Surface and Crustal Response to Lithospheric Removal Processes: Insights from Numerical and Analogue Modeling

Surface and Crustal Response to Lithospheric Removal Processes: Insights From Numerical and Analogue Modeling

O˘guzHakan G¨o˘g¨u¸s

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy Graduate Department of Geology University of Toronto

Surface and Crustal Response to Lithospheric Removal Processes:

Insights From Numerical and Analogue Modeling

O˘guzHakan G¨o˘g¨u¸s

Doctor of Philosophy

Department of Geology

University of Toronto

2010

Geological, geophysical, and geochemical evidence indicates that a signiﬁcant portion of the continental mantle lithosphere may be absent in a number of regions near plate boundaries or plate interiors. Delamination and viscous Rayleigh-Taylor instability (“dripping”) are widely cited to account for the missing lithosphere, however these removal processes are poorly constrained.

This thesis examines the dynamics of delaminating and dripping mantle lithosphere, in particular focusing on the response of the crust to underlying lithospheric removal.

Using forward computational models, I explore whether certain (surface) geological observables may be diagnostic of either removal mechanism. Surface topography associated with delamination has a broad zone of uplift above the lithospheric gap and a mobile zone of subsidence at the delamination hinge, whereas with dripping lithosphere, the topographic expression is symmetric and ﬁxed above the mantle lithosphere downwelling. The pattern of surface crustal deformation is also distinctly asymmetric with delamination compared to dripping lithosphere.

Expanding on these results, I investigate whether present day geologicalgeophysical observables in Eastern Anatolia are consistent with delamination of the mantle lithosphere. Experimental results demonstrate that well-developed plateau uplift, syn-convergent extension, and crustal thinning in the central part of the Anatolian plateau are consistent with a topographic proﬁle at longitude

42◦E and a geologically interpreted zone of syn-convergent extension in eastern Anatolia.

With three-dimensional physical scaled analogue modeling experiments, I consider the process of oceanic plate subduction evolving into continental delamination. Model results show that slower plate convergence with retreating ocean lithosphere subduction can develop into delamination,

ii whereas for the experiments with higher plate convergence, the crust above the consumed mantle lithosphere becomes accreted on the retro-plate similar to ﬂake tectonics. The results suggest that delamination is a process analogous to subduction retreat; however, delamination involves decoupling of the retreating mantle lithosphere slab from the buoyant continental crust.

iii Acknowledgements

I agree after all these years that the doctoral work has been a long and stressful yet, an interesting walk/run of both scientiﬁc successes and failures. However I have developed many other qualiﬁcations along the way; such as, patience, commitment, hard-work, listening(before you an- swer the questions) and defending your work. I owe thanks to many people for helping me to prepare this thesis along the way, people who gave me enough encouragement, support/friendship and motivation to do my best.

First and foremost, I’d like to thank my advisor, Russ Pysklywec, for many reasons. Since the day I arrived the department in September 7th of 2004, he made me feel that I knew I have been at the right place. Russ‘s encouragement, scientiﬁc guidance most importantly his friendship has been immensely helpful. Over the years, I appreciate his special knack for making even the hard problems much easier. The sentence of “Come on O˘guz,you know this” will forever ring in my head. Thank you Russ.

I would also like to acknowledge the members of my dissertation committee, who have all helped me immensely throughout my PhD. Jerry Mitrovica and Alexander (Sandy) Cruden have given me insightful recommendations and given me hard moments on my defenses. However it was not long before I realized contributions were to my work overall. I learned a lot from both of them. I would also like to thank Pierre Robin for his additional encouragement and interest in my research. I owe thanks to Claudio Faccenna from RomaTre university who kindly invited me to his lab in Rome to do my analogue modeling studies there and for many fruitful scientiﬁc discussions. I am also thankful to Clint Conrad from the University of Hawaii for serving as my external examiner and for his recommendations on my thesis.

I would not have come this far if I had not have support/encouragement from my friends at the department. I thank David Boutelier for countless hot-scientiﬁc discussions and sharing his graduate school experiences with me. The PhD programme has never been vacation!.Thanks for reminding this David. I thank Sergio Gelcich, Guido Seraﬁni, Chris Charles, Christoph Schrank,

Lisa Tutty, Duane Smythe, Abin Das, Guillaume Barlet, for giving me a bit of laughter amidst the many times that I felt like “it was enough”. Ed Spooner was the ﬁrst faculty member I met, when

I arrived the department. Our conversations on interesting topics (e.g science, archeology, history,

iv soccer etc) has continued until then. Thanks Ed. My friends/colleagues in the department made the last six years of work both fun and memorable for me. Great memories with you guys.

Outside of department, I spent most of my ”oﬀ time” with friends in Tango, discussing politics- philosophy etc. I have met some of my closest friends in Toronto during the hard and sometimes lonely years of my PhD. I would like to acknowledge Brian Janeway, Hakan Toksoy-Julia Graham,

Inan¸c-S¸ima˙ Yıldırım, Ilker-Ceren˙ So˘gukpınar,Tayfun-Aslı Akku¸s, Defne Berkin, G¨uln¨urOzan,

Bedia Tatlısu and Nesime A¸skınwhom thought me that greatest lesson I learned is friendship in whereever you are and independent of how many degrees you have.

Last but not least, I thank my family members my father Kemal Gö˘gü¸s, my mom Sevgi Gö˘gü¸s, my sister Inci Gö˘gü¸sAras, my brother in law Ekmel Aras and my nephew Can Deniz Aras for their support, continued love and encouragement. They gave me the support I needed to study on this

PhD thesis and to see its ultimate completion from thousands of miles away. I feel very fortunate to have family like you. Well, who would guess that I would meet with such an amazing girl-Corrie

7 months ago at the library and without her companion/support this thesis at the end may not have ﬁnished.

v “The Sleep of Reason Produces Monsters”

Francisco Jos´ede Goya y Lucientes, 1746-1828

vi Contents

1 General Introduction 1 1.1 Introduction ...... 2 1.2 Physical Conditions for Mantle Lithosphere Removal ...... 3 1.2.1 The Unstable Mantle Lithosphere Layer ...... 3 1.2.2 Pre-existing Weak Zone between the Crust and Mantle Lithosphere . . . . . 8 1.3 Thesis Structure ...... 9 1.4 Statement for Authorship ...... 10

Bibliography 12

2 Near Surface Diagnostics of Dripping or Delaminating Lithosphere 19 2.1 Abstract ...... 20 2.2 Introduction ...... 20 2.3 Experimental Results ...... 25 2.3.1 Model Description ...... 25 2.3.2 Dripping and Delaminating Mantle Lithosphere ...... 28 2.3.3 Surface Topography ...... 31 2.3.4 Moho Temperatures ...... 34 2.3.5 P-T Histories ...... 39 2.3.6 Crustal Deformation ...... 42 2.4 Conclusions and Discussion ...... 45

Bibliography 49

3 Mantle Lithosphere Delamination Driving Plateau Uplift and Synconvergent Extension in Eastern Anatolia 54 3.1 Abstract ...... 55 3.2 Introduction ...... 55 3.3 Modeling Delamination ...... 58 3.4 Surface Topography and Crustal Deformation ...... 60 3.5 Delamination Beneath Eastern Anatolia ...... 63 3.6 Conclusions and Discussion ...... 65

Bibliography 66

4 The Surface Tectonics of Mantle Lithosphere Delamination Following Ocean Lithosphere Subduction: Insights From Physical Scaled Analogue Experiments 69 4.1 Abstract ...... 70

vii 4.2 Introduction ...... 70 4.3 Experimental Design ...... 75 4.3.1 Model Description ...... 75 4.4 Experimental Results ...... 84 4.4.1 Experiment DEL-12 (Vp = 0.25 cm/min) ...... 84 4.4.2 Experiment DEL-13 (Vp = 0.38 cm/min) ...... 89 4.4.3 Experiment DEL-11 (Vp = 0.5 cm/min) ...... 92 4.4.4 Experiment DEL-10 (Vp = 1.0 cm/min) ...... 95 4.4.5 Delamination and collision: Retreating Plates Beneath the Continental Crust. 98 4.5 Conclusions ...... 101 4.6 Discussions ...... 102 4.6.1 Comparison with previous models and natural systems ...... 102

Bibliography 105

5 Synthesis of Principle Conclusions 112 5.1 Near Surface Diagnostics of Dripping or Delaminating Lithosphere ...... 113 5.2 Mantle Lithosphere Delamination Driving Plateau Uplift and Synconvergent Exten- sion in Eastern Anatolia ...... 114 5.3 The Surface Tectonics of Mantle Lithosphere Delamination Following Ocean Litho- sphere Subduction; Insights from Physical Scaled Analogue Experiments ...... 115

Bibliography 117

Appendix 120

A Description of the Numerical Model 120

viii List of Tables

4.1 Scaling relationships of the reference experiment (DEL-12) ...... 79 4.2 Experimental parameters of all materials ...... 80

ix List of Figures

1.1 Mechanical thickening of the lithosphere ...... 7

2.1 Delamination vs drip model ...... 24 2.2 Model set-up ...... 27 2.3 Models’ evolution ...... 30 2.4 Surface topography plots ...... 33 2.5 Moho temperatures (DEL) ...... 36 2.6 Moho temperatures (DRIP-1) ...... 37 2.7 Moho temperatures (DRIP-2) ...... 38 2.8 P-T history ...... 41 2.9 Crustal deformation ...... 44

3.1 Tectonic map of Eastern Anatolia ...... 57 3.2 Evolution of the model ...... 59 3.3 Surface topography and crustal thickness ...... 62 3.4 Comparison of observables (model vs nature) ...... 64

4.1 Neotectonic evolution of Eastern Anatolia ...... 74 4.2 Analogue model set-up ...... 78 4.3 Delamination in analogue modeling ...... 83 4.4 Model Results Experiment DEL-12 ...... 88 4.5 Model Results Experiment DEL-13 ...... 91 4.6 Model Results Experiment DEL-11 ...... 94 4.7 Model Results Experiment DEL-10 ...... 97 4.8 Migration of delaminating hinge ...... 100

x Chapter 1

General Introduction

1 1.1 Introduction

Various lines of geological, geophysical, and geochemical evidence have been used to infer that large-scale removal of the mantle lithosphere (i.e., the sub-crustal portion of the lithosphere) has occurred in places around the Earth. Continental mantle lithosphere removal has been postulated for the India-Eurasia collision zone (Bird, 1978; Houseman et al., 1981; Meissner and Mooney,

1998; Kosarev et al., 1999), the Colorado Plateau (Bird, 1979), the Altiplano-Puna plateau in the central Andes (Kay and Kay, 1993; Beck and Zandt, 2002), the Eastern Anatolian collision zone

(Keskin, 2003; Gö˘gü¸sand Pysklywec, 2008a; S¸engöret al., 2008), the Alboran Sea-Rif Betics (Seber et al., 1996; Platt et al., 1998; Fadil et al., 2006), New Guinea (Cloos et al., 2005), the Apennines-

Tyrrhenian (Reutter et al., 1980; Channell and Mareschal, 1989), the Carpathian-Pannonian region

(Houseman and Gemmer, 2007; Lorinczi and Houseman, 2009), and the Sierra Nevada (Ducea and

Saleeby, 1998; Jones et al., 2004; Zandt et al., 2004).

The mantle lithosphere may be colder, and therefore denser than the underlying sublithospheric mantle (i.e., gravitationally unstable) in tectonically younger areas and for younger lithospheric compositions particularly in Phanerozoic times (Poudjom Djomani et al., 2001; O’Reilly et al.,

2001; Lee et al., 2005) (See subsection 1.2). The dense mantle lithosphere may then be removed

(either wholly or partially) and sink into the sublithospheric mantle. When mantle lithosphere removal occurs, the overlying crust can experience a) rapid regional surface uplift, b) high heat

ﬂow, c) magmatism/melt production due to asthenospheric upwelling, and d) localized extension or contraction.

Two main geodynamic removal mechanisms have been proposed: 1) lithospheric “dripping”, whereby the dense mantle lithosphere descends as a viscous Rayleigh-Taylor-gravitational instability

(Houseman et al., 1981); 2) delamination, where the mantle lithosphere may peel away from the overlying crust along the Moho and sink into the sublithospheric mantle (Bird, 1978, 1979). As a ﬁrst-order distinction, the former mechanism involves an “un-platelike” behaviour of mantle lithosphere, while the latter mechanism is associated with a semi-rigid “platelike” response of the mantle lithosphere layer. The term “delamination” is used following its deﬁnition by (Bird, 1979).

Often the term delamination is used more generically to denote any type of mantle lithosphere

2 removal event but this introduces ambiguity relating to what are quite diﬀerent physical processes.

Previous geodynamic modeling studies primarily describe rheological, thermal and mechanical properties/conditions for dripping and delaminating lithosphere (Houseman et al., 1981; House- man and Molnar, 1997; Conrad and Molnar, 1997; Molnar et al., 1998; Schott and Schmeling, 1988;

Conrad and Molnar, 1997, 1999; Conrad, 2000; Elkins-Tanton, 2005; Morency et al., 2002; Morency and Doin, 2004). Several modeling studies of dripping mantle lithosphere have been used to quan- tify the basic evolution of surface topography and crustal deformation for various rheological and buoyancy controlling factors (Neil and Houseman, 1999; Pysklywec and Cruden, 2004). However, the response of the surface to delamination, the distinction between how the crust responds to delamination and dripping, and the quantitative interpretation of these events in speciﬁc geological contexts are not well understood. In addition, recent geological/geophysical studies of the young

Alpine orogenic chains (Tethysides) in the Mediterranean region resulting from collision of the Afro-

Arabian plate with the Eurasian plate (Reutter et al., 1980; Wortel and Spakman, 2000; S¸eng¨or et al., 2003, 2008; Fadil et al., 2006) and the India-Eurasia collision (Kosarev et al., 1999) may suggest that delamination of the mantle lithosphere is linked to preceding ocean subduction events at convergent plate boundary systems. However, previous studies do not clarify the dynamics of this relationship.

In this thesis, my objective is to constrain quantitatively the varying surface/crustal responses of large-scale lithospheric removal processes in both intra-plate and convergent orogenic settings by using forward geodynamic modeling techniques. The thesis consists of ﬁve chapters where chapters

2, 3 and 4 are written as separate journal articles related to the themes described above. Overall, this thesis aims to provide new insight into how lithospheric removal works and how such deep lithosphere processes may drive surface tectonics.

1.2 Physical Conditions for Mantle Lithosphere Removal

1.2.1 The Unstable Mantle Lithosphere Layer

In this work, the intention is not to explore and test the variable physical conditions/parameters

(e.g., lithospheric temperatures, density, rheology) that initiate lithospheric removal processes.

3 Rather, the experimental results focus on the surface and crustal (near-surface) responses to mantle lithosphere removal assuming it has been initiated. However, it is worthwhile to give some insight into the geological/physical conditions/assumptions in our models in the context of how delamination or Rayleigh-Taylor type instabilities of the mantle lithosphere are thought to be initiated.

A primary assumption of mantle lithosphere removal events is that continental mantle lithosphere is denser than the sublithospheric mantle. It has been suggested that Phanerozoic mantle lithosphere is approximately 10-30 kg/m3 denser than the underlying asthenosphere, owing to its colder temperature and possibly because of compositional differences (e.g. Stacey, 1992; O’Reilly et al., 2001; Poudjom Djomani et al., 2001). Houseman et al. (1981) assume density differences between the mantle lithosphere and mantle reach up to 100 kg/m3 in their Rayleigh-Taylor instability experiments. Burov (2007) refers to a 20 kg/m3 density difference between the mantle lithosphere and sublithospheric mantle for gravitational instability experiments along continental rifted margins. It is well documented that seismic velocities (P and S- wave velocities) are systematically higher in the mantle lithosphere than in the sublithospheric mantle, suggesting that the former is denser or chemically distinct or both. At the lithosphere-asthenosphere boundary (LAB), this takes the form of a sharp decrease in seismic velocities (LeFevre and Helmberger, 1989; Rychert et al., 2007). If the seismic velocity change is assumed to be related to density, this suggests a potentially unstable density structure.

Despite the apparent instability of the mantle lithosphere due to thermal buoyancy, the continental mantle lithosphere seems to be predominantly stable. It has been postulated that lithospheric roots/mantle lithosphere became stabilized with respect to convective disruption by the addition of basalt-depleted, low-density mantle material (peridotites) (Jordan, 1978). Jordan (1978) suggests that the stabilization of the Archean cratons in the interval of 2.0 − 2.8 Ga is the earliest formation of the thick/stable lithospheric roots or “continental tectosphere”. It is proposed that this tectosphere overcomes the thermal contraction/density instability by chemical density stabilization of the evolved lithosphere.

More recent geochemical studies interpret that the depletion of the continental lithosphere is not constant through geologic time. O’Reilly et al. (2001) calculates density proﬁles for Archean,

4 Proterozoic and Phanerozoic mantle lithosphere based on heat ﬂow and xenolith compositions, and they suggest that newly formed mantle lithosphere becomes less depleted from Archean to

Proterozoic and Phanerozoic times, in terms of Al and Ca content and in Mg and Fe/Al. They conclude that there is a secular evolution of lithosphere composition from depleted Mg-rich low density Archean mantle to more fertile denser Phanerozoic mantle. Poudjom Djomani et al. (2001) also document that there is a signiﬁcant increase in mean density of the mantle lithosphere under continents (at standard temperature and pressure) from Archean (3.31 × 103 ± 0.016 kg/m3) to

Proterozoic (3.34 × 103 ± 0.02 kg/m3) to Phanerozoic (3.36 × 103 ± 0.02 kg/m3). Thus, younger lithosphere is more prone to be removed whereas old lithosphere may be inherently stable (Poudjom

Djomani et al., 2001; O’Reilly et al., 2001; Lee et al., 2005).

A density increase of the lithosphere can also occur through intrusion/injection of melts (mainly basaltic) within the eclogite stability field. Elkins-Tanton and Hager (2000) suggest that melt intrusion to the mantle lithosphere as an injection above a mantle upwelling may increase the density of the mantle lithosphere to 3340 kg/m3 when the melt solidifies as eclogite. They suggest that this establishes lithospheric instability compared to a density of the underlying sublithospheric mantle of 3280 kg/m3. Also, the accumulation of mafic rocks in the lower crust due to the fractionation and melt extraction can increase the density of the lithosphere. It is estimated that every 10% addition of eclogite through injection can increase the density of the mantle lithosphere by 1% (Kay and

Kay, 1993). A similar process of melt intrusion/solidiﬁcation is proposed by Jull and Kelemen

(2001) for particular tectonic areas associated with hot geothermal gradients such as arc settings, regions undergoing continental extension, and areas underlain by mantle plumes. They suggest that a density increase of the lower crust and mantle lithosphere by 25-100 kg/m3 is plausible and in arc regions, a density contrast of 1 − 5% compared to the underlying mantle has been suggested.

In addition to geochemical/petrological factors, tectonic disturbance of the lithosphere can also promote or initiate lithospheric removal. In particular, various geodynamic studies have shown that lateral compression or extension of the lithosphere can also promote delamination or Rayleigh-

Taylor instability of mantle lithosphere (Figure 1.1). (Houseman et al., 1981; Houseman and

Molnar, 1997; Conrad and Molnar, 1997, 1999; Conrad, 2000; Burov, 2007; G¨o˘g¨u¸sand Pyskly- wec, 2008a). Mechanical thickening of the mantle lithosphere by lateral compression can promote

5 Rayleigh-Taylor instabilities of the mantle lithosphere in three ways (Conrad and Molnar, 1997,

1999; Conrad, 2000): 1) During thickening/shortening there is a signiﬁcant increase in the volume of cold and dense mantle lithosphere layer. 2) If the rocks are weakened by increasing strain rates (non-

Newtonian rheology), horizontal compression can weaken the lithosphere significantly and increase the possibility of a convective instability to grow (Molnar et al., 1998). 3) Non-uniform thickening changes the temperature gradient of the lithosphere and this causes convective/gravitational instabilities to grow with changing density variations. It should also be noted that with Newtonian viscosities the amplitude of the mantle lithosphere instability grows exponentially with time (Chan- drasekhar, 1961). However, with non-Newtonian viscosity the instability grows superexponentially as the effective viscosity of the fluid decreases while strain rates increase (Conrad and Molnar,

1999; Houseman and Molnar, 1997). It should be noted that these Rayleigh-Taylor instabilities must grow over shorter timescales than the rate at which thermal diﬀusion dissipates the density instability.

6 Figure 1.1: Hypothetical evolution of the continental lithosphere during mechanical thickening. A) Mechanical thickening induces the gravitational instability of the cold and dense mantle lithosphere. B) The thickening of the mantle lithosphere may lead to convective removal/dripping instability.

7 1.2.2 Pre-existing Weak Zone between the Crust and Mantle Lithosphere

In our delamination experiments we assume a pre-existing low-viscosity layer between the mantle lithosphere and the crust to help initiate delamination. Previous studies point out the necessity of a weak layer within the crust (lower crust) for delamination (Bird, 1979; Meissner and Mooney, 1998;

Ranalli , 2000; Jull and Kelemen, 2001; Morency and Doin, 2004). Bird (1979) assumed a crustal origin for this layer (“horizontal d´ecollement”) that is of signiﬁcantly lower viscosity than the mantle lithosphere. He estimated that the maximum viscosity of the lower crust to facilitate delamination is approximately 1018 Pa·s. Meissner and Mooney (1998) calculated viscosity-depth curves based on reasonable geotherms and lithospheric rheologies and propose that low-viscosity zones can occur:

1) at the base of the felsic upper crust, 2) within the lower crust, and 3) several tens of kilometers under the Moho. They suggest that lithosphere is possibly delaminated somewhere within the lower crust because the usually observed high velocity (>6.8 km/s) layer within the crust is not observed after delamination. Jull and Kelemen (2001) suggest that temperatures > 500◦ − 700◦C are required for lower crustal instabilities of dense maﬁc compositions to occur. Possible tectonic regions for these conditions include: arc regions, volcanic rifted margins, extensional zones, and regions underlain by mantle plumes (Elkins-Tanton and Hager, 2000). Using numerical modeling,

Morency and Doin (2004) tested various Moho temperatures that allowed delamination between the mantle lithosphere and crust. They ﬁnd that Moho temperatures must be higher than 800◦C to permit eﬀective decoupling between the crust and mantle lithosphere.

Several studies compare the strength of the lithosphere to a jelly sandwich, where weak lower/mid- lower crust occurs between the rigid upper crust and upper mantle (Zuber, 1994; Burov and Watts,

2006). Ranalli (2000) suggests that for most geotherms and crustal compositions, especially if the rheology is quartz controlled (felsic composition), the lower crust is in the ductile ﬁeld and delamination can occur along this layer. Such a layer has also been detected seismologically and is associated with a low velocity seismic zone (Berry and Mair, 1977; Meissner and Mooney, 1991).

According to Ranalli and Murphy (1987) the low velocity seismic zone within the crust is due to high temperature, composition, and the presence of water. Lamontagne and Ranalli (1996) suggest that warm lower crust can have three orders of magnitude lower viscosity than the cold upper crust and underlying mantle lithosphere if it has an appropriate (mostly felsic) composition.

8 Within several tectonic environments, it has been suggested that a weak lower crustal layer would promote decoupling between the mantle lithosphere and crust. For example, a style of “ﬂake tectonics” involves an intra-crustal delamination (Oxburgh, 1972; Price, 1986; De Franco et al.,

2008). Flake tectonics can occur where large sheet-like masses (ﬂakes) of the upper crust of the subducting plate are stripped oﬀ onto the overriding plate along a weak layer within the crust.

However, this type of tectonic behaviour does not involve any type of exposure of the crust to the sublithospheric mantle (and the related thermal-topographic perturbation) associated with mantle lithosphere delamination. Flake tectonics has been suggested for the Eastern Alps (Oxburgh, 1972) and the California Transverse Ranges (Yeats, 1981). In particular cases, such as in large scale strike- slip fault systems, lower crustal viscosities may be closer to asthenospheric viscosities and the zone can be considered to be “intracrustal asthenosphere” (Turcotte et al., 1984).

1.3 Thesis Structure

The thesis is structured as follows:

Chapter 2 investigates several near surface geological observables as diagnostic indicators for dif- ferentiating lithospheric dripping (mechanism — I; Houseman et al., 1981) from mantle lithosphere delamination (mechanism — II; Bird, 1979). Results from two representative dripping models

(partial removal and complete removal) are compared with those of the delamination model. The experimental observables are: surface topography, Moho temperatures, P-T-t paths, and the pattern of crustal deformation. In the discussion section, model results and the comparison between those two diﬀerent removal styles are discussed brieﬂy in the context of possible mantle lithosphere removal in the southern Sierra Nevada.

Chapter 3 tests the viability of mantle lithosphere delamination for the Eastern Anatolian plateau (Keskin, 2003, 2007; S¸eng¨oret al., 2003, 2008) where lithospheric thinning, plateau uplift, high geothermal gradient and synconvergent extension are present. The region is undergoing active plate convergence, and as such the study represents the ﬁrst investigation of lithospheric delamination during plate shortening. We present various model results that characterise the tectonic evolution of the surface (i.e., surface topography, crustal thickness, surface strain rate) and compare

9 these quantitative results with present day observations of surface topography, crustal thickness

(Zor et al., 2003), and surface deformation (Dhont and Chorowicz, 2006) from Eastern Anatolia.

Chapter 4 investigates how continental delamination evolves from oceanic plate subduction.

It has been suggested that the style of formerly subducting ocean lithosphere (i.e, advancing, retreating) can inﬂuence the surface topography, crustal deformation, and style of plate consumption following collision (Royden, 1993). The continuing descent of the previously subducting ocean lithosphere following collision may help to initiate delamination of the mantle lithosphere. In this work, we present experimental results of surface topography, and crustal deformation associated with post-subduction continental delamination in a plate convergent regime.

1.4 Statement for Authorship

Chapter 2 is co-authored by R. N. Pysklywec. O. Gö˘gü¸sis the first author and he ran the thermomechanical numerical experiments, wrote the algorithm for crustal deformation analyses, evaluated/plotted the data and wrote the paper. R. N Pysklywec revised the earlier versions of the manuscript. Before the manuscript was submitted for journal publication, the interpretations of model results were discussed between O. Gö˘gü¸sand R. N. Pysklywec. The paper is published as

“Near-surface diagnostics of dripping or delaminating lithosphere ” in the Journal of Geophysical

Research, vol. 113. B11404, doi: 10.1029/2007JB005123, 2008.

Chapter 3 is co-authored by R. N. Pysklywec in which O. Gö˘gü¸sis the first author. It has been published in Geology, vol. 36(9), 723-726, doi: 10.1130/G24982A.1, 2008 as “Mantle lithosphere delamination driving plateau uplift and synconverent extension in eastern Anatolia”. O. Gö˘gü¸s ran the thermo-mechanical numerical experiments, wrote the algorithm for crustal deformation analyses, evaluated/plotted the data, and wrote the paper. O. Gö˘gü¸salso integrated analyses of regional geological-geophysical data of the eastern Anatolia collision zone. Both authors discussed the geodynamic model results together and interpreted them. The paper was written by O. Gö˘gü¸s and reviewed by Russell N. Pysklywec.

Chapter 4 is co-authored by R. N. Pysklywec, F. Corbi, and C. Faccenna. O. G¨o˘g¨u¸sconducted the analogue/physical modeling experiments in the tectonic modeling laboratory at Roma Tre

10 University. The model set up and evaluation of early modeling results were discussed with F. Corbi and C. Faccenna. O. Gö˘gü¸sevaluated/plotted the data and the interpretation of model results were made with R. N Pysklywec in Toronto. The paper was written by O. Gö˘gü¸sand reviewed by

R. N. Pysklywec, C. Faccenna, and F. Corbi.

11 Bibliography

Anderson, D.L., 2005, Large Igneous Provinces, Delamination, and Fertile Mantle Elements, v. 1,

p. 271-275, doi: 10.2113/gselements.1.5.271.

Anderson, D.L., 2007, The eclogite engine: Chemical geodynamics as a Galileo thermometer, in

Foulger, G.R., and Jurdy, D.M., eds., Plates, plumes, and planetary processes: Boulder, Col-

orado, Geological Society of America Special Paper, v. 430, p. 47-64, doi: 10.1130/2007.2430(03).

Beck, S. L., and Zandt, G., 2002, The nature of orogenic crust in the central Andes: Journal of

Geophysical Research, v. 107, B10, 2230, doi:10.1029/2000JB000124.

Berry, M.J., and Mair, J.A. 1977, The nature of the earth’s crust in Canada: American Geophysical

Union, Geophysical. Monographs, v. 20, p. 319-348.

Bird, P., 1978, Initiation of intracontinental subduction in the Himalaya: Journal of Geophysical

Research, v. 83, p. 4975-4987.

Bird, P., 1979, Continental delamination and the Colorado Plateau: Journal of Geophysical Re-

search, v. 84, p. 7561-7571.

Burov, E. B., and Watts, A. B., 2006, The long-term strength of continental lithosphere: “jelly

sandwich” or “crème-brûlée”?:GSA Today, v. 16, p. 4-10.

Burov, E., 2007, The role of gravitational instabilities, density structure and extension rate in

the evolution of continental margins: Geological Society London, Special Publications v. 282. p.

139-156.

Chandrasekhar, S., 1961. Hydrodynamic and Hydromagnetic Stability, Oxford University Press,

12 Oxford.

Channell, J.E.T., and Mareschal, J.C., 1989, Delamination and asymmetric lithospheric thicken-

ing in the development of the Tyrrhenian rift, in Coward, M.P., et al., eds., Alpine tectonics:

Geological Society London Special Publication v. 45, p. 285-302.

Cloos, M., Sappie, B., Quarles van Uﬀord, A., Weiland, R.J., Warren, P. Q., and Mc Mahon, T.

P., 2005, Collisional delamination in New-Guinea: The geotectonics of subducting slab breakoﬀ:

Geological Society of America Special Paper, v. 400, 48 pages.

Conrad, C.P., and Molnar, P., 1997, The growth of Rayleigh-Taylor type instabilities in the litho-

sphere for various rheological and density structures: Geophysical Journal International, v. 129,

p. 95-112.

Conrad, C.P., and Molnar, P., 1999., Convective instability of a boundary layer with temperature-

and strain-rate-dependent viscosity in terms of available buoyancy: Geophysical Journal Inter-

national, v. 139, p. 51-68.

Conrad, C.P., 2000, Convective instability of thickening mantle lithosphere: Geophysical Journal

International, v. 143, p. 52-70.

De Franco, R., Govers, R., and Wortel, M. J. R., 2008, Dynamics of continental collision: Inﬂuence

of the plate contact: Geophysical Journal International, v. 174, p. 1101-1120.

Dewey, J. F., Ryan, P.D., and Andersen, T.B., 1993, Orogenic uplift and collapse, crustal thickness,

fabrics and metamorphic phase changes: The role of eclogites. In: H.M. Prichard, T. Alabaster,

N.B.W. Harris and C.R. Neary, Editors, Magmatic Processes and Plate Tectonics: Geological

Society Special Publications, v. 76, p. 325-343.

Dhont, D., and Chorowicz, J., 2006, Review of the neotectonics of the Eastern Turkish-Armenian

Plateau by geomorphic analysis of digital elevation model imagery: International Journal of

Earth Sciences, v. 95, p. 34-49, doi: 10.1007/s00531-005-0020-3.

Ducea, M. N., and Saleeby, J. B., 1998, A case for delamination of the deep batholithic crust

beneath the Sierra Nevada, California: International Geology Review, v. 40, p. 78-93.

13 Elkins-Tanton L.T. and Hager, H., 2000. Melt intrusion as a trigger for lithospheric foundering and

the eruption of the Siberian ﬂood basalt: Geophysical Research Letters, v. 27, p. 3937-3940.

Elkins-Tanton, L.T., 2005, Continental magmatism caused by lithospheric delamination, in Foulger,

G.R., Natland, J.H., Presnall, D.C., and Anderson, D.L., eds., Plates, plumes, and paradigms:

Geological Society of America Special Paper, v. 388, p. 449-461, doi: 10.1130/2005.2388 (27).

Fadil, A., Vernant, P., McClusky, S., Reilinger, R., Gomez, F., Sari, D. B., Mourabit, T., Feigl,

K.,and Barazangi, M., 2006, Active tectonics of the western Mediterranean: Geodetic evidence for

roll back of a delaminated subcontinental lithospheric slab beneath the Rif Mountains, Morocco:

Geology, v. 34(7), p. 529- 532.

G¨o˘g¨u¸s,O.H., and Pysklywec, R.N., 2008a, Mantle lithosphere delamination driving plateau up-

lift and synconvergent extension in eastern Anatolia: Geology, v. 36, no. 9, p.723-726, doi:

10.1130/G2498A.1.

G¨o˘g¨u¸s,O.H., and Pysklywec, R.N., 2008b, Near-surface diagnostics of dripping or delaminating

lithosphere: Journal of Geophysical Research, v. 113, B11404, doi: 10.1029/2007JB005123.

Houseman, G.A., McKenzie, D.P., and Molnar, P., 1981, Convective instability of a thickened

boundary layer and its relevance for the thermal evolution of continental convergent belts: Jour-

nal of Geophysical Research, v. 86, p. 6115-6132.

Houseman, G.A., and Molnar, P., 1997, Gravitational (Rayleigh-Taylor) instability of layer with

non-linear viscosity and convective thinning of continental lithosphere: Geophysical Journal In-

ternational, v. 128, p.125-150.

Houseman, G.A., and Gemmer, L., 2007, Intra-orogenic extension driven by gravitational instabil-

ity: Carpathian-Pannonian orogeny: Geology, v. 35, p. 1135-1138. doi: 10.1130/G23993A.1

Jones C.H., Farmer, G.L., and Unruh J., 2004, Tectonics of Pliocene removal of lithosphere of

the Sierra Nevada, California: Geological Society of America Bulletin, v. 116, p. 1408-1422, doi:

10.1130/B25397.1.

Jordan, T.H., 1978. Composition and development of the continental tectosphere: Nature v. 274,

14 p. 544-548.

Jull, M., and Kelemen, P.B., 2001, On the conditions for lower crustal convective instability: Journal

of Geophysical Research, v. 106, p. 6423-6446, doi: 10.1029/2000JB900357.

Kay, R.W., and Kay, M., 1993, Delamination and delamination magmatism: Tectonophysics, v.

219, p. 177-189.

Lamontagne, M., and Ranalli, G., 1996. Thermal and rheological constraints on the earthquake

depth distribution in the Charlevoix, Canada, intraplate seismic zone: Tectonophysics, v. 257,

p. 55-69.

Lorinczi, P., and Houseman, G.A., 2009, Lithospheric gravitational instability beneath the South-

east Carpathians: Tectonophysics, v. 474, p. 322-336, doi:10.1016/j.tecto.2008.05.024

Keskin, M., 2003, Magma generation by slab steepening and break oﬀ beneath a subduction accre-

tion complex: An alternative model for collision-related volcanism in Eastern Anatolia, Turkey:

Geophysical Research Letters, v. 30, no. 24, 8046, doi: 10.1029/2003GL018019.

Keskin, M., 2007, Eastern Anatolia: A hotspot in a collision zone without a mantle plume: in

Foulger, G.R., and Jurdy, D.M., eds., Plates, plumes and planetary processes: Geological Society

of America Special Paper, v. 430, p. 693-722, doi. 10.1130/2007.2430(32).

Kosarev, G., Kind, R., Sobolev, S. V., Yuan, X., Hanka, W., and Oreshin, S., 1999, Seismic evidence

for detached Indian lithospheric mantle beneath Tibet: Science, v. 283, p. 1306-1309.

LeFevre, L. V., and Helmberger, D.V., 1989, Upper mantle P velocity structure of the Canadian

shield: Journal of Geophysical Research, v. 94, p. 17749-17765.

Lee, C.-T., A. Lenardic, C. M. Cooper, F. Niu, and Levander, A. , 2005, The role of chemical

boundary layers in regulating the thickness of continental and oceanic thermal boundary layers:

Earth and Planetary Science Letters, v. 230, p. 379-395.

Meissner, R., Wever, T., Sadowiak, P., 1991. Continental collision and seismic signature Geophysical

Journal International, v. 105, p. 15-23.

15 Meissner, R., and Mooney, W.D., 1998, Weakness of the lower continental crust: a condition for

delamination, uplift, and escape: Tectonophysics, v. 296 p. 47-60.

Molnar, P., Houseman, G.A., and Conrad, C.P., 1998, Rayleigh-Taylor instability and convective

thinning of mechanically thickened lithosphere: eﬀects of non-linear viscosity decreasing exponen-

tially with depth and of horizontal shortening of the layer: Geophysical Journal of International,

v. 133, p. 568-584.

Morency, C., Doin, M.P., and Desmoulin, C., 2002, Convective destabilization of a thickened con-

tinental lithosphere; Earth and Planetary Science Letters, v. 202, p. 303-320.

Morency, C., and Doin, M.P., 2004, Numerical simulations of mantle lithospheric delamination:

Journal of Geophysical Research, v. 109, B03410, doi: 10.1029/2003JB002414.

Neil, E.A., and Houseman, G.A., 1999, Rayleigh-Taylor instability of the upper mantle and its role

in intraplate orogeny: Geophysical Journal International, v. 138, p. 89-107.

OReilly, S.Y., Griﬃn, W.L., Poudjom Djomani Y., Morgan P., 2001, Are lithospheres forever?

Tracking changes in subcontinental lithospheric mantle through time: GSA Today, v.11, p. 4-9.

Oxburgh, E.R., 1972, Flake tectonics and Continental collision: Nature, v. 239, p. 202-204.

Platt, J.P., Soto, J.-I., Whitehouse, M.J., Hurford, A.J., and Kelley, S.P., 1998, Thermal evolution,

rate of exhumation, and tectonic signiﬁcance of metamorphic rocks from the ﬂoor of the Alboran

extensional basin, western Mediterranean: Tectonics, v. 17, p. 671-689.

Poudjom Djomani, Y.H., O’Reilly, S.Y., Griﬃn, W.L., and Morgan, P., 2001, The density structure

of subcontinental lithosphere through time: Earth and Planetary Science Letters, v. 184 p. 605-

621.

Price, R.A., 1986, The southeastern Canadian Cordillera: Thrust faulting, tectonic wedging and

delamination of the lithosphere: Journal of Structural Geology, v. 8, p. 239-244.

Pysklywec, R. N., and Cruden, A. R., 2004. Coupled crust-mantle dynamics and intraplate tecton-

ics: Two-dimensional numerical and three-dimensional analogue modeling, G3: Geochemistry,

Geophysics, Geosystems, 5(10), doi: 10.1029/2004GC000748, 22 pp

16 Ranalli, G., Murphy, D.C., 1987, Rheological stratiﬁcation of the lithosphere: Tectonophysics v.132,

p. 281-295.

Ranalli, G., 2000, Rheology of the crust and its role in tectonic reactivation: Journal of Geody-

namics, v. 30, p. 3-15.

Reutter, K.G., Giese, P. and Closs, H., 1980, Lithospheric split in the descending plate: observations

from the Northern Apennines: Tectonophysics, v. 64, p. 11-19.

Rychert, C. A., S. Rondenay, and Fischer, K. M., 2007, P-to-S and S-to-P imaging of a sharp

lithosphere-asthenosphere boundary beneath eastern North America: Journal of Geophysical

Research, 112, B08314, doi:10.1029/2006JB004619.

Royden, L.H., 1993, Evolution of retreating subduction boundaries formed during continental col-

lision: Tectonics, v. 12, p. 629-638.

Schott, B. and Schmeling, H., 1998: Delamination and detachment of a lithospheric root. Tectono-

physics, v. 296, p. 225-247.

Seber, D., Barazangi, M., Ibenbrahim, A., and Demnati, A., 1996, Geophysical evidence for litho-

spheric delamination beneath the Alboran Sea and Rif-Betic mountains: Nature, v. 379, p.

785-790, doi: 10.1038/379785a0.

S¸eng¨or,A.M.C., Ozeren, S., Zor, E., and Genc, T., 2003, East Anatolian high plateau as a mantle-

supported, N-S shortened domal structure: Geophysical Research Letters, v. 30, p. 8045, doi:

10.1029/2003GL017858.

S¸eng¨or, A.M.C., Ozeren, M.S., Keskin, M., Sakinc, M., Ozbakir, A., and Kayan, I., 2008,

Eastern Turkish high plateau as a small Turkic-type orogen: Implications for post-collisional

crust-forming processes in Turkic-type orogens: Earth-Science Reviews, v. 90, p.1-48,

doi:10.1016/j.earscirev.2008.05.002.

Stacey, F. D., 1992. Physics of the Earth: Brookfeld Press, Brisbane.

Schubert, G., Turcotte, D., and Olson, P., 2001, Mantle convection in the Earth and Planets:

Cambridge university press, London.

17 Turcotte, D.L., Liu, J.Y., and Kulhawy, F.H., 1984, The role of an intracrustal asthenosphere on

the behaviour of major strike-slip faults: Journal of Geophysical Research, v. 89, p. 5801-5816.

Wortel, M.J.R., and Sparkman, W., 2000., Subduction and slab detachment in the Mediterranean-

Carpathian region; Science, v. 290, p.1910-1917.

Yeats, R.S., 1981, Quaternary ﬂake tectonics of the California Transverse Ranges: Geology, v. 10,

p. 16-20.

Zandt, G., Gilbert, H., Owens, T. J., Ducea, M., Saleeby, J., Jones, C. H., 2004, Active Foundering

of a Continental Arc Root Beneath the Southern Sierra Nevada, California: Nature v. 432, p.

41-46.

Zor, E., Sandvol, E., Gurbuz, C., Turkelli, N., Seber, D., and Barazangi, M., 2003, The crustal

structure of the East Anatolian plateau from receiver functions: Geophysical Research Letters,

v. 30, p. 8044, doi: 10.1029/2003GL018192.

Zuber, M.T., 1994, Folding a jelly sandwich: Nature, v. 371, p. 650-651.

18 Chapter 2

Near Surface Diagnostics of Dripping or Delaminating Lithosphere

19 2.1 Abstract

In various geological regions, it has been postulated that the mantle lithosphere has been thinned or completely removed. Two of the primary removal mechanisms that have been put forward include: a) delamination–a wholesale peeling away of a coherent block of the mantle lithosphere; and b) lithospheric “dripping”–viscous Rayleigh-Taylor instability of the mantle lithosphere. Using computational models, we investigate several near-surface observables to determine if these may be diagnostic of either (often ambiguous) removal mechanism. Surface topography associated with delamination has a broad region of uplift above the lithospheric gap and a localized and mobile zone of subsidence at the delaminating hinge. With dripping lithosphere, the topographic expression is symmetric and ﬁxed above the downwelling. Delamination of mantle lithosphere is more eﬃcient than dripping for thermal heating of the crust; the onset is more rapid and the elevated temperatures persist longer. The resultant crustal P-T-t paths show modest pressure variations and high temperature increases with large-scale delamination or dripping. Delamination also causes contraction directly above the (migrating) hinge and distal extension. Dripping lithosphere induces superimposed contraction and extension above and symmetric about the viscous instability. For all the observables, if only a portion of the mantle lithosphere is removed by viscous instability

(delamination inherently removes all of the mantle lithosphere), the diﬀerences between the two removal mechanisms are even more pronounced. For example, with only partial removal of the mantle lithosphere, uppermost mantle lithosphere remains well coupled to the crust, leading to lower surface temperature variations and broad zones of crustal deformation/thickening.

2.2 Introduction

The removal of the mantle lithosphere (i.e., the sub-crustal portion of the lithosphere) has been invoked in a variety of tectonic regimes to account for a range of geological, geophysical and geochemical observations. For example, anomalous heating, topography, and gravity has led to interpretations that mantle lithosphere has been removed during some phase of the plate convergence at the Andean margin (Kay and Kay, 1993; Beck and Zandt, 2002), the India-Eurasia collision (Bird,

1978; Houseman et al., 1981), the New Guinea collisional zone (Cloos et al., 2005), the Eastern

Anatolian Plateau (Keskin, 2003, 2007; S¸eng¨oret al., 2003, 2008), the Carpathian-Pannonian region

20 (Houseman and Gemmer, 2007; Lorinczi and Houseman, 2009). Within the interior of continental tectonic plates, such as at the Sierra Nevada (Ducea and Saleeby, 1998; Jones et al., 2004; Zandt et al., 2004) and beneath the Colorado Plateau (Bird, 1979) the removal of the mantle lithosphere has been proposed to explain regional uplift. These mantle lithosphere removal events are generally based on the density contrast of the mantle lithosphere with respect to the less dense underlying mantle. The density contrast can arise as a result of thermal contraction of the cold mantle lithosphere, although compositional density variations of the lower crust have also been called upon to explain lithospheric instability (Jull and Kelemen, 2001; Elkins-Tanton, 2005).

The geodynamic mechanisms in which mantle lithosphere is removed are still debated, but there are fundamentally two primary removal scenarios that have been put forward. Bird (1978,

1979) proposes a model of mantle lithosphere delamination, where the cold and dense mantle lithosphere peels away as a coherent slice from the crust along the Moho. The removed slice of mantle lithosphere is replaced by hot and buoyant asthenosphere (Fig 2.1A). Generally, delamination is predicated on the idea that the hot-weak lower crust is the most pronounced strength discontinuity in the lithosphere. This results in separation between the strong crust and strong mantle lithosphere portions of the plate. Morency and Doin (2004) used 2-D numerical simulations of convection with a viscoplastic rheology to study the delamination mechanism and in particular consider the geodynamic conditions that cause this type of lithospheric removal. They suggest that delamination begins with localized thinning of the mantle lithosphere and where the highest Moho temperature is achieved, progressive thermomechanical erosion of lithospheric mantle occurs due to asthenospheric upwelling. This leads to sinking of the dense mantle lithosphere slice into the asthenosphere. The authors intimate that delamination is only possible when Moho temperatures exceed approximately 800◦C.

Delamination has been associated with topographic and thermal perturbation of the crust. It is suggested that regional uplift of the Sierra Nevadas, Andes, and Colorado plateau has developed through mantle lithosphere delamination events in these areas (Bird, 1979; Jones et al., 2004; Le

Pourhiet et al., 2006). Uplift is largely the result of isostatic adjustment from replacing dense mantle lithosphere with buoyant asthenosphere, although ﬂexure and deformation of the plate also cause topographic variations (Le Pourhiet et al., 2006). The nature of delamination means that hot asthenosphere comes into direct contact with the crust. The presence of high potassium intrusives

21 and high crustal heat ﬂow has been put forward as the thermal/magmatic signature of delamination

(Kay and Kay, 1993; Ducea and Saleeby, 1998; Manley et al., 2000).

An alternative lithospheric removal mechanism to delamination is viscous convective removal of the mantle lithosphere (e.g. Houseman et al., 1981; England and Houseman, 1989). In this case, some or all of the mantle lithosphere may be removed as cold dense mantle lithosphere “drips” as a viscous Rayleigh-Taylor (RT) gravitational instability (Fig 2.1B). The primary difference from delamination is that the mantle lithosphere is not removed as a coherent lithospheric slice, but deforms in a distinctly “un-platelike” manner as it descends/drips into the mantle. Subsequent studies have demonstrated that the propensity of the lithosphere to experience convective removal is controlled by the viscous rheology of the mantle lithosphere (e.g. Buck and Toksoz, 1983; Lenardic and Kaula, 1995; Houseman and Molnar, 1997; Molnar et al., 1998). In addition, the timescale and character of these viscous perturbations can be influenced by horizontal shortening of the lithosphere (e.g. Conrad and Molnar, 1997; Molnar et al., 1998), and the presence of an overlying crust (e.g. Neil and Houseman, 1999; Pysklywec and Cruden, 2004). The explicit assumption of these models is that there is sufficient perturbation (and at a suitable wavelength) to the mantle lithosphere to initiate Rayleigh-Taylor instability. Furthermore, the growth rate of the instability must outpace thermal diffusion of the cold mantle lithosphere root into the hot mantle.

It is worthwhile to clarify that we use the term delamination following its deﬁnition by Bird

(1979). We differentiate this from the viscous process of Rayleigh-Taylor instability (or “convective removal” or “dripping”) of the mantle lithosphere. Often the term delamination is used to denote any type of mantle lithosphere removal event. This introduces ambiguity relating to what are quite different processes with different surface expressions, as we will demonstrate.

As with delamination, the replacement of the dense mantle lithosphere by more buoyant mantle can cause isostatic surface uplift. This type of mechanism has been used to interpret the Tibetan plateau uplift (England and Houseman, 1989) and the anomalous regional topography across the southern Sierra Nevada (Saleeby and Foster, 2004). Neil and Houseman (1999) demonstrate that in certain situations viscous dripping can also induce thickening of the overlying crust and hence further surface uplift. These isostatic responses are complemented by mantle ﬂow-induced topography as the viscous dripping progresses (e.g. Pysklywec and Cruden, 2004). Saleeby and Foster (2004) suggest that the subsidence of the Tulare Lake basin, in the southern Great Valley is driven by

22 active mantle dripping. The adjacent uplift and tilting of the Sierra Nevada range may be related to thinning of the mantle lithosphere peripheral to the drip (Zandt et al., 2004).

Thus, as mantle lithosphere removal mechanisms, it has been shown or postulated that both delamination and Rayleigh-Taylor instability will produce a variety of thermal and deformational responses of the crust. Consequently, in regions where removal of the mantle lithosphere seems to have taken place–such as the Sierra Nevada, as discussed above–initial interpretation of the surface constraints may lead to ambiguity about which mechanism is active.

The purpose of this paper is to investigate in more detail the crustal and surface observables that may be used as diagnostic elements to diﬀerentiate between the two of mantle lithosphere removal processes: delamination or Rayleigh-Taylor-type convective removal/dripping. We designed a series of forward numerical experiments representing both types of removal mechanisms. We focus on four primary properties during evolution of the model: (1) surface topographic evolution, (2) distribution and style of crustal deformation, (3) thermal evolution of various levels within the crust, and (4) the metamorphic (P-T-t) evolution of lower, middle and upper crust. These represent surface or near-surface observables that may generally characterize the thermal-mechanical evolution of the lithosphere during mantle lithosphere removal. Our intent is that by doing a direct comparison using generic models, the results will provide clearer quantitative information for helping to interpret the deep lithospheric dynamics from surface geology.

23 Figure 2.1: Schematic illustration of geodynamic models for mantle lithosphere removal. A) De- lamination of the mantle lithosphere (Bird, 1979); B) Convective removal of the mantle lithosphere (Houseman et al., 1981).

24 2.3 Experimental Results

2.3.1 Model Description

For our experiments, we used SOPALE, a plane strain, incompressible numerical code to study the thermomechanical behaviour of the coupled crust and mantle. SOPALE is based on the arbitrary

Lagrangian-Eulerian ﬁnite element technique and as such is useful for treating ﬁnite deformations, and for tracking boundaries (surface topography) and internal particles (P-T paths) (Fullsack,

1995; Pysklywec et al., 2002) (see Appendix). Figure 2.2A,B shows the initial conﬁguration of our experiments. In the models, the 160 km thick lithosphere is made up of 40 km thick buoyant crust

3 3 ρ0 = 2840 kg/m , pink) and 120 km thick dense mantle lithosphere (ρ0 = 3300 kg/m ,dark blue)

3 overlying an upper mantle region (ρ0 = 3260 kg/m , grey).

h i −5 Density is also a function of temperature: ρ(T ) = ρ0 (1 − α(T − T0) , where α = 2×10 1/K

◦ is the coeﬃcient of thermal expansion and T0 = 25 C is the reference temperature (Tackley et al., 1994). The assumed density variations are responsible for all motions in the box (gravitational acceleration g = 9.8 m/s2); in addition to this, there are no imposed velocities in the models. The viscous response of the crust is based on a wet quartzite ﬂow law (Gleason and Tullis, 1995) whereas the sub-crustal (mantle) material is governed by a dry olivine rheology (Hirth and Kohlstedt, 1996).

The crustal layer also yields according to a Coulomb failure law (Fig. 2.2), so that the upper crust behaves in a brittle manner.

The numerical (width) x (depth) resolution is 201 x 101 Eulerian nodes and 601 x 301 Lagrangian nodes. However, half of the Eulerian and Lagrangian elements are concentrated in the top 160 km in order to enhance resolution in the lithosphere.

The initial temperature proﬁle is the same in both experiments (Fig. 2.2). Thermal properties

−1 −1 −1 −1 (thermal conductivity, k = 2.25 W·m ·K , heat capacity, cp= 1250 J·kg · K are the same for all materials and we ignore radioactive heat production and shear heating in the model. The top boundary of the box is held at 25◦C and the bottom boundary is held at 1523◦C, also heat

ﬂux across the side boundaries is zero. The 350◦C Moho is based on estimates for the southern

Sierra Nevada of California (Lachenbruch and Sass, 1977), as a region where both dripping and delamination have been postulated.

The model has a free top surface, allowing topography to develop as the model evolves. The

25 mechanical boundary conditions at the other three sides are deﬁned by zero tangential stress and normal velocity (“free-slip”). We have extended the depth of the solution space into the lower mantle so that the sinking mantle lithosphere material moves away from the lithosphere. Although the depth of the box is 1000 km, the eﬀects of the endothermic olivine phase (γ-spinel structure to perovskite) change at 660 km depth (e.g. Christensen and Yuen, 1984) are not implemented because we wanted to allow downgoing mantle lithosphere to move away from the surface to the deeper levels of the mantle.

Each of the experiments has several modiﬁcations to initiate either delamination or viscous dripping. In order to start the delamination, we inserted a low viscosity weak zone, with a viscosity of 5 × 1019 Pa·s., between a section of the crust and mantle lithosphere (Fig. 2.2a). A series of numerical experiments demonstrates that this viscosity is suﬃciently low to decouple the crust from mantle lithosphere, whereas an increase to appreciably 5 × 1020 Pa·s retards the development of the delamination process.

The inclusion of a low-viscosity weak zone to initiate delamination is an approach followed by other studies of the process (Morency and Doin, 2004; Le Pourhiet et al., 2006). The density of

3 the mantle lithosphere over a 180 km wide, 80 km thick zone was increased to ρ0 = 3400 kg/m in order to further facilitate the delamination. Jull and Kelemen (2001) and Elkins-Tanton (2005) suggest that the lower lithosphere can become signiﬁcantly denser and gravitationally unstable due to the transformations such as from granulite to eclogite. According to Jull and Kelemen (2001), the density of the lower crust may become 50-250 kg/m3 denser at minimum pressures of less than 1.5 GPa and eventually the anomalously denser lower crust may descend/sink as a “blob” depending on the nature of the instability.

For the viscous dripping models we introduce a perturbation to the base of the dense mantle lithosphere to initiate the descent. It has been suggested that such perturbations may arise as a result of lithospheric contraction and thickening (Houseman and Molnar, 1997) or eclogitic metamorphism of lower crust (Kay and Kay, 1993; Ducea and Saleeby, 1998).

We emphasize that the intent of this work is not to consider the conditions that control the initiation and development of the dripping or delamination of the mantle lithosphere; this has been studied elsewhere (Conrad and Molnar, 1997; Jull and Kelemen, 2001; Morency and Doin, 2004).

Rather, we set up conditions that will start these removal events and focus on various facets of the

26 resulting near-surface dynamics.

Figure 2.2: Illustration of the model geometry and set up for A) Mantle lithosphere delamination and B) RT-type mantle instability experiments. A viscous flow law of ˙ = Aσn exp[−Q/(RT )] is used for viscous material response where, where ˙ is the strain rate, T is temperature, σ is differential stress. Variables A, n, Q and R are the viscosity parameter, stress exponent, activation energy, and ideal gas constant, respectively. Based on a strain rate of 10−15 1/s and temperature of 1350◦C, the 20 viscous flow law results in an effective viscosity of (ηeff ) = 10 Pa·s for the sub-lithospheric mantle lithosphere. For numerical considerations, viscosity in these models range between 5 × 1019 Pa·s to 1023 Pa·s. For continental crust A = 1.1 × 10−28 Pa−4/s, n = 4, and Q = 223 kJ/mol are used based on wet quartzite (Gleason and Tullis, 1995). For the crust an internal angle of friction φ = 15◦ is used for a Coulomb yield criterion. A = 4.89 × 10−17 Pa−3.5/s, n = 3.5 and Q = 535 kJ/mol are used for mantle material (mantle lithosphere and sub-lithospheric mantle) based on dry olivine (Hirth and Kohlstedt, 1996).

27 2.3.2 Dripping and Delaminating Mantle Lithosphere

In this study, we present our results in dimensionless time. This is motivated by variations in viscosities between the three models that give rise to variations in the dimensional times of events in the models. As a characteristic timescale, we chose the time that it takes the descending mantle lithosphere to traverse the upper mantle and reach the bottom of the solution space. (The characteristic timescales for DEL-1, DRIP-1, and DRIP-2 are 4.5 m.y., 21 m.y., and 5.7 m.y., respectively.)

The non-dimensional times based on (variable) material deformation are more comparable and less inﬂuenced by variations in viscosity between models.

Figure 2.3A shows the evolution of a delamination model. By t = 0.26 the mantle lithosphere is peeling away from the crust. As the lithosphere delaminates, the lower viscosity sub-lithospheric mantle flows into the area vacated by mantle lithosphere. This results in appreciable advection of mantle heat upwards (Fig. 2.3A, inset). The hot sub-lithospheric isotherms are deflected toward the crust as the material progressively intrudes into the mantle lithosphere gap. The delamination progresses rapidly and by t = 0.84, a ∼800 km wide section of mantle lithosphere has been removed, exposing the crust to sub-lithospheric mantle. The width of the breach is dependent on the width of the weak decoupling layer. Rather than peeling away as a single slice, the delaminated fragment of mantle lithosphere eventually detaches into separate fragments as it falls into the mantle. The sublithospheric mantle flow becomes more subdued as the delamination comes to an end and over the longer course of the experiment the material in the mantle lithosphere gap cools down to produce new lithosphere.

To compare to the delamination model, two diﬀerent RT-instability models are shown: DRIP-

1 and DRIP-2. They are physically and geometrically similar models, except that DRIP 2 has non-linear (n=3.5) temperature independent behaviour for mantle lithosphere, whereas DRIP-1 has non-linear (n=3.5) temperature dependent mantle lithosphere rheology. Both models use non-

Newtonian (n=3.5) viscosities. In comparison to Newtonian (n=1) fluids, the non-linear rheology will tend to localize deformation since viscosity is reduced as strain rates increase. This tends to limit the amount of material that is involved in the drip (Houseman and Molnar, 1997). In terms of controlling parameters, For the DRIP-1 , we use the flow law and controlling parameters described in Figure 2.2, whereas the DRIP-2 uses on both sides the same flow law but with parameters Q= 0

28 and A = 10 × 10−38 Pa−n/s. Based on strain rates of 10−13 1/s to 10−17 1/s that are characteristic of ﬂow in the models, this results in a mantle lithosphere viscosity ranging from 2.5 × 1020 Pa·s to

1 × 1023 Pa·s.

In both models, the lithospheric perturbation induces a drip-style downwelling of the mantle lithosphere (Figures 2.3B and 2.3C). However, in the temperature-dependent experiment (Fig.

2.3B) only the lowermost portion of the mantle lithosphere is dripping, as the warmer/weaker region of the lithosphere. Also, this deformation is spread over a rather broad lateral extent. In the temperature-independent case (Fig. 2.3C), most of the mantle lithosphere layer is being deformed, but over a localized lateral extent. Clearly, DRIP-2 is descending quicker than DRIP-1 since there is greater mass of unstable mantle lithosphere with DRIP-2 than DRIP-1. The amount of material participating in the instability has been explained as an “available buoyancy” that controls the growth rate of the drip (Conrad and Molnar, 1999).

29 Figure 2.3: Evolution of the models: A) DEL delamination; B) DRIP-1 viscous dripping with non- linear temperature dependent rheology; C) DRIP-2 viscous dripping with non-linear, temperature independent rheology. Each frame shows material colours (see Fig. 2.2) and deformed Lagrangian mesh. The latter is plotted at one half actual resolution; mesh is initially even rectangular. Inset in (A) shows isotherms of zoomed region. S1 and S2 indicate locations of sections for P-T-t analyses.

30 2.3.3 Surface Topography

As the mantle lithosphere starts to delaminate (t=0.13), it causes subsidence of the crust with an amplitude of approximately -1100 m (Fig. 2.4A). This subsidence is a result of the loading on the surface of the peeling/descending mantle lithosphere. Adjacent topography highs at x = 1300 km and x = 800 km probably arise from surrounding upwelling return ﬂow of sub-lithospheric mantle.

By t = 0.26, the negative surface deflection reaching a maximum amplitude of 3.5 km. A significant length of delaminating slab is pulling down on the crust at the location where the mantle lithosphere is still attached to the surface (Fig. 2.3A). Enhanced uplift occurs on the right side of the depression as hot mantle material is flowing into the lithosphere breach vacated by the delaminating lithosphere both sides of this negative topography, again due to return mantle flow.

The signal of topography is becoming more asymmetrical as the delamination progresses in these very early stages. The amplitude of the negative deﬂection is very high; we ascribe this to the extra forcing by the anomalously high density lithospheric block that was used to initiate delamination.

By t= 1.26 most of the mantle lithosphere has been removed and the surface topography is characterized by a broad uplift (t =1.26; Fig. 2.4A). Now, most of this uplift is associated with the replacement of the lithospheric mantle by buoyant sub-lithospheric mantle, as there is relatively little mantle ﬂow across the vacated lithospheric zone. The exception to this is at x = 400 km where there is still some active delamination that induces enhanced local subsidence/uplift. It is important to note that through the delamination process the locus of the main topography anomaly migrates to the left. That is, the topography is spatially transient, as it responds to the peeling lithosphere.

In the DRIP-1 experiment a negative topography initially develops above the descending RT instability reaching a maximum depression of 600 m (t = 0.30; Fig. 2.4B). This symmetric topography is supported by the actively descending/dripping mantle lithosphere (Fig. 2.3B). Eventually, the subsidence inverts to uplift (t = 0.60 - 0.90). This is a result of the decrease in the downwelling forces as the descending mantle lithosphere is necking and narrowing, and reaches the bottom of the box. The topography is now dominated by isostatic uplift associated with the ﬂow induced crustal contraction and thickening (e.g. Pysklywec and Cruden, 2004). The physical development and the progression of these events of DRIP-2 are similar with DRIP-1: There is initial surface

31 subsidence, followed by uplift as a result of the interplay between the dynamic eﬀects of the mantle ﬂow and crustal thickening. However, with DRIP-2, the more localized removal of mantle lithosphere material is responsible for a more focused band of crustal contraction/thickening and isostatic compensation by the asthenospheric mantle.

As in the delamination model, both drip experiments show an initial phase of subsidence, followed by uplift. However, the drip models show clearly symmetrical topography signals that remain ﬁxed in location above the downwelling mantle lithosphere. This assumes the mantle lithosphere instability does not migrate with respect to the overlying plate.

32 Figure 2.4: Plots of surface topography at three time intervals for models: A) delamination; B) DRIP-1; and C) DRIP-2.

33 2.3.4 Moho Temperatures

The Moho temperatures were tracked in all experiments to consider the thermal expression of the crust to the mantle lithosphere removal style. In the numerical model, we used two diﬀerent methods to illustrate the varying Moho temperatures. Firstly, a time series plot of zonal average and maximum Moho temperatures for each of the experiments was tracked (Figures 2.5 - 2.7A).

This is done by calculating maximum and average temperatures at a depth between 38.4 and 41.6 km and width between 500 km and 1500 km. The time series demonstrates how temperature changes in a “Moho zone” in a Eulerian reference frame. The lateral limits (x = 500 and 1500 km) to the Moho zone were chosen to focus on the location of delamination and drip at the center of the computational box. Secondly, we plotted the temperature of speciﬁc numerical particles along the entire base of the crust (x = 0 - 2000 km) at discrete time periods (Figures 2.5 and 2.7B). The particles started at the initial Moho depth of 42.6 km, notably couple of kilometers deeper than

Eulerian depths, and subsequently track the evolving temperature of these particles in a Lagrangian framework.

For delamination model, until t = 1.00, there is a rapid increase of the maximum Moho temperatures reaching more than 600◦C. (Figure 2.5A). This is a consequence of the upwelling of the asthenospheric material into the lithospheric gap. Subsequently, there is more gradual increase of the maximum and average temperature in this zone as more of the lithosphere delaminates away and the hot mantle temperatures conduct heat into the crust. It reaches a maximum of 1200◦C. by about t = 5.00. In Figure 2.5B, the temperatures of the “Moho particles ” are plotted at three discrete intervals: t = 0.26, t = 0.64 and t = 1.00. At t = 0.26, the base of the crust is heated to

400◦C. As most mantle lithosphere is removed and a broader zone of Moho is heated by t =0.64, a wider swath of these particles has increased temperature up to 500◦C. in average. By t = 1.00, after the main delamination event, the full l range of particles above the lithospheric gap experience elevated temperatures.

Calculated Moho temperatures for DRIP-1 and DRIP-2 show signiﬁcant diﬀerences. While the maximum Moho temperature values for DRIP-1 show very slight changes (Fig. 2.6A), temperatures obtained from DRIP-2 are close to that of the delamination model (Fig. 2.7A). For DRIP-1, the maximum temperatures at the base of the crust do not even reach 400◦C and the average

34 temperature of the Moho actually decreases slightly. There is only subtle warming of the Moho particles over time as they are pulled/pushed to greater depths. Clearly, the crust is quite insulated from the thermal eﬀects associated with the DRIP-1 downwelling, which occurs mostly in the lower portion of the mantle lithosphere (Fig. 2.3B).

The initial stages of the Moho temperatures of DRIP-2 are similar with DRIP-1: As the mantle lithosphere instability grows until ∼t= 1.25 the crust experiences relatively little thermal perturbation (Fig. 2.7A). During this phase, the advection of temperatures downwards with descending mantle lithosphere helps to keep the Moho region relatively cool (Fig. 2.7C). Eventually, the dripping mantle lithosphere detaches and hot sub-lithospheric mantle material flows into the void, resulting in a rapid increase in lower crust zonal temperatures (Fig.2.7A). By t = 2.19, discrete particles in the lower crust show heating to about 600◦C across a 1000 km wide band corresponding to the removal area (Fig. 2.7B). Clearly, the rheology of the mantle lithosphere, which governs the portion of mantle lithosphere that is removed (i.e., DRIP-1 versus DRIP-2), has a first-order effect on the thermal expression of dripping lithosphere in the crust. With almost complete removal the thermal signature is similar to that of delamination.

The high temperatures decrease quite rapidly from their peak at ∼t = 2.5 (Fig. 2.7A). We ascribe this to continual ﬂow of surrounding mantle lithosphere into the lithospheric gap that causes cooling in this zone. The delamination model does not experience this rapid drop in temperature as the stronger (more plate-like) mantle lithosphere does not have the same predilection for lateral

ﬂow.

35 Figure 2.5: For the delamination model: A) Plots of zonal maximum and average temperatures at 42.6 km depth as a function of time (see text for explanation). B) Particle temperatures of “Moho particles ” at t = 0.31, t = 0.76 and t = 1.18.

36 Figure 2.6: For the DRIP-1 model: A) Plots of zonal maximum and average temperatures at 42.6 km depth as a function of time (see text for explanation). B) Particle temperatures of “Moho particles ” at t = 0.54, t = 1.22 and t = 1.72.

37 Figure 2.7: For the DRIP-2 model: A) Plots of zonal maximum and average temperatures at 42.6 km depth as a function of time (see text for explanation). B) Particle temperatures of “Moho particles ” at t = 0.78, t = 1.66 and t = 2.19.

38 2.3.5 P-T Histories

We constructed pressure-temperature-time paths (P-T-t) by tracking individual groups of La- grangian particles within the deforming crustal material domain. We focused on two vertical proﬁles locations: At S1 (x = 1000 km) and S2 (x = 750 km) which correspond to the middle of the box and an intermediate distance across the removal zone (Figure 2.3). For these two locations, we track the pressure and temperature at positions in the upper (z = 7 km), middle middle (z

= 19 km), and lower (z = 39 km) crust. The pressure tracked is actually the lithostatic pressure associated with the burial or uplift of material. That is, it is derived as pressure p = ρ·g·h, where h is the thickness of overburden material above the point, g is gravitational acceleration, and ρ is the density of the overburden material. The additional “dynamic ” pressure is neglible compared to the lithostatic pressures for these types of non-convergent plate models. We note that surface erosion is not implemented in the models.

Figure 2.8A shows the P-T-t path of the mantle lithosphere delamination (DEL) model. S1 (x = 1000 km) of 39 km depth reaches its maximum burial depths rapidly at t = 0.38 with a maximum pressure of 11 kbar. This corresponds to modest extra burial of approximately three kilometers as that portion of lower crust is pulled down by the delaminating mantle lithosphere.

Subsequently, the P-T history is dominated by heating of 350◦C until t = 2.8. This is accompanied by decompression of 2.3 kbar, or about 9 km of tectonic exhumation. The P-T-t trends of the shallower particles at z = 16 km and z = 7 km depths for same section (S1) are similar in that they are characterized by decompression heating. However, the temperature increases are slower and smaller at these depths, as the mantle heat has to conduct through the crust.

The progression of the P-T change is essentially mirrored in the lithospheric section at S2 (x = 750 km), except that the events are delayed by several million years as the mantle lithosphere delaminates in that direction. For example, the maximum burial of the deep crust point to 12 km is reached at t = 0.6 and maximum heating is reached at t = 4.2 This suggests that the metamorphic signal in the “a single metamorphic wave ” with the delaminating mantle lithosphere.

The P-T evolution of DRIP-1 is very diﬀerent. It is dominated by slow pressure increase and very modest changes in temperature (Figure 2.8B). For example, the lower crust at section S1 shows an increase in pressure of 12 kbar over t = 2.0, which represents a burial of 86 km. This is clearly

39 a portion of the crust that is being entrained downward within the mantle lithosphere drip (e.g.

Pysklywec and Cruden, 2004). The package is relatively better thermally insulated as it warms

◦ only 50 C over this time. In comparison, at section S2 the lower crust is buried to a depth of only 59 km. The eﬀects of crustal thickening and burial are most extreme directly above the mantle lithosphere downwelling. Indeed at distances suﬃciently far from the drip, the crust can undergo extension and uplift during these stages instead (Pysklywec and Cruden, 2004). The shallower points, again, demonstrate a similar behaviour, but with more modest pressure increases.

Interestingly, the P-T-t paths associated with experiment DRIP-2 show patterns more akin to the delamination model. Lower crust at S1 increases to a pressure of 12.1 kbar at t = 1.1 (Figure 2.8C), corresponding to the time of vigorous mantle lithosphere dripping (Figure 2.3C). After this

◦ time, there is a pressure decrease to 8 kbar and heating to 600 C until t = 3.3. At section S2 the P-T-t path is diﬀerent: It does not experience the initial burial and most heating occurs prior to the modest ﬁnal uplift. This occurs also in the shallower crustal points.

The variation in behaviour between sections diﬀerentiates the metamorphic signature of crustal material between the delamination model and DRIP-2. The delamination P-T-t paths at the diﬀerent sections were essentially the same, but with a temporal shift associated with the peeling lithosphere. For DRIP-2, the sections show varying behaviour depending on their lateral position since the mantle lithosphere drip is not moving laterally with respect to the surface crust.

40 Figure 2.8: Pressure-temperature paths for particles in the upper (z = 7 km), middle (z = 19 km), and lower (z = 39 km) crust at sections S1 (x = 1000 km) and S2 ( x = 750 km). Paths for model: A) delamination; B) DRIP-1; C) DRIP-2.

41 2.3.6 Crustal Deformation

Lastly, we investigate styles of ﬁnite crustal deformation and structure that may be characteristic of the mantle lithosphere removal mechanisms. Diﬀerences are highlighted by: 1) Plotting the variation of the Moho position and surface topography to illustrate variations in crustal thickness;

2) Displaying the Lagrangian cells in the lithosphere, which are initially even rectangular, and so show accumulated deformation in their contorted state. While the amount and the style of deformation in the crust may be well constrained by the distortion of the Lagrangian meshes in the crust, extreme deformation of these cells such as very highly deformed meshes in the asthenosphere in the later stages of the models can only show the intensity of deformation.

For the delamination experiment at t = 1.0, the deformed Lagrangian cells demonstrate that above the delaminating lithosphere (at x = ∼ 500 km) there is contraction (Figure 2.9A). The downgoing plate is pulling crustal material into the zone above the delaminating hinge. This correlates with thickened crust as the Moho is deﬂected downward by approximately 7.5 km. It is worthwhile to note that the surface topography is negative in this region despite the thickened crust. Apparently, the sub-crustal loading of the delaminating mantle lithosphere is overwhelming isostatic (uplift) eﬀects associated with thickening crust. On the other side of the lithospheric gap

(at x = ∼ 1250 km), there is extension of the crustal elements. The crust here has been thinned by up to 7.5 km. We attribute the extension/thinning to several factors. Firstly, the sublithospheric mantle ﬂow into the gap and laterally may be helping pull apart the crust. Secondly, previously contracted/thickened crust may be in gravitational collapse as the delaminating mantle has moved away. Note again, that although the crust is thin, the surface topography is not anomalously low.

This suggests that surface topography is dynamically supported by underlying mantle ﬂow.

Figure 2.9B shows the variation in the crustal thickness and surface topography at t = 1.0 for

DRIP-1. At this stage, it is observed that dripping mantle lithosphere causes lateral crustal flow towards the center, which results in contraction and crustal thickening, while the distal regions experience broad bilateral or symmetrical type extension with contraction/thickening. The deflection of the Moho directly above the drip is consistent with interpretations of crustal structure for the southern Sierra Nevada from receiver function studies (Zandt et al., 2004). This work suggests that the Moho is characterized by a “V”-shaped profile by the entrainment of the viscous crust into the

42 mantle lithosphere, much like that shown in Figure 2.9B.

DRIP-2 at t=1.0 shows a conﬁned zone of contractional deformation with shortened Lagra- grangian cells within ∼ 100 km of each side of the centre of the downgoing mantle lithosphere (at x

= 1000 km; Figure 2.9C). Just outside this zone, on both flanks, the crust has been extended. This contraction is accompanied by a thin region of crustal thickening as crust is entrained downward into the mantle lithosphere drip. The localized thickening occurs within a broader (∼ 800 km) zone of thinned crust. The crustal extension and thinning may be due to collapse of previously developed topography (e.g., Figure 2.4C) and small-scale mantle flow in the mantle lithosphere gap as material is pulled from the flanks toward the centre of the downwelling. The symmetric and highly localized nature of deformation of the crust in DRIP-2 clearly distinguish it from both the delamination and alternate drip model.

43 Figure 2.9: Plots of Moho position and surface topography across a portion of each model at t = 1.0: A) Delamination, B) DRIP-1, and C) DRIP-2 .Included for each is a frame displaying material colours and Lagrangian mesh (plotted at one half actual resolution). Please note that “ext” refers to extension, “cont” refers to contraction.

44 2.4 Conclusions and Discussion

In conclusion, there are similarities in the surface crustal response to dripping or delaminating mantle lithosphere, but the models demonstrate that there may be several surface (or near-surface) diagnostic expressions of delaminating or dripping mantle lithosphere. The numerical experiments demonstrate that:

Surface topography for both drip models shows initial subsidence with a subsequent phase of uplift, the locus of which stays ﬁxed above the downwelling. Geometrically, the surface topography associated with both drip events is symmetric. In the delamination model, there is paired subsidence and uplift that migrates as the mantle lithosphere peels away from the crust. The surface topographic expression is distinctly asymmetric.

The thermal expression of delamination is characterized by a rapid increase in Moho temperature soon after the lithosphere begins to peel away. The thermal spike (up to near mantle temperature) migrates along the lithospheric gap, and the crustal temperature in this zone continues to increase while sub-lithospheric ﬂow is active. Depending on the style of dripping (i.e., how much mantle lithosphere is removed), there can be appreciable thermal perturbation of the crust

(DRIP-2), or very little (DRIP-1). Delamination seems to be more eﬃcient as a crustal heating mechanism since the mantle lithosphere gap is maintained for a longer time due to the inherent characteristics of the process. That is, with viscous dripping, surrounding cool mantle lithosphere will ﬂow into the lithospheric gap, causing a more rapid decrease of the elevated crustal temperatures. The thermal perturbations associated with dripping mantle lithosphere are centred and

ﬁxed above the downwelling, whereas with delamination the locus of peak temperature migrates as the peeling lithosphere does.

The P-T evolution of the crust above delaminating mantle lithosphere shows a clockwise path dominated by temperature increase and a modest initial pressure increase and gradual decompression. This P-T signature migrates across the crust as the mantle lithosphere delaminates, almost as a type of “a single metamorphic wave ”. With partial dripping of the mantle lithosphere (DRIP-

1), the overlying crust will show very little thermal variation, but pressure increases associated with burial/thickening can occur. With wholesale dripping of the mantle lithosphere, the crust directly above the drip may experience (slower) temperature increase and some pressure increase

45 then decrease as in the delamination model. However, unlike with delamination, where P-T path is essentially repeated (migrated) from one location to another, e.g at points of S1 (x = 1000 km) to

S2, (x = 750 km). The crust at the periphery of the drip will experience a diﬀerent P-T evolution at S1 ( x = 1000 km) and S2(x = 750 km) such as, varying burial and exhumation degrees. Mantle lithosphere delamination induces a zone of contraction and crustal thickening just above the hinge of the peeling mantle lithosphere. A region of extension occurs at the distal end of the lithospheric gap as sub-lithospheric mantle ﬂows up and laterally to replace delaminating lithosphere. The contraction zone sweeps across the crust as the mantle lithosphere peels away. The drip models also drive contraction and crustal thickening centered on the downgoing mantle lithosphere. However, in these models there is symmetric crustal extension on both sides. Depending on mantle lithosphere rheology, the deformation may be very localized to the central axis of the downwelling (DRIP-2) or broadly distributed about it (DRIP-1). Deep entrainment of crust into the descending mantle lithosphere represents an extreme mode of crustal thickening.

It is important to emphasize that our derived P-T-t paths represent the pressure-temperature history of the crustal packages to only the delamination/drip event. There may be pre-/syn-/post- tectonic events in addition to the mantle lithosphere removal event that modiﬁes the P-T-t path.

For example, with simultaneous plate convergence, burial and enhanced pressure increase could occur. The results demonstrate that the crust, follows a similar P-T pattern at three (upper, mid, and lower) levels, but with diﬀerent amplitudes. Not surprisingly, though, the most apparent signature is in the lower crust. Platt et al. (1998) have used P-T paths from rocks extracted from the Alboran seaﬂoor and in the Betics, beneath which mantle lithosphere removal is proposed.

The authors suggest that mantle drip/convective removal is more plausible to the region since

P-T paths indicate the exhumation in the metamorphic rocks for last 30 m.y. due to the crustal extension (following the overthickened lithosphere above the mantle drip) that is responsible for the exhumation of these rocks.

Here we show one delamination and two of the drip models from a range of numerical experiments that were run. Based on the convention of delamination that we are considering, namely that the whole mantle lithosphere must be removed as a coherent slice the type of model is fairly tightly conﬁned. Models with alternate rheologies and densities (but still delaminating) modiﬁed the amplitude and timing of the surface response, but not the general character. On the other hand,

46 the drip models had an unconstrained aspect that was particularly important for altering the effect on the crust: the amount of mantle lithosphere that was viscously removed. We chose DRIP-1 and DRIP-2 as representative models of partial removal and full removal, respectively. Indeed, the results show that care must be taken in intepreting large-scale lithospheric dynamics from surface observables as one geodynamic mechanism can affect the crust in different ways depending on a specific aspect of the removal event. Again, we emphasize that the purpose of this work was not to examine the parameters controlling the removal mechanisms. Rather, we have identified and shown these three experiments as representative models of the fundamental modes of removal and surface response.

One of the assumptions we have made in the models is that the dripping mantle lithosphere does not migrate laterally with respect to the overlying crust. We recognize that it is possible that a viscous mantle lithosphere instability could migrate, possibly within the presence of a larger background mantle flow field or as a result of other unbalances. The perturbation of the dripping instability can possibly start in the localities where the lower crust has experienced eclogitic metamorphism to make the crust sufficiently denser (Jull and Kelemen, 2001). Similarly, the regions with a previously thickened mantle lithosphere due to the plate convergence and a thicker crust is available locations to occur such instabilities (Conrad, 2000). Houseman et al. (2000) suggest that the nature and the proximity of the drip to the convergence zone may change, depending on the rate of the convergence from a multiple sheets like to single downwelling. Three-dimensional physical scaled analogue models of dripping mantle lithosphere showed, for example, lateral move- ment of the instability, probably as a consequence of interaction with adjacent unstable mantle lithosphere (Pysklywec and Cruden, 2004). The attendant surface deformation followed this drip migration. Also, in three-dimensional models a single drip may be surrounded by several higher topographic features in every direction, and the whole evolution of the drip may occur in a radial pattern, whereas two-dimensional models may represent the cross-sectional view of these models.

It has been suggested that a drip structure beneath the southern Sierra Nevadas has shifted over the last 4-5 m.y. (Zandt, 2003). They interpret this from a migrating locus of surface volcanism that they attribute to return ﬂow of hot mantle associated with the downgoing mantle lithosphere.

We note that our results suggest that such an observation may be more consistent with delaminating mantle lithosphere and a further investigation taking into account all the factors described

47 above (in conjunction with other geophysical evidence) may help to diﬀerentiate the mechanism.

Nevertheless, lateral motion of the drip would alter some of the conclusions we make, above, and should be taken into account when interpreting the results.

48 Bibliography

Beck, S. L., and Zandt, G., 2002, The nature of orogenic crust in the central Andes: Journal of

Geophysical Research, v. 107, B10, 2230, doi:10.1029/2000JB000124.

Bird, P., 1978, Initiation of intracontinental subduction in the Himalaya: Journal of Geophysical

Research, v. 83, p. 4975-4987.

Bird, P., 1979, Continental delamination and the Colorado Plateau: Journal of Geophysical Re-

search, v. 84, p. 7561-7571.

Buck, W. R., and. Toksoz, M. N., 1983, Thermal eﬀects of continental collision: Thickening a

variable viscosity lithosphere: Tectonophysics, v. 100, p. 53-69.

Conrad, C.P., and Molnar, P., 1997, The growth of Rayleigh-Taylor-type instabilities in the litho-

sphere for various rheological and density structures: Geophysical Journal International, v. 129,

p. 95-112.

Conrad, C.P., and Molnar, P., 1999., Convective instability of a boundary layer with temperature-

and strain-rate-dependent viscosity in terms of available buoyancy: Geophysical Journal Inter-

national, v. 139, p. 51-68.

Conrad, C.P., 2000, Convective instability of thickening mantle lithosphere: Geophysical Journal

International, v. 143, p. 52-70.

Christensen, U.R., and Yuen, D.A., 1984, The interaction of a subducting lithospheric slab with a

chemical or phase boundary: Journal of Geophysical Research, v. 89, p. 4389-4402.

Cloos, M., Sappie, B., Quarles van Uﬀord, A., Weiland, R.J., Warren, P. Q., and Mc Mahon, T.

49 P., 2005, Collisional delamination in New-Guinea: The geotectonics of subducting slab breakoﬀ:

Geological Society of America Special Paper, v. 400, 48 pages.

Ducea, M. N., and Saleeby, J. B., 1998, A case for delamination of the deep batholithic crust

beneath the Sierra Nevada, California: International Geology Review, v. 40, p. 78-93

Elkins-Tanton, L.T., 2005, Continental magmatism caused by lithospheric delamination, in Foulger,

G.R., Natland, J.H., Presnall, D.C., and Anderson, D.L., eds., Plates, plumes, and paradigms:

Geological Society of America Special Paper, v. 388, p. 449-461, doi: 10.1130/2005.2388 (27).

England, P.C., and Houseman, G.A., 1989, Extension during continental convergence, with appli-

cation to the Tibetan Plateau: Journal of Geophysical Research, v. 94, p. 17,561-17,579.

Fullsack, P., 1995, An arbitrary Lagrangian-Eulerian formulation for creeping ﬂows and its appli-

cations in tectonic models: Geophysical Journal International, v. 120, p.1-23.

Gleason, G.C., and Tullis, J., 1995, A ﬂow law for dislocation creep of quartz aggregates determined

with the molten salt cell: Tectonophysics, v. 247, p.1-23, doi: 10.1016/0040-1951(95)00011-B.

Hirth, G., Kohlstedt, D.L., 1996, Water in the oceanic upper mantle: Implications for rheology,

melt extraction and the evolution of the lithosphere: Earth and Planetary Science Letters, v.

144, p. 93-108, doi: 10.1016/0012-821X(96)00154-9.

Houseman, G.A., McKenzie, D.P., and Molnar, P., 1981, Convective instability of a thickened

boundary layer and its relevance for the thermal evolution of continental convergent belts: Jour-

nal of Geophysical Research, v. 86, p. 6115-6132.

Houseman, G.A., and Molnar, P., 1997, Gravitational (Rayleigh-Taylor) instability of layer with

non-linear viscosity and convective thinning of continental lithosphere: Geophysical Journal In-

ternational, v. 128, p.125-150.

Houseman, G., Neil, E.A., and Kohler, M.D., 2000, Lithospheric instability beneath the Trans-

verse Ranges of California: Journal of Geophysical Research, v. 105 p. 16237-16250 doi:

10.1029/2000JB900118.

Houseman, G.A., and Gemmer, L., 2007, Intra-orogenic extension driven by gravitational instabil-

50 ity: Carpathian-Pannonian orogeny: Geology, v. 35, p. 1135-1138. doi: 10.1130/G23993A.1

Jones C.H., Farmer, G.L., and Unruh J., 2004, Tectonics of Pliocene removal of lithosphere of

the Sierra Nevada, California: Geological Society of America Bulletin, v. 116, p. 1408-1422, doi:

10.1130/B25397.1.

Jull, M., and Kelemen, P.B., 2001, On the conditions for lower crustal convective instability: Journal

of Geophysical Research, v. 106, p. 6423-6446, doi: 10.1029/2000JB900357.

Kay, R.W., and Kay, M., 1993, Delamination and delamination magmatism: Tectonophysics, v.

219, p. 177-189.

Keskin, M., 2003, Magma generation by slab steepening and break oﬀ beneath a subduction accre-

tion complex: An alternative model for collision-related volcanism in Eastern Anatolia, Turkey:

Geophysical Research Letters, v.30, no. 24, 8046, doi: 10.1029/2003GL018019.

Keskin, M., 2007, Eastern Anatolia: A hotspot in a collision zone without a mantle plume: in

Foulger, G.R., and Jurdy, D.M., eds., Plates, plumes and planetary processes: Geological Society

of America Special Paper., 430 p.693-722, doi. 10.1130/2007.2430(32).

Lachenbruch, A.H., and Sass, J.H., 1977 Heat ﬂow in the United States and the thermal regime

of the crust. In: J.G. Heacock, Editor, The Earth’s Crust, its Nature and Physical Properties:

American Geophysical Union, Geophysical. Monograph. Series v. 20, p. 626-675.

Le Pourhiet, L.L., Gurnis, M., and Saleeby, J., 2006, Mantle instability beneath the Sierra Nevada

Mountains in California and Death Valley extension: Earth and Planetary Science Letters, v.

251, p. 104-119. doi: 10.1016/j.epsl.2006.08.028.

Lenardic, A., and Kaula, W. M., 1995, More thoughts on convergent crustal plateau formation and

mantle dynamics with regard to Tibet: Journal of Geophysical Research, v. 100, p. 15,193-15203.

Lorinczi, P., and Houseman, G.A., 2009, Lithospheric gravitational instability beneath the South-

east Carpathians: Tectonophysics, v. 474, p. 322-336.. doi:10.1016/j.tecto.2008.05.024

Manley, C.R., Glazner, A.F., and Farmer, G.L., 2000, Timing of volcanism in the Sierra Nevada of

California: Evidence for the Pliocene delamination of the batholithic root?: Geology, v. 28, p.

51 811-814.

Molnar, P., Houseman, G.A., and Conrad, C.P., 1998, Rayleigh-Taylor instability and convective

thinning of mechanically thickened lithosphere: eﬀects of non-linear viscosity decreasing exponen-

tially with depth and of horizontal shortening of the layer: Geophysical Journal of International,

v. 133, p. 568-584.

Morency, C., and Doin, M.P., 2004, Numerical simulations of mantle lithospheric delamination:

Journal of Geophysical Research, v. 109, B03410, doi: 10.1029/2003JB002414.

Neil, E.A., and Houseman, G.A., 1999, Rayleigh-Taylor instability of the upper mantle and its role

in intraplate orogeny: Geophysical Journal International, v. 138, p. 89-107.

Pysklywec, R.N., Beaumont, C., and Fullsack, P., 2002, Lithospheric deformation during the early

stages of continental collision: numerical experiments and comparison with South Island, New

Zealand: Journal of Geophysical Research, v. 107, 2133, doi: 10.1029/2001JB000252.

Pysklywec, R. N., and Cruden, A. R., 2004. Coupled crust-mantle dynamics and intraplate tecton-

ics: Two-dimensional numerical and three-dimensional analogue modeling, G3: Geochemistry,

Geophysics, Geosystems, 5(10), doi: 10.1029/2004GC000748, 22 pp

Platt, J.P., Soto, J.-I., Whitehouse, M.J., Hurford, A.J., and Kelley, S.P., 1998, Thermal evolution,

rate of exhumation, and tectonic signiﬁcance of metamorphic rocks from the ﬂoor of the Alboran

extensional basin, western Mediterranean: Tectonics, v. 17, p. 671-689.

Saleeby, J., and Foster, Z., 2004, Topographic response to mantle lithosphere removal, southern

Sierra Nevada region, California: Geology, v. 37, p. 245-248.

S¸eng¨or,A.M.C., Ozeren, S., Zor, E., and Genc, T., 2003, East Anatolian high plateau as a mantle-

supported, N-S shortened domal structure: Geophysical Research Letters, v. 30, p. 8045, doi:

10.1029/2003GL017858.

S¸eng¨or, A.M.C., Ozeren, M.S., Keskin, M., Sakinc, M., Ozbakir, A., and Kayan, I., 2008,

Eastern Turkish high plateau as a small Turkic-type orogen: Implications for post-collisional

crust-forming processes in Turkic-type orogens: Earth-Science Reviews, v. 90, p.1-48,

52 doi:10.1016/j.earscirev.2008.05.002.

Tackley, P.J., Stevenson, D.J., Glatzmaier, G.A., and Schubert, G., 1994, Eﬀects of multiple phase

transitions in a 3-dimensional spherical model of convection in Earth’s mantle : Journal of

Geophysical Research, v. 99 (B8), p. 15,877-15,901.

Zandt, G., 2003, The southern Sierra Nevada drip and the mantle wind direction beneath the

southwestern United States, Int. Geol. Rev. 45, 213-224.

Zandt, G., Gilbert, H., Owens, T. J., Ducea, M., Saleeby, J., Jones, C. H., 2004, Active Foundering

of a Continental Arc Root Beneath the Southern Sierra Nevada, California: Nature v. 432, p.

41-46.

53 Chapter 3

Mantle Lithosphere Delamination Driving Plateau Uplift and Synconvergent Extension in Eastern Anatolia

54 3.1 Abstract

Eastern Anatolia is the site of lithospheric thinning, plateau uplift, heating, and syn-convergent extension. Using numerical geodynamic experiments, we test the hypothesis that these tectonic anomalies are all related and the consequence of delamination of the mantle lithosphere. Our

ﬁndings indicate that delamination during plate convergence results in ∼ 2 km high plateau uplift.

The removal of mantle lithosphere induces distinct regions of contraction and thickening as well as extension and thinning of the crust. The latter occurs even within a regime of plate shortening, although it is muted with increasing plate convergence. Detachment of the delaminating slab results in minor surface topographic perturbation, but only above the delamination hinge. The plateau uplift and pattern of surface contraction and/or extension are consistent with a topographic proﬁle at 42◦E and geologically interpreted zone of synconvergent extension at eastern Anatolia.

3.2 Introduction

The Eastern Anatolian plateau has formed as part of a Himalayan-type orogenic system through the collision of the Arabian and Eurasian plates (Fig. 3.1A). Based on receiver function studies, the crust beneath the plateau is only 38-50 km thick (Zor et al., 2003); hence it has been suggested that the high topography is not isostatically supported by a thick crustal root (S¸eng¨oret al., 2003;

Keskin, 2003). Furthermore, seismic data for eastern Anatolia are interpreted as evidence for the complete absence of the mantle lithosphere beneath the plateau (Gok et al., 2007) (Fig. 3.1B) and are consistent with high heat ﬂow and volcanic activity (e.g., Nemrut, Suphan, and Agri-Ararat volcanoes) across eastern Anatolia.

Large-scale plate deformation in the region is dominated by plate convergence with shortening and contraction, but normal fault controlled extensional basins such as, Kagizman-Tuzluca, Hinis,

Karliova and Mus basins are well documented (Dhont and Chorowicz, 2006) within the plateau.

Such extensional features, as well as the presence of the young volcanics are notable because their stress orientations are inconsistent with E-W extensional deformation (Fig. 3.4D). Rather, the inferred extension seems to be oriented approximately N25◦E. GPS measurements slightly to the west of this region also indicate local extension, but directed NNW (Reilinger et al., 2006). Although there is some discrepancy in the precise orientations, the geology and geodesy both suggest extension

55 in the same general direction and contemporaneous with the dominant plate convergence.

Anderson (2005) suggests that topographic uplift with widespread volcanism in eastern Anatolia may be related to lithospheric delamination in the manner defined by Bird (1979). That is, mantle lithosphere is removed as a coherent slice by peeling away along the crust-mantle boundary or at the upper margin of the anomalously dense lower crust (Anderson, 2007). In the light of the observations given above a “slab break-off”model has been proposed by S¸engöret al. (2003) and

Keskin (2003). These studies suggest that break-oﬀ of the northward subducting oceanic Arabian plate in the last 7-8 m.y. has caused domal uplift and volcanic activity in eastern Anatolia through rising mantle. We note that although the S¸eng¨oret al. (2003) and Keskin (2003) do not use the term delamination, they implicitly assume a delamination-style separation of the mantle lithosphere from the crust prior to its detachment beneath the entire plateau.

Alternatively, Ershov and Nikishin (2004) propose a mantle plume scenario for eastern Anatolia.

However, petrological and geophysical evidence — e.g., the migration of the volcanism from north to south within the plateau and its change in the chemistry (calc-alkaline to alkaline) Keskin (2003) and seismic tomographic interpretations of the detached slab beneath the plateau Lei and Zhao

(2007) — does not favor the viability of a plume model.

Here, we propose that all the primary tectonic anomalies for Eastern Anatolia-plateau uplift and heating; but also the notable presence of syn-convergent crustal extension may be related as the coupled response of the crust to active underlying mantle dynamics during plate collision. Using computational geodynamic models, we test whether the geological and geophysical observables are consistent with delamination of the mantle lithosphere. Any mantle lithosphere removal in eastern

Anatolia progresses within a convergent plate regime, so we conduct a series of experiments with variable rates of the imposed convergence of the delaminated slab and with higher yield strength of the mantle lithosphere. The mantle lithosphere is permitted to detach and we consider how slab break-oﬀ modiﬁes the surface tectonic expression.

56 Figure 3.1: A) Topographic map of Eastern Anatolia created with generic mapping tools (GMT) showing the major tectonic boundaries. B) Lithospheric structure beneath Eastern Anatolia mod- iﬁed from S¸eng¨oret al. (2003); Dhont and Chorowicz (2006); Gok et al. (2007) and Keskin (2003).

57 3.3 Modeling Delamination

In our numerical experiments, we used a plane strain viscous-plastic ﬁnite element code, SOPALE,

(Fullsack, 1995; Pysklywec et al., 2002) (see Appendix). The configuration of the model (Fig. 3.2A) is designed as an idealized representation of the continent-continent plate boundary at eastern Ana- tolia. We impose a convergence velocity VAR = 0 to the northern-edge of the Arabian lithosphere and pin the southern-edge of “Eurasian” lithosphere at VEU = 0 (Fig. 3.2A). The top of the box is a free surface. A viscous flow law of ˙ = Aσn exp[−Q/(RT )] is used for viscous material response where, where ˙ is the strain rate, T is temperature, σ is differential stress. A, n, Q and R are the viscosity parameter, stress exponent, activation energy, and ideal gas constant, respectively. For mantle A = 4.89 × 10−17 Pa−3.5/s and Q = 535 kJ/mol are used (Hirth and Kohlstedt, 1996) and a Coulomb yield stress of 120 MPa. For continental crust A = 1.1 × 10−28 Pa−4/s, n = 4, and Q =

223 kJ/mol are used based on wet quartzite (Gleason and Tullis , 1995). Additionally, the upper crust has a brittle Coulomb behavior with an internal angle of friction φ = 15. Density, ρ, is a function of composition and temperature (using α = 2 × 10−5 1/K ).

A low viscosity (5 × 1019 Pa·s) weak zone is inserted between a portion of the crust and mantle lithosphere to initiate the delamination process (e.g., Morency and Doin, 2004). We recognize that anomalously dense (e.g., eclogitized) lower crust may also participate in the removal. However, the model set up is simpliﬁed by assuming that such crust is already descending with the mantle lithosphere.

Figure 3.2A shows the evolution of our reference model that may correspond to the evolution of the mantle lithosphere in eastern Anatolia, where VAR = 3.0 cm/yr. At time t = 1.2 m.y. the mantle lithosphere is delaminating from the crust, exposing a Moho width of ∼ 300 km to the mantle. Subsequently, hot and buoyant sublithospheric mantle flows into the region vacated by the peeling mantle lithosphere. The rapid nature of the delamination means that the high mantle temperatures are efficiently advected upwards, heating the lower crust (Gö˘gü¸sand Pysklywec,

2008a). The delaminated mantle lithosphere is bent steeply into the mantle but remains intact as a coherent plate, i.e., as opposed to a viscous dripping-type removal. Between 2.9 and 7.0 m.y., there is detachment and/or break-oﬀ of this mantle lithosphere slab. At the latest stage shown (t = 7.0 m.y.), the Eurasian mantle lithosphere on the left side undergoes more subdued delamination.

58 59

Figure 3.2: A) Progressive evolution of mantle lithosphere delamination for modeled Eastern Anatolia. B) Plots of surface topography and crustal thickness; “ext”refers to extension, “cont”refers to contraction. Dashed line at 42.6 km represents the initial crustal thickness. 3.4 Surface Topography and Crustal Deformation

At t =1.2 m.y., negative surface topography (∼ 2.8 km ) develops as the crust is pulled down by the dense delaminating mantle lithosphere (Fig. 3.2B). Flanking uplift features arise as a consequence of upward return ﬂow and replacement of less dense mantle in the mantle lithosphere gap (Fig.

3.2A). Crustal contraction, driven by entrainment toward the downgoing mantle lithosphere and the imposed convergence, is subtly visible within the Lagrangian mesh (Fig. 3.2A). Note that, the negative surface topography is observed at the surface, even though the crust has thickened by ∼ 5 km.

At t = 2.9 m.y., positive surface topography dominates with a peak near the delaminating gap where upwelling ﬂow is most vigorous. Inspection of the Lagrangian mesh elements also show that there is localized crustal contraction, with ∼ 12.5 km crustal thickening and as much as 30% shortening near the hinge (Fig. 3.2A,B). At the center of the lithospheric gap, crustal extension of ∼ 30% and thinning of as much as several kilometers are observed (Fig. 3.2A). This extension and thinning are notable as internally driven tectonic processes within the overall convergent plate regime.

By t = 7.0 m.y., the surface topography is characterized by plateau type uplift with an average values (>2 km) (Fig. 3.2B). The subsidence has disappeared since the delaminated slab has detached. (Figs. 3.2A and 3.2B). A zone of much thinner crust persists above the gap, although the extensional forcing from the delamination is being overtaken by the continued lithospheric convergence.

We illustrate how variable rates of plate shortening from VAR = 0 and VAR = 6.0 cm/yr modify the results (Fig. 3.3A and B). In the former case at t = 2.9 m.y., delamination causes an uplift and subsidence pattern above the lithospheric gap and delaminating lithosphere. Zones of crustal extension and thinning as well as contraction and thickening develop; again, paradoxically to the patterns of uplift and subsidence. However, the extension and thinning are much more pronounced with ∼ 15 km crustal thinning in this model since there is no imposed plate convergence. Clearly, the delamination process alone is eﬀective for stretching the crust. A broad topographic uplift develops by t = 7.0 m.y. (Fig. 3.3A), although it is not as regular and “plateau-like” as in the reference model (Fig. 3.2B). With an increased convergence velocity of VAR = 6.0 cm/yr there is

60 accelerated contraction and thickening of the crust during the delamination event (Fig. 3.3B). The surface topography is signiﬁcantly elevated by t = 2.9 m.y. and most of the crust has thickened from its initial value. Any subsidence due to loading from the delaminated slab is overwhelmed by uplift from the widespread crustal thickening. By t = 7.0 m.y., a broad plateau of thickened crust has developed.

In Figure 3.3C, we show a model in which the yield stress of the mantle lithosphere is doubled to σY = 240 MPa compared to the reference model. The stronger mantle lithosphere is less prone to detachment. This does not cause much diﬀerence in the evolution of surface topography and crustal thickness (cf. Figures. 3.3C and 3.2B), except at the hinge zone where the delaminating slab is hanging. At t = 3.8 m.y. the stronger mantle lithosphere is still attached and consequently there is a narrow zone of ∼ 1.3 km surface subsidence at x = 1200 km (Fig. 3.3C). At the same time, with the weaker mantle lithosphere that has detached, this subsidence has diminished to

∼ 500 m (Fig. 3.2B).

61 Figure 3.3: A) Plots of surface topography and crustal thickness when there is no convergence velocity imposed (VAR = 0). B) Plots of surface topography and crustal thickness of model when convergence velocity is increased to VAR = 6.0 cm/yr. C) Plots of surface topography and crustal thicknesses when yield stress of mantle lithosphere is doubled to σY = 240 MPa.

62 3.5 Delamination Beneath Eastern Anatolia

A comparison of model surface topography at 7.0 m.y. and present day surface topography across eastern Anatolia (at 42◦E) demonstrates a similar plateau uplift (Fig. 3.4A). It has been suggested that eastern Anatolia emerged from sea level ∼ 11 m.y. ago- on a similar timescale similar to that of modeled delamination events-. The short wavelength topographic features in the observed proﬁle are related to geomorphologic processes not included in our models. The long wavelength plateau uplift of eastern Anatolia is consistent with delamination removal of mantle lithosphere across a

∼ 500 km wide zone.

Figure 4B demonstrates that both eastern Anatolian and modeled crust is relatively thin across the middle of the plateau, however, only in the latter case is the crust thickened at the plateau

ﬂanks. Several factors may account for this. The models do not include material transformations that could result in removal of the lower parts of the thickened crust (Jull and Kelemen, 2001).

It is possible that anomalously thinner crust in the northern part of the Bitlis suture zone may be due to post-delamination removal of eclogitic- lower crust. Perhaps most signiﬁcantly, our two- dimensional models present an upper bound on the amount of crustal thickening since they do not permit extrusion of material out of the plane. For eastern Anatolia, recent geodetic measurements suggest that as much as ∼ 70% of the Arabian-Eurasian plate convergence is accommodated by lateral extrusion of the Anatolian plate (Reilinger et al., 2006). It may be that if this extrusion is permitted in the models, predicted anomalous topography from delamination may be reduced.

Most interesting is a comparison of horizontal surface strain rate (˙xx) in the model and primary structural features across eastern Anatolia (Fig. 3.4C, D). Within the model there is a zone of surface extension that corresponds with observed anomalous extensional features, such as E-W trending normal fault controlled structures (Mus, Hinis and Karliova basins and the Nemrut, and

Agri volcanic calderas). The northern and southern ends of the extensional zone are associated with contractional deformation. In the south, this zone may be represented with the large-scale

Bitlis-Zagros suture zone, and in the north, such contractional deformation may correspond to development of thrust fault controlled Pasinler and Erzurum basins.

63 Figure 3.4: A) Modeled (reference model; 7 m.y.) and observed surface topography at 42◦E. B) Modeled and observed Zor et al. (2003) crustal thickness both were taken along 42◦E. C) Modeled horizontal surface strain rate (˙xx), where in our convention, positive strain rate is extensional. D) Structural map of Eastern Anatolia. All modeled plots are scaled into a zone of 610 km within the plateau.

64 3.6 Conclusions and Discussion

The geodynamic experiments demonstrate that delamination causes surface uplift as a result of the isostatic and dynamic eﬀect of lithospheric removal. The uplift is enhanced and evened into a plateau by plate shortening. A pulse of (migrating) subsidence can develop as the delaminating lithosphere loads the lithosphere at the hinge zone.

Lithospheric delamination causes distinct zones of contraction and thickening (at the plateau

ﬂanks) and extension and thinning (within the plateau, to the far side of the delaminating hinge) within the crust. The crustal extension and thinning can occur within an overall plate convergent regime, but it becomes more muted with higher rates of plate shortening.

Such syn-convergent extension is a common, yet largely enigmatic phenomena at many collisional environments. For example, it has been observed at the Apennines-Tyrrhenian, Himalayas, and Alboran Sea-Rif/Betics. The Alboran Sea is currently undergoing subsidence rather than plateau uplift, but as our experiments show, surface subsidence is the early stage response to mantle lithosphere delamination before the inversion to uplift (Fig. 3.2B).

Detachment (or break-off) of the delaminated slab modifies the surface topography. However, the effect is confined largely to a narrow region close to the delamination hinge (i.e., within ∼ 100 km). We suggest that the surface effects of detachment (sensu stricto) do not span a large, well- developed continental collision like Eastern Anatolia.

The upwelling mantle flow with delamination also has significant thermal and metamorphic consequences for the crust. We do not focus on the thermal consequences of delamination in this contribution, but different study demonstrates that delamination removal of the mantle lithosphere results in rapid heating of the base of the crust and likely a migrating pulse of high-temperature/low- pressure metamorphism (Gö˘gü¸sand Pysklywec, 2008b). Thus, delamination of mantle lithosphere would possibly reconcile the high heat flow and volcanism that occurs across Eastern Anatolia

(Keskin, 2003).

65 Bibliography

Anderson, D.L., 2005, Large Igneous Provinces, Delamination, and Fertile Mantle Elements, v. 1,

p. 271-275, doi: 10.2113/gselements.1.5.271.

Anderson, D.L., 2007, The eclogite engine: Chemical geodynamics as a Galileo thermometer, in

Foulger, G.R., and Jurdy, D.M., eds., Plates, plumes, and planetary processes: Boulder, Col-

orado, Geological Society of America Special Paper, v. 430, p. 47-64, doi: 10.1130/2007.2430(03).

Bird, P., 1979, Continental delamination and the Colorado Plateau: Journal of Geophysical Re-

search, v. 84, p. 7561-7571.

Dhont, D., and Chorowicz, J., 2006, Review of the neotectonics of the Eastern Turkish-Armenian

Plateau by geomorphic analysis of digital elevation model imagery: International Journal of

Earth Sciences, v. 95, p. 34-49, doi: 10.1007/s00531-005-0020-3.

Ershov, A.V., and Nikishin, A.M., 2004, Recent geodynamics of the Caucasus-Arabia-East Africa

Region: Geotectonics, v. 38, no. 2, p. 123-136.

Fullsack, P., 1995, An arbitrary Lagrangian-Eulerian formulation for creeping ﬂows and its appli-

cations in tectonic models: Geophysical Journal International, v. 120, p.1-23.

Gleason, G.C., and Tullis, J., 1995, A ﬂow law for dislocation creep of quartz aggregates determined

with the molten salt cell: Tectonophysics, v. 247, p.1-23, doi: 10.1016/0040-1951(95)00011-B.

G¨o˘g¨u¸s,O.H., and Pysklywec, R.N., 2008a, Mantle lithosphere delamination driving plateau up-

lift and synconvergent extension in eastern Anatolia: Geology, v. 36, no. 9, p.723-726, doi:

10.1130/G2498A.1.

66 G¨o˘g¨u¸s,O.H., and Pysklywec, R.N., 2008b, Near-surface diagnostics of dripping or delaminating

lithosphere: Journal of Geophysical Research, v. 113, B11404, doi: 10.1029/2007JB005123.

Gok, R., Pasyanos, M., and Zor, E., 2007, Lithospheric structure of the continent-continent col-

lision zone: Eastern Turkey: Geophysical Journal International, v. 169, p. 1079-1088, doi:

10.1111/j.1365-246X.2006.03288.x.

Hirth, G., Kohlstedt, D.L., 1996, Water in the oceanic upper mantle: Implications for rheology,

melt extraction and the evolution of the lithosphere: Earth and Planetary Science Letters, v.

144, p. 93-108, doi: 10.1016/0012-821X(96)00154-9.

Keskin, M., 2003, Magma generation by slab steepening and break oﬀ beneath a subduction accre-

tion complex: An alternative model for collision-related volcanism in Eastern Anatolia, Turkey:

Geophysical Research Letters, v.30, no. 24, 8046, doi: 10.1029/2003GL018019.

Lei, J., and Zhao, D., 2007, Teleseismic evidence for a break-oﬀ subducting slab under Eastern

Turkey: Earth and Planetary Science Letters, v. 257, p. 14-28, doi: 10.1016/j.epsl.2007.02.011.

Jull, M., and Kelemen, P.B., 2001, On the conditions for lower crustal convective instability: Journal

of Geophysical Research, v. 106, p. 6423-6446, doi: 10.1029/2000JB900357..

Morency, C., and Doin, M.P., 2004, Numerical simulations of mantle lithospheric delamination:

Journal of Geophysical Research, v. 109, B03410, doi: 10.1029/2003JB002414

Pysklywec, R.N., Beaumont, C., and Fullsack, P., 2002, Lithospheric deformation during the early

stages of continental collision: numerical experiments and comparison with South Island, New

Zealand: Journal of Geophysical Research, v. 107, 2133, doi: 10.1029/2001JB000252.

Reilinger, R., and 24 others, 2006, GPS constraints on continental deformation in the Africa-

Arabia-Eurasia continental collision zone and implications for the dynamics of plate interactions:

Journal of Geophysical Research, v. 111, B05411, doi: 10.1029/2005JB004051

S¸eng¨or,A.M.C., Ozeren, S., Zor, E., and Genc, T., 2003, East Anatolian high plateau as a mantle-

supported, N-S shortened domal structure: Geophysical Research Letters, v. 30, p. 8045, doi:

10.1029/2003GL017858.

67 Zor, E., Sandvol, E., Gurbuz, C., Turkelli, N., Seber, D., and Barazangi, M., 2003, The crustal

structure of the East Anatolian plateau from receiver functions: Geophysical Research Letters,

v. 30, p. 8044, doi: 10.1029/2003GL018192.

68 Chapter 4

The Surface Tectonics of Mantle Lithosphere Delamination Following Ocean Lithosphere Subduction: Insights From Physical Scaled Analogue Experiments

69 4.1 Abstract

Many postulated lithospheric removal events occur in regions with an earlier history of subduction but the relationship between the two processes has not been explored. In this work, we use physical scaled analogue experiments to investigate the evolution from ocean lithosphere subduction to collision and possible delamination of the mantle lithosphere from the crust. We test how varying the magnitude of plate convergence alters the behavior of the subduction-delamination model. Our experiments show that a subducting/retreating ocean pro-plate can evolve to continental mantle lithosphere delamination. Negative surface topography is supported at the delamination hinge and this migrates back with the peeling lithosphere. With high plate convergence, delamination is suppressed. Rather, the crust and mantle lithosphere split at the collision zone in a form of

ﬂake tectonics as oncoming pro-crust is accreted on top of the retro-plate and the pro-mantle lithosphere subducts below. Localized high topography develops at this zone of crustal accretion and thickening. The results suggest that delamination may be a continental continuation of plate retreat and that lithospheric removal is triggered by the transition from one process to another.

4.2 Introduction

Recent geological and geophysical studies have interpreted that cold and dense mantle lithosphere slab can peel away from the crust as a coherent slice in a plate/slab-like manner. Here, we adopt the term “delamination ” to describe this speciﬁc process, as deﬁned by Bird (1979). The delaminated mantle lithosphere is replaced by hot and buoyant sublithospheric mantle and this may account for the uplift of high topography without a thick lithosphere as well as thermal perturbation of the crust as observed in a number of tectonic regions (Bird, 1979; Seber et al., 1996; Cloos et al., 2005;

Le Pourhiet et al., 2006; G¨o˘g¨u¸sand Pysklywec, 2008a,b).

Delamination of the mantle lithosphere has been interpreted in particular for plate convergent zones. For example, seismological and geochemical evidence indicates completely or partially delaminated mantle lithosphere structure beneath the East Anatolian plateau (Gok et al., 2007;

Keskin, 2007; G¨o˘g¨u¸sand Pysklywec, 2008a) (Figure 4.1), the Altiplano-Puna (Kay and Kay, 1993;

Beck and Zandt, 2002), the Tibetan plateau (Bird, 1978; Meissner and Mooney, 1998; Kosarev et al., 1999), the Rif-Betics (Fadil et al., 2006; Seber et al., 1996), New Guinea (Cloos et al., 2005), the

70 Apennines (Reutter et al., 1980; Channell and Mareschal, 1989), and the Carpathian-Pannonian region (Houseman and Gemmer, 2007; Lorinczi and Houseman, 2009). In these instances the delamination occurs either contemporaneously with or following an active orogenic/collisional event.

Furthermore, the geological “preconditions” for delamination have been qualitatively described in the context of subduction to continental plate convergence (Bird, 1978; Meissner and Mooney, 1998;

Massone, 2005; S¸eng¨oret al., 2008).

There are a relatively few number of geodynamic modeling studies that have considered the geodynamics of delamination, but these provide some insight into the compositional, rheological, and geometric controls on delamination of the mantle lithosphere from the crust. A primary control on delaminationand other similar processes related to removal of the lithosphere— is the presence of a signiﬁcant crustal or mantle density anomaly (Bird, 1978, 1979; Kay and Kay, 1993; Schott and Schmeling, 1988; Jull and Kelemen, 2001). It may be that such density variations in nature can arise from the phase changes within eclogitized lower crust (Dewey et al., 1993; Anderson,

2005, 2007). Phanerozoic lithospheric mantle is also prone to be gravitationally unstable due to its chemical composition. (Poudjom Djomani et al., 2001; O’Reilly et al., 2001). Along the same lines as subduction related eclogitization, Schott and Schmeling (1988) suggest that anomalously dense and thick lithospheric mantle could initiate a delamination event in a plate convergence zone.

They indicate that thickening lithospheric roots must be at least 100-170 km deep to reach suﬃcient density to become potentially unstable. Furthermore, it has been suggested that delamination may facilitate/precede continental subduction and slab break-oﬀ when such density anomalies exist

(Chemenda et al., 2000; Regard et al., 2003; Toussaint et al., 2004; Boutelier et al., 2004; De

Franco et al., 2008).

Even with signiﬁcant density anomalies, the thermal-rheological conditions must also be amenable to delamination. Using temperature-dependent numerical experiments, Morency and Doin (2004) suggest that delamination can only occur when Moho temperatures exceed 800◦C. This results in a mechanical decoupling between the crust and mantle lithosphere along the weak lower crust (Bird,

1979; Meissner and Mooney, 1998; Ranalli , 2000). In numerical experiments with a brittle-viscous crust and mantle lithosphere reaching effective viscosities of 1×1023 Pa·s, a viscosity of 5×1019 Pa·s of the (hot) lower crust effectively decouples the lithosphere (Gö˘gü¸sand Pysklywec, 2008a,b).

An aspect of (continental) delamination that has not been considered is the role of preceding

71 ocean plate subduction in the mantle lithosphere removal event. Regard et al. (2003) argue that the dynamics of previously subducting oceanic slabs at depth can inﬂuence the evolution of the subsequent continental collision event (slab break-oﬀ, continental subduction). Royden (1993) suggests that depending on the style of the previously subducting plate (e.g., whether it was advancing or retreating) there may be various types of surface topographic/crustal responses at the continental collision zone. Although Royden (1993) does not explicitly consider continental delamination, the tectonic framework is discussed in the context of Mediterranean tectonics where delamination has been subsequently suggested (references given above).

In making the link between subduction retreat and delamination, we note that the processes are similar except that delamination occurs beneath the buoyant (orogenic) continental crust. More speciﬁcally, one might argue that delamination is akin to subduction retreat except that the former separates the crust from the mantle lithosphere whereas the latter consumes the surface crust. As an example, the transition from retreat of the subducting slab to delamination is proposed for

Eastern Anatolia (Figure 4.1). Eastern Anatolia is part of the Neo-tethyan oceanic realm where

Bitlis Poturge mantle lithosphere (BPML) was subducted beneath the Pontide arc (S¸eng¨oret al.,

2003; Keskin, 2007). The ocean closed in Late Oligocene-Miocene times when the East Anatolian accretionary complex (EAAC) made its initial contact with the Bitlis-Poturge Massif (BPM). The northward subducting BPML starts to peel away/delaminate from the East Anatolian Accretionary

Complex (EAAC) about 13 Ma. The delamination of the BPML below the EAAC leads to upwelling of the sublithospheric mantle into shallow crustal regions and widespread magmatism

(migrating from North to South) and plateau uplift (Keskin, 2007; S¸eng¨oret al., 2003, 2008).

Recent results from lithospheric-scale numerical experiments of delamination identify a delaminating hinge — the separation point between the peeling mantle lithosphere and crust — that progressively migrates/rolls back while the model is evolving (G¨o˘g¨u¸sand Pysklywec, 2008a,b).

This delamination hinge is analogous to a retreating “S-point” in subduction-like behaviour of the lithosphere (Willett et al., 1993).

Despite the potential relationship between oceanic subduction and continental delamination, previous studies have only considered each of these in isolation. In this study, we investigate how continental delamination evolves directly following a phase of ocean plate subduction. In addition to exploring the style of lithospheric removal, the purpose of this work is to determine how

72 surface topography/deformation evolve during subduction-delamination. The research is carried out using scaled analogue modeling techniques of the coupled upper mantle-crust system. Such modeling techniques have advantages over numerical modeling in inherently considering the three- dimensional behaviour of the system and capturing discrete deformational features (Funiciello et al., 2003, 2006; Schellart, 2004). This is balanced by a limitation of analogue techniques in modeling coupled thermal-mechanical dynamics. Although there is a signiﬁcant body of research that has involved analogue modeling of upper mantle scale subduction processes and Rayleigh-Taylor viscous instabilities, the delamination process has not been explored with such techniques. Initially, we explore the ﬁrst-order physical conditions for delamination in analogue modeling. Then, we present a series of experiments that start with idealized ocean plate subduction and investigate the possible transition to continental delamination.

73 Figure 4.1: Simplified tectonic evolution of the Eastern Anatolian plateau along 42◦E during the last 13 m.y. Modified from Keskin (2007) and S¸engöret al. (2003, 2008).

74 4.3 Experimental Design

4.3.1 Model Description

The experiments were run in a 25 cm (width) x 55 cm (length) x 25 cm (depth) rectangular

Plexiglas tank (Fig. 4.2) and were set up as a scaled upper mantle depth model of pro-plate collision/subduction beneath a retro-plate. Following Willett et al. (1993) the terms “pro” and

“retro” are used to desribe the subduction geometry, where “pro” refers to the subducting plate side and “retro” to the overriding plate side.

The lowest layer/sublithospheric mantle (abbreviated as slm) is modeled using glucose syrup

3 (ρslm = 1425 kg/m , ηslm = 26.75 Pa·s) (Espurt et al., 2008; Guillaume et al., 2009) (Table 1). The 1.3 cm thick pro-plate mantle lithosphere is modeled by a Newtonian viscous silicone putty

(Rhodorsil Gomme + PBDMS + galena ﬁllers) (Funiciello et al., 2003, 2006; Bellahsen et al., 2005).

Silicone putty is a viscoelastic material that behaves viscously at experimental strain rates. That is, the experimental time scale — on the order of minutes (min) — is always higher than the Maxwell relaxation time (order of seconds) (Weijermars and Schmeling, 1986). For the reference experiment

(DEL-12), the density and viscosity of the pro-plate mantle lithosphere (abbreviated as ml) are

3 5 ρml = 1507 kg/m and ηslm = 3.5 × 10 Pa·s, respectively (Table 1). PBDMS (polyborondimethyl- siloxane) is the boron derivative of PDMS (polydimethylsiloxane) and it is used to increase the viscosity of Rhodorsil Gomme (2.9 × 104 Pa·s at 20◦C) by an order of magnitude (Weijermars and

Schmeling, 1986). In our model set up, the lower crust is modeled by glucose syrup (the same material as the sublithospheric mantle) to behave as a weak (presumably hot) lower crust or weak layer within the crust that would be the decoupling zone between the mantle lithosphere and upper crust (Fig. 4.2). The existence of a weak intracrustal layer has been suggested in previous studies

(Turcotte et al., 1984). The upper crust and retro-plate lithosphere are made up of viscous silicone putty (Rhodorsil Gomme + PBDMS + galena ﬁllers) of similar viscosity to the pro-mantle lithosphere, but of lower density (Table 1).

These materials were used to simplify the stratified temperature-dependent Earth rheological profile respecting standard scaling procedures of length, density, stress, and viscosity in a natural gravity field, as described by Weijermars and Schmeling (1986). The writing of the scaling scheme follows Pysklywec and Cruden (2004) and Cruden et al. (2006). Scaling relationships of the reference

75 experiment and general experimental parameters are listed in Table 1 and Table 2, respectively.

The experiments were designed using the physical properties and the length scale of the mantle lithosphere material as the main constraints. For practical purposes we deﬁne a length scale based on a model lithospheric thickness, lm = 0.013 m, which for a natural continental lithospheric thick-

−7 ness ln = 80 km gives a length scale ratio L = lm/ln = 1.63 × 10 (where subscripts m and n refer to nature and laboratory and model scales, respectively). The silicone putty mixture (Rhodorsil

3 Gomme + PBDMS + galena ﬁlters) has a density of ρm = 1507 kg/m that models a natural density

3 of pro-plate mantle lithosphere ρm = 3300 kg/m This sets a density scale of P = ρm/ρn = 0.46. In these gravity-driven experiments at 1 g the gravity scale ratio is G = gm/gn = 1.0. The uncertainty in density measurements is ± 10 kg/m3. The rheological properties of all silicone materials used in this study were measured with a concentric cyclinder viscometer over a range of strain rates relevant to the experiments (Guillaume et al., 2009). The precision of this method is ± 5%. The viscosity of dense mantle lithosphere/silicone putty was measured to be 3.5 × 105 Pa·s. Assuming a viscosity of η = 10 ×24 Pa·s for the natural oceanic mantle lithosphere deﬁnes a viscosity scale

−19 ratio M = (ηm/ηn) = 3.5 × 10 . The timescale ratio for the experiments can now be deﬁned

−12 as T = M/P L = tm/tn = 4.72 × 10 . Based on this scaling ratio, one million years (1 m.y.) corresponds to 148 s in the model.

The sidewalls and the bottom of the box have no-slip boundary conditions and the top surface is a free boundary. Except for the pro-plate (because it is ﬁxed to the indentor), it is assumed that each plate is surrounded by weak fault zones (trench and transform faults). Horizontal shortening is achieved by displacing a rigid indentor at a constant velocity perpendicular to the plate margin.

The indentor extends only into the upper part of the glucose syrup and the syrup is free to move underneath it. The experiments are isothermal, so the role of thermal convection is not taken into account. As such, the flow in the syrup is produced only by plate convergence and the internal dynamics of the subducting/colliding plates. Further, mantle flow is only generated by these internal dynamics and clearly we do not consider any influence of larger scale global flow effects

(Hager and O’Connell, 1978; Ricard et al., 1991). The closed bottom of the Plexiglas tank means that the models represent only upper mantle ﬂow. This assumption has signiﬁcant implications for the dynamics of the slabs in the subsequent modeling results where the bottom boundary of the box is modeled to be the impermeable barrier of the upper-lower mantle discontinuity. Some

76 justification for this assumption includes estimates of mean mantle viscosity that suggest there is an increase in viscosity of up to two orders of magnitude from the upper to lower mantle through this region (Mitrovica and Forte, 1997) that would inhibit flow through this level. Furthermore, the endothermic phase transformation of olivine at 670 km depth may introduce buoyancy effects that may significantly hinder mass flux across the upper mantle-lower mantle boundary (Christensen and Yuen, 1984, 1985).

Oceanic lithosphere subduction is initiated in the experiments by forcing downward the leading edge of the mantle lithosphere into the glucose syrup at 45◦ to a depth of 3 cm (Fig. 4.2). After this, the buoyancy of the slab drives the subduction dynamics as the piston moves the pro-plate towards the retro-plate. We test how varying the magnitude of plate convergence alters the behavior of the subduction-delamination model. Previous studies show that faster plate convergence is associated with advancing plate subduction, whereas slower plate convergence is associated with retreating plate subduction (Bellahsen et al., 2005; Schellart, 2005; Faccenna et al., 2007; Funiciello et al., 2008). Our experiments demonstrate how the continental pro-plate behaves during collision following this subduction phase.

Each experiment was monitored using a sequence of side- and top-view photographs taken during progression of the experiments. The surface topography evolution of the experiments was monitored with a topography scanner (Real Scan USB model 300; measurement uncertainty is

± 0.5 mm).

77 Figure 4.2: Top-view and side-view schematic of the experiment set up. The silicone layers represent the retro-plate and pro-plate lithosphere and glucose syrup represents the sublithospheric mantle and lower crust.

78 Parameter Units Nature Reference model

g gravitational acceleration m·s−2 9.81 9.81

Thickness m

h mantle lithosphere 80000 0.013

H upper mantle 615385 0.10

−7 scale factor for length Lmodel/Lnature = 1.63 × 10

Density kg·m−3

ρlit mantle lithosphere 3300 1507

ρslm sublithospheric mantle 3125 1425

ρrp retro-plate 3052 1400

scale factor for density ∆ρmodel/∆ρnature = 0.46

Viscosity Pa·s

24 5 ηlit mantle lithosphere 10 3.5 × 10

19 ηslm sublithospheric mantle 7.64 × 10 26.75

23 5 ηrp retro-plate 7.42 × 10 2.6 × 10

−19 scale factor for viscosity ηmodel/ηnature = 3.5 × 10

Characteristic time s

13 T tmodel/tnature 148 3.16 × 10 (∼ 1Ma) =

(∆ρgL)nature/(∆ρgL)model ×

−12 ηmodel/ηnature = 4.72 × 10

Characteristic velocity m·s−1

−5 U Umodel/Unature 4.1 × 10 3.81 cm/year

= (tnature × Lmodel)/(tmodel ×

4 Lnature) = 3.45 × 10

Table 4.1: Scaling relationships of the reference experiment (DEL-12).

79 Experiment DEL 4 DEL 5 DEL 6 DEL 9 DEL 10 DEL 11 DEL 12 DEL 13

Piston velocity m·s−1 0 0 0 0 1.6 × 10−4 8.3 × 10−5 4.1 × 10−5 6.3 × 10−5

Pro-plate — Mantle lithosphere

Width Wml m 0.14 0.16 0.15 0.15 0.14 0.14 0.15 0.15

Length Lml m 0.37 0.37 0.38 0.37 0.36 0.35 0.35 0.34

Thickness hml m 0.013 0.013 0.013 0.013 0.013 0.013 0.013 0.013

−3 Density ρml kg·m 1507 1504 1507 1507 1507 1507 1507 1507

5 5 5 5 5 5 5 5 Viscosity ηml Pa·s 3.5 × 10 3.5 × 10 3.5 × 10 3.5 × 10 3.5 × 10 3.5 × 10 3.5 × 10 3.5 × 10

Pro-plate — Upper crust

Width Wuc m 0.093 0.1 0.1 0.1 0.1 0.1 0.095 0.095

Length Luc m 0.195 0.197 0.195 0.2 0.21 0.2 0.2 0.2

Thickness huc m 0.003 0.003 0.003 0.003 0.003 0.003 0.003 0.003

−3 Density ρuc kg·m 1462 1417 1417 1466 1411 1400 1400 1400

5 5 5 5 5 5 5 5 Viscosity ηuc Pa·s 2.6 × 10 2.6 × 10 2.6 × 10 2.6 × 10 2.6 × 10 2.6 × 10 2.6 × 10 2.6 × 10

Pro-plate — Lower crust

Width Wlc m 0.1 0.11 0.12 0.12 0.1 0.1 0.11 0.1

Length Llc m 0.2 0.2 0.22 0.23 0.21 0.21 0.21 0.2

Thickness hlc m 0.003 0.003 0.003 0.003 0.003 0.003 0.003 0.003

−3 Density ρlc kg·m 1422 1423 1420 1420 1424 1420 1420 1420

Viscosity ηlc Pa·s 26.75 26.75 241.93 241.93 26.75 26.75 26.75 26.75

Retro-plate

Width Wret m - - - - 0.15 0.15 0.15 0.15

Length Lret m - - - - 0.15 0.15 0.15 0.16

Thickness hret m - - - - 0.013 0.013 0.013 0.013

−3 Density ρret kg·m - - - - 1411 1400 1400 1400

5 5 5 5 Viscosity ηret Pa·s - - - - 2.6 × 10 2.6 × 10 2.6 × 10 2.6 × 10

Sublithospheric mantle

Thickness hslm m 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1

−3 Density ρslm kg·m 1422 1423 1420 1420 1424 1420 1425 1420

Viscosity ηslm Pa·s 26.75 26.75 26.75 26.75 26.75 26.75 26.75 26.75

Table 4.2: Experimental parameters. See also Figure 4.1 for the description of W, h, L.

80 Two experiments (DEL-4 and DEL-5) are speciﬁcally designed to test the role of upper crustal density where the viscosity of the lower crust/glucose syrup (ηlc= 26.75 Pa·s) and all other parameters are kept constant (Fig. 4.3A and B). For DEL-4, the buoyancy of the upper crust (silicone

3 putty) is chosen to be ρuc = 1462 kg/m . The experiment begins when the edge of the mantle lithosphere is pushed ∼ 3 cm into the sublithospheric mantle. The dense mantle lithosphere slab initially begins to sink into the sublithospheric mantle at a high dip angle. Over time, the dense mantle lithosphere sinks into the model mantle by a retreat-like process. By t = 10 mins, (Fig.

4.3A) the mantle lithosphere slab and crust are both being consumed into the mantle, although this does not occur consistently along the length of the subduction zone: there was more subduction at the lower part of the box frame compared to the upper part of the frame. The subduction results in the intrusion of the transparent sublithospheric mantle above the consumed lithosphere (Fig.

4.3A). The experiment does not show any delamination of the crust from the mantle lithosphere and is essentially a pure continental subduction model. The model is allowed to evolve until t =

15 mins and by that time ∼ 60% of the upper crust is subducted into the sublithospheric mantle.

In experiment DEL-5, the density of the upper crust (silicone putty) is decreased to ρuc = 1417 kg/m3 so that it is buoyant (by ∼ 6 kg/m3) with respect to the glucose syrup (sublithospheric mantle). This buoyancy contrast is smaller than the estimated uncertainty in density measurements (± 10 kg/m3). However, following convention (Espurt et al., 2008; Guillaume et al., 2009) we also evaluated qualitatively the buoyancy diﬀerences (∆ρ) between crust and sublithospheric mantle, and mantle lithosphere/retro-plate and sublithospheric mantle. We determined from simple buoyancy tests that the crust was less dense than the sublithospheric mantle. The dense mantle lithosphere slab sinks into sublithospheric mantle and when the slab reaches the bottom of the box, the mantle lithosphere starts to separate from the light-coloured buoyant upper crust. At t =12 mins, the upper crust and mantle lithosphere have delaminated from each other about 6 cm where the buoyant crust is now underlain by transparent sublithospheric mantle (Fig. 4.3B). The delamination is more advanced in the centre of the model along-strike to the delamination boundary (or

“delaminating hinge”). The experiment continues for about 25 mins and at termination almost

80% of the upper crust has delaminated from the mantle lithosphere.

Previous numerical studies by G¨o˘g¨u¸sand Pysklywec (2008b) suggest that an increase in the lower crustal/weak zone viscosity between crust and mantle lithosphere would prevent delamination

81 or retard the process signiﬁcantly. The next two experiments test the role of a higher lower crustal viscosity (Table 2).

In model DEL-6, we used an alternative glucose syrup material that has the same physical properties as the previous one except its viscosity is ηlc = 242 Pa·s. In this experiment the evolution of the mantle lithosphere slab is similar to DEL-5 as the slab initially begins to sink with high angle and delaminates back as the model evolves. However, the timescale of the lithospheric removal in the model is longer owing to the higher viscosity lower crust. The delamination now begins at t

= 37 mins and the higher viscosity lower crust modiﬁes the morphology of the delaminated slab, holding it at a shallow angle near the surface then steepening to near vertical at depth (Fig 4.3C).

Furthermore, the delamination is more developed at the upper part of the top-view frame (i.e., along the strike of the delamination hinge) compared to the lower part of the top-view frame. We do not attribute this to have geodynamic signiﬁcance; it is most likely an artifact of the imperfect setup of the experiment.

Experiment DEL-9 is the same as DEL-4, except that the lower crustal viscosity is increased to

ηlc = 242 Pa·s (Table 4.2). The mantle lithosphere and crust begin to subduct together just before the slab hits the bottom of the box at t = 14 mins (Fig 4.3D). The experiment lasts about t= 45 mins and by the time it ends, about %40 of the upper crust has subducted. In general, the model is similar to DEL-4: they show continental subduction of a coupled crust and mantle lithosphere, however, in DEL-9 the process is slower, apparently hindered by the higher lower crustal viscosity.

In all, the test experiments confirmed findings from numerical studies on the basic conditions for delamination and provided information on the scaled conditions for designing the physical analogue delamination models. One of the primary factors controlling delamination of the mantle lithosphere from the crust is the buoyancy difference between crust and sublithospheric mantle

(∆ρ = ρuc − ρslm). The density of the upper crust should be lower than that of sublithospheric mantle to develop delamination, assuming a suitable lower crustal viscosity is chosen. We note that there are certainly factors besides viscosity of the lower crust and density contrast of the crust and mantle lithosphere that control delamination. However, a sensitivity analysis to all of these factors is outside of the scope of this contribution.

82 Figure 4.3: Topview (left) and sideview (right) photos of the experiments A) DEL-4 (t = 10 mins) B) DEL-5 (t = 12 mins) C) DEL-6 (t = 40 mins) D) DEL-9 (t = 40 mins). The physical parameters of these experiments are given in the text and in Table 2. DEL-4 and DEL-9 do not show delamination; instead, upper crust and the mantle lithosphere descend together. DEL-5 and DEL-6 show delamination between the upper crust and mantle lithosphere slab.

83 4.4 Experimental Results

Experiments were set up to test the role of plate convergence on continental delamination with pre-existing ocean lithosphere. Here, we show the results of four selected experiments (DEL-12,

DEL-13, DEL-11 and DEL-10) where experimental descriptions are given in Table 2.

4.4.1 Experiment DEL-12 (Vp = 0.25 cm/min)

In this experiment, the pro-plate is pushed at a constant velocity of Vp = 0.25 cm/min by a piston moving perpendicular to the trench (Fig 4.4). The convergent pro-mantle lithosphere subducts beneath the retro-plate at a high angle, almost 90◦, as the imposed convergence begins and the slab hits the bottom of the box at t = 8 mins. After reaching the bottom boundary, the subducted slab lays along the boundary and the (still converging) pro-plate mantle lithosphere begins to peel away from the crust as delamination begins. By t = 16 mins, the mantle lithosphere has delaminated

∼ 5 cm, as the delamination hinge (dh) has migrated to the right. This is visible from the top view

(Fig. 4.4A) as light coloured sub-lithospheric material intrudes beneath the convergent crust (on the right). From the side view, the delamination, although somewhat less apparent, manifests as a developing gap between the sand coloured retro-plate and pro-plate mantle lithosphere plates.

The delamination of the mantle lithosphere progresses and by t = 22 mins, a 7.5 cm wide section of the mantle lithosphere has peeled away, exposing the crust to the sublithospheric mantle (Fig.

4.4A). In the direction along-strike to the plate boundary, the delamination is more developed in the centre of the continental plate compared to the edges. Implicit with this delamination is an active upwards return ﬂow of sublithospheric mantle into the delamination gap. In the top view photo, the increased intrusion of light coloured/transparent sublithospheric mantle under the pro-plate upper crust shows the result of this ﬂow.

At t= 33 mins, a ∼ 10 cm long mantle lithosphere slab is lying at the bottom of the box and the delamination gap has opened to ∼ 10 cm . Clearly, the weight of the mantle lithosphere slab is suﬃcient to drive delamination, even as the plate is being pushed forward at 0.25 cm/min. As explained in the introduction, this is eﬀectively subduction retreat/roll-back but distinguished as delamination since the mantle lithosphere is rolling back beneath a buoyant (and converging, in this case) continental crust.

84 Laser data in Figure 4.4B shows the evolution of surface topography along the X-X0 cross section line for t= 16 mins, 22 mins, and 33 mins. The zero topography line is somewhat arbitrary and we have chosen it to correspond to the topography at the left edge of the retro-plate. The precision of the topography measurements is high enough that the profiles show short-wavelength/small amplitude variations associated with the inherent roughness of the background surface. A semi- periodic variation of the short-wavelength topography is caused by the surface grid lines. We plotted the central (X-X0) cross section for the average topography profile in the experiment since profiles near the edges may not reflect clearly the actual delamination induced topography. Previous studies also show that subduction (here delamination) develops faster in the centre of a model lithospheric plate compared to the edges (Funiciello et al., 2003; Schellart, 2004).

By t = 16 mins, the topography signal across the X-X0 section is characterized by distinctly diﬀerent behaviours at the three diﬀerent regions of the plate surface. (1) Across the retro-plate

(0 - 14.5 cm) the topography is slightly negative, sloping down from zero at the left edge to

∼ −0.13 cm. This tilt is likely the response of the surface to underlying mantle ﬂow where the downgoing slab in the centre of the model drives a negative dynamic topography. This type of dynamic topography is consistent with long-wavelength tilted continental interiors that have been interpreted from stratigraphic analyses and computational modeling (e.g., Pysklywec and Mitrovica,

1999). (2) At the boundary between the pro-plate and retro-plate (at x = 15.0 cm) the surface topography reaches -0.6 cm. This reflects the initial conditions of the model where the distinct crust of the retro-plate and pro-plate are joined after the collision. The feature is not geologically significant. (3) Across the pro-plate, there is more marked subsidence that reaches a maximum of - 0.33 cm at x = 21.0 cm. This corresponds to the location where the delaminating slab is attached the crust suggesting that the subsidence is caused as the crust is pulled down by the weight of the delaminating slab. We note that other models of delamination have shown plateau uplift above the delamination gap (Gö˘gü¸sand Pysklywec, 2008a,b). In these analogue models there is no thermal buoyancy of the sublithospheric mantle to“lift” a plateau along this gap, so plateau-type topography does not develop. The pro-plate to the right of the hinge is also deepened by the weight of the delaminated slab, whereas to the left of the hinge, the subsidence is caused by sublithospheric flow (dynamic topography).

At t = 22 mins, there is a similar pattern of topography across the section, but there is an overall

85 uplift of the signal. That is, across the retro-plate and delamination gap the topography is slightly negative (∼ −0.08 cm). The area is still likely being drawn down by a dynamic topography effect of the sublithospheric flow, but this effect may be reducing as the delaminating slab migrates to the right as it rolls back. In addition, some of the uplift may be owing to continued thickening and isostatic uplift of the crust in the convergent model (see below). The trough in surface topography at the delamination hinge has migrated to the right with the migration of the hinge. Its reduced magnitude is likely a result of isostatic uplift as the crust thickens. Essentially the interplay between mantle lithosphere/sublithosphere loading and isostatic effects is controlling the surface topography.

Computational models of delamination (Bird, 1979; Le Pourhiet et al., 2006; G¨o˘g¨u¸sand Pysklywec,

2008a,b) showed a similar behaviour, but the isostatic processes seem to be less signiﬁcant in these analogue experiments since the lithospheric deformation is broadly distributed (essentially as a fairly simple viscous material) rather than localised (brittle, non-Newtonian rheology in the numerical results).

By t = 33 mins, the broad uplift is well developed. The subsidence at the delamination hinge

(x = 26.0 cm) reaches -0.2 cm but most of the section has elevated to approximately zero. Again, the uplift in topography is a result of broad lithospheric thickening and diminished downward forcing topography at regions distant from the migrating delamination. An interesting feature of the model evolution is that the plate boundarydenoted by the high magnitude negative spike in topographymoves to the left as the delamination hinge moves to the right (Fig 4.4B). This indicates clearly that the plate convergence and delamination are contemporaneously active processes in the experiment

The surface strain along a section Y-Y0 has been plotted in Figure 4.4C. This one-dimensional strain was obtained by measuring the ﬁnite displacement ﬁeld of the each single grid along Y-Y0 cross section direction. At t = 16 mins, the strain is characterized largely by negative strain values which corresponds to shortening. An especially high magnitude surface contraction at grid x = 17 is due to its proximity to the converging piston where deformation seems to be localizing. Later stages of surface strain show a similar pattern of general contraction. The magnitude is increasing as convergence proceeds in the model. However, the deformation does not seem to localize anywhere other than closest to the piston. The lack of localized deformation is also apparent in the top-view photos of the experiment (Fig. 4.4A) where the surface grids remain fairly regular.

86 A notable result is that there does not seem to be evidence of extension across the delamination gap. Previous numerical experiments on delamination have shown that significant syn-convergent extension can develop within a similar delamination gap as a response to divergent sublithospheric upwards return flow (Gö˘gü¸sand Pysklywec, 2008a,b). These experiments showed that the synconvergent extension could be lessened by broader-scale contraction depending on the plate convergence rate. It may be that this analogue experiment is in such a convergence-dominated regime and the viscous rheologies we used here do not promote localized deformation.

87 Figure 4.4: A) Sideview (left) and topview (right) pictures of the experiment DEL-12 (Vp = 0.25 cm/min). B) Surface topography plots along X-X0 cross section for 16 mins, 22 mins, and 33 mins. C) Crustal strain plots along Y-Y0 for 16 mins, 22 mins, and 33 mins.

88 4.4.2 Experiment DEL-13 (Vp = 0.38 cm/min)

For this experiment, the imposed convergence velocity is increased to Vp = 0.38 cm/min. In this experiment, we also wanted to test if the results in the later stages of delamination would signiﬁcantly change when the plate convergence is reduced. Accordingly, the plate convergence is stopped (Vp= 0) at t= 16 min. Aside from this, all other parameters are same with those of the reference experiment (DEL-12).

As in DEL-12, DEL-13 starts with subduction of the pro-plate until the slab reaches the bottom of the box at ∼ t = 6 mins. The mantle lithosphere begins to peel away from the crust at this time when the slab reaches the base of the model (Fig. 4.5A); again indicating that the change in slab dynamics facilitates delamination. We note that in computational experiments of delamination, this is not the case as “free-hanging” lithosphere was able to delaminate (G¨o˘g¨u¸sand Pysklywec,

2008b).

At t = 6 mins, the retro-plate and pro-plate crustal plates have not yet collided–i.e., some intervening “ocean” plate remains, while delamination just begins. By t = 13 mins, the ocean closure is achieved and delamination has progressed so that ∼ 2.5 cm of the mantle lithosphere has separated from the crust. By this stage the crustal plates have almost collided. At t = 24 mins delamination continues and exposes more upper crust to sublithospheric mantle — along an approximately 7.5 cm wide gap at its widest point. We note that in this model, the delamination is not constant along-strike of the plate boundary; rather, it is developing more quickly at the top of the top-view frame (Fig. 4.5A). As before, this would most likely be attributed to small variations in the physical model set-up (e.g., thickness of the lithospheric layers, change in the orientation of the plate).

Surface topography plots along the X-X0 cross section are shown in Fig 4.5B. Again, the zero magnitude of surface topography plot is referenced to the left edge of the retro-plate. For t = 6 mins when delamination is just beginning, the surface topography is similar to experiment DEL-12.

Across the retro-plate (x = 0 - 15.0 cm) the topography ranges from 0 to -0.1 cm. We attribute this subsidence to dynamic topography associated with the subducting slab. In the middle of the model, the high magnitude negative spike corresponds with the plate boundary where there remains a trough at the intervening ocean before the collision. On the pro-plate side there is a deep trough

89 to -0.4 cm that corresponds to the location of the delamination hinge. The rest of the plate is subsided more modestly.

By t = 13 and 24 mins, the average surface elevation at the retroplate is quite similar to the earlier time except that the topographic trough at the delamination hinge has migrated to the right

(to x = 22.0 and 25.0 cm). Compared to DEL-12, there is no gradual uplift of the topography as the model progresses. This is owing to slower development of the delamination — which retains the dynamic topographic support of the subducting slab closer to the retro-plate — and the removal of plate convergence at t =16 mins — which takes away continued crustal thickening and isostatic uplift.

Figure 4.5C plots the amount of strain along the Y-Y0 cross section versus x grid location of the crust. The data indicate that lithospheric shortening is not very well developed (compared to the previous model where there was a lower convergence velocity) because the convergence is turned oﬀ at t = 16 mins. At t = 6 and t = 13 mins, there is variable surface crustal strain and by t = 24 mins, primarily contractional accumulated deformation has developed.

90 Figure 4.5: A) Sideview (left) and topview (right) pictures of the experiment DEL-13 (Vp = 0.38 cm/min). B) Surface topography plots along X-X0 cross section for 6 mins, 13 mins, and 24 mins. C) Crustal strain plots along Y-Y0 for 6 mins, 13 mins, and 24 mins.

91 4.4.3 Experiment DEL-11 (Vp = 0.5 cm/min)

For DEL-11, the imposed pro-plate velocity is further increased to Vp = 0.5 cm/min (Fig. 4.6A). The experiment evolves through three main stages. The ﬁrst stage (not shown) is ocean subduction. Initially, as in the previous experiments, the pro-mantle lithosphere subducts into the sublithospheric mantle at a high dip angle (70◦ - 80◦). At t = 5 mins, the subducted lithosphere hits the bottom of the box while it is also subducting beneath the retro-plate (Fig. 4.6A).

The collision between the pro-plate crust and the retro-plate occurs at t = 18 mins. Following collision and contact of the subducted slab with the bottom of the box, there is no mantle lithosphere delamination as in DEL-12 and DEL-13. In DEL-11, the pro-plate mantle lithosphere attaches to the retro-plate and the plates move retro-ward together while subduction continues. The pro-plate separates from the crust that is accreting onto the retro-crust (t = 24 mins). As the pro-plate slab subducts, it folds over the already subducted slab where a ∼ 5 cm of the subducted slab lays along the bottom of the box. This progresses to t = 27 mins and the crust is decoupled from this subducting mantle lithosphere and continues to accrete onto the retro-plate crust (top-view photos,

Fig. 4.6A).

Owing to the higher plate convergence in DEL-11 compared to the previous experiments, the delamination of the mantle lithosphere from crust does not occur fast enough to develop. Instead, a separation of crust and mantle lithosphere occurs at the model plate collision zone as the pro-plate is driven into the retro-plate. Several studies refer to a similar style of accretion of the pro-plate crust on to the retro-plate during continental collision, terming it “ﬂake tectonics” (Oxburgh, 1972).

Cloos et al. (2005) suggest that the whole crust or some part of it may separate from the mantle lithosphere and become accreted on the pro-plate during the New Guinea continental collision.

(Ellis et al., 1999) suggest a process of the episodic accretion of the pro-plate crust in a continental convergent regime.

Figure 4.6B shows the evolution of surface topography along the X-X0 cross section. The zero topography line in the experiments is a reference level corresponding to the left edge of the retro- plate. Apart from the high frequency signal, at t = 12 mins the surface topography is characterized by a trough of magnitude ∼ −0.2 cm in the middle and a mostly negative topography across the rest of the section. The trough reﬂects both subsidence from the weight/descent of the downgoing

92 slab and the initial low elevation gap between converging retro- and pro-plates.

By t = 24 mins, the topography shows a significant topography high (∼ 0.18 cm) adjacent to the high amplitude trough. The high elevation is associated with accretion of the buoyant crust on top of the retro-plate. The crustal overthrusting and thickening results in an isostatic uplift of the surface; this type of enhanced topographic uplift with crustal accretion is suggested for continental collision zones such as for the Alps (e.g. Ellis et al., 1999). Similarly, Schmid et al. (1996) propose a process for the Italian Alps where the European plate subducts while the Adriatic plate acts a chisel scraping off the upper from the lower crust. Across the rest of the surface, the topography signal is quite noisy, but generally negative. The broad subsidence is the surface response to broad downward flow of sub-lithospheric mantle caused by the subducting slab beneath these model plate regions.

At t = 27 mins, there is a similar signal of topography — i.e., dominated by the accretion- related uplift and subduction trough. However, the paired topography anomalies have moved to the left as the pro-plate and the retro-plate migrate together.

The surface strain along a section Y-Y0 is plotted in Figure 4.6C. The strain shows a similar pattern at the three time intervals with increasing amplitude from the early to late stages as convergence progresses. Towards the collision zone, high amplitude negative strain(∼ −0.28) develops indicating shortening, albeit uneven.

93 Figure 4.6: A) Sideview (left) and topview (right) pictures of the experiment DEL-11 (Vp = 0.50 cm/min). B) Surface topography plots along X-X0 cross section for 12 mins, 24 mins, and 27 mins. C) Crustal strain plots along Y-Y0 for 12 mins, 24 mins, and 27 mins.

94 4.4.4 Experiment DEL-10 (Vp = 1.0 cm/min)

Figure 4.7A shows the evolution of experiment DEL-10 where the imposed pro-plate velocity is increased to Vp = 1.0 cm/min. Analogous to the previous experiment (DEL-11), the advancing slab initially subducts steeply into the mantle ﬂuid and it hits the bottom of the box at t = 4 mins.

At t= 6 mins, collision between the pro- and retro-crust has not yet occurred. It seems that a very small part of the pro-plate mantle lithosphere has decoupled from the crust. This is evident in the top view photo where transparent sub-lithospheric mantle may be intruding between pro- crust and mantle lithosphere. However, it may just be at the margins of the crust (in the along- strike direction to the plate boundary) that there is less detachment. Collision between the plates occurs between t =10 - 20 mins (Fig. 4.7A) and results in the attachment of the two plates and subsequently the migration of the joined plate in the retro-ward direction.

At t = 20 mins, there is signiﬁcant total shortening between the two plates (Fig. 4.7A). The left edge of the retro-plate has already reached the end of the experimental box and the converging plates are constrained between the piston and the box. The retro-plate has begun to extrude towards the top and bottom part of the frame (i.e., in the along-strike direction) to partially accommodate the plate shortening. By this stage a portion of the pro-crust — at the centre in the along-strike direction — has accreted on to the retro-plate. As with DEL-11, there is a separation between pro-crust and mantle lithosphere, and with the high rate of plate convergence the layers are chiseled onto the retro-plate rather than having the mantle lithosphere delaminate and fall back.

Laser data shows the evolution of surface topography along the X-X0 section (Fig. 4.7B). At t = 6 mins, there is a 1.5 cm deep negative surface deﬂection that reﬂects the sinking ocean lithosphere slab initially pushed into sublithospheric mantle. We note that this trough is not geologically important here, but rather a remainder of the initial conditions of the model. By t

= 10 mins, the surface topography looks similar, except the gap/trough between retro-plate and pro-plate is now closing when two plates are converging rapidly. By t = 20 mins, the variation in the surface topography changes signiﬁcantly. There is a topographic high (∼ 0.19 cm) due to the accretion/overthrusting of the buoyant pro-plate crust on top of the retro-plate, with negative deﬂection across the pro-plate as the subducting slab pulls down the surface.

95 The surface strain along section Y-Y0 is plotted in Figure 4.7C. There are fewer gridlines drawn on the surface of the model (as an earlier model in the developing series of experiments) resulting in lower resolution strain measurements. The strain curves for both t = 6 mins and t = 10 mins look similar where crustal grids are not highly deformed before the collision occurs. However, by t = 20 mins shortening is very apparent, with %40 strain across the pro-plate crust having higher deformation near the plate boundary rather than the central section of the model as was observed in the delamination models (DEL-12 and DEL-13).

96 Figure 4.7: A) Sideview (left) and topview (right) pictures of the experiment DEL-10 (Vp = 1.0 cm/min). B) Surface topography plots along X-X0 cross section for 6 mins, 10 mins and 20, mins. C) Crustal strain plots along Y-Y0 for 6 mins, 10 mins, and 20 mins.

97 4.4.5 Delamination and collision: Retreating Plates Beneath the Continental Crust.

The experiments show that depending on the convergence rate, the mantle lithosphere plate beneath the continental crust may delaminate back under the pro-plate or underthrust beneath the retro- plate. In the ﬁrst two sets of experiments DEL-12 and DEL-13, we track the position of the delamination hinge (dh) with respect to the retro-plate (Fig. 4.8A,B). Essentially, this plots the width of the delamination gap “x” that develops in the models after the collision occurs.

For DEL-12, x = 0 at t = 0 - 8 mins as ocean plate subduction proceeds and before the slab reaches the bottom of the box. When delamination begins at t = 8 mins, the hinge moves rapidly for the ﬁrst minute but slows to an almost constant migration rate (from t = 9 - 16 mins). This is followed by an increase in the delamination rate from t = 18 - 20 min before it slows again.

This demonstrates an interesting episodic behaviour to the delamination process as the mantle lithosphere “retreats” beneath the continental crust. Such episodic behaviour of the retreating slab in subduction zones has been suggested in Funiciello et al. (2003); Bellahsen et al. (2005) and

Schellart (2005), and it also explains the episodic nature of some back-arc basins in the central

Mediterranean (i.e., Liguro-Provencal basin and Tyrrhenian sea) (Faccenna et al., 2001).

The migration of the hinge point and opening of the mantle lithosphere gap was also plotted for experiment DEL-13 (Fig. 4.8B). Since there is a variation in the form of delamination along-strike in this model (Fig. 4.4A), the position of the delamination hinge is deﬁned as the middle location

(i.e., along X-X0) between the two delaminated edges of the crust. Delamination proceeds from t

= 6 - 9 mins but stalls at t= 9 - 10 mins.

Between t = 10 - 16 mins, there is an approximately constant growth of the delamination gap.

The migration rate increases after t = 16 mins, corresponding to the time when imposed plate convergence is turned oﬀ in the experiment. This shows again that delamination is sensitive to any simultaneous plate convergence: here, delamination/retreat is facilitated by the reduction in convergence velocity.

These dynamics are analogous to retreat and advance in a subduction system, except they occur beneath a buoyant (possibly orogenic) continental crust. Previous investigations by Funiciello et al. (2003) and Schellart (2004, 2005) focus on the behaviour of a retreating slab and migration of a

98 subduction hinge in a roll-back system with changing plate convergence velocities, plate thickness and densities. Similar experimental results are suggested by Zhong et al. (1995)— intervals of slow plate velocities are associated with rapid subduction migration —, whereas fast plate velocities results in more subdued horizontal motions of the slab

99 Figure 4.8: A) Sideview illustration of experiment DEL-12 showing the migration of the delaminating hinge. B) x (distance) as a function of time. DEL-12 (solid line), DEL-13 (dashed line).

100 4.5 Conclusions

In this study, we investigate the evolution of continental delamination and its surface response in analogue/laboratory models of continental collison with pre-existing ocean lithosphere subduction.

We focus on four delamination experiments with diﬀerent plate convergence velocities.

The experiments demonstrate that continental mantle lithosphere delamination may develop following the transition from retreating oceanic subduction to collision. In the presence of on- going plate convergence delamination is facilitated by a low velocity of plate shortening. In the experiments where delamination occurs, the delamination hinge migrates relatively rapidly in the

ﬁrst 2-3 mins after the slab reaches the bottom of the box.

The lithosphere delaminates in the opposite direction of plate convergence (similar to a roll-back system) and does this episodically, with significant changes in the rate of the hinge migration. We suggest that delamination is an analogous process to subduction retreat, the difference being that the former process involves a separation between the sinking/retreating mantle lithosphere and buoyant (possibly orogenic) continental crust. The behaviour of the experiments would be different if material was not constrained by the bottom of the box; however, we have limited the models to considering upper mantle scale flow.

The surface topography of the experiments where delamination occurs shows a distinct surface subsidence/trough due to the downwards forcing of the delaminating slab. This trough migrates with the delamination hinge as the mantle lithosphere peels back under the crust. For DEL-12 there is a gradual broad-scale isostatic uplift of the surface owing to plate convergence/shortening.

In DEL-13, there is much less isostatic uplift since there is not enough shortening/contraction achieved by t = 16 mins when plate convergence is terminated in the experiment. The surface deformation is characterised by broad contraction and because of this significant shortening there was no evidence of synconvergent extension across the delamination gap as has been noted in other geodynamical modeling studies (Gö˘gü¸sand Pysklywec, 2008b).

With higher plate convergence velocities, mantle lithosphere consumption occurs too rapidly for delamination to occur. In the analogue experiments, the convergent mantle lithosphere folds overtop of previously subducted slab material deeper in the mantle. The pro-plate mantle lithosphere does not advance beneath the retro-plate in the strict sense of an advancing plate, rather the S-point

101 between the plates remains stable and the joined plates migrate in the retro-ward direction with on-going convergence. The pro-plate crust, however, does advance onto the retro-plate. Separation of the buoyant continental crust and mantle lithosphere occurs at the plate collision zone and the on-coming pro-plate crust is accreted/ovethrusted on top of the retro-plate.

The surface in the experiments shows broad subsidence associated with the heavy steeply descending slab. This subsidence is gradually recovered through the model evolution as plate shortening causes isostatic uplift. The broad signal is punctuated by a paired peak and trough at the plate collision zone. These are related to the accreted/thickened crust and initial plate contact zone, respectively. Surface strain in the experiments is predominantly contractional and it is mostly accommodated near the plate boundary.

4.6 Discussions

4.6.1 Comparison with previous models and natural systems

Previous models and speculations on delamination have invoked some type of lithospheric density anomaly to initiate mantle lithosphere delamination (Kay and Kay, 1993; Jull and Kelemen, 2001;

Anderson, 2005; Elkins-Tanton, 2005). Our results demonstrate that pre-existing subduction (in the absence of any other density perturbation) may provide a suﬃcient trigger for delamination of the mantle lithosphere. In this way, delamination may be a natural consequence of a transition from retreating subduction to collision where the weight of a subducted ocean continues to peel away the mantle lithosphere from the continental crust following collision. This may explain the prevalence of delamination in plate convergent zones. For example, in the Mediterranean the

Appenines, Carpathians, and Hellenides are/were retreating subduction systems (Royden, 1993) that may develop/have developed to delamination during the continental collision. Recently, for the eastern Anatolian region, Zor (2008) suggests that the northern branch of the Tethyan ocean lithosphere was retreating with a velocity of 2 cm/year prior to the continental collision. The evolution of the retreating ocean mantle lithosphere in this region possibly may have evolved into delamination style removal of the mantle lithosphere from the crust at the syn-post collision stage.

Thermomechanical numerical experiments show that high plateau-type topography is caused by isostatic uplift associated with replacement of hot and buoyant sublithospheric mantle following

102 delamination (Gö˘gü¸sand Pysklywec, 2008a,b; Le Pourhiet et al., 2006). It was speculated that uplift of the East Anatolian plateau, for example, may be a consequence of underlying delamination of the mantle lithosphere since the Miocene (Gö˘gü¸sand Pysklywec, 2008a). However, the analogue experiments presented here do not have thermal-bouyancy effects so high, wide surface topography does not develop above the delamination gap. Rather, uplift in the analogue models is related to isostatic uplift from lithospheric thickening imposed by the plate convergence; subsidence at the delamination hinge develops as a result of the “heavy hanging” lithosphere. In nature, we would expect these topographic anomalies to be superimposed on a broad-scale plateau uplift that is not treated in the analogue experiments.

The numerical results of Gö˘gü¸sand Pysklywec (2008a) suggest that there is an appreciable crustal extension due to the sublithospheric mantle flow even in a convergent regime. Here, the analogue modeling experiments do not show such syn-convergent type extension. A reason for this may be the differing rheologies between the numerical and analogue experiments as well as the significant plate shortening factor in the analogue experiments. In the numerical experiments, the crust and mantle materials are non-Newtonian and temperature-dependent, so they tend to localise deformation more readily than the Newtonian and temperature-independent analogue materials.

Furthermore, in the numerical experiments, the crust above the delamination gap may undergo thermal weakening as hot sub-lithospheric mantle rises, making the crust more prone to stretching deformation. In the analogue experiments, the crust maintains its strength perhaps allowing it to resist any extensional forcing induced by the sub-lithospheric mantle ﬂow.

An aspect of this work was to consider the first order conditions for mantle lithosphere delamination. We found that similar to previous results a reduced viscosity of the lower crust maybe promote faster delamination (Gö˘gü¸sand Pysklywec, 2008a,b). As well, the increase in the density of the upper crust may promote continental subduction even though the lower crust can still be weak enough to decouple mantle lithosphere from crust.

Our results show that when the plate convergence is high, the mantle lithosphere does not delaminate in the traditional sense following oceanic subduction and continental collision. Instead, the crust and mantle lithosphere separate at the collision zone with the crustal overthrusting/accreting into the orogenic suture zone and the mantle lithosphere descending into the mantle. Similar tectonic modeling results of chisel/ﬂake tectonics are suggested for alpine collision tectonics. Ellis et

103 al. (1999) and Pfiffner et al. (2000) postulate that during the final stage of Alpine collision, the

Adriatic upper crust was peeled off/scraped off from the lower crust and it forms a folded nappe stack. Such overthrusting of the upper crust to the retro-plate is cited as delamination in De Franco et al. (2008). Clearly, though, this type of tectonic behaviour does not involve any type of exposure of the crust to the sublithospheric mantle (and the related thermal-topographic perturbation) that is associated with the strict definition of delamination.

The models do not include a lower mantle layer and we assume that upper-lower mantle boundary as a rigid boundary. An intriguing question in the context of the models is what would be the inﬂuence of a larger scale mantle circulation on the delamination/subduction processes? The high convergence velocity (non-delamination) models have the subducting slab folding over itself owing to the reference frame of the model with a “stationary” mantle beneath moving plates. P- wave tomography in the Mediterranean shows similar structures of subducting slabs folding over themselves (Piromallo and Morelli , 2003). However, we recognize that the morphology of such features in the model is controlled by the assumed reference frame within an active mantle circulation

(Boutelier and Cruden, 2008).

104 Bibliography

Anderson, D.L., 2005, Large Igneous Provinces, Delamination, and Fertile Mantle Elements, v. 1,

p. 271-275, doi: 10.2113/gselements.1.5.271.

Anderson, D.L., 2007, The eclogite engine: Chemical geodynamics as a Galileo thermometer, in

Foulger, G.R., and Jurdy, D.M., eds., Plates, plumes, and planetary processes: Boulder, Col-

orado, Geological Society of America Special Paper, v. 430, p. 47-64, doi: 10.1130/2007.2430(03).

Beck, S. L., and Zandt, G., 2002, The nature of orogenic crust in the central Andes: Journal of

Geophysical Research, v. 107, B10, 2230, doi:10.1029/2000JB000124.

Bellahsen, N., Faccenna, C., Funiciello, F., 2005, Dynamics of subduction and plate motion in

laboratory experiments: Insights into the plate tectonics behavior of the Earth: Journal of

Geophysical Research, v. 110, B01401, doi: 10.01029/02044JB002999.

Bird, P., 1978, Initiation of intracontinental subduction in the Himalaya: Journal of Geophysical

Research, v. 83, p. 4975-4987.

Bird, P., 1979, Continental delamination and the Colorado Plateau: Journal of Geophysical Re-

search, v. 84, p. 7561-7571.

Boutelier, D., Chemenda, A., and Jorand, C., 2004, Continental subduction and exhumation of

high-pressure rocks: Insights from thermo-mechanical laboratory modeling: Earth and Planetary

Science Letters, v. 222, p. 209-216.

Boutelier, D., and Cruden, A.R., 2008, Impact of regional mantle ﬂow on subducting plate geometry

and interplate stress: Insights from physical modelling, Geophysical Journal International, v. 174,

p. 719-732, doi: 10.1111/j.1365-246X.2008.03826.x

105 Channell, J.E.T., and Mareschal, J.C., 1989, Delamination and asymmetric lithospheric thicken-

ing in the development of the Tyrrhenian rift, in Coward, M.P., et al., eds., Alpine tectonics:

Geological Society London Special Publication v. 45, p. 285=302.

Chemenda, A., Burg, J.-P., and Mattauer, M., 2000, Evolutionary model of the Himalaya-Tibet

system: Geopoem based on new modelling, geological and geophysical data: Earth and Planetary

Science Letters, v. 174, p. 397-409.

Christensen, U.R., and Yuen, D.A., 1984, The interaction of a subducting lithospheric slab with a

chemical or phase boundary: Journal of Geophysical Research, v. 89, p. 4389-4402.

Christensen, U.R., and Yuen, D.A., 1985, Layered convection induced by phase transitions: Journal

of Geophysical Research, v. 90, p. 10,291-10,300.

Cloos, M., Sappie, B., Quarles van Uﬀord, A., Weiland, R.J., Warren, P. Q., and Mc Mahon, T.

P., 2005, Collisional delamination in New-Guinea: The geotectonics of subducting slab breakoﬀ:

Geological Society of America Special Paper, v. 400, 48 pages.

Cruden, A.R., Nasseri, M. B. and Pysklywec, R. N., 2006, Surface topography and internal strain

variation in wide hot orogens from three-dimensional analogue and two-dimensional numerical

vise models, in Analogue and Numerical Modelliing of Crustal-Scale Processes, In: S. J. H.

Buiter and G. Schreurs (Eds.), Analogue and Numerical Modelling of Crustal - Scale Processes.

Geological Society, London, Special Publications, v. 253, p. 79-104.

De Franco, R., Govers, R., and Wortel, M. J. R., 2008, Dynamics of continental collision: Inﬂuence

of the plate contact: Geophysical Journal International, v. 174, p. 1101-1120.

Dewey, J. F., Ryan, P.D., and Andersen, T.B., 1993, Orogenic uplift and collapse, crustal thickness,

fabrics and metamorphic phase changes: The role of eclogites. In: H.M. Prichard, T. Alabaster,

N.B.W. Harris and C.R. Neary, Editors, Magmatic Processes and Plate Tectonics: Geological

Society Special Publications v. 76, pp. 325-343.

Ducea, M. N., and Saleeby, J. B., 1998, A case for delamination of the deep batholithic crust

beneath the Sierra Nevada, California: International Geology Review, v. 40, p. 78-93

106 Elkins-Tanton, L.T., 2005, Continental magmatism caused by lithospheric delamination, in Foulger,

G.R., Natland, J.H., Presnall, D.C., and Anderson, D.L., eds., Plates, plumes, and paradigms:

Geological Society of America Special Paper, v. 388, p. 449-461, doi: 10.1130/2005.2388 (27).

Ellis, S., Beaumont, C. and Pﬁﬀner, A., 1999, Geodynamic models of crustal-scale episodic tec-

tonic accretion and underplating in subduction zones: Journal of Geophysical Research, v. 104,

p.15169-15190.

Espurt, N., Funiciello, F., Martinod, J., Guillaume, B., Regard, V., Faccenna, C., and Brusset, S.,

2008, Flat subduction dynamics and deformation of the South American plate: Insights from

analog modeling: Tectonics, v. 27, TC3011, doi:10.1029/2007TC002175.

Faccenna, C., Becker T.W., Lucente F.P., Jolivet L., Rossetti F., 2001, History of subduction and

back-arc extension in the central Mediterranean: Geophysical Journal International, v. 145, p.

809- 820, doi: 10.1046/j.0956-540x.2001.01435.

Faccenna, C., Heuret, A., Funiciello, F., Lallemand S., and Becker, T.W., 2007, Predicting trench

and plate motion from the dynamics of a strong slab: Earth and Planetary Science Letters, v.

257, p. 29-36, doi: 10.1-16/j.epsl.2007.02.016.

Fadil, A., Vernant, P., McClusky, S., Reilinger, R., Gomez, F., Sari, D. B., Mourabit, T., Feigl,

K.,and Barazangi, M., 2006, Active tectonics of the western Mediterranean: Geodetic evidence for

roll back of a delaminated subcontinental lithospheric slab beneath the Rif Mountains, Morocco:

Geology, v. 34(7), p. 529- 532.

Funiciello, F., Faccenna, C., Giardini, D., and Regenauer-Lieb, K., 2003., Dynamics of retreat-

ing slabs: 2. Insights from three-dimensional laboratory experiments: Journal of Geophysical

Research, v. 108 (B4), 2207.DOI: 10.1029/2001JB000896.

Funiciello, F., Moroni, M. , Piromallo, C., Faccenna, C., Cenedese, A., and Bui, H.A., 2006,

Mapping mantle ﬂow during retreating subduction: Laboratory models analyzed by feature

tracking: Journal of Geophysical Research, v. 111, B03402, doi: 10.1029/2005JB003792.

Funiciello, F., Faccenna, C., Heuret A., Di Giuseppek, E., Lallemand S., Becker T.W., 2008, Trench

107 migration, net rotation and slab-mantle coupling: Earth and Planetary Science Letters, v. 271,

p. 233-240, doi: 10.1016/j.epsl.2008.04.006.

Gok, R., Pasyanos, M., and Zor, E., 2007, Lithospheric structure of the continent-continent col-

lision zone: Eastern Turkey: Geophysical Journal International, v. 169, p. 1079-1088, doi:

10.1111/j.1365-246X.2006.03288.x.

G¨o˘g¨u¸s,O.H., and Pysklywec, R.N., 2008a, Mantle lithosphere delamination driving plateau up-

lift and synconvergent extension in eastern Anatolia: Geology, v. 36, no. 9, p.723-726, doi:

10.1130/G2498A.1.

G¨o˘g¨u¸s,O.H., and Pysklywec, R.N., 2008b, Near-surface diagnostics of dripping or delaminating

lithosphere: Journal of Geophysical Research, v. 113, B11404, doi: 10.1029/2007JB005123.

Guillaume, B., Martinod J., Espurt N., 2009, Variations of slab dip and overriding plate tecton-

ics during subduction: Insights from analogue modeling : Tectonophysics, v. 463, p. 167-174,

doi:10.1016/j.tecto.2008.09.043.

Hager, B.H., and. OConnell, R.J., 1978, Subduction zone dips and ﬂow driven by the plates:

Tectonophysics, v. 50, p.111-134.

Houseman, G.A., and Gemmer, L., 2007, Intra-orogenic extension driven by gravitational instabil-

ity: Carpathian-Pannonian orogeny: Geology, v. 35, p. 1135-1138. doi: 10.1130/G23993A.1.

Jull, M., and Kelemen, P.B., 2001, On the conditions for lower crustal convective instability: Journal

of Geophysical Research, v. 106, p. 6423-6446, doi: 10.1029/2000JB900357.

Kay, R.W., and Kay, M., 1993, Delamination and delamination magmatism: Tectonophysics, v.

219, p. 177-189.

Keskin, M., 2003, Magma generation by slab steepening and break oﬀ beneath a subduction accre-

tion complex: An alternative model for collision-related volcanism in Eastern Anatolia, Turkey:

Geophysical Research Letters, v.30, no. 24, 8046, doi: 10.1029/2003GL018019.

Keskin, M., 2007, Eastern Anatolia: A hotspot in a collision zone without a mantle plume: in

Foulger, G.R., and Jurdy, D.M., eds., Plates, plumes and planetary processes: Geological Society

108 of America Special Paper., 430 p.693-722, doi. 10.1130/2007.2430(32).

Kosarev, G., Kind, R., Sobolev, S. V., Yuan, X., Hanka, W., and Oreshin, S., 1999, Seismic evidence

for detached Indian lithospheric mantle beneath Tibet: Science, v. 283, p. 1306-1309.

Le Pourhiet, L.L., Gurnis, M., and Saleeby, J., 2006, Mantle instability beneath the Sierra Nevada

Mountains in California and Death Valley extension: Earth and Planetary Science Letters, v.

251, p. 104-119. doi: 10.1016/j.epsl.2006.08.028.

Lorinczi, P., and Houseman, G.A., 2009, Lithospheric gravitational instability beneath the South-

east Carpathians: Tectonophysics, v. 474, p. 322-336.. doi:10.1016/j.tecto.2008.05.024

Massone, H. J., 2005, Involvement of Crustal Material in Delamination of the Lithosphere after

Continent-Continent Collision: International Geology Review, v. 47, p. 792-804.

Meissner, R., and Mooney, W.D., 1998, Weakness of the lower continental crust: a condition for

delamination, uplift, and escape: Tectonophysics, v. 296 p. 47-60.

Mitrovica, J.X., and Forte, A.M., 1997, The Radial Proﬁle of Mantle Viscosity: Results From the

Joint Inversion of Convection and Post-Glacial Rebound Observables: Journal of Geophysical

Research, v. 102, p. 2751-2769.

Morency, C., and Doin, M.P., 2004, Numerical simulations of mantle lithospheric delamination:

Journal of Geophysical Research, v. 109, B03410, doi: 10.1029/2003JB002414.

OReilly, S.Y., Griﬃn, W.L., Poudjom Djomani Y., Morgan P., 2001, Are lithospheres forever?

Tracking changes in subcontinental lithospheric mantle through time. GSA Today, v.11, p. 4-9.

Oxburgh, E.R., 1972, Flake tectonics and Continental collision: Nature, v. 239, p. 202-204.

Pﬁﬀner, O.A., Ellis, S., and Beaumont, C., 2000, Collision tectonics in the Swiss Alps: insight from

geodynamic modeling: Tectonics v. 19 p. 1065-1094.

Poudjom Djomani, Y.H., O’Reilly, S.Y., Griﬃn, W.L., and Morgan, P., 2001, The density structure

of subcontinental lithosphere through time: Earth and Planetary Science Letters, v. 184 p. 605-

621.

109 Piromallo, C., and Morelli, A., 2003, P wave tomography of the mantle under the Alpine- Mediter-

ranean area: Journal of Geophysical Research, v. 108 (B2), 2065 10.129/2002JB001757.

Pysklywec, R., and J.X. Mitrovica, 1999, The role of Subduction-Induced Subsidence in the Evo-

lution of the Karoo Basin: Journal of Geology, v. 107, p. 155-164.

Pysklywec, R. N., and Cruden, A. R., 2004. Coupled crust-mantle dynamics and intraplate tecton-

ics: Two-dimensional numerical and three-dimensional analogue modeling, G3: Geochemistry,

Geophysics, Geosystems, 5(10), doi: 10.1029/2004GC000748, 22 pp

Ranalli, G., 2000, Rheology of the crust and its role in tectonic reactivation: Journal of Geody-

namics, v. 30, p. 3-15.

Reutter, K.G., Giese, P. and Closs, H., 1980, Lithospheric split in the descending plate: observations

from the Northern Apennines: Tectonophysics, v. 64, pp. 11-19.

Regard, V., Faccenna, C., Martinod, J., Bellier, O., and Thomas, J.-C., 2003, From Subduction to

Collision: Control of Deep Processes on the Evolution of Convergent Plate Boundary: Journal

of Geophysical Research, v. 108, 2208, doi:10.1029/2002JB001943.

Ricard, Y., Doglioni, C. and Sabadini, R., 1991. Diﬀerential rotation between lithosphere and

mantle: a consequence of lateral mantle viscosity variations: Journal of Geophysical Research,

v. 96, p. 8407-8415.

Royden, L.H., 1993, Evolution of retreating subduction boundaries formed during continental col-

lision: Tectonics, v. 12, p. 629-638.

Schellart, W.P., 2004, Kinematics of subduction and subduction-induced ﬂow in the upper mantle:

Journal of Geophysical Research, v. 109, B07401, doi: 10.1029/2004JB002970.

Schellart, W.P., 2005, Inﬂuence of the subducting plate velocity on the geometry of the slab and

migration of the subduction hinge: Earth and Planetary Science Letters,v. 231, p. 197-219, doi:

10.1016/j.epsl.2004.12.019.

Schmid, S.M., Pﬁﬀner, O.A., Froitzheim, N., Schnborn, G., and Kissling, E., 1996, Geophysical-

geological transect and tectonic evolution of the Swiss-Italian Alps: Tectonics, v. 15, p. 1036-1064.

110 Schott, B. and H. Schmeling, 1998: Delamination and detachment of a lithospheric root. Tectono-

physics, v. 296, p. 225-247.

Seber, D., Barazangi, M., Ibenbrahim, A., and Demnati, A., 1996, Geophysical evidence for litho-

spheric delamination beneath the Alboran Sea and Rif-Betic mountains: Nature, v. 379, p.

785-790, doi: 10.1038/379785a0.

S¸eng¨or,A.M.C., Ozeren, S., Zor, E., and Genc, T., 2003, East Anatolian high plateau as a mantle-

supported, N-S shortened domal structure: Geophysical Research Letters, v. 30, p. 8045, doi:

10.1029/2003GL017858.

S¸eng¨or, A.M.C., Ozeren, M.S., Keskin, M., Sakinc, M., Ozbakir, A., and Kayan, I., 2008,

Eastern Turkish high plateau as a small Turkic-type orogen: Implications for post-collisional

crust-forming processes in Turkic-type orogens: Earth-Science Reviews, v. 90, p.1-48,

doi:10.1016/j.earscirev.2008.05.002.

Toussaint, G., Burov, E., and Jolivet, L. 2004, Continental plate collision: Unstable vs.stable slab

dynamics: Geology, v. 32, p. 33-36.

Turcotte, D.L., Liu, J.Y., and Kulhawy, F.H., 1984, The role of an intracrustal asthenosphere on

the behaviour of major strike-slip faults: Journal of Geophysical Research, v. 89, p. 5801-5816.

Weijermars, R., and Schmeling, H., 1986, Scaling of Newtonian and non-Newtonian ﬂuid dynamics

without inertia for quantitative modeling of rock ﬂow due to gravity (including the concept of

rheological similarity): Physics of the Earth and Planetary Interiors, v. 43, p. 316-330.

Willett, S., Beaumont, C., and Fullsack, P., 1993, Mechanical model for the tectonics of doubly

vergent compressional orogens: Geology, v. 21, p. 371-374, doi: 10.1130/0091-7613(1993)021.

Zhong, S., and Gurnis, M., 1995, Mantle convection with plates and mobile, faulted plate margins:

Science v. 267, p. 838-843.

Zor, E., 2008, Tomographic evidence of slab detachment beneath eastern Turkey and the Caucasus:

Geophysical Journal International, v. 175, p. 1273-1282.

111 Chapter 5

Synthesis of Principle Conclusions

112 The main conclusions of the thesis can be subdivided into three categories; each category representing the conclusions of a chapter.

5.1 Near Surface Diagnostics of Dripping or Delaminating Litho-

sphere

Various surface responses among the drip and delamination models provide diagnostic indicators of the style of mantle lithosphere removal:

1. Surface topography for the delamination model is markedly asymmetric with a paired subsidence and uplift response that migrates as the mantle lithosphere peels away or rolls-back from the crust. The uplift of the surface above the delamination gap owing to the inﬂux of hot mantle material is broad and it can be interpreted as a type of plateau feature. The surface subsidence at the delamination hinge is caused by the negative buoyancy of the delaminating slab. In contrast, the surface topography of both drip models is associated with a symmetric pattern that remains

ﬁxed above the mantle lithosphere downwelling. Uplift of the surface following drip detachment tends to be localized, rather than developing a plateau-type structure. The set-up of the models is such that the mantle drips do not migrate laterally with respect to the overlying crust. However, we do not preclude the possibility that the mantle lithosphere instability may move in nature (e.g., apparent southwestern motion of the mantle drip beneath Sierra Nevada of eastern California)

(Zandt, 2003; Zandt et al., 2004).

2. Moho temperature plots were also used to be make the comparison between delamination and drip models. The fast removal of the mantle lithosphere during the delamination process results in the sudden broad increase of the Moho temperature since hot-sublithospheric mantle ﬂow actively replaces the entire thickness of the cold mantle lithosphere. For dripping models, based on the quantity of the mantle lithosphere removal (i.e., the thickness of the mantle lithosphere that is removed), there is a diﬀerence in surface heating. When complete removal occurs (DRIP-2) there may be substantial increase in the Moho temperatures, whereas with partial removal the increase in the Moho temperatures is relatively subdued. In nature, it has been suggested that high heat

ﬂow due to delamination/dripping results in various types of volcanism (often instantaneous) and large igneous provinces (Anderson, 2005). For instance, volcanism in the Andes (Kay and Kay,

113 1993), Sierra Nevada of Eastern California (Ducea and Saleeby, 1998; Manley et al., 2000; Jones et al., 2004; Elkins, 2005), Colorado plateau, (Bird, 1979), Eastern Anatolian plateau (Keskin, 2003,

2007) have all been attributed to various styles of lithospheric removal.

3. The P-T-t paths from various horizons within the crust (lower, middle and upper crustal) were plotted to make a comparison of first-order metamorphic implications between the delamination and drip models. The P-T evolution of the crust above delamination is clockwise with a rapid increase temperature. The pressure increases rapidly for a short period, then experiences gradual decompression. The P-T-t paths associated with experiment DRIP-2 show patterns more akin to the delamination model (increase in pressure at the beginning and then decompression, contemparenous with significant increase in temperature). DRIP-1 does not show a temperature increase and can not be classified as having a distinguishing P-T path. The pressure increase in

DRIP-1 is dramatic and continuous since a signiﬁcant amount of crust is thickened and buried.

4. The crustal deformation pattern associated with delamination is asymmetric where there is crustal thickening/contraction above the delaminating hinge and crustal thinning/extension above the lithospheric gap. The horizontal crustal tectonics are caused by entrainment with the delamination sublithospheric mantle flow. The distribution of crustal deformation for drip models is symmetric with contraction/thickening above the dripping lithosphere and flanking extension on both sides of the drip owing to the lateral mantle flow.

5.2 Mantle Lithosphere Delamination Driving Plateau Uplift and

Synconvergent Extension in Eastern Anatolia

In this work, plate convergence is evaluated as a factor for inﬂuencing delaminating mantle lithosphere and the associated surface response. We suggest that mantle lithosphere delamination drives various crustal/tectonic processes during plate convergence that are consistent with the primary anomalous tectonic features of eastern Anatolia:

1. Delamination causes surface uplift as a result of the isostatic eﬀect of lithospheric removal.

The uplift is enhanced and evened into a plateau by plate shortening. In the absence of plate convergence, the uplift is less plateau-like since delamination-driven extension/thinning of the crust becomes more eﬃcient. A pulse of (migrating) subsidence can develop as the delaminating litho-

114 sphere loads the lithosphere at the hinge zone. Detachment of this delaminating lithosphere modiﬁes the topography near the delamination hinge. Our modeled surface topography at t = 7.0 m.y. is consistent with the proﬁle of the present day topography across Eastern Anatolia at 42◦E (Fig.

1B). Consequently, we attribute the anomalous plateau uplift in the region to delamination of the mantle lithosphere and slow plate shortening. Eastern Anatolia started to emerge from sea level

∼ 11 m.y. ago and in the last 6-7 m.y. has experienced enhanced volcanic activity and uplift (Ke- skin, 2003; Sengor et al., 2003). These events are also consistent with the proposed delamination of the mantle lithosphere beneath the plateau.

2. Delamination causes distinct zones of contraction/thickening (at the plateau/gap ﬂanks) and extension/thinning (within the delamination gap, to the far side of the delaminating hinge) within the crust. In the presence of plate convergence, the contraction and thickening is ampliﬁed but still localized to the sites determined by the delamination. Notably, the results demonstrate that crustal extension and thinning can occur within an overall plate convergent regime, depending on the rate of plate shortening. This presents a mechanism for explaining synconvergent extensional tectonics, which are an unexplained aspect of a number of orogenic regions. We compare horizontal surface strain rate (˙xx) data from the model to large-scale structural elements in eastern Anatolia. The delamination-related extension in the models corresponds with a series of normal fault bounded extensional basins across eastern Anatolia region, e.g., Mus basin, Hinis basin, and Karliova basin

- in front of the Bitlis suture zone (Dhont and Chorowicz, 2006).

5.3 The Surface Tectonics of Mantle Lithosphere Delamination

Following Ocean Lithosphere Subduction; Insights from Phys-

ical Scaled Analogue Experiments

Physical scaled analogue modeling experiments of continental mantle lithosphere delamination were conducted to explore the potential transition from ocean lithosphere subduction to continental collision and delamination. The experiments focused on how changes in the imposed plate convergence velocities changed the behaviour of the post-collisional tectonics and determined:

1. Mantle lithosphere delamination can occur with slow plate convergence where the slab peels

115 oﬀ/rolls back similar to a retreating ocean slab subduction. The results suggest that continental delamination may be a natural progression from prior ocean plate subduction and illustrate that the removal of mantle lithosphere does not necessarily require a signiﬁcant density heterogeneity to initiate. The tectonic regions of Alpine tectonics and Mediterranean basins (Royden, 1993; Seber et al., 1996; Faccenna et al., 2001) may represent suitable examples of this transition.

2. The experiments show that when the plate convergence is higher, the mantle lithosphere is less prone to delaminate from the crust. With higher plate convergence the consumed mantle lithosphere can drape forward instead. The pro-plate crust separates from the mantle lithosphere only at the collision zone and is overthrusted/accreted on top of the retro-plate. Similar tectonic modeling results of flake tectonics are suggested for alpine collision tectonics associated with folded nappes such as the Swiss Alps (Oxburgh, 1972; Ellis et al., 1999; Pfiffner et al., 2000). The surface topography is high where the crustal accretion occurs, whereas right at the plate suture zone there is subsidence as the downgoing mantle lithosphere pulls down the surface.

3. The evolution of surface topography with analogue experiments suggests that there is an appreciable amount of surface depression caused by vertical forcing of the delaminating mantle lithosphere. The surface depression migrates as the mantle lithosphere delaminates underneath the crust. This ﬁnding is similar to numerical results where a free-hanging delaminated slab — whether it is broken oﬀ or not — creates this type of migrating surface trough. With the analogue modeling results there is broad isostatic uplift of the surface topography due to the imposed plate convergence.

However, this is a relatively small magnitude uplift compared to the previous numerical models of delamination, in particular there is no well-defined plateau uplift. This is a consequence of the absence of thermal effects in the analogue models so that the mantle material moving into the delamination gap is not sufficiently buoyant to raise a plateau.

4. The surface deformation of the analogue experiments is dominated by the shortening of the crust and they do not indicate any type of lithospheric extension as was observed in numerical thermo-mechanical experiments of delamination (Gö˘gü¸sand Pysklywec, 2008a). Since the rheologies are different between the numerical (non-Newtonian, temperature-dependent) and analogue (Newtonian, temperature-independent) experiments, the lithospheric layers (crust and mantle lithosphere) respond to the stresses differently.

116 Bibliography

Anderson, D.L., 2005, Large Igneous Provinces, Delamination, and Fertile Mantle Elements, v. 1,

p. 271275, doi: 10.2113/gselements.1.5.271.

Beck, S. L., and Zandt, G., 2002, The nature of orogenic crust in the central Andes: Journal of

Geophysical Research, v. 107, B10, 2230, doi:10.1029/2000JB000124.

Bird, P., 1978, Initiation of intracontinental subduction in the Himalaya: Journal of Geophysical

Research, v. 83, p. 4975-4987.

Bird, P., 1979, Continental delamination and the Colorado Plateau: Journal of Geophysical Re-

search, v. 84, p. 75617571.

Dhont, D., and Chorowicz, J., 2006, Review of the neotectonics of the Eastern Turkish Armenian

Plateau by geomorphic analysis of digital elevation model imagery: International Journal of

Earth Sciences, v. 95, p. 3449, doi: 10.1007/s00531-005-0020-3.

Ducea, M. N., and Saleeby, J. B., 1998, A case for delamination of the deep batholithic crust

beneath the Sierra Nevada, California: International Geology Review, v. 40, p. 7893.

Elkins-Tanton, L.T., 2005, Continental magmatism caused by lithospheric delamination, in Foulger,

G.R., Natland, J.H., Presnall, D.C., and Anderson, D.L., eds., Plates, plumes, and paradigms:

Geological Society of America Special Paper, v. 388, p. 449461, doi: 10.1130/2005.2388 (27).

G¨o˘g¨u¸s,O.H., and Pysklywec, R.N., 2008a, Mantle lithosphere delamination driving plateau up-

lift and synconvergent extension in eastern Anatolia: Geology, v. 36, no. 9, p.723-726, doi:

10.1130/G2498A.1.

117 Ellis, S., Beaumont, C. and Pﬁﬀner, A., 1999, Geodynamic models of crustal-scale episodic tec-

tonic accretion and underplating in subduction zones: Journal of Geophysical Research, v. 104,

p.15169-15190.

Faccenna, C., Becker T.W., Lucente F.P., Jolivet L., Rossetti F., 2001, History of subduction and

back-arc extension in the central Mediterranean: Geophysical Journal International, v. 145, p.

809 820, doi: 10.1046/j.0956-540x.2001.01435.

Jones C.H., Farmer, G.L., and Unruh J., 2004, Tectonics of Pliocene removal of lithosphere of the

Sierra Nevada, California: Geological Society of America Bulletin, v. 116, p. 1408 1422, doi:

10.1130/B25397.1.

Jull, M., and Kelemen, P.B., 2001, On the conditions for lower crustal convective instability: Journal

of Geophysical Research, v. 106, p. 64236446, doi: 10.1029/2000JB900357.