A likely geological record of deep tremor and slow slip events from a subducted continental broken formation

Fluids in subduction zones play a key role in controlling seismic activity, drastically affecting the rheology of rocks, triggering mineral reactions, and lowering the effective stress. Fluctuating pore pressure is one important parameter for the switch between brittle and ductile deformation, thus impacting seismogenesis. Episodic tremor and slow slip events (ETS) have been proposed as a common feature of the geophysical signature of subduction zones. Their geological record, however, remains scanty. Only the detailed and further characterization of exhumed fossil geological settings can help fill this knowledge gap. Here we propose that fluctuating pore pressure linked to metamorphic dehydration reactions steered cyclic and ETS-related brittle and ductile deformation of continental crustal rocks in the subduction channel of the Apennines. Dilational shear veins and ductile mylonitic shear zones formed broadly coevally at minimum 1 GPa and 350 °C, corresponding to ~ 30–40 km depth in the subduction zone. We identify carpholite in Ca-poor metasediments as an important carrier of H2O to depths > 40 km in cold subduction zones. Our results suggest that the described (micro)structures and mineralogical changes can be ascribed to deep ETS and provide a useful reference for the interpretation of similar tectonic settings worldwide.


Results
A likely geological record of deep ETS in continental metasediments. The Apennines formed due to the convergence and successive collision of the European and African plates from the Late Eocene (e.g. 20 , Supplementary Fig. 1). The internal Northern Apennines are characterized by unmetamorphosed continental and oceanic units stacked with their metamorphic counterparts. The metamorphosed oceanic units derive from the Ligure-Piemontese oceanic basin, whereas the continental units originate from the former Tethyan passive distal continental margin of the stretched Adria plate (e.g. 21,22 ). Initial subduction of oceanic crust was followed by subduction of this distal continental margin, as attested to by metamorphic high pressure-low temperature conditions between 0.8-1.6 GPa and 300-500 °C (summary of pressure and temperature (P-T) conditions in 23 ). In particular, carpholite and lawsonite in the metamorphic parageneses from metamorphosed continental and oceanic units suggest relatively cold geothermal gradients of c. 8-10 °C/km during subduction (e.g. 22,24 ).
The metamorphosed continental units in the study area are composed of a several 100-m-thick Middle-Low Triassic metasedimentary clastic sequence comprising alternating metaconglomerate, metaquartzarenite and metapelite layers and is regionally referred to as Verrucano Formation 25 (Supplementary Fig. 1d). The Verrucano Formation (traditionally called "Verrucano") is indeed of regional importance as it crops out over an area of ~ 12,000 km 2 in the Northern Apennines. In the study area, the Verrucano is intensely deformed, with mechanically competent metaconglomerate and metaquartzarenite layers stretched and boudinaged and enveloped by less competent metapelite (Figs. 1a,b and 2a,b; GPS coordinates in Supplementary Table 1). As a whole, it forms a so-called "broken formation", defined as a "disrupted rock unit with a block-in-matrix fabric, which contains no exotic blocks but only native components" 26,27 (Fig. 1c,d). The observed structures crop out over an area of ~ 1 km 2 .
Diffuse ductile deformation is therein characterized by shear zones at all scales, with a mylonitic blueschist facies foliation defined by aligned carpholite, phengitic muscovite (Si content > 3 apfu) and quartz wrapping around clastic quartz grains and producing asymmetric pressure shadows composed of quartz, carpholite and muscovite ( Fig. 2c-f and Supplementary Fig. 2; details in Supplementary). The metaconglomerate and metaquartzarenite may contain quartz clasts displaying evidence of compenetration at sites of high normal stress and asymmetric pressure shadows occurring at sites of low normal stress (i.e., perpendicular and parallel to the foliation, respectively). In between two compenetrated grains, phyllosilicates may locally occur and in the pressure shadows quartz is finer grained and overgrows epitactically the clasts (Fig. 2d-f). These microstructures evidence deformation mainly by dissolution-precipitation creep (e.g. 28 ).
Localized brittle deformation is also common and is expressed by tension gashes, dilational step-overs and dilational shear veins occurring in the three lithotypes. Dilational step-overs and tension gashes occur mainly in the stronger lithotypes oriented at high angle to the foliation (~ 70-80°; Fig. 2g,h), whereas dilational shear veins are subparallel to the foliation and are more common in metapelite ( Fig. 3a-e). In the metaconglomerate and metaquartzarenite, dilational shear veins may be up to 1 m long, a few decimetres thick and can form up to ~ 10% of the entire volume of the rock. In the metapelite, dilational shear veins can be several metres long and up to a few decimetres thick and locally represent 50% of the rock volume. In all lithotypes vein geometry is from straight to winding (Figs. 1, 2g,h and 3a-e).
The veins are composed of quartz and carpholite fibres, some decimetres long, and minor carbonate (Figs. 2g,h and 3a,b). The fibres are iso-oriented, and parallel and perpendicular to the vein boundaries in the dilational shear veins and tension gashes, respectively. They are also parallel to the host-rock stretching lineation, which is defined by phengitic muscovite, quartz and carpholite grains. Both vein fibres and stretching lineation plunge gently to the E-NE, with only a few scattered vein fibres plunging to the N; the mylonitic foliation dips gently to the E (Fig. 4a). Dilational shear veins exhibit crack-and-seal textures. At the outcrop, fibres are locally cut across by thin sets of parallel fractures oriented perpendicularly to the fibre long dimension (i.e. perpendicular to the stretching direction of the fibres) and perpendicular to the foliation, marking successive growth increments (Figs. 2h and 3b).
In thin section, quartz fibres appear as up to a few centimetres long and display undulose extinction (Figs. 3e and 4b). They are monocrystalline, as visible in the optical scans of the thin sections with the gypsum plate inserted, and have high aspect ratios (up to 30; Supplementary Fig. 3a). They are locally recrystallized, leading to a significant grain size reduction down to some tens of microns and low aspect ratios for the new grains. Quartz fibres contain abundant fluid inclusion bands with both liquid and vapor phases 29 oriented perpendicular to the long dimension of the grains (i.e. perpendicular to the foliation; Fig. 4b,c and Supplementary Fig. 4a-d). Spacing between the inclusion bands is in the order of a few tens of microns, although some variability exists between adjacent bands. Similar microfractures are also present in the carpholite fibres (see next sections), although along those fractures fluid inclusions bands are not observed. This type of microstructure has been interpreted by 12,30 as resulting from incremental crack-and-seal growth of the dilational shear veins (see also sketch of Fig. 3c). In the text below, these veins will be referred to as dilational hydroshear veins, as defined by 31   www.nature.com/scientificreports/ veins at high angle to the foliation along the low-shear strength foliation planes that are locally exploited as slip surfaces, according to the model of 12,32 ( Fig. 3c-e). Summarizing, these microstructures reflect the incremental epitaxial growth of the fibres by repeated brittle failure and sealing of the fracture coupled to slip along the phengitic muscovite-rich bands during overall ductile deformation of the host rock (see sketch in Fig. 3c). Locally, quartz bands display incipient boudinage, with muscovite in the boudin necks (Figs. 3c and 4b and Supplementary Fig. 4a-d). Carpholite fibres are also boudinaged and finer-grained carpholite and quartz occur in the boudin necks, suggesting that boudinage formed coevally with the documented incremental cracking and sealing (Fig. 4d,e and Supplementary Fig. 4e-h). Some veins are locally intensely folded by non-cylindrical folds, both at the meso-and microscale (Figs. 1f and 5). Carpholite is stable within fold hinges and is oriented parallel to the axial plane, supporting the conclusion that carpholite was stable at the time of folding (Fig. 5e,f). Additionally, folding is limited to some bands, leaving surrounding veins undeformed and suggesting that several episodes of dilational hydroshear vein formation occurred (Fig. 5g,h). Locally, folded dilational hydroshear veins are cut by a younger generation of quartz and carpholite veins (Fig. 6). Based on the presented meso-and microstructural data, several episodes of vein formation are to be postulated and cyclic brittle and ductile deformation are likely to have alternated at similar P-T conditions (see below).

Dilational hydroshear veins forming at 30-40 km of depth.
The presence of syn-deformational carpholite in the dilational hydroshear veins suggests that fracturing and veining occurred at blueschist facies conditions, i.e. at high pressure and low temperature within the subduction channel (e.g. 24,33 ). Quantitative compositional X-ray mapping reveals the presence of microcracks in the carpholite grains oriented perpendicular to the long dimension of the grains (i.e., perpendicular to the stretching direction) that are sealed by a second generation of carpholite with higher X Mg content (X Mg = Mg/(Fe 2+ + Mg); Fig. 7a-c and Supplementary Table 2). This is consistent with the quartz microstructure constraining vein growth by incremental crack-seal, with repeated fracturing and sealing of the dilational hydroshear veins by mineral precipitation, as proposed by 12,30 (Figs. 3b,c and 4b-e and Supplementary Fig. 4).
Thermodynamic modelling was carried out using an X-ray mapping approach that allowed us to extract the local bulk composition of the studied microstructures 34,35 (see "Methods"; Fig. 7 and Supplementary Fig. 3 and Table 3). Results constrain P-T conditions of at least 1 GPa and 300-350 °C for the formation of both the highpressure veins (Sample A, Fig. 3a-e) and the mylonitic foliation in the metapelite (Sample B, Fig. 3f-i), similarly to what previously reported for this geological unit 29 (Fig. 8a,b; details available in Supplementary).

Discussion
The Verrucano bears evidence of brittle dilational hydroshearing and ductile shearing alternating cyclically during subduction at temperatures typical of the brittle-ductile transition zone (350-450 °C in 36,37 ). Importantly, this zone can occur at greater depths than usual in cold subduction zones, where, as in the case of the Northern Apennines, a low geothermal gradient of 8-10 °C/km was present (data from this study and from 22,23,29 ). Both veins and foliation formed at pressure conditions typical of c. 30-40 km depth. This depth corresponds to the lower limit of the seismogenic zone of some subduction megathrusts, where deep ETS are generally reported (e.g. 5,11,38 ; Fig. 8c). Examples of ETS occurring in subduction settings with comparable geothermal gradients as in the Northern Apennines include the seismologically active modern subduction zones of New Zealand, Costa Rica, Alaska and SW Japan as well as fossil and exhumed subduction settings such as the Cyclades in Greece and the Italian Western Alps (Fig. 1 of 10 and Fig. 3 of 11 and references therein). Additionally, deep ETS are also reported from seismologically active modern subduction zones involving continental crust. This is the case of the southern Central Range of Taiwan, characterised by warmer geothermal gradients compared to the Northern Apennines. In this geological context, deep ETS occur between 15 and 45 km and are linked to metamorphic dehydration reactions happening within the subducting continental crust and producing high pore pressure and failure along low dipping thrust faults 39,40 .
Dilational hydroshear veins in subducted metasediments are thus interpreted as possible evidence of ETS and suggest the presence of pore pressure transiently exceeding the least principal compressive stress, roughly equivalent to near-lithostatic values in subduction settings 12,32,36 . Hence, we propose that the studied brittle structures likely correspond to the geological record of tremors documented by the seismologic data, while ductile deformation relates to slow slip in the geodetic data, as suggested by 7,12,15,32,41 . Moreover, our estimated P-T range corresponds to conditions typical for dehydration reactions in Ca-poor metasediments. Between 0.5 GPa-200 °C and 1.9 GPa-550 °C, the main hydrous minerals in Ca-poor metasediments are kaolinite (containing 14 weight percent (wt%) of H 2 O bounded in the crystal structure), pyrophyllite (5 wt%), chlorite (10-14 wt%), carpholite (10-11.5 wt%), chloritoid (7 wt%), muscovite (4.3 wt%) and paragonite (4.6 wt%; Fig. 8a,b) www.nature.com/scientificreports/ A direct correlation exists between the carpholite modal amount (volume percent-vol%) and the H 2 O content of the mineral phases (wt%) above 0.8 GPa (Supplementary Fig. 5). In our samples, carpholite decreases from 42 and 63 vol% (sample A and B, respectively) to 0 vol% between 1 GPa-300 °C and 1.8 GPa-500/550 °C ( Supplementary Fig. 5c,h). This decrease matches the content of H 2 O in solids, from 7-8 wt% for sample A and B, respectively, to 4.6 wt% for both samples ( Supplementary Fig. 5b,g). For our P-T conditions, the decrease in carpholite is balanced by chloritoid increase (Supplementary Fig. 5e,l). However, the net aqueous fluid release from the sample is positive, due to a lower amount of H 2 O per formula unit of chloritoid compared to carpholite.
Carpholite-rich veins can, therefore, act as a syn-deformation trap for H 2 O released from chlorite-out reactions, because a significant amount of H 2 O can be stored in the carpholite structural formula. Successively, this aqueous fluid can be released deeper down in the subduction channel, at a temperature higher than 350 °C, as the carpholite-out reaction is temperature dependent (Fig. 8a,b). Therefore, especially in cold subduction zones, carpholite can be stable down to ~ 60 km. Our estimates of 2.4-3.4 wt% of H 2 O release due to the carpholite-out reaction are maximum amounts. We intentionally selected samples that represent the higher spectrum of the carpholite vol% visible in the field, and we modelled dilational hydroshear veins and metapelite characterised by carpholite-rich bands (Supplementary Table 3). However, ~ 0.5-1 wt% of H 2 O release can be regarded as representative for the Verrucano broken formation, assuming a 5-15 vol% of carpholite as average value.
We thus propose carpholite to be one important H 2 O carrier in poor-Ca metasediments. This possibility has been overlooked 2 or considered minor 44 , likely due to the restricted carpholite stability field, which is limited to cold subduction zones and poor-Ca metasediments, and to the ready destabilization of carpholite as temperature increases at the beginning of exhumation due to the relaxation of depressed isotherms 24 . We suggest carpholite to have destabilized in the studied fossil subduction channel starting only a few km deeper than the investigated samples, thus affecting the local rheological behaviour of the subduction interface. Most likely, also the dehydration of nearby oceanic units contributed to the fluid budget of the studied subduction interface, as above 350 °C the lawsonite-out dehydration reaction occurs (e.g. 2,[45][46][47][48]. These data suggest that the described metamorphic dehydration reactions provide batches of aqueous fluid whenever a subducting volume of rocks crosses a dehydration reaction, leading to pore pressure fluctuations and brittle-ductile cyclicity, as proposed by 14,45,49,50 and as also suggested for the Northern Apennines 51 . The studied subduction interface is characterized by a broken formation, with more competent blocks in a weaker matrix. This is a scale-invariant feature, which is readily recognised from the macro-and meso-to the micro scale (Figs. 1, 2a,b and 3c-e). In metaconglomerate and metarenite, detrital quartz grains behave as rigid objects in a weaker ductile matrix composed of newly crystallised quartz, carpholite and muscovite grains growing in the strain shadows (Fig. 2c-f). Carpholite is stronger than quartz and is locally fractured and boudinaged, with quartz filling the boudin necks (Fig. 4d,e and Supplementary Fig. 4e-h). Quartz layers are more competent than the adjacent muscovite bands and display incipient boudinage (Figs. 3c and 4b,c). Such a "block in matrix" behaviour is similar to what described in 52 for shallower geological settings. We cannot exclude that a "block in matrix" structure could have also been acquired at shallow structural levels at the onset of subduction 27 and/or during the exhumation path "en route" to surface. However, the described ubiquitous foliation and dilational hydroshear veins and, more importantly, the presence of synkinematic carpholite and phengitic muscovite in boudins, boudin necks and matrix (Figs. 2a,b and 3c-e) constrain formation of these structures to the peak P-T recorded by these rocks. A later retrograde greenschist facies overprint is locally observed at the micro-and meso scale, where both blueschist facies foliation and veins are deformed by upright folds and S-C-C′ planes associated with chlorite ( Supplementary Figs. 6 and 7; details available in Supplementary).
Episodic jamming of more competent blocks can occur during deformation, forming a transient load-bearing, more competent network of competent blocks and minor deformation in the surrounding weak matrix 53 . Jamming can cause fracturing of the blocks and surrounding matrix and distributed ductile deformation in the matrix 53 . Nevertheless, the peculiarity of the studied field site is that structures ascribable to both brittle and Metapelite with quartz and carpholite dilational hydroshear veins, some metres in length. View parallel to foliation plane; dashed light blue and blue lines indicate fibres and stretching lineation, respectively. (b) Detail of dilational hydroshear vein with individual quartz and carpholite fibres up to 10 cm long. Carbonate is locally intergrown with the previous minerals. Note crack-seal growth increments marked by fractures oriented parallel to the vein boundaries and highlighted by the dashed yellow line. (c) Polished slab of (b) cut parallel to both stretching lineation and quartz and carpholite fibres (X = L s ) and perpendicular to the foliation (Z parallel to the pole of the mylonitic foliation). Incipient lateral segmentation and boudinage of the quartz band is evident along the bottom of the hydroshear vein. The sketch illustrates the formation mechanism of dilational hydroshear veins, with inclusion bands marking growth increments oriented perpendicular to the long dimension of the crystals and vein boundaries; assumed stress trajectories (σ 1 and σ 3 ) are shown (based on 31 ). (d,e) Thin section optical scans of (c) with quartz bands composed of fibres some centimetres in length. Some fibres show undulose extinction. Mica-rich bands appear dark due to micron-sized inclusions of hematite and graphite. Carpholite fibres vary from colourless to brownish due to incipient retrogression. Plane-polarized light and crossed-polarized light, respectively. (f) Stretching lineation in metapelite defined by quartz and carpholite fibres on a foliation plane (dashed light blue lines) deformed by upright fold (axial plane trace: dashed light green line); view parallel to foliation plane. (g) Detail of quartz (white) and carpholite (green) fibres intergrown with muscovite (silvery). Weathered carbonate (brown) is located inside a vein. (h,i) Thin section optical scans of (f) displaying a folded foliation defined by quartz-rich-and white mica and carpholite-rich bands. Chlorite, with green absorption colours, locally replaces carpholite (further details in Supplementary Figs. 3e-h and 6). Planepolarized light and crossed-polarized light, respectively. www.nature.com/scientificreports/ www.nature.com/scientificreports/ ductile deformation are found in all lithotypes, irrespective of their competence (Fig. 8d), comparable to what observed in metasediments deformed at similar temperatures but at much shallower depths 12 . This style of deformation is different from what reported by 15,41 , where brittle deformation is reported as only occurring in the more competent blocks, and ductile deformation is mostly limited to the surrounding weaker matrix.
In the Verrucano, metapelite and phyllosilicate-rich layers acted as seals that maintained overpressure and allowed for preferential fluid flow parallel to the mesoscopic planar anisotropy defined by the foliation and lithological boundaries 50 . Such conditions are highly favourable for the opening of dilational hydroshear veins,  (Fig. 6). Concluding, we propose that the studied geological structures may represent the fossil record of deep ETS, which would have led to cyclic brittle and ductile deformation within the subduction channel at > 30 km depth. Pore pressure is proposed as the main trigger of cyclicity, repeatedly and transiently reaching near-lithostatic values due to metamorphic dehydration reactions occurring in both oceanic and continental rocks. We identify the carpholite-out reaction as one such reaction for metasediments poor in Ca, and carpholite as one of the carriers of H 2 O down to 30-60 km depth in cold subduction zones. Phyllosilicate-rich rocks can act as relatively low-permeability barriers that maintain overpressure and allow for fluid flow to preferentially occur parallel to the foliation and lithological boundaries. Our results suggest reconsidering the role of quartz-carpholite veins forming coevally with metamorphic foliation as a possible record of deep ETS in similar geological settings of other convergent orogens, including the Alps 54 , Tukey 55 , Crete 56 , Oman 57 , the Svalbard 24 and New Caledonia 58 . Electron probe micro-analyser (EPMA) and X-ray compositional map elaboration. EPMA analysis was performed on carbon-coated thin sections using a JEOL JXA-8200 electron microprobe at the Department of Earth Sciences of the University of Milano (Italy). Backscattered electron images (BSE) were acquired using an accelerating voltage of 15 keV, a beam current of 5 nA, and a working distance of 11 mm. Point analyses and X-ray compositional maps were acquired using wavelength-dispersive spectrometers. Point analyses were acquired first, before the X-ray compositional maps on the same area. Analytical conditions of point analysis were a 15 keV accelerating voltage, a 5 nA beam current and a beam ø of ~ 1 μm. Nine oxide compositions were measured, using the following standards: grossular (SiO 2 /Al 2 O 3 /CaO), fayalite (FeO), forsterite (MgO), K-feldspar (K 2 O), omphacite (Na 2 O), ilmenite (TiO 2 ), and rhodochrosite (MnO). Analytical conditions for Xray map acquisition were a 15 keV accelerating voltage, a 100 nA specimen current, and 50 ms dwell times. Nine elements (Si, Ti, Al, Fe, Mn, Mg, Na, Ca, and K) were measured at the specific wavelength in two passes. X-ray www.nature.com/scientificreports/ maps were processed using XMapTools 34 and intensity X-ray maps were standardized to concentration maps of oxide weight percentage using spot analyses as the internal standard.

Methods
Thermodynamic modelling. The Gibbs free energy minimization algorithm Theriak-Domino 59 was used to compute the isochemical equilibrium phase diagrams, mineral isopleths, and diagrams of the H 2 O content in solids and modal amount of the hydrous mineral phases. The thermodynamic database of 60 , based on 61 , was used. All Gibbs free energy minimizations were computed with an excess in pure H 2 O fluid. Local bulk compositions were obtained using standardized X-ray maps, following the procedure described in 35 , removing the Fe content of hematite, as in 60 . Fe 3+ was ignored because of its subtle content or absence in the studied mineral phases (Supplementary Table 2). Ca and Mn were removed from the input composition because of their minor content (Supplementary Table 3). Plots of vol% of the hydrous mineral phases and wt% of H 2 O in solids were computed along a prograde P-T path valid for the Northern Apennines, based on data from this study and from 23 ( Supplementary Fig. 5).

Data availability
All data generated or analysed during this study are available within this published article (and its Supplementary Information files).