Crustal evolution of the Paleoproterozoic Ubendian Belt (SW Tanzania) western margin: A Central African Shield amalgamation tale

a Department of Earth Science, Tohoku University, Aoba, Sendai 980-8578, Japan b Center for Northeast Asian Studies, Tohoku University, Aoba, Sendai 980-8576, Japan c Frontier Research Institute for Interdisciplinary Sciences, Tohoku University, Aoba, Sendai 980-0845, Japan d University of Dar es Salaam, Geology Department, P.O. Box 35052, Dar es Salaam, Tanzania e Graduate School of International Resource Sciences, Akita University, Akita 010-8502, Japan f Center for Fundamental Education, Okayama University of Science, Okayama 700-0005, Japan


Introduction
Studies on the growth and evolution of continental crust through time are fundamental to understand Earth's history. Magmatism, accretion of magmatic provinces and record of continental growth show a widely recognized episodic nature and that is apparently tied to the supercontinent cycles (e.g. Ernst et al., 2016;Pastor-Galán et al., 2019). The supercontinent cycle explains how continents amalgamate periodically into superplates that likely influence the evolution of the lithosphere, biosphere, atmosphere and hydrosphere (Nance et al., 2014;Condie et al., 2011;Rogers and Santosh, 2003). Due to the limitations of the geological record's oldest pieces, the study of radiogenic isotopes in rocks is crucial for determining the evolution and age of the continental crust (Armstrong, 1981;Hawkesworth and Kemp, 2006). The global record of magmatic ages is likely dominated by accretion of supercontinents (Pastor-Galán et al., 2019) and the most recent data show at least five major magmatic events connected to the assemblage of supercontinents observed at~250 Ma (Pangea),~560 Ma (Gondwana),~1 Ga (Rodinia),~1.8 Ga (Columbia/Nuna) and~2.6 Ga (Kenorland?) (Runcorn, 1962;Hawkesworth et al., 2010;Meert, 2012;Pastor-Galán et al., 2019).
During the Paleoproterozoic-Mesoproterozoic Transamazonian-Eburnean orogenies, several nuclei conformed the Congo-Sao Francisco shield (including Sao Francisco, Congo, Gabon, Angola, Kasai and Tanzania Cratons; Meert and Santosh, 2017). This shield would remain together during the amalgamation of Rodinia, Gondwana and Pangea to partially break up with the opening of the Atlantic in the Mesozoic (e.g. Torsvik et al., 2012). The Ubendian-Usagaran Belt, in east Africa, is sandwiched between the Archean Tanzania Craton and the Bangweulu Block, highly re-worked during the Pan-African orogeny (Neoproterozoic). This belt is thought to represent the Paleoproterozoic assembly of these two constituents of the Congo shield at around 1.8 Ga, during the amalgamation of Atlantica within the frame of the Columbia supercontinent cycle (e.g. Meert and Santosh, 2017). It contains a large volume of granitoids and metagranitoids with possible ages between 1.9 and 1.8 Ga (e.g. Bahame et al., 2016). However, absolute ages for granitoids along the boundary of Bangweulu Block and Ubendian Belt are scarce (Lenoir et al., 1994). Thus, the origin of the Bangweulu Block is disputed. Some authors proposed that it was an active continental margin related to the Ubendian Belt (Andersen and Unrug, 1984), whereas others supported that it formed as the result of overthickened crustal anatexis (Hanson, 2003). Recently, inferred from the absence of crustal thickening and subsequent unroofing, some studies suggested that the Bangweulu Block and Irumide Belt were metacratonized margins resulted from an attempted subduction of a passive continental margin. This subduction triggered a loss of cratonic rigidity and increased susceptibility to magmatism during subsequent deformation (De Waele et al., 2006a;Liégeois et al., 2013).
The tectonic evolution of the metagranitoids in the western margin of the Ubendian Belt is underexplored and major unknowns are: What are the geochemical characteristics of the granitoid forming magmatism and how it can constrain to processes involved in the petrogenesis, and its geodynamic setting? When did the magmatism occur? What is the extent of late Neoproterozoic metamorphism? Geochronological and geochemical studies of metagranitoids that try to decipher intracratonic crustal evolution are critical importance in understanding stabilization of continental crust.
In this research, we present new U-Pb zircon ages and whole-rock major and trace elements geochemistry of metagranitoids along two traverses across the Ufipa Terrane of the Ubendian Belt and northeastern portion of the Bangweulu Block. Our new data aim to contribute to the understanding of crustal evolution of the western margin of the Ubendian Belt, as well as, new insights on Paleoproterozic intense magmatic events at cratonic margin and global reworking of continental crust during late Neoproterozoic time.

Geological background
Although plate kinematic constraints are limited, most researchers agree that Columbia (a.k.a. Nuna and Paleopangea, see Meert, 2012 for the naming debate) amalgamated from late Paleoproterozoic to early Mesoproterozoic (e.g. Meert and Santosh, 2017). Its maximum packing probably occurred at~1.5 Ga (Salminen et al., 2016) and its main constituents nuclei were: Ur, Nuna (sensu Meert and Santosh, 2017) and Atlantica (Rogers and Santosh, 2002;Zhao et al., 2004;Merdith et al., 2017). Atlantica nucleus was formed after series of collisions that joined the Archean cratons of West Africa, Congo, Kalahari, Amazon, São Francisco, Rio de la Plata and Sahara (Rogers, 1996;Neves, 2011;Grenholm, 2019). The break-up of the Columbia apparently began at 1.45 Ga and continued until at least 1.38 Ga Pisarevsky et al., 2014). This break-up marked the beginning of the next supercontinent cycle, Rodinia. Rodinia supercontinent assembled between~1.1 Ga and~0.9 Ga (e.g. Li et al., 2008) and broke-apart at 0.8 Ga, when the Gondwana (a.k.a. Pannotia, see Nance and Murphy, 2019) cycle started. The assembly of the Gondwana commenced at 630 Ma with the amalgamation of eastern and western Gondwana and ended about at Late Ediacaran-Early Cambrian times. Eastern Gondwana comprised much of the present-day continents of Australia, India and East Antarctica. Whereas, western Gondwana was composed of Congo, São Francisco, Kalahari, Rio de la Plata, West Africa cratons (Meert, 2003;Collins and Pisarevsky, 2005;Oriolo et al., 2017). At 600 Ma, the major event marking the formation of Gondwana, the Pan-African orogeny, began. The Pan-African orogeny is one of the most extensive orogenies and continental crustal reworking events occurred in the Neoproterozoic involving a long series of tectonothermal events throughout Eastern and Western Gondwana (Kröner and Stern, 2005;Meert, 2003;Veevers, 2003).

Overview of the Ubendian Belt
The Ubendian Belt occupies approximately 600 × 150 km area from eastern Congo to western Tanzania and from northern Malawi to northeast Mozambique, with NW-SE trend ( Fig. 1; Daly, 1988;Lenoir et al., 1994). The belt consists mainly of NW to NNW trending Paleoproterozoic granitoids and medium to high-grade metamorphosed granitoids with migmatite, mafic granulite, pelitic-psammitic gneisses, rare eclogites and ultramafic rocks, intruded by a few carbonatites with ages ranging from Mesoproterozoic to Mesozoic. On its northern end, the belt is uncomfortably overlain by rift related metasedimentary rocks of the Mesoproterozoic Kibaran Belt and Neoproterozoic Malagarasi Platform (Deblond et al., 2001;Legler et al., 2015).
Three distinct Proterozoic orogenic events were recorded in the Ubendian Belt that includes: (1) the occurrences of Paleoproterozoic high-pressure rocks in the belt (Ring et al., 1997;, suggest that subduction and collisional events were occured between 2.0 and 1.86 Ga. These events are concomitant with the assembly of the Atlantica amalgamation within the Columbia superconti-nent cycle. (2) During the Mesoproterozoic (~1.4 Ga to~1.2 Ga), the Ubendian Belt, like the neighboring Kibaran Belt, underwent extension regime and opening of an oceanic basin (Boniface, 2019), whereas the history from~1.2 to~1.0 Ga, coeval with the Irumide Belt collision, probably due to far-field effect of the Rodinia assembly (Tack et al., 2010).
(3) Occurrence of the Neoproterozoic eclogites  in the belt evinces a regional reworking during the Pan-African orogenic event and the amalgamation of Gondwana. Relicts of the abovementioned events are uniformly distributed throughout the belt.
The Wakole Terrane is excluded from the grouping, as it solely consists of rift-basin sediments Mesoproterozoic metasedimentary rocks without Paleoproterozoic traces (Boniface et al., 2014). The main features of internal terranes of the Ubendian Belt are briefly presented in Table 1.

The Bangweulu Block
The Bangweulu Block is a Paleoproterozoic cratonic region with an Archean basement component, surrounded by Proterozoic mobile orogenic belts (Fig. 1). The southern and western sides are bounded by the Mesoproterozoic Kibaran and Irumide Belts (1.35-1.0 Ga), whereas southwestern side bounded by the Neoproterozoic Lufilian Belt (880-765 Ma). The Bangweulu basement consists of three major units: (1) eastwest trending crystalline schist belt, (2) metavolcanics and granitoid batholiths, and (3) discordant intrusions. A large part of the basement is covered by the Paleoproterozoic metasedimentary sequences of the Mporokoso Group (Andersen and Unrug, 1984;De Waele et al., 2006b). The east-west trending schist belts are distributed in the northeastern and eastern part and farther west of the block and mainly consist of migmatitic biotite gneiss, biotite-epidote and muscovite gneiss, associated with mica-schist, pelitic schist, amphibolite and recrystallized quartzite.
The metavolcanics and granitoid batholiths form a corona like pattern at the margins of the Bangweulu Block intruding the schist belts ( Fig. 1). Geochemically, the metavolcanics, and granitoids are characterized by high-K, calc-alkaline feature, and show negative Nb, Ta, Zr, Hf, Ti and P anomalies accompanied with positive anomalies of LILE (Large Ion Lithophile Element). Andersen and Unrug (1984) attributed the geochemical signatures of the Bangweulu metavolcanics and granitoids to their formation at the volcanic arc setting. However, De Waele et al. (2006a) suggested metacratonisation of the Bangweulu Block continental margins during the~2.03 Ga Usagaran phase and subsequent several major reactivations in the late Usagaran phase, (~1.94 Ga), Ubendian phase (~1.87 Ga), and Lukamfwa phase (~1.6 Ga). U-Pb zircon ages for these granitoids range between 1.87 and 1.86 Ga. The εNd values are of -4.3 and -7 (volcanics) and -5.9 and -6.0 (granitoids) which gave Nd model ages, which estimate the mantle extraction age (e.g. McCulloch and Wasserburg, 1978), of 3.16-2.37 Ga (De Waele et al., 2006a;De Waele and Fitzsimons, 2007;Schandelmeier, 1983).
Whereas granites from east and south central Bangweulu Block give present-day εNd compositions of -23, -16 and model age of 3.62-2.88 Ga (Debruyne et al., 2014). These granitoids, associated with volcanic units, and mainly consist of pyroclastic rocks, minor hypabyssal intrusions and andesitic-rhyolitic flows. Volcanic rocks exhibit similar geochemical feature and rhyolites have crystallization U-Pb zircon age of 1. 88-1.86 Ga (De Waele and Fitzsimons, 2007).
To the west from the Bangweulu Block, the Copperbelt Domes include metagranitoids and metavolcanic units similarly emplaced between 2.06 and 1.87 Ga (De Waele et al., 2006b), that were suggested westward continuation of the block within the Neoproterozoic Lufilian Belt (Rainaud et al., 2005). To the south and southeast, basement of the Mesoproterozoic Irumide Belt comprises a Neoarchean unit with igneous crystallization age of~2.61 Ga, and various~2.03 and~1.97 Ga gneisses (De Waele et al., 2006a;Debruyne et al., 2014). The evidence of the Archean basement in the Bangweulu Block is recorded by the Mesoarchean (~3.2 Ga) xenocrystic and detrital zircon within a~1.8 Ga lapilli tuff (Rainaud et al., 2003) and 2.76 Ga tonalitetrondjhemite-granodiorite suite (De Waele et al., 2006a, 2006b).

Study area
The study area straddles the boundary of the Ufipa Terrane and the northeastern margin of the Bangweulu Block along the eastern shore of Tanganyika Lake (Fig. 2). The boundary is delineated by a shear zone and metamorphic grade transition from chlorite-muscovite-biotite schist from the central part of the Bangweulu Block to sillimanitecordierite gneisses and amphibolite towards the Ubendian Belt (Andersen and Unrug, 1984;De Waele et al., 2006b). Recognizing this contact is not easy, as the contact zone is intruded by numerous granitic intrusions of the Paleoproterozoic Kate-Ufipa Complex. The Kate-Ufipa Complex consists mainly of deformed plutonic rocks, including coarsegrained porphyritic granite, biotite-granite, granodiorite, plagiogranite, tonalite and syenogranite. Volcanic equivalents of the Kate-Ufipa Complex is represented by the Kipili volcanics. Lithologies show spatial variation, with dacitic and rhyodacitic tuffs and the few andesites dominating to the south, whereas rhyolitic tuffs, flows, and ignimbrites become more frequent to the north and east (Lenoir et al., 1994;Nanyaro et al., 1983). Based on metamorphic mineral assemblages in gneisses, Boniface and Appel (2018) subdivided the Ufipa Terrane into   the northern and southern parts. The Northern Ufipa Terrane consists mainly of kyanite-bearing biotite-granitic gneisses, semi-metapelitic to metapelitic gneisses, and metagranites, with minor eclogites and high-pressure granulites; the occurrence of kyanite-bearing gneiss and eclogites suggest relatively high-pressure metamorphic regimes. In contrast, the Southern Ufipa Terrane is characterized by low pressure metamorphic regimes, composed of sillimanite-bearing biotite-rich granitic gneisses, semi-pelitic to metapelitic migmatites. Metapelites and gneisses of the both Northern and Southern Ufipa Terranes yielded zircon and monazite metamorphic ages of 1.96-1.90 Ga and of 599-537 Ma. We collected deformed granitoid samples from both the Northern and Southern Ufipa Terrane and weakly deformed granitoids from the Kate-Ufipa Complex (KRD060-062) as well as granoporphyries from the Kipili metagranitoids (KRD055-056, 059). Locations and lithologies of the studied samples are shown in Table 2 and Fig. 2.

Whole-rock geochemistry
A total of sixteen metagranitoid samples was selected for wholerock major and trace element chemistry. The analyses were carried out at Activation Laboratories Ltd., Canada, using Code 4Lithoresearch Lithogeochemistry Package; the package uses lithium metaborate/ tetraborate fusion with inductively coupled plasma optical emission spectrometry (FUS-ICPOES) and inductively coupled plasma mass spectroscopy (FUS-ICPMS) for the major and trace element analyses, respectively.

U-Pb geochronology
Zircons were separated from the nine whole-rock samples using conventional rock-crushing, heavy liquid and magnetic techniques, and by handpicking under a binocular microscope. The handpicked zircons are mounted in epoxy resin (Struers Specifix-40) discs and polished until the zircon interiors were exposed and examined by transmitted and reflected light with an optical microscope. A Metkon Forcipol 1 V and a 3 M aluminum oxide lapping film was used for the polishing. To reveal their internal structures, cathodoluminescence (CL) images were obtained using a Hitachi S-3400 N SEM, equipped with a Gatan model MiniCL system housed at the Graduate School of Science, Tohoku University, Japan (Fig. 10). The CL observation was conducted using 25 kV accelerating voltage and a 90 nA probe current. Distinct domains within the zircons were selected for analysis, based on the CL images. In situ zircon U-Pb dating was carried out in the Okayama University of Science by using a Thermo Fisher Scientific iCAP-RQ single-collector quadrupole coupled to a Teledyne Cetac Technologies Analyte G2 ArF excimer laser ablation (LA) system equipped with a HelEx 2 volume sample chamber. The laser ablation of zircons was conducted at the condition of laser spot size of 25 μm, fluence of 1.8 J/cm 2 and repetition rate of 5 Hz. Other conditions of LA-ICPMS method are referred to Aoki et al. (2019) and Aoki et al. (2020). The zircon 91,500 (1065 Ma; Wiedenbeck et al., 2004) was used as an external standard for age calibration, and the NIST SRM 612 (Jochum et al., 2011) silicate glass was applied for instrument optimization, and the Plešovice zircons (337 Ma; Sláma et al., 2008) were measured as secondary standards for quality control. U-Pb ages and concordia diagrams were calculated and drawn using the programs IsoplotR (ver. 3.75; Vermeesch, 2018), concordia age of each sample incorporates errors on the decay constants and includes evaluation of concordance of apparent ages. The concordia ages and errors are at the two-sigma level.

Whole-rock geochemistry
The major and trace element geochemical data of the studied metagranitoids including granoporphyries are given in Table 3. On the basis of major element characteristics, the samples can be divided into high-K calc-alkaline and calc-alkaline granitoids regardless of deformation. Hence, hereafter weakly foliated granitoids, Grt-Bt-Ky augen gneisses and weakly foliated granoporphyries are referred to high-K metagranitoids, and Grt-Bt-Ky-Sil-Crd gneisses are referred to calc-alkaline metagranitoids.

Alteration and element mobility
Since studied rocks have been affected by metamorphism and sericite alteration, prior to petrogenetic interpretation of major and trace element content, it is essential to ensure whether they have triggered significant post-magmatism element mobility. All of the metagranitoids and volcanic rock samples have low Loss of Ignition (LOI) values (all < 2.36%), and show their relatively low contents of weathering or alteration minerals (e.g. carbonate, hydrous minerals). Moreover, poor correlation has been shown between LOI and Eu anomaly (= 0.2; Eu/Eu*) and Sr (=0.1), indicating Sr or Eu anomaly is not significantly affected by post-emplacement processes. To assess the element mobility, we compare high field strength elements (HFSE) and rare earth element (REE), considered as immobile during metamorphic processes, against the large ion lithophile elements (LILE), which are mobile during secondary processes (Polat and Hofmann, 2003).
The bivariate plots for the Zr (sensitive indicator of immobility) versus Na 2 O and Ce show scattered points and poor correlation (R2 < 0.3), which suggests it was likely affected by secondary processes (not shown), whereas Zr versus REE (represented by Sm) (Fig. 4a), HFSE (represented by Hf) (Fig. 4b), LILE (represented by Rb) (Fig. 4c) demonstrated a relatively linear trend with moderate to good correlations (R2 = 0.46-0.96), suggesting that they are not much affected by secondary processes. We interpreted our geochemical analyses avoiding those elements that show mobility during secondary processes. When plotting samples on major element discrimination diagrams and on bivariate diagrams, samples exhibit two different trends (Figs. 5 and 6), especially regarding K 2 O.

Zircon U-Pb geochronology
Analytical results are presented in Supplementary data 1, representative zircon CL images are given in Fig. 10 and concordia plots of the data are presented in Fig. 11. The obtained ages are shown in a decreasing order.
Sample KRD063 has zircons that are colorless or light brown, mostly subhedral, stubby, up to 200 μm in length (Fig. 10a). These zircons display weak oscillatory zoning or without zoning in CL image. The measured Th/U ratios ranging from 0.04 to 1.03. Five zircons possess Th/U ratios lower than 0.1 (Fig. 12a) with concordia age of 570 ± 5.7 Ma (MSWD = 3.8; Fig. 11a), and one zircon 1914.2 ± 19 Ma, to be considered to represent the inherited zircon age of the metagranite.
Sample KWB052 is a metagranite with colorless or light brown zircons and euhedral to anhedral shapes, about 100-200 μm in size, aspect ratios from 1 to 3, some zircons manifest patchy zoning or homogeneous core surrounded by bright rims in CL images (Fig. 10b). Patchy zoning may reflect the strain experienced by zircon during final magmatic emplacement (Paquette et al., 1995). From sixteen analyses, which Th/U ratios ranging from 0.02 to 2.1, two of them yielded concordia age of 565 ± 7.7 Ma (Fig. 11b). Zircons with low Th/U concordia age reflects the metamorphic event (Fig. 12a, b).
Sample KWB048 has clear prismatic euhedral and subhedral zircons up to 300 μm, whose aspect ratios are from 1 to 3 (Fig. 10d). Many zircons have bright CL core and distinct oscillatory zoning, some zircons have thin bright rims. Seventeen analyses provide a concordia age of 1880 ± 4.6 Ma (MSWD = 1.7; Fig. 11d), measured Th/U ranging from A. Ganbat, T. Tsujimori, N. Boniface et al. Gondwana Research 91 (2021) 286-306 1.2 to 2.6. The concordia age is interpreted as the crystallization age of the metamonzogranite. Sample KWB053 contained colorless zircons euhedral to subhedral, no longer than 200 μm in length, stubby, with aspect ratios 2 or less, zircons exhibit fine-scaled oscillatory zoning in CL-image (Fig. 10e). Twenty-seven zircons were analyzed, and their Th/U ratios ranging from 0.89 to 2.6. All analyzes are plotted on concordia diagram, providing common concordia age of 1876.4 ± 7.8 Ma (MSWD = 0.88; Fig.  11e). We interpreted this age as crystallization age of the metagranodiorite.
Sample KYB065 contained transparent to light brown zircons, up to 200 μm in length, varies from stubby to elongated in shape, with aspect ratios from 1 to 3. In CL images, zircons show wide oscillatory zoning, some zircons have homogeneous xenocrystic core (Fig. 10f). Thirty-six grains were chosen to analyze. The measured Th/U ratios ranging from 0.6 to 2.1. All zircons are plotted on or close to the concordia, describing concordia age of 1874.5 ± 3.9 Ma (MSWD = 0.68; Fig. 11f), interpreted to be the metagranite crystallization age.
Sample KWB049 contained transparent to light brown zircons, up to 150 μm in length, mostly stubby with aspect ratios not exceeding 2, prismatic euhedral in shape. In CL images, zircons show narrow oscillatory zoning, but some zircons contain xenocrystic core (Fig. 10g). Thirty zircons were chosen to analyze. The measured Th/U ratios are ranging from 0.9 to 1.9. All zircons are plotted close or on the concordia, describing common concordia age of 1859.6 ± 9.6 Ma (MSWD = 0.88; Fig. 11g), interpreted to be the time of crystallization of the metamonzodiorite.
Sample KRD060 has stubby to slightly elongate, euhedral to subhedral, clear zircons with lengths up to 200 μm and aspect ratio about 1.5. Many zircons have CL-bright zones in xenocrystic core with faint oscillatory rims (Fig. 10h). The Th/U varies range from 0.7 to 2.3. Concordia age of ten grains yielded 1858.5 ± 15 Ma (MSWD = 1.9, Fig. 11h), which is interpreted as the crystallization age of metamonzogranite.
Sample KRD062 has brown but mostly clear, euhedral or subhedral stubby and rounded zircons 100-150 μm in length. Twenty-four grains were selected to be analyzed for this sample (Fig. 10i). Th/U ratios range from 0.99 to 2.5. All analyzes are plotted on or very close to the concordia, yielding concordia age of 1854.7 ± 9.7 Ma (MSWD = 2.6; Fig. 11i), which is considered as crystallization age of the metagranodiorite.
A. Ganbat, T. Tsujimori, N. Boniface et al. Gondwana Research 91 (2021) 286-306 200 μm in length, homogeneous with no obvious zoning, partially enclosed in the major primary minerals. Results of analysis are presented in Supplementary data 2 and concordia plots of the data are presented in Fig. 13. Eight spots of sample KYB065 define an isochron line with an intercept age of 1878 ± 55.9 Ma on a Tera-Wasserburg plot (Fig. 13a). Eleven spots of sample KRD062 build an isochron line with age of 1864 ± 79.9 Ma (Fig. 13b). Those apatite ages overlap of zircon U-Pb ages of the same samples. U-Pb analyses of apatite exhibit a broad discordia array that scatters and initial 207 Pb/ 206 Pb ratio varying between 0.721 and 0.757.

Geochemical constraints for Paleoproterozoic convergent margin
The whole-rock major element compositions show that the weakly foliated Kate-Ufipa Complex associated with granoporphyries from the Bangweulu Block and metagranitoids from the Northern Ufipa Terrane, share similar geochemical features, in spite of the net boundary of the Bangweulu Block and the Ufipa Terrane demarcated by shear zones (e.g. De Waele et al., 2006b). Major element variation is consistent with the subdivision of Boniface and Appel (2018), differing between the North and the South Ufipa Terranes. Although calc-alkaline metagranitoids from the South Ufipa Terrane exhibit different major element characteristics, their trace element and REE compositional trend is similar (Table 2; Fig. 7). The geochemical resemblance among metagranitoids including granoporphyries implies that these rocks were formed under similar igneous processes in akin tectonic settings.
Granitoids including intermediate rocks may form in various ways, such as the (1) fractional crystallization of underplated basaltic magmas (Grove and Brown, 2018), (2) mixing of crustal derived silicic and mantle-derived mafic magmas (e.g. Guo et al., 2012), (3) partial melting of the lower crust (Bryan et al., 2008), and (4) water-induced melting of the mantle wedge metasomatized by subduction derived fluids or melts (e.g. Hawkesworth and Kemp, 2006). Melting by fractional crystallization or partial melting generates systematic trends between SiO 2 and other major elements (e.g. Xu et al., 2009). Such trends are shown by high-K calc-alkaline samples (Fig. 6). Fractionation of clinopyroxene and hornblende causes decreases in MgO, FeO, and CaO with increasing SiO 2 . Decreases in Al 2 O 3 and Sr concentrations with increasing SiO 2 are  Sun and McDonough (1989). The grey shaded areas indicate geochemical patterns of Paleoproterozoic granitoids and metavolcanics from western Ubendian Belt adopted from Kazimoto et al. (2014aKazimoto et al. ( , 2014b and Tulibonywa et al. (2017) respectively. The reference geochemical patterns of continental arc magmas. The comparison data of Upper Continental Crust is from Rudnick and Gao (2005).  (Schandl and Gorton, 2002). See details in text.
A. Ganbat, T. Tsujimori, N. Boniface et al. Gondwana Research 91 (2021) 286-306 correlated with the fractionation of plagioclase, and decreases in P 2 O 5 and TiO 2 may reflect the fractionation of accessory minerals such as apatites and rutiles. However, if metagranitoids were formed by fractional crystallization of basaltic or andesitic magmas, then the volume of basaltic or andesitic rocks exposed in the study area should greatly exceed that of the granitoids. This is not the case, and voluminous metagranitoids and granoporphyries are exposed in our study area. In addition, a mixing model for mantle-derived mafic melt and crustal derived silicic magma is not suitable for the petrogenesis of studied samples. In such case, mixing process would cause a decrease in the silica content of the rocks and enrichment in Cr, Co, and Ni, presence of MME (mafic magma enclaves), which is not observed (Cantagrel et al., 1984).
Trace element characteristics such as high Zr/Nb ratio (aveage = 26), Th/Nb ratio (average = 1.9), Hf/Ta ratio (average = 10.7) and depletion of Ti-Nb-Ta are typical of arc magma, which originate as a result of partial melting of the upper mantle (Gill, 2011;Pearce and Peate, 1995). Metagranitoids and granoporphyries have high SiO 2 , but low MgO, FeO T , CaO, Sc, Cr, Co, and Ni contents. Their LREE and LILE enrichments coupled with HFSE depletions and wide range of HREE concentrations (Fig. 7a, b). This composition suggests involvement of crust and heterogeneous source possibly indicating derivation from the partial melting of the lower continental crust, subducting slab and the upper mantle. On widely using geodynamic discrimination diagram for granites, invented by Pearce et al. (1984) which use elements such as Rb, Nb, Y, Yb and Ta (Fig. 8a, b), the calc-alkaline metagranitoids are mostly plotted on VAG (Volcanic Arc Granites), whereas two samples are plotted on WPG (Within Plate Granite) field. On another trace element diagrams of Th/Hf versus Ta/Hf and Th/Ta versus Yb (Fig. 8c,  d), the rocks belong to the active continental margin field.
Recent studies recognized that magmatism formed in slab failure setting could be masked and misinterpreted as an arc Whalen, 2014, 2017;Hildebrand et al., 2018). Because of the buoyancy difference between oceanic and continental crust, it is expected some sort of slab break-off after every closure of an oceanic basin. Therefore, slab failure is the rupture and separation of subducting plates and an integral phenomenon of the plate tectonic and a natural consequence of subduction (e.g. Atherton and Ghani, 2002;Davies, 2002). Slab failure rocks originated from the deep mantle processes that likely include partial melting of the mafic upper portion of the torn slab, leaving a garnetbearing, plagioclase-free eclogitic residue (Hildebrand and Whalen, 2017). Due to their location and short-lived, generally voluminous character, slab break-off magmatism is usually confused in magmatic arc activity. The depth of break-off largely controls the volume of magmatism and associated metamorphism of the orogen. The rebound of the partially subducted continent will cause intense uplift and exhumation of deeper crustal levels (Duretz et al., 2011;Duretz and Gerya, 2013). Break-off occurring in shallow depths creates narrow orogens, lowergrade metamorphism, and intense rates of exhumation. In contrast, deep break-off creates broad orogens with higher grade metamorphism and more subdued rebound and exhumation (Duretz et al., 2011). In this case, the resultant uplift and exhumation might cause the erosion of the arc in the overriding plate, leaving only minor vestiges of the upper crust (Hildebrand and Whalen, 2014). Whalen and Hildebrand (2019) developed new approaches to discriminate different arc magmatism by using the differences in trace element concentrations of large number of well-studied samples. Their new trace element tectonomagmatic plots for I-type rocks (restricted for S-and A-types) discriminates between slab failure and arc rocks. They tested empirically the diagrams, observing the distinction of trace elements such as Sr/Y, Nb/Y, Gd/Yb and La/Yb to arrive at separative values (Fig. 9). The strong partitioning of HREE into residual garnet and absence of Srplus Eu-hosting plagioclase in slab failure magmatism, exhibit a distinctive high La/Yb, Gd/Yb, Sm/Yb and Sr/Y ratios. Instability of a Ti-rich phase such as rutile, plus residual garnet, produce their high Nb/Y and Ta/Yb ratios. High Sm/Yb in slab failure magmatism reflects their greater depth of melting (Putirka, 1999).
On the newly developed diagrams, metagranitoids mostly belong to slab failure field on La/Sm versus Sm/Yb, La/Yb versus Gd/Yb, La/Yb versus Ta + Yb, Ta/Yb versus Ta + Yb, Gd/Yb versus Nb + Y and Sm/Yb versus Nb + Y (Fig. 9). Predominance of felsic over mafic magma composition; the diverse signatures of granites (e.g. volcanic-arc granite, within-plate granite); and heterogeneity of origin can be interpreted by slab failure setting (Massawe and Lentz, 2020;van Staal et al., 2009). Presence of Paleoproterozoic high-pressure metamorphic rocks in the Ufipa Terrane suggest that slab failure occurred at a great depth (Hildebrand and Whalen, 2014) and it can explain the absence of "real" arc magmatism along the Bangweulu Block and the Ufipa Terrane.

Geochronological constraints on magmatism
Zircons with high Th/U ratios (> 0.2) are of magmatic origin, whereas those having a low Th/U ratio (< 0.1) have undergone a secondary process such as metamorphism and hydrothermal alteration A. Ganbat, T. Tsujimori, N. Boniface et al. Gondwana Research 91 (2021) 286-306 (e.g. Hartmann et al., 2000). High Th/U ratios (> 0.15) have also been recorded in recrystallized zircon, and these zircons are grown during high-temperature metamorphism (e.g. Yakymchuk et al., 2018). The dated zircons with high Th/U ratios and clear oscillatory zoning (Fig.  12a) are most likely to be magmatic origin (Pidgeon, 1992). Zircon LA-ICPMS U-Pb data from the metagranitoids and granoporphyries on the boundary of the Bangweulu Block and the Ufipa Terrane of the Ubendian Belt show concordia ages at 1889.9 ± 4, 1880.5 ± 4.6, 1876.4 ± 7.8, 1874.5 ± 3.9, 1859.6 ± 9.6, 1858.5 ± 15 and 1854.3 ± 9.7 Ma and emplaced during a time period about 40 Myr. Ages show no systematic younging along the studied area, suggesting geochronological similarity along the Bangweulu Block boundary and the Ufipa Terrane. These ages support the limited ages by the pioneering works of Schandelmeier (1983), who reported Rb-Sr isochron age for granite of the Ufipa Terrane (1838 ± 86 Ma), and TIMS U-Pb zircon age (1864 ± 32 Ma) (Fig. 14a) by Lenoir et al. (1994). Ages of the Ufipa metagranitoids are in good correlation with magmatic ages of different terranes of the Ubendian Belt (Fig. 14b), and a metamorphic age of granulite-facies monazite and zircon from metapelites in the Ufipa Terrane. We attribute this high-grade metamorphism to the heat advection by high-K and calc-alkaline magmas into a continental margin. This scenario has also been applied to the Paleoproterozoic granulites of the Katuma Terrane (Kazimoto et al., 2015). Extensive volcano-plutonic rock ages in the margins of the Bangweulu Block (Fig. 14c;De Waele et al., 2006b). For example, the felsic to mafic lavas extruded in the Irumide Belt during 1.88-1.85 Ga. Similarly, at the Mansa area, southwestern margin of Bangweulu Block, granitoids and felsic lavas are emplaced at 1.89-1.86 Ga (De Waele and Fitzsimons, 2007). Our datasets are also consistent with a 1.89-1.86 Ga subduction event within the Usagaran Belt, which is also supported by arc Fig. 11. U-Pb concordia diagrams of zircons from the metagranitoids and granoporphyries, showing U-Pb isotope ratios. White spots indicating not calculated spots.

Geochronological constraints on metamorphism
The mineral assemblage, gneissic texture and banding of the studied metagranitoids imply that our study area experienced low to intermediate metamorphic conditions after their magmatic emplacement. Particularly, samples KWB049 and KYB065 contained sodic plagioclase, hornblende, epidote, garnet and kyanite. It suggests that metamorphism    A. Ganbat, T. Tsujimori, N. Boniface et al. Gondwana Research 91 (2021) 286-306 occured under epidote-amphibolite to amphibolite facies conditions. Lack of metamorphic zircon or overgrowth rim hampered the dating of metamorphic events for these samples. Seven zircons of two metagranitoids from the Northern Ufipa Terrane have low Th/U ratios, similar to the metamorphic zircons (Fig.  12a), suggesting tectonothermal overprints of Neoproterozoic collisional metamorphism at 570.4 ± 5.7 and 565 ± 7.7 Ma (Fig. 12b). Our data coincide with the previous findings. Several parts of the Ubendian Belt were re-worked during 590-500 Ma. U-Pb SHRIMP zircon ages of 593 ± 20, 548 ± 39, and 524 ± 12 Ma are obtained from kyanite-free eclogites in the Northern Ufipa Terrane ( Fig. 15; . Apatite is widely used for constraining time-temperature information on low temperature processes because of their comparably low closure temperature of 375-600°C (Cochrane et al., 2014;Kirkland et al., 2018). Apatite ages of samples from the Kate-Ufipa Complex and the Southern Ufipa Terrane (Fig. 2) yielded discordia lines with intercept ages between 1.88 ± 56 and 1.86 ± 80 Ga, respectively. Although, uncertainty of apatite ages is large, they are yielded obvious Paleoproterozoic ages, which is within the range of zircon ages. Smaller grain sized apatite (100-200 μm) in our samples probably constrain younger age of magmatic component given Pb-in-apatite diffusion an spherical geometry characteristics (Cherniak, 2005). Apatite typically incorporates substantial amounts of common Pb compared to U (i.e. low 238 U/ 204 Pb) when it crystallizes (Chew and Spikings, 2015;Kirkland et al., 2017). This results in a slow increase with time in the ratio of radiogenic Pb to non-radiogenic Pb (e.g. 206 Pb radiogenic/ 206 Pb initial). High common Pb content in the analyzed apatite array may imply interaction with a distinct, more radiogenic crustal source (Fig.  13). The grain shapes and textural relationships of apatite are indicative of a primary magmatic origin and its U-Pb age is interpreted as the time of last resetting during a thermal event at~1.86 Ga. All apatite samples yield similar high upper intercepts with 207 Pb/ 206 Pb ratios comparable to 1.9-1.8 Ga common Pb of the Pb evolution model by Stacey and Kramers (1975), implying the incorporation of Pb into the crystal structure from the magmatic environment during apatite crystallization. Paleoproterozoic apatites were only confirmed from the Southern Ufipa Terrane and the Kate-Ufipa Complex. These suggest that the Pan-African overprint effect was not equal throughout the Ufipa Terrane. Neoproterozoic tectonothermal events was not intense enough to anneal apatites in the Southern Ufipa Terrane and the Kate-Ufipa Complex. In contrast, the Neoproterozoic overprint was severe to reset zircon U-Pb age in the Northern Ufipa Terrane.

Geodynamic setting and implications for the Central African Shield amalgamation
The Paleoproterozoic geodynamic evolution of the Ubendian Belt is dominated by an active continental margin followed by collision (e.g. Boniface and Tsujimori, 2021). Metagranitoids from the Ufipa Terrane of the Ubendian Belt and northeastern portion of the Bangweulu Block emplaced between 1.89 and 1.85 Ga show silimar trace element pattern. They are formed by break-off of the subducting oceanic slab. Slab failure magmatism is possibly generated by melting of a torn subducting slab and the upper mantle above the subducting oceanic crust. It is followed by ascent and evolution of these magmas at the interface of the upper mantle and lower crust, which lead to partial melting of lower crust. While ascending, the magmas differentiated and mixed with the crustal magmas, resulting in the development of 1.89-1.85 Ga calc-alkaline magmas of the Ufipa Terrane and the northeastern Bangweulu Block (Fig. 16a). Consequently, the Mporokoso Group sediments are deposited from volcanic and plutonic source between 1.88 and 1.85 Ga (Fig.  16b).
Subduction and accretion in the Ubendian Belt to the Tanzania Craton apparently initiated within the Katuma and Lupa Terranes, and proceeded by fusion of the Upangwa and Nyika Terranes as they recorded the oldest magmatism ( Fig. 2; Fig. 13b). The activation of subduction culminated between 1.89 and 1.86 Ga, associated with crustal thickening and gold mineralization Kazimoto et al., 2014b;Lawley et al., 2013bLawley et al., , 2014. The subducting oceanic slab formed the Paleoproterozoic 1.89-1.87 Ga eclogites with MORB affinity in the Ubende Terrane is another evidence of the oceanic basin closure. Subsequent collision and accretion between the Tanzania Craton and the Bangweulu Block occurred at~1.83-1.82 Ga . The Ubendian Belt together with the Usagaran Belt evidences an important tectonothermal event: assembly of the Tanzania Craton and the Bangweulu Block with the Congo Craton and forming the Central African Shield as long-lived convergent margin of Archean craton fragments during Paleoproterozoic period (Alessio et al., 2019;Begg et al., 2009) and their participation to the Columbia supercontinent amalgamation.
A different subduction and accretion event in the Ubendian Belt is constrained by 570-565 Ma metamorphic zircon age and eclogite occurrence of the Usagaran Belt. Geochemical compositions indicate that precursors of these eclogites are associated with volcanic rocks of subducted oceanic crust and back-arc basin . These volcanic lavas attained their eclogite facies metamorphism after an episodic accretion between the Bangweulu Block and the Tanzania Craton along the Ubendian Belt, commenced at around 590 Ma . Closure and accretion were suggested by monazite crystal overgrowth rim of pelitic gneiss, yielding ages between 566 and 556 Ma Appel, 2017, 2018). This period is interpreted as subducting passive margin. The final continent-continent collision is marked by 530 and 520 Ma eclogites ( Fig. 16c; . The 590-520 Ma regional tectonothermal event A. Ganbat, T. Tsujimori, N. Boniface et al. Gondwana Research 91 (2021) 286-306 of the Ubendian Belt contemporaneous with Neoproterozoic consolidation of the Central African Shield and final Gondwana assembly by the Pan-African orogeny (e.g. Alessio et al., 2019). Eclogite with MORB and back-arc affinity coupled with the Paleoproterozoic active continental margin magmatism of the Ubendian Belt suggest that the Bangweulu Block took a part in the amalgamation of still large Central African Shield as a small separate unit surrounded by ocean during the assembly of the Columbia in Paleoproterozoic and the Gondwana in Neoproterozoic times.

Conclusions
Metagranitoids, including granoporphyries, from the South and North Ufipa Terranes of the Ubendian Belt and northeastern part of the Bangweulu Block were studied to constrain their geochemical and geochronological features. Based on geochemical, and geochronological data we propose that: (1) Paleoproterozoic metagranitoids and granoporphyries in the Ufipa Terrane and northeastern portion of the Bangweulu Block cannot be distinguished by geochronology and trace element geochemistry from the each other.
(2) The Ufipa Terrane of the Ubendian Belt has a protolith equivalent to Paleoproterozoic I-type granites. New U-Pb data suggest at least~40 Myr for the magmatic pulse, between 1.89 and 1.85 Ga.
(3) Geochemical features of metagranitoids such as enrichment in LREE and LILE (e.g. Rb, Ba, Cs, Pb), depletion of HFSE (Ti, Ta, Nb), high La/Yb, Sm/Yb, Gd/Yb and Sr/Y ratios and wide range of HREE interpreted in terms of their formation by melting of the torn slab and upper mantle above the subducting oceanic crust, followed by ascent to the lower crust and subsequent partial melting of lower crust.
(4) Zircons with low Th/U ratio yielded~570 Ma ages, suggesting a metamorphism during the Pan-African Orogeny in the Northern Ufipa Terrane. The protolith of the Ufipa Terrane is originated from the collided crustal rocks of the Bangweulu Block. Tectonothermal effect of Neoproterozoic collision was unequal throughout the Ufipa Terrane as we found non-annealed Orosirian age apatite in the Southern Ufipa Terrane and the Kate-Ufipa Complex.
(5) The Bangweulu Block and the Ubendian Belt participated in amalgamation of the Central African Shield as separated continents, surrounding oceanic crusts during the Paleoproterozoic Eburnean orogeny and the Neoproterozoic Pan-African orogeny.

Declaration of Competing Interest
None