Abstract

Recognizing detrital contributions from sediment source regions is fundamental to provenance studies of active and ancient orogenic settings. Detrital zircon U-Pb geochronology of unconsolidated sands from modern rivers that have source catchments with contrasting bedrock signatures provides insight into the fidelity of U-Pb age signatures in discriminating tectonic provenance and downstream propagation of environmental signals. We present 1705 new detrital zircon U-Pb ages for 15 samples of unconsolidated river sands from 12 modern rivers over a large spatial extent of Ecuador (∼1°N–5°S and ∼79°–77°W).

Results show distinctive U-Pb age distributions with characteristic zircon age populations for various tectonic provinces along the Andean convergent margin, including the forearc, magmatic arc, and internal (hinterland) and external (foreland) segments of the fold-thrust belt. (1) Forearc and magmatic arc (Western Cordillera) river sands are characterized by Neogene–Quaternary age populations from magmatic sources. (2) Rivers in the hinterland (Eastern Cordillera) segment of the Andean fold-thrust belt have substantial populations of Proterozoic and Paleozoic ages, representing upper Paleozoic–Mesozoic sedimentary and metasedimentary rocks of ultimate cratonic origin. (3) River sands in the frontal fold-thrust belt (Subandean Zone to Oriente Basin) show distinctive bimodal Jurassic age populations, a secondary Triassic population, and subordinate Early Cretaceous ages representative of Mesozoic plutonic and metamorphic bedrock.

Detrital zircon U-Pb results from a single regional watershed (Rio Pastaza) spanning the magmatic arc to foreland basin show drastic downstream variations, including the downstream loss of magmatic arc and hinterland signatures and abrupt introduction and dominance of selected sources within the fold-thrust belt. Disproportionate contributions from Mesozoic crystalline metamorphic rocks, which form high-elevation, high-relief areas subject to focused precipitation and active tectonic deformation, are likely the product of focused erosion and high volumes of local sediment input from the frontal fold-thrust belt, leading to dilution of upstream signatures from the hinterland and magmatic arc.

INTRODUCTION

Detrital zircon U-Pb geochronology has become a fundamental provenance tool providing critical records of mountain building, large-scale drainage evolution related to tectonic and climatic drivers, and source-to-sink evolution of depositional systems and sedimentary basins (e.g., Dickinson and Gehrels, 2003, 2008; Weislogel et al., 2006; Cawood et al., 2012; Laskowski et al., 2013; Blum and Pecha, 2014; Sharman et al., 2015; Horton, 2018; Ducea et al., 2018; Wang et al., 2018). Discrimination of competing sedimentary source regions and terranes is essential to detrital zircon provenance studies (e.g., DeCelles et al., 2000; Horton et al., 2010a, 2010b, 2015; Gehrels, 2014; Saylor and Sundell, 2016; Sickmann et al., 2018). However, despite the utility of detrital zircon U-Pb data sets, there are important limitations involved in interpretation of such data. At a fundamental level, do detrital zircon data accurately reflect the distribution of source units exposed in their river catchments? The commonly implicit assumption that detrital contributions are proportional to source unit exposure area within a catchment (Saylor et al., 2013) can be compromised in situations where there are disproportionate contributions from different bedrock units, due to: (1) heterogeneous zircon fertility (Moecher and Samson, 2006; Dickinson, 2008; Spencer et al., 2018); (2) uneven bedrock erodibility (Capaldi et al., 2017); (3) contrasting lithologic breakdown by weathering, erosion, and climate (Nesbitt et al., 1996; Amidon et al., 2005); and (4) varied mechanical durability during transport (e.g., Garzanti et al., 2013).

In this study, we present a large data set of 1705 new zircon U-Pb ages from unconsolidated sand samples collected from modern rivers across Ecuador in order to characterize detrital zircon U-Pb signatures of the different tectonic provinces and explore downstream variations in detrital signatures within a single large river system. Sampled rivers span the coastal forearc to retroarc foreland basin of Ecuador (∼80°–77°W) and have Andean mountainous catchment areas ranging from small (∼200 km2) to large (∼13,500 km2), covering a cumulative surface area of ∼50,000 km2 (Figs. 1 and 2). First, the results show that U-Pb age distributions of modern rivers in Ecuador clearly discriminate the major tectonic provinces (forearc, magmatic arc, and hinterland and foreland segments of the fold-thrust belt). Second, U-Pb age distributions along the large Rio Pastaza catchment (∼13,500 km2) show a downstream loss of important age populations, likely due to pronounced input from singular sources due to focused erosion driven by high precipitation and/or active tectonic processes.

GEOLOGIC SETTING

Ecuador is divided into various tectonic provinces with distinctive rock units of different age and tectonic history. These provinces are parallel to the regional tectonic strike of the northern Andes, with roughly north-south orientations that broadly coincide with major topographic boundaries (Figs. 1 and 2). Here we classify these tectonic provinces, according to their relative position, as different components of a Cordilleran-style orogen along an ocean-continent convergent plate boundary. We recognize that the precise boundaries between successive forearc, magmatic arc, retroarc fold-thrust belt, and foreland basin provinces are likely subject to internal complexities wherein these boundaries may deviate such a simplified framework at finer spatial scales. From west to east, the major tectonic provinces include the coastal forearc region, the magmatic arc (Western Cordillera), the internal hinterland segment of the fold-thrust belt (Eastern Cordillera), the external frontal segment of the fold-thrust belt (Subandean Zone), and the Oriente foreland basin.

Forearc and Magmatic Arc (Western Cordillera)

The coastal forearc and magmatic arc (Western Cordillera) region (Figs. 1 and 2) is composed of allochthonous mafic and ultramafic Upper Cretaceous rocks of oceanic plateau and/or intraoceanic island arc origin that accreted to the margin ca. 75 Ma, overlain by Upper Cretaceous–Paleogene sedimentary rocks and Paleogene–Quaternary magmatic rocks of both volcanic and plutonic origin (Kerr et al., 2002; Jaillard et al., 2004; Vallejo et al., 2009; Schütte et al., 2010; Spikings et al., 2015). Paleocene–Miocene (66–5.3 Ma) magmatic rocks include dominantly granodioritic intrusions and andesitic to rhyolitic lava flows, ignimbrites, and tuffs (Schütte et al., 2010, and references therein). Pliocene–Quaternary (5.3–0 Ma) magmatism is manifest as a volcanic chain that persists from Ecuador (∼2.5°S) northward into Colombia. In Ecuador alone, >80 recognized volcanoes (mostly stratovolcanoes) with Pliocene–Quaternary activity have produced massive amounts of volcanic material ranging from andesitic to rhyolitic lava flows, ignimbrites, and tuffs (Hall and Wood, 1985; Barberi et al., 1988; Hall et al., 2008; Bernard and Andrade, 2011; Jackson et al., 2019).

Hinterland Fold-Thrust Belt (Eastern Cordillera)

The Eastern Cordillera represents the internal hinterland segment of the fold-thrust belt (Figs. 1 and 2) composed of sublinear north-trending belts of variably metamorphosed and variably deformed upper Paleozoic–Lower Cretaceous sedimentary, igneous, and volcanic rocks. In general, these include: (1) upper Paleozoic–Lower Cretaceous low- to intermediate-grade metasedimentary rocks formed from pelitic protoliths ultimately derived from cratonic sources; (2) Triassic–Jurassic low- to high-grade metaigneous rocks and Triassic S-type granitoids; and (3) Jurassic–Lower Cretaceous volcanic and low-grade metavolcanic rocks (Aspden and Litherland, 1992; Aspden et al., 1992; Pratt et al., 2005; Ruiz et al., 2007; Spikings et al., 2015). Mesozoic metamorphism is linked to a complex, spatially variable history of sedimentary burial, crustal shortening, and terrane collision (Litherland et al., 1994; Kerr et al., 2002). Finally, several isolated Quaternary volcanic centers have developed on top of metamorphic basement in the Eastern Cordillera (Hall et al., 2008), but represent a limited cumulative area (<5%–10%) of this tectonic province.

Frontal Fold-Thrust Belt (Subandean Zone to Oriente Basin)

The Subandean Zone to Oriente Basin region represents the outer, frontal segment of the fold-thrust belt and associated flexural foreland basin (Figs. 1 and 2). The Oriente Basin is part of the larger hydrocarbon-bearing Subandean retroarc foreland basin system and is continuous southward with the Peruvian Marañón Basin and northward with the Colombian Putumayo Basin (Dashwood and Abbotts, 1990; Higley, 2001). The Cretaceous–Cenozoic sedimentary succession of the Oriente Basin is flat-lying to the east and locally deformed in the west by active fold-thrust structures of the frontal Subandean wedge-top depozone (Bès de Berc et al., 2005; Baby et al., 2013).

Mesozoic basin-fill strata have cratonic derivation and include sandstones, siltstones, organic-rich shales, and subordinate limestones (Tschopp, 1953; Dashwood and Abbotts, 1990; Barragán et al., 2004; Ruiz et al., 2007; Horton et al., 2010a; Vallejo et al., 2017). Cenozoic strata include coarser clastic material (sandstones, siltstones, and conglomerates) derived from Andean sources to the west (Christophoul et al., 2002; Jaillard et al., 1997; Martin-Gombojav and Winkler, 2008; Horton, 2018; Gutiérrez et al., 2019). The Cretaceous–Cenozoic succession unconformably overlies Jurassic magmatic arc rocks, which are exposed exclusively in the Subandean Zone as non-metamorphosed granites, monzogranites, and granodiorites (Litherland et al., 1994; Romeuf et al., 1995; Spikings et al., 2001; Pratt et al., 2005; Spikings et al., 2015).

METHODS

Detrital Zircon Geochronology

Uranium-lead (U-Pb) geochronological analyses of detrital zircon grains were performed on 15 unconsolidated sand samples collected from 12 rivers in Ecuador (Figs. 1 and 2; Supplemental Table S11). To reduce potential intersample bias that may result from hydrodynamic sorting and grain-size dependence (e.g., Malusà et al., 2016; Ibañez-Mejia et al., 2018), modern fluvial bar deposits of medium sand size were routinely targeted for sampling. Individual zircon crystals were density separated using a Wilfley water table and Frantz magnetic separator, followed by heavy liquid treatment using methylene iodide (density 3.33 g/cm3). Samples containing residual pyrite were washed in nitric acid. The final heavy fractions were then poured onto tape (rather than picked by hand), mounted in epoxy resin, polished, and imaged by backscatter electron imaging. Individual zircon grains were then chosen at random for analysis. U-Pb analyses were performed at the Arizona LaserChron Center (Tucson, Arizona, USA) by laser ablation inductively coupled plasma mass spectrometry on an Element2 HR ICPMS (Gehrels et al., 2006, 2008; Gehrels, 2014). Zircon standards used were Sri Lanka, FC-1, and R33. Analyses exhibiting >20% discordance, >5% reverse discordance, or >10% internal uncertainty were removed from further consideration. The reported ages represent 206Pb/238U ages for zircons <900 Ma and 206Pb/207Pb ages for zircons >900 Ma. For very young zircons (<2 Ma), a U-Th disequilibrium correction was applied to 206Pb/238U ages following the approach of Crowley et al. (2007), assuming a magma Th/U value of 2.3. Uncertainties are reported at the 1σ level. Probability distribution plots for data visualization were created using detritalPy software (Sharman et al., 2018). Grain dimensions were acquired for a subset of analyzed zircons, with measurements of long and short axes, and confirm that diagnostic age populations are not restricted to particular grain sizes.

Major Age Populations

Major age populations in the northern Andes have been identified from geochronological studies of crystalline rock units and detrital zircon U-Pb results from sedimentary units defining the Andean belt and adjacent sectors of northern South America (e.g., Teixeira et al., 1989; Litherland et al., 1994; Cordani and Teixeira, 2007; Cardona et al., 2010; Horton et al., 2010b, 2015; Saylor et al., 2013). Considerations of both Andean rock units and cratonic signatures are important due to the prevalence of upper Paleozoic–Mesozoic sedimentary rocks units composed of clastic material derived from the Amazonian craton (e.g., Dashwood and Abbotts, 1990; Cooper et al., 1995; Horton et al., 2010b; Vallejo et al., 2017; Gutiérrez et al., 2019).

In this study, major age populations (Fig. 3) are defined on the basis of significant age groups recognized in Cretaceous–Cenozoic basin fill in forearc and retroarc basins (Witt et al., 2017; Gutiérrez et al., 2019). Undisputed exposures of Precambrian basement rocks are restricted to cratonic provinces farther east and do not exist in the study area; thus Precambrian ages present in studied samples must have been originally derived from the craton and were subsequently incorporated into younger rock units. Basement age groups may be divided into Andean (<650 Ma) and cratonic (>900 Ma) affinities, recognizing that older Precambrian ages are common in upper Paleozoic–Mesozoic sedimentary and metasedimentary rocks of the northern Andes. Andean populations are dominated by Mesozoic–Cenozoic (250–0 Ma) ages common in magmatic units with subordinate Paleozoic (450–250 Ma), Famatinian orogeny (550–450 Ma), and late Neoproterozoic (650–550 Ma) ages of crystalline basement origin (Chew et al., 2007, 2008; Cardona et al., 2010; Gutiérrez et al., 2019). Ages >900 Ma are of cratonic affinity, with an original derivation from Precambrian cratonic provinces to the east, including: >2300 Ma (Central Amazonian); 2250–2150 Ma (Maroni-Itacaiúnas); 2150–1800 Ma (Ventuari-Tapajós); 1800–1550 Ma (Rio Negro–Jurena); 1550–1300 Ma (Rondonia–San Ignacio); and 1300–900 Ma (Sunsás) ages (Fig. 1, lower right inset) (Teixeira et al., 1989; Cordani et al., 2005; Cordani and Teixeira, 2007; Chew et al., 2007, Santos et al., 2008; Bahlburg et al., 2009; Bettencourt et al., 2010; Cardona et al., 2010).

RESULTS

Detrital zircon U-Pb geochronological results for 15 river sand samples from 12 modern Ecuadorian rivers at 1°N–5°S are presented with a series of diagrams defining the age distributions (probability density functions) in the context of the aforementioned diagnostic age populations of the northern Andes. A composite age distribution (Fig. 3) is defined by 2057 zircon U-Pb analyses, the sum of results from our 15 river samples (1705 new analyses) and 352 published analyses of four additional river samples (Pepper et al., 2016). The composite results reveal diverse age populations spanning the Precambrian through the Cenozoic, with Andean ages constituting 79% of all analyses (65% Cenozoic–Mesozoic and 14% Andean basement), and cratonic ages (>900 Ma) constituting the remaining 21% (Fig. 3; Table S1 [footnote 1]).

Among Cenozoic ages, 51% are Pliocene–Quaternary, 12% are Miocene, and 36% are Paleogene (dominantly Eocene–Oligocene). Among Mesozoic ages, over half (54%) are Jurassic, and the remainder are equal contributions of Cretaceous and Triassic (each 23%). Older Precambrian ages (>900 Ma) decrease in proportion with increasing age, consistent with decreasing proximity to the oldest cratonic core in the distal easternmost regions (Fig. 1). Among all cratonic ages in the cumulative data set (n = 423 analyses), 50% are 1300–900 Ma (Sunsás), 18% are 1550–1300 Ma (Rondonia–San Ignacio), 13% are 1800–1550 Ma (Rio Negro–Jurena), 13% are 2150–1800 Ma (Ventuari-Tapajós), and only 6% are >2150 Ma. The age distributions are presented for the different tectonic provinces according to sample location, river name, and catchment size (Table S2 [footnote 1]).

Forearc and Magmatic Arc (Western Cordillera)

Six new samples of rivers with individual catchment areas ranging from 197 to 3555 km2 in the forearc and magmatic arc and three published samples represent a cumulative catchment area of 18,222 km2 (Figs. 1 and 2). New geochronological results (n = 653 analyses) together with results from three published samples (n = 283 analyses; Pepper et al., 2016) show an overall dominance of Cenozoic, primarily Neogene (23–0 Ma) and late Paleogene (dominantly ca. 40–23 Ma) ages, with limited pre-Cenozoic ages (Fig. 4). Jurassic ages are virtually absent (<1% of analyses), and Paleozoic and older ages are also nearly absent in seven of the nine samples (Fig. 4).

Hinterland Fold-Thrust Belt (Eastern Cordillera)

Six new samples from rivers with individual catchment areas ranging from 341 to 8371 km2 from the Eastern Cordillera hinterland segment of the fold-thrust belt represent a cumulative catchment area of 21,671 km2 (Figs. 1 and 2). Geochronological results (n = 578 analyses) exhibit overall cosmopolitan age distributions dominated by Paleozoic–Precambrian ages (constituting 61% of analyses) (Fig. 5). Cenozoic ages are generally restricted to the Paleogene (dominantly ca. 50–30 Ma), as Neogene ages (23–0 Ma) are nearly absent, constituting only 2% of analyses. Mesozoic ages are principally Triassic (200–250 Ma), with minor Jurassic–Cretaceous (200–66 Ma) ages. Paleozoic (450–250 Ma), Famatinian (550–450 Ma), and Neoproterozoic (650–550 Ma) ages are present in five of six samples in the Eastern Cordillera province (all but sample 11: EC1535). Older Precambrian age populations are present in all samples of this region, with the most prevalent, the Sunsás group (1300–900 Ma), appearing in all samples, followed by Rondonia–San Ignació (1550–1300 Ma), Rio Negro–Jurena (1800–1550 Ma), and Ventuari-Tapajós (2150–1800 Ma).

Frontal Fold-Thrust Belt (Subandean Zone to Oriente Basin)

Three sand samples from rivers with individual catchment areas ranging from 4069 to 13,473 km2 in the Subandean Zone represent a cumulative catchment area of 19,351 km2 (Figs. 1 and 2). The new geochronological results (n = 474 analyses) together with those from one published sample (n = 69 analyses; Pepper et al., 2016) show age populations generally restricted to the Mesozoic, and Jurassic ages in particular, which constitute 59% of all analyses for this region and exhibit a clear bimodal distribution with age peaks corresponding to the Early Jurassic (29% of analyses; 200–174 Ma) and Late Jurassic (29% of analyses; 165–145 Ma) (Fig. 6). Secondary populations are Triassic (250–200 Ma) followed by limited Early Cretaceous (145–110 Ma) ages. Late Cretaceous and Cenozoic ages are virtually absent, as are Paleozoic and older populations, except in one sample (sample IV: RTENEC) where limited pre-Mesozoic ages are present.

DISCRIMINATION OF TECTONIC PROVINCES

Comparison of cumulative U-Pb age distributions by tectonic province reveals several diagnostic populations (Fig. 7). (1) The coastal forearc to magmatic arc (Western Cordillera) is characterized by Pliocene–Quaternary (5.3–0 Ma) and Paleogene (dominantly ca. 40–23 Ma) age populations. (2) The hinterland fold-thrust belt (Eastern Cordillera) is characterized by Paleogene (dominantly ca. 50–30 Ma), Triassic (250–200 Ma), minor Early Jurassic (200–145 Ma), Andean basement (650–250 Ma), and older Precambrian (>900 Ma) populations; Paleogene ages in the Eastern Cordillera hinterland are older than those in the forearc and magmatic arc. (3) The frontal fold-thrust belt (Subandean Zone to Oriente Basin) is characterized by two prominent Jurassic age peaks and a secondary Triassic age population that exhibits a broad distribution spanning most of the Triassic (200–250 Ma).

In this study, the sediment sources are explicitly known, in the sense that zircon grains in the modern rivers are derived from surface waters within individual stream catchments with known surface geology. Nevertheless, to independently verify diagnostic age peaks within the sediment source areas, we applied the non-negative matrix factorization (NMF) approach of Saylor et al. (2019). This inverse technique generates a series of age distributions representative of potential sources. The NMF model results for the 19 modern river samples (Fig. S1 [footnote 1]) demonstrate the diagnostic Mesozoic–Cenozoic age populations within the Subandean Zone (two Jurassic age peaks), Eastern Cordillera (Eocene, Early Cretaceous, and broad Triassic age peaks), and Western Cordillera (Pliocene–Quaternary, Miocene, Oligocene, and Late Cretaceous age peaks).

A statistical test for similarity among the 19 modern river samples using multidimensional scaling (MDS; Vermeesch, 2012) shows clear clustering of sample U-Pb age distributions into zones representing the three contrasting tectonic provinces—the forearc to magmatic arc (Western Cordillera), the hinterland segment of the fold-thrust belt (Eastern Cordillera), and the foreland segment of the fold-thrust belt (Subandean Zone to Oriente Basin) (Fig. 8). Only two samples have age distributions that appear outside of their respective fields. The first is a sample (sample 11: EC1535) from the Rio Paute in the Eastern Cordillera that appears to share both Eastern and Western Cordillera signatures in the MDS plot. This case is readily explained by the spatial overlap of the catchment into both the Eastern and Western Cordillera regions (Figs. 1 and 2), each accounting for roughly half of the total 5682 km2 catchment area for this sample.

The second apparent outlier in the MDS plot is from the Rio Pilaló in Western Cordillera (sample 3: EC1505), which appears to cluster very well with results diagnostic of the Eastern Cordillera (Fig. 8). The U-Pb age distribution for this sample (Fig. 4; n = 111 analyses) consists of a dominant Precambrian population (>900 Ma, 60%), primarily defined by Sunsás (1300–900 Ma) ages that constitute 49% of analyzed grains (54 of 111 analyses), with a subordinate range of ages >1300 Ma. The second-largest population is defined by Paleozoic (450–250 Ma, 13%) grains followed by 6% Famatinian (550–450 Ma), 5% late Neoproterozoic (650–550 Ma), and 6% early Neoproterozoic (900–650 Ma) ages. Although Neogene ages are hallmarks of the overall age distributions for the Western Cordillera region (Figs. 4 and 7), the Rio Pilaló sample is nearly devoid of Neogene zircons (Fig. 4, sample 3). This sample from the Rio Pilaló has the second-smallest catchment area of rivers in this study, representing only 251 km2, and a catchment composed of only Paleogene rocks (Table S2 [footnote 1]; Figs. 1 and 2). Detrital zircon geochronological results from Vallejo et al. (2019) for Paleogene stratigraphic units equivalent to those in this catchment show similar detrital U-Pb age distributions. These distributions record sedimentary input from source units in the Eastern Cordillera (Vallejo et al., 2019).

DOWNSTREAM TRENDS IN A LARGE CATCHMENT

River sand samples in the Subandean Zone represent large watersheds with western headwaters in the high-elevation Andean hinterland and magmatic arc (Fig. 2). Tributaries and main-stem rivers cross the Eastern Cordillera hinterland where U-Pb age distributions have characteristic Paleozoic and Precambrian age signatures (Figs. 5 and 6). The westernmost headwaters of the Rio Pastaza drain considerable portions of the magmatic arc in which the catchments are dominated by Pliocene–Quaternary volcanic and volcaniclastic rocks (Fig. 1; Barberi et al., 1988) (e.g., Rio Pastaza and Rio Napo watersheds, samples 15: EC1601 and 13: EC1602).

Given this scenario, it is surprising that U-Pb distributions of Subandean rivers lack the Pliocene–Quaternary and Precambrian–Paleozoic age populations diagnostic of the magmatic arc and hinterland regions, respectfully (Figs. 6 and 7). Lacking evidence for any significant sediment storage or depositional sink within upstream sectors of the catchments, we attribute the unexpected U-Pb distributions to dilution by disproportionately large detrital contributions from crystalline Mesozoic units. Five samples from the Rio Pastaza catchment demonstrate this abrupt downstream loss of important U-Pb age populations and overwhelming dilution by Mesozoic ages (Fig. 9). Measurements of the long and short axes of individual analyzed grains (Table S3 [footnote 1]) demonstrate that variations in age populations among these samples cannot be attributed to grain-size fractionation.

Upper Catchment: Magmatic Arc

The Rio Pastaza is formed by two longitudinal tributaries (the Rio Chambo and Rio Patate) within the magmatic arc. The Rio Chambo drains the northern half and the Rio Patate the southern half of the Pastaza’s upper catchment. Outcrops within the upper catchment are almost entirely Neogene–Quaternary volcanic and volcaniclastic rocks, and >20 volcanoes with documented Pliocene–Quaternary activity (Bernard and Andrade, 2011). A sample from the Rio Chambo (sample 4: EC1509) represents 2016 km2 of the Rio Pastaza’s upper catchment. The U-Pb signature of the sample is characterized by a unimodal Pliocene–Quaternary (5.3–0 Ma) age distribution (Fig. 9), consistent with sediment sources in the catchment.

Middle Catchment: Hinterland

Two river sand samples from the middle part of the Rio Pastaza’s catchment (Fig. 9, samples 9: EC1510 and 10: EC1513) within the hinterland fold-thrust belt province show a near absence of Pliocene–Quaternary zircons, and an abundance of Precambrian zircons >900 Ma, which constitute 47% and 42% of analyses for the two samples, respectively. Secondary age populations in the two mid-catchment samples are late Neoproterozoic–Famatinian (650–450 Ma; 12% and 10%, respectively) and Paleozoic–Triassic (450–200; 11% and 13%, respectively) (Fig. 9). While these U-Pb age distributions are consistent with those of other river sand samples from the Eastern Cordillera hinterland region (Fig. 5), the sample distributions bear no resemblance to that of the upper catchment sample, and are severely disproportionate to the relative exposure areas. This relationship is particularly striking because the large upstream catchment area is dominated by magmatic arc rocks, with >85% Cenozoic rocks (including >50% Pliocene–Quaternary volcanic rocks).

Lower Catchment: Proximal and Distal Subandean Zone

In the lower part of the Pastaza catchment, two samples represent both the proximal (western) part of the Subandean Zone and the distal (eastern) part of the Subandean Zone to Oriente Basin. A lower catchment sample from the proximal Subandean Zone (Fig. 9, sample 14: EC15191606) represents a 9252 km2 catchment area in which Cenozoic rocks make up 78% of exposed rock units (including 47% Quaternary volcanic rocks). Paleozoic–Triassic rocks constitute 3%, Jurassic rocks 4%, and Cretaceous rocks the remaining 15% of exposures in the catchment area. The U-Pb age distribution for the sample shows dominantly Mesozoic (250–66 Ma) ages (84%) that are principally Jurassic (145–200 Ma; 74%), scarce Precambrian (>900 Ma) ages (14%), and a virtual absence of Cenozoic (66–0 Ma) ages. Jurassic populations in these samples exhibit a bimodal distribution of Early (200–174 Ma; 36%) and Late Jurassic (165–145 Ma; 38%) ages.

The most downstream sample (sample 15: EC1601) is located near the transition from distal Subandean Zone to foreland basin and represents 13,473 km2, or 98%, of the total 13,700 km2 catchment of the Rio Pastaza when it debouches onto the Amazon Plain in the Pastaza megafan (Figs. 1 and 2; Bès de Berc et al., 2005; Bernal et al., 2011). The detrital zircon U-Pb age distribution of this sample is comparable with those of the previous two lower catchment samples, with overwhelming proportions of Mesozoic (250–66 Ma) ages (67%). In this sample, despite Jurassic crystalline rocks accounting for only 7% of its total upstream drainage area, Jurassic (145–200 Ma) ages constitute almost half (45%) of the results (96 of 212 analyses). Furthermore, while Cenozoic rocks make up 72% of exposed geologic units in the catchment, and more specifically 50% Pliocene–Quaternary volcanic rocks, Cenozoic zircons are nearly absent in this sample.

DISCUSSION

Provenance Signatures of Tectonic Provinces

The detrital zircon U-Pb results presented here for modern river sands enable discrimination of diagnostic sediment source regions that correspond to various tectonic provinces in the Ecuadorian segment of the northern Andes. Specifically, the U-Pb age distributions clearly identify the geochronological fingerprints of the major tectonic provinces recognized in cordilleran-type orogenic systems along ocean-continental subduction margins—namely the forearc, magmatic arc, and hinterland and foreland segments of the retroarc fold-thrust belt.

The U-Pb results from modern rivers in the west indicate that catchments spanning the Ecuadorian forearc and Western Cordillera are dominated by Neogene–Quaternary age populations, as expected for rocks related to the Andean magmatic arc. In contrast, the hinterland segment of the Andean fold-thrust belt (Eastern Cordillera) is defined by Proterozoic–Paleozoic age signatures diagnostic of the exposed Mesozoic sedimentary and metasedimentary rocks of ultimate cratonic origin that define this elevated region. Farther east, in frontal segments of the fold-thrust belt, rivers draining the Subandean Zone to Oriente Basin show distinctive bimodal Jurassic age populations, a secondary Triassic population, and subordinate Early Cretaceous ages representative of the major exposures of Mesozoic plutonic and metamorphic bedrock.

Paleozoic and Precambrian age signatures are generally restricted to and characteristic of the Eastern Cordillera, the elevated hinterland segment of the Andean fold-thrust belt in Ecuador (Fig. 2). As mentioned, most Precambrian ages, particularly those >900 Ma, were originally sourced from South American cratonic provinces to the east (Figs. 1 and 2). Therefore, the >900 Ma zircons are direct signatures of Eastern Cordillera detritus, indicating that individual grains have a long history involving initial derivation from the craton, Paleozoic–Mesozoic deposition in basin areas that included the present Eastern Cordillera, followed by Cenozoic erosion and transport within modern rivers.

Collectively, these results underscore the broad applicability of detrital zircon U-Pb methods to the northern Andes in general, and the Ecuadorian Andes in particular, for assessing the introduction and exhumation of sediment source regions and changes in regional paleodrainage patterns (e.g., Horton et al., 2010a, 2010b, 2015; Gutiérrez et al., 2019). These data provide a baseline for future studies of the Mesozoic–Cenozoic tectonic evolution of the northern Andes, as well as source-to-sink sediment routing and identification of critical provenance signatures within sedimentary basin systems within northern South America (Horton et al., 2015; Anderson et al., 2016; Horton, 2018).

Complex Downstream Variations in Source Signals

Variations in detrital zircon U-Pb age distributions of river sands from the large Rio Pastaza catchment reveal complexity in downstream signal propagation (Fig. 10). The characteristic Pliocene–Quaternary age population of the magmatic arc region in the upper catchment is lost at the expense of hinterland signatures in the middle catchment. In the lower catchment of the Subandean Zone, hinterland signatures are subdued at the expense of Subandean signatures, specifically Jurassic ages, even though Jurassic rocks account for a rather small proportion (7%) of rock units within the catchment area that is represented by the most-downstream sample.

The bimodal Jurassic U-Pb age distribution seen in downstream samples matches the ages of Jurassic igneous units exposed in the catchment, including the variably foliated Late Jurassic Azafrán plutonic complex within the Eastern Cordillera and the nonmetamorphosed Early Jurassic Abitagua Granite in the Subandean Zone (Litherland et al., 1994; Pratt et al., 2005; Cochrane 2013; Spikings et al., 2015). High surface slopes and high topographic relief across Paleozoic–Mesozoic igneous and metamorphic rock exposures in the Eastern Cordillera and proximal (western) Subandean Zone suggest that large, focused contributions of clastic material from these crystalline units result in their overrepresentation within U-Pb age distributions and the dilution of competing age signatures (Fig. 10C). In the middle catchment, the Rio Pastaza deeply and narrowly incises the Eastern Cordillera, with several large knickpoints with over 60–80 m of vertical relief (Fig. 10) and a steep stream gradient (22 m/km on average), suggesting that intense, focused erosion of rock units in this reach is sufficient to drown out the Pliocene–Quaternary zircon age signatures of the upper catchment.

Focused incision and erosion could be a product of spatially concentrated precipitation and/or active tectonic processes. Tectonic activity in the Subandean Zone since the Pleistocene has caused significant changes to drainage patterns, including several diversions of the Rio Pastaza by thrust-related fold growth and uplift of the Abitagua Granite (Fig. 10), as well as back-tilting due to fold growth above frontal structures farther east. 14C dating of fluvial terraces at various locations along the Rio Pastaza indicate rapid and potentially increasing rates of incision within hinterland and Subandean regions, with minimum average rates ranging from 5.0 to 6.7 mm/yr since ca. 18,000 yr B.P. (Bès de Berc et al., 2005). Although the highest rates may be influenced by dating of the youngest terraces (e.g., Gallen et al., 2015), on the basis of spatial relationships to active or recently active structures, Bès de Berc et al. (2005) argued that terrace development is mostly governed by active tectonics rather than climatic changes.

In considering downstream variations in U-Pb age distributions, it is possible that the Jurassic igneous units have high zircon concentrations or enhanced “zircon fertility” (e.g., Moecher and Samson, 2006; Dickinson, 2008) relative to Cenozoic sedimentary and Paleozoic–Cretaceous metamorphic units in the catchment (Fig. 10A), thereby contributing disproportionately large amounts of zircon. However, this is unlikely to be the singular cause for the observed dilution, on the basis of U-Pb age results from the Rio Zamora catchment to the south, where ∼50% of the catchment area is occupied by the Early Jurassic nonmetamorphosed Zamora Granite, a southern equivalent to the Abitagua Granite exposed in the Pastaza catchment (Spikings et al., 2015). A sample from the Rio Zamora (sample EC1572) does contain a large Jurassic U-Pb age population, but the age distribution is not as diluted by Jurassic zircons as would be expected based on the large surface area of the igneous body in the catchment if zircon fertility was the only contributing factor.

Another possible factor of disproportionate sediment contribution is the extreme spatial heterogeneity in mean annual precipitation between the eastern Amazonian lowlands and the high Andes (Fig. 10). Easterly trade winds carrying considerable moisture over the Amazon Basin are blocked by the topography of the Andes, forcing adiabatic expansion, condensation, and rainout, creating rainfall hot spots along the eastern flank of the Andes (e.g., Bookhagen and Strecker, 2008; Chavez and Takahashi, 2017). This wringing out of air masses at the mountain front leads to a sharp precipitation gradient in the Rio Pastaza catchment, with wet conditions (>3 m/yr) along the lower reaches of the catchment (<2 km elevation) and semiarid conditions in the upper Rio Pastaza valley (>3 km elevation), which receives <0.75 m/yr (Fig. 10; INAMHI, 2014). This sharp divide in mean annual precipitation occurs at 2–3 km elevation and coincides with the exposure levels of the Late Jurassic Azafrán pluton (Eastern Cordillera) and a precipitation maximum within the Early Jurassic Abitagua Granite (Subandean Zone). This spatially variable climatic framework may contribute to concentrated erosion within the fold-thrust belt and disproportionately large volumetric contributions from these Jurassic granites (Fig. 10).

A further consideration is the possible influence of temporally variable climatic phenomena such as El Niño–Southern Oscillation (ENSO; average reoccurrence between 2 and 7 yr) on modern river detrital zircon age distributions. Historically, La Niña events have been correlated with increased sediment and water discharge on the Rio Pastaza beyond background variability (Bernal et al., 2012); such flooding events could have led to further heterogeneous erosion along the stream channel, or rare mobilization of sediment sources in upstream reaches of the catchment.

Finally, are the number of analyses sufficient to capture all of the age populations present in downstream segments of the Rio Pastaza? Considering a commonly used threshold in detrital zircon provenance studies, whereby a sample size of 117 individual grain ages is considered to produce 95% confidence that populations >5% are identified (Vermeesch, 2004), two Rio Pastaza samples fall below this threshold—Eastern Cordillera sample 9: EC1510 with 105 grains, and sample 10: EC1513 with 91 grains (Fig. 9). However, the first of these (sample 9: EC1510) yields 95% confidence that populations >10% are identified (threshold of n = 103 grains), and the second, that populations >20% are identified at the same confidence level (threshold of n = 89) (Vermeesch, 2004). Considering that the upstream sample age distribution (Fig. 9, sample 4: EC1509) contains 92% Neogene ages (23–0 Ma)—which is much greater than a 20% fraction—we consider it highly unlikely that minor variations in the number of grains analyzed can explain the downstream loss of critical age populations.

Overall, drastic downstream changes in U-Pb age distributions along the Rio Pastaza provide a cautionary example whereby if it is not explicitly known that these samples are derived from the same large river system, they would likely be interpreted as having drastically different provenance.

CONCLUSIONS

Detrital zircon U-Pb geochronological results for 15 new samples (a total of 1705 analyses) of modern unconsolidated sand characterize the detrital fingerprints of 12 Andean rivers across the forearc, magmatic arc, and fold-thrust belt of Ecuador (1°N–5°S). These results show that river sediments from different tectonic provinces have diagnostic U-Pb age distributions defined by characteristic age populations. (1) In the forearc to magmatic arc (Western Cordillera), U-Pb age distributions are characterized by Cenozoic populations with an overall dominance of Pliocene–Quaternary ages. (2) In contrast, the hinterland segment of the fold-thrust belt (Eastern Cordillera) has a diverse (cosmopolitan) U-Pb age distribution characterized by large relative proportions of Proterozoic and Paleozoic ages representing upper Paleozoic–Mesozoic sedimentary and pelitic metamorphic rocks of ultimate cratonic derivation, and a near absence of Neogene–Quaternary ages. (3) For the frontal segment of the fold-thrust belt (Subandean Zone to Oriente Basin), U-Pb age distributions show limited diversity restricted to the Mesozoic, and are characterized by a large and bimodal Jurassic population with Early and Late Jurassic age peaks, a secondary cosmopolitan Triassic population, and subordinate Early Cretaceous ages.

Although samples from the Rio Pastaza headwaters yield age distributions consistent with all three tectonic provinces spanning the upper reaches of this catchment, these characteristic signatures are lost downstream as the river is overwhelmed by contributions from local sources. This downstream dilution of age distributions in the large Rio Pastaza watershed may be caused by focused erosion of fertile crystalline rocks due to an interplay between active tectonics and orographically induced precipitation focused at the range front.

ACKNOWLEDGMENTS

This research was supported by a U.S. National Science Foundation (NSF) Graduate Research Fellowship (grant 2016197896) awarded to LJJ, and NSF grant EAR-1338694 and National Geographic Society grant 9909-16 awarded to BKH. We thank E. Gabriela Gutiérrez, Sarah W.M. George, and Bernardo O. Beate for assistance in the field and insightful conversations. We thank reviewers R. Cecil, A. Laskowski, and Associate Editor B. Yanites for comments that helped improve the manuscript.

1Supplemental Material. Table S1: U-Pb geochronological results for modern river sand samples of Ecuador. Table S2: Sample information for modern river sand samples of Ecuador. Figure S1: NMF model results for modern river sand samples of Ecuador. Table S3: Grain-size measurements for modern river sand samples of the Rio Pastaza, Ecuador. Please visit https://doi.org/10.1130/GES02126.S1 or access the full-text article on www.gsapubs.org to view the Supplemental Material.
Science Editor: Andrea Hampel
Associate Editor: Brian J. Yanites
Gold Open Access: This paper is published under the terms of the CC-BY-NC license.