Age constraints on island-arc submarine basalts from geomagnetic paleointensity

High-resolution dating of volcanic rocks is the foundation for understanding the evolution of volcanoes and for estimating possible hazards. However, dating is often difficult for submarine volcanoes, where radiocarbon or other dating is frequently unavailable or imprecise. Here, we report paleointensity results from submarine basalts around Izu-Oshima Island, a typical island-arc volcano, and their bearing on age constraints. Basaltic lava samples were collected from submarine ridges located southeast of Izu-Oshima Island. Rock magnetic data indicate that the samples contain Ti-rich titanomagnetite with blocking temperatures of around 250–400 °C. The magnetic properties of the samples do not change significantly when heated in Ar or vacuum. We apply the Tsunakawa-Shaw method to estimate absolute paleointensity. Samples from different submarine ridges show distinct behavior. One ridge (SE1) shows moderate paleointensity of about 37 μT, while another ridge (SE3) records relatively strong magnetic fields of about 60 μT. Comparing those results with regional paleointensity data, we estimate the age of the SE1 ridge to be younger than 0.5 ka or around 1.4 ka. The other ridge (SE2) exhibits different paleointensity for two samples obtained from different localities, implying that the ridge consists of multiple eruptions. These results demonstrate that paleomagnetism can improve the dating of submarine volcanic rocks.


Introduction
Submarine volcanic activity can cause severe disruption to both civilization and natural systems (e.g., Imamura et al. 2022;Ohno et al. 2022).However, submarine eruption history is generally not well known compared to subaerial volcanoes.It is particularly difficult to reconstruct a high-resolution submarine volcanic activity; while detailed radiocarbon dating of associated materials is sometimes applicable to document the activity of subaerial volcanoes over the past few tens of thousands of years (e.g., Kawanabe 2012;Ishizuka et al. 2015), this method is essentially unavailable for submarine volcanic rocks.
Geomagnetic paleosecular variations of both direction and intensity have been utilized to estimate or refine numerical ages for recent volcanic and sedimentary materials (e.g., Greve and Turner 2017;Nitta et al. 2020;Kanamatsu et al. 2022).Submarine samples often lack orientation control, meaning that we cannot determine the paleomagnetic directions.Nevertheless, it has been demonstrated that age estimation for unoriented basaltic glass samples from mid-ocean ridges is possible using paleointensity only (Carlut and Kent 2000;Carlut et al. 2004;Bowles et al. 2006;2014).Basaltic rocks on the seafloor generally undergo low-temperature oxidation which hampers absolute paleointensity determination due to overprint by thermochemical magnetic remanence (e.g., Grommé et al. 1979;Riisager et al. 2003).On the contrary, the timing of pervasive oxidation is still under debate.From mid-ocean ridge basalt, apparently successful absolute paleointensity results were reported from young pillows with ages spanning a few months (Kent and Gee 1996) to 100 ka (Grommé et al. 1979;Prévot et al. 1983).
In addition, marine magnetic anomalies around a spreading ridge revealed patterns consistent with absolute and relative paleointensity data up to 800 ka (Gee et al. 2000), suggesting that most remanence carrier minerals and their thermoremanent magnetization (TRM) may be preserved.Rocks from arc settings have not been examined systematically yet, but we may expect young samples to provide reliable absolute paleointensity.
In this paper, we estimate the paleointensity of submarine lavas around Izu-Oshima Island, a typical island-arc volcano.We compare the results with regional data and global models.Together with age constraints by petrography and geochemistry, we show that paleomagnetism can refine age estimates of these submarine lavas.

Izu-Oshima volcano
Izu-Oshima volcano is located about 100 km south of Tokyo.The top of the volcano forms an island (Izu-Oshima Island), which is inhabited by ~ 6000 people.The volcano is at the northernmost part of the Izu-Bonin arc.Written records indicate that the Izu-Oshima volcano has erupted every 100-150 years from the seventh century CE until the mid-nineteenth century, and then every 30-50 years to the present with smaller scale eruptions (Kawanabe 1998).Geological data show that the magmatic activity in the current summit region of Izu-Oshima island goes back to ~ 20 ka.The activity during the past ~ 20 kyr has produced satellite cones and eruptive fissures trending NW-SE both on land and in the sea.Petrographic and geochemical analyses of submarine ridges suggest that these edifices were also fed by Izu-Oshima volcano magmas (Ishizuka et al. 2014).Earlier activity began around 50 ka, and its remnant is exposed on the NE flank of the island (Kawanabe 1998).
For submarine activities, limited age constraints have been obtained from petrography and geochemistry.Petrography of both subaerial and submarine volcanic rocks distinguishes plagioclase-phyric and sparsely phyric basalts.Kawanabe (1991) proposed that these two petrography types approximately characterize eruptions before and after a large eruption at 1.8 ka, respectively (Nakamura 1964;Kawanabe 2012), even though the topmost part of the pre-eruption basalt includes aphyric basalt (Kawanabe 1991).Radiocarbon-dated subaerial Izu-Oshima magma spanning the last 14 ka reveals large variation in trace element compositions (in particular Ba, La, Yb, and Ce) and Sr, Nd, and Pb isotope ratios (Ishizuka et al. 2015).Those variations were interpreted to reflect a mixing of frontal arc Izu-Oshima magma with back-arc magma with a smaller slab-derived fluid influx, similar to an adjacent volcanic field.Specifically, the composition was nearly pure frontal-arc component at 14 ka, followed by an increase in the back-arc component until the highest contribution (~ 40%) at around 5.4 ka.Then, the composition rapidly returned to the frontal-arc composition.The submarine rocks reveal a similar compositional range (Ishizuka et al. 2014), suggesting that the geochemical correlation to the subaerial samples can provide crude age estimates.

Samples
We measure samples from submarine ridges (chains of cones) southeast of the Izu-Oshima volcano (Fig. 1).Because the samples are from topographic high, we consider they originated from the corresponding ridge.The samples were obtained using robotic arms of ROV Hyper-Dolphin during NT09-06 (R/V Natsushima) cruise in 2009.We use samples from three ridges namely SE1, SE2, and SE3 (Fig. 1 and Table 1); SE1 and SE3 samples were collected at sites close to each other on each ridge, while SE2 samples were obtained from two sites about 2 km apart.The original rock samples were between 3 and 30 kg each.The activity of the submarine ridges was inferred to be < 14 ka, based on geological observations (Ishizuka et al. 2014).ROV observations and seismic profiles show no sediment cover on the submarine volcanic ridges, suggesting young ages.In particular, the seismic profiles of the submarine ridges indicate that the lavas and volcaniclastics along the ridges are at the same stratigraphic level as the uppermost layer of the southeastern slope of the Izu-Oshima edifice (Irozaki-oki group; Okamura et al. 1999).This stratigraphic correlation indicates that the ridges are as young as subaerial satellite cones (< 14 ka).The samples from SE1 and SE2 are sparsely phyric, suggesting they are younger than ~ 1.8 ka.On the contrary, the samples from SE3 show more abundant plagioclase phenocryst, suggesting it is older than ~ 1.8 ka (Kawanabe 1991;Ishizuka et al. 2014).Their geochemical data were reported by Ishizuka et al. (2014).Samples from each ridge revealed similar compositions, arguing

Methods
To infer magnetic mineralogy, we measured the temperature dependence of magnetic susceptibility k(T).Measurements were conducted using the KLY-4 with a furnace CS-3 (AGICO).Powdered samples were heated in flowing Ar up to 700 °C and cooled to 50 °C.Selected samples were also measured in air.Additional measurements were also conducted after the paleointensity experiments.We also measured magnetic hysteresis using a vibrating sample magnetometer (VSM) Model 29/3902 (Princeton measurements) from − 1 to + 1 T. Coercivity of remanence was determined by a backfield demagnetization of isothermal remanence acquired at 1 T. We used the Tsunakawa-Shaw paleointensity method (Yamamoto et al. 2003) which build upon the Shaw method (Shaw 1974).The method compares the intensity (equivalent moment) of natural remanent magnetization (NRM) and laboratory TRM during alternating field (AF) demagnetizations.Thermal alteration of samples is checked and corrected using the AF demagnetization of anhysteretic remanent magnetization (ARM) before and after heating (Rolph and Shaw 1985).When calculating magnetic moment of each remanence, the remanence vector after the highest AF demagnetization step is subtracted to eliminate the effect of any remaining high coercivity component.The moment of the two ARMs is referred to as M ARM0 and M ARM1 , respectively.An alteration correction to the moment of the laboratory TRM (M TRM1 ) is done at each AF demagnetization step as follows: The paleofield intensity (B anc ) is estimated as follows: where slope NRM-TRM1* is the gradient from plots of the moment of NRM (M NRM ) and M TRM1 * , and B lab is the laboratory field intensity.
The validity of the correction is checked by imparting another set of TRM and ARM (Tsunakawa and Shaw 1994), whose intensities are labeled as M TRM2 and M ARM2 , respectively.M TRM2 is corrected similarly to (1) as A large deviation from unity in the slope of the M TRM1 -M TRM2 * plot would indicate that the correction is invalid for the corresponding specimen. (1) Samples were cut into 3 or 4 mutually oriented cube specimens of around 1 × 1 × 1 cm 3 size.We could not cut out cubes from the sample HPD#997-R08; instead, we used unoriented small chips mounted on a glass plate with OMEGA CC high-temperature cement.
Paleointensity experiments were conducted in Japan Agency for Marine-Earth Science and Technology.We use spinner magnetometers ASPIN (Natsuhara Giken Co., Ltd.) and JR-6A (AGICO) and cryogenic magnetometers (2G Enterprises).The moment resolutions are around 10 -8 Am 2 , 10 -10 Am 2 , and 10 -11 Am 2 , respectively.AF demagnetization was conducted progressively up to a peak AF of 150 mT using a two-axis sample tumbling demagnetizer DEM-95 (Natsuhara Giken Co., Ltd.).Before a progressive AF demagnetization, we applied a low-temperature demagnetization by putting samples in liquid nitrogen in a magnetic shield to reduce remanence carried by multidomain magnetite (Yamamoto et al. 2003).We imparted TRMs with an applied field of 50 μT using a thermal demagnetization oven TDS-1 (Natsuhara Giken Co., Ltd.).Heating was conducted in a vacuum (< 10 Pa), which would minimize alteration during heating for some samples (Mochizuki et al. 2004).We chose 50 μT to mimic the expected paleointensity for the recent past in Japan (Kitahara et al. 2018(Kitahara et al. , 2021)).The maximum temperature was set as 500 °C based on the susceptibility measurements, as detailed later.The holding time was ~ 15 min for the first heating and ~ 1 h for the second.We imparted ARMs using DEM-95 with a DC bias field of 70 μT.The DC field intensity was chosen so that ARM and TRM become comparable magnitudes based on pilot specimens.
We filtered the paleointensity estimates by the following criteria, which are similar to those of Ahn and Yamamoto (2019): NRM by a progressive AF demagnetization on the orthogonal vector plot.) is between 0.95 and 1.05.The ChRM vector was calculated by principal component analysis (PCA) (Kirschvink 1980).

Rock magnetism
Figure 2 shows k(T) curves.All examined samples heated in Ar show a susceptibility drop at around 300 °C with nearly reversible behavior.Samples HPD#997-R08 and HPD#1004-R06 show a broader reduction in magnetic susceptibility extending to ~ 450 °C.From visual inspections, we interpret that the Curie temperatures of the samples are roughly 300-450 °C.In contrast, samples heated in air reveal highly irreversible curves (Fig. 2g).Based on these results, we performed Tsunakawa-Shaw experiments with a maximum temperature of 500 °C in a vacuum to impart full TRM.The heating curves for samples after the paleointensity experiments show a small contribution of phase(s) with a Curie temperature up to ~ 510 °C (Fig. 2h).This contribution disappears when heated in Ar to 700 °C.
Hysteresis parameters are summarized on a Day plot (Day et al. 1977;Dunlop 2002) in Fig. 3.Most samples are within the "pseudo-single domain" region to the singledomain region, suggesting a relatively small grain size of magnetic minerals.(Day et al. 1997;Dunlop 2002).MD stands for multidomain, PSD for pseudo single domain, and SD for single domain

Paleointensity
Each specimen reveals near single-component of NRM during AF demagnetization; however, close examination often reveals a small but persistent overprint demagnetized by fields up to 15-40 mT (Fig. 4a, b).Stability against AF demagnetization is variable, with the median destructive field ranging from 14 to 67 mT (Table 2).Except for the mutually unoriented specimen of HPD#997-R08, the specimens exhibit internally consistent directions for each samples (Fig. 4c).
Representative paleointensity results are shown in Figs. 5, 6.The M NRM -M TRM1 plot is mostly straight.ARMs after each heating also reveal minimal alterations, as indicated by high linearity and slopes around 1 in the M ARM1 -M ARM0 and M ARM2 -M ARM1 plots (Table 3).Thirteen out of 22 specimens pass the selection criteria.Of nine failed specimens, eight specimens showed slope TRM1 -TRM2 * larger than 1, and one revealed R TRM1 -TRM2* below 0.995 (Table 3; Fig. 6).
Ridges exhibit different paleointensity characters.Specimens from two sites on the ridge SE1 show a paleofield intensity of 33-55 μT, while specimens from two sites on the ridge SE3 consistently show a higher value of 58-64 μT (Table 3; Fig. 7).At the sample level, HPD#997-R08 show a large scatter (Table 3; Fig. 7), even though we do not detect a major difference in rock magnetic properties relative to the other samples.Based on these observations, we discard the results from HPD#997-08 and use HPD#997-R09 data as representative paleointensity for SE1, which yield the average paleointensity of 36.7 μT.For SE2, the sample HPD#999-R10 yields a paleointensity of 35-40 μT, while HPD#1004-R06 reveals 54-62 μT.

Magnetic mineralogy
A prerequisite for absolute paleointensity experiments is that NRM is TRM.The linear relationship in the M NRM -M TRM1 plots (Fig. 5) suggests that NRM has similar coercivity distribution with TRM.Additional lines of evidence were obtained from rock magnetic results.Basaltic rocks may undergo low-and high-temperature oxidation, causing secondary thermochemical remanence.Low-temperature oxidation would produce titanomaghemite below its Curie temperature.In our samples, the reversible k(T) curves and minimal changes in ARMs during paleointensity experiments argue against a significant contribution from titanomaghemite, as this mineral would invert into hematite or magnetite upon heating.This observation suggests that the samples examined here preserve TRM acquired during initial cooling.Previous paleomagnetic studies of subaerial lavas from Izu-Oshima Island reported two Curie temperatures of 500-600 °C and 250-350 °C (Tanaka 1982;Yoshihara et al. 2003;Mochizuki et al. 2004).Based on microscopic observations, Mochizuki et al. (2004) suggested that these Curie temperatures correspond to Ti-poor and Ti-rich titanomagnetite, respectively, and that the Ti-poor titanomagnetite originates from high-temperature oxidation of the Ti-rich titanomagnetite.Such Ti-poor titanomagnetite sometimes acquires thermochemical remanence, which may bias paleointensity estimates (Yamamoto et al. 2003;Mochizuki et al. 2004;Fabian 2009).In contrast, our submarine basalts only show the lower Curie temperature, suggesting the lack of high-temperature oxidation of titanomagnetite.On the basis of the absence of significant low-and high-temperature oxidation, we consider that the NRMs measured here mainly reflect TRMs, even though we cannot completely rule out some contribution from thermochemical remanence.Heating in a vacuum at 500 °C for ~ 1 h produces some phase(s) with Curie temperature of ~ 510 °C (Fig. 2h).This behavior may reflect slight oxidation of titanomagnetite to titanomaghemite during the experiment.Titanomaghemite may be reduced back to titanomagnetite by small impurity in Ar gas during the subsequent susceptibility measurements.On the contrary, simple oxidation does not explain the result that the new Curie temperature is different from that of the samples heated in air (490-500 °C; Fig. 2g).A possible mechanism to explain these observations is nanoscale chemical clustering (Jackson and Bowles 2018;Bowles et al. 2019).It is reported that some titanomagnetites exhibit Curie temperature enhancement after annealing.However, the reported annealing temperatures tend to be lower than those of our experiments.It is also difficult to account for the fact that the bulk of the samples keep the original Curie temperature.Regardless of the origin, the success in the Tsunakawa-Shaw experiments, in particular the high linearity of M NRM -M TRM1 * plots, suggests that the high Curie temperature phase(s) does not contribute significantly to the remanence.

Age constraints
To obtain age constraints from our paleointensity data, we compare them with regional compilations from the GEOMAGIA50.v3database (Brown et al. 2015) as well as global field models (Fig. 8).From the GEOMAGIA50.v3, we select data from Japan and South Korea.The data selections are based on the following four criteria: (1) the paleointensity experiments should have accompanying alteration checks, (2) the number of accepted paleointensity results should be 3 or larger, (3) the standard deviation of the paleointensity should be smaller than 10 μT, and (4) the uncertainty of the age should be smaller than 50 yr (Fig. 8a).We also filter out the data with no error estimates for either age or paleointensity.These criteria return 29 records (Yoshihara et al. 2003;Mochizuki et al. 2004;Yamamoto and Hoshi 2008;Yu 2012;Hong et al. 2013;Kitahara et al. 2018Kitahara et al. , 2021)).Besides, we also incorporated recent archeointensity data from Tema et al. (2023) which meet the above criteria.Data were converted into virtual axial dipole moment (VADM) using the site locations (Table 4).
The SE1 samples indicate relatively low paleointensity values of ~ 37 μT (Table 3, Fig. 7).At the same time, their sparsely phyric petrography suggests that they are likely to be younger than 1.8 ka (Kawanabe 1991(Kawanabe , 2012)).This petrologic constraint is also consistent with the  nearly pure frontal-arc composition, which may occur either < 2 ka or > 10 ka (Ishizuka et al. 2015).Previous paleointensity data from Japan and Korea indicate higher paleointensity between 2 and 1.5 ka and between 0.9 and 0.6 ka, respectively (Hong et al. 2013;Yoshihara et al. 2003;Kitahara et al. 2018Kitahara et al. , 2021) ) (Fig. 8a).Thus, our paleointensity data narrow down the possible ages of SE1 to be either younger than 0.5 ka, or around 1.4 ka.The two samples from the SE2 ridge do not show coherent paleointensities.While these samples show slight difference in Curie temperature and hysteresis parameters (Figs. 2, 3), there is no clear correlation between these rock magnetic properties and paleointensity in the studied samples.Although the number of specimens that pass the selection criteria is not high for SE2, the nominal paleointensities from specimens that failed the criteria show a similar trend (Fig. 7).In contrast to SE1 and SE3, the two samples are obtained from separate sites.We attribute the difference to a change in geomagnetic field between the eruptions, implying that the SE2 ridge consists of products of more than one eruption event.
The sparsely phyric petrography and frontal-arc chemical composition point to relatively young (< 1.8 ka) ages for both samples.The sampling site of HPD#1004-R06 locates on the extension of the subaerial Y4 fissure (Fig. 1), whose activity was correlated to a written record at 1421 CE (Kawanabe 1998).Because paleointensity of Y4 lavas (60.6 ± 6.0 μT; Yoshihara et al. 2003) is comparable to that of HPD#1004-R06 (57.3 and 62.2 μT from two specimens), we speculate that HPD#1004-R06 may be simultaneous to the Y4 eruption at 1421 CE.The paleointensity estimated from HPD#999-R10 is close to the SE1 average, which may be correlated to either < 0.5 ka or around 1.4 ka (Fig. 8a).
The global field models reveal possible complications in the age constraints.Here, we consider SHA.DIF.14 k (Pavón-Carrasco et al. 2014) and CALS3k.4(Korte and ConsTable 2011).The important difference between these two models is that the former is based on volcanic and archeological data, while the latter utilizes sedimentary data as well.For the Izu-Oshima location, these two models predict distinct field behavior between 1.8 and 0.4 ka (Fig. 8b).Generally, SHA.DIF.14 k model agrees better with the paleointensity data; however, it overestimates the paleointensity between ~ 1.0 and 1.5 ka.The data for this period represent the most recent archeointensity studies (Kitahara et al. 2018, 2021, Tema et al. 2023). Tema et al. (2023) proposed that the younger paleointensity data may be biased toward high values, because they are obtained mainly by the Coe-Thellier method (Coe 1967), which may be insufficient to exclude the multidomain contributions.Consequently, they argue that CALS3k.4model may be more reliable.If this is the case, paleointensity around Japan does not change much during the last 1.8 ka, and the resolving power of paleointensity as a dating tool would be limited.At the same time, some of the South Korean data (Hong et al. 2013) obtained by the IZZI-Thellier method (Yu et al. 2004) as well as our two paleointensities obtained from the sample HPD#1004-R06 are higher than the prediction of CALS3k.4.Reexamination of submarine and subaerial Japanese volcanic rocks is necessary to resolve this issue.

Conclusions
Submarine basaltic lavas from an island-arc volcano provide geomagnetic paleointensity records.Magnetic minerals are Ti-rich titanomagnetites with Curie temperatures of 300-450 °C without detectable low-or high-temperature oxidation.The Tsunakawa-Shaw paleointensity method with vacuum heating is effective in minimizing laboratory alteration during paleointensity experiments in our samples.Two ridges show distinct paleointensity estimates, while another ridge shows lower and higher values.Compared with regional paleosecular variation data, our paleointensity data are consistent with petrographical and geochemical age constraints and narrow the possible age range.For better age constraints, refinement of reference paleointensity data, particularly before 2 ka, is necessary.Our results demonstrate that basaltic rocks from submarine arc volcanoes can retain primary TRM without significant thermochemical overprint for at least a few thousand years.Systematic paleomagnetic studies should be part of future investigations of submarine volcanoes.

Fig. 1
Fig. 1 Sample locations with seafloor bathymetry and simplified subaerial geological map modified from Ishizuka et al. (2014).Bathymetry was obtained by a high-resolution multi-narrow beam echo-sounder survey.Major bathymetry contours are shown in black and 100 m apart, while gray contours are spaced 20 m.Color lines indicate submarine ridges identified by Ishizuka et al. (2014) based on bathymetry.Yellow stars show sample locations.Eruptive fissures on Izu-Oshima Island are shown by black lines with eruption stage names (Kawanabe 1998).Different colors on Izu-Oshima Island indicate the distribution of lavas at different stages.Inset shows the location of Izu-Oshima Island 2. On the M NRM -M TRM1 * diagram, a linear segment exists within the coercivity range defining the ChRM.The segment should cover at least 30% of the total NRM intensity and should have a correlation coefficient (R NRM -TRM1* ) larger than 0.995.3. On the M TRM1 -M TRM2 * diagram, a linear segment with a correlation coefficient (R TRM1 -TRM2* ) larger than 0.995 is recognized in the coercivity range covering that of the linear segment on the M NRM -M TRM1 * diagram.The high coercivity end of the linear segment on the M TRM1 -M TRM2 * diagram coincides with that of the linear segment on the M NRM -M TRM1 * diagram.The slope of the linear segment on the M TRM1 -M TRM2 * diagram (slope TRM1 -TRM2*

Fig. 2 Fig. 3
Fig.2Temperature dependence of magnetic susceptibility (k).Data are normalized by the room temperature susceptibility before heating (k 0 ).Orange solid lines show heating curve, and blue dashed lines show cooling curve.a-f Heating in Ar. g Heating in air for the sample HPD#1000-R09.h Heating in Ar after paleointensity experiment involving heating to 500 °C in vacuum for the sample HPD#1000-R10

Fig. 4
Fig.4NRM demagnetizations.Orthogonal vector plots for samples a HPD#997-R09 and b HPD#1000-R09.Black and red circles represent projections onto the horizontal plane, and white and blue circles to the vertical plane.Colored symbols show steps used to identify ChRM.Dotted lines are the ChRM directions.c ChRM direction plotted by equal area projection.Open circles show specimen data.Orange circles show mean directions, with α 95 indicated by dashed ovals.Note that the samples were not geographically oriented, and the direction of each sample is arbitrary.Plotted with MagePlot(Hatakeyama 2018)

Fig. 5
Fig. 5 Example of paleointensity results that met the criteria (HPD#997-R09-01).a M NRM -M TRM1 diagram.b M ARM0 -M ARM1 diagram.c M TRM1 -M TRM2 diagram.d M ARM1 -M ARM2 diagram.Gray triangles in a and c show M TRM1 and M TRM2 before the correction, while circles indicate the corrected TRM intensities (see text for details).Red circles show data used to calculate paleointensity

Fig. 6
Fig. 6 Example of rejected paleointensity results (HPD#1000-R09-04).a M NRM -M TRM1 diagram.b M ARM0 -M ARM1 diagram.c M TRM1 -M TRM2 diagram.d M ARM1 -M ARM2 diagram.Gray triangles in a and c show M TRM1 and M TRM2 before the correction, while circles indicate the corrected TRM intensities (see text for details).This specimen failed because the slope in c is larger than 1.05

Fig. 7 Fig. 8
Fig. 7 Summary of the paleointensity results.Solid small circles show data that passed the selection criteria, and gray crosses show data that failed

Table 2
NRM demagnetization dataL lower field end for the PCA calculation.H higher field end for the PCA calculation, MAD maximum angular deviation, MDF median destructive field