Elastic anisotropies of rocks in a subduction and exhumation setting

. Subduction and exhumation are key processes in the formation of orogenic systems across the world, for example, in the European Alps. For geophysical investigations of these orogens, it is essential to understand the petrophysical properties of the rocks involved. These are the result of a complex interaction of mineral composition and rock fabric including mineral textures (i.e. crystallographic preferred orientations). In this study we present texture-derived elastic 10 anisotropy data for a representative set of different lithologies involved in the Alpine orogeny. Rock samples were collected in the Lago di Cignana area in Valtournenche, in the Italian Northwestern Alps. At this locality a wide range of units of continental and oceanic origin with varying paleogeographic affiliations and tectono-metamorphic histories are accessible. Their mineral textures were determined by time-of-flight neutron diffraction at the Frank Laboratory of Neutron Physics at the JINR in Dubna, Russia. From these data the elastic properties of the samples were calculated. The data set includes 15 representative lithologies from a subduction-exhumation-setting. In subducted lithologies originating from the oceanic crust, the elastic anisotropies range from 1.4 to 5.0% with average P-wave velocities of 7.01-8.24 km/s and VP/VS-ratios of 1.71-1.76. In the metasediments of the former accretionary prism the elastic anisotropies range from 4.7 to 8.2 %. This tectonic setting displays average P-wave velocities of 6.47-7.23 km/s and VP/VS-ratios of 1.60-1.76. Continental crust which is incorporated in the collisional orogen shows elastic anisotropies ranging from 1.8 to2.8 % with average P-wave velocities of 20 6.42-6.51 km/s and VP/VS-ratios of 1.56-1.60. Our results suggest that mafic and felsic rocks in subduction zones at depth may be discriminated by a combination of seismic signatures: lower anisotropy and higher VP/VS ratio for mafic rocks, higher anisotropy and lower VP/VS ratio for felsic rocks and metasediments. geodynamic models the texture-derived properties are the P- and S-wave velocities, the VP/VS-ratio and the elastic anisotropies

The diffraction data are processed by Rietveld Texture Analysis (RTA) (Von Dreele, 1997;Matthies et al., 1997, Lutterotti 95 et al., 1997. This method allows for the determination of the textures of all mineral phases in the sample by calculation of the orientation distribution function (ODF) using the E-WIMV algorithm (Lutterotti et al. 2004, Chateigner et al. 2019). The RTA is performed in the software Mineral Analysis Using Diffraction (MAUD: Lutterotti, 2010, Wenk et al., 2010. For further discussion on the functions and limitations of MAUD, see the publication of Wenk et al. (2012). For RTA, knowledge of the mineral assemblage is of great importance. For this reason, thin sections of all samples were made and 100 analyzed by polarization microscopy.
To calculate the P-wave velocities and other elastic properties of the samples the program BEARTEX (Wenk et al. 1998) was used. To calculate the bulk-rock elastic moduli, the orientation distribution function (ODF) of the mineral phases, knowledge of their single-crystal elastic constants and their volume fractions is necessary. The elastic moduli were then calculated using the Christoffel equation (1) (Christoffel, 1874;Mainprice and Humbert, 1994): 105 In it the phase velocity of P-, S1-or S2-waves is represented by V, the Kronecker delta by and the Christoffel Tensor by Γ . The Equation of the Christoffel Tensor (2) is as follows: In it represents the density-corrected elastic (stiffness) moduli and 1 represent the direction cosines of the wave 110 propagation direction.
For the averaging, it has been shown by Hill (1952) that the widely used averaging schemes of Voigt (1928) and Reuss (1929) give upper and lower bounds of the elastic moduli, respectively. Furthermore, other averaging schemes have been proposed by Matthies and Humbert (1995) and Matthies (2012). However, since the values of the parameters are still uncertain, the Voigt averaging scheme was used consistently throughout this study. It is known that the recalculated 115 velocities are therefore maximum velocities only.
The ODF of all phases are exported from MAUD. The volume percentages of the phases are estimated from thin-sections and calculated in MAUD by RTA. The single-crystal elastic constants were taken from literature (quartz: Heyliger et al., 2003;albite: Brown et al., 2006;muscovite: Vaughan and Guggenheim, 1986;calcite: Dandekar, 1968;dolomite: Humbert & Plique, 1972;hornblende: Aleksandrov and Ryzhova, 1961; epidote/zoisite/clinozoisite: Aleksandrov et al., 1974;garnet: 120 Zhang et al., 2008;glaucophane: Bezacier et al., 2010;omphacite: Bhagat et al., 1992). In some cases, single-crystal elastic constants are not available for all phases (chlorite, actinolite, barroisite and clinozoisite), so similar minerals were selected as approximations. For samples containing chlorite we have selected the single-crystal elastic constant of muscovite (Vaughan and Guggenheim, 1986) to approximate chlorite. For clinozoisite the single-crystal elastic constants of epidote (Aleksandrov et al., 1974) were selected, for the blueschist-facies amphibole barroisite those of glaucophane (Bezacier et al., 2010) and for actinolite those of hornblende (Aleksandrov and Ryzhova, 1961). In most cases, the contents of the phases for which approximations were necessary, were below 10% of the total sample volume. At this low content, the CPOs of these minerals are of low relevance for the physical properties of the rock (Mainprice and Ildefonse, 2009). In RTA, amphiboles are generally very difficult to distinguish, since they are extremely similar crystallographically. For this reason, the crystallographically best fitting amphibole structure was picked for refinement, which is not always the chemically fitting 130 one in the sample. This has an insignificant influence on the calculated CPO due to the crystallographic similarity of the amphibole group (C2/m space group; Reynard et al. 1989).
It is known that at low confining pressures, close to the Earth's surface, the elastic properties of rocks are greatly influenced by microcracks and crack fabrics, as well as a shape preferred orientation (SPO). This has been shown in multiple studies (Kern et al., 2002, 2008, Pros et al., 2003, Ivankina et al., 2005, Ullemeyer et al., 2006. Further studies have even shown 135 that the influence of microcracks is still visible up to a confining pressure of 1 GPa (Christensen, 1974, Ullemeyer et al., 2011. In this study the elastic properties are calculated solely on the basis of mineral texture and single-crystal data without taking microcracks into account. Since the microcracks are closed at great depth, the calculated elastic properties can be seen as representative of these conditions.

Composition and microfabrics of the samples studied 140
In this study we have related the subduction and exhumation settings to several structural positions, each containing characteristic lithologies (figure 2). These positions are (1) the oceanic crust of the subducting plate, (2) the sedimentary cover of the subducting plate as well as the accretionary prism and finally (3) the continental crust of a microcontinent. The eclogite, blueschist, amphibolite and greenschist of the Zermatt-Saas zone belong to the first of these. The metasediments of the Combin zone represent the second position and finally the gneisses of the Dent-Blanche represent the position of the 145 deformed continental crust. However, the samples do not always represent the peak conditions of metamorphism but contain minerals grown during various steps of the P-T path. In most samples presented here a clear chlorite-forming greenschist overprint is visible, even if the rock originated at higher metamorphic conditions. This is especially true for the metabasic rocks in this study, which often contain hornblende, actinolite and chlorite, indicating the amphibolite-and greenschist-facies overprint, respectively.  (table 1). The stretching lineation is formed by elongated barroisite and feldspar grains in the foliation plane. On the microscopic scale a foliation formed by barroisite can clearly be seen. The grain-size distribution is bimodal, with a finer grained matrix of amphibole and feldspar, and larger, strongly retrogressed, inclusion-rich and fractured garnets therein. 165

Amphibolite
The composition of the amphibolite (MJS26; table 1)

Greenschist
In the greenschist MJS36, the main constituent minerals are chlorite (30 vol.-%), clinozoisite (30 vol.-%), and actinolite (20 vol.-%). Feldspar is visible throughout the matrix (table 1, figure 3b). Actinolite defines the foliation of the sample, while chlorite grains show a weaker preferred orientation. In the foliation, amphibole grains can be seen forming a stretching 175 lineation. The greenschist is fine-grained and shows a very homogeneous mineral distribution.

Metasediments
The composition of the metasediments (MJS18, MJS20, MJS22) is variable (table 1) shows a pronounced SPO of mineral fishes aligned in the foliation plane. Where chlorite is present it is interwoven with white mica between the quartz and feldspar layers.

Eclogite 200
In the eclogite (MJS17) omphacite shows a distinct CPO and has the highest texture strength of all phases in the sample with an F2 index (Bunge, 1982; see complete list in appendix A) of 3.49. The c-axes display the highest intensity clusters with a maximum parallel to the stretching lineation. The a-axes display maxima in a girdle perpendicular to the stretching lineation.
The omphacite b-axes form a weak girdle perpendicular to the stretching lineation with a clear maximum normal to the foliation plane (figure 4). The glaucophane in eclogite has a weak CPO with an F2 index of 1.19. The c-axes display a 205 maximum parallel to the stretching lineation. The a-axes develop a girdle perpendicular to the stretching lineation with a maximum paralleling the kinematic y-axis. The glaucophane b-axes display two maxima, the stronger of which being normal to the foliation plane and a secondary maximum parallel to the stretching lineation (figure 4). Clinozoisite has weak textures with an F2 index of 1.01. The c-axes display a maximum paralleling the kinematic y-axis. The main b-axes maximum is parallel to the stretching lineation (figure 4). The texture of white mica in the eclogite is quite weak (F2 index of 1.02). 210 However, the basal planes show an alignment to the foliation plane (figure 4). Garnet and albite show random textures.

Blueschist
In the blueschist (MJS16), barroisite displays the strongest textures with an F2-index of 2.36. The strongest textures are seen in the c-axes, which exhibit a maximum parallel to the stretching lineation. The a-axes in barroisite display maxima in a girdle distribution perpendicular to the lineation direction with a maximum parallel to the kinematic y-axis. The b-axes display maxima in a similar girdle perpendicular to the stretching lineation with highest densities normal to the foliation plane (figure 4). Clinozoisite and epidote CPOs are very similar and relatively weak. Both show an F2 index of 1.02 and baxes with a maximum almost parallel to the stretching lineation. A-axes and c-axes show girdles perpendicular to the lineation (figure 4). Garnet and albite exhibit random textures.

Amphibolite 220
In the amphibolite MJS26, the strongest textures are found in the hornblende, with an F2 index of 1.62. The hornblende caxes show the strongest CPO with a distinct maximum parallel to the stretching lineation. The b-axes are distributed in a girdle normal to the stretching lineation, with a maximum normal to the foliation plane. The a-axes also display a girdle normal to the stretching lineation (figure 5). For chlorite, the F2 index is 1.10. Chlorite c-axes show a girdle perpendicular to the lineation with a maximum parallel to the foliation normal (figure 5). In albite the texture strength is low, with an F2 225 index of 1.07. However, a-axes display a girdle perpendicular to the stretching lineation and a weak maximum normal to the foliation plane while b-axes display a broad, weak maximum parallel to the stretching lineation (figure 5). The textures in garnet and clinozoisite are nearly random.

Greenschist
In the greenschist MJS36, the strongest textures are found in chlorite with an F2 index of 1.34. Chlorite basal planes are 230 oriented parallel to the foliation plane (figure 5). The textures of actinolite are of intermediate strength with an F2 index of 1.21. The c-axes display a strong maximum parallel to the stretching lineation and a weak girdle distribution in the foliation plane. The a-axes show highest intensities with a maximum normal to the foliation plane. The b-axes display a slightly weaker texture with a maximum parallel to the foliation normal, connected by a weak, irregular girdle normal to the stretching lineation. Furthermore, secondary maxima are slightly oblique to the stretching lineation (figure 5). Albite has an 235 F2 index of 1.18 and a-axes display the strongest CPO with a maximum parallel to the foliation normal. The b-axes are concentrated in a girdle in the foliation plane with weak maxima therein. The c-axes display a weak CPO with diffuse maxima (figure 5). Clinozoisite shows a random CPO.

Metasediments
In the metasediment samples (MJS18, MJS20, MJS22) the CPOs are quite distinct and often stronger than in the samples 240 In the calcschist MJS22 all mineral phases display medium texture strength (appendix A). In calcite the F2 index is 1.34. The c-axes display a strong maximum parallel to the foliation normal. A-axes are concentrated in the foliation plane with a 260 maximum parallel to the stretching lineation (figure 6). In quartz the textures are stronger than in calcite, with an F2 index of 1.54. The quartz c-axes display a weak sinistrally rotated girdle with a strong maximum offset from the kinematic y-axis by ca. 30°. A weaker density cluster is found perpendicular to this girdle. The a-axis maximum is located close to the periphery of the pole figure slightly sinistrally rotated away from the x-direction. The poles of the prism and rhomb planes display weaker maxima in girdles paralleling the c-axes girdle (figure 6). Chlorite has an F2 index of 1.52 and displays a strong 265 alignment of its basal plane in the foliation (figure 6).

Gneiss
In the gneisses (MJS34, MJS35) CPOs are similar to those of the metasediments (figures 6 and 7).
In MJS34 the muscovite and chlorite display basal planes parallel to the foliation plane. The textures are strongest for chlorite with an F2 index of 2.15. Muscovite has an F2 index of 1.27. Quartz shows an F2 index of 1.18 and its c-axes are 270 distributed in small circles around the foliation normal. The a-axes display a weak and diffuse girdle distribution in the foliation plane. In albite a very weak (F2 index of 1.03) but distinct CPO can be seen. The a-axes display a maximum parallel to the stretching lineation (figure 7).
In MJS35 both muscovite and chlorite display clear alignment of the basal planes in the foliation plane. Their F2 indices are 1.13 and 1.53, respectively. The CPOs in quartz are weaker (F2 index of 1.06) than in MJS34. C-axes display an asymmetric 275 crossed girdle distribution. The a-axes display diffuse intensity clusters at the periphery around the stretching lineation. In albite the CPO is nearly random (F2 index = 1.01), with a faint maximum of the a-axes in the lineation direction (figure 7). https://doi.org/10.5194/se-2021-3 Preprint. Discussion started: 1 February 2021 c Author(s) 2021. CC BY 4.0 License.

Elastic anisotropy
P-wave anisotropies for the various lithologies were calculated from the ODFs of constituent mineral phases, single-crystal data of the respective phase and their volume fraction within the sample using the software BEARTEX (Wenk et al., 1998). 280 Graphical representations of S-wave velocity distributions are not shown in this study, since the differences between maximum and minimum velocities in a sample are on the order of 0.01 km/s and deemed insignificant.

Blueschist
In the blueschist the average VP is 7.6 km/s, with a maximum of 7.8 km/s and minimum of 7.5 km/s. The average VS is 4.5 290 km/s, resulting in a VP/VS-ratio of 1.71. The P-wave anisotropy is of medium strength with 3.7 % and defined by barroisite

Amphibolite 295
The amphibolite has an elastic anisotropy of 3.1 %. The VP range from 7.0 km/s to 7.3 km/s, with an average of 7.1 km/s.

Greenschist 300
In the greenschist MJS36, the average VP is 7.0 km/s, with a maximum of 7.2 km/s and a minimum of 6.8 km/s. The average VS is 4.1 km/s, resulting in a VP/VS-ratio of 1.73. The P-wave anisotropy is high with a value of 5 %. The highest VP are
In the dolomitic micaschist (MJS18) the average VP of 7.2 km/s is high for a metasedimentary sample. The VP anisotropy is 6.7%, the highest determined in the metasediments. The average VS is 4.1 km/s, resulting in a VP/VS-ratio of 1.76. 310 Maximum VP is parallel to the stretching lineation lowest VP is found parallel to the foliation normal (figure 8).
In the calcareous micaschist (MJS20) the average velocities are lower than in the previously described sample. The average VP is 6.7 km/s and the average VS is 4.1 km/s, resulting in a VP/VS-ratio of 1.63. The P-wave anisotropy in the calcareous micaschist is very strong at 8.2%. Highest VP is concentrated in the foliation plane, lowest VP is found parallel to the foliation normal (figure 8). 315 In the calcschist (MJS22) the average VP is 6.5 km/s with an anisotropy of 4.7%. The average VS is 4.0 km/s, resulting in a VP/VS-ratio of 1.60. Highest P-wave velocities are concentrated in a broad girdle parallel to the foliation plane with lowest P-wave velocities perpendicular to it (figures 6 and 8).

Gneiss
In the gneiss samples (MJS34, MJS35) the P-wave anisotropies and velocity-distributions are quite similar (figure 8). 320 In MJS34 the average VP is 6.5 km/s and the average VS is 4.1 km/s resulting in a VP/VS-ratio of 1.6. The P-wave anisotropy of 2.8% is low, and the velocity-distribution displays fast velocities in the foliation plane and a minimum velocity perpendicular to it (figure 8).
In MJS35 the average VP is slightly lower than in MJS34 at 6.4 km/s. The average VS is 4.1 km/s, resulting in a VP/VSratio of 1.56. The P-wave anisotropy is 1.8%. Similar to MJS34, the highest velocities are found in the foliation plane. 325 However, a weak maximum is visible almost parallel to the stretching lineation (figure 8).

Discussion
The elastic anisotropy of rocks is influenced by several factors, among which are its mineral composition, its grain fabric, possible microcracks, and the CPO of its constituent minerals. The CPO constrains a large part of the anisotropy and mostly results from rock deformation. According to Mainprice and Ildefonse (2009) only mineral phases which make up more than 330 10% of the sample, have an important influence on the elastic properties of the rock in question. In our calculations we have, however, included all mineral phases down to a volume fraction of 5 %. The CPOs of omphacite, glaucophane, barroisite and hornblende have a strong effect on the elastic anisotropy of the eclogite, the blueschist and the amphibolite. In previous studies, the CPO geometry of omphacite has been interpreted as being the result of corresponding strain geometry (Helmstaedt et al., 1972;Kurz et al., 2004;Neufeld et al., 2008;Müller et 340 al., 2011;Keppler et al., 2016). L-type (lineation dominated) fabrics are interpreted as the result of constrictional strain, Stype (foliation dominated) fabrics as the result of flattening strain and the transitional fabrics as indicators of plain strain deformation (e.g. Helmstaedt et al., 1972). SL-and LS-type fabrics have been described by Godard and Van Roermund (1995). Omphacite in MJS17 displays an L-type fabric, with c-axes paralleling the stretching lineation and a-and b-axes paralleling the kinematic y-and z-axes, respectively (figure 4). Keppler (2018) suggests that this correlation of strain 345 geometry to CPO geometry found in omphacite is applicable for amphiboles as well. The amphibole CPO in three of the four metabasites (eclogite, blueschist, and amphibolite; figure 4 and 5) can therefore also be interpreted as L-type or LS-type fabrics, since the c-axes are aligned in lineation direction, while a-axes and b-axes are closer to a girdle distribution perpendicular to the lineation. In the greenschist the actinolite c-axes display a maximum paralleling the stretching lineation with a girdle extending into the foliation plane, while the a-axes display a maximum perpendicular to the foliation plane 350 (figure 5), which points to a transitional SL-type fabric.
Since the omphacite, glaucophane and barroisite c-axes are parallel to both the stretching lineation and one another, the assumption can be made that they formed in the same prograde deformation event. A further possibility would be retrograde mimetic overgrowth of omphacite by glaucophane (McNamara et al., 2012). There is considerable debate on timing of CPOformation in eclogites, with proponents of formation during subduction, during exhumation, or both, including concurrent 355 switches in strain geometry (Zulauf 1997;Kurz et al., 2004;Kurz, 2005;Neufeld et al., 2008;Müller et al., 2011;Keppler et al., 2016;Keppler, 2018). The CPOs of clinozoisite and epidote frequently have similarities to those of omphacite and the amphiboles. The clinozoisite b-axes display a maximum paralleling the stretching lineation, matching the omphacite and amphibole c-axes. Keppler et al. (2017) suggest concordant CPOs due to opposing crystallographic axis definitions and further conclude that clinozoisite CPOs can likewise be classified in S-type, L-type and intermediated fabrics (Keppler et al. 360 2018;Puelles et al. 2017). In this case the CPOs of clinozoisite would suggest L-type fabrics hinting at constrictional strain and matching those already described for omphacite and the amphiboles. Albite preferentially displays weak CPOs in the metabasites. However, in the amphibolite and greenschist the weak albite texture of the b-and c-axes corresponds to the stretching lineation, while the a-axis maximum is aligned nearly normal to the foliation plane (figure 5). The slip-systems in plagioclase are highly complex and a correlation with the strain geometry is difficult (Hacker and Christie, 1990;Prior and Wheeler, 1999). When chlorite or white mica are present, they mostly display a strong alignment of the basal planes in the foliation plane. The similarities in the CPOs of chlorite, amphibole and albite in the greenschist suggest a formation during the same deformation, which likely took place during the retrograde exhumation.

Elastic anisotropies
The elastic anisotropies and the VP-distribution in the metabasites are primarily controlled by the influences of omphacite, 370 amphibole and chlorite. These minerals result in two variations of elastic anisotropies and VP-distributions. On the one hand there is a group of results with lower anisotropies (eclogite: 1.4 %; blueschist 3.7%; amphibolite: 3.1 %) and highest VP paralleling the lineation direction and on the other hand there is the greenschist, with a higher anisotropy (5%) and VPdistributed in a girdle in the foliation plane.
The compositions, anisotropies and VP-distributions of the first group are primarily controlled by omphacite and amphibole 375 and secondarily by clinozoisite, epidote and garnet. In the amphiboles, highest P-wave velocities in the single-crystals correlate almost perfectly with the c-axes, only deviating by a few degrees. Since the c-axes of the amphiboles in these metabasites parallel the stretching lineation, this results in a velocity maximum in lineation direction. The influence of omphacite is slightly more complicated, since in the omphacite single-crystals the highest velocity is not parallel to the caxis and instead can be found between the c-and a-axes. This discrepancy has little influence in this case since it also results 380 in a velocity maximum parallel to the stretching lineation, but it should be kept in mind when considering anisotropy strength and distribution. Epidote and clinozoisite CPOs in the blueschist likely contribute to the same pattern of high velocities in lineation direction. Even though garnet has a near random CPO and does not contribute directly to the anisotropy, the high garnet content in the eclogite sample (30 vol.-%, table 1) and the very high isotropic P-wave velocities in garnet single-crystals increase the average seismic velocities in MJS17. In the blueschist and the amphibolite in which the 385 garnet is only present to 5-10% this effect is negligible. Albite can also influence the bulk anisotropy of the metabasites due to its strong single-crystal velocity anisotropy (Brown et al., 2006;Brown et al., 2016). The highest velocities in albite are parallel to the b-and c-axes (Brown et al., 2006), both of which display weak maxima parallel to the stretching lineation in the metabasites, further contributing to the high velocities parallel to the stretching lineation ( figure 4 and 5). In the amphibolite, the chlorite basal planes display a girdle distribution perpendicular to the stretching lineation with a maximum 390 perpendicular to the foliation (figure 5). This CPO reflects a rotation of the chlorite basal planes around an axis paralleling the stretching lineation. Due to the lowest velocities in sheet silicates normal to the basal plane, the CPO of chlorite in this sample contributes to the low velocity girdle perpendicular to the stretching lineation. The eclogite contains white mica, however, due to its very low volume fraction of 5 % it is considered insignificant for the bulk anisotropy (Mainprice and Ildefonse, 2009). 395 In the greenschist the influence of chlorite dominates the composition and the anisotropies as well as the VP-distribution. In possible explanations for this. The first of these is that in many studies, blueschist samples are collected from highly deformed bands resulting in stronger anisotropies. In our study, however, a sample with a fabric that is representative for the lithology within the Alps was selected, resulting in weaker overall mineral CPOs and a lower anisotropy. Furthermore, in the two studies mentioned the amphibole in the blueschist was glaucophane, while in the case of the sample in this study it is mostly barroisite. This was addressed in part when calculating the elastic properties of the blueschist since no single-crystal 415 data for barroisite were available so the single-crystal data of glaucophane (Bezacier et al. 2010) were selected as a suitable substitute. However, glaucophane may still react differently to deformation than barroisite, possibly resulting in stronger

CPO development 430
The small circles of quartz c-axes around the foliation normals of samples MJS18 and MJS20 might indicate flattening strain (Lister and Hobbs, 1980;Schmid and Casey, 1986) (figure 6). This is also supported by the distribution of quartz a-axes, but the textures are quite weak. The c-axis pole figures of MJS18 and MJS20 are indicative of basal <a> slip and rhomb <a> slip in quartz (cf. Schmid and Casey, 1986) (figure 6). In MJS22 the quartz c-axes display an inclined single girdle pointing to simple shear. This distribution is also reflected by the quartz a-axes, prism and rhomb planes and indicative of multi-slip 435 activity, possibly a combination of rhomb <a> slip, prism <a> slip and perhaps some basal <a> slip (e.g., Schmid and Casey, 1986;Stipp et al., 2002) (figure 6). In all metasediments the basal planes of white mica and chlorite display a strong alignment in the foliation plane. In all three metasedimentary samples the CPOs of calcite c-and a-axes exhibit axial symmetry with respect to the foliation normal as a symmetry axis ( figure 6). This symmetry in calcite has first been interpreted as indicative of pure shear deformation (Wenk et al., 1987;Kern and Wenk, 1983). However, more recently, 440 there was considerable debate on this topic (De Bresser, 1989;Ratschbacher et al., 1991;Erskine et al., 1993;Burlini et al., 1998;Leiss et al., 1999;Bestmann et al., 2000). The maximum of the calcite a-axes can either parallel the stretching lineation or the kinematic y-axis or even display multiple maxima between the two (Ratschbacher et al., 1991;Punturo et al., 2005;Bestmann et al., 2000;Bestmann et al., 2006;Trullenque et al., 2006). Very few studies have focused on dolomite CPOs. Delle Piane et al. (2009) show that dolomite CPOs very closely reflect those of calcite, but deformation itself strongly 445 influence the texture development when both phases deform together. Dolomite in MJS18 displays very similar CPOs to calcite, with the same axial symmetry in the c-and a-axes (figure 6).

Elastic anisotropies
The elastic anisotropies of the metasediments (MJS18, MJS20 and MJS22) range from 4.7-8.2 % and are among the highest presented in this study (figure 8). These anisotropies directly correlate to the pronounced foliation and high mica and chlorite 450 contents found in the metasediments, as well as the strong influence of calcite and dolomite CPOs on the samples. This correspond to elastic anisotropy data from similar metasediments reported in other studies such as Weiss et al. (1999) with 4.56-6.53 %, Erdman et al. (2013) with 2.3-11.4 %, andKeppler et al. (2015) with 5.1-7.4 %.
In our study, the orientation of white mica and chlorite (10-40 vol.-%) basal planes and the very high single-crystal anisotropy of white mica, which is used for both sheet silicates, mainly contribute to the elastic anisotropy in the 455 metasediments. The VP distribution in muscovite single crystals is dominated by its highest velocity parallel and lowest velocity normal to the basal plane (figures 6 and 8). As micas preferentially align with increasing strain in the foliation, the result is a strong tendency for high P-wave velocities in the foliation plane in deformed mica-rich rocks.
The composition of the metasediments varies substantially. Calcite and dolomite combined range from 15 vol.-% up to 70 vol.-% (table 1). The CPOs of both of these minerals strongly influence the P-wave anisotropy of the metasediments. They result in high velocities in the foliation plane with a maximum parallel to the stretching lineation. Further they contribute to the velocity minima normal to the foliation plane. This is the result of the calcite single crystal data in which the c-and aaxes are the slow and fast directions, respectively (Puntero et al., 2005). The single-crystal anisotropy of dolomite is quite similar to that of calcite. Hence, both similarly contribute to the bulk anisotropy of sample MJS18. The contribution of quartz to the P-wave anisotropy is weak due to the near random CPO observed for the poles of the rhombs, which represent 465 the fastest direction in the quartz crystal. Furthermore, the three-fold symmetry of the quartz single crystal makes any influence on the velocity distribution complicated to assess. The garnet present in MJS20 displays a random texture and therewith has no significant influence on the elastic anisotropy.
The VP/VS-ratios in the metasediments range from 1.60 to 1.76. The highest ratio is found in sample MJS18 at 1.

CPO development
The small circles around the foliation normal, visible in the quartz c-axes and the distribution of a-axes of MJS34 indicate flattening strain with combined activity of rhomb <a> slip and basal <a> slip (Lister and Hobbs, 1980;Schmid and Casey, 1986) (figure 7). In MJS35, the type I crossed girdle in the quartz c-axes is indicative of plane strain with combined basal <a>, rhomb <a> slip and prism <a> slip (e.g., Schmid and Casey, 1986). White mica and chlorite CPO in both MJS34 and 480 MJS35 bear a strong relation to the foliation plane (figure 7). Albite in MJS34 and MJS35 shows very weak CPOs, but albite a-axes in both display a weak but visible maximum paralleling the stretching lineation (figure 7). Due to the random distribution of the b-axes and c-axes a clear assignment cannot be made, however this CPO pattern is closest to an axial A type, indicating that the activated plagioclase slip system is [100](010) and/or [100](001) (Satsukawa et al., 2013).

Elastic anisotropies 485
The elastic anisotropy in the gneisses MJS34 and MJS35 is primarily constrained by the CPOs of mica and chlorite (25 vol.-% and 15 vol.-% respectively, table 1) and secondarily by quartz and albite. However, the influence of quartz is far weaker than that of the sheet silicates as the pole figure maxima of quartz do not match well to the bulk elastic anisotropy diagram.
The strong single-crystal anisotropy of muscovite and chlorite in combination with their strong CPO, results in the high velocities found in the foliation plane and the corresponding perpendicular velocity minima (figures 7 and 8). The VP/VS-490 ratio of the gneiss samples ranges from 1.56 to 1.60. As in the metasediments this low ratio is the result of the very low ratio  ; table 1). The velocity distribution is very similar to that of the metasediments, with high velocities in the foliation plane and low velocities perpendicular to this plane (figure 8), but the elastic anisotropy is lower. When comparing the results of our study to those previously published, we see a very wide range of P-wave 495 anisotropy data reported in gneisses. Values range from low to intermediate anisotropies such as 1.76-8.73 % by Weiss et al. (1999), 1.72-2.99 % by Ivankina et al. (2005), 0.9-3.2 % by Ullemeyer et al. (2006), and 7.9 % by Kern et al. (2008), up to high to very high anisotropies such as 4.1-20.1 % by Erdman et al. (2013) and8.0-11.3 % by Llana-Fúnez andBrown (2012). As gneisses predominantly display weak CPOs similar to our results, our low elastic anisotropy values are common for gneisses. 500

Elastic anisotropy data in the context of a subduction and exhumation setting
In this study we examine the elastic properties of a wide array of lithologies in a subduction and exhumation setting. As previously mentioned and is visible in figure 2 we have divided this setting into several structural positions. These are the oceanic crust of the subducting plate undergoing different metamorphic facies, the sedimentary cover of the subducting plate, which is at least partly scraped off and transferred into the accretionary prism and finally the deformed continental 505 crust. This continental crust can be part of a microcontinent (as in the case of the Dent Blanche unit) or can be the continental crust on the opposite side of the ocean, which enters the subduction zone subsequently to the oceanic crust leading to continental collision. This scenario is presented in the schematic cross-section in figure 2. The structural positions vary in a range of criteria such as degree of deformation and mineral composition. The different degrees of deformation and metamorphism lead to different texture strengths and variable mineral content. When comparing the different structural 510 positions, samples can be related in order of SiO2-content, with quartz-and feldspar-rich felsic rocks on one end of the spectrum, and the olivine-rich ultramafic rocks on the opposite end of the spectrum. This has been discussed extensively in a very insightful study by Almqvist and Mainprice (2017) for the earth's crust. Generally, a trend of lower velocities in the felsic and higher velocities in the mafic to ultramafic rocks can be observed. In this study we further discuss another criterion of great importance on the elastic properties which is the influence of CPO of the constituent minerals on the elastic 515 anisotropy. This might differ from the compositional effect in the spectrum from felsic, over mafic to ultramafic. Concerning elastic anisotropies, the sheet-silicate content and its alignment caused by deformation show a pronounced influence.
During the prograde metamorphism in subduction zones the basaltic oceanic crust undergoes blueschist-facies metamorphism leading to the formation of glaucophane (in some cases barroisite), epidote, omphacite and some accessory minerals. Glaucophane and feldspar usually dominate the composition, of which glaucophane greatly influences the 520 anisotropy and the velocity distribution. Blueschists display strongly varying elastic anisotropies of 3.7-24.6 % (Cao et al., 2013;Bezacier et al., 2010). The deformation during subduction leads to CPO formation in amphibole. In most cases plane strain is dominant. The orientation of the c-axes of amphibole parallel to the stretching lineation results in the highest velocity parallel to the lineation. Dominant constriction in the blueschist of the present study produces a similar pattern (see https://doi.org/10.5194/se-2021-3 Preprint. Discussion started: 1 February 2021 c Author(s) 2021. CC BY 4.0 License.
figures 4 and 8). At greater depth the oceanic crust reaches eclogite facies metamorphic conditions. There, omphacite and 525 garnet are the rock-forming minerals (table 1) and omphacite has the strongest influence on the elastic anisotropies. The elastic anisotropies in eclogites are lower than those in the blueschists and display values of 0.4-10.2 %, due to the higher elastic anisotropy of amphiboles compared to pyroxenes (Llana-Fúnez and Brown, 2012;Keppler et al., 2015;Cao et al., 2013;Bezacier et al., 2010;Wang et al., 2009). The highest P-wave velocities remain parallel to the stretching lineation, due to the omphacite CPO yielding an alignment of [001] in lineation direction during plane strain as well as constriction (see 530 figures 4 and 8).
During exhumation the high pressure rocks frequently experience retrogression as they pass through the amphibolite facies, where they are overprinted or completely retrogressed. In the amphibolite-facies rocks, hornblende together with feldspar and epidote dominates the composition. These minerals result in anisotropies of 0.59-6.4 % (Siegesmund et al., 1989;Ivankina et al., 2005;Ullemeyer et al., 2006;Keppler et al., 2015), of which the amphiboles have the strongest influence. 535 The velocity distribution in the amphibolites is very similar to those observed in the blueschists and eclogites, with highest velocities found parallel to the stretching lineation (figure 8). Further along the exhumation path, the rocks enter greenschist facies conditions, where further retrogression is possible. In the greenschist facies, chlorite is formed and together with actinolite, epidote and feldspar, makes up the greenschists. The high chlorite content results in intermediate to high anisotropies of 5 %, presented in this study and a velocity distribution with high velocities in the foliation plane ( figure 8). 540 This is directly caused by the sheet silicate chlorite and provides greenschist VP patterns very similar to those of the metasediments and the deformed continental crust. Very few studies have been made on the elastic properties of greenschists and further research is necessary to better constrain this lithology. The present study, however, shows that greenschist facies metabasites constitute a special case concerning their elastic properties. While VP is still higher than that of metasediments and comparable to the other metabasites, the velocity distribution resembles that of the metasediments. Due to their frequent 545 occurrence in subduction settings and collisional orogens this needs to be considered in seismic investigations.
In the overlying metasediments VP anisotropies display values of 2.3-11.4 % (Erdman et al., 2013;Keppler et al., 2015). The composition of these metasediments in the present study is highly variable. However, quartz, feldspar, mica, chlorite, calcite and dolomite are the most abundant mineral phases and also those controlling the elastic properties. The velocity distribution shows highest velocities in the foliation plane with a velocity-minimum normal to it, as seen in the greenschists, due to the 550 high contents in sheet silicates as well as of calcite and dolomite.
In the deformed continental crust the composition is dominated by feldspar, quartz, mica and chlorite, represented by the gneisses in this study. In the gneisses the elastic anisotropies are similar to those observed in the metasediments, with values of 1.72-20.1 % (Ivankina et al., 2005;Ullemeyer et al., 2006;Erdman et al., 2013;Kern et al., 2008;Weiss et al., 1999)  boundaries of the predominantly mafic subducting plate and the felsic continental margins. It can further assist to better constrain the anatomy of subduction channels.

Common issues for calculated elastic properties
Precise single-crystal data are not available for all mineral phases, but our fitting represents a good approximation. Firstly, the selected substitute minerals are very similar to those in the samples, especially with respect to their crystallography. 575 Secondly, the goal of our models is not to simulate specific rock samples, but the rock types that are representative distinct sections of the subduction and exhumation cycle. Hence, for the missing mineral phases substitute single crystal data were selected, which we considered to be acceptable matches due to the metamorphic and petrological context they can be found in. As mentioned above, this was done for clinozoisite, for which we selected the single-crystal data of epidote (Aleksandrov et al., 1974), as well as for chlorite, for which we selected the data of muscovite (Vaughan and Guggenheim, 1986). 580 Furthermore, the almandine garnet single-crystal data of Zhang et al. (2008) was used for all garnet-bearing samples, even if different garnet compositions were present in the samples. A similar simplification was made for the amphiboles. In samples with glaucophane or barroisite, we used glaucophane single-crystal data by Bezacier et al. (2010), while for all other amphiboles we utilized the single-crystal elastic constants for hornblende from Aleksandrov and Rhyzova (1961).
Throughout this study the Voigt averaging scheme was used consistently. This model is based on the "equal strain" 585 assumption and represents the upper bound for the stiffness of a polycrystalline material. As a result, the velocities presented are therefore maximum velocities only. This does not correspond to the "equal stress" assumption of the Reuss model that represents the lower stiffness boundary. Neither method takes grain shape, layering, microcracks, fractures or pores into account (Vasin, et al. 2017). Other methods such as the self-consistent method GeoMIXself (GMS) by Matthies (2010;2012) can take CPOs, morphologies and SPOs of the mineral grains into account. However, only based on CPO data the 590 Voigt averaging scheme is deemed appropriate to estimate representative samples. Furthermore, the Voigt averaging scheme https://doi.org/10.5194/se-2021-3 Preprint. has widely been used on the subject and ensures therefore comparability with most other published texture derived elastic anisotropy data.
A further issue, which requires discussion is the influence of microcracks on the elastic properties of the rocks (Siegesmund and Vollbrecht, 1991;Siegesmund et al., 1993;Ullemeyer et al. 2011, Vasin et al., 20142017). The elastic anisotropies 595 presented in this study are texture-derived and do not take the presence of microcracks into account. Instead, they depend on the mineral composition, single-crystal mineral properties and the CPOs of the respective phases. This means that the calculated data reflect the elastic anisotropies of rocks at great depth, in which microcracks are closed due to high confining pressure. The influence of microcracks can well be observed in studies in which CPO-derived calculated results have been compared to elastic anisotropy measurements in laboratory experiments at high confining pressure. Some of these studies 600 have demonstrated that the influence of microcracks is still detectable up to confining pressures of 1 GPa (Christensen, 1974;Ullemeyer et al., 2011). However, with increasing pressure and closing of the microcracks, the results far better resemble the CPO-derived results and the influence of microcracks decreases significantly (Ivankina et al., 2005;Lokajicek et al., 2017;Vasin et al., 2017).

Conclusion 605
Using time-of-flight neutron diffraction, texture data from a wide range of compositionally variable and complex rocks were collected. These are representative of different positions for subduction and exhumation in a collisional orogen. From these data texture-derived elastic anisotropies and other petrophysical properties were modelled. The following conclusions can be made: 1. Composition and CPO have a strong effect on the rock texture and the related elastic properties in subduction and 610 exhumation settings. Felsic rocks with high sheet silicate and calcite or dolomite contents preferentially display the fastest Pwave velocities in the foliation plane independently of the orientation of the stretching lineation. The mafic rocks with high amphibole and pyroxene contents in this study mostly display highest P-wave velocities parallel to the stretching lineation, with exception of greenschist, in which the high chlorite content results in the fast velocities in the foliation plane. and calcareous schists display higher elastic anisotropies (1.8-8.2 %), lower VP/VS-ratios (1.56-1.76) and fast P-wavevelocities distributed in the foliation plane.