The Zircon Story of the Niger River: Time‐Structure Maps of the West African Craton and Discontinuous Propagation of Provenance Signals Across a Disconnected Sediment‐Routing System

The Niger River drains a large part of the West African Craton, where rocks ranging in age from Paleoarchean to recent offer an unexcelled opportunity to map the diverse time structures of sediment sources and provide essential information for provenance diagnoses. In this study, U‐Pb zircon dating is complemented with bulk‐sand geochemical (Zr, Hf, REE) and Nd‐Hf isotope data to pin‐point parent rocks of zircon grains and draw inferences on sediment generation across sub‐Saharan western Africa. In Upper Niger sand, zircon ages pass from exclusively Archean in Guinea headwaters to dominantly Paleoproterozoic in the Inner Delta in Mali, testifying to the progressive dilution by tributary sediment derived from the Birimian domain. Zircon ages abruptly change to dominantly Neoproterozoic downstream of the Inner Delta, becoming indistinguishable from those in Saharan eolian dunes across the Sahel. Most of the sediment generated in the headwaters is thus dumped in the marshlands and bedload is reconstituted downstream by recycling eolian sand. Zircon grains in the Lower Niger yielded virtually the same U‐Pb ages as in Benue sediment, indicating an overwhelming supply from the Benue tributary. In the Niger Delta, however, Archean zircons reappear, and both εNd and εHf values become notably more negative, indicating extensive reworking of sand deposited along the coastal plain at earlier times of wetter climate, when artificial barriers to the sediment flux did not exist in the upper to middle Niger River catchment.


10.1029/2023JF007342
3 of 26 Sahelian Niger only in winter, reaching the Middle Niger well after the white flood characterized by waters laden with kaolinitic sediments that shortly follows the local summer rain season (Ogunkoya, 2023).The Benue River, instead, has only one flood season between May and October.Consequently, discharge in the Lower Niger peaks in November, with a slight rise in winter caused by floodwaters from the Upper Niger.The water budget remains negative across the Sahel from Mali to Niger, and the annual flow entering Nigeria has decreased considerably in the last decades.The water flux is restored across Nigeria, where large contributions from tributaries, including the Kaduna (length 575 km, basin area ∼66,000 km 2 ) and especially the Benue (length 1,400 km, basin area ∼340,000 km 2 ), fuel an average water discharge of ∼270 km 3 /year at the Niger Delta (FAO & IHE Delft, 2020).
The suspended sediment and dissolved loads for the entire Niger River were estimated as ∼40 and ∼10 million tons/a, respectively (Hay, 1998;Milliman & Farnsworth, 2011).Based on gauged data (NEDECO, 1961), annual bedload and suspended load were assessed to be 0.3 and 4.6 million tons for the Middle Niger and 0.6 and 11 million tons for the Benue River, respectively (Adegoke et al., 2017).Middle Niger load has been reduced by impoundment in major artificial reservoirs, including Kainji Lake in western Nigeria (dam completed in 1968), the hydropower Jebba Dam ∼100 km downstream (completed in 1984;Yue et al., 2022), and the Shiroro Dam on the Kaduna tributary ∼230 km upstream of the confluence with the Niger River (completed in 1990; Daramola et al., 2022).The ∼450 km-long Lower Niger forms the largest delta in Africa (∼19,000 km 2 ), characterized by high distributary density and extensive development of tidal flats and freshwater marshes (Allen, 1965;Coleman et al., 2008;Kuenzer et al., 2014).The wave-modeled lobate shoreline reflects high nearshore wave action and sediment transport by divergent bi-directional longshore currents (Anthony, 2015).(Caby, 1987;Ferré et al., 2002).

Geology of the Niger Catchment
The Niger River cuts across the southern part of the West African Craton, welded during the Eburnean (Birimian) orogeny around 2.1 Ga.Archean rocks represent the core of the Man Shield in Guinea, the basement of the Taoudeni basin in Mali, and the core of the Nigeria basement complex (Figure 1).
After drawing a great arc across the southern edge of the Sahara Desert, the Niger River in Niger flows between the edge of the Paleoproterozoic Birimian rocks of the Leo Shield in the west, widely exposed in Burkina Faso and including medium-high grade gneiss and migmatite granite, and the Neoproterozoic covers and Cenozoic sediments of the Iullemeden-Sokoto Basin encircled by Cretaceous epicontinental marine deposits in the east (Figure 1).Downstream, the Niger River flows along the northernmost margin of the Dahomeyide Belt, accreted in the Neoproterozoic during the collision between the West African Craton and the Nigerian Shield (Affaton et al., 1991;Attoh et al., 1997;Caby, 1987).Gneisses, metagabbros and metadiorites mostly yielded Neoproterozoic ages clustering between 650 and 550 Ma (Affaton et al., 2000;Kalsbeek et al., 2012), but Paleoproterozoic gneisses dated as ∼2.1 Ga also occur (Ganade et al., 2016;Glodji, 2012).
In Nigeria, the Middle Niger flows along the Bida Basin (Rahaman et al., 2019) as the Benue River flows along the Benue Trough (Abubakar, 2014), both being failed arms of the Cretaceous rift system associated with the opening of the northern South Atlantic Ocean (Burke et al., 1971;Fairhead et al., 2013;Obaje et al., 2020).The Bida rift trough separates the SW Nigeria basement, drained by the Oli tributary of the Niger River and by the Ogun River draining directly into the Gulf of Guinea, from the NW Nigeria basement, drained by the Ka and Kaduna tributaries of the Niger River.The Benue rift trough separates the eastern part of the North Nigeria basement drained by the Okwa, Mada, Dep (Ankwe), and Shemanker tributaries of the Benue River from the Bamenda Massif in the southeast, drained by the Katsina-Ala tributary (Figure 1).
Ring complexes comprising alkali-feldspar granites of mostly Jurassic age (190-145 Ma;"Younger granites";Obaje, 2009), associated with rhyolite and minor gabbro and syenite, occur in the Jos Plateau in the eastern part of the North Nigeria basement (Bowden & Kinnaird, 1984;Ngako et al., 2006).Ring complexes display a southward-younging trend, with the oldest intrusions dated as ∼330 Ma in the north (Rahaman et al., 1984;Vincent et al., 2022).The volcanic activity is documented by the intrusion of trachyte-phonolite plugs in the Benue Trough during the Miocene (22-11 Ma), followed by widespread eruption of basaltic lava since 7 Ma (Grant et al., 1972).

Methods
Thirty-four samples of river sand were collected from active fluvial bars through the entire Niger catchment, 11 in the headwaters in Guinea and Mali in June to early July 2022 (six from the mainstem, five from major tributaries), three from the mainstem across the Sahel in Mali and Niger, eight from the Middle Niger (four each from the mainstem and tributaries), seven from the Benue (three from the mainstem, five from tributaries), four from the Lower Niger and distributaries in the Niger Delta, and one from the Ogun coastal river.Data from three samples collected from Saharan dunes on both sides of the Niger course, illustrated in Pastore et al. (2021), were also considered.Full information on sampling sites is provided in Appendix S1 in Supporting Information S1 and Google Earth® file NigerDZ.kmz.

Petrography, Heavy Minerals, Geochemistry
Petrographic analyses were carried out by counting 450 points in each thin section of bulk-sand samples using the Gazzi-Dickinson method (Ingersoll et al., 1984).Sand was classified according to the three main groups of framework components (Q, quartz; F, feldspars; L, lithic fragments), considered if exceeding 10%QFL and listed in order of abundance (classification scheme of Garzanti (2019)).Heavy-mineral analyses were conducted by point counting of ∼200 transparent minerals with density >2.90 g/cm 3 on grain mounts of the 15-500 μm fraction obtained by wet sieving of sand samples (Garzanti & Andò, 2019).The equivalent diameter (d e ) was determined by image analysis as the geometric mean of the long and short axes of each zircon grain identified in the studied samples by semi-automatic Raman spectroscopy.The mean size (in ϕ units), sorting (σϕ), and skewness (Sk) of the d e distribution were calculated following Folk and Ward (1957).

Detrital-Zircon Geochronology
From the heavy-mineral separation of 32 fluvial sands obtained by wet sieving (15-500 μm fraction) and of three bulk samples of eolian-dune sand, detrital zircons were concentrated with standard magnetic techniques, directly mounted in epoxy resin without any operator selection by hand picking, and identified by automated phase mapping (Vermeesch et al., 2017) with a Renishaw QONTOR™ Raman microscope.U-Pb zircon ages were determined at the London Geochronology Centre using an Agilent 7900 laser-ablation inductively coupled-plasma mass-spectrometry system, employing a NewWave NWR193 Excimer Laser operated at 10 Hz with a 25 μm spot size and ∼2.2 J/ cm 2 fluence.No cathodoluminescence imaging was performed, and the laser spot was always placed blindly in the middle of zircon grains to treat all samples equally and avoid bias in intersample comparison ("blind-dating approach" discussed in Garzanti et al. (2018)).The mass spectrometer data were converted to isotopic ratios using GLITTER 4.4.2software (Griffin et al., 2008), employing Plešovice zircon (Sláma et al., 2008) as primary age standard and GJ-1 (Jackson et al., 2004) and 91500 (Wiedenbeck et al., 2004) as secondary age standards, obtaining average ages of 592.3 ± 0.9 (MSWD 2.4, n = 65) and 1,022.4± 3.1 (MSWD 0.8, n = 16), respectively.A NIST SRM612 glass was used as a compositional standard for U and Th concentrations.GLITTER files were post-processed using IsoplotR 5.3 (Vermeesch, 2018).Concordia ages were calculated as the maximum likelihood intersection between the concordia curve and the error ellipse of 207 Pb/ 235 U and 206 Pb/ 238 U (Ludwig, 1998).The discordance cutoff was set at −5/+15 (Vermeesch, 2021).Zircon-age data are plotted as kernel density estimates (KDE) using the provenance package of Vermeesch et al. (2016).Statistical techniques used to illustrate our data set include multidimensional scaling (MDS), which produces a map of points in which the distance among samples is approximately proportional to the Kolmogorov-Smirnov dissimilarity of their compositional or chronological signatures; the goodness of fit is evaluated using the "stress" value of the configuration (0.2 = poor; 0.1 = fair; 0.05 = good; Vermeesch, 2013;Vermeesch & Garzanti, 2015).The complete geochronological data set of 6677 ages, 4299 of which are considered concordant (64% on average but ranging from only 28%-42% in Guinean headwaters to 78%-89% for Benue tributaries), is provided in Appendix S2 in Supporting Information S1.
Forward mixing calculations of zircon-age populations were made by both inverse Monte Carlo modeling with Kolmogorov-Smirnov and Kuiper test statistics (Sundell & Saylor, 2017) and Wasserstein statistics, a refined method more sensitive to the tails of the distribution (Lipp & Vermeesch, 2023).The best fit is obtained for the synthetic mixing proportion yielding the minimum Wasserstein distance to the spectrum of the outlet sample.

Upper Niger
In uppermost Niger sand, all ages are Archean (11% Neoarchean, 88% Mesoarchean, 1% Paleoarchean) (Figure 2).The youngest zircon age is 2.65 Ga (Liberian), the oldest is 3.39 Ga (Leonian), and the main peak is at 2.88 Ga, revealing provenance exclusively from the Man Shield.Only 42% of the obtained ages could be considered concordant.Archean ages prevail over Paleoproterozoic ages in Milo sand, whereas Paleoproterozoic ages prevail in Tinkisso sand and are overwhelming in Sankarani, Baoulè and Bagoé sands in the northeast (Table 2).

Table 1 Continued
and subordinately Mesoarchean ages in the Inner Delta to dominantly late Neoproterozoic ages downstream indicates that Upper Niger sediments are efficiently sequestered in the Inner Delta, and that Sahelian Niger sediments are virtually entirely reconstituted from erosion of eolian dunes across the edges of the Sahara sand sea (Figure 3).

Niger River in Nigeria
Neoproterozoic zircon grains are dominant in river sand throughout Nigeria, reflecting provenance from the Pan-African Trans-Saharan Belt (Figure 4).Central age, standard deviation, and proportions are calculated with IsoplotR 5.3 (Vermeesch, 2018), implementing the discrete mixture modeling algorithms of Galbraith and Laslett (1993).

Table 2 U-Pb Age Peaks of Detrital Zircon Identified in Each Studied Sample
Gulf of Guinea.Ogun sand contains a larger proportion of Paleoproterozoic zircons (41%, mainly Rhyacian with peak at ∼2.1 Ga) associated with late Neoproterozoic grains (33% Ediacaran, 20% Cryogenian).

Age Discordance and Degree of Metamictization
The percentage of concordant ages decreases regularly in samples with older zircon grains, which reflects a higher degree of metamictization consequent to the progressive accumulation of radiation dose since crystallization (Resentini et al., 2020).A lower degree of crystallinity owing to metamictization favors lead diffusion and weathering.A great number of very old zircon grains in Guinea show a linear trend of Pb diffusion/common lead mixing, and their highly metamict state is confirmed by the correlation with elements (e.g., lanthanum) not originally present in the zircon lattice but detected in significant amounts in weathered radiation-damaged crystals (Andersen & Elburg, 2022;Andersen et al., 2019).
Only 28%-42% of the obtained ages are concordant in Upper Niger headwater sands draining the Archean Man Shield in Guinea (samples 6250, 6253, 6254, 6257).This percentage reaches 48%-60% in sand mostly generated in the Paleoproterozoic Baoulé-Mossi (Birimian) domain in Mali, and systematically higher values downstream of the Inner Delta (61%-75% in Saharan dunes, 76%-78% in the Sahelian Niger; 75% ± 10% in Nigeria), where most zircons yielded Neoproterozoic ages.The percentage of concordant ages increases with grain size in Middle Niger (from 55% to 64% in very fine sand to 70%-82% in coarser sand), Benue (from 70% in very fine sand to 78%-89% in coarser sand), and Lower Niger samples (from 72% in very fine sand to 78%-81% in coarser sand).Such a distinct behavior is not an artifact caused by sample preparation or by mixing age domains during analysis, but it is a genuine characteristic of moderately to strongly metamict Upper Niger zircon populations, in contrast with the moderately high crystallinity of zircon grains entrained by the Sahelian Niger, Middle Niger, Benue River, and Lower Niger (Figure 6).

Zircon and Zr Concentrations
Zircon concentration can be independently estimated based on mineralogical and geochemical data in the diverse reaches of the Niger drainage basin (Table 1).In the Upper Niger catchment, integrated petrographic and heavy-mineral data indicate that zircon represents 0.14% ± 0.09% of the sand shed from largely Archean rocks in Guinea, about twice as much as for sand shed from the Paleoproterozoic Baoulé-Mossi (Birimian) rocks exposed in the Bani catchment (0.06% ± 0.03%).Because of extensive recycling of zircon-poorer Proterozoic sandstones of the Siguiri and Taoudeni basins, zircon concentration in Upper Niger sand decreases downstream in Mali, attaining a median value of 0.05%.This is confirmed by geochemical data: Zr concentrations are invariably higher in Guinea tributaries (Zr 346-1,428 ppm, average 749 ppm) than in the Bani catchment (Zr 98-313 ppm) and decrease steadily from Guinea headwaters (Zr 746 ± 77 ppm) to Mali (Zr mostly <100 ppm).
Zircon and Zr concentrations decrease further in Sahelian Niger sand, where they are as low as in Saharan dunes (zircon 0.03% ± 0.01% vs. 0.03% ± 0.03%, Zr 87 ± 12 vs.72 ± 17 ppm), and increase again progressively in Nigeria, indicating that zircon fertility of granitoid basement rocks is higher than for sedimentary covers by up to an order of magnitude.Zircon and Zr concentrations are 0.06% ± 0.06% and 174 ± 138 ppm for Middle Niger mainstem and tributaries, and highly variable in the sand of Benue mainstem and tributaries (0.07% ± 0.06% and mainly 88 ± 30 ppm but locally up to 624 and 1,708 ppm in Mada and Dep sands, respectively).In the Lower Niger sand, zircon and Zr concentrations vary strongly from 0.008% to 37 ppm to 0.12% and 630 ppm.

HREE Patterns
Because zircon contains significant amounts of the heaviest REEs, HREE patterns in Niger sand are largely controlled by zircon content.The Gd N /Yb N ratio correlates negatively with Zr concentration (r = −0.53,sign.lev.<0.1%) and Zr and Hf concentrations correlate best with the concentration of Yb and Lu (r = 0.89-0.92).HREE fractionation varies notably (Gd N /Yb N from 0.4 in Dep sand to 3.3 in Kaduna sand), and it is lower in Upper Niger (1.2 ± 0.5), Saharan-dune (Gd N /Yb N 1.3 ± 0.1), and Middle Niger sands (Gd N /Yb N 1.5 ± 0.1) than in Benue (Gd N /Yb N 2.0 ± 0.7) and Lower Niger sands (Gd N /Yb N 1.9 ± 0.6).In a very fine to fine quartz-rich sand deposited by a tractive current, zircon grains are expected to be ∼0.6 ϕ finer than bulk-sample grain size (Garzanti et al., 2008). 10.1029/2023JF007342 13 of 26 By considering that the Lu/Hf ratio is much lower in zircon (∼0.002;Kinny & Maas, 2003) than in the upper continental crust (∼0.055;Taylor & Mclennan, 1995), the average Lu/Hf value of 0.026 (although with wide variations among samples from 0.006 to 0.072) implies that zircon accounts for at least half of the Lu budget overall.
Within our sample set, the Zr/Hf ratio is much lower in the Upper Niger catchment (42 ± 3) than in Saharan dunes (60 ± 2) and ranges widely in Nigeria (between 46 and 55 in the Middle Niger catchment, and between 41 and 61 in the Benue catchment).Downstream of the Benue confluence, Zr/Hf increases from 51 to 52 in medium sand of the Lower Niger to 61-69 in fine sand of deltaic distributaries, which is significantly higher than in Ogun sand (47).
The seven highest Zr/Hf ratios were obtained for pure quartzose Saharan-dune sands (58-62; zircon accounting for 0.01%-0.06% of total sample), quartz-rich feldspatho-quartzose to pure quartzose delta-distributary sands (61-69; zircon accounting for 0.05%-0.06% of total sample), and feldspatho-quartzose Dep and K-feldspar-rich feldspatho-quartzose Mada sand (57-61; zircon accounting for 0.08%-0.25% of total sample), indicating common granite protosources.Ratios close to the expected value for mud (Zr/Hf 36 ± 3) were obtained from the three very coarse silt samples collected upstream and downstream of the Niger/Benue confluence.In sand, ratios significantly lower than expected were obtained across the Upper Niger catchment (Zr/Hf 42 ± 3) but in Nigeria only for coarse quartzose Okwa sand (Zr/Hf 41), signaling the only local occurrence of evolved granites in the Trans-Saharan Belt.

Zirconium Budget
In zircon crystals grown in a variety of igneous and metamorphic rocks, the concentration of Zr is on average ∼49% (close to the stoichiometric value of 49.76%), with the Zr/Hf ratio ranging widely between 20 and 70 (Bea et al., 2006).Most common minerals have Zr concentrations well below that of the average upper continental crust (i.e., ∼200 ppm; Rudnick & Gao, 2003) and among them only titanite, some amphiboles, or allanite may occasionally reach a Zr concentration well above 1,000 ppm (Deer et al., 1997).Such minerals, however, are not sufficiently abundant to have a significant impact on Zr and Hf budgets (Garzanti & Andò, 2007).Among common silicates, only quartz may be abundant enough, despite its low Zr and Hf concentration (indicatively 35 and ∼1 ppm, respectively; Garzanti et al., 2010).
Based on Zr and Hf concentration in different detrital minerals (data after Bea et al. (2006)), zircon is estimated to account on average for 87% of the Zr budget, and feldspars for less than 3% for Sahelian and Middle Niger sands but up to ∼5% for uppermost Niger and Benue sands.Quartz is estimated to account for 11% of the Zr budget on average, but even for a half or more in a few pure quartzose samples.No correlation was observed between the Zr/Hf ratio and the estimated relative contributions of zircon and quartz.

Neodymium and Hafnium Isotopes
Isotopic ratios represent very useful additional tools to unravel provenance and constrain sediment budgets.Specifically, hafnium isotopes in bulk sand represent a valuable complement to detrital-zircon ages, considering that no less than 65%-70% of upper crustal hafnium is estimated to be hosted in zircon (Van de Flierdt et al., 2007).

The Zircon Budget
In this section, U-Pb age distributions are considered as fingerprints of different source-rock domains and used to quantitatively assess the provenance of zircon grains in different parts of the Niger catchment.It must be underscored that the results of such an exercise cannot be directly translated into a sediment budget for a variety of reasons (Vezzoli et al., 2016).First, these estimates would need to be corrected for zircon concentrations, which may vary strongly among different potential sediment sources and can hardly be assessed with the required precision (Malusà et al., 2016).Second, sediment composition cannot be safely assumed as grain-size invariant, and results based on bedload sand may not be applicable to the mud fraction carried in suspension, which represents the largest part of the sediment flux by far.Third, inhomogeneities in sediment mixing in the fluvial system are caused by different grain-size distributions and/or seasonal variations in tributary versus mainstem sediment fluxes, local reworking of loose deposits (e.g., terraces, alluvial fans at confluences, eolian dunes) and anthropogenic effects (e.g., dam construction).Consequently, the estimated relative contributions suffer from the intrinsic variability of complex natural phenomena and are non-unique and uncertain (Garzanti et al., 2012;Resentini et al., 2017), even more so when they are based on a rare mineral such as zircon that occurs only in a limited range of source rocks (chiefly granitoids/metagranitoids and sandstones/metasandstones) and represents far less than a ten thousandth of the sediment load.Moreover, the unmixing of detrital-geochronology data can be done in various ways and under a range of assumptions, which, on the other hand, allows us to verify the robustness of the outcomes and obtain an empirical estimate of the associated uncertainty (Table 3).

Zircon Provenance Budget
Zircon mixing proportions in various reaches of the Niger course were estimated by combining different approaches (Table 3).Zircon grains in our Inner Delta sample are calculated to be derived slightly more from the Bani River (51%-70%) than from the Upper Niger (30%-49%), and most zircon grains in the Sahelian Niger (88%-94%) appear to be recycled from Saharan dunes, with only a minority (≤12%) derived from upstream of the Inner Delta.This reflects transport disconnection between the Upper Niger and the Sahelian Niger because of sediment dumping in the Inner Delta (Figure 7).

Inferences on Sand Provenance
The zircon provenance budget (Table 3), combined with zircon and zirconium concentrations appraised from integrated petrographic, heavy-mineral, and geochemical data in different parts of the Niger drainage basin (Table 1), allows us to tentatively draw semi-quantitative indications on bulk-sand provenance.If our assumptions are correct, then a zircon concentration of 0.05%-0.06% in both Upper Niger and Bani sands implies that our Inner Delta sample was fed in subequal proportions by the Upper Niger and its Bani tributary.Three explanations are viable: (a) sand production is more efficient in the Bani catchment than in the Upper Niger catchment, which is however in contrast with notably greater water discharge and sediment yields for the Upper Niger (7-8 t km −2 a −1 ) than for the Bani River (3 t km −2 a −1 ; Olivry, 1995Olivry, , 1998)) (Sundell & Saylor, 2017) A zircon concentration of 0.06%-0.07%is estimated for both Middle Niger mainstem and tributaries and Benue mainstem and tributaries, indicating that most (∼90%) Lower Niger sand is supplied by the Benue River rather than by the Middle Niger.The predominance of Benue supply is confirmed by forward mixing calculations based on Nd isotope data (Table 1), suggesting that 89% ± 9% of the REEs in Lower Niger sand are derived from the Benue River and 11% ± 9% from the Middle Niger.Hf isotope data confirm a predominant, although less overwhelming, prevalence of Benue contribution (77% ± 5%).More efficient sediment production in the Benue catchment than in the Middle Niger catchment is indicated unless a substantial amount of Middle Niger sediment is sequestered in large reservoirs upstream (e.g., Kainji and Shiroro lakes; Daramola et al., 2022;Ogunkoya, 2023).

Segmented Sediment-Transport Systems: Causes and Consequences
In most modern river systems, a continuous sediment flux from the headwaters to the mouth has been prevented by extensive human activities, including the construction of countless small, medium, and large dams since the dawn of civilization.The second half of the last century has seen an acceleration of dam construction worldwide for flood regulation, water supply and hydropower, including huge dams capable of sequestering virtually all sediment load in the reservoir behind.As a result, sediment delivery to the ocean from big rivers such as the Nile or the Indus has been drastically reduced to virtually zero (Stanley & Warne, 1998;Syvitski et al., 2013).This is partly true also for the Niger River, where dams built both on the Upper Niger and Middle Niger have reduced the sediment flux downstream by an undetermined amount.
Segmentation of fluvial sediment transport, however, is by no means an exclusively Anthropocene phenomenon and it has natural causes as well.Natural barriers to sediment-transport continuity include large lakes found in both orogenic and anorogenic settings.For instance, detritus generated in the headwaters of rivers sourced from the entire East African rift, from the Shire River in Malawi to the White Nile in Uganda, is trapped in large natural lakes.Similarly, detritus generated in the headwaters of Alpine rivers is blocked in lakes shaped by glaciers along both the northern and southern flanks of the European Alps (Hinderer, 2001).Another effective barrier to sediment transport is represented by vast marshlands formed in subsiding areas of cratonic interiors, such as the Sudd on the White Nile in South Sudan (Garzanti et al., 2015) and the Pantanal on the Paraguay River in South America (Assine et al., 2015), or by inland deltas such as those formed by the Okavango River in Botswana ( McCarthy & Ellery, 1998) and by the Niger River in Mali (Olivry et al., 1994).The temporaneous or permanent segmentation of a fluvial sediment-transport system implies that detritus generated in the highlands is temporarily or permanently stored in the continental interiors and cannot make it to the ocean.The consequent major impact, not only on sediment budgets but also on sediment composition (Garzanti et al., 2021), must be given due consideration in provenance analysis.

Time Structure of Source Rocks
In provenance analysis, the diverse time-structure maps of source rocks obtained from the determination of isotopic ratios in sediments constitute a valuable complement to the description of their lithological structure as obtained by traditional petrographic and mineralogical approaches (Garzanti et al., 2018).As a function of the targeted mineral and isotopic system, each technique reproduces a different age distribution across source areas, thus providing complementary information on their evolution through successive episodes of crustal growth.The age map based on U-Pb zircon dating, which provides an overview of igneous and metamorphic crystallization ages, is here complemented with two maps based instead on mantle extraction ages (i.e., on the inferred time of separation of continental crust from the mantle based on a geochemical model of the evolving crust-mantle system) calculated from Sm-Nd and Lu-Hf isotopic ratios (Figure 9).
The West African Craton drained by the Niger River contains three main geological domains: the Archean (Leonian-Liberian) Man Shield, the middle Paleoproterozoic (Eburnean) Leo Shield (Baoulé-Mossi domain), and  the Neoproterozoic (Pan-African) Trans-Saharan Belt (Table 4).The distribution of zircon-crystallization ages and bulk-sediment model ages in sediment derived from these three main protosources and transported across the Niger drainage basin are highlighted below.

Archean Protosources
Zircon grains eroded from basement rocks of the Man Shield mostly crystallized during the Liberian episode of crustal growth and yielded an average sample age of 2877 ± 11 Ma.A subordinate number of zircons yielded older Leonian ages (average sample age 3203 ± 19 Ma) (Table 2).Archean zircons represent the entire population in headwater Niger sand and are progressively diluted downstream by zircon grains supplied by younger Proterozoic rocks all along the Niger sediment-routing system.Downstream of the Inner Delta, Leonian zircons show up only sporadically in Saharan dunes but are traced in a very minor amount as far as the Niger Delta (Table 2).Liberian zircons are significant in Saharan dunes and in the sand of the Sahelian Niger and the Sokoto tributary of the Middle Niger, but they show up downstream only in the Niger Delta.It is noteworthy that a minor population of younger, late Neoarchean zircons, occurring in Saharan dunes, Sahelian Niger, and Sokoto tributary sands, also shows up again in the Lower Niger and Niger Delta.Because Archean zircons are negligible throughout the Benue catchment, the reappearance of a few but significant Leonian, Liberian, and late Neoarchean zircons in the Niger Delta indicates reworking of sediment deposited along the coastal plain at earlier times of presumably wetter climate, when neither natural nor artificial efficient barriers to the sediment flux existed along the entire Niger sediment-routing system from Guinea headwaters to the ocean.

Paleoproterozoic Protosources
Zircon grains eroded from igneous and metamorphic rocks of the Baoulé-Bossi domain crystallized during the Eburnean (Birimian) episode of crustal growth and yielded an average sample age of 2072 ± 14 Ma).Vermeesch, 2018), implementing the discrete mixture modeling algorithms of Galbraith and Laslett (1993).Full data set of 496 ages from 107 literature sources and information on adopted methods and references are provided in Appendix S3 in Supporting Information S1.    2).These data reflect extensive Pan-African metamorphic recrystallization of Archean gneisses and Paleoproterozoic schists representing the core of the Nigeria basement complex, and widespread intrusion of the "Older granites" during growth of the Trans-Saharan Belt.
In  2).Early Calymmian magmatism is in fact reported across the study area, from the Leo Shield in Burkina Faso (Baratoux et al., 2016) to southeastern Nigeria (Ukwang et al., 2012).
Zircons younger than 500 Ma do not occur in the Upper Niger but are present in Saharan dunes and Sahelian Niger, where Permian-Carboniferous to Jurassic zircons become significant (Table 2).Jurassic zircons derived from the "Younger granites" widely exposed in the Jos Plateau occur in sand of the Middle Niger and of its tributaries (their absence in Kaduna sand may be explained by sequestration in the Shiroro reservoir), of the Benue River and of most of its tributaries, and of the Lower Niger as far as the Niger Delta.Cretaceous/Eocene zircons were detected in a few tributaries (Ka, Katsina-Ala), in the Lower Niger, and in the Niger Delta, where a few zircons as young as the Pleistocene reflect recent eruptive events along the Cameroon Line.

Conclusions
In After the extensive worldwide construction of large dams in the last century, the segmentation of fluvial sediment transport has become the rule for most modern river systems, although this is by no means an exclusively artificial phenomenon.In anorogenic settings, very efficient natural barriers to sediment transport include rift-related lakes as well as vast marshlands formed in subsident regions of the cratonic interiors, such as the Inner Delta in Mali or the Sudd marshes formed by the White Nile in South Sudan.These temporary or even permanent barriers prevent a large part of the detritus generated in the highlands from reaching the ocean, with consequent impact on both sediment budgets and sediment composition, a further unknown that must be given due consideration when solving the complex system of equations called provenance analysis.

Figure 1 .
Figure 1.The Niger River in southern West Africa.The main geological domains and their bedrock age signatures are indicated, together with sampling sites and sample numbers.The Adrar des Iforas, Air Mountains, and Nigeria and Cameroon basements are all part of the Pan-African Trans-Saharan Belt, which binds the West African, Saharan, and Congo cratons(Caby, 1987;Ferré et al., 2002).
Along the Upper Niger course, detrital zircons of Archean age are progressively diluted by Paleoproterozoic zircons supplied by tributaries draining the Birimian Baoulé-Bossi domain, which become predominant in the Inner Delta.Archean zircons display a major Liberian peak (2.88 Ga, n = 433) and a minor Leonian peak (3.20 Ga, n = 20); Paleoproterozoic zircons display a latest Rhyacian peak (2.07 Ga; n = 560).A few Pan-African ages appear as the rivers cut across the Neoproterozic strata of the Taoudeni Basin (Ediacaran peak at ∼570 Ma; n = 31).

Figure 2 .
Figure 2. U-Pb age spectra of detrital zircons (white panels = Niger River; blue panels = tributaries; yellow panels = eolian dunes).Headwater Niger sand yielded exclusively Archean (>2.65 Ga) zircons derived from the Man Shield, Bagoé sand overwhelmingly ∼2.1 Ga Birimian zircons.The MDS map in the bottom right panel highlights the disconnection of sediment transport across the Inner Delta; sediment load is reconstituted by recycling of Saharan dunes downstream (Figure 3).

Figure 3 .
Figure 3. Fluvial-eolian interaction at the southern edge of the Sahara Desert.The flow of the Niger and Bani rivers is hampered at the entrance of the dune field, and sediment derived from Mesoarchean to Paleoproterozoic rocks exposed in the headwaters is dumped in the Inner Delta marshes.Downstream, the reduced discharge is partitioned in many streamlets that meander around E/W-trending dunes, while bedload is reconstituted by recycling of eolian sand with a late Neoproterozoic zirconage signature.

Figure 4 .
Figure 4. U-Pb age spectra of detrital zircons (white panels = Niger and Benue rivers; pink panels = Middle Niger tributaries; green panels = Benue tributaries; lilac panel = coastal river).Ediacaran zircons are dominant throughout the Benue catchment (but for the Dep River carrying mostly Tonian grains), whereas Paleoproterozoic zircons commonly occur in Middle Niger tributaries and especially in the Ogun River sourced in the SW Nigeria basement.

Figure 5 .
Figure5.The size distribution of detrital zircons in the Niger catchment (n° = number of analyzed zircon grains; boxes extend from the 25th to the 75th percentile, median is indicated by central red mark and outliers by red crosses; gray bars show average size of bulk 15-500 μm fraction).In a very fine to fine quartz-rich sand deposited by a tractive current, zircon grains are expected to be ∼0.6 ϕ finer than bulk-sample grain size(Garzanti et al., 2008).

Figure 6 .
Figure6.The degree of zircon metamictization revealed by v 3 (SiO 4 ) Raman peak position(Nasdala et al., 1995; n° = number of analyzed zircon grains; boxes extend from the 25th to the 75th percentile, median is indicated by central red mark and outliers by red crosses).Archean to Paleoproterozoic zircons in Upper Niger sand yield a higher percentage of discordant ages (53% ± 11% vs. 25% ± 9% in other sands) and display a higher degree of metamictization [i.e., v 3 (SiO 4 ) peak shifted to lower wavenumbers].
Epsilon values become progressively more negative downstream of the Middle Niger, both in the mainstem (ε Nd and ε Hf down to −20 and −33, corresponding to Paleoproterozoic model ages) and in tributaries (from ε Nd −13 and ε Hf −25 in Sokoto sand to ε Nd −21 and ε Hf −36 in Kaduna sand), reflecting increasing detritus from the Nigeria basement complex.Epsilon values are less negative and more homogeneous across the Benue catchment (ε Nd −13 ± 2 and ε Hf −19 ± 3, corresponding to Tonian/Ectasian T Nd,CHUR , mostly Calymmian/Statherian T Nd,DM , mostly Tonian/ Stenian T Hf,CHUR , and mostly Ectasian/Calymmian T Hf,DM model ages) than in the Lower Niger (where they tend to become more negative downstream from ε Nd −14 ± 4 and ε Hf −16 ± 3 to ε Nd −22.2 ± 0.1 and ε Hf −25 ± 1).

Figure 7 .
Figure 7. Multidimensional scaling maps based on U-Pb zircon-age spectra highlight the segmentation of the Niger sediment-transport system (closest and second-closest neighbors are linked by solid and dashed lines, respectively).(a) The Upper Niger is disconnected from the Sahelian Niger downstream of the Inner Delta, and zircons in the Lower Niger are overwhelmingly supplied by the Benue River.(b) Zircons in the Upper Niger are supplied by Archean (Man Shield) and Paleoproterozoic (Baoulé-Mossi) protosources (data from Table4), whereas zircon grains are mostly derived from the Neoproterozoic (Trans-Saharan Belt) protosource downstream of the Inner Delta.

Figure 8 .
Figure 8. Evolution of the zircon-age signal along the segmented Niger sediment-routing system and zircon-provenance budget (inset).Zircon grains reaching the Inner Delta are derived in subequal proportions from the Upper Niger and Bani rivers, whereas zircon grains reaching the Niger Delta appear to be overwhelmingly supplied by the Benue River.

Figure 9 .
Figure 9. Time-structure maps of the Niger catchment.(a) Along the Upper Niger, Archean zircons from the Man Shield are rapidly diluted by Paleoproterozoic zircons from the Baoulé-Mossi domain.Neoproterozoic zircons are invariably dominant downstream of the Inner Delta.(b, c) T Nd,DM and T Hf,DM model ages, as old as late Eoarchean/Paleoarchean in uppermost Niger sand, decrease progressively to earliest Proterozoic/latest Archean in the Inner Delta and next abruptly to mid-Proterozoic in the Sahelian Niger.In Nigeria, they increase again downstream the Middle Niger, decrease below the Benue confluence, and finally increase in the Niger Delta.
Orosirian-Rhyacian zircons, predominant in all studied tributaries of the Upper Niger and Bani rivers, decrease sharply in abundance downstream of the Inner Delta but remain common in the Sahelian Niger and Middle Niger tributaries, and are most common in Ogun sand derived from the SW Nigeria basement.Instead, Paleoproterozoic zircons are minor in Benue sand, and particularly rare in the sand of its northern tributaries draining the eastern part of the North Nigeria basement, scarcely affected by the Paleoproterozoic orogenic event (Figure1).As for Archean zircons, also middle Paleoproterozoic zircons increase along the Lower Niger and are significant in Niger Delta distributaries, confirming that the sand generated in the upper parts of the Niger catchment and deposited at earlier times is being extensively reworked across the delta plain.This trend is confirmed by isotope values, which become more negative in fine to medium sand of the Lower Niger and Niger Delta distributaries (ε Nd from −11.6 to −18.4 and ε Hf from −22.3 to −25.5; the Pan-African domain, from the Sahara to the Niger Delta, T Nd,DM model ages range from Paleoproterozoic (Orosirian-Rhyacian) to early Mesoproterozoic (Calymmian) and T Nd,CHUR model ages from Paleoproterozoic (late Orosirian) to earliest Neoproterozoic (early Tonian).T Hf,DM and T Hf,CHUR model ages are mid-Paleoproterozoic (Orosirian-Rhyacian) for Middle Niger sand, Mesoproterozoic (early Calymmian) for Benue sand, and intermediate (late Paleoproterozoic) for Lower Niger and delta distributary sands.All model ages increase progressively along the Middle Niger, reflecting sediment supply from tributaries draining the western part of the Nigeria basement complex, decrease abruptly downstream of the Benue confluence, indicating overwhelming supply from the Benue River, but increase again in the Niger Delta.This is a further indication of extensive reworking across the delta plain of older sediments generated in larger proportions than today in the Niger catchment upstream of the Benue confluence.7.4.Other Protosources Subordinate Mesoproterozoic protosources are indicated by a minor early Calymmian cluster of zircon ages (1576 ± 35 Ma) showing up in the sand of the Inner Delta, Saharan dunes, Middle Niger tributaries, and a few Benue tributaries (Dep and Katsina-Ala; Table Upper Niger sand, U-Pb detrital-zircon ages are exclusively Archean in Guinea headwaters draining the Man Shield (all ages >2.65 Ga, peak at ∼2.88 Ga) and dominantly Paleoproterozoic (peak at ∼2.07 Ga) in the Inner Delta in Mali, testifying to the progressive downstream dilution by sediment supplied by tributaries draining the Paleoproterozoic Birimian domain.Pan-African ages first appear as the Niger River cuts across Neoproterozoic strata of the Taoudeni Basin.The zircon-age signal changes drastically to dominant Neoproterozoic downstream of the Inner Delta and becomes indistinguishable from that of Saharan eolian dunes across the Sahel (peak at ∼0.6 Ga).An abrupt change is observed also for ε Nd values, which become significantly less negative in the Sahelian Niger than in the Upper Niger and Bani rivers especially for sand but also for clay.Because of hampered flow as the Niger River enters the dune-covered arid lowlands of the Taoudeni Basin, Inner Delta marshlands do not trap only most of the sand generated in Guinea and SW Mali headwaters, but a large part of silt and clay as well.Downstream across the Sahel, sediment load is progressively reconstituted by extensive recycling of dune sand with a late Neoproterozoic zircon-age signature and by eolian dust with less negative ε Nd .Neoproterozoic zircons supplied in abundance by tributaries draining the Pan-African Trans-Saharan Belt remain dominant throughout Nigeria.Downstream of the Benue confluence, Lower Niger sand yielded very few Paleoproterozoic and no Archean zircons, as in Benue sand, indicating overwhelming supply from the Benue tributary.In the Niger Delta, however, small populations of Paleoarchean (Leonian), Mesoarchean (Liberian) and late Neoarchean zircon reappear, and both ε Nd and ε Hf values become more negative.This indicates extensive reworking of sand deposited along the coastal plain at earlier times of presumably wetter climate, when neither artificial nor major natural barriers to the sand flux existed along the entire Niger River from Guinean headwaters to the ocean.

Table 2
). Archean ages decrease from 6% to 2%, Orosirian to Rhyacian ages decrease from 26% to 20%, and Jurassic ages increase from 1% to 4%.The drastic change from dominantly mid-Paleoproterozoic Niger sand generated entirely from the Archean Kénéma-Man domain in Guinea (ε Nd −41 and ε Hf −67, corresponding to earliest Paleoarchean and latest Eoarchean T Nd,CHUR and T Nd,DM model ages, and to mid-Mesoarchean and latest Paleoarchean T Hf,CHUR and T Hf,DM model ages; Table1).Because of mixing with sediment derived from Paleoproterozoic and finally Neoproterozoic rocks, epsilon values become progressively less negative downstream the Upper Niger (ε Nd −23 and ε Hf −43 in Inner Delta sand, corresponding to Orosirian and late Siderian T Nd,CHUR and T Nd,DM model ages and to Rhyacian and latest Neoarchean T Hf,CHUR and T Hf,DM model ages; Table1).Epsilon values are less negative in the sand of the Bani tributary draining the Paleoproterozoic Birimian domain in northern Ivory Coast and southernmost Mali (ε Nd −19 and ε Hf −38, corresponding to model ages ranging from early Calymmian to Rhyacian; Table1).Epsilon values abruptly become less negative in Sahelian Niger sand (ε Nd and ε Hf from −23 and −43 in the Inner Delta to −10 and −26 in Niger, corresponding to model ages decreasing from Orosirian/late Neoarchean to Tonian/Orosirian).Such a decrease reflects extensive recycling of Saharan dunes (ε Nd −12 ± 1 and ε Hf −32 ± 5, corresponding to Tonian/Stenian T Nd,CHUR and Calymmian T Nd,DM model ages and to Ectasian/Orosirian T Hf,CHUR and Paleoproterozoic T Hf,DM model ages; Table1) and ultimate provenance from the Neoproterozoic Trans-Saharan Belt.A significant change downstream of the Inner Delta is observed also for ε Nd values in clay (own data).

Table 3
; (b) larger dams on the Upper Niger (e.g., Selingué Dam completed in 1982 on the Sankarani tributary, Markala Dam completed in 1947 on the mainstem; Ferry et al., 2012) may have sequestered a significant part of Upper Niger sand in the reservoir; (c) our sample is not representative of the average composition of Inner Delta sand.Zircon Provenance Budgets Based on the Iterative Method of Garzanti et al. (2019) With Wasserstein (Wa) Statistics (Lipp & Vermeesch, 2023), and on Inverse Monte Carlo Modeling With Kuiper (Ku) and Kolmogorov-Smirnov (KS) Test Statistics Geological domain Jurassic 2σ Pan-African 2σ Tonian 2σ Birimian 2σ Liberian 2σ Leonian 2σ

Table 4
Compilation of Geochronological (U-Pb, Rb-Sr, Sm-Nd, and K-Ar) Ages of Igneous and Metamorphic Rocks Exposed in Diverse Domains of the West African Craton

Table 1
Late Neoproterozoic zircons (Ediacaran peak at 597 ± 13 Ma) first appear in Upper Niger sand as the river enters the Taoudeni Basin in Mali and invariably prevail in Sahelian Niger, Middle Niger, and Lower Niger sands.Neoproterozoic zircons shed by the Pan-African Trans-Saharan Belt are overwhelming throughout the Benue catchment, Ediacaran ages being dominant except for Dep sand where zircon grains are mostly older (Tonian; 912 ± 12 Ma).Although invariably subordinate, zircons of Tonian age (878 ± 34 Ma) widely occur in the Niger catchment from the Sahara to the Niger Delta and are sporadically detected even in the Upper Niger catchment (Tinkisso and Bani sands; Table Sediment sampling in Guinea and Mali was financed and coordinated by IRD-Bamako and carried out with the fundamental help of Boubacar Baïlo Diakité and Diarra Diallo of the National Hydraulic Directorate of Conakry (DNH-Guinea) and of Boubacar Sidibé and Adama Mariko of the National Hydraulic Directorate of Bamako (DNH-Mali).Other samples in Mali and Niger were collected by Luca Baglioni (student at the University of Milano-Bicocca) and Ibrahim Mamadou (Ph.D student at the Abdou Moumouni at Nyamei and Paris 1 universities).Samples in Nigeria were collected by geology students, professionals, and researchers at the Federal University of Petroleum Resources Effurun (Ilo David Adeyemi, Ogefere Onoharigho, Igbinosa Osaze Temple, Solomon Awakessien, Efenudu Benjamin), Nasarawa State University (Christabel Otite), and University of Lagos (Olusegun Adeaga).