Evolution of the Dragon Horn detachment fault. We have mapped the evolutionary history for a portion of the ridge flank based on the half spreading rates measured along magnetic profiles L2 and L3 and in the near bottom magnetic data. If we assume that OCC formation is initiated at the rift valley wall, typically located 3 to 6 km distant from the AVR accretionary axis (Fig. 6) and that OCC2 is still actively slipping, based on in situ observations, then we can reconstruct the history of OCC formation.
Starting with Profile L2, which has the best definition of both OCC1 and OCC2, we begin at the present day and restore the section sequentially through time. If we assume the present-day fault at OCC2 is active and continuing to slip as indicated by the seismic activity23, then we can restore the fault slip using the estimated half spreading rate of 7.97 km/My calculated earlier from the L2 profile (Fig. 2). The termination (T2) of OCC2 is located approximately 5.9 km from the present-day axis of spreading. Given the lateral slip distance on OCC2 of 2.6 km results in an initiation age of 0.33 My BP for OCC2 (Fig. 6a). There is approximately 3.4 km between the breakaway (B2) of OCC2 and the termination (T1) of the older OCC1. If we assume the older OCC1 formed at the same distance from the spreading axis as the present day OCC2 i.e. 5.9 km, this would translate into 0.43 My of spreading, suggesting that OCC1 (T1) stopped slipping approx. 0.76 My BP (i.e. 0.33 + 0.43 My). Restoring the slip on OCC1 of 5.7 km at 7.97 km/My gives a time duration of 0.72 My. Adding this time to the age at the end of slip on OCC1 gives an initiation time of 1.48 My BP for the formation of the OCC1 breakaway B2 (Fig. 6a).
Along profile L3, the older OCC1 is not well-defined, however, for the younger OCC2 we have noted that the Brunhes appears to be extended further here than the other profiles (Fig. 2), which is likely due to the formation of a stranded block (T2’-B2’, with ~ 500 m width) on the surface of OCC2. As noted earlier we have also identified the Jaramillo event in the high resolution near bottom magnetic data (Fig. 5), which provides additional constraints on spreading half-rates. We estimated a half spreading rate of 7.6 km/My for the period between Jaramillo and the Brunhes chron, but a faster 9.1 km/My for the Brunhes chron to the AVR, which encompasses the OCC2 fault. If we use the fast spreading half rate of 9.1 km/My for the slip on OCC2, again assuming it is currently active, we get a period of time of 0.31 My, slightly shorter than the estimate from profile L2 (Fig. 6b). If the fault initiated 0.31 My BP then there would have been 0.47 My of previous crustal spreading to create the Brunhes chron, which at the 9.1 km/My rate would have resulted in 4.3 km of crustal accretion. Given that the fault termination (T2) is located 3.5 km from the axis (AVR) and adds the width of stranded block, then the fault clearly initiated within the normal polarity Brunhes chron, about 300 m from the Brunhes/Matuyama boundary. This is consistent with the observation that the present-day fault breakaway (B2) is ~ 300 m within the Brunhes chron (Fig. 5a). As noted in Fig. 1, Profile L3 does not have a clear expression of OCC1, which is present further west on Profile L2. Calculations from L2 suggest OCC1 initiated 1.48 My BP and terminated at 0.76 My BP, which would have obviated the accretion of Jaramillo-aged crust if all the extension was taken up by slip on OCC1 along profile L2.
If we assume the periods when the OCCs were actively slipping were times of reduced magmatic supply, our evolutionary history for the Dragon Horn segment would imply that magmatic episodes of spreading occurred between 2.8 My and 1.48 My BP and again between 0.76 My and 0.33 My BP. Similarly, slip on the old detachment fault (OCC1) appears to have lasted approx. 0.72 My, which is within previous estimates of DF slip in other SWIR segments 0.6 to 1.5 My.24. The young DF of OCC2 has been active for ~ 0.33 My and may continue for another ~ 0.4 My if the evolution time of DF1 is considered representative of a complete cycle.
Detachment faulting system and control on hydrothermal activity. We believe that the most recent OCC2 at the Dragon Horn segment formed in two main steps (Fig. 6c). The original detachment fault DF2 formed at the breakaway B2 (0.31 ~ 0.33 My BP) and originally stopped at termination T2’. Sometime afterwards, the main slip on the detachment fault stepped further to the south and initiated a second breakaway B2’ on a slip plane that strands a block (T2’ - B2’) on the surface of the scarp face. The stranded block correlates with a modestly high magnetic anomaly along the middle of the detachment surface. This block used to form part of the hanging wall but is now a stranded footwall block. The upper portion of the OCC2 surface, above T2’, is the fossil footwall, while the lower section of the OCC2 slip surface B2’ to T2 is now the exposed active fault (Fig. 6c).
The inactive LQ-3 hydrothermal vent field, located on top of this stranded block on the mid-slope bench between B2’ and T2’ and was likely active following the detachment fault began to initiate at B2 that is 0.33 My Bp, and may have become inactive as it was stranded on its block and became separated from its fluid circulation system and heat source. If we assume continuous slip on the OCC2 and partition time between the old and present-day slip surfaces simply based on the lateral distance, then we can estimate when LQ-3 might have stopped venting. The inactive fault surface (B2-T2’) is ~ 0.4 km wide. Using the interval half spreading rate of 9.1 km/My this would translate into the current slip occurring ~ 46 kyr when the main OCC detachment jumped from its previous location to its present location and formed the stranded block. We speculate that the LQ-3 vent site was actively venting up to 46 kyr prior to the faulting event that formed the stranded block and since that time the vent site has become inactive.
At the western edge of OCC2, the stranded block discussed above causes a lateral offset in the trace of the termination T2 (Fig. 4a). This offset in the trace of the termination means that the active fault surface of OCC2 is revealed upslope as a disconnected portion of the termination T2 at this western edge. Just down slope from T2 is the hanging block that hosts the LQ-1 vent field (Fig. 4a). The LQ-1 vent field includes the M and S zones, with the M zone probably being younger than the S zone. HOV Jiaolong investigations show that there are more inactive vents in the S zone compared with the M zone, and that the temperature of active vents in S are generally lower than those in M zone (Supplementary Fig. 3). The distribution of vent zones suggests a progressive migration of activity along a hydrothermal channel or fault zone25 from the S zone to the M zone.
We speculate that the LQ-1 vent field may have formed at the same time as the LQ-3 vent site. It may have become inactive after the LQ-3 stranded block formed, but then came back to life more recently as activity is now ongoing. This speculation is supported by detailed mineralogical patterns found in the chimneys and dating of the sulfides. A relict chimney collected from the M zone shows abundant medium grained chalcopyrite forms granular aggregates around the inner channels and fine- to medium-grained pyrite and sphalerite to be intergrown with minor chalcopyrite around the outer part of the chimney26. This indicates that the chimney had a low temperature environment early in its formation as an outer pyrite-rich layer formed followed by a period of high temperature that formed the inner chalcopyrite-rich lining. We speculate that hydrothermal venting at LQ-1, although located on a hanging wall block, maybe linked to this stranded block formation. For this case, the faulting activity forming the stranded block has apparently led to a rejuvenation of the thermal pathways for fluid flow and enhanced fluid discharge through the hanging wall block. Furthermore, as discussed earlier, if we assume the DF2 system may continue to be active for another 0.4 My and allow hydrothermal circulation to tap heat over an extended period this would suggest that LQ-1 may continue to grow as a large sulfide deposit.
Additional hydrothermal activity appears to have occurred continuously during the evolution of the DF system inferring that the inactive LJ-E vent field may have been active sometime during 0.76–1.48 My BP when DF1 was active. The LJ-W hydrothermal anomaly was investigated27 but no vent has been detected. We suggest that because heat is mainly being mined by the hydrothermal system related to the younger DF2 system, only a small amount of hydrothermal activity is focused through the path of the older DF1 system to the LJ-W site.
Intermittent detachment faulting and episodic magmatic accretion. High resolution multiscale magnetics allows us to constrain the relative balance between periods of detachment faulting and magmatism to better describe accretionary processes on an ultraslow spreading ridge. We find that detachment fault OCC2 initiated 5.9 km from the AVR on profile L2 and has been slipping on the fault for the past 0.33 My during which no appreciable magmatic accretion has occurred. The 3.4 km of accreted crust between OCC2 and the previous OCC1 represents 0.43 My of accretion between 0.33 My and 0.76 My after OCC1 stopped slipping. We have hypothesized that the old OCC1 formed at the same distance from the AVR as the present day OCC2, i.e, 5.9 km, which given the 5.7 km of slip on OCC1 means that OCC1 initiated 1.48 My BP. Again, we assert that no appreciable magmatic accretion occurred during the period between 0.76 MyBP and 1.48 MyBP. This magmatic episodicity linked with the record of fault initation and slip means that the accretionary record will have significant hiatuses with respect to distance from the AVR. As shown in Fig. 6a, the age of crust between OCC1 and OCC2 i.e. B2 and T1 is 1.79 to 2.22 My, while the crust between the AVR and OCC2 (T2 termination) is highly affected with ages of 0-0.33, 0.33 to 0.76 and 1.48 to 1.79 My. Ref.14 has proposed the similar viewpoint according to U-series eruption ages of volcanic rocks collected from SWIR (11°–15°E), whereas our study shows that we can constrain these processes based on a detailed magnetic framework for detachment faulting systems on an ultraslow-spreading ridge. Multi-scale magnetic surveys are a useful approach for constructing a framework to accurately describe the timing of magmatic and tectonic processes involved in the crustal accretion at ultraslow-spreading ridges.