Modelling the influence of pre-existing brittle fabrics on the development and 1 architecture pull-apart basins 2 3 Giacomo Cortia,*, Rosa Nencinib, Pietari Skyttäc 4 5 a Consiglio Nazionale delle Ricerche (CNR), Istituto di Geoscienze e Georisorse (IGG), UOS Firenze, Via G. 6 la Pira 4, 50121 Firenze, Italy 7 *Corresponding Author, giacomo.corti@igg.cnr.it (tel. +39 055 2757524) 8 b Dipartimento di Scienze della Terra, Università degli Studi di Firenze, Via G. la Pira 4, 50121 Firenze, Italy 9 c Department of Geography and Geology, Geology Section, FI-20014 University of Turku, Finland 10 11 12 Abstract 13 We use new analogue modeling experiments to analyze the development of pull-apart basins in an 14 upper crust characterized by the presence of pre-existing discrete fabrics. As in previous models, 15 lateral movement of rigid basal plates induced strike-slip deformation of a sand-pack. Local 16 extension allowing the formation of a pull-apart basin was produced by the step-over geometry of 17 the master faults; in this area, a basal silicone layer was introduced to distribute deformation and 18 reproduced a weaker crust in the basin. Conditions of neutral, overlapping and underlapping 19 interacting master faults were reproduced. The model upper crust, modelled by a sand mixture, was 20 characterized by the presence of pre-existing structures; the orientation of these inherited 21 heterogeneities was systematically varied in different experiments. Model results indicate that –22 depending on their orientation with respect to the strike-slip displacement- reactivation of the pre-23 existing structures can occur both within and at the margins of the pull-apart basins. Inside the basin, 24 reactivation occurs when the pre-existing structures are orthogonal or sub-orthogonal to the strike-25 slip displacement; in this case, the pre-existing fabrics delay the development and linkage of cross-26 basin faults and increase the complexity of the deformation pattern giving rise to a new set of faults 27 characterized by an atypical trend. Pre-existing fabrics oblique to the local extension direction may 28 be partly reactivated in the central part of the basin as segments of cross-basin faults. At the margins 29 of the pull-apart, reactivation occurs if the fabrics spatially coincide with the lateral boundaries of the 30 silicone layer. In these conditions, reactivation allows a faster development of the border faults, which 31 are less segmented than in the homogenous models; this also results in a more regular final 32 geometry of the pull-apart 33 34 35 Keywords: pull-apart basins; pre-existing structures; reactivation; analogue modelling 36 37 38 1. Introduction 39 Pull-apart basins form where geometrical irregularities such as bends or step-overs in the 40 main strike-slip fault system produce zones of local extension. They represent important features of 41 strike-slip tectonics, and more than 160 basins around the globe have been attributed to transcurrent 42 motion across segmented systems (Mann, 2007). Localised extensional deformation within these 43 basins is typically accompanied by prominent subsidence and thinning of the crust and lithosphere, 44 which may eventually lead to the break-up of the continental lithosphere (e.g., Mann et al., 1983; 45 Umhoefer, 2011). Therefore, these basins are an important component of plate tectonics. 46 Analogue modelling has been proven to be a powerful technique to understand the evolution 47 and architecture of pull-apart basins (e.g., Dooley and Schreurs, 2012). Indeed, much of the current 48 knowledge on these basins comes from the results of analogue models (e.g., Soula, 1984; Faugère 49 et al., 1986; Hempton and Neher, 1986; Raynaud, 1987; McClay and Dooley, 1995; Richard et al., 50 1995; Dooley and McClay, 1997; Rahe et al., 1998; Dooley et al., 1999; Basile and Brun, 1999; Sims 51 et al., 1999; Atmaoui et al., 2006; Smit et al., 2008a, b; Wu et al., 2009; Mitra and Paul, 2011; Dooley 52 and Schreurs, 2012; Corti and Dooley, 2015), which provided significant insights into the 53 development and fault pattern of these basins in a homogeneous brittle or brittle-ductile crust, as 54 summarized below. 55 56 1.1. General characteristics of pull-apart basins 57 Pull-apart basins are laterally limited by segments of transcurrent faults (standard strike-slip 58 faults, SSFs), which constitute the principal deformation zone (PDZ; Fig. 1). The basins are bounded 59 by fault systems with normal or oblique-slip kinematics (basin sidewall faults, BSFs) which form in 60 the direction roughly orthogonal to the strike-slip displacement and are directly connected to the 61 PDZs (Fig. 1). The floor of pull-apart basins is cut by a system of faults (cross-basin faults, CBFs) 62 that link the offset PDZs, accommodate lateral displacement and commonly localise intrabasin 63 subsidence and may separate depocenters within the basin (Fig. 1). The characteristics of the offset 64 between the SSFs are the most important parameter controlling the architecture of pull-apart basins, 65 which can be lozenge, lazy-Z or rhomboidal shaped depending on the offset angle (defining 66 underlapping, neutral and overlapping interactions, see below; e.g., Dooley and Schreurs, 2012 and 67 references therein). Other parameters (e.g., the horizontal separation between the offset SSFs, the 68 ratio of brittle and viscous thickness, the strain rate and resulting coupling between brittle and ductile 69 layers; e.g., Dooley and Schreurs, 2012) are important in controlling the evolution and deformation 70 pattern of pull-apart basins. 71 In cross section, basin morphology displays significant along-strike variations, with transition 72 from narrow V- and U-shaped grabens to a more symmetric, boxlike geometry passing from the 73 basin terminations to the centre (e.g., Dooley and McClay, 1997; Dooley and Schreurs, 2012; Corti 74 and Dooley, 2015). 75 76 1.2. The role of pre-existing structures in strike-slip tectonics and pull-apart development 77 The above-mentioned modelling works focused on the architecture and evolution of pull-apart 78 basins in a homogeneous crust, with no pre-existing discrete structural heterogeneities. However, 79 natural examples and modelling studies have shown that inherited brittle fabrics such as shear 80 zones, faults, foliated rocks, and dykes inside the upper crust exert an important control on the 81 evolution of strike-slip systems. For instance, recent work by Rotevatn and Peacock (2018) has 82 shown that pre-existing segmented normal faults have a large control on the characteristics of the 83 deformation during later phase of strike-slip motion. In agreement with these findings, multiphase 84 deformation analogue models by Richard and Krantz (1991) indicate that faults formed during a first 85 dip-slip stage are reactivated at depth during later strike-slip deformation. Similar results were 86 obtained by Dooley and Schreurs (2012) who showed a control exerted by extensional basins formed 87 during a first phase of extension on the distribution of later transcurrent motion. Models by Viola et 88 al. (2004) illustrate how the reactivation of pre-existing discrete fabrics in the upper brittle crust is 89 able to influence the pattern and evolution of strike-slip faults, a process which has likely controlled 90 the Late Oligocene–Neogene evolution of the Giudicarie fault system in the Italian Eastern Alps. Koyi 91 et al. (2008) have illustrated reactivation of pre-existing discrete fabrics in centrifuge models of 92 simple shear deformation. Besides inherited discrete fabrics, pre-existing heterogeneities of variable 93 nature (e.g., weak bodies such as salt diapirs) have been shown to exert an important control on the 94 pattern of deformation during strike-slip tectonics e.g., (Dooley and Schreurs, 2012). 95 Field examples indicate possible influence of inherited fabrics on the development and 96 architecture of several pull-apart basins. For instance, in the Coso geothermal field, hosted in a 97 transtensional pull-apart basin, the polyphase history of deformation may have involved fabric 98 reactivation (e.g., Dewey et al., 2008), with pre-existing structures at the margins of the basin 99 influencing the development of BSFs (Fig. 2a; Dooley and Schreurs, 2012). In the Cinarcik basin, 100 Sea of Marmara, Turkey, pre-existing basement structures may have controlled the architecture of 101 deformation, resulting in a complex structural pattern deviating from the classical pull-apart 102 architecture (Sugan et al., 2014). Several other examples testify the influence of inherited fabrics on 103 the development of pull-apart basins (e.g., the Erzincan and Merzifon-Suluova basins, North 104 Anatolian Fault Zone, Turkey, Temiz, 2004; Rojay and Koçyiğit, 2012; basins on the Yunnan-105 Myanmar region, Indochina, Morley, 2007). Recent work by Piippo et al. (2019) and Skyttä et al. 106 (2019) indicates that the evolution and architecture of pull-apart basins developed in the 107 Palaeoproterozoic Perapohja Belt in northern Finland was strongly influenced by the structural 108 anisotropy of the Archaean crust. Skyttä et al. (2019) support these observations by means of 109 analogue modelling of a pull-apart basin developing in an upper crust characterized by the presence 110 of inherited, discrete brittle fabrics (Fig. 2b). The results of this model indicate that pre-existing fabrics 111 reactivate during pull-apart development influencing the fault pattern and giving rise to additional 112 fault sets, with atypical trend, affecting the basin floor and causing a prominent segmentation of the 113 CBFs (Fig. 2b). 114 115 1.3. Aims of this work 116 Skyttä et al. (2019) present a single analogue model with pre-existing structures orthogonal 117 to the trend of the strike-slip faults; therefore, the conditions under which the pre-existing structures 118 were reactivated and the role of parameters such as the orientation of inherited fabrics with respect 119 to the trend of the main strike-slip faults on reactivation were not systematically analyzed. 120 In this work, we fill this gap and present new analogue modeling experiments aimed at 121 systematically investigating the influence of pre-existing structures (with different orientations with 122 respect to the trend of the main strike-slip faults) on the development and architecture of pull-apart 123 basins of variable master fault geometries (under- and overlapping, neutral). We show that that -in 124 specific conditions- reactivation of the pre-existing structures can occur both within and at the 125 margins of the basins, strongly affecting the pattern and evolution of pull-aparts. 126 127 2. Model set-up and experimental series 128 We conducted 15 modelling experiments at the Tectonic Modelling Laboratory of the Institute 129 of Geosciences and Earth Resources - National Research Council of Italy and the Department of 130 the Earth Science of the University of Florence (Table 1). The experiments were ran under normal 131 gravity conditions (1g) and the set-up consisted of two crustal blocks that relatively moved past each 132 other to simulate simple shear deformation (Fig. 3). Motion was imposed by motor-driven lateral 133 displacement of a basal Plexiglas plate whose geometry was characterised by an offset of the main 134 strike-slip fault system in order to produce a zone of localised extension and give rise to a pull-apart 135 basin (Fig. 3). Since there is only one mobile plate, the experiments are asymmetric pull-apart basin 136 models. 137 The models consisted of a single 4 cm-thick sand-pack simulating the brittle behaviour of the 138 upper crust; this sand-pack was unconfined in all directions above the basal plates. For this layer, 139 we used a mixture of Quartz and K Feldpsar sands (70% - 30% in weight, respectively), 140 characterised by an angle of internal friction of 39°, density of 1400 kg m-3 and cohesion of 10 141 Pa (Montanari et al., 2017). A basal, 1 cm-thick layer of silicone (Polydimethylsiloxane, PDMS) was 142 placed in the offset area in order to distribute deformation in the basin (Fig. 3). This silicone layer, 143 analogous the use of a basal rubber sheet between rigid plates (e.g., Dooley and Schreurs, 2012) 144 or of a weak zone within the offset area between two rigid blocks (Corti and Dooley, 2015), was 145 intended to simulate weaker crust in a pull-apart. 146 Top-view photos of the models were taken at regular intervals in order to monitor the evolution 147 of surface deformation. Experiments were repeated at least twice; in all cases, the first order results 148 were similar. At the end of each experiment, the models were soaked in water and cut in a set of 149 cross sections to analyse their 3-D internal geometry. 150 151 2.1. Type of experiments 152 In the experiments, the geometry of the basal plate was varied to reproduce the three 153 conditions of neutral (Series 1 models), overlapping (Series 2 models) and underlapping (Series 3 154 models) master faults (Fig. 3a), identical to those investigated by many previous experimental works 155 (see for instance Dooley and Schreurs, 2012 and references therein). For each of these series, we 156 made manual vertical cuts to the sand pile with a knife at regular intervals (4 cm) to reproduce the 157 presenceof pre-existing structures (with width of 3-4 mm) within the upper crustal layer (e.g., Viola 158 et al., 2004; Bellahsen and Daniel, 2005); the basal polymer was not affected by these cuts. In 159 different experiments, the orientation of these pre-existing heterogeneities was varied with respect 160 to the trend of the main strike-slip faults (Fig. 3b; Table 1). For comparison, homogenous models 161 (i.e., with no pre-existing cuts) were also preformed (Table 1). 162 163 2.2. Scaling 164 The models were built with a geometric scaling ratio of ∼3.3×10−6, such that 1 cm in the 165 experiments corresponded to ∼3 km in nature. This allowed modelling ∼12 km of lateral 166 displacement of a ∼12-km-thick upper crust. For a density of natural upper crustal materials of 2700-167 2800 Kg m-3 (resulting in a density model to nature ratio *=model/nature of ∼0.5), a gravity ratio of 1 168 (the gravity is the same in nature and experiments) and the above reported geometric scaling ratio 169 (h*∼3.3×10−6) give a model to nature ratio of stresses (*=*g*h*) of ∼2×10−6 (Hubbert, 1937; 170 Ramberg, 1981). Since the scaling ratio of cohesion is c*=*, the cohesion of the sand mixture (10 171 Pa) scales down to a natural cohesion of rocks of ∼5 MPa. 172 Since the silicone at the base of the models does not correspond to any specific layer in 173 nature but it is an experimental expedient to distribute deformation, the scaling of velocity is not 174 critical in these models. However, the viscosity of the PDMS at room temperature (=2x104 Pa s) 175 and the velocity of deformation applied to the models (v=5 cm/hr) can be scaled down considering 176 the relation *=*v*/h* (Hubbert, 1937; Ramberg, 1981). According to this, natural viscosities 177 between ∼4,4x1019 and 1,5x1021 Pa s correspond to a natural velocity in the range ∼1–35 mm yr−1, 178 well correlating with natural strike-slip systems (see Corti and Dooley, 2015 and references therein). 179 180 3. Model results 181 3.1. Series 1 models: Neutral master faults 182 The models with neutral master faults display initial development of Riedel shears and E-W 183 strike-slip faults in the PDZs, together with NNW-SSE trending BSFs which accommodate 184 subsidence of the pull-apart basin above the basal PDMS (Figs. 4-7; Figs. SI 1-2). Within the basin, 185 deformation is normally accommodated by early development of CBFs, which obliquely cut across 186 the depression and connect the offset PDZs (e.g., Fig. 4). The pull-apart basin in the homogenous 187 model is characterised by the presence of normal faults (internal faults; IFs) that delineate minor 188 horsts which abut against CBFs. Owing to the transcurrent component of motion, the southern and 189 northern margins of the basin are characterised by a typical en-echelon arrangement of major faults 190 (Figs. 4-7). The length/width ratio of the basin at the end of deformation is 1.1. 191 The evolution and architecture of the model with N-S pre-existing structures (Fig. 5) display 192 significant differences: the inherited fabrics are reactivated during the initial phases of deformation 193 (reactivated faults, RFs); this reactivation also affects the development of the CBFs. In contrast to 194 the homogenous model, the CBFs are unable to rapidly link to form a single throughgoing fault 195 connecting the PDZs. Rather, CBFs in this model are initially (1cm of displacement) highly 196 segmented and their linkage occurs for 2cm of displacement only. Once linked, the CBFs cut the 197 RFs, which are progressively rotated clockwise to attain a final average NNE-SSW orientation (Figs. 198 4, 6). The final length/width ratio of the basin is 1.1. 199 The model with E-W fabrics (Figs. 6, 7) has a similar evolution with respect to the 200 homogenous model, with early development and linking of the CBFs. In this model, however, 201 influence of the pre-existing structures is evident at the southern and northern margins of the basins 202 where, for increasing deformation, major linear faults develop. This is in contrast to the homogenous 203 model where, as explained above, the southern and northern margins are characterised by 204 segmented, en-echelon major faults. The length/width ratio of the basin at the end of deformation is 205 1.2. 206 The model with N45°W-oriented pre-existing structures (Fig. 7; Fig. SI 1) shows reactivation 207 of an inherited fabric in the central part of the pull-apart; in this case, the reactivated fabric 208 corresponds to segments of CBFs. The evolution and pattern of deformation is rather similar to that 209 of the homogenous model, although the development of internal normal faults seems to be 210 influenced by the pre-existing fabrics, as testified by the trend of some of these faults which tends to 211 parallelize the inherited NW-SE heterogeneities. The length/width ratio of the basin at the end of 212 deformation is 1.0. 213 Similarly, to the two previous models (with N-S and N45°W-oriented fabrics) the development 214 of CBFs in the model with N45°E-oriented pre-existing structures (Fig. 7; Fig. SI 2) seems to be 215 slightly delayed with respect to the homogenous model. However, in this case no apparent influence 216 of the inherited fabrics on the fault pattern within or outside the pull-apart basin is observed. The final 217 length/width ratio of the basin is 1.0. 218 219 3.2. Series 2 models: Overlapping master faults 220 The homogenous model with overlapping master faults displays early development of NNW-221 SSE trending normal faults bounding the subsiding pull-apart basin, and CBFs connecting the PDZs 222 (Figs. 8, SI 3-7). For increasing lateral displacement, a set of BSFs with NNE-SSW trend normally 223 develops at the margins of the basal PDMS (Fig. 8). Sets of NW-SE-trending internal faults give rise 224 minor horsts within the basin. Similar to series 1 models, the southern and northern margins of the 225 basin are characterised by a typical en-echelon arrangement of major faults (Figs. 8, SI3). The 226 length/width ratio of the basin at the end of deformation is 1.0. 227 Fabric reactivation in the model with N-S pre-existing structures (Figs. 8, SI4) inhibits again 228 the rapid linkage of the CBFs; in this case, these structures remain highly segmented during 229 deformation and their linkage does not clearly occur even for 4cm of displacement. The reactivation 230 of the inherited fabrics also inhibits the development of NE-SW faults at the margins of the PDMS; 231 therefore, differently from the homogenous model, the BSFs are roughly N-S oriented in this model. 232 The RFs within the basin are progressively rotated clockwise with progressive displacement (Figs. 233 8, SI4). The final length/width ratio of the basin is 1.0. 234 As in Series 1 models, reactivation of E-W inherited fabrics (Figs. 8, SI5) is evident at the 235 northern margin of the basins where a major linear fault develops at the beginning of deformation. 236 This contrasts with the segmented, en-echelon major faults characterising the corresponding margin 237 in the homogenous model. No similar linearity of faults is observed in the southern margin, likely 238 because the pre-existing fabric did not spatially correspond to the margins of the basal PDMS. The 239 evolution of the basin is otherwise similar to that of the homogenous one, with NNE-SSW-trending 240 BSFs developing for displacement >2cm, systems of internal faults and CBFs giving to a complex 241 fault pattern affecting the pull-apart depression. The length/width ratio of the basin at the end of 242 deformation is 1.0. 243 Pre-existing N45°W fabrics (Figs. 8, SI6) are partly reactivated in the central part of the pull-244 apart during early stages of deformation. Differently from the homogenous model, the NNE-SSW-245 trending BSFs are less developed and their trend is roughly N-S in this model. Notably, in this case, 246 the CBFs display a linear geometry which contrast with the significant curvature or segmentation of 247 the CBFs in all other overlapping models. The length/width ratio of the basin at the end of deformation 248 is 1.0. 249 The model with N45°E-oriented pre-existing structures display an early development of NE-250 SW-trending BSFs, which correspond to the reactivation of inherited fabrics parallel to the margins 251 of the PDMS (Figs. 8, SI7). These structures progressively accumulate deformation for increasing 252 lateral displacement. This model results in more distributed extensional deformation and higher 253 number of internal faults. The final length/width ratio of the basin is 0.9. 254 255 3.3. Series 3 models: Underlapping master faults 256 Similarly, to the previous corresponding experiments, the homogenous model with 257 underlapping master faults displays initial development of NW-SE trending normal faults (BSFs) 258 delimiting the subsiding pull-apart basin and systems of CBFs connecting the offset PDZs (Figs. 9, 259 SI 8-12). Internal faults are longer, more linear and fewer than in the corresponding neutral and 260 overlapping models (Figs. 9, SI8). Curved, en-echelon normal faults again characterise the southern 261 and northern margins of the basin (e.g., Figs. 9, SI8). The length/width ratio of the lozenge-shaped 262 basin at the end of deformation is 1.2. 263 Reactivation of N-S pre-existing structures during early stages of deformation is similar to 264 what observed in previous models, causing again initial segmentation of the CBFs (Figs. 9, SI9). 265 The RFs within the basin are progressively rotated clockwise with increasing deformation (Figs. 9, 266 SI9). The final length/width ratio of the basin is 1.1. For this model, we also present transversal 267 cross-sections (Fig. 10), which document significant along-strike variations in basin morphology. At 268 the basin terminations, deformation is characterised by narrow V-shaped grabens and negative 269 flower structures; the basin centre displays instead a trapezoidal shape with high angle normal faults 270 (dip >70°) delimiting the depression and inner minor horsts and grabens. Fault reactivation is visible 271 within the pull-apart (see sections D-D’, E-E’, G-G’), whereas fabrics are not reactivated outside the 272 basin. 273 As in previous Series 1 and 2 models, pre-existing E-W fabrics are reactivated at both 274 northern and southern margins of the basin where major linear faults develop (Figs. 9, SI10). Due to 275 the interaction between linear E-W faults and the NW-SE-trending BSFs, the basin is characterised 276 by a very regular shape, with a final length/width ratio of 1.4. 277 No pre-existing structures in the models with N45°W-oriented (Figs. 9, SI11) and N45°E-278 trending (Figs. 9, SI12) fabrics are reactivated within or at the boundaries of the pull-apart basin. The 279 only differences with respect to the homogenous model are a slightly more complex fault pattern 280 (mostly in terms of higher number of more segmented internal normal faults) in the model with 281 N45°W-oriented fabrics (Figs. 9, SI11) and a single CBF and less segments BSFs in the model with 282 and N45°E-trending fabrics (Figs. 9, SI12). The length/width ratio of the basin at the end of 283 deformation is 1.3 for both models. 284 285 4. Discussion 286 The current experiments document a strong control exerted by pre-existing discrete brittle 287 fabrics on the evolution and internal architecture of pull-apart basins. Whereas the standard 288 homogenous models (i.e., with no pre-existing structures) reproduced the typical deformation pattern 289 of pull-aparts characterized by major normal/oblique slip faults at the margins of the subsiding basin 290 and by transcurrent faults obliquely cutting the basin floor and connecting the main strike-slip fault 291 segments (see section 1.1; Figs. 6-8; e.g., Dooley and Schreurs, 2012 and references therein), 292 models with inherited fabrics documented the development of additional fault sets and atypical 293 evolution of structures. The influence or Reactivation of pre-existing structures may occur within or 294 at the margins of the pull-apart basins, depending on the orientation of the inherited fabrics with 295 respect to the strike-slip displacement, as explained below. 296 297 4.1. Pre-existing fabrics orthogonal to the strike-slip displacement 298 Pre-existing fabrics orthogonal to the strike-slip displacement (N-S-trending fabrics) are 299 always reactivated within the basin floor (Fig. 11), giving rise to fault segments characterised by a 300 dominant left-lateral displacement (Fig. 12). This kinematics and the trend of these structures (at a 301 high angle to the strike-slip faults) suggest that these faults can be compared to antithetic Riedel (R’) 302 shears. Therefore, displacement on these fabrics occurs because they are in a favourable orientation 303 to be reactivated as antithetic Riedel shears, in agreement with previous analysis (e.g., Ranalli and 304 Yin, 1990) showing that reactivation occurs when the pre-existing structures are favourably oriented 305 with respect to the local tress field. Similar structures, although with slightly different orientation, have 306 been observed in pull-apart experiments (e.g. Basile and Brun, 1999) or in distributed strike-slip 307 brittle-ductile experiments (e.g., Dooley and Schreurs, 2012). 308 Reactivation occurs during the early stages of strike-slip motion and leads to the development 309 of N-S fault segments, which are not observed in any homogenous model. Deformation along the N-310 S striking structures delays the development of CBFs, which cannot rapidly link to form a continuous 311 system obliquely cutting the basin floor as observed in the homogeneous models (Fig. 11). This 312 testifies that the inherited fabrics, once reactivated, may inhibit the propagation of newly formed 313 faults, as evidenced in previous works (e.g., Teufel and Clark, 1984). In all models, the reactivated 314 N-S faults accumulated only limited amount of deformation: their activity decreased rapidly during 315 progressive strike-slip motion until they were deactivated, cut by CBFs and progressively rotated 316 clockwise. Their final orientation (variable from roughly NNE-SSW to NE-SW) gives rise to a fault 317 system which is atypical with respect to those observed in classical homogenous models. As a result, 318 the complexity of deformation patterns increased as normally observed in cases of fault reactivation 319 (e.g., Morley, 1999; Peacock and Sheperd, 1997; Rotevatn and Peacock, 2018). 320 321 4.2. Pre-existing fabrics oblique to the strike-slip displacement 322 Reactivation of NW-SE- and NE-SW-trending fabrics is strongly dependent on their position 323 and orientation with respect to the margins of the basal silicone layer (which in turn controls the 324 distribution of deformation in the upper crust). Reactivation is favoured when the inherited structures 325 are spatially coincident with the margin of the PDMS layer (i.e., the margins of a weaker crust in the 326 basin): in this case, the strain localisation effect generated by the pre-existing weak fabric adds to 327 the strength contrast between the PDMS and the surrounding sand. A favoured reactivation in case 328 of spatial coincidence between inherited structures and location of development of newly formed 329 faults have been already observed in previous experimental works (e.g., Viola et al., 2004). If this 330 condition is satisfied, reactivation occurs if the trend of pre-existing fabrics is similar to the angle of 331 offset between the two PDZs (i.e., the inherited fabrics are parallel to the long side of the basal 332 PDMS layer), as exemplified in Figs. 13, 14. Reactivation at the basin margins strongly affects the 333 development of BSFs by accelerating their development, decreasing their segmentation and 334 therefore promoting a more linear geometry (Figs. 13, 14), but does not influence significantly the 335 architecture of deformation within the basin floor. However, partial reactivation of NW-SE-trending 336 fabrics is observed in the central part of the pull-apart basin in models with neutral and overlapping 337 master faults (Figs. 7, 8; Figs. SI 1, SI 6); in this case, reactivation occurs because the pre-existing 338 structures are favorably oriented to directly link the offset master faults and are reactivated as 339 segments of CBFs. 340 341 4.3. Pre-existing fabrics parallel to the strike-slip displacement 342 Pre-existing fabrics parallel to the strike-slip displacement (E-W-trending fabrics) are typically 343 reactivated at the northern and southern margins of the basin, when there is a spatial coincidence 344 between inherited structures and the short side of the basal PDMS layer (Figs. 13, 14). In this case, 345 major linear faults develop, contrasting with the more segmented, en-echelon nature of the BBFs in 346 corresponding margins of the homogenous model. This also results in a more regular shape of the 347 basin, as exemplified by model with underlapping master faults (see Figs. 9, Si10). 348 349 4.4. Summary of model results 350 The current models document an important control exerted by pre-existing fabrics on the 351 evolution and the structural pattern of pull-apart basins, with reactivation of pre-existing occurring 352 both within and at the margins of the basins. Inside the basin, reactivation occurs when the pre-353 existing structures are orthogonal to the local strike-slip displacement, i.e., favorably oriented to be 354 reactivated as antithetic Riedel shears (N-S-trending fabrics); in this case, the pre-existing fabrics 355 delay the development and linking of cross-basin faults and increase the complexity of the 356 deformation pattern. Indeed, reactivation gives rise to a new set of faults characterized by an atypical 357 trend, absent in homogenous models. Partial reactivation may also occur when the pre-existing 358 structures are favorably oriented to directly link the master faults (NW-SE-trending fabrics); in these 359 conditions, the inherited heterogeneities are reactivated as segments of cross-basin faults in the 360 central portion of the pull-apart basin. At the margins of the pull-apart, reactivation occurs if the 361 fabrics spatially coincides with the boundaries of a basal silicone layer introduced to distribute 362 deformation. In these conditions, reactivation allows a faster development of the border faults, which 363 are less segmented than in the homogenous models; this also results in a more regular final 364 geometry of the pull-apart. Overall, in line with previous analysis of fault reactivation (e.g., Ranalli 365 and Yin, 1990), these modelling results support that reactivation occurs for favourably oriented pre-366 existing fabrics and is favoured by a significant strength contrast (in this occurring at the margins of 367 the pull-apart basin). 368 369 4.5. Implications for natural pull-apart basins 370 These results have important implications for the evolution and architecture of natural pull-371 apart basins with respect to e.g. the basin architecture, infill stratigraphy, development of pathways 372 for fluid and magma migration and hydrocarbon traps. As explained in section 1.2, many field 373 examples indicate multiphase deformation and possible influence of inherited brittle fabrics (such as 374 shear zones, faults, foliated rocks, and dykes inside the upper crust) on the development of these 375 basins (e.g., the Erzincan and Merzifon-Suluova basins, North Anatolian Fault Zone, Turkey, Temiz, 376 2004; Rojay and Koçyiğit, 2012; Cinarcik basin, Sea of Marmara, Turkey; Okay et al., 2000; Sugan 377 et al., 2014; basins on the Yunnan-Myanmar region, Indochina; Morley, 2007). Pre-existing 378 structures in these natural cases may have contributed to give rise a complex structural pattern 379 deviating from the classical pull-apart architecture and resulting in fault sets with atypical trend, as 380 observed in the current models. One important outcome of our experimental results is that the 381 inherited fabrics within the basin are reactivated during early stages of deformation, and the activity 382 of these faults decrease for increasing lateral displacement and these structures are later 383 deactivated and cut by other fault sets. Processes such as syn-deformation addition of sediments 384 syn-tectonic sedimentation may mask the surface appearance of these structures during progressive 385 deformation, which however may be present at depth. This in turn may influence features such as 386 the potential for these basin to host traps for hydrocarbons or the migration of fluids within the pull-387 apart, with important economic implications. As an example, in the Coso geothermal field, hosted in 388 a transtensional pull-apart basin, the polyphase history of deformation may have involved fabric 389 reactivation (e.g., Dewey et al., 2008; Dooley and Schreurs, 2012), which may have had an influence 390 on the pattern of geothermal fluid transfer. In this natural example, fabric reactivation may have also 391 influenced the development of BSFs (Fig. 2a; Dooley and Schreurs, 2012), as observed in our 392 models. 393 In general, these experiments support that reactivation of pre-existing structures in pull-apart 394 basins may be of significant importance, and may influence several aspects of these basins 395 (evolution, architecture, physiography, etc.) and related processes (e.g., potential for hydrocarbons 396 traps, fluid migration, volcanism). 397 398 Conclusions 399 New analogue models of development of pull-apart basins in an upper crust characterized 400 by the presence of pre-existing discrete fabrics indicate that –depending on their orientation with 401 respect to the strike-slip displacement- these latter may be reactivated both within and at the margins 402 of the basins. Specifically: 403 -pre-existing fabrics orthogonal to the strike-slip displacement are always reactivated within 404 the basin floor giving rise to a new set of faults orthogonal or sub-orthogonal to the strike-slip 405 displacement. In this case, the pre-existing fabrics delay the development and linking of cross-basin 406 faults and increase the complexity of the deformation pattern; 407 -pre-existing fabrics oblique to the strike-slip displacement are reactivated at the margins of 408 the pull-apart if the fabrics spatially coincides with the boundaries of a basal silicone layer introduced 409 to distribute deformation and corresponding to a weaker crust in the basin. In these conditions, 410 reactivation allows a faster development of the border faults, which are less segmented than in the 411 homogenous models; this also results in a more regular final geometry of the pull-apart; 412 -pre-existing fabrics oblique to the strike-slip displacement are partly reactivated within the 413 basin when they are favorably oriented to directly link the offset master faults; in this case, the 414 inherited heterogeneities are reactivated as segments of cross-basin faults in the central portion of 415 the pull-apart basin; 416 -pre-existing fabrics parallel to the strike-slip displacement are reactivated at the northern 417 and southern margins of the basin, when there is a spatial coincidence between inherited structures 418 and the short side of the basal silicone layer. In this case, the inherited fabrics control the 419 development major linear faults, which also results more regular final geometry of the pull-apart. 420 Overall, these results are in line with previous analysis of fault reactivation. They support 421 indeed that reactivation occurs for favourably oriented pre-existing fabrics and is favoured by a 422 significant strength contrast at the margins of the pull-apart, and result in the development of atypical 423 fault sets characterising the deformation pattern. 424 425 Acknowledgments 426 This work benefited from EPOS TCS MSL TNA access to the Tectonic Modelling Lab of Florence, 427 supported by European Community HORIZON 2020 research and innovation program under grant 428 agreement N 676564. We thank Daniele Maestrelli for discussions on the elaborations presented in 429 Fig. 12. Tim Dooley is warmly thanked for the fruitful discussions and the criticism on the paper. An 430 anonymous reviewer is thanked for the detailed comments. We also thank the Editor Stephen 431 Laubach for suggestions. 432 433 Figure captions 434 Figure 1. Typical fault pattern of a pull-apart basin (from Corti and Dooley, 2015). BSFs: basin 435 sidewall faults; CBFs: cross-basin faults; SSFs: standard strike-slip faults. 436 Figure 2. Fabric reactivation in natural and experimental pull-apart basins. a) Summary map of the 437 Coso pull-apart system (modified from Dooley and Schreurs, 2012), showing reactivation of pre-438 existing N-S fabrics at the margins of the basin. b) Evolution of an analogue model of pull-apart 439 development with pre-existing brittle discrete fabrics (modified from Skyttä et al., 2019). 440 Figure 3. Set-up of the experimental series. a) Geometry of the basal plate and angle of offset 441 between the master fault segments, giving rise to conditions of neutral (Series 1 models; offset angle 442 A=90°), overlapping (Series 2 models; offset angle A=135°) and underlapping (Series 3 models; 443 offset angle A=45°) master faults. The layer of Polydimethylsiloxane (PDMS) at the base of the model 444 is indicated with the greenish colour. b) Orientation of the pre-existing cuts (red lines) in the different 445 models (taking the experimental series with neutral master faults as exemplificative). Note that this 446 orientation is referred to cardinal points, where the North-South direction is perpendicular to the 447 strike-slip displacement. 448 Figure 4. Evolution of the homogenous model with neutral master faults illustrated as top-view 449 photos (top) and schematic fault pattern (bottom). IFs: internal faults; other abbreviations as in Fig. 450 1. 451 Figure 5. Evolution of model with neutral master faults and N-S pre-existing fabrics illustrated as 452 Fig. 4. RFs: reactivated faults; other abbreviations as in Figs. 1 and 4. 453 Figure 6. Evolution of model with neutral master faults and E-W pre-existing fabrics illustrated as in 454 Fig. 4. 455 Figure 7. Summary of experimental results of Series 1 models (neutral master faults) 456 Figure 8. Summary of experimental results of Series 2 models (overlapping master faults) 457 Figure 9. Summary of experimental results of Series 3 models (underlapping master faults) 458 Figure 10. Transversal cross-sections for model with underlapping master faults and N-S pre-459 existing heterogeneities. 460 Figure 11. Comparison between homogeneous models (left column) and models with N-S pre-461 existing fabrics (right column) with neutral, overlapping and underlapping master faults (top, middle 462 and bottom panels, respectively) for 1cm of horizontal displacement. Colour coding as in previous 463 figures. 464 Figure 12. Kinematics of faults in the models, exemplified by experiment with underlapping master 465 faults and N-S-trending pre-existing fabrics. Top panel: slip vectors along single faults calculated by 466 using Particle Image Velocimetry (see Philippon et al., 2015 for details of calculations); bottom panel: 467 interpretation of fault kinematics. 468 Figure 13. Comparison among homogeneous models (left column) and models with pre-existing 469 fabrics (right panels) in model with overlapping (top panels) and underlapping (bottom panels) 470 master faults for 1cm of horizontal displacement. Pre-existing fabrics are oriented NE-SW and E-W 471 in case of overlapping and underlapping master faults, respectively. Colour coding as in previous 472 figures. 473 Figure 14. Summary of the influence of pre-existing fabrics on the architecture of pull-apart basins, 474 as exemplified by models with neutral master faults. Note the atypical fault pattern with NNE-SSW 475 to N-S trend in the model with N-S fabrics (central panel), not present in the homogenous model 476 (upper panel); also note the linear faults bordering the northern and southern margins of the basin 477 in the model with E-W pre-existing fabrics (bottom panel), which contrast with the segmented, en-478 echelon major faults characterising the same margins in the homogenous model. 479 480 References 481 Atmaoui, N., Kukowski, N., Stöckhert, B., König, D., 2006. Initiation and development of pull-apart 482 basins with Riedel shear mechanism: insights from scaled clay experiments. International Journal of 483 Earth Sciences, 95, 225-238. http://dx.doi.org/10.1007/s00531- 005-0030-1. 484 Basile, C., Brun, J.P., 1999. Transtensional faulting patterns ranging from pull-apart basins to 485 transform continental margins: an experimental investigation. Journal of Structural Geology 21, 23–486 37. 487 Bellahsen N., Daniel J.M., 2005. Fault reactivation control of normal fault growth: an experimental 488 study. Journal of Structural Geology, 27, pp 769-780. 489 Corti G. and Dooley T. P., 2015. Lithospheric-scale centrifuge models of pull-apart basins. 490 Tectonophysics 664, 154-163. 491 Dewey, J.F., Taylor, T.R., Monastero, F.C., 2008. Transtension in the brittle field: the Coso region, 492 southern California. International Geology Review 50 (3), 193–217. 493 Dooley, T.P., McClay, K., 1997. Analog modeling of pull-apart basins. AAPG Bulletin, 81 (11), 1804–494 1826. 495 Dooley, T.P., McClay, K., Bonora, M., 1999. 4D evolution of segmented strike-slip fault systems: 496 applications to NW Europe. In: Fleet, A.J., Boldy, S.A.R. (Eds.), Petroleum Geology of Northwest 497 Europe: Proceedings of the 5th Conference, Geological Society, London, 215–225. 498 Dooley, T.P., Schreurs, G., 2012. Analogue modelling of intraplate strike-slip tectonics: a review and 499 new experimental results. Tectonophysics 574–575, 1–71. 500 Faugère, E., Brun, J., Van Den Driessche, J., 1986. Asymmetric basins in pure extension and in 501 wrenching: experimental models. Bulletin du Centre de Recherches Elf Exploration Production 10, 502 13–21. 503 Hempton, M., Neher, K., 1986. Experimental fracture, strain and subsidence patterns over en 504 echelon strike-slip faults: implications for the structural evolution of pull-apart basins. Journal of 505 Structural Geology 8, 597–605. 506 Hubbert M.K., 1937. Theory of scale models as applied to the study of geologic structures. Bulleting 507 of the Geological Society of America, 48, 1459—1520. 508 Koyi, H., Ghasemi, A., Hessami, K., Dietl, C., 2008. The mechanical relationship between strike-509 slip faults and salt diapirs in the Zagros fold-thrust belt. Journal of the Geological Society 165 (6), 510 1031–1044. http://dx.doi.org/10.1144/0016-76492007-142. 511 Mann, P., 2007. Global catalogue, classification and tectonic origins of restraining and releasing 512 bends on active and ancient strike-slip fault systems. In: Cunningham, W.D., Mann, P. (Eds.), 513 Tectonics of Strike-slip Restraining and Releasing Bends. Geological Society, London, Special 514 Publications 290, 13–142. 515 Mann, P., Hempton, M.R., Bradley, D.C., Burke, K., 1983. Development of pull-apart basins. Journal 516 of Geology 91, 529–554. 517 McClay, K., Dooley, T., 1995. Analogue models of pull-apart basins. Geology 23, 711–714. 518 Mitra, S., Paul, D., 2011. Structural geometry and evolution of releasing and restraining bends: 519 insights from laser-scanned experimental models. AAPG Bulletin 95 (7), 1147–1180. 520 http://dx.doi.org/10.1306/09271010060. 521 Montanari D., Agostini A., Bonini M., Corti G. and Ventisette C.D., 2017. The use of empirical 522 methods for testing granular materials in analogue modelling. Materials, 10, 635. 523 Morley C.K., 1999. How successful are analogue models in addressing the influence pf pre-existing 524 fabrics on rift structure?. Journal of Structural Geology, 21, 1267-1274. 525 Morley, C.K., 2007. Variations in Late Cenozoic-Recent strike-slip and oblique-extensional 526 geometries, within Indocina: The influence of pre-existing fabrics. Journal of Structural Geology, 29, 527 36-58. 528 Okay A.I., Kaşhar-Özcan A., Imren C., Boztepe-Güney A., Demirbağ E. and Kuşçu I., 2000. Active 529 faults and evolving strike-slip basins in the Marmara Sea, northwest Turkey: a multichannel seismic 530 reflection study. Tectonophysics, 321, 189-218. 531 Peacock D. C., Sheperd J., 1997. Reactivated faults and transfer zones in the Sourthern Coalfield, 532 Sydney Basin, Australia. Australian Journal of Earth Sciences, 44, 265-273, 533 https://doi.org/10.1080/08120099708728309. 534 Philippon M., Willingshofer E., Sokoutis D., Corti G., Sani F., Bonini M., Cloetingh S., 2015. Slip re-535 orientation in oblique rifts. Geology, 43, 147–150. 536 Piippo, S., Skyttä, P., Kloppenburg, A., 2019. Linkage of crustal deformation between the Archaean 537 basement and the Proterozoic cover in the Peräpohja Area, northern Fennoscandia. Precambrian 538 Research 324, 285–302. 539 Rahe, B., Ferill, D.A., Morri, A.P., 1998. Physical analogue modeling of pull-apart basin evolution. 540 Tectonophysics 285, 21–40. 541 Ramberg, H., 1981. Gravity, deformation and the Earth’s crust. Academic Press, London, 452 pp. 542 Ranalli G., Yin Z.M., 1990. Critical stress difference and orientation of faults in rocks with strength 543 anisotropies: the two dimensional case. Journal of Structural Geology, 12, 1067-1071. 544 Raynaud S., 1987. Les premiers stades de la déformation dans une zone de relais entre 545 décrochements: exemples naturels et expérimentau.. Bulletin de la Société Géologique de France 546 8, III (3), 583–590. 547 Richard, P., Naylor, M.A., Koopman, A., 1995. Experimental models of strike-slip tectonics. 548 Petroleum Geoscience 1, 71–80. 549 Rotevatn A., Peacock D.C.P., 2018. Strike-slip reactivation of segmented normal faults: Implications 550 for basin structure and fluid flow. Basin Research, 1-16. 551 Rojay B., Koçyiğit, A., 2012. An Active Composite Pull-apart Basin Within the Central Part of the 552 North Anatolian Fault System: the Merzifon-Suluova Basin, Turkey. Turkish Journal of Earth 553 Sciences, 21, 473–496. 554 Skyttä P., Piippo S., Kloppenburd A., Corti G., 2019. 2.45 break-up of the Archean continent in 555 Northen Fennoscandia: Rifting dynamics and role of inherited structures within the Archean 556 basement. Precambrian Research, Vol. 324, pp.303-323, doi: 10.1016/j.precamres.2019.02.004. 557 Sims, D., Ferrill, D.A., Stamatakos, J.A., 1999. Role of a ductile décollement in the development of 558 pull-apart basins: experimental results and natural examples. Journal of Structural Geology 21, 533–559 554. 560 Smit, J., Brun, J.-P., Cloetingh, S., Ben-Avraham, Z., 2008a. Pull-apart basin formation and 561 development in narrow transform zones with application to the Dead Sea Basin. Tectonics 27, 562 TC6018. http://dx.doi.org/10.1029/2007TC002119. 563 Smit, J., Brun, J.-P., Fort, X., Cloetingh, S., Ben-Avraham, Z., 2008b. Salt tectonics in pull-apart 564 basins with application to the Dead Sea Basin. Tectonophysics 449, 1–16. 565 Soula, J.-C., 1984. Genèse de bassins sedimentaires en regime de cisaillement transcurrent: 566 modèles expérimentaux et exemples géologiques. Bulletin de la Société Belge de Géologie 93 (1–567 2), 83–104. 568 Sugan, M., Wu, J.E.L., McClay, K., 2014. 3D analogue modelling of transtensional pull-apart basins: 569 comparison with the Cinarcik basin, Sea of Marmara, Turkey. Bollettino di Geofisica Teorica ed 570 Applicata, 55, 699-716. 571 Temiz, H., 2004.The role of thrust ramp reactivation in pull-apart mechanism of the Erzincan basin, 572 North Anatolian Fault Zone, Turkey. Geodinamica Acta, 17, 219–228. 573 Teufel L.W., Clark J.A., 1984. Hydraulic fracture propagation in layered rock: experimental studies 574 of fracture containment. Soc. Petrol. Engrs J. 24, 19-32. 575 Umhoefer, P.J., 2011. Why did the Southern Gulf of California rupture so rapidly? Oblique 576 divergence across hot, weak lithosphere along a tectonically active margin: GSA Today 21 (11), 4–577 10, doi:10.1130/G133A.1. 578 Viola G., Odonne F., Mancktelow N.S., 2004. Analogue modelling of reverse fault reactivation in 579 strike-slip and traspressive regimes: application to the Giudicaries fault system, Italian Eastern Alps. 580 Journal of Structural Geology, 36, 401-418. 581 Wu, J., McClay, K., Whitehouse, P., Dooley, T., 2009. 4D analogue modelling of transtensional 582 pull-apart basins. Marine and Petroleum Geology 26 (8), 1608–1623. 583 584 585 Experimental series Model Type of master fault interaction Orientation of pre- existing fabrics Displacement (cm) 1 1 Neutral - 4.1 1 2 Neutral N-S 4.0 1 3 Neutral N45°W 4.0 1 4 Neutral N45°E 4.2 1 5 Neutral E-W 4.0 2 6 Overlapping - 4.0 2 7 Overlapping N-S 4.1 2 8 Overlapping N45°W 4.0 2 9 Overlapping N45°E 4.1 2 10 Overlapping E-W 4.0 3 11 Underlapping - 4.1 3 12 Underlapping N-S 4.1 3 13 Underlapping N45°W 4.1 3 14 Underlapping N45°E 4.0 3 15 Underlapping E-W 4.2 586 Table 1. Characteristics of the different experiments 587 588 589 Supplementary Figure captions 590 Figure SI1. Evolution of model with neutral master faults and NW-SE pre-existing fabrics. 591 Figure SI2. Evolution of model with neutral master faults and NE-SW pre-existing fabrics. 592 Figure SI3. Evolution of the homogenous model with overlapping master faults. 593 Figure SI4. Evolution of model with overlapping master faults and N-S pre-existing fabrics. 594 Figure SI5. Evolution of model with overlapping master faults and E-W pre-existing fabrics. 595 Figure SI6. Evolution of model with overlapping master faults and NW-SE pre-existing fabrics. 596 Figure SI7. Evolution of model with overlapping master faults and NE-SW pre-existing fabrics. 597 Figure SI8. Evolution of the homogenous model with underlapping master faults. 598 Figure SI9. Evolution of model with underlapping master faults and N-S pre-existing fabrics. 599 Figure SI10. Evolution of model with underlapping master faults and E-W pre-existing fabrics. 600 Figure SI11. Evolution of model with underlapping master faults and NW-SE pre-existing fabrics. 601 Figure SI12. Evolution of model with underlapping master faults and NE-SW pre-existing fabrics. 602 603