Constraints on the Timing and Lower Crustal Accretion at the Schulz Massif, Mohns Ridge, Arctic Mid Ocean Ridges

The ultraslow‐spreading Mohns Ridge is a key supersegment of the Arctic Mid‐Ocean system, where it represents a boundary between the Jan Mayen hotspot in the south and the highly anomalous Knipovich Ridge to the north. Understanding the timing and mode of Plio‐Pleistocene seafloor spreading along this ridge segment is critical for establishing the recent geodynamic evolution of the Norwegian‐Greenland Sea. To investigate magmatic accretion at this ultraslow spreading ridge, we collected samples from the Schulz Massif, which is located off‐axis at 73.4°N and exposes gabbroic intrusives and mantle peridotite. The petrology and petrography of these samples indicate that the exposed crustal section underwent multiple episodes of magmatism, which are characterized by distinct crystal sizes and geochemistry. To calibrate the age of the seafloor, we combined high‐resolution and high‐precision single zircon U‐Pb geochronology. Our data suggest that seafloor spreading has been nearly symmetrical for the last ∼4.6 Myr with a time‐averaged half‐spreading rate of ∼7.4 mm yr−1. Crystal size analysis of olivine in porphyric intrusions suggests that the crustal section was fed crystal‐laden melts with recurrence rates predicted to stabilize fault‐dominated seafloor spreading. Our combined geochronological and crystal size approach gives a critical perspective on the mode of seafloor spreading in the Mohns Ridge and allows insights into accretionary mechanisms and crustal structures during symmetric seafloor spreading.


Introduction
The divergent margin between the Eurasian and North American plates is composed of several ridge segments with unique spreading characteristics and has been a target of interest since they were first discovered in the 60 s (Johnson & Heezen, 1967).Rifting and separation began in the early Eocene and are characterized by a complex evolution involving variations in spreading direction and rate, oblique spreading, and ridge jumps (Talwani & Eldholm, 1977;Vogt et al., 1982).The recent evolution of the Mohns Ridge is of special interest because of its very slow spreading rate, strongly asymmetrical mountain topography, highly depleted mantle domain, and close connection with the anomalous Knipovich Ridge (Bruvoll et al., 2009;Johansen et al., 2019;Pedersen, Thorseth, et al., 2010;Sanfilippo et al., 2021).An accurate determination of the age of the ocean crust, spreading rates, and details about magmatic construction could provide more insight into the complex evolutionary history of this area.
The age of the ocean floor is typically determined using seafloor magnetic anomalies and plate models which are suitable for understanding the long-term evolution of ocean basins.However, this approach is not straightforward at the Mohns Ridge where magnetic anomalies within about 70 km of the present-day ridge axis are poorly constrained, offering an age resolution between 5 and 10 million years, too low to identify changes in seafloor parameters and dynamics occurring on smaller temporal scales (Gaina et al., 2002;Mosar et al., 2002;Müller et al., 2016).Another approach to determine the age of the ocean floor is the U-Pb dating of zircon (ZrSiO 4 ).Gabbroic rocks containing zircon have been documented on the ocean floor at slow-spreading ridges where the spreading regimes allow for the uplift of lower ocean crust in tectonic windows and oceanic core complexes.Consequently, U-Pb dating of zircon grains from slow (Baines et al., 2008;Grimes et al., 2008Grimes et al., , 2009;;John et al., 2004;Lissenberg et al., 2009Lissenberg et al., , 2016;;Rioux et al., 2015Rioux et al., , 2016;;Schoolmeesters et al., 2012;Schwartz et al., 2005;Tani et al., 2011), and fast (Hayman et al., 2019;Rioux et al., 2012) mid-ocean ridges have provided the best constraints on the age of the ocean floor.In addition, the use of U-Pb and trace element chemistry of zircon can give insights into magmatic processes (e.g., temperatures, chemical composition, redox conditions), and the mode of crustal accretion and uplift of lower crustal sequences (e.g., timescales of crystallization, cooling rates).
In a previous effort (Bjerga, Stubseid, Pedersen, & Pedersen, 2022), we dated a set of zircon grains using LA-ICP-MS to determine whether the core complex had inherited continental zircon.However, a more precise determination of the crystallization age of the magmatic section requires the addition of several techniques, such as secondary ion mass spectrometry (SIMS) and isotope dilution thermal ionization mass spectrometry (ID-TIMS) (Schaltegger et al., 2015).By using zircon geochronology with precise ages and known crystallization locations, it is possible to determine time-averaged spreading rates, leading to a better understanding of spreading dynamics.Calculating spreading rates using U-Pb dating, however, relies on several assumptions.The first assumption is that the age of the samples is accurately determined, which is controlled by the accuracy of the analytical methods and any potential post-spreading alteration that has affected the U-Pb dates.The second assumption is that the samples are representative of the overall spreading rate and that the samples have not been transported from another location.The third assumption is that the spreading rate has been constant over time.This implies that there have been no major tectonic events that have affected the position of the samples.
To address questions about the timing and mode of spreading along the Mohns Ridge, we used a remotely operated vehicle (ROV) to sample an off-axis oceanic core complex constituted by a gabbro-peridotite association called the Schulz Massif located at 73.4°N (Figure 1).We successfully recovered zircon grains from two gabbroic samples and dated them with U-Pb geochronology using ID-TIMS and SIMS.We used zircon trace element compositions and O-isotopes, together with the geochemistry of the gabbroic lithologies, to infer the origin and magmatic evolution of the exposed gabbroic crust.Quantitative crystal size distribution (CSD) analysis of olivine in olivine-phyric gabbros provides information on the subvolcanic plumbing system, allows us to speculate on the timescales of magma recurrence rates, and link the construction of larger intrusive bodies with those of smaller individual intrusions.Our data and observations suggest that much of the lower ocean crust now exposed at the Schulz Massif formed through repeated injections of variably crystal-laden melts.

Geology of the Schulz Massif and Samples
The Schulz Massif (73.4°N) is a bathymetric high located approximately 20 km off-axis at the intersection between the slow-spreading Mohns and Knipovich Ridges (Figure 1).The summit rises approximately 600 m below sea level and up to 2,000 m above the surrounding seafloor.It has a fault slope of approximately 13°toward the east, which is composed of soft sediments and local rock outcrops.The fault surface terminates in a sedimentary basin with ∼800 m of sediment (Bruvoll et al., 2009).The distance from the breakaway in the west to the termination of the fault surface in the east is approximately 8 km.The massif exhibits a cross-section from lava to mantle rocks, and it was proposed to represent an oceanic core complex locally dislocated by later faulting (Pedersen, Rapp, et al., 2010;Pedersen, Thorseth, et al., 2010).
Investigations of the massif and surrounding area with the research vessel G.O. Sars were performed in 2008, 2009, and 2016 and include a series of ROV dives and dredges, which have also been described in earlier efforts (Bjerga, Stubseid, Pedersen, Beinlich, & Pedersen, 2022;Bjerga, Stubseid, Pedersen, & Pedersen, 2022; Figure 1).Mantle peridotites and basalts were recovered from several locations of the massif and their distribution and characteristics have been described in Bjerga, Stubseid, Pedersen, Beinlich, and Pedersen (2022).Dive GS16-ROV6 was deployed at the fault termination and we visited three different locations, spaced between 3 and 4 km apart, where we transected upwards from the foot of the massif.Large parts of the area are covered by fine-grained sediment and scattered debris.At all three locations visited during dive GS16-ROV6 mantle peridotites and gabbros were recovered from within a few meters of each other (Figures 1 and 2).Some samples were collected from loose rocks while others were taken from exposed bedrock (Figures S1-S4 in Supporting Information S1 shows photos from representative sample sites).Although not all samples come from in situ bedrock, it seems unlikely that they have been transported far.The exposure of bedrock occurs at several locations, but manganese crust makes it difficult to see any structures.Near the base of site 2 (Figure 1), a large sample of peridotite (GS16A-ROV6-004) crosscut by gabbroic veins and a mafic dike was collected (Figure 2).Close by, we recovered gabbroic rock sample GS16A-ROV06-006 from a planar feature that dips into the sediment.At location 3, two coarse-grained gabbros were recovered together with a microgabbro containing abundant sulfides, most likely indicating that high-temperature hydrothermal fluids have percolated through the fault.
A dredge haul (GS08-DR1) crossed several eastward dipping fault scarps and recovered fragments of serpentinized peridotite, gabbro, and mafic dikes (Figure 2).Gabbros are intruded by mafic dikes and show cataclastic to mylonitic structures.Cataclastic textures (Figure 3) are common in the recovered gabbros, suggesting that brittle deformation followed the high-temperature shearing leading to the uplift of the crustal section.All recovered peridotites from the dredge are undeformed and display mesh textures after olivine, except for one sample which appears heavily deformed (Bjerga, Stubseid, Pedersen, Beinlich, & Pedersen, 2022).
During dive GS16A-ROV02, we crossed a fresh-looking talus to reach a fault scarp exposing massive bedrock at ∼2,100 mbsl (meter below sea level).Here, several olivine-phyric gabbros (Figure 2) were recovered from 2,200 to 2,300 mbsl (Figure S4 in Supporting Information S1).A piece of microgabbro from the end of transect GS16-ROV3, sample GS16-ROV3-005, has abundant sulfides, suggesting that high-temperature hydrothermal fluids have also percolated through this fault.The transect GS16-ROV4 passes over a gently sloping, sedimented surface which separates two ridge parallel structures interpreted as normal faults, into a more steeply dipping talus ramp toward the top of the Massif (Figure 1).A poorly sorted, matrix-supported breccia (GS16A-ROV4-001) with highly vesicular basaltic clasts up to 10 cm in size was sampled.In-place lavas are exposed near the summit of the breakaway where the average slope is about 35°.Aphyric basaltic sample GS16A-ROV4-003 and highly vesicular basalt sample GS16A-ROV4-004 were taken from a steeply dipping wall.
Dive GS07-ROV9, approximately ∼800 m from the top of the breakaway on the NE face, found primarily gabbroic lithologies, whereas GS07-ROV10 recovered pillow basalts and mafic dikes.A prominent topographic high directly east of the massif, where a sheeted dike complex was observed in a steeply dipping wall during ROV dive GS08-ROV6, is interpreted as a hanging wall above the massif.Based on the distribution of samples and geological observations, it appears that the mantle peridotites were intruded by gabbros and mafic dikes at various stages of the evolution of the massif.

Petrography and Mineral Chemistry
Based on cross-cutting relationships, texture (grain size), and chemical compositions, we identified three categories of magmatic rocks exposed at the Schulz Massif.The first category, G1, is coarse-grained (∼1 cm grains) plagioclase + clinopyroxene gabbro.Olivine is locally present as small grains within plagioclase (Figure 4a).Amphibole and chlorite have replaced much of the primary magmatic pyroxene and primary calcic plagioclase has been partially albitized.The primary magmatic textures are commonly obliterated by deformation and grain size reduction (Figure 3).Medium-grained (up to c. 0.7 cm grains) olivine-phyric gabbros make up the second category G2 of magmatic lithologies (Figure 4b; Figures S5 and S6 in Supporting Information S1).Subhedral to euhedral plagioclase laths (up to 0.2 cm grains) and pyroxene are present in the finer-grained matrix between porphyric olivine together with Fe-Ti oxides.Finally, microgabbros with grain sizes <2 mm and dolerites mark the third category, G3 (Figure 4c).

Whole-Rock Geochemistry
The whole-rock geochemical data are available in Table S2.Coarse-grained gabbros have SiO 2 contents between 49.9 and 51.9 wt.% and Al 2 O 3 between 15.7 and 17.6 wt.%.Both CaO and Al 2 O 3 correlated negatively with MgO content, corresponding to the accumulation of Ca-rich plagioclase.They have low concentrations of Zr and Hf similar to common gabbroic lithologies from ODP Hole 735B (Figure 5).Two types of chondrite normalized REE-patterns are present (Figure 6) (a) typical cumulate with low La n /Sm n (0.38-0.46) and positive Euanomalies, and (b) MORB-like with La n /Sm n (0.72) and slightly negative Eu-anomaly.In primitive-mantle normalized trace element diagrams, they display positive K, Sr, and Pb anomalies and negative Th and Nb anomalies (Figure 7). 87Sr/ 86 Sr ranges from 0.7036 to 0.7037 and is similar to basalts from the Schulz Massif (Bjerga, Stubseid, Pedersen, Beinlich, & Pedersen, 2022).

U-Pb Geochronology
The isotope dilution-thermal ionization mass spectrometry (ID-TIMS) methods were applied to five zircon grains from sample GS16A-ROV6-016, but only two of them yielded results.The two successful grains show variable U contents (172 and 26 ppm) and Th/U (1.43 and 0.13) but relatively low precision due to the small amount of Pb available and the low 206 Pb/ 204 Pb ratios (31.5 and 23.4).They yielded uncorrected 206 Pb/ 238 U dates of 4.97 ± 0.08 and 4.66 ± 0.16 Ma (Figure 8a: Table S3).Given the young age, low U concentrations, and low degree of radiation damage, it is unlikely that substantial Pb-loss affected the zircon.
Five grains from GS16A-ROV6-016 from 85 to 275 μm in length were analyzed by SIMS and yielded uncorrected 206 Pb/ 238 U dates between 3.89 ± 1.59 and 4.86 ± 0.75 Ma (Figure 8b: Table S4) with a mean weighted average 206 Pb/ 238 U age of 4.54 ± 0.13 Ma (MSWD = 0.68, n = 7).Cathodoluminescence imaging shows that most grains exhibit either oscillatory or sector zoning (Figure 9), a characteristic feature of magmatic zircon.The concentration of Pb is very low, meaning that the ages are sensitive to Pb-corrections.The grains have between 25 and 434 ppm of U, and between 16 and 328 ppm of Th, with Th/U ranging between 0.56 and 1.04.

Zircon Trace Element Geochemistry and Ti-In Zircon Temperature
Analyzed zircon grains show a range of trace element compositions comparable to zircon from continental settings (Figure 10; Table S5).In chondrite normalized REE diagrams, they display negative Eu and positive Ce anomalies with Eu/Eu* (Eu N /√[(Sm N × Gd N )]) and Ce/Ce* (Ce N /√[(La N × Pr N )]) of 0.08-0.16and 8.7-185.2,respectively.All the grain plot in the "magmatic field" of zircon (i.e., (Sm/La) N > 100 and La < 1 ppm, Hoskin, 2005), indicating that they have not been significantly impacted by hydrothermal alteration.Two grains have distinctly higher U/Yb ratios of 0.6 and 2.8 that plot close to arc-like zircon (Figure 10).Ti concentrations range from 22 to 92 ppm yielding uncorrected Ti-in zircon temperatures (Ferry & Watson, 2007) between 816 and 983°C and a mean of 892°C.

Textural Quantification Using Crystal Size Distribution
Data from three olivine gabbros with olivine abundances ranging from 38.3% to 52.2% are presented in Table S6 and have been used to generate CSD plots.The CSDs are similar, with minor differences in size and curvature (Figure 11) with shallow slopes ranging from 0.674 to 0.775 and maximum crystal sizes (L max : calculated as the average of the four largest grains of each sample) of 7.4-7.9mm.Complex crystallization histories and mixing of crystal populations result in kinked and strongly curved CSDs.The approximately linear CSDs for larger grain sizes here suggest a simpler crystallization history.We, therefore, propose that the similarities in the shape of the three CSD curves for larger grain sizes reflect similar crystallization and storage histories for the three studied samples.Variations in the slope rate may be related to growth time (residence), growth rate, accumulation of crystals, or textural coarsening (Higgins, 2006).Higgins (2006) showed that increased growth rates and residence time favor an increase in larger crystals and thus lead to a progressive shallowing of the slope but not the intercept.Boorman et al. (2004) demonstrated that the modification of cumulus textures may occur during compaction-driven recrystallization and crystal aging.There is no evidence for compaction-related processes in the studied olivine-phyric gabbros, and it is unlikely that the samples experienced any significant post-cumulus processes that would lead to complex crystal morphology.We conclude that the granular olivine represents the primary magmatic texture, unmodified by subsequent recrystallization and/or textural equilibration.As a first-order, we therefore suggest that the use of the largest size segments yields reasonable results regarding nucleation and crystal growth.
We can estimate the maximum timescale of crystal growth by combining the L max with an estimated growth rate.However, precisely defining the growth rate of crystals is in most cases not possible and it thus represents a considerable source of uncertainty.Based on previously published growth rates which range from 10 3 to 10 10 mm/s, we consider intermediate growth rates of 1 × 10 6 -1 × 10 8 mm/s to be a realistic estimate of the average growth rate of the granular olivine.By choosing a maximum growth rate of 1 × 10 7 mm/s combined with the largest observed crystal size (7.9 mm), we find a lower limit on the timescale of magma residence for the largest crystals to be 916 days (Figure 11).An alternative approach is to estimate the crystallization time of a population of crystals from the slope of the CSD (Marsh, 1998).By combining the slope values and growth rate of 1 × 10 7 mm/s, we found a characteristic residence time of olivine grains between 191 and 209 days (Figure 11, Table S6).
The time constraints determined by crystal sizes have weaknesses and uncertainties that must be discussed.Specifically, the estimates assume a constant crystal growth rate and a steady-state system.This is less likely to be achieved for intrusions in magma chambers than for extrusive rocks.The granular morphology of olivine is typically interpreted as reflecting slower cooling rates at low degrees of undercooling under near-equilibrium conditions (Donaldson, 1976;Faure et al., 2003).However, this view is challenged by Welsch et al. (2014), who, based on phosphorus zoning in olivine crystals from a range of environments, suggested that early olivine growth may be initiated as skeletal morphology which is progressively filled in with time.This suggests that the growth of polyhedral olivine may be more rapid than previously expected.Experimental data by Mourey and Shea (2019) showed growth rates up to ∼10 3 mm/s for low-to-moderate degrees of undercooling (25-60°C).Also, experiments and models suggest that the growth rate may vary over time and eventually slows down as the degree of undercooling decreases and crystallinity increases (e.g., Hammer, 2008;Huber et al., 2009).Additionally, because of the possible influence of coarsening or crystal accumulation, these values should be considered as maximum values.Finally, the dissolution of the olivine crystals would cause crystals not to record the complete crystallization history.

When and How Does Zircon Form in Slow-Spreading Environments
Although we were not able to locate zircon in the thin section, the textures and internal zonation of zircon grains (Figure 9) are typical of magmatic growth.The age of the grains combined with isotopic and trace-element data indicates that they formed from mantle-derived basaltic melts during crustal accretion.According to Boehnke et al. (2013), the saturation of zircon in basaltic magmas at temperatures higher than 1,000°C requires >5,000 ppm Zr, which is over 1,000 times more than the content of the sample GS16A-ROV06-016.The presence of zircon in the coarse-grained gabbroic sample GS16A-ROV6-016 derived from mafic melts, therefore, warrants a discussion on how and when they form.The sample averaged temperature of 892 ± 142°C (2SD) recorded by the zircon using a Ti-in zircon thermometer (Ferry & Watson, 2007) is in the upper range of that seen in magmatic ocean zircon (Grimes et al., 2009) and within the range of solidus temperatures of 860 ± 30°C estimated from Mid-Atlantic Ridge gabbros using the amphibole-plagioclase thermometer (Coogan et al., 2001).The saturation and crystallization of zircon are controlled primarily by the temperature and melt composition.Specifically, melts with a high content of tetrahedron-forming elements (e.g., Si) and low temperatures favor crystallization.For instance, Borisov and Aranovich (2019) derived an empirical equation for Zr solubility and used the data from Nandedkar et al. (2014) to infer that zircon could crystallize at Zr content ∼100 ppm near solidus conditions, which are slightly above 700°C.Importantly, they showed that while the crystallization of zircon in dry basaltic melts is very unlikely, the crystallization in evolved hydrous basaltic melts is plausible.At 830°C, the expected Zr saturation is anticipated at 400 ppm, while at 750°C it is expected to be at 215 ppm (Borisov & Aranovich, 2019).The presence of water effectively accomplishes two things, firstly, it lowers the solidus temperature of the melt, and second, it increases the SiO 2 content, both favorable for the crystallization of zircon.The H 2 O content of analyzed basalts (n = 67) from the Mohns Ridge is high and varies between 0.2 and 2.2 wt.%, with a median of 0.6 wt.% (Dixon et al., 2017;Kelley et al., 2013;Michael, 1995;Neumann & Schilling, 1984).This suggests that the melts feeding the crustal sections along the Mohns Ridge could contain significant volatiles which may promote crystallization of zircon from interstitial evolved melt pockets (e.g., Borisov & Aranovich, 2019).
Zircon sourced from a depleted mantle has a diagnostic low U/Yb of less than 0.1 (Grimes et al., 2015), which is lower than the mean 0.16 (excluding the two anomalous analyses with U/Yb > 0.6) seen in the Schulz Massif zircon grains.Ocean arc zircon has U/Yb around 0.1, and is plotted in the overlapping field of continental and oceanic zircon in Figure 10 ( Grimes et al., 2015).Fractionation of titanite ± amphibole leads to an increase in U/ Yb and may explain the arc-like U/Yb characteristics.However, the high Ti-concentrations do not support a scenario of amphibole-present fractionation (Grimes et al., 2015).Instead, U/Yb appears to increase with the lowering of Gd/Yb, suggesting that the fractionation of titanite may explain the most extreme values of U/Yb (Grimes et al., 2015).The zircon data support the presence of a unique chemical signature in the mantle beneath Mohns and Knipovich Ridges that may result from it being composed of high proportions of depleted ancient lithologies mixed with recycled crustal components (e.g., Sanfilippo et al., 2021) or a subduction zone influence (e.g., Yang et al., 2021).Although more samples need to be examined, our data suggest that the zircon has inherited characteristics from mantle-derived melts, making it an important petrogenetic tool for differentiating between mantle sources.
The studied zircon grains have variable U and Th concentrations, with four grains displaying high Th/U ratios between 1.13 and 1.43.Variations in the U concentration and Th/U ratios of zircon may result from the growth of zircon in different interstitial domains with a difference in zircon/melt ratios and differences in the concentration of U in the melt.Alternatively, increasing temperatures are also known to cause an increase in zircon U/Th (Kirkland et al., 2015).However, nearly all zircon from slow-spreading ocean crust analyzed to date has much lower Th/U, suggesting the influence of other factors.At the superfast spreading East Pacific Rise, Hayman et al. (2019) suggested that high Th/U developed due to changes in the zircon Th and U partition coefficients due to oxidation of the melt.The inferred high water content of the melt could therefore promote differences in the partition coefficients leading to high Th/U.If changes in the Th/U content were due to the introduction of water from the assimilation of altered crustal rocks or seawater, we would expect to see a lowering of the δ 18 O values.Mean weighted δ 18 O of 5.2 ± 0.04‰ is, however, remarkably similar to the average values of 5.3 ± 0.8‰ seen in mantle-derived melts (Cavosie et al., 2009;Grimes, Ushikubo, et al., 2011;Valley et al., 2005), which together with the trace element characteristics of the analyzed grains indicate a magmatic control on the Th/U.This is also consistent with the observation that Th/U remains unchanged although the ƒO 2 (calculated after Loucks et al. ( 2020)) of zircon ranges from reduced (ΔFMQ 2.90) to slightly oxidized (ΔFMQ + 0.02).Consequently, we interpret that zircon formed locally either in evolved melt pockets or at zircon-saturated mineral-melt interfaces (Bea et al., 2022).In summary, the petrological observations, isotopic constraints, and zircon characteristics all support the idea that the magmatic suite formed in an axial magma chamber beneath the ridge axis.Consequently, the dated zircon grains should yield a representative age of the ocean crust at the Schulz Massif.

Constraining the Timing of Magmatism
In the study area, sea-surface magnetic anomalies and plate spreading models (Müller et al., 2008) together with previous LA-ICP-MS U-Pb geochronology (Bjerga, Stubseid, Pedersen, & Pedersen, 2022) suggest that the samples come from the crust that formed approximately 4 Ma and was transported off-axis at a (half) spreading rate of ∼8 mm/year.The high-precision U-Pb data here allows us to place further constraints on the timing and duration of crustal growth at the Schulz Massif.There are two interpretations for the difference between the Thcorrected high-precision TIMS dates; (a) they represent an analytical scatter about the true crystallization which results in a Th-corrected 206 Pb/ 238 U mean weighted average age of 4.984 ± 0.003 Ma (n = 2, MSWD = 20.0)which would be the most accurate time of crystallization of the crustal section; (b) the ages reflect actual variability in zircon crystallization.The latter is consistent with the prolonged zircon crystallization observed in previous studies of ocean gabbro (Lissenberg et al., 2009;Rioux et al., 2012Rioux et al., , 2015;;Schwartz et al., 2005) and implies that assimilation of preexisting gabbroic rocks is an important process during accretion of slow-spreading crust.However, because the two dates do not overlap within uncertainty, we combine them with a set of SIMS dates.When using high-resolution techniques like SIMS, the youngest analysis cannot be used as an indicator of the last crystallization event unless the grains are chemically treated.Instead, calculating the mean weighted average of multiple young grains provides a more reliable estimate of the sample age (e.g., Schaltegger et al., 2015).SIMS 206 Pb/ 238 U mean weighted average age of 4.54 ± 0.13 Ma (n = 7, MSWD = 0.68) is within the uncertainty of the youngest TIMS date (4.66 ± 0.16 Ma) and suggests that the main crustal forming event occurred ∼4.6 Myr.This is ∼0.5 Myr older than the previously determined age (4.15 ± 0.13 Ma) from LA-ICP-MS analysis from sample GS16A-ROV6-016 (Bjerga, Stubseid, Pedersen, & Pedersen, 2022).
If we know the crystallization age of the rock (4.6 Ma, as the mean of TIMS and SIMS analysis), the distance to the present-day ridge axis, and the depth of crystallization, we can determine the time-integrated spreading rates.Because of the sample location's position 20 km from the present-day rift valley (Figure S10 in Supporting Information S1), we lack accurate constraints on the transport path and consider two potential locations (Figure S10 in Supporting Information S1).In the first scenario, the samples formed directly beneath the present-day spreading axis, which would yield a horizontal travel distance of 29.7 km.In the second scenario, the rocks crystallized closer to the rift flank, yielding a horizontal transport path of 25 km.Na-partitioning in clinopyroxene is strongly pressure-dependent and can, if not affected by post-cumulus modifications (Coogan et al., 2000;Lissenberg & MacLeod, 2017;Sanfilippo et al., 2020), be used to estimate the depth of crystallization (Neave & Putirka, 2017).Calculated pressures of crystallization using Na-content of clinopyroxene from microgabbro GS16A-ROV3-005 range from ∼1.4 to ∼3 kbar, corresponding to depths of ∼4.5 and 10 km, respectively.This is in the range of crystallization depths determined from zircon thermochronology, tectonic reconstructions, and microseismic studies at other slow-spreading ridges (Grimes, Cheadle, et al., 2011;Lissenberg et al., 2016;Schoolmeesters et al., 2012) and the crustal thickness at the central parts of the Mohns Ridge (4 ± 0.5 km, Klingelhöfer et al., 2000).Because the zircon U-Pb dates come from a coarsegrained gabbro that was recovered from a structurally lower level (∼3.5 km from the breakaway) and has clinopyroxene with comparable low Na-content (i.e., Na 2 O ∼ 0.2 wt.%), we interpret that it formed at similar depths.Consequently, we regard the 4.5 km depth estimated from the clinopyroxene-liquid thermobarometer as a reasonable estimate for the depth of crystallization.
Transporting the rock complex from the central rift to its present position with a half spreading rate of 8 mm/year would take 3.13-3.71Ma.However, the transport path for the samples will also be affected by the fault geometry (Grimes et al., 2008), with steeper fault angles causing longer transport distances.While the complete exhumation history of the Schulz Massif remains unknown, the present-day fault dip above the sedimentary basin is ∼13°-16°, with an estimated dip below the surface of around 20°-25° (Bruvoll et al., 2009).However, it is possible that the initial fault was much steeper (∼60°-70°), as is observed in other oceanic core complexes (de Martin et al., 2007;Smith et al., 2006).In the following spreading calculations, we do not attempt to estimate a footwall rotation.The spreading rate calculations have an uncertainty of about 0.20 mm/year per km and around 0.16 mm/ year per 0.1 Ma year age difference.This means that the effect of the exhumation history may potentially add 0.2-0.4mm/year to the estimated spreading rate (∼3-6% difference).
Based on the calculations, assuming crystallization occurred in the central part of the axial volcanic ridge in the present-day rift valley (scenario 1) at a depth of 4.5 km, the 206 Pb/ 238 U dates (4.6 Ma) suggest a half spreading rate of 7.4 mm/year, slightly lower than the 7.8 mm/year estimated for the last 10.3Ma from regional models and seafloor magnetic anomalies (Mosar et al., 2002;Müller et al., 2008), but consistent with spreading rates of 7.4 mm/year for the last 1.3 Ma derived from sedimentary stratigraphy (Bruvoll et al., 2009).Shifting the sample location closer to the rift flank (scenario 2) would cause a significant decrease in the half spreading rate (6.4 mm/ year), requiring substantial spreading asymmetry for ∼5 Ma that is not indicated by the magnetic data (Müller et al., 2008) or sedimentary record (Bruvoll et al., 2009).Accordingly, our data are consistent with axial gabbro accretion beneath the ridge axis at mantle depths, and our U-Pb data confirm the very slow-spreading rates of the Mohns Ridge.It also suggests that the spreading has been nearly symmetrical for at least 5 Myr.Considering a depth of melt emplacement that approximates 4.5 km, and the spatial association with mantle peridotite exposed on the seafloor (Figures 1 and 2; Bjerga, Stubseid, Pedersen, Beinlich, & Pedersen, 2022), it seems likely that the plutonic rocks crystallized in the lithospheric mantle supporting a crustal architecture characterized by discrete gabbroic bodies in peridotite (Cannat, 1996;Coogan, 2013;Schwartz et al., 2005).

Mode of Seafloor Spreading Along the Mohns Ridge
Slow-spreading crust forms from episodic additions of magma and the cooling of crystal mushes (e.g., review by Coogan, 2013).The crystal population in the olivine-phyric gabbroic rocks (Figures 2b, 4b, and 10) may have several origins, and the textural data presented here, therefore, allows an attempt to reconstruct the deep crystallization history of the ultraslow-spreading Mohns Ridge.As summarized by Jerram and Martin (2008), rocks seldom consist solely of crystals that grew during the final emplacement of magma.Olivine in mafic lithologies commonly has two sources: xenocrystic olivine from the mantle and primary olivine crystallizing directly from the melt.Because hydrothermal reactions have completely altered the olivine, we cannot distinguish between different origins as is normally done by a combination of chemical and deformational features (e.g., Vinet & Higgins, 2010).An alternative approach is to use the composition of serpentinite as a proxy for the original olivine composition.During the serpentinization of olivine, excess iron commonly enters secondary minerals such as magnetite, leading to an increase in the Mg# in the serpentine phase.The sparse amounts of magnetite in the sample, however, suggest that the Mg# of serpentine should be very close to the Mg# of the original olivine.Mg# in the studied serpentine varies from Mg# = 0.82 to 0.91 and is closely correlated with the content of Al and Fe because of intergrowth with phyllosilicates.The serpentinite, which is closest in composition to the pure serpentinite composition (i.e., Si/Mg ∼0.67 and Mg/Fe ∼3) has Mg# = 91, values typically associated with mantle olivine.One possibility is that these lithologies are formed by the complex interactions between melts and rocks in the mantle, which have been used to explain the formation of Ol-rich troctolites in other slow-spreading crustal sections and in ophiolites (Drouin et al., 2009(Drouin et al., , 2010;;Ferrando et al., 2018;Sanfilippo et al., 2016;Suhr et al., 2008).Alternatively, the olivine-phyric rocks intruded as crystal-phyric magmas and the whole-rock chemistry results from olivine addition and accumulation in the plumbing system similar to that proposed for olivine-rich lithologies drilled south of the Kane Transform (Meurer & Gee, 2002) or sampled as picrites at the Macquarie Island (Husen et al., 2016).This scenario, while not entirely unambiguous, is better supported by the overall whole rock geochemistry (Figures S7 and S8 in Supporting Information S1) since olivine-rich gabbros formed through melt-rock reactions consistently exhibit lower TiO 2 concentrations than volcanic rocks at the same MgO-content.Therefore, we propose that the petrography and geochemistry of samples collected from the Schulz Massif are indicative of two modes of magmatism: the development of larger plutonic bodies and small injections of melts with varying crystal loads.
Our observation that crystal-laden melts fed the crust during seafloor spreading has implications for understanding the magmatic construction.The transition from movable magmas to rigid, locked-in crystal mushes, depends on the number of crystals versus melt.Consequently, crystal growth must be related to the recurrence rates of magmatic injections and therefore serve as a key factor for how seafloor spreading takes place (e.g., Olive & Dublanchet, 2020).The different timescales recorded using zircon geochronology and CSD, therefore, highlight an important aspect of oceanic magmatic systems; while zircon U-Pb ages represent a protracted crystallization of larger magmatic bodies and/or assimilation of older crustal material (Grimes et al., 2008;Lissenberg et al., 2009;Rioux et al., 2012;Schwartz et al., 2005), CSD reveals the crystallization history of individual intrusions.If the timescale of crystal growth estimated from the CSD is accurate and representative, we can in a simplified manner determine the rate of crustal accretion.By applying an average crystal growth time of 209 days for crystals of 1.8 mm in length, an average-sized magmatic layer in the slow-spreading crust (1 m) would take 317 years to crystalize at growth rates of 1 × 10 7 mm/s (Figure 12, Table S6).Using these estimates to constrain the duration of crustal accretion, the crystallization of a 1-m-thick sill every 317 years yields a continuous growth rate of 0.3 cm/year, significantly lower than rates obtained from zircon geochronology at the Atlantis Bank (1.1 cm/year; Grimes et al., 2008).By assuming that 1 m is composed of 50% olivine and 50% melt, consistent with the volume of olivine derived from the CSD analysis, we would expect an intrusion every 159 years (yielding a continuous growth rate of 0.6 cm/year).In contrast, if we assume a slightly faster growth rate of olivine crystals of 2 × 10 7 mm/s, the formation of a 1-m magmatic layer consisting of 50% olivine and 50% melt would take 79 years (Figure 6), yielding a continuous growth rate of 1.3 cm/year (Figures 12 and 13).
Based on the adopted growth estimates, several interpretations are possible.For instance, adopting a slower growth rate implies that the crustal accretion through the formation of smaller sill intrusions occurs at timeaveraged rates that are ∼3 times slower than the construction of larger gabbroic bodies.However, olivine growth rates depend not only on the temperature and the diffusion rate of the crystal-forming elements (i.e., Mg, Fe, and Si) but also on the water content of the melt where higher water content leads to faster growth times.67 published analyses of H 2 O content in basalts from the Mohns Ridge show a range between 0.2 and 2.6 wt.% and a mean of 0.7 wt.% (Dixon et al., 2017;Kelley et al., 2013;Michael, 1995;Neumann & Schilling, 1984), which suggests that the crystallizing magmas may have also had a high content of water.Therefore, growth rates of 2 × 10 7 mm/s, yielding a continuous growth rate of 1.3 cm/year, may better reflect the time-average growth rates.Adopting this growth rate suggests that we could expect an intrusion every ∼80 years.The similarity between these results and zircon data (e.g., continuous crustal growth rate of ∼1.6 cm/year (Grimes et al., 2008)) provides independent constraints on crustal accretion rates and timescales of magmatic processes using two fundamentally different approaches.Interestingly, 50-to a 100-year time frame suggested by our CSD data for 1m-thick individual intrusions is consistent with those predicted by Olive and Dublanchet (2020) for the development and stabilization of long-lived faulting (recurrence rates between 50 and 120 years).Previous studies have identified various factors which are central to the development and growth of lower crustal sections at mid-ocean ridges, such as the degree of amagmatic extension, depth of dike intrusion, and the timescale of sill intrusions.Our findings provide additional insights into some of the crustal forming mechanisms that are important for the construction and exposure of lower crustal sections at the slow-spreading mid-ocean ridges.Our data and observations support the idea that crustal accretion at slow-spreading ridge segments occurs episodically and at variable depths.A significant part of the lower crust at the Schulz Massif appears to have been formed through small-scale magmatic injections, and the intrusion of these crystal-laden dikes occurred at timescales predicted to stabilize long-lived faulting, suggesting that the recurrence rates of magmatic intrusions may be a strong control over the mode of seafloor spreading.

Conclusions
The study has investigated the timescales and mode of magmatic accretion at an exposed lower crustal section ∼20 km off-axis in the Mohns Ridge.Based on high-resolution and high-precision U-Pb dating, we found that the main gabbroic crustal section crystallized near the ridge axis ∼4.6 Ma.Using the crystallization time estimated from zircon geochronology and the depth of crystallization derived from Na-content in clinopyroxene, we find a time-averaged seafloor spreading rate of mm yr 1 (half-rate), implying that seafloor spreading has been symmetric during the last ∼5 Ma.Petrography and CSD analyses show that olivine-laden melts intruded the crustal section with recurrence rates of approximately 80 years to form the crustal section, yielding a continuous growth rate of 1.3 cm/year for smaller intrusive events.

Figure 1 .
Figure 1.Geographic overview of the study area and sample sites.White squares show the location of remotely operated vehicle (ROV) dives, while the dashed black lines connected by white squares show the dive tracks.Dashed black lines connected by circles show the dredge track which was carried out up-slope.Pie diagrams show the number of lithologies (number of samples) that were collected in the GS16 ROV survey.The labels GS07, GS08, and GS16 correspond to the year of the survey and the notation DR and ROV indicate the type of survey conducted, either dredge or ROV, respectively.The vertical exaggeration of the bathymetry is 2:1 and the bathymetry has a resolution of 25 m.

Figure 2 .
Figure 2. Examples of magmatic rocks from the Schulz Massif.(a) Coarse-grained plagioclase + pyroxene gabbro with large centimeter-sized plagioclase crystals.(b) Olivine-phyric gabbro with olivine crystals of varying size in a finer-grained matrix made up of plagioclase + clinopyroxene.(c, d) Finer grained basaltic intrusions cross-cut mantle peridotite and gabbro.

Figure 3 .
Figure 3. Dynamic recrystallization leading to anastomosing zones of grain size reduction.(a) Variably recrystallized plagioclase and clinopyroxene in gabbroic sample GS08-DR1-E.(a) is seen in plane polarized light, while figure b is seen in crossed polarized light.(c) Cataclastic textures in gabbroic sample GS08-DR1-M4.Seen in plane polarized light.

Figure 8 .
Figure 8. Concordia and mean weighted mean plots for isotope dilution thermal ionization mass spectrometry (ID-TIMS) and secondary ion mass spectrometry (SIMS) analyses of zircon from sample GS16A-ROV6-016.(a) ID-TIMS ratios and mean-weighted average.(b) Uncorrected SIMS ratios.The blue dotted line represents the regression intercept age forced through the assumed common Pb composition (0.83 ± 0.1).Data-point error ellipses are 2σ.

Figure 9 .
Figure 9. Cathodoluminescence (CL) images of zircon from GS16A-ROV6-016 with secondary ion mass spectrometry U-Pb and O-isotope analyses indicated with red and blue circles, respectively.The scale is given by the red spot, which is 24 μm.

Figure 10 .
Figure 10.Discrimination diagram for zircon based on U/Yb.Fields for continental and ocean zircon are from Grimes et al. (2007, 2015).

Figure 12 .
Figure 12.Timescale of growth for single grain olivine.The red dotted line shows the estimated crystal residence time for the largest crystal size at timeaveraged growth rates of 10 7 mm/s.

Figure 13 .
Figure 13.Timescales of olivine crystallization derived from crystal size distribution data.(a) Timescale for crystallization of 1 m of olivine at different olivine growth rates.Red dotted line highlights the time to crystallize 1 m of olivine crystals at growth rates of 1 × 10 7 and 2 × 10 7 mm/s as discussed in the text.(b) Timescales for growth of individual magmatic layers.The red dotted lines indicate the time to crystallize individual magmatic layers at our preferred time-averaged growth rate of 2 × 10 7 mm/s.