Abstract

The Taza-Guercif Basin is on the southern margin of the former Rifean Corridor, one of the major Miocene marine connections between the Atlantic Ocean and Mediterranean prior to the onset of the Messinian Salinity Crisis. As the first basin in the corridor to emerge during corridor closure, the basin is a key location for understanding this major marine event. To constrain the mechanisms for corridor closure, we contribute 499 zircon U-Pb crystallization ages and 98 zircon fission-track (ZFT) cooling ages from the stratigraphy of the Taza-Guercif Basin. The U-Pb age signature of the Taza-Guercif Basin is dominated by Pan-African (700–560 Ma) and West African craton (2200–1800 Ma) ages, and contains a significant abundance of Mesoproterozoic ages recently characterized in Mesozoic sediments from the Rif and Middle Atlas mountains. The ZFT ages record a significant Triassic-centered cooling population (275–150 Ma), well-defined Variscan (ca. 330 Ma) and post Pan-African (498 Ma) cooling peaks, and a scattering of Precambrian cooling ages. The cooling ages suggest a source in the Middle Atlas; this is consistent with the U-Pb crystallization ages. Furthermore, there is no discernable change in either the U-Pb or ZFT populations during basin emergence. Together, these observations suggest that the Middle Atlas mountains were a consistent source of sediment to the Taza-Guercif Basin and played a significant role in the closure of the Taza-Guercif Basin and possibly the Rifean Corridor.

INTRODUCTION

In the late Miocene, the connection between the Mediterranean Sea and Atlantic Ocean was tectonically severed, leading to deep evaporative drawdown of Mediterranean sea level such that the entire basin approached desiccation in an event known as the Messinian Salinity Crisis (MSC) (Hsü et al., 1973; Lofi et al., 2011; Roveri et al., 2014). The MSC sequestered 6% of global ocean salinity into evaporite deposits (Hsü et al., 1977); created a deep, dry, and hot basin that altered global atmospheric circulation (Murphy et al., 2009); opened passageways for faunal migration between Europe, Africa, and Arabia (Agustí et al., 2006); and ended in the largest flood the Earth has ever experienced (Hsü et al., 1977; Garcia-Castellanos et al., 2009; Pérez-Asensio et al., 2013; Cornée et al., 2014). The combined effects of the MSC make it one of the most important oceanic events in the past 20 m.y. (Krijgsman et al., 1999a).

It is widely accepted that the MSC was initiated through the late Miocene severing of the Betic and Rifean marine corridors (e.g., Krijgsman et al., 1999a; Krijgsman and Langereis, 2000; Duggen et al., 2004, 2005; Braga et al., 2006; Jolivet et al., 2006) that connected the Mediterranean Basin with the Atlantic Ocean through Spain and Morocco, respectively. Within the Rifean Corridor, the Taza-Guercif Basin of Morocco was one of the first basins to emerge during the progression of corridor closure (Krijgsman et al., 1999b; Garcés et al., 2001; Warny et al., 2003; Sissingh, 2008), although the exact nature of its closure is contended. Regional tectonics are considered the primary driver of corridor closure because changes to glacio-eustatic sea level (Hodell et al., 1989, 1994) and sedimentation rates (Krijgsman et al., 1999a) are generally considered to be insufficient to isolate the Mediterranean Sea. Possible tectonic mechanisms contributing to corridor closure include craton-ward thrusting in the Rif mountains following the cessation of slab rollback in the western Mediterranean (Jolivet et al., 2006), subduction-delamination uplift of the Rif mountains on the African continental margin (Duggen et al., 2003, 2004, 2005), and thermal uplift of the Middle Atlas mountains above thinned lithosphere (Babault et al., 2008; Barbero et al., 2011).

The Taza-Guercif Basin is between the Rif and Middle Atlas mountains and underwent structural deformation related to both orogens (Bernini et al., 2000). Rif orogenesis was driven by the tectonic collision of the allochthonous Alboran domain (Internal Zones; Fig. 1) against the margin of Morocco and deformed through southward-propagating reverse faults (see Chalouan et al., 2008). The Atlas Mountains are controlled by inversion of normal faults from Triassic–Jurassic extension associated with the opening of the Tethys Ocean (see Frizon de Lamotte et al., 2008). The central and eastern Rifean Corridor formed above and between Rif and Middle Atlas structures such that shortening and uplift associated with both orogenic belts are possible contributors to corridor closure. The Taza-Guercif Basin is just east of the abutment of the Rifean frontal thrust against the structurally uplifted Tazzeka spur of the Middle Atlas mountains (Fig. 1). This structural juxtaposition is probably where the marine connection between the Atlantic Ocean and Mediterranean Sea was first severed.

The tectonic interference of the two mountain belts makes determining the contribution of uplift from each belt to corridor closure difficult. Our approach to solving this problem is to perform a detailed detrital zircon provenance study of the Taza-Guercif Basin stratigraphy as proxy for surface uplift in the Rif and Middle Atlas mountains, the need for which was previously recognized (Gomez et al., 2000). Herein we present the results of the first provenance study of the Taza-Guercif Basin using a combination of detrital-zircon U-Pb crystallization ages and fission-track cooling ages. In Pratt et al. (2015), the detrital zircon signatures of key elements of the Rif and Middle Atlas mountains were determined using the same methodology, and provide detrital zircon signatures for comparison to the Taza-Guercif sediments. The differences and similarities between the signal in the Taza-Guercif Basin and those of the bounding orogens are used to evaluate the closure of the Rifean Corridor and, by extension, the initiation of the MSC.

GEOLOGIC BACKGROUND

The Miocene Taza-Guercif Basin is in the former Rifean Corridor of Morocco (Fig. 1). The basin formed in the foreland of the south-vergent Cretaceous–Holocene Rif mountain fold and thrust belt and is on top of reactivated Middle Atlas Triassic–Jurassic rift-related faulting (Bernini et al., 2000). The Rif mountains are part of the Betic-Rif orocline that rims the Mediterranean Sea from southern Spain, across the Gibraltar arc, and into Morocco (Fig. 1). This orogenic system formed due to the Cretaceous–Miocene dissection and accretion of a microcontinent, named the Alboran block in the Rif, during slab rollback in the eastern Mediterranean (Chalouan et al., 2008, and references therein). In a different scenario, the Middle Atlas mountains formed from the Late Cretaceous–Neogene inversion of a failed Triassic–Jurassic rift system (Frizon de Lamotte et al., 2008). Both orogenic systems accommodated the strain in the north African margin due to the convergence with Eurasia, and it is within the context of African-Eurasian convergence that the Taza-Guercif Basin was formed. Hydrocarbon exploration within the Taza-Guercif Basin has provided ample well and seismic data that along with outcrop studies have constrained the structural and stratigraphic evolution of the Taza-Guercif Basin (e.g., Bernini et al., 2000; Krijgsman et al., 1999a; Gelati et al., 2000; Sani et al., 2000), from which the following summary is simplified.

The Neogene evolution of the Taza-Guercif Basin began in the Tortonian with the formation of a graben system superimposed on the northeastern continuation of the Middle Atlas mountains. The graben-forming extension is attributed to foreland flexure associated with loading in the external Rif thrust that reactivated Middle Atlas rift structures (Bernini et al., 2000; Sani et al., 2000) or distal effects of a sinistral shear zone associated with west-southwest–vergent thrusting in the central Rif (Gomez et al., 2000). The onset of extension was marked by the deposition of discontinuous continental conglomerates and breccias of the Tortonian Draa Sidi Saada Formation. Shallow-marine sedimentation followed with the deposition of the Ras el Ksar Formation, which is composed of rippled sandstones, biolithites, and alternating siltstones and mudstones. These strata recorded the assimilation of the Taza-Guercif Basin into the Rifean marine corridor.

Continued subsidence led to the deposition of the open-marine Melloulou Formation, the base of which consists of thick and uniform marine marls of its Blue Marl subunit. In the basin depocenter, two turbidite sandstone systems interfinger with the blue marls: the finer and more thinly bedded El Rhirane turbidites and the coarser, more thickly bedded Tachrift turbidites. Current marks suggest a paleoflow from the south, indicating a source in the Middle Atlas (Gelati et al., 2000). The Tortonian-Messinian boundary (ca. 7.2 Ma) is at or near the top of the Tachrift turbidites (Krijgsman et al., 2000). The turbidites are overlain by the Gypsiferous Marl subunit, which was deposited after rapid shallowing of the basin between 7.2 and 7.1 Ma (Krijgsman et al., 1999a). The post-shoaling marl contains abundant gypsum crystals indicating evaporative conditions and is several hundred meters thick (Fig. 2), indicating the continuation of basin subsidence (Krijgsman et al., 1999a; Gelati et al., 2000).

Compressional forces became dominant in the Taza-Guercif Basin during the early Messinian, caused by the encroachment of the south-vergent Rifean thrust front that inverted preexisting Middle Atlas shear zone structures (Bernini et al., 1999, 2000; Gomez et al., 2000). At the same time, the westward connection with the Rifean corridor was progressively restricted, leading to lowering sea levels in the basin. The onset of basin emergence is marked by the unconformity overlying the Melloulou Formation and the deposition of the overlying Kef ed Deba Formation.

The Kef ed Deba Formation consists of transitional marine facies that are capped by fluvial-deltaic conglomerates and fossiliferous sandstones (Gelati et al., 2000). These regressive transitional marine facies of the Kef ed Deba Formation are truncated by a stark regional unconformity that marks the final emergence of the Taza-Guercif Basin between 6.7 and 6.0 Ma (Krijgsman et al., 1999a, 1999b). Above the unconformity, continental deposition began with the lacustrine carbonates and fluvial conglomerates of the Pliocene Bou Irhardaiene Formation (Gelati et al., 2000). The postemergence succession evolved under the influence of the transpressional Middle Atlas shear zone in the absence of further Rifean forcing (Bernini et al., 2000).

METHODS

Sampling

The analyzed samples obtained from Taza-Guercif Basin stratigraphy represent all formations with the exception of the monogenetic conglomerates of the Draa Sidi Saada Formation. In all formations, medium to coarse sandstones were targeted for high zircon yields, but often only finer grained sandstones were available. In total, nine sandstone samples were selected for analysis (locations are shown in Figs.1C and 2).

Three samples were collected from the Ras el Ksar Fm. TGB1 (sample prefix TGB—Taza-Guercif Basin) was sampled in the Bab Stout region in the north of the basin ∼10 m upsection of the contact with the underlying Draa Sidi Saada Formation (34.2402°, –3.8564°). The outcrop consists of a 1–2 m pitted and bioturbated beige fine- to medium-grained sandstone that is tabular and laterally continuous. Sample TGB2 was collected ∼5 m upsection from TGB1, separated by fine-grained slope-forming rock, and similarly is a 1–2-m-thick beige fine- to medium-grained pitted and bioturbated sandstone. Sample TGB14A was collected from a rust-colored medium-grained sandstone on the uplifted southern margin of the basin near the contact with the underlying Jurassic substratum in the vicinity of Bou Rached (TGB14A; 33.9224°, –3.5761°). The sampled bed displays a lenticular base that cuts into a marly bed below. The bed pinches out laterally and where observed is overlain by planar-bedded sandstones.

Sample TGB12A was collected from the El Rhirane turbidites between the Melloulou and Zobzit Rivers (TGB12A; 33.9944°; –3.7620°). The El Rhirane turbidite sandstones crop out in this area across a low rise, forming small step-like exposures separated by slightly thicker intervals of mudstones. The sample is from an ∼40-cm-thick tabular bed of rust-colored medium-grained sandstone.

Samples TGB3 and TGB4 were collected east of the Zobzit River within a coarsening-upward succession of the Tachrift turbidites that form a ridge overlooking the village of Timalit (TGB3 and TGB4; 33.9982°, –3.7392°). Each sample was taken from the uppermost bed of two of different coarsening-upward intervals. Sampled beds are ∼1.5 m thick and composed of beige medium-grained upward-fining structureless sandstones. Sample TGB3 is the uppermost of the 2 samples, occurring ∼5 m upsection from TGB4.

Samples TGB5 and TGB6 were acquired from more arenaceous intervals within the Gypsiferous Marl subunit exposed east of the Zobzit River on the eastern limb of the Safsafat anticline (TGB5 and TGB6; 34.0012°, –3.6593°). Similar to the El Rhirane turbidites, the exposures occur on a low rise with sandstones interbedded within marl forming discrete step-like ridges. TGB5 was acquired from a dark red to brown ∼50-cm-thick fine-grained planar-bedded quartz-rich sandstone. TGB6 was sampled ∼5–6 m upsection in a 20–30-cm-thick dark red to brown poorly sorted medium- to coarse-grained sandstone.

Sample TGB10 was collected from the Kef ed Deba Formation within a red sandstone interval cropping out on the southern exposure of the eastern limb of the Safsafat anticline (TGB10; 34.0055°, –3.649089°). The sandstone bed is fine grained, ∼50-cm-thick, planar bedded, and contains abundant pelecypods.

TGB7 is the stratigraphically highest sample analyzed in this study and was collected from an ∼70-cm-thick rippled fine- to medium-grained sandstone intercalated with the conglomerates of the Bou Irhardaiene Formation ∼3 m above a regional unconformity in the Taza-Guercif Basin (TGB7; 34.0070°, –3.6492°). The surrounding pebble-cobble conglomerates are matrix poor and observed clasts consist almost entirely of carbonate.

Detrital Zircon U-Pb Geochronology

Seven samples (TGB1, TGB4, TGB5, TGB7, TGB10, TGB12A, TGB14A) were analyzed using laser ablation–single-collector–inductively coupled plasma–mass spectrometry (LA-SC-ICP-MS) at the University of South Carolina’s Center for Elemental Mass Spectrometry. Our procedure was detailed in Pratt et al. (2015) and is only briefly summarized here. Following standard rock disaggregation and zircon separation techniques, samples were ablated with a PhotonMachines 193 nm ArF excimer laser and the ablated material was plumbed to and analyzed in a Thermo Scientific Element2 high-resolution SC-ICP-MS. Our analysis measured the signal intensities of 202Hg, 204(Pb + Hg), 206Pb, 207Pb, 208Pb, 232Th, and 238U. Each sample analysis targeted ∼100 unknown zircons and incorporated an analysis of the natural zircon standard SL2 (563.2 ± 4.8 Ma; Gehrels et al., 2008) after every fifth unknown, and R33 (419.3 ± 0.4 Ma; Black et al., 2004) after every twentieth unknown. Each individual grain analysis included a 10 s blank integration followed by 28 s of analysis during 10 Hz ablation by a 25 μm circular spot, followed by 15 s for post-ablation chamber wash-out.

Data reduction was performed in the Iolite software add-on (Paton et al., 2011) for WaveMetric’s IgorPro software utilizing the U-Pb Geochronology3 Data Reduction Scheme. Data were corrected for background signals, downhole fractionation, and instrument drift to produce the final isotopic ratios and ages (Supplemental Table 11). Final analyses with 2σ error >20% in either the 238U/206Pb or 207U/206Pb ages were disregarded. Resulting analyses with 238U/206Pb ages older than 500 Ma were subjected to a concordance filter whereby any grains with >30% normal discordance or >5% reverse discordance were similarly disregarded. Grains with 238U/206Pb ages younger than 500 Ma were retained regardless of discordance. Herein we use 238U/206Pb ages for grains with ages younger than 1.3 Ga, and 207Pb/206Pb ages for grains with 238U/206Pb ages older than 1.3 Ga. This age division was chosen to reduce individual age uncertainty and to avoid skewing a population that extends across the more conventional 1.0 Ga age-method division.

Detrital Zircon Fission-Track Geochronology

Zircon fission-track (ZFT) thermochronology provides ages at which zircons cool below the closure temperature, which is typically ∼250 °C for average zircons (Fleischer et al., 1975; Tagami and O’Sullivan, 2005). These ages are determined using the density of fission-induced damage trails, or fission tracks, in the crystal lattice of the zircons and the rate at which 238U undergoes spontaneous fission. Above the closure temperature, fission tracks are repaired and below the closure temperature, fission tracks are retained. Thus reheating above the closure temperature resets the fission-track age (Naeser, 1979; Bernet and Garver, 2005). The distribution of fission-track ages from detrital zircons in a sedimentary sample can provide a unique signature separate from the U-Pb crystallization ages. The combination of the two signatures better discriminates between potential source rocks and is particularly valuable in in areas, such as Morocco, where crystallization ages are likely to be uniform but have a more heterogeneous low-temperature thermal history.

The ZFT analysis of our samples was performed using the scanning electron microscope high-density fission-track (SEM-HDFT) technique (e.g., Montario and Garver, 2009). This technique utilizes higher magnifications and gentler etching techniques to allow the counting of zircon fission tracks at high densities (3 × 108 tracks/cm2), thereby unlocking cooling ages in highly radiation damaged grains, as are common to old and/or high [U] crystals. The closure temperature for radiation-damaged grains is between ∼150 and 195 °C. Details of the technique can be found in Montario and Garver (2009); our procedure was presented in detail in Pratt et al. (2015), and is only briefly summarized here. This method preferentially targets nearly metamict zircon grains and caution should be applied when comparing directly to results from conventional optical fission-track analysis.

Zircons were extracted from the same separates produced for the U-Pb analysis to produce grain mounts for four samples (TGB3, TGB6, TGB12A, and TGB14A) in addition to three mounts of natural standards for zeta calibration: the Buluk Tuff (16.4 ± 0.2 Ma fission-track age), Fish Canyon Tuff (27.9 ± 0.5 Ma), and Peach Springs Tuff (18.5 ± 0.1 Ma). Etching proceeded for 5–7 h in a KOH:NaOH eutectic at 228 °C. The mounts were then flattened, cooled, and cleaned prior to affixation with an ∼0.2 mm mica flake. The mounts and 3 glass dosimeters with a 238U content of 12.3 ppm were irradiated in Oregon State University’s TRIGA Mark II Reactor. Following irradiation, the mica detectors were etched in 48% HF for ∼18 min. The mounts and corresponding mica detectors were mounted as mirror images on petrographic slides with an ∼8–10-nm-thick carbon coating and imaged at ∼3000–10,000× using the secondary electron detector of a Zeiss EVO50 tungsten-filament SEM.

Natural and induced fission-track densities were determined using spatial analysis of the secondary electron images in the ImageJ software package (http://imagej.nih.gov/ij/). The number of spontaneous fission tracks preserved in the zircon, induced fission tracks in the mica, and image area were entered into ZetaAge (Brandon, 1996) and DensityPlotter (Vermeesch, 2009, 2012) to determine grain ages, kernel density estimates (KDE), and mixture model peaks. The mixture model attempts to fit a discrete number of cooling events to the fission-track ages where peaks are located at the age of the event and distribution of ages about the peaks is due to partial resetting (Brandon, 1996). To create a mixture model DensityPlotter chooses the number of age components by minimizing the Bayes Information Criterion and determines their value using a hybrid deterministic and Markov chain Monte Carlo method (Vermeesch, 2009, 2012). The KDE creates a curve of the relative probability that a randomly selected grain from an infinite population would occur at a given age based on the density distribution of sampled ages (Vermeesch, 2012).

RESULTS

Zircon U-Pb Results

We analyzed 638 zircons in the course of this study. After error and concordance filtering, analysis yielded 500 individual zircon U-Pb ages from 7 samples, TGB14A, TGB1, TGB12A, TGB4, TGB5, TGB10 and TGB7 (see Supplemental Table 1). Age spectra for individual samples as well as the composite of the Taza-Guercif Basin are shown in Figure 3.

The composite of Taza-Guercif Basin samples contains ages ranging from 3570 to 68 Ma and contains 2 dominant populations. The largest population contains 228 ages between 693 and 518 Ma, representing 45.6% of all analyzed grains. The second-largest population with 78 grains consists of ages that are between 2185 and 1790 Ma, and represents 15.6% of the total population; <9% of all grains are younger than the 518 Ma lower limit of the primary age peak, and only 7% are older than 2185 Ma. Individual samples are all generally very similar to the composite spectrum (Fig. 3). For brevity, the populations in each individual sample are reported Figure 3, and the differences between individual samples are discussed later in the text.

ZFT Results

SEM high-definition fission-track (SEM-HD-FT) analysis yielded results from 99 zircon grains from 4 samples from the Taza-Guercif Basin (TGB14A, TGB12A, TGB3, and TGB6) (see Supplemental Table 22). Greater grain counts were obtained for TGB12A and TGB3, which contain 35 and 40 grains, respectively, whereas the remaining 2 samples measured 11 and 13 grains. Analysis of a fifth sample, TGB7, was attempted, but yielded too few appropriate zircons after those for U-Pb analysis were separated. The 95% confidence interval for individual ages is regularly in excess of 100 m.y. above and below the calculated age. KDEs and mixture model peaks were generated for each sample and are shown in Figure 4. The mixture model attempts to fit a discrete number of cooling events to the fission-track ages where peaks are located at the age of the event and deviation about the peaks is due to partial resetting. The KDE uses the density of sampled ages to compute a curve of the probability that a randomly chosen grain would fall at a given age if the population were samples infinitely.

Ras el Ksar sample TGB14A (n = 13) produced KDE peaks at 198, 259, and 506 Ma with mixture model peaks occurring at 228 ± 21 Ma, 510 ± 55 Ma, and 1037 ± 343 Ma (Fig. 4). The mixture model is in good agreement with the KDE peaks with the exception that the mixture model lumped the two youngest peaks defined by the KDE. Statistically, the 2 KDE peaks are inseparable based on the 95% confidence interval of individual ages.

El Rhirane turbidite sample TGB12A (n = 35) displays KDE peaks at 237, 323, and 489 Ma, and mixture model peaks occur at 203 ± 36, 323 ± 27, and 606 ± 78 Ma (Fig. 4). The KDE and mixture model peaks occur at similar ages for the youngest two peaks. A mismatch occurs with the third peak where the mixture model combines the grains represented by the 489 Ma KDE peak with all older ages.

The spectra for the Tachrift turbidite sample TGB3 (n = 40) yielded 3 strong KDE peaks at 181, 323, and 489 Ma. Mixture model peaks for TGB3 were calculated as 172 ± 19 Ma, 323 ± 37 Ma, and 592 ± 50 Ma (Fig. 4). The mixture model and KDE peaks produce similar young peaks, while the oldest peaks in each model deviate significantly. The issue is the same as TGB12A, where the mixture model lumps the oldest half (52% ± 13%) together to define one peak.

Sample TGB6 from the Gypsiferous Marl (n = 11) displays three dominant KDE peaks at 223, 335, and 957 Ma. Mixture model peaks are calculated as 171 ± 26, 336 ± 29, and 959 ± 300 Ma (Fig. 4). The two models are in near perfect agreement, with the only difference occurring in the age of the youngest peak. However, the difference in these ages is within the 95% confidence interval for individual grain ages.

DISCUSSION

Basin-Scale Provenance

Examining the detrital zircon U-Pb crystallization age data from the Taza-Guercif Basin as a whole reveals a dominant Pan-African signal as demonstrated by the robust ca. 700–520 Ma population (Fig. 3). These Pan-African ages represent a period of orogenesis and magmatism that occurred during the coalescence of Gondwana in the latest Proterozoic and earliest Paleozoic (Stern, 1994, 2002). This population combined with the ca. 2.2–1.8 Ga Paleoproterozoic U-Pb crystallization ages composes a signature typical of sediment derived from the West African craton and its Pan-African margin (Fig. 5; Nance and Murphy, 1994; Abati et al., 2010; Avigad et al., 2012). The smaller population at ∼950 Ma crystallization age is not typical of the West African craton crystalline basement, but has been identified in Ediacaran–Cambrian exposures in the Anti-Atlas mountains (Avigad et al., 2012), in the Internal Zones of the Betic-Rif arc in Spain (Fig. 1; Platt and Whitehouse, 1999; Zeck and Whitehouse, 1999, 2002; Zeck and Williams, 2001), as well as in Mesozoic exposures in the Middle Atlas and Rif mountains (Fig. 5; Pratt et al., 2015).

While it is unsurprising that these northwest African samples show a dominant signal from the West African craton, the presence of Mesoproterozoic ages as represented by a small population peak ca. 1190 Ma is significant. To date, Mesoproterozoic zircons have only been identified in Morocco within the Mesozoic samples mentioned above. In general, Mesoproterozoic zircon crystallization ages are rare in the region, occurring only in small abundances in other exposures across North Africa and not coevally with the ca. 1190 peak observed here (Thomas et al., 2010; Linnemann et al., 2011).

In the thermochronologic data, all of the studied Taza-Guercif Basin strata contain a ZFT cooling age population centered on the Triassic (Fig. 4), spanning from the Permian to the Late Jurassic (ca. 285–160 Ma). The age range of this cooling population coincides with late Variscan magmatism that affected the Moroccan Meseta (Michard et al., 2008, and references therein), the Triassic rifting of the Atlas basins and the Central Atlantic Magmatic Province lava flows in Morocco (Knight et al., 2004; Verati et al., 2007; Frizon de Lamotte et al., 2008), and Early Jurassic magmatism that occurred in the High Atlas (Frizon de Lamotte et al., 2008, and references therein) with coeval hydrothermal events in the Middle Atlas (Hamidi et al., 1997; Auajjar and Boulègue, 2002; Dekayir et al., 2005).

All samples except for TGB14A also contain a population of grains with ZFT ages ca. 330 Ma, coeval with the early period of Variscan tectonics and magmatism that affected the Moroccan Meseta and the Anti-Atlas mountains (Michard et al., 2008). In the Taza-Guercif Basin composite ZFT spectrum, this peak is the best defined and of greatest amplitude (Fig. 6). This suggests that the zircons of this population were not affected by subsequent annealing during the latter phases of the Variscan orogeny or during the Mesozoic extension that followed.

Samples TGB14A and TGB3 contain a significant KDE cooling age peak in the Cambrian, whereas sample TGB12A displays a minor peak during this time (Fig. 4). The absence of this population in the remaining sample, TGB6, may simply result from its comparatively small grain count, as it otherwise closely resembles TGB12A. These Proterozoic cooling ages may reflect distinct cooling events, but were likely subject to various degrees of reheating and partial resetting and as such are not diagnostic.

Evaluation of the Middle Atlas and Rif Mountains as Zircon Sources

In the late Miocene both the Rif and Middle Atlas mountains were deforming adjacent to the Taza-Guercif Basin. The Middle Atlas mountains were undergoing broad thermal uplift (e.g., Teixell et al., 2005; Barbero et al., 2007; Babault et al., 2008) while the encroachment of the Rif thrust front reactivated Middle Atlas structures and caused shortening in the Rif domain (Bernini et al., 1999, 2000; Gomez et al., 2000). Due to their immediate proximity to the Taza-Guercif Basin and their elevation, these two mountain ranges are the most likely sources of sediment to the basin. The currently available data and geologic histories of these potential source areas are compared and evaluated in relation to the both the detrital zircon U-Pb crystallization ages and the ZFT cooling ages obtained from the sediments of the Taza-Guercif Basin.

Middle Atlas

In the Middle Atlas, the only available U-Pb or ZFT data come from the Jurassic (Bathonian–Callovian) Bou Rached sandstones sampled in the northern Middle Atlas near the contact with Miocene sediments of the Taza-Guercif Basin (Fig. 1C; Pratt et al., 2015). These sandstones extend beneath the Draa Sidi Saada and Ras el Ksar Formations and compose part of the Jurassic substratum, as confirmed by borehole and seismic data (Sani et al., 2000). In some portions of the basin, the sandstones are capped by upper Jurassic limestones and in other parts, particularly the margins of the basin, they immediately underlie the Miocene basin fill beneath a dramatic angular unconformity.

Despite this relationship, the composite U-Pb crystallization age spectrum of the Taza-Guercif Basin does not match that of the Middle Atlas Bou Rached Jurassic sandstones (Fig. 5), although the latter may be a minor contributor to the former. Detrital zircon U-Pb age spectra from the Bou Rached sandstones contain a 500–400 Ma population that accounts for <0.5% of grains in the Taza-Guercif composite age-distribution curve while composing >10% of the grains in the Bou Rached sandstones. The Bou Rached sandstones display a characteristic gap in U-Pb ages between 560 and 500 Ma not present in the Taza-Guercif Basin samples and contain a larger proportion of Archean ages (Fig. 5).

The two successions also contain different cooling age distributions. The Bou Rached sandstones contain Ordovician–Silurian cooling peaks that are absent in the Taza-Guercif Basin samples. Another difference is observed in the youngest cooling peaks in the Taza-Guercif Basin samples that are Triassic–Early Jurassic in age, whereas those of the Bou Rached sandstones are Middle Jurassic and Early Cretaceous. This difference is within the possible range of error for individual grains as determined by the 95% confidence interval and may reflect method and not geology.

While the Bou Rached sandstones are in closest proximity to the sampled sediments of the Taza-Guercif Basin, the majority of the northern Middle Atlas outcrops consist of the lower Jurassic Lias platform carbonates. These rocks are an unlikely source for detrital sediments in the Taza-Guercif Basin as the dominant carbonate lithology is incapable of generating the coarser arenaceous turbidites and transitional marine sandstones found within the Taza-Guercif Basin, although the possibility of a contribution from interbedded quartzose lithologies cannot be excluded.

Farther south in the Middle Atlas, the Jurassic carbonates are overlain by lower Cretaceous continental red beds that were subsequently covered in the Late Cretaceous by sediments deposited during a regional transgression (Faure-Muret and Choubert, 1971; Frizon de Lamotte et al., 2008). The Cretaceous continental and marine successions are preserved in the southern half of the Middle Atlas and are mostly absent to the north. This missing cover may represent a viable distal source from the Middle Atlas to the Miocene Taza-Guercif Basin.

Rif

Limited data in the Rif mountains are available for the Ketama and Tisiren units that represent the External Zones and Maghrebian Flysch domain, respectively (Fig. 1B). The U-Pb crystallization ages from these domains are nearly identical to one another. The Taza-Guercif Basin samples contain U-Pb crystallization age signatures that closely resemble those of the Rifean samples (Fig. 5). The majority of U-Pb ages shared between the Rif samples and the Taza-Guercif Basin samples correspond to the Pan-African orogeny and West African cratonic signatures while the source of the shared Mesoproterozoic population is poorly constrained. Given the commonality of the Pan-African and West African craton ages across the region and the uncertainty of the temporal and spatial distribution of the Mesoproterozoic ages, this correlation is insufficient to conclude a source-sink relationship.

Analysis of the ZFT cooling age distributions suggests that the sampled domains in the Rif are not the source for the Taza-Guercif Basin sediments (Fig. 5). A combination of the ZFT cooling ages obtained in the Tisiren and Ketama units could provide the Variscan age distributions found in the Taza-Guercif Basin. However, both sampled Rif units lack the strong Triassic-centered cooling age population that is ubiquitous in the Taza-Guercif Basin ZFT spectra.

The Internal Zones of both the Rif and Betic mountains are equivalent and composed of rocks from the allochthonous Alboran domain (Chalouan et al., 2008, and references therein). There are no comparable data for Internal Zones of the Rif (Fig. 1B); however, there are data for the Internal Zones of the Betic Mountains of Spain. Comparison of the zircon crystallization and cooling age data from the Betic Alboran domain (Platt and Whitehouse, 1999; Zeck and Whitehouse, 1999, 2002; Zeck and Williams, 2001) and those of the Taza-Guercif Basin indicate that the Betic Alboran domain was not the source for the Taza-Guercif Basin. (1) The Taza-Guercif Basin sediments lack the Paleogene to early Miocene ZFT cooling age ages commonly recorded in the Alboran domain, although this may be a product of selection bias for radiation-damaged grains in our ZFT analysis. (2) The Variscan U-Pb crystallization ages that occur in abundance in the Internal domain are scant in the Taza-Guercif sediments (Fig. 5). (3) The U-Pb zircon data from the Alboran domain lack the Mesoproterozoic crystallization ages that occur in all Taza-Guercif samples.

Triassic-Centered Detrital Cooling Ages and Their Possible Sources

The Triassic cooling age population that is consistent in ZFT data across the Taza-Guercif Basin samples distinguish them from the available data in the Bou Rached sandstones of the Middle Atlas as well as the Internal Zones, Maghrebian Flysch domain, and External Zones of the Rif (Figs. 4 and 5; Pratt et al., 2015). Even though the Bou Rached sandstones lack this key ZFT population, the most likely source for the Triassic cooling populations is other strata in the High and Middle Atlas mountains (Fig. 1B). Triassic extension and rifting in the Tethyan domain formed the High and Middle Atlas rift basins (Frizon de Lamotte et al., 2008) and domains were marked by magmatism in the Mesozoic.

Magmatic events that may be recorded in the ZFT ages include the Central Atlantic Magmatic Province magmatism that affected Morocco and is restricted in age to between 200 and 196 Ma (Knight et al., 2004; Verati et al., 2007). The cooling age population also encompasses two phases of Jurassic–Early Cretaceous magmatism that followed in the High Atlas at 175–155 Ma and 135–110 Ma (Souhel, 1996; Frizon de Lamotte et al., 2008). These ages could derive from either the High or Middle Atlas mountains. Triassic Central Atlantic Magmatic Province–related magmatism occurred in both regions and while only the High Atlas was affected directly by the Jurassic magmatism (Frizon de Lamotte et al., 2008), the Middle Atlas underwent coeval hydrothermal heating (Hamidi et al., 1997; Auajjar and Boulègue, 2002; Dekayir et al., 2005). These magmatic episodes are consistent with the range of ages present in the Triassic-centered cooling age population (Fig. 3), and the combination of several events could explain the broad distribution of ages as seen in the Taza-Guercif Basin composite (Fig. 6).

A source in the Middle and High Atlas domains is supported by the available thermochronological data outside the Middle Atlas mountains that imply that Morocco was mostly subsiding during the Triassic–Early Jurassic and that regional exhumation is an unlikely cause of the cooling recorded in the Triassic–Early Jurassic ZFT ages. The western Anti-Atlas and Western Meseta were located along the West Moroccan arch, the common rift shoulder of Tethyan (Atlas) and Atlantic rifting (Frizon de Lamotte et al., 2008, 2009), and are the most likely terranes in Morocco to record Triassic–Early Jurassic exhumation.

Temperature-time models constructed from apatite fission-track (AFT) data of Variscan granitoids and metamorphic rocks within the Western Meseta predict cooling below 120 °C by ca. 300–250 Ma (Saddiqi et al., 2009; Barbero et al., 2011) and record subsidence and burial through the Triassic and Jurassic that reached temperatures insufficient to reset the ZFT system (Ghorbal et al., 2008; Saddiqi et al., 2009). These data imply that the zircons within the Variscan and older rocks of the Western Meseta basement most likely retained their Variscan and older cooling ages. Variscan cooling ages are also retained in the central and western Anti-Atlas, confirmed by ca. 340–310 Ma ZFT ages obtained from the Proterozoic basement that cooled steadily in the absence of regional exhumation (Sebti et al., 2009; Ruiz et al., 2011; Oukassou et al., 2013).

Middle Atlas Source (?)

The U-Pb zircon crystallization age results of this study as well as those from previous Moroccan detrital-zircon studies (Abati et al., 2010; Avigad et al., 2012; Pratt et al., 2015) suggest a similar U-Pb signature for the majority of Moroccan stratigraphy, obscuring robust source-sink correlations using only detrital zircon U-Pb crystallization ages. As a result we primarily rely on our smaller (and larger error) ZFT cooling age data set, integrated with regional geologic relationships, to interpret the provenance of the Taza-Guercif Basin stratigraphy. The ZFT data indicate that the currently sampled stratigraphy of the Rif and Middle Atlas mountains is not the dominant source of the Taza-Guercif Basin sediments because they do not share the Triassic-centered cooling population. Therefore, despite the similar U-Pb zircon crystallization age distributions between the Taza-Guercif Basin and the Rif (Pratt et al., 2015), we propose that the dominant zircon source to the Taza-Guercif Basin was the lower Cretaceous continental successions of the Middle and High Atlas mountains. This interpretation is based on five lines of reasoning.

  1. The age of at least part of the source stratigraphy should be the same age or younger than the 150–125 Ma ZFT ages found in the basin as these grains are inherited from the sources. If these cooling ages were acquired in the source in situ, it is highly unlikely the older cooling ages would have been retained.

  2. The Middle and High Atlas were the loci of Triassic–Jurassic extension and associated magmatic and hydrothermal heating that are the inferred source of the corresponding Triassic–Jurassic cooling peak.

  3. A Late Cretaceous transgression inundated the majority of Morocco, burying older Middle and High Atlas stratigraphy that may have contained more locally derived sediment containing the Early Jurassic zircon cooling age signal.

  4. The Lias carbonates and the upper Cretaceous fine-grained lithologies are unlikely sediment sources for the coarse and zircon-rich sandstones sampled in the Taza-Guercif Basin.

  5. This model is consistent with observed paleocurrent indicators in the El Rhirane turbidites that suggest a southern Middle Atlas source (Gomez et al., 2000). Although now mostly preserved on the margins of the uplifted Atlas Mountains, these Early Cretaceous sandstones likely covered a much wider area in the Cretaceous and were subsequently eroded during the uplift of the Atlas. These rocks, along with the Lias strata, supplied sediment to the Taza-Guercif Basin, generating the marls that dominate the basin, as well as the intercalated turbidite sandstones within.

Provenance Evolution

Although the composite U-Pb crystallization age spectrum of the Taza-Guercif Basin suggests a dominant Middle Atlas source of sediment for the Taza-Guercif Basin, the presence or absence of provenance shifts and trends in the basin fill from opening to closure is pertinent to constraining the tectonic severing of the Rifean Corridor. The following discussion explores the provenance of each individual sample moving upsection.

The Ras el Ksar formation samples TGB1 and TGB14A, from the north and south of the basin, respectively, demonstrate at least a partial divergence in sediment provenance during deposition at the opening of the Taza-Guercif Basin. Sample TGB1 from the Bab Stout area contains a Paleozoic U-Pb zircon crystallization signature that accounts for 25% of single-grain ages within the sample (the largest such signature in the Taza-Guercif Basin samples) and a younger Neoproterozoic peak than present in the other samples. The sample has the lowest percentage of Mesoproterozoic grains, lacking the Ectasian–Tonian (ca. 1300–900 Ma) peaks common to other Taza-Guercif Basin samples. These features indicate that the sample may have been partially sourced from Middle Jurassic sandstones equivalent to the Bou Rached sandstones. These samples may also have sourced from the west through the Rifean Corridor, because the sample location for TGB1 is nearer to the paleoconnection with the Saiss basin (Fig. 1). The Ras el Ksar Formation sample TGB14A contains fewer grains with Paleozoic U-Pb crystallization ages than the average for the Taza-Guercif Basin samples, and contains the two-peaked Ectasian-Tonian (ca. 1300–900 Ma) populations (Fig. 3), suggesting a provenance different from that of the northern sample (TGB1).

The stratigraphically upsection turbidite units of the Melloulou Formation (TGB3 and TGB12A) contain similar U-Pb crystallization age spectra, each containing a Pan-African signature with a dominant peak ca. 615 Ma and a subordinate peak ca. 534 Ma. The spectra differ in the older ages, yet the similarity is striking for ages younger than 700 Ma, and is not shared with any of the other Taza-Guercif Basin samples. The age of mixture model peaks generated for the ZFT cooling age data are similar between the two units. Significant differences occur only in the Mesozoic cooling populations, where the El Rhirane turbidite sample contains a peak in the Triassic and the Tachrift turbidite sample contains a peak in the Early Jurassic. This degree of variation is within the 95% confidence interval of error for individual ZFT analyses and it is possible that these peaks represent the same population. The KDE peak ages for the ZFT spectra are also similar between samples, sharing the Variscan peak and with a Triassic peak for the El Rhirane unit and an Early Jurassic peak for the Tachrift unit. The two Mesozoic peaks are within the 95% confidence interval of each of the individual ages that compose the populations. Despite the differences in the proportions of each cooling population between the samples, they occur at the same ages. This suggests that they were derived from the same general source region. The minor differences between the samples probably reflect changes in the relative exposure of different rocks within the source area or a modest shift in the locus of erosion. Paleocurrent indicators in the El Rhirane turbidites indicate a southern source (Bernini et al., 1999), consistent with erosion of Middle Atlas or High Atlas Mesozoic cover.

Upsection in the post-shallowing Gypsiferous Marl unit of the Melloulou Formation, the U-Pb zircon crystallization age spectrum of sample TGB5 contains Tonian- and Ectasian-age peaks and a distinctive tri-peaked Pan-African age distribution. ZFT cooling ages of the Gypsiferous Marl unit sample TGB6 generally coincide with that of the El Rhirane turbidite sample TGB12A, containing Mesozoic cooling ages represented by a KDE peak at 223 Ma and a mixing model peak at 171 Ma, along with a large ca. 335 Ma Variscan peak in both the KDE and mixing model. These data suggest that the Gypsiferous Marl unit sandstones were derived from the same source region as the El Rhirane and Tachrift turbidites.

U-Pb zircon crystallization age spectra from the upsection Kef ed Deba Formation sample TGB10 and the continental Bou Irhardaiene Formation sample TGB7 show no significant changes from the Gypsiferous Marl sandstone. Both TGB10 and TGB7 contain a tri-peaked Pan-African distribution, while the magnitudes of the Ectasian–Tonian and the Paleoproterozoic peaks vary slightly. This minor variance may be attributed to changing exposures in the source region, hydrodynamic sorting, or simple statistics of random grain selection.

Implications for the Closure of the Rifean Corridor

Our crystallization and cooling age data from detrital zircons of the Taza-Guercif Basin suggest that there were no major shifts in provenance between deposition of (1) the El Rhirane and Tachrift turbidites, deposited when the basin reached its deepest bathymetry, and (2) the post-shallowing Gypsiferous Marl unit. Thus, the shallowing of the basin did not shift the source for the Taza-Guercif Basin, and it continued to subside throughout the deposition of the thick Gypsiferous Marl unit.

The shallowing was most likely the result of a restricted marine connection to the west and/or the cessation of tectonic subsidence (Bernini et al., 1999; Gomez et al., 2000). Basin uplift associated with the advancement of the Rif thrusts and olistostrome emplacement is an alternative explanation for the onset of shallowing (Krijgsman et al., 1999; Krijgsman and Langereis, 2000). Detrital zircons from Taza-Guercif Basin strata deposited before and after shallowing show no obvious provenance shifts in either the fission-track cooling ages or U-Pb crystallization ages that would suggest increased sediment supply accompanying uplift in the adjacent Rif mountains. The presence of the Masgout and Tazzeka areas as basement highs along the Msoun arch (Gomez et al., 2000; Sani et al., 2000) during the Neogene filling of the Taza-Guercif Basin may have prevented Rif sediments from penetrating the Melloulou-Zobzit embayment, particularly during the deposition of the Gypsiferous Marl unit under shallow conditions.

The shift to an intracontinental basin ca. 6.7–6.0 Ma marked by the unconformity between the Kef ed Deba and Bou Irhardaiene Formations did not significantly affect the detrital zircon U-Pb crystallization age distribution in the basin. The presence of the distinct tri-peaked Pan-African zircon crystallization age signal in the Gypsiferous Marl, Kef ed Deba Formation, and Bou Irhardaiene Formation would not likely have been maintained across their bounding unconformities through a major shift in provenance. This implies that the final emergence of the basin did not affect the provenance of basin sediments.

While the subsidence of the Taza-Guercif Basin was controlled by loading from the Rif interacting with Middle Atlas structures and then sediment loading (Bernini et al., 1999, 2000; Gomez et al., 2000; Sani et al., 2000), the availability of sediments that filled the basin appears to have been controlled by the relative uplift of Middle Atlas mountains. Thus, an apparently consistent source of sediments derived from the Middle Atlas agrees well with models for domal surface uplift of the region beginning ca. 15 Ma (Barbero et al., 2007; Babault et al., 2008), with intermittent fault-controlled uplift providing coarse turbidite sedimentation.

The role of thermal uplift of the Middle Atlas has also been proposed to contribute to the closing of the Saiss basin west of the Taza-Guercif Basin and the contact of the Rif and Middle Atlas mountains (Babault et al., 2008). The Ras el Ksar Formation outcrop where TGB14A was collected now is at an elevation of ∼580 m, ∼260 m higher than the Ras el Ksar Formation outcrops in the north of the basin where TGB1 was collected. Today, the northern Middle Atlas mountains drain to the north through the old Taza-Guercif Basin depocenter, a scenario that seems relatively unchanged since the Neogene opening of the Taza-Guercif Basin.

CONCLUSIONS

Our analysis of the detrital zircon in the Taza-Guercif Basin provides the first attempt to constrain the provenance of the basin sediments under the dual tectonic influence of the Rif and Middle Atlas mountains to elucidate the forces that severed the Rifean Corridor. The conclusions from this analysis follow.

Inherited Triassic-centered ZFT cooling ages present in all of the Taza-Guercif Basin samples indicate that the primary source for the basin derived from the post-Jurassic cover of the Middle and/or High Atlas mountains, the proximal position of the former offering a more likely match.

The sediment provenance at the opening of the Taza-Guercif Basin was not uniform across the basin. The northern Ras el Ksar Formation contains significant zircons with crystallization ages between 500 and 400 Ma, a population found in the Middle Jurassic sandstones of the Middle Atlas (Pratt et al., 2015), while the southern Ras el Ksar Formation contains a provenance signal similar to that throughout the rest of the basin fill.

The turbidites of the overlying Melloulou Formation differ slightly from one another but both reflect an overall Middle Atlas provenance. Variations may have resulted from minor shifts in the locus of erosion or changes in the relative exposure of different source rocks. These changes may reflect tectonic activity in the Middle Atlas, a scenario that has been previously proposed (Bernini et al., 1999; Gelati et al., 2000).

Sediment provenance did not change with the deposition of the Gypsiferous Marl unit after rapid shallowing of the Taza-Guercif Basin and Rifean Corridor. The U-Pb zircon crystallization age spectra do not differ significantly onward through the Pliocene Bou Irhardaiene Formation, suggesting that the Taza-Guercif Basin received the majority of its sediment from a consistent source, suggested here to be the Middle Atlas mountains to the south.

There is no indication of increased sedimentation from the Rif mountains as they encroached upon the Rifean Corridor in the Messinian. This suggests that the thermal and contractional uplift of the Middle Atlas played a role in isolating the Taza-Guercif Basin prior to the onset of the MSC in the Mediterranean. With similar effects of Middle Atlas uplift documented in the Saiss basin (Babault et al., 2008), across the Rif-Middle Atlas contact, uplift of the Middle Atlas may have contributed substantially to the closure of the Rifean Corridor.

1 Supplemental Table 1. Detrital zircon U-Pb data. Please visit http://dx.doi.org/10.1130/GES01192.S1 or the full-text article on www.gsapubs.org to view Supplemental Table 1.
2 Supplemental Table 2. Detrital zircon fission-track data. Please visit http://dx.doi.org/10.1130/GES01192.S2 or the full-text article on www.gsapubs.org to view Supplemental Table 2.