Pre-existing basement faults controlling deformation in the Jura Mountains fold-and-thrust belt: Insights from analogue models

Abstract Pre-existing faults in the mechanical basement are believed to play an important role in controlling deformation of the thin-skinned Jura Mountains fold-and-thrust belt, which constitutes the northernmost extension of the European Alps. We use brittle-viscous analogue models to investigate the influence of frontal and oblique basement steps on the subsequent evolution of structures during thin-skinned shortening. Vertical offset between two rigid baseplates (simulating the mechanical basement) causes the formation of reverse faults and grabens in the overlying brittle layers that are not reactivated during subsequent thin-skinned shortening. However, baseplate steps localise deformation, causing a temporary frontward propagation of deformation in an early stage and inhibiting propagation afterwards. Downward baseplate steps induce very strong deformation localisation and foster the formation of fault-bend folds. Models featuring upward steps develop step-controlled pop-up structures with imbricated fronts and viscous ramps that shorten dynamically with progressive contraction. We find that deformation localisation increases both with higher step-throws and lower obliquity (α) of the strike of the step (e.g. frontal step α = 0°). With increasing step-throws, α = 30° and α = 45° oblique upward-steps lead to a characteristic imbrication of the brittle cover with laterally confined thrust-slices and step-parallel oblique-thrusts, which rotate up to 15° about a vertical axis over time. Step-controlled backthrusts preceding the formation of thrust-slices do not show notable rotation and hence constitute excellent indicators for the orientation of oblique upward-steps. The topographic patterns of oblique-step models resemble individual thin-skinned structures of the Internal Jura (i.e. Pontarlier and Vuache fault zones, the nappe system SE of Oyonnax and the Chasseral anticline), strongly suggesting that pre-existing NNE-SSW and NW-SE striking oblique upward-steps in the basement controlled deformation in the overlying cover. Our model results may be applied to other thin-skinned fold-and-thrust belts worldwide that formed above pre-existing basement structures.

Alpine Tethys, as well as during the evolution of the ECRIS and the Alpine flexural foreland basin system in Cenozoic times (Illies, 1962;Laubscher, 1986;Madritsch et al., 2009;Meier, 2010;Reisdorf and Wetzel, 2018;Schumacher, 2002;Wetzel et al., 2003;Ziegler, 1988a;Ziegler and Dèzes, 2007). This led to: (1) depositional differences of Mesozoic and Cenozoic sediments along fault zones, (2) mechanical heterogeneities in the deformed sedimentary cover and (3) substantial vertical offsets of the basement and the cover. In this paper, we test the influence of the latter two predispositions on the evolution of the JFTB.

Lineaments in the Jura Mountains
In the Eastern Jura, folds frequently deviate from ideal cylindrical geometries, being bow-shaped, sigmoidal shaped or simply oblique to the transport direction of the JFTB in map view (Allenbach and Wetzel, 2006;Laubscher, 2008a;Meier, 2010). Such anomalous fold axes connect to lineaments with a rough NNE-SSW (Rhenish) orientation (Fig. 3a, b). The existence and implications of Rhenish lineaments in the Eastern Jura have been debated for more than a century (Laubscher, 2008a;Steinmann, 1892). Nowadays, anomalous fold orientations in the Eastern Jura are interpreted as thinskinned structures pre-conditioned by NNE-SSW striking normal faults in the basement, that are southward continuations of Eo-Oligocene URG faults (Allenbach and Wetzel, 2006;Boigk and Schöneich, 1974;Laubscher, 2008aLaubscher, , 2008bMeier, 2010;Steinmann, 1902), forming preferentially along Palaeozoic Rhenish structural trends.
Also in the central External Jura, the Mesozoic cover deformed along lineaments (see Fig. 3c), which are oblique to the general northwest-directed tectonic transport of the JFTB. In Fig. 3c, lineaments strike NE-SW, NNE-SSW (Rhenish) and ENE-WSW (Erzgebirgian). Along these lineaments, we observe alignments of short, dissected folds (Fig. 3c). Analogous to the situation in the Eastern Jura  (Ustaszewski and Schmid, 2006). See Fig. 1a for location. b) Crosssection across the Ferrette Zone (Ustaszewski and Schmid, 2006).

Basement-fault induced cover structures
To the west of the JFTB, in the autochthonous Bresse Graben close to Cormoz (Fig. 1a), Bergerat et al. (1989) interpreted an ECORS seismic line, showing a relatively steep normal fault of 75° in the crystalline basement, above which a reverse fault disrupts the Mesozoic cover (Fig. 5a). There is also a minor extensional collapse in the Jurassic series, above the higher western basement block. The Triassic salt system accommodates offsets in the basement (Fig. 5a). Normal faults in the basement that induce reverse faults in the cover are also documented in the URG (Ustaszewski et al., 2005), north of Ferrette (Fig. 5b, for location see Fig. 1a). Note that the cross-sections in Fig. 5 show extensional systems in connection with graben formation of the ECRIS, but the dominant structures in the Mesozoic cover are reverse faults (Fig. 5). We also like to point out that basement faults in Fig. 5a do not continue as a single fault into the cover. The salts in Fig. 5a distribute basement induced deformation, i.e. decoupling (e.g. Withjack and Callaway, 2000), causing different structures in the cover than in the basement. In order to test the influence of basement faults on the structures in the JFTB, we performed a series of analogue model experiments, in particular testing different throws and strikes of basement faults. Fig. 6 Sketches illustrating different ramp types of a regional décollement above pre-existing vertical steps in the basement. a) Schematic drawing showing different types of upward ramps (after Apotria et al., 1992) above corresponding basement steps.
The angle α specifies the angular deviation of an oblique step from a frontal step and therefore, the angle α is 0° for a frontal step and 90° for a lateral step. Transport directions are indicated with white arrows. b) A low-angle thrust-ramp accommodates a vertical upward step. c) A frontal downward ramp covers a vertical downward step in the basement.

Material properties in analogue models
We apply a 1:20,000 model-to-nature scaling ratio for our brittle-viscous models, so that 70 mm thick dry quartz sand simulates ca. 1400 m of uniform brittle Mesozoic cover. One model tested a reduced cover of 40 mm (corresponding to 800 m in nature). The sand has a homogeneous density (ρ) of ca.
1560 kg/m 3 when sieved from 30 cm height (Klinkmüller et al., 2016), its grain size (ø) ranges between 60-250 µm and its angle of internal friction is 36.1° (Zwaan et al., 2016(Zwaan et al., , 2018a, see Table 1). The sand surface is flattened with a scraper at every centimetre during model preparation. One model is scanned using X-Ray Computed Tomography (CT), for which we create a layering visible on CTimagery by adding a very thin corundum layer (< 1 mm) on top of the sand pack after each scraping.
The corundum sand has a density (ρ) of 1890 kg/ m 3 pure halite (Bergerat et al., 1990, see Table 2). The density profile of our models also prevents diapirism, taking account of the general absence of salt diapirs in the JFTB. Table 1 Material properties after Panien et al. (2006) and Zwaan et al. (2016Zwaan et al. ( , 2018a.

Granular materials Quartz sand Corundum sand
Grain size range (

Model scaling
Analogue models are scaled with the equation (Eq. 1, Hubbert, 1937;Ramberg, 1981) where σ * , ρ * , g * and L * represent the ratios of stress, density, gravity and length between the model and the natural example. The gravity ratio is g * = 1 and the density ratio ρ * corresponds to ca. 0.6 ( In nature, the viscosity of pure dry halite is around 10 18 Pa·s, whereas the effective viscosity of halite with small grain sizes at high temperatures is 10 17 Pa·s, ranging up to 10 20 Pa·s for large grain sizes at low temperatures (van Keken et al., 1993). Present-day maximum overburden of the basal décollement level in the Jura Mountains hardly exceeds 2.3 km (e.g. borehole Treycovagnes-1 in Sommaruga, 1997;Sommaruga et al., 2012), which at a temperature gradient of 35°C / km calculates to relatively low maximum temperatures of about 80°C. In addition, the Triassic salt systems of the Jura Mountains commonly contain various amounts of clay, anhydrite and gypsum, which further increases the viscosity of the salt. We therefore aim to simulate a rock salt décollement with an elevated viscosity (> 10 18 Pa·s).
The maximum displacement of the central Jura Mountains is about 30 km (Affolter, 2004;Laubscher, 1965;Philippe et al., 1996). Since the main Jura formation stage lasted roughly 8 Ma (Becker, 2000), we estimate an average natural displacement velocity of 3.75 mm per year. We adjust the model velocity accordingly to 10 cm/h to obtain a velocity ratio (v * ) of 2.34·10 5 (Table 2). This translates to a strain rate ratio (  (Bergerat et al., 1990, see positions in Fig. 1a). The densities given for the décollement (ρd) and cover (ρc) in the Jura correspond to densities of the Upper Triassic salt system and Jurassic limestone-marl layers, respectively..

Experimental set-up
The experimental apparatus is 117 cm long for frontal step experiments and 113 cm long for oblique steps and flat-base models, at a constant width of 50 cm (Fig. 7a). The sandbox frame consists of 2.1 cm thick rigid pine-plywood with two mobile parts: a horizontal mobile frontal baseplate and a vertical mobile backstop (Fig. 7b). The bottom of the apparatus includes two baseplates ( Fig. 7a, b). The baseplate underneath the backstop remains fixed whereas the horizontal mobile baseplate to the front can be moved vertically with maximum offsets of 20 mm (400 m in nature) upwards or downwards. An upward offset of the frontal baseplate constitutes an upward step in the direction of transport, whereas a downward offset produces a downward step (Fig. 6b, c). By using different baseplate configurations, the orientation of the plate contact is varied, specified as angle α that gives the deviation from a frontal step (Fig. 7a). This allows the simulation of abrupt frontal steps (α = 0°) or oblique steps (α = 30˚ and 45˚), analogous to steps in the basement that pre-date JFTB tectonics (Fig. 6).
The mobile backstop can be moved horizontally towards the front of the apparatus, with a maximum displacement (D max ) of 39 cm ( Fig. 7a, b), corresponding to a shortening of 7800 m in nature. Models featuring frontal steps (except model U10 with a 10 mm upward step, Table 3) are terminated at D = 35 cm. Note that we define the left and right hand side of the model apparatus with regard to the displacement direction of the backstop (Fig. 7a). The backstop is installed 1.3 cm above the fixed baseplate, so that only the sand is compressed without actively squeezing the PDMS/corundum-sand mixture (Fig. 7b).
The sidewalls, front wall, backstop and baseplates are covered with a PVC foil (formerly "Alkor" foil With this experimental set-up, we focus on testing discrete oblique and frontal steps. It is beyond the scope of this study to model the evolution of the whole Jura arc above a fractured basement, or to test local variations of the cover and basal décollement (e.g. thickness or rheological changes). The granular material simulating the Mesozoic cover of the JFTB does not allow for modelling overturned forelimbs of anticlines, since the sand collapses at slopes higher than ~35°. Our models do not include redistribution of granular material by erosion and sedimentation, which can affect the tectonic evolution of structures (e.g. Bonnet et al., 2008;Caër et al., 2018).

Modelling procedure
Every experimental run starts with a flat baseplate setting. The PDMS/corundum-sand mixture is filled in and a minimum of two days of resting time ensures an even distribution of the viscous layer. The sand layers are sieved in afterwards. A 4 x 4 cm grid of dark corundum sand (Table 1) is added to the final sand surface, which allows tracking surface displacements and rotations (Fig. 7c).
Models that contain a step are conducted in two phases: first, the mobile plate is moved upwards or downwards (max. 20 mm, Fig. 7d). In the second phase (or first and only phase for models without a step), the backstop pushes the sand across the produced step (Fig. 7d). There was no resting time between the two phases. A total of 19 models were completed (Table 3). The models are classified into a downward-step (D), flat-base (F) and an upward-step (U) series (Table 3).
J o u r n a l P r e -p r o o f

Model analysis
Bird's-eye perspective photographs of the model surface evolution are taken at 1-minute intervals, which translates to a displacement of 1⅔ mm between each image (about 33 m in nature). Videos of all models, except model U20-45°-CT, are provided in the supplementary materials (Schori et al., 2020). After every model run, the sand is removed and the surface of the viscous layer is

Flat-base models
The flat-base models F1 and F2 serve as a reference for the other models that include a baseplatestep. The first structure in the flat-base models is an imbricate stack that forms in front of the backstop (referred to as backstop imbricate BI, see Fig. 8a-c). The subsequent thrust sheets T1 and T2 forming in front of the BI show leading pop-up structures (Fig. 8c). The BI and thrust sheet T1 are closely spaced and in an evolved stage of T1, its backthrust interferes with the previous structure (Fig. 8c). In each flat-base model, thrust sheets T1 and T2 are identical in their internal architecture and dimension.
In order to compare the structural wedge-development of models, wedge-propagation plots are used Oblique view photo of model F1. c) Interpreted cross-section sketch across flat-base model F2 at ~32 cm displacement, illustrating principal structures and parameters. Structural units BI, T1 and T2 are accentuated in different shades of grey. T1 is in an evolved stage, backthrusting the imbricate stack BI whereas T2 just emerges. Note that the wedge front and the length of structures are measured at the model surface. d) Wedge-propagation plot illustrating the difference between flat-base models F1 and F2, of which the latter has a thicker viscous layer that reduces the basal friction. A sudden increase of L is caused by the formation of new structures (BI, T1 and T2). Note that lengths of thrust sheets (lT1 and lT2) between models differ. BI adds thrustslices by frontal accretion, which leads to a jagged graph between X = 24 and 28 cm. BI: Backstop imbricate-stack, T1, T2: Thrust sheet 1 and 2, lT1, lT2: length of thrust sheets T1 and T2 at the surface.

Evolution of structures related to basement step formation
During vertical movement of the mobile base plate in phase 1, a major reverse fault rooting at the upper step-edge commonly forms in the brittle sand cover. Sub-parallel to that, a minor reverse fault with less offset is seen on CT imagery of model U20-45°-CT, rooting at the lower step edge (Fig. 9a).

Frontal downward-steps
Frontal downward-step models (Fig. 10a-c) illustrate that with increasing throw of downward steps, thrust sheet T1 forms at less displacement (D) of the backstop and additionally, T1 becomes increasingly larger. In comparison to flat-base model F1, our downward steps therefore lead to a J o u r n a l P r e -p r o o f notable forward advancement of the deformation front ( Fig. 10a-c). In an evolved stage however, localisation of deformation at the step prevents downward ramping onto the lower baseplate and inhibits an overall propagation of deformation in comparison to flat-base models (Fig. 10b, c).
After the step-controlled front of T1 has formed in models D10 and D20, backthrusts appear in a backbreaking sequence (see videos in supplementary material, Schori et al., 2020). The overall structure at the front of T1 corresponds to a fault-bend fold with a single thrust-ramp, where backthrusts nucleate (Fig. 11a). The thrust-ramp at the front of T1 roots at the upper edge of the baseplate step and continuous upwards, over the sand of the lower baseplate (Fig. 11a).

Frontal upward-steps
The small upward step of 5 mm (100 m in nature) in model U05 does not trigger thrusts and the overall wedge propagation plot of model U05 is quite similar to flat-base model F1 (Fig. 10d). Upward steps of 10 and 20 mm height (corresponding to 200 and 400 m in nature) localise deformation (step-controlled thrusts), and the deformation front is attracted forwards leading to an enlarged thrust sheet T1 (Fig.   10e, f). In all our upward-step models, a long viscous ramp smoothens the abrupt step between the two baseplates (Fig. 11b) enabling a propagation of deformation onto the higher baseplate (Fig. 11b).
Backthrusting becomes increasingly prominent with higher steps (Fig. 10e, f). For one thing, stepcontrolled leading backthrusts (Fig. 9b) precede the formation of thrust-sheet T1 in models U10 and U20 (Fig. 10e, f). For another thing, the backthrust of the pop-up structure shows concentrated displacement (Fig. 11b) and remains constantly active during shortening of T1. Wedge-propagation plots of models U10 and U20 (Fig. 10e, f) show jagged graphs at the front of thrust sheet T1. This is due to thrust slices that are added at the toe of T1, nucleating beyond the step above the higher frontal baseplate (Fig. 11b). The accretion of thrust-slices distributes deformation among several thrust planes so that individual fore-thrusts absorb notably less displacement than the long-time active J o u r n a l P r e -p r o o f backthrust (Fig. 11a). The overall structure at the front of thrust sheet T1 in model U10 is a pop-up with dominant backthrusting (Fig. 11b) and an imbricated front.
In contrast to downward steps (Fig. 10a-c), thrust sheet T1 in the upward step models forms at about the same displacement (D) of the backstop as observed in flat-base model F1 (Fig. 10d-f) and therefore, forward advancement of the deformation front is due to enlarged thrust sheets only. Note that frontal-step models do not form thrust sheet T2, except for model U05, where the low upward-step of 5 mm does not trigger thrusts (Fig. 10d).

Oblique downward-step model
Model D10-30° is the only experiment featuring an oblique downward-step (10 mm, at α = 30°, Fig.   12). No downward ramp forms at the transition from the upper to the lower baseplate. Instead, a large fault-bend fold develops that thrusts the sand on the lower baseplate (Fig. 11a). Due to the oblique step, we observe important lateral variations. The deformation front sufficiently close to the oblique step is advanced forward between Y = 0 to 30 cm in Fig. 12a, attracted by the step. As the step controls the deformation front of this segment, the thrust is parallel to the step (Fig. 12a). In flat-base model F1, the position of the regular thrust-front of thrust-sheet T1, which is unaffected by steps, is at X = 10 cm (Fig. 12a). The downward-step model forms a relay thrust that connects the regular thrust position and the forward promoted, step-controlled thrust front. This leads to an initial front of T1, which is a composite of two thrusts that are about 84° oblique to each other (Fig. 12a). We observe an overall angular deformation front that tapers in the direction of transport. Note that in model D10-30˚, no thrust is perpendicular to the direction of displacement (Fig. 12a-c). Fig. 12 Top-view photos showing the evolution of structures in model D10-30°, which has a 30° oblique downward-step with a throw of 10 mm. a) The front of thrust sheet T1 nucleates at the step between Y = 0 to 30 cm and therefore, it is parallel to the step. b) A thrust slice is added on the right hand side of the model subsequently. c) Individual oblique thrusts connect to a single thrust front. Due to its large size, the fault-bend fold collapses at the centre forming a graben, parallel to the oblique step.
In our experiment, deformation is step-controlled on the left side (Y = 0 to 30 cm) and afterwards, a thrust slice is added on the right-hand side (Fig. 12b, Y = 25 to 50 cm). The reverse fault remaining from baseplate-offset in phase 1 is passively transported forwards without significant rotation (see lateral position 0 to 25 cm in Fig. 12a, b). In an evolved stage of the experiment, a new graben J o u r n a l P r e -p r o o f Journal Pre-proof structure forms due to extensional collapse roughly in the centre of the large oblique fault-bend fold (Fig. 12c). This graben is sub-parallel to the downward oblique step.
In summary, we observe three step-parallel structures, which are (1) a reverse fault related to vertical baseplate offset in phase 1, (2) an oblique thrust-front of T1 on the left (Fig. 12a, b) controlled by the oblique step during phase 2 and (3) extensional collapse normal faults forming in response to localisation of deformation at the oblique step in phase 2 (Fig. 12c).

Oblique upward-step models
In the two oblique upward-step series with oblique angles (α) of 30° and 45°, we vary the step throw between 5 and 20 mm (Fig. 13). All oblique upward-step models form oblique viscous-ramps between the lower and higher baseplates (Fig. 11b). Above these continuous ramps, different oblique structures form in the brittle sand cover. These structures reveal step-controlled thrust segments that nucleate near the baseplate step and are therefore parallel or sub-parallel to the step. With increasing step throw, step-controlled thrusts on the sand surface are more numerous (Fig. 13).

30° oblique-step series
At a low oblique upward-step height of 5 mm (100 m in nature), thrust sheet T1 remains a continuous structure with virtually no imbrication (Fig. 13a). The frontal thrust of T1 in Fig. 13a is step-controlled on the left, roughly between Y = 0 to 15 cm. In an evolved model stage, this leads to a continuous, bow-shaped front of thrust sheet T1 (Fig. 13a, late stage). Note that flat-base model F1 (Fig. 8a) is notably less bow-shaped than U05-30° (Fig. 13a), illustrating that the bow shape is not a mere cause of boundary effects.
Model U10-30° (Fig. 13b), with a 30° oblique-upward step and a throw of 10 mm (200 m in nature), develops thrust slices that pinch out laterally and reveal curved fronts, resulting in distinct crescentshaped structures (Fig. 13b, late stage). These thrust-slices are added at the toe of the wedge, alternatingly on the left and right hand side. Crescent-shaped thrust-slices nucleate in the viscous layer on the upper plate, commonly beyond the step. With increasing throw of the oblique step, curved thrusts become less numerous in favour of angular thrusts (Fig. 13c, d). Therefore, at step throws of 15 and 20 mm (300 to 400 m in nature), individual thrust-slices are reminiscent of triangles ( Fig.   13c,d). The left side of angular thrusts (left limb of triangle) mirrors the orientation of the step and nucleates at the upper step-edge, thus representing a step-controlled feature (Fig. 13c, d).
At the highest oblique upward-step of 20 mm (400 m in nature), step-controlled thrust segments are initially sub-parallel to the step (Fig. 13d, early stage), but rotate clockwise with progressive deformation (Fig. 13d, late stage). The thrust segment between Y = 20 to 25 cm in Fig. 13d (early stage), forms as a relay oblique-thrust that is ca. 80° oblique to the step and also distinctly oblique to the direction of transport (Fig. 13d). The relay thrust connects the step-controlled thrust front above the upper baseplate with the trailing regular thrust-front on the lower baseplate.

J o u r n a l P r e -p r o o f
Journal Pre-proof

Fig. 13 Evolution of step-controlled structures with increasing throws for 30° (a-d) and 45° oblique upward-steps (e-h). a) A 30°
oblique step with a low throw (5 mm) leads to a continuous structure which is step-controlled between Y = 0 to 15 cm. b) With increasing step throw, imbrication increases and thrusts are curved. c) Angular thrusts with step-controlled left sides are accompanied by curved thrusts. d) At a 30° oblique step and a high throw of 20 mm, the thrust front between Y = 0 to 20 cm is step-controlled, whereas on the right between Y = 25 to 50 cm, the regular thrust front lags behind. A relay oblique thrust connects both fronts. e-g) At 45° oblique upward-steps with throws of 5 to 15 mm, the front of T1 is continuous but asymmetric, being step-controlled on the left side. h) At a high 45° oblique step with a throw of 20 mm, imbrication of T1 occurs. features ( Fig. 13e-g, early stage). Only at a high step throw of 20 mm (400 m in nature), the 45° oblique-step model shows imbrication of thrust sheet T1 (Fig. 13h).

Difference between 30° and 45° oblique steps
We observe a fundamental difference between 30° and 45° oblique-step models at intermediate step heights of 10 and 15 mm (200 to 300 m in nature, Fig. 13b, c, f, g). The overall wedge-length evolution plots of 30° oblique-step models show a notable influence of the step (Fig. 14a, b), whereas the wedge-length plots of 45° oblique-step models resemble flat-base model F1 (Fig. 14c, d). For one thing, this difference is because 30° oblique steps cause a strong frontal imbrication of thrust sheet T1 at a step height of 10 and 15 mm (Fig. 13b, c), which is not the case for 45° oblique steps with the same throws (Fig. 13f, g). For another thing, instead of adding thrust-slices to T1, deformation propagates forwards in the 45° models and a new thrust-sheet T2 forms similar to flat-base models ( Fig. 14c, d). inactive after only 1 to 3 cm displacement and subsequently, they are dissected during imbrication of thrust sheet T1 (Fig. 13c, d, late stage). Since leading backthrusts are step-controlled features that hardly experience subsequent rotation (Fig. 13d, f), they constitute excellent indicators for the orientation of oblique steps, even in an evolved stage of models (Fig. 13).

Lineament rotation
Step-controlled thrusts start forming on the left side of our oblique-step models and append sequentially along the step (Fig. 15a). This is typically accompanied by crescent-shaped thrust slices that are added on the left wedge front (Fig. 15a, b), seemingly to compensate an asymmetric deformation front, and resulting in an overall thrust front that becomes more perpendicular to the direction of transport. When triangular thrust-slices append, their step-controlled left segments connect to a single lineament (Fig. 15a) that shows an overall clockwise rotation about a vertical axis over time.
This rotation is strongest for high upward steps in models U20-30° and U20-45°, where we measure a rotation of 12° and 15°, respectively of step-controlled fore-thrusts (see Fig. 15a, b). Table 4 shows observed rotations at different step throws. In contrast to step-controlled fore-thrusts, step-controlled leading backthrusts do not reveal any noticeable rotations (see Fig. 15a, b, sequence number 2). Fig. 15 Line drawings of top-views illustrating the relative sequence of structures in models U20-30° and U20-45°. Structures are (1) reverse faults forming in phase 1, (2) step-controlled leading backthrusts, (3) initial pop-up structures and (4) first triangular thrust-slices. Subsequent structures vary in between the two models. a) In the model U20-30° featuring a 30° oblique step, structure (5) is another triangular thrust-slice, (6) and (7) follow as crescent-shaped thrust-slices whereas structures (8) are thrusts associated with small buckle folds. b) In comparison to this, at the 45° oblique step of U20-45°, structures (5) to (7) are crescent-shaped thrust-slices.

J o u r n a l P r e -p r o o f
J o u r n a l P r e -p r o o f Table 4 Overview of observed clockwise rotations of step-controlled fore-thrusts over time. A negative throw denotes a downward step whereas positive throws indicate upward steps. α: Obliquity of step. Step

CT-scanned oblique upward-step model
Model U20-45°-CT is a CT-scanned rerun of model U20-45°, featuring a 20 mm (400 m in nature) upward-step of 45° obliquity. The CT scans allow the reconstruction of cross-sections at different model stages (Fig. 16). Raw CT-scan sections interpreted for Fig. 16 are compiled in the supplementary material of this article (Schori et al., 2020).
Structures of phase 1 (Fig. 16a) are fragmented laterally and at depth during horizontal shortening, and transported forwards embedded in the front of thrust sheet T1 (compare to Fig. 16c and d). In contrast to this, late collapse structures due to sediment stacking are situated at the back of thrust sheet T1 and not at the front (e.g. Fig. 16d, cross-section B4 and C4).
The front of thrust sheet T1 is step-controlled on the left side (Fig. 16b, cross-sections D2 and E2) whereas the right hand side is yet unimpeded by the step (Fig. 16b, cross-section A2). By comparison, pop-up structures in cross-sections D2 and E2 (Fig. 16b) lie slightly more frontwards, attracted by the upward oblique-step, which results in an overall curved thrust front, visible on the top view image in Fig. 16b.
The thrust-slice added in cross-sections B3 to E3 (Fig. 16c) corresponds to a triangular thrust-slice in top-view images of Fig. 16c, d. Cross-sections C3 and D3 (Fig. 16c) illustrate that on the left side, the 16d). In upward-step models, the forward flowing viscous material accumulates in two locations. First, thickening occurs underneath the front of the pop-up structure of T1 (Fig. 16d, section A4). Secondly, upward steps cause an accumulation of viscous material on the lower baseplate, between the step and the roots of BI, thereby forming a long viscous ramp with moderate inclination (Fig. 16d, C4 to E4). Since the length of the ramp depends on the spacing between the step and BI, the length of the viscous ramp varies laterally (Fig. 16d, C4 to E4) and furthermore, over time with ongoing displacement of the backstop (Fig. 16a to d). Note that if the step-throw exceeds the thickness of the viscous layer (e.g. step height 15 or 20 mm, 300 or 400 m in nature), the formation of the viscous ramp ultimately leads to in-situ uplift of a few millimetres. Above the edge of the upper baseplate, the viscous layer thins out down to a few millimetres (Fig. 16d, C4 to E4).

Oblique upward-step model with reduced cover
Model U20-45°-RC tests a 20 mm (400 m in nature) oblique upward-step with a reduced sand cover of 40 mm instead of 70 mm (Fig. 17a, b). The thinner sand cover leads to shorter and more numerous thrust sheets and generally smaller thrust-slices in comparison to a cover thickness of 70 mm (Fig.   17c), which is directly related to the geometric relationship between layer thickness and thrust-fault dip-angle (Allemand and Brun, 1991). However, similar step-controlled features as observed in models with a high oblique upward-step and a sand thickness of 70 mm are present, such as triangular thrustslices and a step-controlled leading backthrust (Fig. 17a). In the model, an overall sigmoidal structure is the result of triangular thrust-slices of variable size that grow sequentially (Fig. 17a, b). The overall step-controlled front of connected triangular thrust-slices rotates by about 15° clockwise (Fig. 17b).
J o u r n a l P r e -p r o o f  (Fig. A.1 and Fig. A.2), and the original data are compiled in the supplementary material (Schori et al., 2020).
An overview of experiments and the first order influence of upward and downward steps is summarised in (Fig. 18). Flat-base models (Fig. 18a, b) develop two similar thrust sheets (T1 and T2) with notable less stacking of granular material than observed in step models and there is no imbrication of thrust sheets. Vertical offset of baseplates (Fig. 18c)  sheet T1 (Fig. 18d). An oblique downward-step (Fig. 18e) controls the formation of fault-bend folds with oblique frontal thrusts. Numerical and analogue downward-step models of Caër et al. (2018) show that with sufficient shortening at step-controlled thrusts under the right mechanical conditions (high friction angle of cover sediments and low friction angle of décollement layers), deformation can propagate onto the lower baseplate.
The localisation of deformation at upward steps is less efficient than for downward steps. In comparison to downward steps, upward steps ( Fig. 18f) feature smooth viscous ramps that enable a propagation of deformation onto the higher baseplate. Thickening of viscous material is therefore indicative for upward steps. The brittle cover shows dominant backthrusts and frontal imbrication (Fig.   18f). Oblique upward-step models (Fig. 18g) reveal a more complex imbrication of brittle structures, featuring asymmetric thrust slices and step-controlled thrust segments that rotate up to 15° about a vertical axis. In contrast to this, oblique viscous ramps remain continuous structures.
Higher oblique steps increase deformation localisation and step-controlled thrust segments become more important, overall leading to angular structures (Fig. 18g, U20-30° and U20-45°). 45° oblique steps cause less localisation of deformation (less imbrication and faster wedge-propagation) than 30° oblique steps. Consequently, the efficiency of deformation-localisation does not depend on step height alone, but also on the angle of obliquity (α) of the upward-step.
In summary, the effect of deformation localisation at steps is (i) stronger at high steps than at low steps, (ii) stronger at downward steps than at upward steps and (iii) stronger at frontal steps than at oblique steps.
J o u r n a l P r e -p r o o f 7.2 Comparison to previous oblique-ramp models Fig. 19 a) Oblique and lateral-ramp analogue sandbox-models of Philippe (1995). b) Cross-section through the lateral ramp model illustrating a basal wooden plate that enforces a rigid ramp of 30°. A thin layer of microbeads provides the basal décollement. c) Model U20-45°-RC featuring a thin brittle layer shows remarkably similar structures as the 45° oblique upwardstep model of Philippe (1995) in an early stage.
Previous models on oblique ramps in the Jura Mountains were carried out by Philippe (1995, see Fig. 19a). In contrast to our models, the baseplate geometry in Philippe (1995) provides an enforced, rigid ramp with an inclination of 30° (Fig. 19b). The brittle cover is thinner above the higher basal wooden plate (Fig. 19b) and no precursory baseplate-offset phase is simulated. Furthermore, the basal décollement is modelled using a thin layer of microbeads (Fig. 19b) and therefore, no viscous flow occurs. However, the top-view pattern of the early stages of the 45° oblique-ramp model of Philippe (1995) is very similar to the pattern forming in the 45° oblique-step model with reduced brittle cover U20-45°-RC of this paper (Fig. 19c). In addition to 45°-oblique rigid ramps, Philippe (1995) tested 60° oblique ramps as well as lateral ramps (Fig. 19a), showing that with increasing obliquity, triangular thrust-slices disappear in favour of thrust-sheets with a rather straight front perpendicular to the transport direction.
Our experiments demonstrate that deformation localisation at oblique steps is the reason for triangular structures. Therefore, the absence of such triangles in 60° oblique-step models (Fig. 19a) suggests reduced deformation localisation with increasing step-obliquity (α). Extrapolated, this suggests that lateral steps show the least localisation of deformation and frontal upward-steps the most.

J o u r n a l P r e -p r o o f
Journal Pre-proof

Local and regional rotations of structures
Step-controlled fore-thrusts in our models reveal local rotations about a vertical axis of up to 15° over time (e.g. Fig. 17, Fig. 15), related to localisation of deformation at individual baseplate steps. The sketch in Fig. 20 shows a schematic overview of rotations of step-controlled thrusts deduced from observations in our models, with a transport direction and basement-step orientations matching typical cases in the JFTB. Note that step-controlled oblique structures are local anomalies within a regional trend of structures. The Central and Eastern Jura show a regional clockwise rotation of 7° to 8° (Laubscher, 1965(Laubscher, , 1961), whereas regional counter-clockwise rotation is proposed for the Southern Jura (Affolter, 2004;Philippe, 1994). These regional rotations are more broadly linked to lateral displacement gradients across the Jura arc (Affolter, 2004). Regional and local rotations in the JFTB have a different origin; they were competing and have to be distinguished. Fig. 20 Rotation scheme for thin-skinned thrusts that are controlled by pre-existing oblique upward-steps, derived from observations of our analogue models and those by Philippe (1995). The sketch fits typical directions that may occur in the JFTB, with basement faults striking NNE-SSW, ENE-WSW and NW-SE, and a transport direction to the NW.

Natural examples of step-controlled structures
Our analogue models show that sufficiently high oblique steps ultimately lead to step-controlled oblique structures, such as sigmoidal, curved or angular thrusts viewed from above (e.g. Fig. 17b, Fig.   18g). In the tectonic map of Fig. 21a, we point out exemplary zones in the JFTB, where we propose oblique structures, associated with pre-existing basement faults that acted as structural controls.
These zones are described with sketches ( Fig. 21b) outlining the proposed tectonic linkage between basement faults and corresponding thin-skinned structures in the cover. Note that sketches in Fig. 21b illustrate thin-skinned structures of the JFTB, which formed during Mio-Pliocene times above a basal décollement in Triassic evaporites. Therefore, oblique structures were also transported several kilometres north-westwards after their formation, at least in the Internal Jura (see displacements indicated in Fig. 21a).
J o u r n a l P r e -p r o o f and does not represent a late structural feature of the Jura Mountains. This agrees with previous suggestions by Laubscher (1961).
Zone II to the north of the town of La Chaux-de-Fonds (Fig. 21a) shows angular and overlapping thrusts. Therefore, similar to zone I, we propose NNE-SSW striking oblique basement-steps that triggered the observed structures in the Mesozoic cover (Fig. 21b). The sketch for zones I and II in Fig.   21b correlates with structures of analogue models of Philippe (1995) featuring 60° oblique upwardsteps, where laterally confined triangular thrust-slices are absent (Fig. 19a).

Zone III -Oyonnax-Vuache
Zone III (Fig. 21a) lies in the Southern Jura Mountains and an enlargement is depicted in Fig. 22a. The Vuache fault zone in the southern part of Fig. 22a reveals oblique ramps that are indicative of stepcontrolled structures and a NW-SE trending basement step. A connection to pre-JFTB tectonic features has also been proposed in previous studies (Laubscher, 1981;Philippe et al., 1996).
A distinct triple-nappe system south of Oyonnax comprises two triangular nappes and a more crescent-shaped nappe (Fig. 22a), which bears resemblance to our analogue model U20-45°, featuring a 20 mm (400 m in nature) 45° oblique upward-step (Fig. 22b). In analogy to model U20-45°, we propose a NW-SE trending basement fault for the structures that formed in zone III (Fig. 21b).
Within the 45° oblique-step series, triangular thrust-slices occur at the highest throw of 20 mm only ( Fig. 13h) and accordingly, we infer a basement step for the natural example of at least 400 m. This is in agreement with a seismic line interpretation across Oyonnax-Champfromier , which reveals a basement high with a throw of several hundreds of metres, SE of Oyonnax (Fig. 22c).
Whether this basement high formed by post-Jura inversion or rather represents a horst that predates the Jura Mountains (see tectonic solutions adressed in Guellec et al., 1990), remains debated.
However, as thrust geometries in Fig. 22a are strongly suggestive of an oblique-step in the basement, our study supports the latter solution. In addition, a direct comparison between the natural example and our analogue model (Fig. 22a, b) suggests a probable structural evolution as depicted in Fig. 22a.
Note that a pre-JFTB normal fault in the basement does not rule out a subsequent inversion of the basement in recent times.
J o u r n a l P r e -p r o o f

Zone IV -Chasseral Anticline
Zone IV (Fig. 21a) lies in the Internal Jura and comprises the Chasseral anticline, which contains an oblique fold-segment in respect to JFTB transport directions (see magnification in Fig. 23a). The overall structure in Fig. 23a is remarkably similar to our 45˚ oblique upward-step model with a reduced cover U20-45°-RC (Fig. 23b). The comparison between the model (Fig. 23b) and the natural example ( Fig. 23a) suggests that the curved shape of the Chasseral anticline is connected to a NNE-SSW trending basement fault. This supports previous suggestions by Lüthi (1954), Laubscher (2008b) and Meier (2010). We furthermore deduce a high basement step of about 400 m for the natural example, as the best-fit model U20-45°-RC features a 20 mm upward step. Based on our model (Fig. 23b), we also deduce a rough relative sequence of structures for the natural example (Fig. 23a). Note that the basement structure that controlled the oblique Chasseral anticline lies between 9.5 to 16 km to the southeast, according to different displacement models of Affolter and Gratier (2004) and Philippe et al. (1996), respectively.
J o u r n a l P r e -p r o o f

Summary of natural structures
We propose the presence of pre-JFTB normal faults in the basement with NW-SE (Hercynian) and NNE-SSW (Rhenish) orientations that provided trigger lines for thin-skinned structures in the Internal Jura (Fig. 22, Fig. 23). The orientations of these basement faults suggest Palaeozoic Variscan fault systems, which were reactivated during Mesozoic and Cenozoic times. This led to important basement offsets that controlled the thin-skinned deformation in Miocene and Pliocene times. The tectonic events that come into question for fault reactivations are the opening of the Alpine Tethys rift system in Jurassic times (Stampfli et al., 1998;Ziegler, 1988b), Eo-Oligocene graben formation in connection with the ECRIS (Illies, 1972;Michon, 2000) and Eocene to Miocene evolution of the Alpine flexural foreland basin (Burkhard and Sommaruga, 1998;Laubscher, 1992).
Our models suggest that basement offsets in the examples in the Internal Jura (Fig. 22, Fig. 23) must reach at least 400 m in order to control the formation of triangular nappes or thrust-slices during Mio- Lacombe and Bellahsen, 2016;Madritsch et al., 2008;Ustaszewski and Schmid, 2007). Recent inversion may be related to tectonic underplating in connection with an equilibration of the Alpine wedge (Mosar, 1999) and/or lithospheric long-wavelength buckling (Bourgeois et al., 2007;Lefort and Agarwal, 1996).

Conclusions
Thin-skinned deformation of the JFTB during Mio-Pliocene was locally pre-conditioned by steps (faults) in the pre-Mesozoic basement (Laubscher, 1961). By means of brittle-viscous analogue sandbox-models, we study the formation of step-controlled structures. Our models involve the formation of steps between rigid baseplates, overlain by a viscous layer at the base of a brittle sand layer. The brittle sand layer is subsequently pushed horizontally across the step. We subsequently compare our model results to natural examples. The following important observations characterise the deformation style:  Initial vertical baseplate offset induces reverse faults and extensional collapse structures in the brittle cover that nucleate at the step and above the higher plate. A reactivation of these initial structures is not observed during subsequent horizontal compression.
 During subsequent horizontal compression, pre-existing steps in the rigid footwall trigger stepcontrolled thrusts, because they localise deformation in a different position due to stress perturbations. Localisation of deformation results in a fast forward propagation of deformation in an early stage but inhibits propagation later.
 Downward steps lead to the formation of fault-bend folds with considerable duplication of the brittle cover. In contrast, upward steps control pop-up structures with imbricated fronts, dominant backthrusts and long viscous ramps with moderate inclinations.
 The effect of deformation localisation at baseplate steps is stronger at high steps than at low steps, stronger at downward steps than at upward steps and stronger at frontal steps than at oblique steps.
Imbrication and angular thrusts, pointing into the direction of transport, are more abundant with increasing step-height. Angular thrusts are composed of two oblique-thrust segments, with one side that is step-controlled and the other side forming a relay thrust that connects to the regular the case of oblique upward-steps) and over time. Accumulation of viscous material occurs in front of upward steps and towards the front of the wedge. Oblique viscous ramps are large continuous structures whereas the overlying brittle layer may show strong imbrication.
We find good correlations of model structures with oblique structures in the Internal Jura ( Fig. 22 and   Fig. 23). Correlations allow inferring the rough orientation and throw of pre-existing basement steps in nature that controlled thin-skinned deformation. In particular, our models support a NNE-SSW striking oblique upward-step with a throw of 400 m that controls the Chasseral anticline and a NW-SE striking oblique upward-step with a throw of 400 m controlling the Oyonnax-Vuache nappe system.
Stepcontrolled structures of the Internal Jura are transported several kilometres north-westwards, away from the basement step that induced deformation. Therefore, step-controlled structures need a palinspastic restoration to identify the position of a corresponding basement step. Model-to-nature comparison of step-controlled oblique structures can also be a powerful tool to assess the relative age of basement structures. An important basement high was for example documented on a seismic line across the Internal Jura , but interpretations did not unambiguously reveal whether the basement high formed before or after the JFTB. Correlations with our models strongly support that the basement high predates the JFTB and controlled characteristic oblique structures in the cover. Examples from the JFTB highlight the importance of analysing structures of fold-and-thrust belts in all three dimensions, and our model results may be of use for interpreting fold-and-thrust belts from around the globe.