Sedimentary provenance in the southern sector of the São Francisco Basin , SE Brazil

Manuscript ID: 20170061. Received on: 04/28/2017. Approved on: 01/22/2018. ABSTRACT: We present new Sm-Nd, U-Pb and Hf isotope geochronological data for the siliciclastic rocks in the southern sector of the São Francisco Basin. An abrupt change in the Sm-Nd data is observed from the Carrancas Formation’s oligomitic conglomerates (TDM ages between 2.7 and 3.3 Ga; εNd(550 Ma) values between -25.09 and -17.98), its finer facies, and the finer facies of the the Moema Laminite (TDM ages from 1.4 to 2.1 Ga; εNd(550 Ma) values between -9.46 and -5.59). No further significant changes in the Sm-Nd data occur farther upwards in the Bambuí Group (Sm-Nd TDM ages within the 1.3-2.0 Ga interval; εNd(550 Ma) values from -9.53 to -4.09), showing a lack of reorganization in the source areas throughout the deposition of the basin. This is yet another argument to dismiss an unconformity within the Bambuí Group. The presence of the index fossil Cloudina sp. in the lower Sete Lagoas Formation makes the glaciation probably Late Ediacaran in age. U-Pb ages for detrital zircons of the Bambuí Group range from the Archean to the early Ediacaran, but the current data is insufficient to distinguish between the contribution from sources in the Brasília and Araçuaí belts.


INTRODUCTION
The geological history of the São Francisco Basin (east-central Brazil; Fig. 1) is a puzzle and has many implications for the West Gondwana assembly at the Ediacaran-Cambrian boundary.A large data set shows that its sedimentary rocks were deposited in a foreland basin fixed over the homonymous craton as a response to the crustal load of the western Brasília Belt (Fig. 1).Intriguingly, the timing of the deposition of most of the sediments is constrained to be younger than 560 Ma (Warren et al. 2014, Paula-Santos et al. 2015), which is much younger than the orogen's metamorphic peak (~630 Ma; Pimentel et al. 1999, Pimentel 2016).As such, this represents early stages of subsidence in the basin.Although the Brasília Belt has a wide range of magmatic rocks spanning through the Ediacaran, the plutonism dated at ca. 540 Ma is considered anorogenic (Pimentel 2016), meaning that refinements in the timing regarding the subsidence and filling of the São Francisco Basin still need to be done.Also, by ca.550 Ma the Araçuaí orogen was already built at the eastern margin of the craton (Fig. 1; Pedrosa-Soares et al. 2011), but its contribution to the basin filling is not fully understood.Seismic sections show that the belt mainly deforms the sediments and there is hardly any sedimentation wedge over the craton (Zalán & Romeiro-Silva, 2007).Exposed rocks that could potentially be sources of sediments dated at ca. 560 Ma are described in the Southern Brasília Orogen (Coelho et al. 2017) and the Ribeira Belt (Heilbron et al. 2008) and they go to the south of the São Francisco craton, so that they may also have contributed to the basin filling.
Furthermore, researchers have not achieved an agreement about the age of the glaciation recorded at the base of the Bambuí Group, the main geological unit of the São Francisco Basin.This unit is comprised of carbonate-siliciclastic rocks that overlie the basal glaciogenic deposits of the Jequitaí Formation and its correlatives in the cratonic and marginal belts areas.The basal unit of the Bambuí Group is the Sete Lagoas Formation (Fig. 2), which is mostly composed of carbonate rocks with very negative δ 13 C values at the base (Alvarenga et al. 2007, 2014, Vieira et al. 2007, Caxito et al. 2012, Kuchenbecker et al. 2016, Paula-Santos et al. 2017).This geological record is in agreement with the Snowball hypothesis (Hoffman et al. 1998, Hoffman & Schrag 2002) that postulates at least three periods of extreme climatic shifts during the Cryogenian and Ediacaran Figure 1.A geological sketch of West Gondwana at ca. 540 Ma, a map of the São Francisco Craton (modified from Alkmim et al. 2006) and location of the samples studied in this work.Samples from the Formiga city area include the Moema Laminite rocks (MG03 27A, 22 and 23A,Tab. 1), siltites from the Sete Lagoas Formation (samples MG05-06, 09, 10, 12 and 13; Tab.1), and siltites from the Serra de Santa Helena Formation (samples MG03- 24 and 25,Tab. 1).Samples from the SAFFRAN quarry are the Moema Laminite siltites (samples MG02-14b, and MG04-41A).In the detailed map, samples MG08-08 and P12 belong to the Carrancas Formation, samples P4, P5, P6, FN-03 and GPP-M are from the Sete Lagoas Formation, samples P2SH, SH-05, SH-06 and SC-SH are from the Serra de Santa Helena Formation, and samples FAS-LJ and LJ-A belong to the Lagoa do Jacaré Formation.( Halverson et al. 2005).These are named Sturtian (~720 Ma), Marinoan (~635 Ma) and Gaskiers (~582 Ma).Babinski et al. (2007) dated the cap carbonates of the Sete Lagoas Formation at 740 ± 22 Ma (Pb-Pb isochron age) and correlated the glaciation of the São Francisco Basin to the Sturtian period.On the other hand, Caxito et al. (2012) and Alvarenga et al. (2014) proposed a Marinoan age for the glaciation, based on isotope chemostratigraphy on the overlying carbonate rocks.However, the presence of the index fossil Cloudina sp. at the middle portion of the Sete Lagoas Formation (Warren et al. 2014) has an important implication for the propositions of these Cryogenian ages.Either an unconformity yet to be found separates the cap carbonates from the rest of the Bambuí Group, or the lowermost part of the Sete Lagoas Formation records the entire Ediacaran period.Based on a stratigraphic and chemostratigraphic analysis, Kuchenbecker et al. (2016) concluded that the existence of an unconformity is unlikely, and proposed that the basal diamictites may be Late Ediacaran in age.
Within this debate we present new Sm-Nd data for the siliciclastic rocks of the Bambuí Group and its basal diamictites.Additionally, U-Pb SHRIMP ages and Hf isotope data of detrital zircon grains retrieved from a sandstone sample of the Serra de Santa Helena Formation (Fig. 2) are also presented.This work is an effort to track the sedimentary sources of the São Francisco Basin and its relationship with the surrounding orogens, as well as to shed light on the age of the diamictites.

GEOLOGICAL SETTING
The São Francisco Basin is comprised of Neoproterozoic-Cambrian sedimentary rocks that cover more than 300,000 km 2 of the homonymous craton in east-central Brazil (Fig. 1).There are exposures of the basement (> 1.8 Ga), which occur in the southern and north-eastern areas of the craton (Teixeira et al. 2000, Alkmim & Martins-Neto 2012).
The base of the sedimentary cover is composed of sparse glacial units of the Jequitaí Formation, interpreted as of continental (Rocha-Campos & Hasui 1981, Karfunkel & Hoppe 1988, Martins-Neto & Hercos 2002) or glaciomarine origin (Rocha Campos et al. 1996, Cukrov et al. 2005, Chaves et al. 2010, Uhlein et al. 2011).In spite of several tectonic discontinuities, the glacial unit has correlating units in the marginal belts: the glacial formations of the Macaúbas Group in the Araçuaí Belt and the Cubatão Formation of the Ibiá Group in the Brasília Belt (Uhlein et al. 1999, Dardenne 2000, Martins-Neto et al. 2001, Pedrosa-Soares et al. 2011, Babinski et al. 2012).In the cratonic area, the Carrancas Formation (Dardenne 1978, Vieira et al. 2007, Uhlein et al. 2012) and the Moema laminites (Rocha-Campos et al. 2011) are at the base of the Bambuí Group in the southern sector of the São Francisco Basin, although the glacial origin of the first unit is debatable (Caxito et al. 2012, Uhlein et al. 2016).
The Bambuí Group is the main unit of the São Francisco Basin and is composed of carbonate-siliciclastic rocks that either overlie the glacial unit or are in sharp contact atop the basement (Dardenne 1978).It records a marine transgression in a foreland basin set over the craton in response to the Brasília and Araçuaí belts crustal overload.The unit crops out throughout the craton and the western Brasília Belt, but not in the eastern orogen (Martins-Neto et al. 2001, Zalán & Romeiro-Silva 2007, Alkmim & Martins-Neto 2012, Reis & Suss 2016).Shear zones with well-developed foliation and stretching lineation separate the basement and sedimentary cover, and the metamorphism in the studied area is low graded.Dardenne (1978) subdivided the Bambuí Group into five units (Fig. 2), from base to top: ■ the Sete Lagoas Formation that is composed of limestones and dolostones with interbedded marls and siltites (thickness ≤ 500 m); ■ the Serra de Santa Helena Formation, which is comprised of mainly siltstones with interbedded shales, sandstones and limestones (640 m); ■ the Lagoa do Jacaré Formation that consists of oolithic limestones, siltstones and marlstones (350 m); ■ the Serra da Saudade Formation, which displays siltstones, green shale and minor occurrences of limestones (100 m); ■ the Três Marias Formation, which is constituted of arkosic sandstones and siltstones (100 m).
The thicknesses here shown were extracted from Iglesias and Uhlein (2009).
The age of the rocks of the São Francisco Basin remains controversial.The basal carbonate rocks of the Sete Lagoas Formation display very negative δ 13 C values, which are similar to other Neoproterozoic cap carbonates (Alvarenga et al. 2007, 2014, Vieira et al. 2007, Caxito et al. 2012, Kuchenbecker et al. 2016).Babinski et al. (2007) obtained a Pb-Pb isochron age of 740 ± 22 Ma in these carbonates, and correlated the glaciation to the Sturtian global event.Other researchers did not agree with this dating estimate and claimed that it was a Marinoan age (~635 Ma), based mainly on the presence of a thin basal pink dolostone, which is recognized worldwide as the start of the Ediacaran (Shields 2005), in addition to the presence of a negative carbon excursion found in these carbonates and in Sr isotope chemostratigraphy (Caxito et al. 2012, Alvarenga et al. 2014).But the recent findings of Cloudina sp.(Warren et al. 2014) and detrital zircons with U-Pb around 560 Ma (Paula-Santos et al. 2015;Fig. 2) in the Sete Lagoas Formation, show that most of the Bambuí Group is Late Ediacaran / Early Cambrian in age.Unless a major unconformity between the cap carbonates and the rest of the deposits of the São Francisco Basin is assumed, the glaciogenic units are also likely to be Late Ediacaran in age (Kuchenbecker et al. 2016).Although some authors point to a sedimentation hiatus at such a stratigraphic position, because of seismic, sedimentary and isotopic features (Martins & Lemos 2007, Zalán & Romeiro-Silva 2007), studies on complete borehole (Kuchenbecker et al. 2016) and field (Perrella Jr. et al. 2017) sections refute that it represents a major regional unconformity.Therefore, a significant time gap at the base of the Sete Lagoas Formation is yet to be described.

Description of the samples
Two conglomerate samples from the Carrancas Formation (Tab. 1) were collected (Fig. 2).The outcrops consist of a monomitic to oligomitic matrix-supported conglomerate, deposited in basement valleys.The clasts are sub-angular to sub-rounded, with sizes ranging between 2 and 20 cm (largest side).They are dominantly composed of igneous rocks from the valley walls and have subordinated quartz vein fragments (Fig. 3A).The matrix has a greenish colour and is comprised of quartz and feldspar silt/sand with minor occurrences of carbonate cement.The size of the clasts decreases upwards and the conglomerate grades to orange / pink siltites with planar lamination (Fig. 3B).One siltite sample was collected.
Nine samples of the Moema Laminites were collected from two different localities.At the SAFFRAN quarry (Fig. 2), mudstone (sample MG04-41A) and varvite facies (MG02-14B) of the glacial succession were sampled.Striations in the bedding plane suggest that the sea ice touched the bottom of the basin during deposition (Rocha-Campos et al. 2011).The remaining samples were collected between Formiga and Bom Despacho cities (Fig. 2), where a sole glacial advance is recorded in a deformed tillite, overlain by a post-glacial laminate (Rocha-Campos et al. 2011).Varvite (n = 2; Tab.1), siltite (n = 3) and mudstone facies (n = 2) of the glacial succession were sampled.
Ten samples from the Sete Lagoas Formation were collected, covering the three Chemostratigraphic Intervals (CI) proposed by Paula-Santos et al. (2017).One marl sample (P6, Tab. 1) interbedded in the CI-1 cap carbonates was retrieved in the Sambra quarry (for detailed stratigraphy see Vieira et al. 2007).We also collected eight samples of siltites from centimetric (5 to 15 cm thick) siliciclastic layers interbedded in the impure carbonate rocks of the CI-2 (Fig 2, Tab. 1).Finally, one marlstone interbedded in the dark grey limestones of the CI-3 (GPP-M) was sampled (Fig. 3C).
Eight samples from the Serra de Santa Helena Formation collected in several locations in the south of the São Francisco Basin were used in this study (Fig. 2).Five of them are siltites, which are the dominant lithotype of the unit (Fig. 3D), and three are sandstones found in some locations.
Finally, two samples from the Lagoa do Jacaré Formation were collected.Sample LJ-A is a fine sandstone interbedded in oolitic dark grey limestones (Fig. 3E), whereas sample FAS-LJ is a siltite (Fig. 3F) described as being within tilted limestones with carbonate intraclasts.

ANALYTICAL PROCEDURES
Thirty-two samples were analysed for Sm-Nd isotopic compositions.Samples P12 from the Carrancas Formation, GPP-M from the Sete Lagoas Formation, P2SH from the Serra de Santa Helena Formation, and LJ-A and FAS-LJ from the Lagoa do Jacaré Formation yielded enough zircon grains for U-Pb dating and Hf isotope analyses.Other samples did not provide enough of these minerals for U-Pb provenance.All of the sample preparation steps and isotope analyses were carried out at the Geochronological Research Center (CPGeo) of the University of São Paulo.
For the Sm and Nd isotopic analyses, whole rock samples were powdered and dissolved with HF, HNO 3 and HCl.Purification of Sm and Nd was performed by standard ion exchange chromatography procedures using RE and LN Eichrom resins.Isotope data were obtained by Thermal Ionization Mass Spectrometry (TIMS) using a Triton mass spectrometer (Thermo Fischer, Germany).The measured 143 Nd/ 144 Nd ratios were normalized to 0.7129 (DePaolo 1981).The mean value of the 143 Nd/ 144 Nd of the reference material JNdi-1 measured in the laboratory was 0.512102 ± 0.000002 (1σ).The Sm-Nd model ages are reported based on the depleted mantle model (DePaolo 1981) and are considered mean crustal residence ages.The f Sm/Nd and ε Nd(550 Ma) parameters were calculated according to Hamilton et al. (1983) for an age of 550 Ma for the sediments (see item 5).
Detrital zircon grains from samples were separated using standard heavy liquid and magnetic procedures.The U-Pb analyses were performed in a multi collector SHRIMP IIe (ASI, Australia).Grains were arranged in an epoxy mount together with the Temora zircon standard (417 Ma;Black et al. 2003), polished to expose their cores, and covered with gold coating for internal cathodoluminescence imaging and dating.To determine the age, five scans through the mass stations were made.U concentrations were calibrated using the SL13 standard (U = 238 ppm, Williams 1998) and a 206 Pb/ 238 U ratio was calibrated using the Temora standard.Decay constants used for age calculations are suggested by Steiger and Jäger (1977).The 204 Pb from each grain was measured for common lead correction.are referred to as 206 Pb/ 238 U ages, whereas the older ones are referred to as 207 Pb/ 206 Pb ages.Uncertainties for the measurements are reported at one sigma level.Hf isotopic analyses were carried out in a multi collector ion coupled plasma mass spectrometer (MC-ICPMS Neptune, Thermo Fischer, Germany), which positioned spots over the ones made by SHRIMP.The average 176 Hf/ 177 Hf for the Mud Tank standard measured in the laboratory was 0.282470 ± 0.000008 (2σ, n = 85), and the average 176 Hf/ 177 Hf for the GJ standard was 0.282015 ± 0.000002 (2σ, n = 334).The initial 176 Hf/ 177 Hf (T1) ratio was calculated using the 206 Pb/ 238 U ages that were previously obtained by SHRIMP.The decay constant 176 Lu = 1.867 × 10 -11 yr -1 of Söderlund et al. (2004) was used.Present chondrite ratios were 176 Hf/ 177 Hf = 0.282772 and 176 Lu/ 177 Hf = 0.0332 (Blichert-Toft & Albarède 1997), and present day depleted mantle ratios were 176 Hf/ 177 Hf = 0.283225 and 176 Lu/ 177 Hf = 0.038512 (Vervoot & Blichert-Toft 1999).Model ages (TDM) were calculated using a two stage model at average crustal 176 Lu/ 177 Hf = 0.015 (Griffin et al. 2002).

Sm-Nd isotope data
Table 1 exhibits the Sm-Nd isotopic results.The conglomerate samples from the Carrancas Formation yielded the oldest Sm-Nd T DM ages: 2.7 and 3.3 Ga (Tab.1; Fig. 4) in samples MG08-04 and MG03-27A, respectively.The ε Nd(550 Ma) values of these rocks are also the most negative, -25.09 and -17.98 for the respective samples above, suggesting sources of long crustal residence time.These values are consistent with the data reported by Guacaneme (2015) for a conglomerate of the same unit in the Inhaúma city area (Tab.1).The author obtained a Sm-Nd T DM age of 2.5 Ga and an ε Nd(550 Ma) value of -23.90.The transition to the siltite facies of the unit is marked by an abrupt shift to a younger age (Fig. 4).Sample P12 displays a Sm-Nd T DM age of 1.5 Ga and an ε Nd(550 Ma) value of -8.26.Although sources of long crustal residence were still active, contribution from younger rocks is more important in the fine-grained facies.The Moema Laminite fine-grained rocks also have model ages that are considerably younger than the conglomerates.The Sm-Nd T DM ages vary from 1.4 to 2.1 Ga with ε Nd(550 Ma) values between -9. .
Overall, the samples from the overlying Bambuí Group yielded Sm-Nd isotopic results that were similar to the finegrained sediments from the Carrancas Formation and the Moema Laminites (Fig. 4).The Sm-Nd T DM ages are within the 1.3-2.0Ga interval and the ε Nd(550 Ma) values range from -9.53 to -4.09 (Tab.1), which also suggests sources of long crustal residence time.The marl sample from the basal CI-1 of the Sete Lagoas Formation has a model age of 1.7 Ga.Obtained ages for the CI-2 are within the 1.5-1.8Ga interval, which is younger than the ages obtained for the same CI by Paula-Santos et al. (2015; 1.8-2.0Ga) in the south-eastern limit of the São Francisco Basin.The marlstone sample (GPP-M) of the upper CI-3 has model ages of 1.6 Ga, which is also younger than the one obtained by those authors (1.8 Ga).The Serra de Santa Helena Formation rocks yielded Sm-Nd T DM ages between 1.5-2.0Ga and the Lagoa do Jacaré Formation samples FAS-LJ and LJ-A have Sm-Nd T DM ages of 1.3 and 1.6 Ga, respectively.The average of the ε Nd(550 Ma) values for the Sete Lagoas Formation is slightly more negative than the average from the combined Serra de Santa Helena and Lagoa do Jacaré formations: -7.85 and -5.93, respectively.

U-Pb ages and Hf isotope data of detrital zircons
Attempts to retrieve detrital zircon grains from the samples were made, but only four samples provided enough minerals for U-Pb and Hf isotope provenance studies: 1. GPP-M from the upper Sete Lagoas Formation (CI-3, Tab.2); 2. P2SH from the Serra de Santa Helena Formation (Tab.3); 3. FAS-LJ (Tab.4); 4. LJ-A (Tab.5), both from the Lagoa do Jacaré Formation.
Only samples P2SH and FAS-LJ provided more than 30 zircon grains within a 90-110% concordance range.Therefore, the calculation of the zircon population probability was performed for only these samples.The Hf isotope data was obtained for the Late Cryogenian / Early Ediacaran grains from sample P2SH as an effort to track the sources of the zircons (Tab.6).The errors are at 1-sigma level.The errors are at 1-sigma level; <LD: below detection limit.
Additionally, eleven grains from sample P12 of the Carrancas Formation were dated, but only six of the ages were within the concordance range (Tab.7).The remaining five grains yielded discordant ages and high common lead concentrations.The concordant U-Pb ages vary from 1922 to 485 Ma.Sample GPP-M provided colourless subeuhedral and reddish rounded grains with the largest side sizes ranging from 80 to 200 μm and igneous oscillatory zoning (Fig. 5).Thirty-two grains were analysed, but twenty-two of them yielded high common lead concentrations and high analytical errors.Only eleven grains yielded ages within the 90-110% concordance range (Tab.2).The Concordia diagram (Fig. 6A) show age concentrations at 800, 1200 and 2000 Ma, but two sets of grains also plot discordias with upper intercepts close to 2200 and 2800 Ma, and lower intercepts towards Neoproterozoic, suggesting Pb loss at this time.
The retrieved grains from sample P2SH are yellowish, euhedral to sub-euhedral, with the largest side of ca being 100 μm, and with igneous oscillatory zoning (Fig. 5).Thirty of the thirty-nine grains that were dated yielded low common Pb concentrations and concordant ages (Tab.3).The U-Pb ages obtained have high analytical errors, even for the zircons with low common lead concentrations (Tab.3), and they vary between 603 and 726 Ma.The Concordia diagrams and the age distribution histogram show a single dominant population aged around 630 Ma (Fig. 6B and Fig. 7A).A Concordia age of 631 ± 5 Ma was calculated (n = 28; Fig. 5B), and the ages older than 690 Ma were suppressed for better results.The ε Hf(t) values of the grains are negative and vary from -13.94 to -5.36 and the Hf T DM ages are between 1.9 to 2.4 Ga (Tab.6; Fig. 8).
Zircon grains retrieved from sample FAS-LJ (n = 52) are colourless, sub-euhedral to sub-rounded and their largest side ranges from 80 to 150 μm.The grains display mainly igneous oscillatory zoning (Fig. 5).Forty-two grains yielded concordant ages (Tab.4).The main age populations (30% of the grains each) are dated at ca.The errors are at 1-sigma level.The errors are at 1-sigma level; <LD: below detection limit.

Sedimentary Provenance and tectonic implications
Our Sm-Nd data suggest little variation in the sedimentary sources active during the deposition of most of the basal glaciogenic units and the Bambuí Group   feeding in response to the marine transgression that flooded the craton.
Provenance studies in the Carrancas Formation conglomerates reveal a wide range of clast assemblage varying from monomitic (Guacaneme et al. 2017) to oligomitic (Kuchenbecker 2011, Vieira et al. 2007) and polimitic (Caxito et al. 2012, Uhlein et al. 2016).The occurrence of the unit described by Guacaneme et al. (2017) in Inhaúma city area is emblematic, as the conglomerate is clearly deposited in a basement valley and only has clasts of the leucogranite, which can be observed in the valley walls (Fig. 9).The U-Pb data of detrital zircons of the Carrancas conglomerate also point to a dominant Archean source (Kuchenbecker 2011, Guacaneme 2015) with only a few zircon grains older or younger than this main source, which is consistent with the age of the basement in the southern sector of the basin (Teixeira et al. 2000).These studies agree with our Sm-Nd T DM ages and negative ε Nd(550 Ma) values (Tab.1; Fig. 4), that suggest Archean basement sources with very long crustal residence time.Similar results were also obtained for the Carrancas Formation by Uhlein et al. (2016).This data set is compatible with a very close igneous source for the conglomerates and a small area of sediment sources, derived mainly from the basement valley walls in which they were deposited (Vieira et al. 2007, Guacaneme et al. 2017).As transgression progressed and flooded the craton, the area of sedimentation extrapolated the limits of these valleys, resulting in deposition of the fine sediments in a larger area and apparently increasing the number of source areas that were initially restricted to the valley walls (Fig. 9).These new sources were potentially younger than the Archean ones.This occurrence is recorded in the transition to the finer facies in the Carrancas Formation, and in the Moema Laminites and their younger Sm-Nd T DM ages and less negative ε Nd(550 Ma) values.The diversification of sources may be also recorded in the U-Pb ages observed in detrital zircon grains of the Carrancas Formation siltites (sample P12, Tab. 7).The most concordant ages (discordance ≤ 3%) are from the Staterian (~1900 Ma) and Ediacaran (600 Ma) periods, with one grain dated at 485 Ma.This young age could bring about several problems concerning the basin evolution, age of glaciation and even the appearance of the Cloudina sp.fossil in the Bambuí Group.But we note that these data are not statistically significant for a provenance study and are not reliable for interpretation.We show the data here so other researchers may have the opportunity to better evaluate them in the future.
On the other hand, the transition from the finer facies of the Carrancas Formation and the Moema Laminites to the Bambuí Group is not marked by any significant changes in the Sm-Nd T DM ages and ε Nd(550 Ma) values.Sm-Nd isotopic data from bulk rock sediments are not adequate to constrain the timing of the deposition, as the T DM ages represent a mixture of sources and rocks with younger detrital zircons that may yield older T DM ages (Paula-Santos et al. 2015).Nevertheless, the striking similarity of data from those units suggests that sources with similar origin were active during most of the deposition of the São Francisco Basin.The database of U-Pb dating in detrital zircons from the Bambuí Group also show a wide range of sources spanning from the Archean to the Early Cambrian (Rodrigues 2008, Pimentel et al. 2011, Kuchenbecker 2014, Paula-Santos et al. 2015, this work), which is consistent with our interpretation for the Sm-Nd data.This adds another argument for the similarity of the sources through the deposition of the lower units and the Bambuí Group.However, tracking whether these sources are located in the Brasília or in the Araçuaí Belt is not just a single task.The T DM ages between 1.3 and 2.1 Ga and ε Nd(550 Ma) values between -9.53 to -4.09 are compatible with the Sm-Nd signatures of igneous rocks from both the western Brasília (Pimentel et al. 1999, Pimentel 2016) and the eastern Araçuaí belt (Pedrosa-Soares et al. 2011).Our exclusively negative ε Hf(t) values do not help solve the question as they mainly point to sources with a long crustal residence time.
The U-Pb ages of the detrital zircon grains from sample P2SH (Serra de Santa Helena Formation) are also dubious.The data indicate a single dominant source of ca.630 Ma (Figs. 6B and 7A).Igneous rocks with this U-Pb age range are also found in both orogens.Collisional magmatism from the Late Cryogenian / Early Ediacaran periods is reported in the Brasília Belt (Valeriano et al. 2008, Seer & Moraes 2013) and in the Rio Doce magmatic arc of the Araçuaí Belt (Pedrosa-Soares et al. 2007, Tedeschi 2013).
The Archean and Paleoproterozoic zircon grains from the studied samples are derived from sources of the São Francisco Basin basement (Teixeira et al. 2000).Sources in the 1400-1050 Ma interval occur in the Mesoproterozoic basement of the Brasília Belt (Pimentel et al. 1999, Pimentel 2016) and in the Araçuaí Belt.Zircon grains from crystal-tuff (Guadagnin et al. 2015) were dated at ca. 1.4 Ga and volcanoclastic rocks (Chaves et al. 2013) with ages close to 1.2 Ga occur in the Espinhaço Supergroup.Detrital zircon populations within this interval age were also obtained in the Macaúbas Group (Babinski et al. 2012) and in the Espinhaço Supergroup (Valladares et al. 2004, Valeriano et al. 2004, Chemale Jr. et al. 2012), both of which are located in the eastern belt, and in the Paranoá, Canastra, Araxá and Ibiá groups of the western belt (Pimentel et al. 2011, Pimentel 2016).These units could have been reworked and they possibly provided sediments to the São Francisco Basin.
The Tonian sources dated around 950 Ma in sample FAS-LJ were also observed in samples GPP-M and LJ-A and may come from the Mayubian and Zanadian Groups, West-Congo orogen (Tack et al. 2001).However, sources with these ages are also found in the Goiás Magmatic arc of the Brasília Belt (Pimentel & Fuck 1992).
Sources dated around 800 Ma could be represented by the metagranitoids of the Arenópolis arc of the Brasília Belt dated at ca. 790 Ma (Laux et al. 2005).Sources with similar ages are also described in the Rio Negro Complex, located from the Ribeira Belt to the south of the São Francisco craton (Tupinambá et al. 2000, Heilbron & Machado 2003).
Finally, zircon grains younger than 540 Ma can also have their sources in the G4 and G5 supersuites of the Araçuaí Belt (Pedrosa-Soares et al. 2011) or in the Novo Brasil wedge of the Brasília Belt (Araújo 2014).In conclusion, our data do not allow for the orogen the sediments came from to be determined.
Studies using seismic sections across the São Francisco Basin suggest that the Brasília Belt is the main source of sediments for the Bambuí Group, with the Araçuaí belt working mainly as a deformer of the unit (Zalán & Romeiro-Silva 2007).Nonetheless, some authors suggest a contribution of sediments from the Araçuaí Belt to the upper Três Marias Formation (Chiavegatto 1992, Martins-Neto & Alkmim 2001).The Late Ediacaran and Cambrian ages also suggest a contribution from both orogens with important implications for the São Francisco Basin evolution.They imply a problematic time gap between the start of the São Francisco Basin subsidence and its filling.The continental collisions at the western margin of the São Francisco craton at ca. 630 Ma (Valeriano et al. 2004, Pimentel 2016) would mark the beginning of the accommodation space in the basin.Unless an unconformity between older units and the Late Ediacaran succession is assumed, the basin would have starved for ca.80 (or 200) m.y., which is not plausible.Some authors point to a possible unconformity between the basal cap carbonates of the Bambuí Group and the rest of the unit (Martins & Lemos 2007, Zalán & Romeiro-Silva 2007, Uhlein et al. 2017).However, sedimentary and isotopic studies in complete borehole and field sections do not support a sedimentation gap of such magnitude at this stratigraphic level (Kuchenbecker et al. 2016, Perrella Jr. et al. 2017).Therefore, we argue that refinements between the subsidence and the filling histories of the São Francisco Basin are required.
Additionally, if the Araçuaí Belt is indeed source to the Bambuí Group, our U-Pb data and other data published so far for the basin may have implications for the eastern orogen.In a recent compilation of the U-Pb ages of detrital zircon from the Bambuí Group, Kuchenbecker (2014) observed that the Neoproterozoic sources of the unit have predominantly a Late Cryogenian to Early Ediacaran age (~630 Ma) with very few Late Ediacaran to Early Cambrian ages, which is consistent with the data obtained in this work.These younger sources are widely exposed in the Araçuaí belt and are grouped in the G3, G4 and G5 supersuites (Pedrosa-Soares et al. 2011).Considering the presence of the Cloudina sp.(Warren et al. 2014) and the age of the occurrence of this index fossil between 550-542 Ma (Grotzinger et al. 2000), it seems that G3 to G4 rocks were scarcely exposed during the deposition of the Bambuí Group, although they were already crystallized.
To summarize, the geochronological data obtained in this work point to only one major reorganization in the source feeding the basin during the evolution of the São Francisco Basin.This event occurs between the conglomeratic and finer facies of the Carrancas Formation and of the Moema Laminites.It is interpreted as an expansion of the sedimentation and the area feeding the basin due to the initial marine transgression that flooded the São Francisco craton.In this scenario, the deposition initially confined within the valleys extrapolated these areas, depositing sediments derived from a larger area.Unfortunately, our data do not allow to distinguish sources that were located in the Brasília Belt, the Araçuaí belt or even in the Ribeira Belt.This question has yet to be answered and is crucial to solve the geological puzzle involving the evolution of the basin and its surrounding mobile belts.

Possible implications for the age of the glaciation
The age of the glacial unit at the base of the São Francisco Basin remains controversial, as several proposals from the Early Cryogenian (Sturtian age; Babinski et al. 2007) (Kuchenbecker 2014) suggest that the deposition of the Bambuí Group may have spanned through the Cambrian.Assuming any of the Cryogenian ages for the glacial deposits, the Late Ediacaran age for the base of the Bambuí Group implies that either the first 100 m of the Sete Lagoas Formation records 85 or 200 Myr of deposition or an unconformity yet to be found exists between the CI-1 and CI-2 carbonates.The first option is highly unlikely, as such durations are not compatible with a second order or a third order sequence (Vieira et al. 2007, Perrella Jr. et al. 2017) subdivision of the lower Sete Lagoas Formation (Vail et al. 1991).Here, we examine the second option and its implications.
Although some works have discussed a possible unconformity at the base of the Sete Lagoas Formation (Martins & Lemos 2007, Zalán & Romeiro-Silva 2007, Uhlein et al. 2017), stratigraphic (Kuchenbecker et al. 2016), sequence stratigraphic (Vieira et al. 2007, Perrella Jr. et al. 2017) and isotope chemostratigraphic data (Paula-Santos et al. 2017, Perrella Jr. et al. 2017) argue against the existence of an unconformity at the base of the Bambuí Group, and suggest a continuous deposition through the cap carbonates to the Cloudina bearing strata.
The C isotope data of the carbonates from the Sapé profile of Perrella Jr. et al. (2017) is especially important, as it shows that the recoveries of the δ 13 C values from very negative to 0‰ are not sharp when the geological record is not condensed or incomplete.Additionally, Paula-Santos et al. (2017) suggested that the initial increase in 87 Sr/ 86 Sr isotope ratios from 0.7074 to 0.7082 in the cap carbonate (Babinski et al. 2007) is not a peak, but rather, they stabilize close to 0.7085 in the CI-2.This also suggests that there is no major hiatus or a long-lived sedimentation history between the cap carbonates and the remaining lower Sete Lagoas Formation.Furthermore, sequence stratigraphy studies point to no sequence boundary between the cap carbonates and the Cloudina bearing strata (Vieira et al. 2007, Perrella Jr. et al. 2017) as previously suggested (Martins & Lemos 2007).Both successions are grouped into one single second or third order sequence.A significant unconformity was also not observed in a complete borehole section at the south of the São Francisco Basin (Kuchenbecker et al. 2016).
In summary, several arguments suggest that no major unconformity or long-lived sedimentation period occurred between the cap carbonates and the rest of the Bambuí Group.We add to these arguments our Sm-Nd data that also point to no major reorganization of the basin sources during this stage.Assuming that no hiatus in the deposition exists, the glaciation may be Late Ediacaran in age, following what was recently proposed by Kuchenbecker et al. (2016).As no extremely negative δ 13 C values similar to the Shuram-Wonoka anomaly (Halverson et al. 2005, Le Guerroué et al. 2006, Halverson & Hurtgen 2007) are reported for the Bambuí Group, we argue that the glaciation may be post-Gaskiers in age (≤ 582 Ma).Glaciations of this period are recorded worldwide (e.g., Frimmel et al. 2006, Hebert et al. 2010, Germs & Gaucher 2012, Ivanov et al. 2013, Etemad-Saeed et al. 2016), so the Bambuí would not be a sole case.
Although we find the possibility of an unconformity difficult to assume, it is fair to analyse the Marinoan and Sturtian propositions.The strongest argument for a Marinoan age proposition is the presence of a thin, patchy, pink cap carbonate (Caxito et al. 2012, Alvarenga et al. 2014).Such dolostone strata are described worldwide as being atop Late Cryogenian diamictites and are considered the base of the Ediacaran period (Shields 2005).Indeed, this is a robust argument, as such rocks are described in many works (e.g., Martins & Lemos 2007, Caxito et al. 2012).However, it is important to note that such cap dolostones are not spread through the whole basin.In many locations, the cap carbonates are pure limestones (Vieira et al. 2007, Kuchenbecker et al. 2016, Paula-Santos et al. 2017) and, although they display other "Marinoan" features (e.g.aragonite pseudomorphs), it is expected that such global strata would be found all over the São Francisco Basin.
Other three pieces of evidence argue in favour of a Marinoan age for the glaciation.First, in the 87 Sr/ 86 Sr ratios, around 0.7075 usually reported for the Sete Lagoas Formation (Alvarenga et al. 2007, 2014, Misi et al. 2007, Paula-Santos et al. 2015), which were interpreted to reflect global seawater values after the Marinoan period (Caxito et al. 2012, Alvarenga et al. 2014).This interpretation was ruled out by Paula-Santos et al. (2017), who suggested that these Sr isotopic ratios are the result of restriction and local factors, and therefore cannot be used for global correlations.Furthermore, those ratios are found in carbonates atop the Cloudina bearing strata.The second piece of evidence is the negative C isotope excursion observed in the cap carbonates (Alvarenga et al. 2007, Kuchenbecker et al. 2016), which is more typical of post-Marinoan cap carbonates (Hoffman & Schrag 2002, Halverson et al. 2005).We note that such an excursion was recently reported for cap carbonates overlying glacial deposits with a maximum depositional age of 560 Ma in Iran (Etemad-Saeed et al. 2016) and is also expected for many glacial periods through the Ediacaran (Hebert et al. 2010).The third piece of evidence is the very positive δ 13 C values of carbonate rocks of the Carrancas Formation that were correlated to the pre-Marinoan "Keele Peak" (Uhlein et al. 2016).Again, these δ 13 C values are also observed in Ediacaran carbonate successions and precede negative C isotope excursions up to the Ediacaran-Cambrian Boundary (Pelechaty et al. 1996, Saylor et al. 1998, Hebert et al. 2010).So, we note that whereas most of the isotopic evidence could serve as arguments for a Marinoan glaciation, it could also be used to sustain a proposal for the Late Ediacaran.
The Sturtian age proposition for the basin glaciation relies on the whole rock Pb-Pb isochron age of 740 ± 22 Ma, obtained in the cap carbonates at the Sambra Quarry (Babinski et al. 2007).This is the only absolute age available for the cap carbonates.The isochron is of good quality, refuting any possibility of analytical problems.Nevertheless, this age would imply a ca.200 m.y.gap between CI-1 and CI-2, which seems inconsistent with other stratigraphic and isotopic evidence.Alternatively, we may be dealing with two cap carbonate levels in the basin, which would correspond to two different diamictites of distinct ages.However, this hypothesis has no evidence to support it, and should be further investigated.
In conclusion, although neither a Sturtian nor a Marinoan age for the São Francisco Basin glaciation can be discarded, the evidence herein discussed suggests that an unconformity at the base of the Sete Lagoas Formation is unlikely to exist.Therefore, the presence of the Cloudina index fossil and of detrital zircon grains younger than 540 Ma in the Bambuí Group suggest that such glaciation is most likely Late Ediacaran in age.

CONCLUSIONS
The new geochronological data obtained in this work provided further insights in the evolution of the São Francisco Basin.Based on the Sm-Nd, U-Pb and Hf isotope data of siliciclastic rocks we state the following: ■ the Sm-Nd isotope data from the lower glacial units and from the Bambuí Group show only one major reorganization of the sedimentary sources through the evolution of the São Francisco Basin.This is observed in the transition from the conglomerates to the finer sediments in the Carrancas Formation and the Moema Laminites, in which the T DM ages abruptly fall from 2.7-3.3Ga to ages younger than 2.1 Ga and the ε Nd(550 Ma) values increase from lower than -17 to higher than -10.These isotope data reflect the change in the sedimentation initially confined to basement valleys, with sediments coming exclusively from the Archean rocks of the valley walls, to a wider basin area with deposition of the finer sediments and a larger source area.During the latter stage, the São Francisco Basin received sediments not only from the homonymous craton basement, but also from rocks located in the marginal mobile belts; ■ T DM ages between 1.3 and 2.1 Ga and ε Nd(550 Ma) values between -9.53 to -4.09 are constantly observed from the siltites and claystones of the Carrancas Formation and the Moema Laminites up to the Lagoa do Jacaré Formation of the Bambuí Group, without significant variations.This adds another argument to support the lack of an unconformity between the lower glacial units and the Bambuí Group.Based on the presence of the index fossil Cloudina sp. in the lower Sete Lagoas Formation, the glaciation recorded in the São Francisco Basin is most likely Late Ediacaran in age.Nevertheless, older age propositions should not be entirely ruled out until further and more robust evidence is found; ■ our geochronological data do not allow tracking whether the Ediacaran sources are located in the Brasília or in the Araçuaí Belt.Nonetheless, with ages younger than 540 Ma occurring in the Bambuí Group, better refinements in the timing between the flexural subsidence and the filling of the basin are required.

Figure 3 .
Figure 3. Field pictures showing: (A) the Carrancas Formation conglomerate in the Inhaúma city area (the pencil is 12 cm long); (B) The Moema Laminite in SAFFRAN quarry (hammer is 23 cm tall); (C) marl layer interbedded in the Sete Lagoas Formation carbonate rocks (sample GPP-M; hammer is 30 cm tall); (D) Serra de Santa Helena Formation siltites in the Lagoa Santa city area; (E) siltites and sandstones interbedded in oolithic limestones from the Lagoa do Jacaré Formation; (F) tilted siltites of the Lagoa do Jacaré Formation (sample FAS-LJ).

63
Figure 5. Cathodoluminescence images of the detrital zircon grains retrieved from samples of the Bambuí Group.

Figure 8 .
Figure 8. εHf evolution diagram for the detrital zircon grains of sample P2SH (Serra de Santa Helena Formation).

Figure 9 .
Figure 9.A sketch showing the change in the source areas during the deposition of the Carrancas Formation.The early rift borders and walls are dominant sources during the beginning of transgression (A), whereas these areas largely expand with the progression of transgression and deposition of the finer facies (B)

Table 1 .
Sm-Nd isotope data from the samples of the São Francisco Basin.

Table 2 .
U-Pb isotope data by SHRIMP in detrital zircon grains from sample GPP-M (Sete Lagoas Formation).

Table 3 .
U-Pb isotope data by SHRIMP in detrital zircon grains from sample P2SH (Serra de Santa Helena Formation).

Table 5 .
U-Pb isotope data by SHRIMP in detrital zircon grains from sample LJ-A (Jagoa do Jacaré Formation).

Table 6 .
Hf isotope data obtained by LA-ICPMS in the detrital zircons from sample P2SH (Serra de Santa Helena Formation).

Table 7 .
U-Pb isotope data by SHRIMP in detrital zircon grains from sample P12 (Carrancas Formation).