Across-strike asymmetry of the Andes orogen linked to the age and geometry of the Nazca plate

The across-strike asymmetry of orogens is rooted in tectonically imposed spatial gradients in lateral and vertical rock advection (Willett et al., 1993). As mountains grow and develop orographic-precipitation effects, high precipitation rates on the wet side are thought to force mountain range asymmetry by increasing the erosional efficiency and driving mountain crest migration toward the dry side (Willett, 1999). In the Andes orogen (Fig. 1), orographic effects have sustained extreme precipitation gradients since Miocene times (Mulch et al., 2010; Insel et al., 2012). Their coincidence with systematic variations in crustal shortening, exhumation, and across-strike asymmetry of orogen mass has been attributed to a first-order control of orographic precipitation on asymmetric erosion (Masek et al., 1994; Horton, 1999; Montgomery et al., 2001; Strecker et al., 2007; Mora et al., 2008). However, in orogens bounded by subduction zones, the properties and dynamics of the subducting slab (i.e., age and geometry) partly control the spatial distribution of strain and uplift in the upper plate (Jordan et al., 1983; Gephart, 1994; Ouimet and Cook, 2010; Capitanio et al., 2011; Horton, 2018). An open question we address is: How much of the acrossstrike asymmetry along the Andes is rooted in these tectonic drivers compared to precipitation distribution?

Decoupling the purported climatically driven orogenic asymmetry from tectonic control is a difficult task because there are few techniques or proxies for constraining the lateral advection of rocks in large orogens such as the Andes (e.g., Eizenhöfer et al., 2019). We approach this question by seeking evidence of the diagnostic relationships between topography, erosion rates, precipitation rates, and across-strike asymmetry as predicted by fluvial incision and critical taper theories. The Andes provide a unique natural laboratory given the geometries and age range of subducting slabs that are found in sections of the continent where winds originate from the west or the east, allowing a separation of the effects of tectonics and direction of the orographic effect. Using data on the age and geometry of the Nazca slab as proxies for tectonic controls over orogenic wedge evolution and pro-wedge and retro-wedge precipitation rates, we statistically assessed the partitioning between tectonic and climatic contributions to orogen asymmetry. We used a compilation of 375 basin-wide erosion rates to assess the statistical findings based on bedrock fluvial incision theory.

Fluvial incision theory predicts that higher precipitation rates increase erosional efficiency and decrease the steady-state topographic steepness required for a given uplift rate (e.g., Whipple, 2009). This topographic feedback can change orogenic asymmetry by driving mountain crest migration but only under spatially uniform uplift rates (e.g., Goren et al., 2014). If predetermined by a tectonic uplift gradient, only extreme orographic effects such as in the Andes can impact but not reverse an orogen’s asymmetry (Willett, 1999; Goren et al., 2014).

From critical taper theory, high erosional efficiency reduces or limits the total and retrowedge widths for a given tectonic accretionary flux (Dahlen, 1990), which necessarily accelerates rock uplift rates (Whipple, 2009). If orographic precipitation drives this positive feedback, the diagnostic evidence would be high erosion rates and reduced orogenic widths coinciding with regions of high orographic precipitation (Whipple and Meade, 2004). Given the east-verging nature of the Andes growth, high erosional efficiency and precipitation rate in the pro-wedge is expected to favor pro-wedge widening, whereas high erosional efficiency in the retro-wedge should limit the propagation of the deformation front (Horton, 1999; Willett, 1999; McQuarrie et al., 2008). If orographic precipitation is the principal driver of orogenic asymmetry in the Andes, the retro-wedge width should be systematically narrower and the orogen systematically more asymmetric (Fig. 1C). In this study, we assessed these diagnostic relationships in the Andes mountains.

We used multilinear models to statistically assess the partial tectonic and climatic contributions to orogen asymmetry. In a convergent margin, the subduction zone megathrust balances the forces in the subduction zone–orogen interface (Meade and Conrad, 2008). Thus, orogen asymmetry is calculated as the ratio of trenchorthogonal distances from the trench to the peak mean topography (i.e., pro-wedge) and from the trench to the retro-wedge deformation front (i.e., total width). A value of 0.5 reveals a symmetrical orogen, and values below or above 0.5 reveal a narrower or wider pro-wedge, respectively. We positioned the deformation front according to the geologic map of South America (Gómez et al., 2019) and corrected to include retrowedge basement uplifts in Peru and Argentina (Fig. 1; Text S1 in the Supplemental Material¹). The average position of the crest was identified in 100-km-wide, approximately trench-normal topographic swath profiles over añ500-m-resolution digital elevation model (DEM) (Tozer et al., 2019) by averaging the location of the highest 1% peaks in mean elevation (Fig. 1; Fig. S1). We excluded the Altiplano-Puna plateau given that its crustal deformation reflects ductile behavior and lateral crustal flow (e.g., Ouimet and Cook, 2010) and likely deviates from the critical taper predictions for orogenic wedges.

We used multilinear models of the form Y = r + β₁X₁ + β₂X₂, where Y is orogen asymmetry, r are the model residuals, X₁ and X₂ are predictors of Y (as described below), and β₁ and β₂ are the regression coefficients (the regression slopes). At-trench slab age (Seton et al., 2020) and slab dips (0–125 km depth) (Hayes et al., 2018) of the Nazca slab are two metrics we used as the tectonic predictors, based on the literature (Capitanio et al., 2011; Horton, 2018). For climatic predictors, we used modal pro- and retrowedge precipitation rates from a 34 yr annual average precipitation rate data set (1 km resolution; Karger et al., 2017) (Figs. S2 and S3). Four models use a tectonic predictor (slab age or slab dip as X₁) and a climatic predictor (pro- or retrowedge precipitation rate as X₂). A fifth model uses slab age as X₁ and slab dip as X₂. Predictors were centered at a mean of 0 and normalized to a standard deviation to yield comparable regression slopes (b₁ and b₂) and inform the relative dependence of Y on X₁ and X₂. The residuals and regression results were then used to detect nonlinear dependencies of Y on precipitation rates (see Text S2). Multilinear analysis requires linear correlations with the response variable. Thus, we used log slab dip and log precipitation rate for improved linearity. None of the X₁-X₂ pairs used here are affected by severe collinearity (i.e., R² > 0.8).

We used previously compiled basin-wide erosion rates (Codilean et al., 2018), morphometric averages (hillslope gradient and river steepness, k_sn), and average precipitation rates (Karger et al., 2017) to assess potential precipitation-erosion rate links in 375 basins (see Text S3). We extracted basin-wide topographic data from 90 m and 30 m DEMs (NASA, 2013) for basins greater and smaller than 5 km², respectively, using topographic tools (Schwanghart and Scherler, 2014; Forte and Whipple, 2019). We explored potential climatic imprints on erosion rates using the enhanced river steepness (k_snq), a metric that accounts for climate using river discharge, q (computed as precipitation ´ the in-stream flow accumulation grid). If climate influences erosional efficiency, k_snq should correlate with erosion rates better than k_sn (Adams et al., 2020).

The time span of the precipitation rate data set used here is undoubtedly short compared to the time scales of cosmogenic nuclide integration (10²–10⁵ yr) and mountain building (10⁷ yr). However, the orographic-precipitation effects and thus the relative spatial distribution of precipitation rates are long-lasting features of the Andes and established at least since the Miocene (Strecker et al., 2007; Mulch et al., 2010; Insel et al., 2012). Thus, basin-wide averages likely reflect the long-term differences between the studied regions. Lastly, any possibility of ¹⁰Be erosion rates having been reset by past climate (e.g., Insel et al., 2010) is likely diluted in the large data set (n = 375).

Orogen asymmetry in the Andes varies between 0.29 and 0.89 and is correlated negatively with slab age and positively with slab dip and pro-wedge precipitation rate, and not correlated with retro-wedge precipitation rate (Fig. 2). Slab age alone explains 61% of the variability in orogen asymmetry, followed by slab dip and pro-wedge precipitation rates (each 55%). The directions of the observed relationships are consistent with scenarios in which each parameter individually influences the asymmetry (Figs. 1C and 2A–2D). Multilinear models composed of tectonic and climatic variables explain 65% of the variability in orogenic asymmetry but do not represent a significant improvement from the univariate model using slab age (Figs. 2A and 2E). The fully tectonic model (i.e., slab age as X₁ and slab dip as X₂) explains 70% of the variability in orogen asymmetry. Precipitation rates are, therefore, secondary predictors at best.

Slab age and dip have higher regression slopes in all the models tested and thus are more important predictors of orogen asymmetry than pro- or retro-wedge precipitation rates (Figs. 2E–2I). Albeit weakly, orogen asymmetry is nonlinearly related to pro-wedge precipitation after adjusting for slab age (Fig. 2J). Here, regression slopes are steepest at pro-wedge precipitation rates <~500 mm/y. This is also true for pro- and retro-wedge precipitation adjusted to slab dip (Figs. 2L and 2M). Beyond ~500 mm/y, the nonlinear slopes are insignificant and reveal that high orographic precipitation rates have no statistical effect on orogen asymmetry.

Erosion rates are weakly correlated with precipitation rates, moderately correlated with river steepness (k_sn), and mostly explained by hillslope gradients (R² ~45%) (Figs. 3A and 3B). Dividing the data into precipitation bins reveals no systematic increase in erosion rate (Figs. 3B and 3C), contrary to what would be expected if precipitation rates had limited the growth of the orogen. We observe a statistically significant improvement in the regression against erosion rate going from k_sn (R² = 0.29) to k_snq (R² = 0.54) (Figs. 3D and 3E). The improvement is greatest for the driest basins (precipitation of 0–250 mm/y) but insignificant for wetter ones (Fig. 3F). This observation lends process-based empirical support to the statistical observations.

Slab ages are nearly symmetrical about the Altiplano-Puna plateau and describe acrossstrike orogen asymmetry in a statistically significant, unifying trend (Fig. 2A). This observation suggests a primary control of orogen asymmetry rooted in the subduction zone. Older segments of the subducting Nazca slab require a thicker crust in the South American plate to balance tectonic forces with high vertical stresses near the trench (Capitanio et al., 2011). In areas next to an older slab (and thicker continental crust), compressive forces propagate further into the orogen over time and drive retro-wedge growth (Capitanio et al., 2011). The wedge width data presented herein suggests this process leads to greater accumulation of mass in the retro-wedge area (Fig. 2A; Fig. S3) and is evident in the across-strike profiles along the Andes (Fig. 4). That slab dip can also partly explain orogen asymmetry suggests it is of secondary importance compared to slab age, consistent with previous studies (Jordan et al., 1983; Gephart, 1994; Capitanio et al., 2011; Horton, 2018).

Changes in the slab shape and age accompany changes in the force balance of the subduction zone as well as the shape of the mantle wedge and lithosphere-asthenosphere boundary underneath the orogen, all of which form a bottom-up control on the direction of tectonic transport and orogen structure (Schellart, 2017; Barrionuevo et al., 2021). Together with our results, this mechanism suggests that the orogenic wedge asymmetry in non-collisional orogens is subject to subduction zone dynamics.

The spatial coincidence of heavy orographic effects and asymmetric mass distribution along the Andes orogen have long been thought to originate in its sustained control on precipitation distribution (e.g., Horton, 1999; Montgomery et al., 2001; McQuarrie et al., 2008). Our analysis of k_snq-erosion relationships and statistical observations suggests that precipitation influences erosional efficiency mostly in drier regions but not otherwise (Fig. 3F). These results reveal that it is at precipitation rates <~500 mm/y that precipitation changes more likely influence erosional efficiency and orogen shape (Figs. 2J–2M). This nonlinear dependence on climate may be related to critical erosion thresholds, which are more likely exceeded in drier climates (e.g., DiBiase and Whipple, 2011). We conclude that high orographic precipitation has minimal influence on erosional efficiency and, therefore, cannot drive changes in wedge widths, which are set by bottom-up tectonic processes. Unlike previously thought (Montgomery et al., 2001), the influence of climate on the asymmetry of the Andes is limited and, therefore, not primary.

P. Val was funded by the U.S. National Science Foundation (award 1651243 to J.K. Willenbring). We thank F. Capitanio, K. Murray, N. Perez, S. Willett, and three anonymous reviewers. We also thank H. Reis, N. Lyons, and D. Peifer for insightful discussions.