes–Matuyama reversal (Channell et al., 2010). However, the

uncertainty associated to the astronomical tuning is not precisely known so that this difference may not be statistically significant. In a first approximation, to calculate the duration of the reversal from our age data, we assume a linear and constant lava accumulation rate for the entire section. Such an assumption is a very first approximation mostly because of the sporadic character of volcanic eruptions and because the age uncertainties are of the same order of magnitude than the duration of event to be dated, i.e. the reversal itself. The studied section emplaced during a high extrusion rate period as confirmed by the absence of thick red baked soils between the flows and the overlapping of the radiometric ages within errors. Such observations are consistent with a high and regular accumulation rate although we cannot exclude that the transitional lava flows (grey zone in Fig. 6) erupted very rapidly in succession as statistically the top has an overlapping uncertainty with that of the lower boundary. A linear regression calculated using ISOPLOT 3.76 software (Ludwig, 2012) and taking all data into account (Fig. 6g) indicates an accumulation rate of 0.92 ± 0.67 flow/ka equivalent to 1 flow every 1.09 ± 0.80 ka. It is characterized by a MSWD (Mean Square Weighted Deviation, i.e. ratio of the observed scatter of the points from the best-fit line to the expected scatter from the assigned errors and error correlation) value of 1.11 and a P (probability of fit) value of 0.36. These values show that our assumption of a linear and constant accumulation rate is statistically reasonable. We obtain a duration of 7.6 ± 5.6 ka for the emplacement of the seven successive transitional flows. Other estimations of the duration of this reversal have been previously published but it should be noted that our calculation is the first one defined with uncertainties. It is longer on the average than the previous estimate of 3200 years (Zhu et al., 1994)

C. Kissel et al. / Earth and Planetary Science Letters 406 (2014) 59–71

Fig. 7. Path of the virtual geomagnetic poles for the Upper Jaramillo transition from a) Pared de Güímar (Tenerife) (this study); b) the North Atlantic sediments (Channell and Lehman, 1997; Mazaud et al., 2009) and c) Tahiti (Chauvin et al., 1990). The VGPs from Tahiti are reported with different colors corresponding to the different sections or groups of sections (see text).

but similar to the intensity low corresponding to the transition in the sediments (about 8 ka at mid height) (Channell et al., 2009; Mazaud et al., 2009). 5.3. Intensity and geometry of the transitional field The VADM and VDM values are very similar (Fig. 5e). VDM values are a little larger than the VADM values above the transition, due to the non-purely reverse inclinations of flows TT35 to 37 and TT25. They vary from about 6–7 × 1022 A m2 outside the reversal to about 2–2.5 × 1022 A m2 during the transition, representing a drop of about 60%. This drop is expected in case of reversal and shows that all these flows are transitional ones. No significant difference is observed in the calculated dipolar moment values (VADM, VDM) before and after the reversal. This is consistent with the paleointensity pattern observed from sediments (Channell et al., 2009) although in contradiction with the saw-tooth pattern invoked for the general behavior of the intensity through geomagnetic reversals (Valet and Meynadier, 1993). From a geometrical point of view, when the virtual geomagnetic poles (VGPs) are plotted on a planisphere (Fig. 7a), they are first lying in the northern hemisphere, close to the Alaskan coastline (flows TT22, TT50). Then, they jump with no precisely defined path to the southern hemisphere (TT23), over Antarctica. A loop toward La Réunion Island is then described by four flows (TT51; 24; 25 and 26) before coming back to the South Pole with some scatter around it, within the limits of paleosecular variation (McElhinny and McFadden, 1997). Only the two flows TT28 and TT29 have VGPs more distant than the others from the South Pole. Their latitudinal deviation exceeds 30◦ and they lie in the southeastern Pacific.

69

This VGP path presents similarities with the one constructed from the North Atlantic sedimentary sequences (Channell and Lehman, 1997; Mazaud et al., 2009) (Fig. 7b). The first part, constituted by the anomalous directions and low intensities from TT22 and TT50 belong to the first loop over Americas observed in the North Atlantic sediments before the directions return to normal ones. During the second part of the reversal, the sedimentary VGPs gently move from north to south whereas the volcanic VGPs move to La Réunion island from high southern latitudes with no transitional lavas with latitudes that are equatorial. The sedimentary and volcanic VGPs are both along the same longitudinally confined path between La Réunion Island and the south pole. Both VGPs therefore tend to lie over the two longitudinal bands described for the Brunhes–Matuyama reversal (Clement, 1991), and as a persistent phenomenon for reversals over the last 12 Ma (Laj et al., 1991). The observation of these longitudinal positions in the UJR record from the Canarian volcanic sequence indicates that these paths are real geometrical features of the earth magnetic field during the UJR, at least for this North Atlantic sector and do not result from sedimentary artifacts (Langereis et al., 1992; Quidelleur et al., 1995). It is also interesting to note that the accumulation of VGPs at low southern latitudes observed in sedimentary sequences (Mazaud et al., 2009), corresponds to the position of the volcanic VGPs from flows TT24, TT25 and TT26. Of course, a major uncertainty in the description of this kind of behavior, both in sediments and lavas, is linked to the lack of knowledge of rapid possible changes in accumulation rates, both in sediments and in lavas. However, it would be an unlikely coincidence if an increased eruptive rate in the Canaries were coeval with an increased sedimentary accumulation rate in North Atlantic. Our data therefore suggest that the observed VGP clusters have also a real geomagnetic meaning along these longitudinal paths. With respect to the three phases described and discussed by Valet et al. (2012), the first transitional flows TT22 and TT50 could correspond to the precursor and flows TT23 to 26 and TT50 to the reversal itself or to a post-reversal rebound. No information is given about the intensity pattern by Valet et al. (2012) but the low PI values obtained here for all the anomalous flows suggest that these flows describe the transition itself. The VGP path obtained by Chauvin et al. (1990) in Tahiti from the only other available volcanic record of the UJR is also complex (Fig. 7c). The sequence contains twenty-three transitional flows sampled along five different sections combined into a single sequence on the basis of field observations or on relationships between altitude, bedding and paleomagnetic directions (Chauvin et al., 1990). The Tahitian VGPs are reported in Fig. 7c with colors reflecting the different sections or group of sections. A rather complicated loop is first described in Punaruu sections P-C and MU with VGPs directly heading south as far as New Zealand and Australia with no intermediate points (red in Fig. 7c). The single PI value obtained from this Punaruu sequence is from this first loop (Chauvin et al., 1990). It is followed by section M1 combined with flows CS, CT and CU (green dots) going back to the north with a VGP cluster close to those observed from North Atlantic sediments and close to the two Canarian flows (TT22 and TT50) on the western coast of central and north America. The following normal to reverse path observed at Tahiti (blue in Fig. 7) contains more points with the first most detailed part over northwestern Pacific. All the VGPs obtained from the Punaruu section are therefore distributed around the Pacific Ocean with a first loop followed by a rather simple normal to reverse path as for the North Atlantic records. It is interesting to note here that the description of a volcanic record varies from one author to the other. For example what we describe as the reversal in the Punaruu record, consistently with Singer (2014), is attributed to a post-reversal rebound

70

C. Kissel et al. / Earth and Planetary Science Letters 406 (2014) 59–71

by Valet et al. (2012) whereas what we interpret as the first loop is the main reversal for the same authors. The database of the Upper Jaramillo reversal is still far too poor to simulate or constrain new modeling. The only point which can be examined at this moment is the degree of symmetry of the transitional field. Gubbins and Love (1998) presented a simple model based on an antisymmetry of the field about the equator, a 180◦ symmetry about the polar axis and flux concentrations on two longitudes. According to this model, a normal to reverse polarity change with polarward motion of the fluxes should be described by VGP paths west of the sites located in the North Atlantic region and east of the Tahiti site. Although the mechanism given by Gubbins and Love (1998) is consistent for the well documented Bruhnes/Matuyama reversal, it does not seem to correctly describe the volcanic records for the Upper Jaramillo because we observe an inverse relationship. So, even if we significantly contribute to the database UJR with a new volcanic record geographically almost opposite to Tahiti, additional data are definitely needed before any robust suggestion can be made about the mechanisms of this normal to reverse field reversal. 6. Conclusions This combined radiometric dating and paleomagnetic study conducted on the “Pared de Güímar” in Tenerife (Canary Islands) provides a new record of the Upper Jaramillo reversal from a lava sequence. It is based on 25 successive lava flows ranging in age between 1009 ± 22 ka and 971 ± 21 ka. Eight of the flows are normal, 7 are transitional, and 10 are reverse. The intensity profile exhibits identical values of about 25–35 μT before and after the reversal consistent with the present day value. Transitional paleointensity values are as low as 8 μT. The nine age determinations (K/Ar or 40 Ar/39 Ar or pooled K/Ar–40 Ar/39 Ar) obtained through the sequence and the assumption of a constant extrusion rate suggest the extrusion of one flow per 1.09 ± 0.80 ka. This extrusion rate is consistent with the high eruptive rate recognized during the activity of the NERZT (Carracedo et al., 2011) and suggests a duration for the reversal itself of 7.6 ± 5.6 ka. The transitional directions correspond to a VGP path longitudinally confined over North America and South Indian Ocean, consistently with VGPs defined from sedimentary records of the same reversal. The coincidence in volcanic and sedimentary VGP paths evidently make this a real geomagnetic characteristic. In addition, the volcanic VGPs coincide with clusters observed in sedimentary records suggesting that, in fact, the field has lingered at some positions during the reversal in a stop-and-go type of behavior. Compared to the VGP path described from the only other volcanic sequence recording the UJR reversal and located almost opposite to the Canary Islands, in Tahiti, there are differences in VGPs locations and similarities such as a loop preceding the reversal itself. Although the new sequence we report on from the upper Jaramillo transition from the Canary Islands contributes to the database, the latter is still not sufficient to draw definite conclusion about the morphology and dynamic of the reversing field during the upper Jaramillo transition. The intensity low which characterizes the interval during which the field is transitional still remains the best tie-point for chronostratigraphy, centered in age at 996 ± 7 ka (2σ , relative to FC sanidine at 28.02 Ma). Acknowledgements This work has received the financial support of the French Commissariat à l’Energie Atomique et aux Energies Alternatives (CEA) and by the Centre National de la Recherche Scientifique (CNRS) in

France and by the Plan Nacional I + D + I, Projects CGL2005-00239 and CGL2008-02842/BTE in Spain. Acknowledgements are due to J.L. Joron (Lab. P. Süe, CEA Saclay), who performed the irradiations of the samples, to J.F. Tannau (LSCE) for his constant technical help and to N. Smialkowski for the sample preparations. A. Mazaud provided the numerical data from the North Atlantic sediments and he helped with the use of the ISOPLOT software. We are specially grateful to A.P. Koppers who contributed a lot in the improvement of the manuscript. This is LSCE contribution n◦ 5401. Appendix A. Supplementary material Supplementary material related to this article can be found online at http://dx.doi.org/10.1016/j.epsl.2014.09.003. References Berggren, W.A., Kent, D.V., Flynn, J.J., Van Couvering, J.A., 1985. Cenozoic geochronology. Geol. Soc. Am. Bull. 96, 1407–1418. Carracedo, J.C., Guillou, H., Nomade, S., Rodríguez-Badiola, E., Pérez-Torrado, F.J., Rodríguez-González, A., Paris, R., Troll, V.R., Wiesmaier, S., Delcamp, A., Fernández-Turiel, J.L., 2011. Evolution of ocean island rifts: the Northeast rift zone of Tenerife, Canary Islands. GSA Bull. 123, 562–584. Channell, J.E.T., Lehman, B., 1997. The last two geomagnetic polarity reversals recorded in high-deposition rate sediment drifts. Nature 389, 712–715. Channell, J.E.T., Xuan, C., Hodell, D.A., 2009. Stacking paleointensity and oxygen isotope data for the last 1.5 Myr (PISO-1500). Earth Planet. Sci. Lett. 283, 14–23. Channell, J.E.T., Hodell, D.A., Singer, B.S., Xuan, C., 2010. Reconciling astrochronological and 40 Ar/39 Ar ages for the Matuyama–Brunhes boundary and late Matuyama Chron. Geochem. Geophys. Geosyst. 11, Q0AA12. http://dx.doi.org/10.1029/2010GC003203. Charbit, S., Guillou, H., Turpin, L., 1998. Cross calibration of K–Ar standard minerals using an unspiked Ar measurement technique. Chem. Geol. 150, 147–159. Chauvin, A., Roperch, P., Duncan, R.A., 1990. Records of geomagnetic reversals from volcanic islands of French Polynesia, 2, Paleomagnetic study of a flow sequence (1.2 to 0.6 Ma) from the Island of Tahiti and discussion of reversal models. J. Geophys. Res. 95, 2727–2752. Clement, B.M., 1991. Geographical distribution of transitional VGP’s: evidence for non-zonal equatorial symmetry during the Matuyama–Brunhes geomagnetic reversal. Earth Planet. Sci. Lett. 104, 48–58. Dinares-Turell, J., Sagnotti, L., Roberts, A.P., 2002. Relative geomagnetic paleointensity from the Jaramillo Subchron to the Matuyama/Brunhes boundary as recorded in a Mediterranean piston core 2002. Earth Planet. Sci. Lett. 194, 327–341. Doell, R.R., Dalrymple, G.B., 1966. Geomagnetic polarity epochs: a new polarity event and the age of the Brunhes–Matuyama boundary. Science 152, 1060–1061. Dunlop, D.J., 2002. Theory and application of the Day plot (M rs / M s versus H cr / H c ) 1. Theoretical curves and tests using titanomagnetite data. J. Geophys. Res. 107 (B3), 2056. http://dx.doi.org/10.1029/2001JB000486. Gubbins, D., Love, J.J., 1998. Preferred VGP paths during geomagnetic polarity reversals: symmetry considerations. Geophys. Res. Lett. 25 (7), 1079–1082. Guillou, H., Kissel, C., Laj, C., Carracedo, J.C., 2013. Dating the Teide Volcanic Complex: radiometric and palaeomagnetic methods. In: Carracedo, J., Troll, R.V. (Eds.), Teide Volcano; Geology and Eruptions of a Highly Differentiated Oceanic Stratovolcano, Series: Active Volcanoes of the World. Springer. 279 pp. Harrison, R.J., Feinberg, J.M., 2008. FORCinel: an improved algorithm for calculating first-order reversal curve distributions using locally weighted regression smoothing. Geochem. Geophys. Geosyst. 9, Q05016. http://dx.doi.org/10.1029/ 2008GC001987. Izett, G.A., Obradovich, J.D., 1994. 40 Ar/39 Ar constraints for the Jaramillo Normal Subchron and the Matuyama–Brunhes geomagnetic boundary. J. Geophys. Res. 99, 2925–2934. Kirschvink, J.L., 1980. The least square lines and plane analysis of palaeomagnetic data. J. R. Astron. Soc. 62, 319–354. Kissel, C., Laj, C., 2004. Improvements in procedure and paleointensity selection criteria (PICRIT-03) for Thellier and Thellier determinations: application to Hawaiian basaltic long cores. Phys. Earth Planet. Inter. 147, 155–169. Koppers, A.A.P., 2002. ArAr CALC—software for 40 Ar/39 Ar age calculations. Comput. Geosci. 28, 605–619. Kuiper, K.F., Deino, A., Hilgen, F.J., Krijgsman, W., Renne, P.F., Wijbrans, J.R., 2008. Synchronizing rock clocks of Earth history. Science 320, 500–504. http:// dx.doi.org/10.1126/science.1154339. Laj, C., Mazaud, A., Weeks, R., Fuller, M., Herrero-Bervera, E., 1991. Geomagnetic reversal paths. Nature 351, 447. Laj, C., Kissel, C., Herrero-Bervera, E., Garnier, F., 1996. Relative geomagnetic field intensity and reversals for the last 1.8 Myr from a central Pacific core. Geophys. Res. Lett. 23, 3393–3396.

C. Kissel et al. / Earth and Planetary Science Letters 406 (2014) 59–71

Langereis, C.G., VanHoof, A.A.M., Rochette, P., 1992. Longitudinal confinement of geomagnetic reversal paths as a possible sedimentary artifact. Nature 358, 226–230. Leonhardt, R., Heider, F., Hayashida, A., 1999. Relative geomagnetic field intensity across the Jaramillo subchron in sediments from the California margin: Ocean Drilling Program Leg 167. J. Geophys. Res. 104 (B12), 29133–29146. Lisiecki, L.E., Raymo, M.E., 2005. A Pliocene–Pleistocene stack of 57 globally distributed benthic δ 18 O records. Paleoceanography 20 (1), 1003. 1–17. Ludwig, K.R., 2012. Using Isoplot/Ex, version 3.75: a geochronological toolkit for Microsoft Excel. Berkeley Geochronology Center Special Publication 5. Mazaud, A., Channell, J.E.T., Xuan, C., Stoner, J.S., 2009. Upper and lower Jaramillo polarity transitions recorded in IODP Expedition 303 North Atlantic sediments: implications for transitional field geometry. Phys. Earth Planet. Inter. 172, 131–140. McElhinny, M.W., McFadden, P.L., 1997. Palaeosecular variation over the past 5 Myr based on a new generalized database. Geophys. J. Int. 131 (2), 240–252. Nomade, S., Renne, P.R., Vogel, N., Deino, A.L., Sharp, W.D., Becker, T.A., Jaouni, A.R., Mundil, R., 2005. Alder creek sanidine (ACs-2): a Quaternary 40 Ar/39 Ar dating standard tied to the Cobb Mountain geomagnetic event. Chem. Geol. 218, 315–338. Quidelleur, X., Holt, J., Valet, J.P., 1995. Confounding influence of magnetic fabric on sedimentary records of a field reversal. Nature 274, 246–249. Renne, P.R., Swisher, C.C., Deino, A.L., Karner, D.B., Owens, T.L., DePaolo, D.J., 1998. Intercalibration of standards, absolute ages and uncertainties in 40 Ar/39 Ar dating. Chem. Geol. 145, 117–152. Shackleton, N.J., Berger, A., Peltier, W.R., 1990. An alternative astronomical calibration of the lower Pleistocene timescale based on ODP site 677. Trans. R. Soc. Edinb. Earth Sci. 81, 251–261. Singer, B.S., 2014. A Quaternary Geomagnetic Instability time-scale. Quat. Geochronol. 21, 29–52. http://dx.doi.org/10.1016/j.quageo.2013.10.003. Singer, B.S., Hoffman, K.A., Chauvin, A., Coe, R.S., Pringle, M.S., 1999. Dating transitionally magnetized lavas of the late Matuyama Chron: toward a new 40 Ar/39 Ar

71

timescale of reversals and events. J. Geophys. Res. 104 (B1), 679–693. Singer, B.S., Brown, L.L., Rabassa, J.O., Guillou, H., 2004. 40 Ar/39 Ar chronology of late Pliocene and early Pleistocene geomagnetic and glacial events in southern Argentina. In: Channell, J.E.T., et al. (Eds.), Timescales of the Paleomagnetic Field. In: American Geophysical Union Geophysical Monograph Series, vol. 145. AGU, Washington, DC, pp. 175–190. Spell, T.L., Harrison, T.M., 1993. 40 Ar/39 Ar geochronology of post-Valles caldera rhyolites, Jemez volcanic field, New Mexico. J. Geophys. Res. 98, 8031–8051. Spell, L.T., McDougall, I., 1992. Revisions to the age of the Matuyama–Brunhes boundary and the Pleistocene polarity timescale. Geophys. Res. Lett. 19, 1181–1184. Spell, T.L., McDougall, I., 2003. Characterization and calibration of 40 Ar/39 Ar dating standards. Chem. Geol. 198, 189–211. Steiger, R.H., Jäger, E., 1977. Subcommission on geochronology: convention on the use of decay constants in geo- and cosmochronology. Earth Planet. Sci. Lett. 36, 359–362. Tauxe, L., Deino, A.D., Beherensmeyer, A.K., Potts, R., 1992. Pinning down the Brunhes/Matuyama and upper Jaramillo boundaries: a reconciliation of orbital and isotopic time scales. Earth Planet. Sci. Lett. 109, 561–572. Thellier, E., Thellier, O., 1959. Sur l’intensité du champ magnétique terrestre dans le passé historique et géologique. Ann. Geophys. 15 (3), 285–376. Valet, J.-P., Meynadier, L., 1993. Geomagnetic field intensity and reversals during the last four million years. Nature 366, 234–238. Valet, J.P., Fournier, A., Courtillot, V., Herrero-Bervera, E., 2012. Dynamical similarity of geomagnetic field reversals. Nature 490, 89–93. http://dx.doi.org/10.1038/ nature11491. Verosub, K.L., Herrero-Bervera, E., Roberts, A.P., 1996. Relative geomagnetic paleointensity across the Jaramillo subchron and the Matuyama/Brunhes boundary. Geophys. Res. Lett. 23 (5), 467–470. Zhu, R., Carlo Laj, C., Mazaud, A., 1994. The Matuyama–Brunhes and Upper Jaramillo transitions recorded in a loess section at Weinan, north–central China. Earth Planet. Sci. Lett. 125, 143–158.