The Magallanes-Austral retroarc foreland basin of southernmost South America presents an excellent setting in which to examine interpretive methods for large detrital zircon data sets. The source regions for retroarc foreland basins generally, and the Magallanes-Austral Basin specifically, can be broadly divided into (1) the magmatic arc, (2) the fold-and-thrust belt, and (3) sources around the periphery of foreland flexural subsidence. In this study, we used an extensive detrital zircon data set (30 new, 87 previously published samples) that is complemented by a large modal provenance data set of 183 sandstone petrography samples (32 new, 151 previously published) and rare earth element geochemical analyses (130 previously published samples) to compare the results of empirical (multidimensional scaling) and interpretive (age binning based on source regions) treatments of detrital zircon data, ultimately to interpret the detailed evolution of sediment dispersal patterns and their tectonic controls in the Magallanes-Austral Basin. Detrital zircon sample groupings based on both a priori age binning and multidimensional scaling are required to maximize the potential of the Magallanes-Austral Basin data set. Multidimensional scaling results are sensitive to differences in major unimodal arc-related U-Pb detrital zircon ages and less sensitive to differences in multimodal, thrust belt–related age peaks. These sensitivities complicate basin-scale interpretations when data from poorly understood, less densely sampled sectors are compared to data from better-understood, more densely sampled sectors. Source region age binning alleviates these biases and compares well with multidimensional scaling results when samples from the less well-understood southern basin sector are excluded. Sample groupings generated by both multidimensional scaling and interpretive methods are also compatible with compositional provenance data. Together, this integration of provenance data and methods facilitates a detailed interpretation of sediment dispersal patterns and their tectonic controls for the Late Cretaceous to Eocene fill of the Magallanes-Austral retroarc foreland basin. We interpret that provenance signatures and dispersal patterns during the retroarc foreland phase were fundamentally controlled by conditions set by a predecessor extensional basin phase, including (1) variable magnitude of extension with latitude, (2) the composition of lithologies emplaced on the antecedent western flank, and (3) long-lasting structural discontinuities associated with early rifting that may have partitioned dispersal systems or controlled the location of long-lived drainage networks.


Retroarc foreland basins and their source regions tend to be areas of relatively high heterogeneity in terms of both crustal composition and geochronologic fingerprints because they form next to areas characterized by active magmatism juxtaposed with compressional deformation of older crustal material (Nie et al., 2012), and they are receptacles for detritus derived from the resulting mix of sources. As such, they inherently offer an opportunity to evaluate methods for using provenance data to determine the scale at which ancient sediment dispersal systems can be resolved based on source signatures and their controls on shifting provenance signatures through time (e.g., Horton et al., 2010; Nie et al., 2012; Saylor et al., 2011; Horton, 2018). Nie et al. (2012) proposed three broadly defined source regions for retroarc foreland basins (Fig. 1), including (1) the continental magmatic arc, (2) the fold-and-thrust belt, and (3) the stable craton and continental block uplifts on the periphery of the basin. Sediment provenance methods that employ geochronology to derive signatures of sedimentary source regions (e.g., U-Pb detrital zircon analysis) have intrinsic leverage in retroarc foreland basins because coeval magmatic arc material is generally entirely younger than the material that is uplifted by compressional tectonism in the fold-and-thrust belt or older continental block material on the periphery of the basin (e.g., Fig. 1; Horton et al., 2010; Nie et al., 2012; Saylor et al., 2011; Horton, 2018). The tectonic history of the Magallanes-Austral retroarc foreland basin in southernmost Patagonia (Fig. 2) generated a basin configuration wherein all three retroarc foreland source regions have distinguishable geochronology-based provenance signatures (Malkowski et al., 2017).

The geochronology provenance data set available from the Magallanes-Austral Basin includes U-Pb detrital zircon (DZ) data from 30 new and 87 published samples (117 total, n = 16,392 individual grain analyses). Geochronology data sets of this size are increasingly common in sedimentary geology and basin analysis and require an interpretive scheme that goes beyond the visual comparison of age spectra. In this study, we examined two methods for visualizing and interpreting large DZ data sets in retroarc foreland basins: (1) multidimensional scaling, which is a quantitative assessment of the relative statistical dissimilarities between samples, and (2) interpretive age binning based on source region information. We demonstrate that the utility of either method depends on the sample density across the basin, both temporally and spatially, as well as the geochronological complexity of zircon age spectra in terms of the number and age of dominant U-Pb age peaks. We combined the results of the DZ interpretive methods with modal provenance data from 32 new and 151 published sandstone point counts and 130 published rare earth element (REE) geochemical analyses, as well as paleocurrent information (824 new and 4556 published measurements), to evaluate the evolution of and tectonic controls on sediment dispersal in the Magallanes-Austral Basin from Late Cretaceous into early Cenozoic time.


The Magallanes-Austral Basin sits on the remnants of the southwestern margin of the Gondwana supercontinent, in what is now southern Patagonia (47.5°S–56°S; Fig. 2). The oldest rocks exposed at the current latitude of the Magallanes-Austral Basin are Paleozoic metasedimentary complexes known as the Eastern Andean Metamorphic Complex (Figs. 2 and 3A; Faúndez et al., 2002). East-dipping subduction initiated along the margin (at Magallanes-Austral Basin paleolatitudes) during Permian to Triassic time (Fig. 3B; Hervé et al., 2003, 2013). Permian–Triassic subduction near the Magallanes-Austral Basin is recorded by accretionary complexes that are located on the western side of the modern Andes, referred to as the Duque de York complex (Figs. 2 and 3B; Rapalini et al., 2001; Faúndez et al., 2002; Hervé et al., 2003).

Jurassic extension occurred at the paleolatitude of the Magallanes-Austral Basin beginning ca. 190 Ma associated with the early rifting and separation of southwestern Gondwana (Fig. 3C). Extension produced three phases of silicic volcanism that were progressively emplaced from the north and east toward the south and west (Fig. 3CD; Pankhurst et al., 2000). The first phase (ca. 200–178 Ma) is recorded by the North Patagonia Massif (Fig. 3C), and the second phase (172–157 Ma) is recorded by the Deseado Massif (Fig. 3C; Pankhurst et al., 2000). The third phase of volcanism (represented by the Tobífera and El Quemado Formations, 157–145 Ma; Fig. 3D) was concentrated in the southern portion of the extensional margin and was associated with the initiation of the north-south–trending Rocas Verdes extensional basin (Fig. 3D; Biddle et al., 1986; Pankhurst et al., 2000; Malkowski et al., 2015b). The remnant Rocas Verdes basin is recorded by large grabens that offset continental basement (largely Eastern Andean Metamorphic Complex rocks; Biddle et al., 1986; Stern and de Wit, 2003), which can be observed both in outcrop and in the subsurface (Biddle et al., 1986; Poiré and Franzesa, 2010; Fosdick et al., 2011). The Rocas Verdes basin effectively “unzipped” from south to north, where both the magnitude of extension was higher and the duration of extension was longer in the south (Malkowski et al., 2015b), ultimately resulting in the emplacement of oceanic crust in the southern part of the basin (Figs. 2C and 3D; de Wit and Stern, 1981; Mukasa and Dalziel, 1996). In contrast, the northern part of the basin was less extended and remained floored by attenuated Eastern Andean Metamorphic Complex basement (Figs. 2C and 3D; Ramos et al., 1982). The remaining history of the Rocas Verdes basin is represented by mud-dominated marine deposition across the entire basin (Figs. 2B and 3E; Biddle et al., 1986; Wilson, 1991).

A transition to a compressional tectonic regime occurred in Albian time along the entire western margin of South America (Fig. 3F; Maloney et al., 2013). This shift to compression inverted the Rocas Verdes basin and initiated the Magallanes-Austral Basin (Fig. 3F). Based on outcrop, core, and seismic observations, clastic deposition appears to have been continuous during Cretaceous time, with no significant basinwide hiatus in deposition from the end of Jurassic extension through the transition to compression and Cretaceous flexural subsidence (Figs. 2B and 4; Wilson, 1991; Arbe, 2002; Sickmann et al., 2018). The only apparent basinwide unconformity occurs between Upper Cretaceous and Middle Eocene deposits (Fig. 2B). Due to north-south crustal heterogeneities inherited from differential extension (Figs. 2C and 3D–3E), the subsequent compression-related evolution of the Magallanes-Austral Basin was also longitudinally variable (e.g., Malkowski et al., 2015a, 2017). Because the southern portions of the Magallanes-Austral Basin were the most attenuated and weakened during extension, they subsided more deeply in response to topographic loading of the fold-and-thrust belt (Figs. 2C and 4; Fosdick et al., 2014). According to this variability in extensional and subsequent compressional history, we defined three basin sectors following established regional nomenclature (e.g., Arbe and Hechem, 1984; Malkowski et al., 2015a): (1) the Austral sector (northern part of the basin; 48.5°S to 50.2°S; Fig. 2); (2) the Magallanes sector (central part of the basin; 50.2°S to 52.5°S; Fig. 2); and (3) the southern Fuegian sector (southern part of the basin; 52.5°S to 56°S; Fig. 2).

Based on outcrop and geochronological studies, the transition to foreland basin subsidence and deposition occurred earlier in the northern part of the Magallanes-Austral Basin than in the south (Fig. 4; Malkowski et al., 2015a). Depositional systems associated with this transition indicate shallower water depths in the north compared to southern temporal equivalents due to a long-lived, southward-dipping depositional gradient in the basin (Figs. 2B and 4; Arbe, 2002; Sickmann et al., 2018). Due to lithospheric weaknesses that were inherited from the extensional phase of the basin, the Magallanes-Austral Basin foredeep appears to have been deeply subsided and “pinned” for an extended period of time with a subdued forebulge (Fosdick et al., 2014).

At any latitude in the basin, the stratigraphic evolution of the Late Cretaceous to Eocene Magallanes-Austral Basin can be distilled into three general depositional successions. We use the term “depositional succession” to broadly define related stratigraphic packages within the basin fill, with no intended reference to sequence stratigraphic nomenclature. The three successions represent: (1) an underfilled period of deposition in environments similar to those that were present during foreland initiation (e.g., shallow-marine to deep-marine environments, depending on the location; Wilson, 1991; Fildani and Hessler, 2005; Malkowski et al., 2015a); (2) a period of rapid filling, during which the shelf-slope break prograded more than 100 km southward in 5–10 m.y. (e.g., Hubbard et al., 2010; Schwartz et al., 2016; Daniels et al., 2017; Sickmann et al., 2018); and (3) a period of coupled overfilling and tectonic uplift of the Late Cretaceous foredeep, recorded by the development of a regional unconformity, after which foreland deposition resumed in Cenozoic time (Macellari et al., 1989; Arbe, 2002; Fosdick et al., 2015; Sickmann et al., 2018). Next, we describe the foreland evolution of the Austral, Magallanes, and Fuegian sectors of the Magallanes-Austral Basin according to the depositional successions that are preserved in each sector. The most detailed descriptions of stratigraphic evolution are available for the Austral (northern) and Magallanes (central) sectors of the Magallanes-Austral Basin, whereas the Fuegian (southern) sector is less constrained (Figs. 2B and 4).

Austral Sector (48.5°S to 50.2°S)

Jurassic extension in the Austral (northern) sector of the Magallanes-Austral Basin was only sufficient to attenuate Paleozoic metasedimentary basement and did not result in emplacement of oceanic crust (Figs. 2C and 3D; Ramos et al., 1982). During subsequent compression, total shortening in the northern sector was 10% or less (Coutand et al., 1999). Sustained coarse clastic deposition in the northern sector first occurred ca. 115–110 Ma (Fig. 4; Ghiglione et al., 2015; Malkowski et al., 2015a). This initial foreland depositional succession in the Austral sector, hereafter referred to as A1, is recorded by a series of fluvial and shallow-marine units including the Rio Belgrano, Rio Tarde, Lago Viedma, and Puesto El Alamo formations (Figs. 2B and 4; Arbe, 2002). The southward transition to deep-marine environments occurs near 49.5°S, where the shallow-marine Lago Viedma Formation is interpreted to transition into the deep-marine Cerro Toro Formation (Fig. 4; Malkowski et al., 2017). Deposition of A1 occurred for 15–30 m.y. (Fig. 4; Albian–Cenomanian into the Santonian; Arbe, 2002; Malkowski et al., 2017).

The middle Austral depositional succession, A2, is marked by downlapping of deep-marine slope clinoforms onto the deep-marine deposits of A1 (Fig. 4B; Macellari et al., 1989; Arbe, 2002). The transition to A2 in the northern, shallow-marine portion of the Magallanes-Austral Basin is poorly constrained stratigraphically (e.g., Sickmann et al., 2018). The A2 succession prograded approximately southward, recorded by the Santonian to Campanian Alta Vista Formation (deep marine), the Campanian La Anita Formation (shallow marine), and the Campanian to Maastrichtian (?) Cerro Fortaleza Formation (terrestrial; Fig. 4; Macellari et al., 1989; Arbe, 2002; Sickmann et al., 2018).

The transition into the final Austral succession examined in this study, the terrestrial to shallow-marine A3, is poorly constrained. Previous workers have suggested the presence of two unconformities in this succession, one between the Cerro Fortaleza Formation and the overlying La Irene Sandstone (braided fluvial) and Chorillo Formation (terrestrial to shallow marine) and a second between the La Irene Sandstone–Chorillo Formation and the tidally influenced Man Aike Formation (Fig. 4A; Macellari et al., 1989; Arbe, 2002). Recent geologic mapping of this area suggests that the exact locations of unconformities in the section are not resolvable with presently available data sets (Sickmann et al., 2018). For the purposes of this study, we considered everything above the Cerro Fortaleza Formation to be part of succession A3 (Figs. 2B and 4). This included strata that are likely Upper Maastrichtian to Eocene (Sickmann et al., 2018).

Magallanes Sector (50.2°S to 52.5°S)

Extension in the Magallanes (central) sector was higher in magnitude than in the Austral sector and was sufficient to generate partially developed oceanic crust (Figs. 2C and 3D; Stern and de Wit, 2003; Calderón et al., 2007). During Late Cretaceous compression, total shortening in the Magallanes sector of the basin was likely twice that of the Austral sector (20%; Fosdick et al., 2011). The first record of coarse clastic sedimentation in this central sector occurred at ca. 100 Ma (Fig. 4A; Fosdick et al., 2011) but does not appear to have been sustained until ca. 93 Ma (Fig. 4A; Fildani et al., 2003). The initiation of sustained coarse clastic deposition in the Magallanes sector marks the Cenomanian base for succession M1. M1 includes deep-marine deposits of Cenomanian to earliest Campanian age (Fig. 4A).

The initial deposits of M1 are the deep-marine turbidite lobes of the Cenomanian to Turonian Punta Barrosa Formation (Figs. 2B and 4; Wilson, 1991; Fildani et al., 2003). The Punta Barrosa Formation is overlain by the Coniacian to Santonian Cerro Toro Formation (Fig. 4), which is dominated by deep-marine shale and includes the Lago Sofia Conglomerate member, an assemblage of conglomeratic deep-marine channels (Arbe and Hechem, 1984; Hubbard et al., 2008; Bernhardt et al., 2012; Malkowski et al., 2018). The channels are oriented north-south along the axis of the basin, with at least three tributaries that enter the channel belt from the west (Fig. 4C). The northernmost tributary occurs at 50.6°S (Fig. 4C; Arbe and Hechem, 1984; Malkowski et al., 2018).

The transition to the middle Magallanes succession, M2, is marked by deposition of the Campanian to Maastrichtian Tres Pasos and Dorotea Formations, which record rapid, southward progradation of deep-marine slope and delta clinoforms (Fig. 4; Romans et al., 2010; Schwartz and Graham, 2015; Schwartz et al., 2016; Daniels et al., 2017). The Tres Pasos Formation preserves a continental-scale, failure-dominated (mass transport deposits) and channelized deep-marine slope, while the overlying Dorotea Formation preserves the genetically related deltaic topsets of the slope (Fig. 4B; Shultz et al., 2005; Schwartz and Graham, 2015). Together, they record relatively rapid (<10 m.y.) shoaling from deep-marine to terrestrial environments in the Magallanes sector during latest Cretaceous time (Romans et al., 2010; Schwartz et al., 2016).

The third Magallanes succession, M3, includes the Eocene Man Aike Formation (Fig. 4A). At Magallanes latitudes, there is a regional, low-angle unconformity between the Dorotea Formation (M2) and the Man Aike Formation (Fig. 4A; Schwartz and Graham, 2015). The Man Aike Formation records coarse-grained, shallow-marine deposition and is stratigraphically and facies-equivalent to the Man Aike Formation in the Austral sector (e.g., Malumián et al., 2000; Sickmann et al., 2018).


Data Sources and Acquisition

This study compiled four data sets: (1) U-Pb DZ analyses (Fig. 5; Table 1; Data Repository items DR1, DR2, DR31), (2) sandstone petrography (Fig. 6; GSA Data Repository items DR4 and DR5), (3) REE geochemistry (Fig. 7), and (4) paleocurrent measurements (Fig. 8; Data Repository item DR6). Each data set, with the exception of REE geochemistry, consists of a combination of new and published data. The data sets compiled for this study included data collected from the entire length of outcrop exposure in the Magallanes-Austral Basin between 47°S and 55°S, encompassing samples from the earliest foreland phase (Albian) through the lowest strata preserved above the basinwide Paleogene unconformity (ca. Eocene). The densest sampling for all methods is between 48.5°S and 52°S (Austral and Magallanes sectors), which contains the best characterized stratigraphy in the basin (e.g., Fig. 4).

U-Pb Detrital Zircon Analyses

Analytical Methods

The DZ data set includes data from 30 new samples from the Austral and Magallanes sectors and 87 published samples from all three sectors, comprising a total of 16,392 individual U-Pb analyses (Table 1). New U-Pb DZ samples collected for this study were analyzed at the University of California–Santa Cruz (UCSC) and the Arizona LaserChron Center (ALC) by laser ablation–inductively coupled plasma–mass spectrometry (LA-ICP-MS). Detrital zircon grains were extracted from 2 to 3 kg sandstone samples using standard hydrodynamic and density separation techniques (e.g., Schwartz et al., 2016). Samples were analyzed at the ALC either on a Nu (multicollector) or Element 2 (single-collector) mass spectrometer. LA-ICP-MS methods and data reduction for samples analyzed at the ALC followed the methods of Gehrels et al. (2008) and Gehrels and Pecha (2014). U-Pb results were filtered by: <10% uncertainty (1σ), minimum 206Pb/204Pb = 200, maximum discordance = 20% (Element 2) and = 30% (Nu), and maximum reverse discordance = 5% for ages older than 400 Ma. Samples analyzed at UCSC were analyzed on an Element XR (single-collector) mass spectrometer with a Photon Machines Analyte 193 nm ArF excimer laser hosted in the Department of Earth and Planetary Sciences (Sharman et al., 2013). LA-ICP-MS methods and data reduction for samples analyzed at UCSC followed the methods of Sharman et al. (2013; additional information is available in Data Repository item DR3). All new ages reported for this study are 206Pb/238U if younger than 1 Ga and 206Pb/207Pb if older than 1 Ga, and reported errors are 1σ. Most published DZ data in the compilation were also collected via LA-ICP-MS, although small numbers of data were collected using a sensitive high-resolution ion microprobe (SHRIMP; Table 1; e.g., Fildani et al., 2003; Romans et al., 2010). New samples presented in this study were collected from Cenomanian to Eocene strata of the northern Magallanes sector and southern Austral sector between 50.8°S and 49.4°S to fill in large gaps in sample coverage for these areas. Readers are referred to Table 1 for sample locations of all samples, and to Data Repository items DR1 and DR2 for raw U-Pb data associated with all new analyses presented in this study.


Plots of cumulative DZ age distributions for each sample, color coded by basin sector, sector average cumulative distributions, and sector average kernel density estimate (KDE) plots show that the three basin sectors display different general provenance trends (Fig. 5). These distributions show that samples from the southernmost Fuegian sector are dominated by relatively young ages (200–45 Ma) with a relatively narrow total range of ages. Magallanes sector samples contain a broad range of ages (1000–45 Ma) with few samples dominated by a single narrow age range (Fig. 5). Austral sector samples show a wide range of age distributions without an obvious trend (Fig. 5). Sector average KDEs show that the Austral sector is the only sector with a significant Early Jurassic age peak (ca. 180 Ma), and the Fuegian sector is almost devoid of ages older than ca. 150 Ma.

Sandstone Petrography

Analytical Methods

The sandstone petrography data set includes data from 32 new samples from the Austral and Magallanes sectors and 151 published samples from all three sectors of the basin (Fig. 6). An inventory of all published Magallanes-Austral Basin sandstone petrography data from six sources (Crane, 2004; Fildani and Hessler, 2005; Romans et al., 2010; McAtamney et al., 2011; Varela et al., 2013; Malkowski et al., 2017) shows that the only data bins common to all studies are total quartz, total feldspar, and total lithic fragments. Therefore, for the purpose of a basinwide provenance analysis, we are limited to the use of total quartz, feldspar, and lithic fragment (QtFL; sensu Dickinson et al., 1983) ternary discriminators. More detailed petrographic analyses are available from five studies (Crane, 2004; Fildani and Hessler, 2005; Romans et al., 2010; Varela et al., 2013; Malkowski et al., 2017) from the densely sampled basin transect between 48.5°S and 52°S (southern Austral sector through the Magallanes sector; area outlined in Fig. 4). However, the methods used in each of these studies for identifying specific lithic fragments differed slightly based on the goals of each study. Since original thin sections were not available to recount for this study, we employed only monocrystalline quartz, feldspar, and total lithic (QmFLt; Dickinson et al., 1983) ternary discriminators to examine the data to alleviate differences in specific lithic fragment identifications. New point count data for this study were acquired following the methodology of Dickinson (1970). At minimum, 400 grains were counted per sample to minimize compositional dependence on grain size (after Ingersoll et al., 1984), and samples with greater than 10% matrix (or pseudomatrix) were excluded from analysis. See Data Repository items DR4 and DR5 for a detailed discussion of pointing count methods and raw data.


Fuegian sector samples plot largely in the basement uplift and arc fields of Dickinson et al. (1983): see Figure 6. Austral and Magallanes samples plot across recycled orogenic and arc fields (Fig. 6; Dickinson et al., 1983). Samples plotted in QmFLt space show that Magallanes samples cluster slightly more tightly than Austral samples across recycled orogenic and arc fields (Fig. 6). Fuegian samples were excluded from the QmFLt plots because total lithic proportions were not provided for previously published samples.

REE Analysis

REE geochemistry data from 130 samples were adapted from Faúndez et al. (2002), Crane (2004), Fildani and Hessler (2005), Hervé et al. (2007), Romans et al. (2010), and McAtamney et al. (2011). All data were from bulk-rock analyses. We plotted all data as chondrite-normalized averages for each unit in question (Fig. 7). Those units included source-region rocks (e.g., ophiolites, batholith rocks, silicic volcanic rocks, and metamorphic complexes) and Rocas Verdes basin and Magallanes-Austral Basin sedimentary deposits. Rocas Verdes basin and Magallanes-Austral Basin sedimentary deposit average compositions included data from both shales and sandstones (Fig. 7).

Paleocurrent Measurements

The paleocurrent data set included 824 new individual paleocurrent measurements from the Magallanes and Austral sectors combined with 4556 published individual measurements from all three sectors of the basin (Fig. 8; Data Repository item DR6). New paleocurrent measurements were collected across the southern part of the Austral sector. New paleocurrents from shallow-marine and terrestrial deposits were measured from trough, tangential, and planar cross-stratification, whereas new paleocurrents from deep-marine deposits were measured from sole marks (flute and tool marks). The host formations and depositional successions that are assigned to paleocurrent measurements followed the stratigraphic framework and nomenclature of Sickmann et al. (2018).


The terrestrial deposits of the Campanian Cerro Fortaleza Formation (A2; Sickmann et al., 2018) display a wide range of paleocurrent directions with a slight southward dominance, whereas fluvial deposits of the Maastrichtian–Paleogene(?) La Irene Sandstone (Macellari et al., 1989) display a strong east-southeast paleocurrent direction (A3; Fig. 8). The shallow-marine deposits of the Cenomanian to Turonian (A1; Malkowski et al., 2017; Sickmann et al., 2018) Lago Viedma and Puesto El Alamo Formations display paleocurrents ranging from southwest to north, and the Campanian (A2; Macellari et al., 1989; Sickmann et al., 2018) La Anita Formation displays southwest to southeast paleocurrents (Fig. 8). The deep-marine Santonian to Campanian Alta Vista Formation (A2; Macellari et al., 1989; Arbe, 2002) displays strongly southward-directed paleocurrents (Fig. 8). The compilation of new and published paleocurrent measurements from the Magallanes and Austral sectors shows a strong south-southeast direction of transport for deep-marine deposits and a broad range of south to northeast directions of transport for shallow-marine deposits with a dominant trend toward the southeast (Fig. 8). See Data Repository item DR6 for all new paleocurrent measurements.

Detrital Zircon Interpretive Methods and Results

We applied and compared two different approaches for interpreting U-Pb DZ age distributions from samples throughout the Magallanes-Austral Basin: (1) a quantitative intersample statistical comparison of similarity/dissimilarity between U-Pb age distributions (e.g., Saylor and Sundell, 2016; Vermeesch, 2017), and (2) incorporation of knowledge (e.g., geochronology and geodynamic context) of source regions to manually bin U-Pb age data into age groups for comparison of relative proportions (e.g., Horton et al., 2010; Nie et al., 2012; Saylor et al., 2011, among others). For the quantitative comparisons of this study, we employed nonmetric multidimensional scaling (MDS) plotted using the Python-based software detritalPy (Sharman et al., 2018). Multidimensional scaling is a method that is used to simplify a matrix of intersample dissimilarity measures that would otherwise be too large or cumbersome to interpret into a representation of points in dimensional space (generally two or three dimensions for DZ data), such that the distance between any two samples (points) in unitless space is reflective of their relative dissimilarity (Fig. 9; Vermeesch, 2013). DetritalPy employs code from the Scikit-learn library (sklearn.manifold.mds) and calculates stress as the sum of the squared distance of the disparities and the distances for all constrained points. The exact coordinate values for the location of each sample in an MDS plot carry no specific meaning; only the relative distribution and distance between sample points indicate their unitless similarity/dissimilarity. The distribution of samples in this space is dependent on the samples that are included in the analysis and will change as samples or groups of samples are excluded or included. The more samples with similar distributions included in the plot, the more distinct, tightly clustered groups will form. See Data Repository item DR7 for a plot of all DZ samples employed in this study as uninterpreted stacked KDE plots.

For the purposes of this study, we employed nonmetric multidimensional scaling that plots points in unitless space based on a ranking of their relative dissimilarities using either the D value of the Kolmogorov-Smirnov (K-S) test or the V value of the Kuiper test (Figs. 9A and 9B). The D statistic that is generated from the K-S test is calculated based on the greatest point of deviation between the cumulative distributions of two samples, while the V statistic of the Kuiper test is generated based on an evaluation of the two greatest points of deviation in two cumulative distributions, one positive and one negative (Fig. 10). All Magallanes-Austral Basin samples plotted in MDS space using both dissimilarity metrics showed a tight cluster composed of approximately half of the samples, with the other half of the samples spread around the cluster (Figs. 9A and 9B).

For sample comparisons based on age bins generated with detailed knowledge of source regions, we generated a ternary discrimination scheme with poles based on three broad retroarc foreland source regions: (1) the magmatic arc, (2) the fold-and-thrust belt, and (3) peripheral sources (e.g., Fig. 1). We began to derive the ternary discrimination scheme with a conceptual examination of the predecessor tectonic history of the Magallanes-Austral Basin region (Fig. 11). Based on the predecessor history and previous studies of the structural evolution of the thrust belt (e.g., Hervé et al., 2003; Fosdick et al., 2011; Süssenberger et al., 2017), we determined that thrust belt sources included uplifted Tobífera and El Quemado Formations (young Chon Aike volcanics; 157–145 Ma) and Paleozoic metasedimentary basement (Eastern Andean Metamorphic Complex; older than 200 Ma; Figs. 11B and 11C), which can be tracked with input of their respective characteristic U-Pb ages. Peripheral sediment input can be traced with older Chon Aike ages (200–157 Ma) from the North Patagonia and Deseado Massifs (Figs. 11B and 11C), and arc ages include ages younger than 145 Ma. A composite KDE curve for all Magallanes-Austral Basin individual U-Pb DZ analyses showed that that these age bins can account for all age peaks observed in the detrital samples (Fig. 11C). It is important to note that these age bins only reflect the chronology of pre–Magallanes-Austral Basin–aged source regions and do not directly reflect zircon grains that were recycled during uplift and recycling of older foreland sediments during incorporation into the fold-and-thrust belt, as is common in foreland basin settings (DeCelles and Giles, 1996). However, based on thermochronological studies from the Magallanes sector, recycling of foreland basin strata does not appear to have been significant until latest Cretaceous (Maastrichtian) to Cenozoic time (Fosdick et al., 2015). Similar data for the degree of structural recycling of older foreland sediments in the Austral sector are unavailable, although based on its lower magnitude of shortening, recycling is likely not as pertinent in the Austral sector.

The binning of ages to generate this ternary discrimination scheme necessitated the delineation of exact boundaries between age classes, which resulted in two potential overlaps of source region signatures across bin boundaries. The first is the boundary between arc ages and young Chon Aike ages (a component of the fold-and-thrust belt age bin). The transition from extension-related magmatic activity to subsequent arc magmatism was gradual during the latest Jurassic (ca. 150 Ma) and into the earliest Cretaceous (ca. 140 Ma), making a discrete delineation of extension- versus compression-related magmatic ages difficult. We placed the boundary between these two source ages at 145 Ma in a natural valley in the basinwide U-Pb KDE (Fig. 11). This is reinforced by the Chon Aike geochronology work of Pankhurst et al. (2000) and Malkowski et al. (2015b), which suggested a minimum U-Pb age for Chon Aike volcanism of ca. 145 Ma, and the Patagonia Batholith geochronology of Bruce et al. (1991) and Hervé et al. (2007), which showed few U-Pb ages older than ca. 145 Ma (Fig. 11).

The second area of overlap is within the Chon Aike silicic volcanic ages. Jurassic extension gradually progressed from north to south between ca. 190 Ma and 147 Ma. Geochronologic investigations of northern Chon Aike volcanism in the North Patagonia Massif are sparse but have produced ages as old as ca. 190 Ma. We defined the older Chon Aike volcanics age bin from 200 to 157 Ma to account for analytical error and to reflect the transition from the peak at 182 Ma in the Magallanes-Austral Basin composite U-Pb age curve to a virtual absence of grains between 200 Ma and 220 Ma (Fig. 11C). This gap in ages older than 200 Ma suggests that the upper age bound on Chon Aike volcanics can be established with high confidence. The boundary between old Chon Aike volcanics (North Patagonia Massif, Deseado Massif; peripheral sources) and young Chon Aike volcanics (El Quemado and Tobífera Formations; thrust belt sources) at 157 Ma is based on the available geochronology of Pankhurst et al. (2000) and Malkowski et al. (2015b) and is similar to the delineation outlined in the DZ provenance analysis of Cenomanian Magallanes-Austral Basin stratigraphy by Malkowski et al. (2017).


Basinwide Multidimensional Scaling and DZ Ternary Results

Groups of the same samples are highlighted on all plots in Figure 9 (and in Table 1), including samples with >50% grain ages attributed to thrust belt sources (157–145 Ma; older than 200 Ma), samples with >50% grain ages attributed to arc sources (younger than 145 Ma; with hand-drawn contours up to >90% on MDS plots), and samples with >15% ages attributed to peripheral, older Chon Aike volcanic ages (200–157 Ma). The DZ ternary discrimination scheme is inherently based on these interpreted age bins. Therefore, interpretations based on the relative position of each sample as a unique point in ternary space assume that these age bins are useful in describing Magallanes-Austral Basin source regions. The relative locations of samples as points in MDS space are independent of this interpretation. The results of each (two different MDS plots, DZ ternary) can be compared accordingly.

Austral samples (blue squares in Fig. 9) spread out widely across all interpretive spaces, Magallanes samples (red circles in Fig. 9) tend to cluster tightly with scattered outliers, and Fuegian samples (yellow triangles in Fig. 9) group broadly together in each space, except for one or two outliers that plot within or near the tight Magallanes-dominated cluster. Although there is some overlap, approximately half of Fuegian samples plot in a distinct group away from Magallanes and Austral samples in MDS space (Fig. 9). It is evident from all three plots that parts of Fuegian sector sources were markedly different from Austral and Magallanes sector sources in their arc signatures. Sediment in the Fuegian sector was arc-dominated (evident in all three plots; Figs. 9A–9C), and the Fuegian arc was different from the Austral/Magallanes sector arc (evident in MDS plots; Figs. 9A and 9B). The Austral sector has a higher proportion of arc-dominated samples compared to the Magallanes sector. However, based on MDS results (Figs. 9A and 9B), the Austral and Magallanes sector arcs were similar in their geochronology.

Because Magallanes-Austral Basin DZ samples vary so widely in their modality and absolute values of their age spectra (e.g., multiple age peaks of different ages in KDEs; Fig. 10), they elucidate the interplay between geologic and statistical controls on the ways in which samples plot in MDS space. As an example, we can examine the tight clustering of thrust belt–dominated samples (>50% ages 157–145 Ma; older than 200 Ma) against the wide spread of arc-dominated samples (>50% ages younger than 145 Ma). Both the K-S and Kuiper tests are particularly sensitive to small deviations between large age peaks in U-Pb spectra (e.g., Maastrichtian APEN25 vs. Cenomanian Pb0104 vs. Cenomanian to Turonian CP09B; Fig. 10). Maastrichtian sample APEN25 from the Fuegian sector plots far from Cenomanian sample CP09B from the Austral sector and Cenomanian sample Pb0104 from the Magallanes sector in both MDS spaces, whereas close-in-age samples CP09B and Pb0104 plot closely together. DZ samples of sediment derived largely from magmatic arcs tend to be dominated by a small number of large age peaks that are close in age to the depositional age of the sediment (Cawood et al., 2012). The age peak that dominates Maastrichtian APEN25 is entirely younger than the depositional age of both CP09B and Pb0104 (Cenomanian to Turonian; Fig. 10). In evaluating arc-dominated DZ samples with K-S– or Kuiper-based MDS, the relative groupings of samples may be as much a function of the age of the sample as similarity of source regions.

Thrust belt–dominated samples such as MP96 (Magallanes sector; Cenomanian–Turonian), EQ16 (Turonian; Austral sector), 0822 (Fuegian sector; Campanian), and EP5 (Austral sector; Eocene) cluster tightly together, regardless of sample age or sector. This occurs in spite of the fact that, although similar, the individual age peaks and proportions in each of the spectra are different (e.g., KDEs on Fig. 10). This is a function of the fact that Magallanes-Austral Basin fold-and-thrust belt sources likely produced relatively consistent ages throughout the basin’s Cretaceous history, as it was always largely composed of Eastern Andean Metamorphic Complex and young Chon Aike volcanics (whereas results in a retroarc foreland basin with a more compositionally complex basement would likely be different, showing more change in thrust belt signatures through time). It is also a function of the fact that cumulative distributions that increase more gradually (i.e., are composed of a larger number of smaller age peaks) are less statistically dissimilar based on the K-S and Kuiper tests than are distributions with large single deviations (e.g., arc-dominated samples). Therefore, MDS results show useful information related to Magallanes-Austral Basin source regions and sediment dispersal systems, but they are also biased by their inherent dependence on imperfect statistical discriminators. For an in-depth discussion of the theory and more detailed statistical explanations of the controls and biases in MDS plots, see Wissink et al. (2018).

Ultimately, in the Magallanes-Austral Basin, with the currently available DZ data set, MDS results do not produce satisfactory groupings of samples such that we can be confident that we have captured the full range of basin-scale source region heterogeneity. Specifically, the wide scatter in arc-dominated samples suggests that to use K-S– or Kuiper-based MDS to interpret results, we need a much higher sampling density, both spatially and temporally, to discern clusters of similar arc-dominated dispersal systems. Additionally, the Fuegian sector samples likely complicate the MDS plot groupings because they are fewer in number (N = 16) than the Austral and Magallanes samples (N = 50 and N = 52, respectively) and are from a part of the basin that is less well understood. The binning of ages using a simplified geologic history and the ternary discrimination scheme derived from that binning are therefore useful simplifications of the data that are required to interpret basin-scale dispersal patterns.

Removing Fuegian Samples

The complications that arise from including samples from a less densely sampled, less well-understood part of the basin are illustrated by plotting Austral and Magallanes sector samples by themselves (Fig. 12). At the sector scale, samples can be broadly divided into three fields: arc-dominated, thrust belt–dominated, and peripherally influenced fields. Note that in order to define mutually exclusive fields, the peripheral Chon Aike pole is exaggerated by a factor of two on the ternary diagram to remove a slight overlap with arc-dominated samples (>50% ages that are younger than 145 Ma) that can be seen on the ternary plot in Figure 9C. We display cumulative density curves for all samples alongside MDS and ternary results to display the data without interpretation. Each data display carries a different level of interpretive bias. The cumulative distribution function plot carries no interpretation other than the basin sector from which the sample was taken, the MDS plot displays empirical statistical manipulation of the data coded again by sector, and the ternary diagram is explicitly based on interpreted provenance fields.

Delineation of the same three provenance fields in all three data displays demonstrates that: (1) given sufficiently distinct source region geochronology signatures and accurate source region knowledge, the results of MDS and interpretive binning broadly agree. (2) At a broad scale, Magallanes-Austral Basin provenance signatures can be described by relative contributions from three simplified source regions (arc, thrust belt, periphery). (3) Finally, when employing MDS, it is best to compare parts of the basin that are described by similar sample densities. With a comparable sample density from the Fuegian sector (N = ∼50), a similarly concise MDS plot may be possible at the basin scale if multisample groupings of similar Fuegian samples formed, as opposed to sporadic spread of a small number of individual, dissimilar samples.

Kuiper V versus K-S D Multidimensional Scaling

At the basin scale, K-S D value-based MDS results seem to group samples more consistently with respect to their retroarc source regions than do Kuiper V value-based MDS results (Figs. 9A and 9B). In particular, samples with >15% peripheral older Chon Aike ages scatter across the Vmax plot and are separated by the thrust belt– and arc-dominated groups (Figs. 9A and 9B). These same samples do not group tightly on the Dmax MDS plot, but they at least plot on the same side without other sample groupings in-between. Wissink et al. (2018) determined that K-S–based MDS produced the least consistent results based on comparisons to other metrics. In the case of the Magallanes-Austral Basin DZ data set, the K-S D value sensitivity to differences in large, single age peaks works as an advantage in distinguishing samples with significant proportions of the 200–157 Ma age range (e.g., samples LH157, ZC12, and RG164; Fig. 10), as these ages are rare-to-absent in most samples in the data set.

Basin- and Sector-Scale Sediment Dispersal Patterns

To interpret the basin- and sector-scale dispersal patterns of the Late Cretaceous to Eocene Magallanes-Austral Basin, we brought together the results of both geochronologic and compositional provenance data. Consistent with the results of DZ analyses, sandstone petrography results indicate that the Austral, Magallanes, and Fuegian sectors were sourced by varying proportions of fold-and-thrust belt, arc, and peripheral material (Figs. 5, 6, and 9). Next, we summarize interpretations for basin-scale dispersal patterns and their tectonic controls for each sector.

Fuegian Sector

Fuegian sector samples suggest arc dominance based on DZ signatures that contain ages almost exclusively younger than 150 Ma (Fig. 5); they plot near the arc pole on the Magallanes-Austral Basin DZ ternary plot, within the >50% arc ages MDS group (Fig. 9), and in the basement uplift and arc fields of Dickinson et al. (1983; Fig. 6 herein). We interpret the arc dominance in the Fuegian sector to be the result of a combination of two factors (Fig. 13). First, mafic and ultramafic rocks incorporated into the fold-and-thrust belt (Fig. 2C) would likely not produce significant or distinguishable zircon populations relative to the arc. Extension in this sector of the predecessor Rocas Verdes basin was sufficient to generate fully developed oceanic crust, likely leaving only a sliver of Paleozoic basement on the western margin of the rift (Fig. 2C; Stern and de Wit, 2003). Zahid and Barbeau (2010) concluded through heavy mineral analysis that an ophiolite was uplifted as a sediment source at Fuegian latitudes, and that Paleozoic basement was eventually incorporated into the fold-and-thrust belt but was not subaerially exposed until the Eocene. Second, because most of the material consumed to accommodate shortening at Fuegian latitudes was dense oceanic crust, the fold-and-thrust belt may not have been topographically high standing (Fig. 13; e.g., Gealey, 1977). It was previously interpreted that a large proportion of Cretaceous shortening that occurred in the Fuegian sector was accommodated by west- to south-dipping subduction and/or underthrusting of Rocas Verdes oceanic crust (Gealey, 1977; Klepeis et al., 2010). Ophiolite that was obducted into the fold-and-thrust belt would have acted as a dense topographic load (Fig. 13; e.g., Fosdick et al., 2014; Ghiglione et al., 2014). Because there is no evidence of >15% 200–157 Ma grains in Fuegian sector samples, we conclude that there was no significant input from peripheral sources before the Eocene (Fig. 9C). A more detailed evaluation of the temporal evolution of this sector of the basin beyond that already interpreted by Barbeau et al. (2009), Zahid and Barbeau (2010), and McAtamney et al. (2011) is not possible with available sample density and stratigraphic constraints.

Magallanes Sector

Magallanes samples appear to reflect mixed arc and thrust belt sources with a trend toward thrust belt dominance based on DZ signatures that show Late Cretaceous (ca. 80 Ma) through Triassic and Paleozoic (older than 200 Ma; Fig. 5) peaks and that plot between the arc and thrust belt poles of the DZ ternary diagram, slightly favoring the thrust belt pole and largely within the tight >50% thrust belt MDS group (Fig. 9). A similar trend for the Magallanes sector is reflected in petrography signatures, which cluster within/between arc and recycled orogenic sources of Dickinson et al. (1983; Fig. 6 herein). Jurassic extension in the Magallanes sector was sufficient to generate oceanic crust in the southern parts of the Magallanes sector (south of 51°S), but it left a large amount of silicic volcanic material and Paleozoic basement on the western margin of the basin (Fig. 2C). Although total shortening was up to 50% less in this sector than in the Fuegian sector, fold-and-thrust belt sources are abundantly evident in DZ and sandstone petrography data (Figs. 5, 6, and 9). We interpret this to be because the fold-and-thrust belt at Magallanes latitudes was constructed largely of Paleozoic basement (Eastern Andean Metamorphic Complex) and silicic volcanic material (e.g., Fosdick et al., 2011), as opposed to dense ophiolite, and therefore had a higher-standing topographic expression and higher zircon fertility (Fig. 13). This fold-and-thrust belt either structurally dammed input from arc sources or had a wide enough area of erodible exposure to dilute arc signatures. Input of contemporaneous DZ from the arc is recorded in this sector by the presence of DZ maximum depositional ages that match closely with interbedded ashes and biostratigraphy (e.g., Fildani et al., 2003; Bernhardt et al., 2012; Malkowski et al., 2015a, 2017). The Magallanes sector does not show any significant input from peripheral sources (Fig. 9C).

Because the Magallanes sector has excellent chronostratigraphic constraints and a high sample density, a detailed evaluation of its temporal evolution is possible. Plotting Magallanes samples in interpreted ternary space through time, there is little change between successions (Fig. 14A). All samples except one (MP97) plot on the line between arc and thrust belt ages, skewed slightly toward thrust belt dominance (Fig. 14A). Similarly, petrography signatures through the deep-marine M1 and deep-marine to shallow-marine M2 show little change (data are unavailable for shallow-marine M3). These results suggest that fold-and-thrust belt sources developed early in the history of the Magallanes sector and remained compositionally similar throughout, although as noted by Romans et al. (2010), the proportion of young Chon Aike (Tobífera Formation) ages increases up section. The occurrence of some arc-dominated samples throughout this central sector (particularly in its older stratigraphy) suggests that although fold-and-thrust belt sources were uplifted early and were prevalent at these latitudes, some drainages still tapped primarily arc material. Such arc-dominated drainages appear to be less common in the younger portions of the stratigraphy, perhaps as a function of continued fold-and-thrust belt development through the Late Cretaceous (Fig. 4). A more pronounced example of this trend (Figs. 15 and 16) is apparent in the northern Austral sector and is discussed in detail below.

Austral Sector

Austral samples plot broadly between arc, thrust belt, and peripheral sources in DZ ternary space and broadly across the MDS space (Fig. 9). The composite KDE for the Austral sector DZ signatures shows a significant Early Jurassic age peak (ca. 180 Ma; Fig. 5). Austral petrography signatures also plot widely across arc and recycled orogenic source regions (Fig. 6). Roughly one third of Austral sector DZ samples have arc-derived signatures, one third have thrust belt–derived signatures, and one-third have peripherally derived signatures (Figs. 9 and 11). We interpret that the fold-and-thrust belt in this part of the basin was not a topographic barrier to large volumes of arc detritus and only locally had sufficient erodible exposure to dilute arc signatures (Fig. 13). This is consistent with the interpretation that the Austral sector experienced the least amount of shortening during the foreland phase, exposing less fold-and-thrust belt material and capturing the flexural periphery of the basin (Fig. 2; Coutand et al., 1999). There must have also been areas at Magallanes and Fuegian latitudes that could similarly be considered to be on the periphery of flexural subsidence, such as the western edge of the Rio Chico–Dungeness Arch (Figs. 13 and 17). However, these portions of the basin fill are currently in the subsurface (e.g., Biddle et al., 1986). Fill patterns interpreted from subsurface seismic geometries suggest that the entire basin filled from both the west (from the fold-and-thrust belt and magmatic arc) and from the north and east (from peripheral sources; Biddle et al., 1986), but peripheral detritus is most abundant in the Austral outcrop belt due to its position on the periphery of flexural subsidence and the modern orientation of the outcrop belt (Fig. 17).

Within the Austral sector, there is a shift in peripheral sediment input signature through time. Figure 17 highlights all Magallanes-Austral Basin samples with >5% peripheral Chon Aike–age grains. Cenomanian to Santonian samples show a ratio of ∼75% North Patagonia Massif ages (190–178 Ma) to 25% Deseado Massif ages (172–157 Ma; Figs. 16 and 17). Samples of Santonian to Eocene age show the opposite proportions, with only 33% North Patagonia Massif ages to 67% Deseado Massif ages (Fig. 16). This up-section trend refines previous work on the connection between peripheral sediment sources and Magallanes-Austral Basin dispersal systems (e.g., Riccardi, 1987, 1988; Varela et al., 2013; Ghiglione et al., 2015; Malkowski et al., 2017). Riccardi (1987, 1988) suggested an Early Cretaceous marine connection between the Rocas Verdes basin and the San Jorge basin, and they also postulated that a fluvial connection between the two basins may have persisted until the Turonian, after which time uplift of the Deseado Massif permanently separated the two basins.

The Austral sector also shows a marked shift toward thrust belt input through time (Figs. 15 and 16). Shallow- to deep-marine succession A1 displays a wide array of dominant DZ signatures that pull toward the thrust belt pole through shallow-marine to terrestrial successions A2 and A3 (Fig. 15A). Petrography signatures show a similar shift from arc fields into the recycled orogenic field in the petrography ternary space of Dickinson et al. (1983) from A1 through to A3 (Fig. 15B). This transition is also well illustrated by a shift from a bimodal average DZ signature with primarily arc and peripheral Chon Aike ages in A1 through a decreased peripheral Chon Aike peak in A2 to a multimodal distribution with a pronounced young Chon Aike peak in A3 (Fig. 16). These trends likely reflect the evolution of a fold-and-thrust belt source and a southward shifting of peripherally derived dispersal systems through time (Figs. 12 and 15).

Implications for Retroarc Foreland Systems

The Magallanes-Austral Basin demonstrates the importance of the interplay among three factors in retroarc foreland basin sediment dispersal and composition (Fig. 13). These are: (1) degree of upper-plate shortening, (2) composition of shortened upper-plate material, and (3) position in the depocenter. The occurrence of arc-dominated provenance signatures across mechanically different parts of the Magallanes-Austral Basin illustrates the interplay between items 1 and 2 (Fig. 13). Arc domination can occur as a result of shortening that is insufficient to uplift thrust belt sources (early Austral sector; Fig. 13), uplift of thrust belt lithologies that are not conducive to producing certain provenance markers (Fuegian sector; Fig. 13), or the formation of a thrust belt composed of material that is too dense to have a significant topographic expression (Fuegian sector; Fig. 13). The spatial distribution of sediment derived from any given fundamental retroarc foreland source component (arc, thrust belt, periphery; Fig. 1) is strongly controlled by the distributions of accommodation and depositional environments across the basin. Although the Magallanes-Austral Basin filled simultaneously from the west, north, and east (Figs. 13 and 15; Biddle et al., 1986), peripheral sediment sources are only recognized in the north around the limit of flexural subsidence (Fig. 17). Accommodation in the eastern part of the basin appears to have acted as a buffer to peripheral sediment input to the western side of the basin (Figs. 13 and 17). This is also likely true in the reverse sense (Fig. 13). Provenance signatures from Upper Cretaceous strata that onlap the Rio Chico–Dungeness Arch (currently in the subsurface; Biddle et al., 1986) likely will not show significant thrust belt input. This accommodation buffer likely best applies to underfilled basin phases and those dominated by longitudinal as opposed to basin-scale transverse sediment dispersal.

Delineating the Spatial Boundaries of Sediment Dispersal Systems

Provenance differences between the Austral and Magallanes sectors have been previously suggested for shallow- to deep-marine A1 and deep-marine M1 based on DZ and sandstone petrography data (Malkowski et al., 2017) and for shallow-marine to terrestrial A2 and deep- to shallow-marine M2 based on sandstone petrography data (Macellari et al., 1989). The provenance differences between the two sectors are evident when comparing the spatial and stratigraphic distributions of arc-dominated, thrust belt–dominated, and peripherally influenced samples in the Austral and Magallanes sectors (Fig. 18). There is an apparent transition zone near the north shore of Lago Argentino between units that are arc and peripherally sourced in the north to units that are largely thrust belt–dominated to the south, with fewer arc-dominated samples in the south with respect to DZ signatures (Fig. 18). Differences are also evident in the petrographic signatures of both sectors (Fig. 6; Macellari et al., 1989). Within this transition zone, units with contrasting provenance signatures are interspersed stratigraphically (Fig. 18). Abrupt spatial variability in provenance trends suggests either that outcropping Austral and Magallanes sediment dispersal systems were in some way partitioned from each other (e.g., Lihou and Allen, 1996) or that down-system dilution by thrust belt sources decreased the arc-dominated signal between the two sectors (e.g., Saylor et al., 2011).

Neither end member, partitioning nor dilution, can be conclusively demonstrated as the dominant control for this provenance shift with the available data, and it is likely that both could play a role here. However, there is sufficient evidence to suggest that structural partitioning due to reactivated predecessor rift structures could have played some role (if not the dominant role) in this provenance signature shift. An east-west–oriented crustal discontinuity or “transfer zone” in the vicinity of Lago Argentino (Figs. 18 and 19) has been previously proposed to explain the west-to-east offset of the contacts between Rocas Verdes basin and Magallanes-Austral Basin stratigraphy (red dashed line; Fig. 18), and the contact between Cretaceous and Cenozoic Magallanes-Austral Basin stratigraphy (blue dashed line; Fig. 18), as well as several regional faults below the resolution of the Figure 18 map (e.g., Likerman et al., 2013; Ghiglione et al., 2014). Transfer zones commonly form in rift settings (e.g., Morley et al., 1990; Gawthorpe and Hurst, 1993) when the locus of extension shifts laterally, which must be accommodated by a zone of strike-slip deformation oriented parallel to the direction of extension (Fig. 19). In the case of the Rocas Verdes basin, which opened east-west and created a north-south–elongated basin (Figs. 3D and 3E), any required transfer zones would have been oriented approximately east-west. Once reactivated during regional compression, such inherited structural discontinuities could have served to either control the location of large regional drainage networks or structurally partition depocenters, as is observed in the Alpine foreland system in Europe (e.g., Lihou and Allen, 1996).

An abrupt lithostratigraphic transition occurs near the latitude of Lago Argentino, where at least three depositional elements of the Magallanes sector are not evident in outcrop north of the hypothesized transfer zone: (1) the Cenomanian to Turonian Punta Barrosa turbidite lobe deposits (Malkowski et al., 2015a; Figs. 4 and 18); (2) the Turonian to Campanian Cerro Toro turbidite conglomerate channels of the Lago Sofia Member, which are interpreted to be tributaries to an axial channel system that entered the basin from the west ∼40 km south of Lago Argentino (Arbe and Hechem, 1984; Malkowski et al., 2018; Figs. 4 and 18); and (3) the Campanian to Maastrichtian Dorotea Formation, which records a tide-influenced delta that shows strong east to southeast paleocurrents, thereby projecting headwaters toward the west-northwest, not the Austral sector (Schwartz and Graham, 2015; Figs. 4 and 18). The coincident location of all three lithostratigraphic units, located just south of the hypothesized transfer zone, and the evidence that the updip equivalents to the Lago Sofia conglomerate and the Dorotea delta project to the west-northwest, rather than outcropping Austral sector deposits, are suggestive of long-lived structural control on Magallanes-Austral Basin dispersal pathways.

Additionally, with or without structural partitioning, in projecting the paleocurrents of outcropping Austral shallow-marine successions, it is not clear that they would feed outcropping Magallanes sector deposits downdip (Fig. 18). For example, paleocurrent diagram 5 on Figure 18 from the La Anita Formation shows bimodal paleocurrents, perhaps reflecting longshore and offshore transport. The dominant mode is strongly southeast-directed, projecting the “downdip” continuation of the system into parts of the Magallanes-Austral Basin that are currently in the subsurface, east of the outcropping Magallanes sector. Similarly, paleocurrent diagrams 3, 4, and 7 show strong south to east-southeast transport in the terrestrial Cerro Fortaleza Formation and the La Irene Sandstone, none of which would project to outcrops in the Magallanes sector. For the purposes of this study, we prefer the interpretation that there was inherited structural influence at the boundary between the Austral and Magallanes sediment dispersal systems (Fig. 20). This structural discontinuity may have simply served to dictate regional drainage patterns and outlets (e.g., the La Sofia tributaries and the Dorotea delta) as shown on Figure 20, or it may have more profoundly influenced the partitioning of distinct depocenters throughout the Late Cretaceous history of the basin. Regardless of the impact of structural partitioning, paleocurrents alone suggest that the Upper Cretaceous shallow-marine deposits preserved in outcrop in the Austral sector are not direct updip equivalents of Upper Cretaceous deep-marine deposits in the Magallanes sector. A more thorough test of structural partitioning versus dilution end-member models requires a denser provenance data set from deep-marine deposits between Lago Viedma and Lago Argentino, and ideally deposits that are currently in the subsurface east of the Magallanes outcrop belt, where shallow-marine paleocurrents suggest that the outcropping Austral dispersal system would project downdip.

Depositional Evolution of the Magallanes and Austral Sectors

We present five paleogeographic reconstructions that are based on our current understanding of the tectonic history of the region, interpretations of the influence of tectonic activity on paleophysiography, and our new compilation of provenance data (Fig. 20).

95–90 Ma (A1/M1)

By 95 Ma, a subaerially exposed fold-and-thrust belt was present west of the foredeep, as suggested by young Chon Aike– and Eastern Andean Metamorphic Complex–aged grains in shallow- to deep-marine A1 and deep-marine M1 DZ signatures (Figs. 18 and 20). A1 signatures contain a significantly lower proportion of fold-and-thrust belt ages, suggesting that the subaerial exposure and/or topographic expression of the fold-and-thrust belt decreased to the north (Figs. 18 and 20). Austral sector depositional systems received sediment from peripheral sources to the north and east, while transverse sediment inputs were dominated by arc detritus (Fig. 20).

In the Magallanes sector, only metamorphic complex sources were exposed in the fold-thrust belt at this time; Tobífera/El Quemado volcanics had not yet been incorporated into the fold-and-thrust belt (Romans et al., 2010). The Magallanes sector of the basin rarely received sediment from the Austral sector, except perhaps in rare instances of longitudinal transport by long-runout deep-marine turbidite systems or possibly longshore drift, and only in the northernmost portion of the sector (e.g., sample MP97; Fig. 18). We omit the initial stages of the northern Austral sector due to the significant data gap and lack of stratigraphic control north of ∼49.5°S.

90–85 Ma (A1/M1)

The paleogeography of the Austral sector remained relatively unchanged through ca. 85 Ma (Fig. 20). Deposition of more than 1 km of shallow-marine deposits (Lago Viedma and Puesto El Alamo Formations) at 49.5°S suggests that sediment supply kept pace with subsidence to maintain depositional systems at or near sea level throughout the Cenomanian–Coniacian in most of the Austral sector (Arbe, 2002; Malkowski et al., 2017).

In the Magallanes sector, the Cerro Toro conglomerate channels initiated during this time, and the first appearance of young Chon Aike detritus is evident in provenance signatures (Romans et al., 2010; Bernhardt et al., 2012; Malkowski et al., 2018). Based on the presence of Tobífera/El Quemado detritus in the basin fill, these units were involved in thrusting (Fosdick et al., 2011; Ghiglione et al., 2014). The Magallanes dispersal systems apparently remained distinct from Austral dispersal systems throughout this time (Fig. 18 cross section).

85–80 Ma (A2/M1-M2)

Southward slope progradation began in the Austral sector by late Santonian/early Campanian time (Sickmann et al., 2018). Austral deposits no longer exhibited strong arc dominance compared to those in A1 (Figs. 15 and 16). This transition suggests that dispersal systems that tapped batholith sources during this time were fed through more thrust belt exposures between the arc and the foredeep (Fig. 20). We suggest that the abrupt progradation of sediment dispersal systems and the transition away from arc-dominated source signatures may have been controlled by thrusting events in the late Coniacian to Santonian (Fig. 4). Fosdick et al. (2011) interpreted the development of a Tobífera duplex structure during this time. This is admittedly speculative in that it relies on alignment of the timing of provenance changes, thrust belt activity, and progradation of depositional systems and is based on current structural interpretations and stratigraphic constraints. Peripheral sediment input during this time shifted from North Patagonia Massif–dominated to Deseado Massif–dominated input, suggesting a shift in drainage patterns north of the Magallanes-Austral Basin, perhaps due to capture by the San Jorge basin and uplift in the Deseado Massif (Figs. 15 and 20).

In the Magallanes sector, Cerro Toro deposits prograded southward (Bernhardt et al., 2012), but source regions remained relatively unchanged. Southward-prograding slope clinoforms began to appear in this sector at the very end of this time interval (ca. 83 Ma; Daniels et al., 2017; Sickmann et al., 2018). Based on paleocurrent, DZ, and petrography data, we interpret that the portions of these clinoforms that are observed in outcrop were likely not fed from outcropping Austral sector deposits (Fig. 20). Instead, we hypothesize that shallow-marine and terrestrial systems in the Austral sector fed deep-marine systems that are now present only in the subsurface to the east of the Magallanes outcrop belt.

80–70 Ma (A3/M2)

By the late Campanian to early Maastrichtian, Cenomanian foreland basin deposits were incorporated into the fold-and-thrust belt (Fig. 20; Fosdick et al., 2011). Provenance signatures in A3 suggest that dispersal systems that were previously dominated by arc sources shifted to largely draining the fold-and-thrust belt, but the timing remains relatively unconstrained. Paleocurrents indicate east-southeast–directed transport, suggesting that the Austral systems did not feed the Magallanes sector (Fig. 20). Interpretations of the conditions in the Austral sector during the late Campanian and through the Maastrichtian remain largely speculative because rocks of this age are poorly preserved (Sickmann et al., 2018).

Southward-prograding clinoforms reached the Magallanes sector by ca. 81 Ma (Daniels et al., 2017). However, the M2 slope-delta system was likely independent from the 85–80 Ma southward-prograding A2 slope. We interpret that a newly coalesced drainage network fed the outcropping Dorotea delta and Tres Pasos slope clinoform system (Fig. 20). We base this interpretation on (1) differences in DZ provenance signatures between A2 and M2, which show that A2 was largely arc dominated and peripherally influenced, and M2 was strongly thrust belt dominated; and (2) depositional ages, where the A2 system is 5–10 m.y. older than the upper M2 system, which contains Dorotea–Tres Pasos clinoforms (e.g., Daniels et al., 2017; Sickmann et al., 2018). The braided fluvial La Irene Sandstone of A3 is excluded as a candidate for the northward continuation of the shallow-marine part of the M2 system based on paleocurrent directions and petrographic signatures, which were interpreted by Macellari et al. (1989) and are reinforced by this study. The poorly understood Chorillo Formation remains a candidate for the northward continuation of M2, specifically the Dorotea delta, based on lithostratigraphic correlations (Fig. 20).

Ca. 45 Ma (A3/M3)

We omit a graphical representation of the basin configuration between 70 and 45 Ma, a time period that is represented by a regional Paleocene to middle Eocene unconformity that is present in both outcrop and the subsurface. Following the development of the regional unconformity, deposition resumed at ca. 45 Ma with deposition of the Man Aike Formation (Fig. 20; Macellari et al., 1989), a very coarse-grained, shallow-marine unit that was affected by strong tides. During the Eocene, older foreland deposits were uplifted into the fold-and-thrust belt and were recycled into the Cenozoic basin (Fosdick et al., 2014). We suggest that frontal thrust structures of the fold-and-thrust belt may have sufficiently propagated eastward by middle Eocene time to place the Man Aike depositional systems in a wedge-top setting. This is supported by regional geologic mapping (Malumián, 2000) and balanced structural cross sections (Fosdick et al., 2011) that indicate the possibility of blind thrusts beneath the Magallanes sector by Eocene time. A wedge-top structural setting has previously been implied by several authors based on the apparent angularity between middle Eocene deposits and underlying Cretaceous strata (e.g., Malumián et al., 2000; Fosdick et al., 2011; Schwartz et al., 2012).


The Magallanes-Austral Basin is a data-rich retroarc foreland basin with unique source region fingerprints during its Cretaceous evolution that lends itself well to evaluating comparative approaches for analyzing large provenance data sets. An examination of the large (N = 117 samples; n = 16,392 individual ages) DZ geochronology provenance data set available from the basin provides an example of the necessity for simplifying interpretive schemes, ideally employing both detailed source region knowledge and independent quantitative metrics such as MDS. With respect to incorporating source region knowledge, the results of this study reinforce the utility of examining provenance signatures in retroarc foreland basins with respect to relative contributions of three broad tectonic source regions, i.e., the magmatic arc, the fold-and-thrust belt, and uplifted sources from the periphery of the basin. Interpretive schemes incorporating this source region interpretation and MDS results yield similar sample groupings. Multidimensional scaling results themselves are strongly influenced by the statistical discriminator employed, the modality of the age spectra being compared, and sample density. Small differences in the peak ages of arc-dominated, more unimodal, samples are exaggerated as compared to differences between thrust belt–dominated, more multimodal samples using both K-S D values and Kuiper V values. Additionally, the inclusion of samples in MDS analysis from a less densely sampled, less well-understood portion of the basin skews sample groupings, obfuscating trends in better-understood basin sectors. Multidimensional scaling and source region–based DZ sample groupings and sandstone petrography results are broadly compatible. Basin sectors with arc-dominated DZ signatures show arc-dominated petrographic signatures; basin sectors with mixed arc and thrust belt dominance in DZ signatures show mixed petrographic signatures between magmatic arc and recycled orogen sources.

The arc versus thrust belt versus peripheral dominance of provenance signatures at any point in the basin can be explained by controls inherited from a predecessor rift phase. Southernmost Fuegian sector samples represent arc-dominated locations due to a history of extensive rifting that left little Paleozoic basement on the western margin to be uplifted during compression. Material in the southernmost fold-and-thrust belt was dense and low riding and did not produce zircon or framework silicate grains that could dilute arc DZ and petrography signatures. The outcropping portion of this sector was too far removed from peripheral sources to receive their detritus. The central Magallanes sector samples are largely thrust belt–dominated due to less extensive rifting in this sector that left Paleozoic basement and Jurassic silicic volcanic material on the western margin of the basin. Compression was subsequently sufficient to uplift fold-and-thrust belt material as a topographic barrier that could dam arc input or that had sufficient exposure and framework silicate and zircon fertility to dilute arc source signatures. The sector appears to have only received occasional pulses of peripherally derived detritus. Northern Austral sector samples display the most cosmopolitan provenance signatures as a function of the proximity of this sector to all three source regions. Compression in this sector appears to have been insufficient to uplift abundant fold-and-thrust belt material to dam or dilute arc signatures until Eocene time.

Finally, there is evidence that inherited east-west–oriented transfer zones from the rift phase may have played a role in partitioning sediment pathways. Provenance signatures between the Austral and Magallanes sectors change abruptly at their boundaries. Paleocurrent indicators suggest that outcropping Austral dispersal systems project east of outcropping Magallanes dispersal systems, placing the downdip equivalents of Austral dispersal systems in the subsurface. We suggest that this may have been controlled by a long-lived structural discontinuity in the vicinity of Lago Argentino (50°S–50.5°S) that was inherited from a rift-phase transfer structure. This discontinuity offsets stratigraphic contacts and structural features and may have played a role in dictating the location of regional drainage networks, dispersal pathways, or even localized depocenters.


Funding for this work was provided by the Stanford McGee Levorsen Grant and the Stanford Project on Deep Water Depositional Systems (SPODDS). The Arizona LaserChron Center is funded in part by the National Science Foundation Instrumentation and Facilities Division (NSF grant EAR-1338583). Field assistance was provided by Glenn Sharman, Aaron Reimchen, Devon Orme, and Jared Gooley. Discussions with Jared Gooley, Brian Horton, and Kristina Butler significantly improved this manuscript. Reviews by Ben Daniels, Joel Saylor, Ryan Leary, an anonymous reviewer, and Science Editor Damian Nance also significantly improved this work.

1GSA Data Repository Item 2019290, DR1: U-Pb detrital zircon data from Arizona Laserchron; DR2: U-Pb detrital zircon data from University of California Santa Cruz; DR3: Notes on analytical methods for University of California Santa Cruz DZ analyses; DR4: Petrography data; DR5: Detailed petrography methods; DR6: Paleocurrent data; DR7: All DZ samples used in this study plotted as individual KDEs, is available at http://www.geosociety.org/datarepository/2019, or on request from editing@geosociety.org.
Gold Open Access: This paper is published under the terms of the CC-BY-NC license.