Previous geodynamic models of continental collision show that the behavior of the lithosphere can be strongly influenced by the presence of surface erosion. That said, absent from these investigations are the effects of sediment deposition. We quantitatively investigate this process using thermal-mechanical numerical experiments of the coupled processes of tectonic deformation and crustal mass flux. The models demonstrate that the inclusion of the effects of sediment deposition can change the style of deformation of the crust and consequently, the evolution of the underlying deforming mantle lithosphere. In the absence of sediment deposition, the early stages of collision are accommodated by subduction of lower crust and mantle lithosphere along a discrete shear zone beneath the overriding plate. Following this initial stage of subduction, the subducting lower crust and mantle lithosphere retreat from the collision zone, permitting the sub-lithospheric mantle to upwell and come into contact with the thickened upper crust. When sediment deposition is imposed subduction-like consumption of the subducting plate remains stable. The presence of sediment deposition introduces a negative Vy-component in the overriding plate in the area adjacent to the collisional zone. The negative Vy-component leads to a greater degree of coupling between the colliding continental plates and decoupling of the overriding upper crust and lower crust/mantle lithosphere. The results demonstrate the first quantitative insights into the feedback between surface deposition and tectonics.
 Geological observations/interpretations as well as crustal- and lithosphere-scale geodynamic models show the behavior of the lithosphere to be strongly influenced by climate-controlled surface processes. The first crustal-scale numerical treatments of climate-tectonic interactions showed that surface denudation, occurring over distances and at rates comparable to tectonic mass flux, can significantly modify the evolution of a continental orogen [e.g.,Beaumont et al., 1992; Willett, 1999]. Later crustal-scale experiments demonstrated that during continental collision the coupled processes of surface erosion and tectonic deformation could produce a modified nature of folding and higher topography than in the absence of erosion [Simpson, 2006]; highly localized crustal deformation in response to concentrated erosion [Stolar et al., 2007]; topographic asymmetry with shallower slopes facing the subducting plate and an asymmetric pattern of exhumation [Willett, 1999]; and localized crustal shortening as well as the development of an intracontinental mountain range [Avouac and Burov, 1996]. The geodynamic investigation of the influence of climate on the tectonic evolution of collisional orogens is not restricted to numerical treatments. Scaled analogue experiments clearly show that the location, mode and rate of surface erosion influence the style, pattern and rate of rock deformation [Whipple, 2009]. Extending this theme to the deep lithosphere (i.e., sub-crustal lithosphere), lithosphere-scale models of continental collision showed a remarkably deep reach of surface erosional processes. For example, in the presence of surface erosion a stable subduction-like plate consumption of continental lithosphere may be maintained, whereas in the absence of erosion, continental subduction is inhibited by the accumulation of buoyant crust and can lead to retreat of the initially subducting plate [Pysklywec, 2006]. In summary, these studies (and others) have developed an important new understanding of tectonic solid Earth systems with respect to the surface forcings of the active hydrosphere and atmosphere.
 The concepts of climate-tectonic interactions derived from this research have been used to explain various enigmatic features of, specifically, collisional zones. The asymmetric elevation profile of the Southern Alps of New Zealand, where the retro-wedge is characterized by wetter and steeper slopes, may be explained by the dominance of tectonic advection over fluvial incision in controlling topography [Willett, 1999]. Pysklywec suggests that very high surface erosion rates on South Island, New Zealand may explain the presence of subduction of sub-crustal lithosphere in response to the convergent component of oblique collision. This is in sharp contrast to the deformation beneath the Transverse Ranges of California where subduction-like behavior is not occurring [Houseman et al., 2000], possibly due to relatively subdued surface processes. In the Olympic Mountains of Washington State, high exhumation (∼14 km) in the center of the orogen, with decreasing exhumation towards to the periphery of the orogen may be caused by enhanced pro-wedge erosion [Willett, 1999]. As a last example, in the southern portion of the Himalayan-Tibetan orogen the South Tibetan Detachment System, a series of north-dipping, east–west striking normal faults, may be the result of exhumation of the greater Himalayan sequence driven by channel flow and ductile extrusion dynamically linked through the effects of surface denudation [Beaumont et al., 2001].
 Most numerical models of continental collision include the combined effects of surface erosion and sediment deposition [e.g., Burov and Toussaint, 2007; Yamato et al., 2008]. Burov and Toussaint for instance showed that the presence of moderate erosion/sedimentation may reduce the resistance of the major thrust and subduction channel to subduction, whereas very strong or weak erosion/sedimentation may enhance plate coupling and promote pure-shear thickening or buckling. Despite the presence of erosion/sedimentation in most numerical models of continental collision, how, specifically, sediment deposition may modify the evolution of the continental lithosphere during pate collision is not well understood. As the conjugate to surface erosion, it may be expected that deposition of the removed mass may similarly influence the lithospheric tectonics. Here, we quantitatively investigate this process using thermal-mechanical numerical experiments of the coupled processes of tectonic deformation and crustal mass flux. A free surface, prescribed erosional laws (e.g., empirically derived relief-dependent erosion) and sediment deposition dependent on the amount of material eroded make up the top boundary of the model domain and allow topography to develop self consistently with the underlying geodynamics. We conduct a series of experiments with varying convergence rates, initial MOHO temperatures and degrees of sediment deposition to show how the inclusion of the effects of sediment deposition produce a modified style of continental plate behavior during collision, as one example of tectonic activity where climate-tectonic interactions are most often invoked. The results give some of the first insights into how deep tectonics is influenced by climate-controlled erosion-deposition processes.
 We solve the coupled conservation equations of mass, momentum and internal energy that govern the behavior of plane-strain (i.e., 2D), visco-plastic incompressible media by using the arbitrary Lagrangian-Eulerian finite element method [Fullsack, 1995]. Idealized continental collision is modelled by introducing new lithosphere at the right (pro-) margin of the box at a velocity of 1.5 , while lithosphere at the left (retro-) margin of the box is held fixed (Figure 1a). To ensure mass balance, an outward flux of 0.19 is distributed evenly along the sides of the sub-lithospheric mantle (S.L.M.) A small weak seed (10 km × 10 km) is inserted into the upper mantle lithosphere to localize the initial deformation. An empirical law for relief-dependent erosion rate, E, in tectonically active regions [Montgomery and Brandon, 2002] is used and applied to the entire width of the model domain: E = Eo + , where k = 0.6 is a rate constant, Sc= 37° is a limiting hill-slope gradient, Eo = 0.05 is the background erosion rate due to weathering and S is mean surface slope. We model deposition by: 1) summing the material eroded at each Eulerian node on the respective slope faces of the model orogen; 2) dividing the amount of material eroded on each slope face by the number of Eulerian cells in the basins (i.e., depressions relative to the initial height of the model) adjacent to their respective slope faces; and 3) adding the result to each Eulerian node in the depressions adjacent to the respective slope faces. For more information about the numerical model, material properties and thermal setup, the reader is referred to the auxiliary material.
 The results of RUN1 (Figures 1b–1e) show how idealized continental collision may proceed when continental lithosphere is subject to varying surface erosion. After 285 km of imposed convergence (Δx = 285 km), corresponding to 19 m.y., retro-plate upper crust and lower crust/mantle lithosphere separate at the collision zone. The upper crustal material is thrust up and the lower crust/mantle lithosphere are thrust down along retrovergent shear zones. Although a portion of the subducting retro-plate has detached and descended into the S.L.M. (Figure 1b), at this stage in the model's evolution the localized deformation is generally stationary (i.e., the subducting retro-plate material shows no propensity for retreat). Owing to the cumulative density difference between the subducting retro-side material (i.e., lower crust and mantle lithosphere) and the overriding plate (δρ = 53 ) and to the decoupling between the upper and lower crust in the area adjacent to the shear zone, by Δx = 577 km the subducting retro-side material has started to retreat from the shear zone. The initiation of retreat has allowed a portion of upper crustal material to enter the subduction zone. By Δx = 600 km, subducting retro-plate lower crust and mantle lithosphere continue to retreat from the subduction zone. By this stage in the model's evolution, the retreat of subducting retro-plate material has sufficiently progressed to allow upwelling S.L.M. to come in contact with eclogitized retro-plate lower crust that has accumulated in the subduction zone. At the end of the model's evolution (Δx = 855 km), the subducting retro-plate material has retreated from the subduction zone by ∼250 km, allowing the hot S.L.M. to upwell and come in contact with the thickened upper crust.
 In experiment RUN2 where sediment deposition is imposed, subduction-like consumption of the retro-plate remains stable throughout the model's evolution (Figures 1f–1i). The differences in the fundamental behavior of the two experiments may be understood by considering how sediment deposition alters the kinematics of the pro-plate in the area adjacent to the collision zone. Initially experiments RUN1 and RUN2 are similar, but by Δx = 577 km, important differences develop. In RUN2, by Δx = 577 km the incoming pro-plate has a y-component (vertical) of velocity (Vy) of ∼−0.1 (velocity profile taken at the black vertical line on the pro-plate lithosphere). The negative Vy-component is due to sediment deposition in the basins adjacent to the model orogen and the resultant eclogitization of the pro-plate lower crust. In essence, sediment deposition is pushing down the incoming pro-plate, leading to eclogitization of the lower crust and a greater degree of coupling with the retro-plate (Figures 1g and 2f). It is important to note that the formation of lower crustal high density eclogite is fundamental to this process. That said, due to the negative Vy-component, the pro-plate has descended to deeper levels in the computational box. This has lead to higher crustal temperatures and consequently a zone of high strain-rate at the pro-plate upper crustal/lower crustal interface (Figure 2g). Effectively, this zone of high strain-rate decouples the pro-plate upper crust from the underlying lower crust/mantle lithosphere, leading to a more diffuse style of upper-crustal deformation (Figure 2g). By Δx = 600 km, the Vy-component of the pro-plate persists, continuously coupling the retro- and pro-plates and disallowing the minor component of retro-plate retreat that is present in RUN1 at the same stage of the model's evolution. By Δx = 855 km, the continued sedimentation has permitted the retro- and pro-plates to remain coupled. By continuing to create a negative Vy-component in the pro-plate, sedimentation has also resulted in eclogitization of the pro-plate's lower crust along a horizontal distance of ∼250 km. It is interesting to note how the average Vy-component in the pro-plate in the area adjacent to the collisional zone is dependent on the deposition % (e.g., 20% deposition means that 20% of the material eroded on the orogen's slopes is deposited in the adjacent basin).Figure 3 shows that for increasing deposition %'s the average Vy-component of the pro-plate in the area adjacent to the collisional zone increases in negativity. This occurs because with increasing deposition % more mass is deposited in the basins adjacent to the model orogen, exerting a greater downward force on the pro-plate.
 Experiments RUN1 and RUN2 illustrate how sedimentation can control deep lithosphere processes; namely active sedimentation prevents continental lithosphere retreat during continental collision. We ran a suite of experiments that test the sensitivity of this model to varying convergence rates and deposition %'s as well as varying initial MOHO temperatures and deposition %'s. Figure 4aillustrates how the horizontal distance along which the pro-plate's lower crust is eclogitized increases with decreasing convergence rates and increasing deposition %. Because the numerical experiments undergo the same amount of shortening (Δx = 855 km), the scaled duration of the experiments increases with decreasing convergence rates. Consequently, once the pro-plate's lower crust undergoes eclogitization it will undergo a greater degree of eclogitization at lower convergence rates because the numerical experiments with lower convergence rates require more time to achieve total shortening of Δx = 855 km. The degree of eclogitization of the pro-plate's lower crust increases with increasing deposition % because the negativity of the Vy-component in the pro-plate increases with increasing deposition %.Figure 4billustrates how the amount of retro-plate deep lithosphere retreat increases with decreasing convergence rates and decreasing deposition %. The amount of deep lithosphere retreat increases with decreasing convergence rates because the retro-plate is subjected to the pull of the subducting/retreating retro-plate material for greater amounts of time and because the incoming plate is too slow to fill the gap as the retro-plate retreats; retro-plate retreat increases with decreasing deposition % because the decreasing negativity of the pro-plate's Vy-component that accompanies a decrease in deposition % leads to increasing retreat. Perhaps most interesting is how deposition % trumps convergence rate in controlling the degree of pro-plate lower crust eclogitization and retro-plate deep lithosphere retreat. We emphasize that this is the result of the Vy-component applied to the incoming pro-plate as a result of sediment deposition.
Figure 5aillustrates how the horizontal distance along which the pro-plate's lower crust is eclogitized increases with increasing MOHO temperatures and increasing deposition %. Increasing the initial MOHO temperature increases the eclogite stability area (pressure ≥1.2 GPa and temperature ≥500°C) in the computational domain. Similar to the suite of experiments where the convergence velocity was varied, increasing deposition %'s increase the horizontal distance along which the pro-plate lower crust is eclogitized because the negativity of the Vy-component in the pro-plate increases with increasing deposition %.Figure 5billustrates how, for initial MOHO temperatures between 475°C and 625°C the amount of retro-plate deep lithosphere retreat increases with increasing initial MOHO temperatures. Increasing the initial MOHO temperature decreases the viscosity of the upper crust, increasing the degree of decoupling between the retro-plate upper crust and lower crust/mantle lithosphere. This increases the propensity of the retro-plate deep lithosphere to retreat from the collisional zone. With an initial MOHO temperature of 700°C the high temperatures in the model place more of the upper mantle lithosphere into the viscous regime and result in a thinner layer of plastic mantle lithosphere. The initially high MOHO temperature results in rapid Rayleigh-Taylor downwellings of the underthrusting deep retro-plate material [e.g.,Pysklywec et al., 2002]. Consequently, retreat of subducting deep retro-plate material occurs at a slower rate and results in less retreat than the models with initially cooler MOHO conditions. With an initial MOHO temperature of 400°C the deep upper crust in the area adjacent to the collision zone is in the plastic regime, producing an initially coupled upper crust and lower crust/mantle lithosphere. As the model progresses in time the comparatively low deep lithosphere temperatures result in gravitational instability greater than that which can be supported by coupling of the upper crust and lower crust/mantle lithosphere. The result is deep-lithosphere retreat independent of deposition %'s.
4. Discussion and Conclusions
 The numerical geodynamic experiments demonstrate that the effects of sediment deposition may alter the style of deep lithosphere deformation during continental collision. Sediment deposition in the basins adjacent to the model orogen significantly influences crustal mass flux within an actively deforming orogen. This consequently alters crustal evolution (particularly in the overriding plate) at the lower crust/mantle lithosphere interface and the style of mantle lithosphere deformation a the plate boundary. The presence of sediment deposition introduces a negative Vy-component in the overriding plate in the area adjacent to the collisional zone. The negative Vy-component leads to: a) a greater degree of coupling between the colliding continental plates, mitigating retreat of deep retro-plate lithosphere; and b) decoupling of the pro-plate upper crust and lower crust, producing a more diffuse style of upper-crustal deformation. In the absence of sediment deposition, the negative Vy-component in the overriding plate vanishes. As a result, 1) the resulting lack of coupling between the colliding plates combined with the weight of the subducting retro-plate material promote retreat of the subducting plate material from the collision zone; and 2) the strong coupling between the crust and mantle lithosphere of the overriding plate results in more localized crustal deformation. It is worth mentioning that in most “real-world” cases deposition histories and the amount of sedimentation are better constrained than denudation rates and the amount of material eroded. Consequently, present models of continental collision account for the effects of sedimentation better than for erosion and the risk of underestimating the impact of sedimentation on collision is limited.
 In these experiments we have focused on continental orogenesis as the tectonic regime where climate-tectonic interactions are studied. The two-dimensional nature of the models enables the surface-tectonic dynamics to be treated at a high numerical resolution. However, we recognize that out-of-plane sediment transport and tectonic motion will alter the results. Retreat/non-retreat of the mantle lithosphere is the deep tectonic observable that we have focused on for illustrative clarity of the interpretations. Other tectonic processes (topography, crustal structure, thermal evolution) are similarly strongly modified by varying surface deposition. The results demonstrate the initial qualifications of first-order significance of climate-controlled deposition and tectonic interactions.
 We thank two anonymous reviewers for comments/suggestions that greatly improved this paper. R.G. was funded by an Alexander Graham Bell Canada Graduate Scholarship and R.N.P. was funded by NSERC.
 The Editor thanks the two anonymous reviewers for assisting in the evaluation of this paper.