The Tres Arroyos Granitic Aplite-Pegmatite Field (Central Iberian Zone, Spain): Petrogenetic Constraints from Evolution of Nb-Ta-Sn Oxides, Whole-Rock Geochemistry and U-Pb Geochronology

: Abundant Li-Cs-Ta aplite-pegmatite dykes were emplaced in the western Central Iberian Zone of the Iberian Massif during the Variscan Orogeny. Their origin and petrogenetic relationships with the widespread granitoids have led to a currently rekindled discussion about anatectic vs. granitic origin for the pegmatitic melts. To deal with these issues, the aplite-pegmatite dykes from the Tres Arroyos area, which constitute a zoned pegmatitic ﬁeld related to the Nisa-Alburquerque granitic batholith, have been studied. This work comprises a complete study of Nb-Ta-Sn oxides’ mineralogy, whole-rock geochemistry

The intermediate bodies occur further away from the NA batholith than the barren aplite-pegmatites (up to~500 m-far), constituting subhorizontal leucocratic dykes that intrude discordantly into the host Neoproterozoic-Lower Cambrian metasediments (Figures 1c and 2b). These aplite-pegmatites seem to be scarcer than the barren bodies, since only seven bodies have been identified. They show variable thickness (~1-8 m) and no internal concentric zoning. Nevertheless, some bodies display a rhythmic layering, where coarse crystals show comb-textures ( Figure 2c). Intermediate aplite-pegmatites are constituted mainly by albite and quartz, with minor K-feldspar and Al-micas. Topaz, Li-Al-phosphates, Nb-Ta-Sn-oxides, and fluorapatite appear as accessory phases ( Table 1).
The lithium-rich dykes correspond to highly evolved aplite-pegmatites, comprising a dozen bodies that appear up to 1 km away from the monzogranite (Figures 1c and 2d). These dykes display high contents of Li-rich micas, feldspars, topaz, and amblygonite-montebrasite series. These bodies are characterized by the lack of internal concentric zoning, showing instead a more or less complex layering (Figure 2e,f). The intermediate bodies occur further away from the NA batholith than the barren aplitepegmatites (up to ~500 m-far), constituting subhorizontal leucocratic dykes that intrude discordantly into the host Neoproterozoic-Lower Cambrian metasediments (Figures 1c and 2b). These aplitepegmatites seem to be scarcer than the barren bodies, since only seven bodies have been identified. They show variable thickness (~1-8 m) and no internal concentric zoning. Nevertheless, some bodies display a rhythmic layering, where coarse crystals show comb-textures ( Figure 2c). Intermediate aplite-pegmatites are constituted mainly by albite and quartz, with minor K-feldspar and Al-micas. Topaz, Li-Al-phosphates, Nb-Ta-Sn-oxides, and fluorapatite appear as accessory phases ( Table 1).
The lithium-rich dykes correspond to highly evolved aplite-pegmatites, comprising a dozen bodies that appear up to 1 km away from the monzogranite (Figures 1c and 2d). These dykes display high contents of Li-rich micas, feldspars, topaz, and amblygonite-montebrasite series. These bodies are characterized by the lack of internal concentric zoning, showing instead a more or less complex layering (Figure 2e,f).

Materials and Methods
The studied oxide minerals were selected from the three aplite-pegmatite types distinguished in the Tres Arroyos field, and to complete the sampling, a few of the studied specimens were collected from the dumps. Over 70 microprobe analyses were obtained from polished thin sections using a JEOL Superprobe JXA-8900 M electron microprobe (EM) at the "Centro de Microscopía Electrónica Luis Bru" from the Complutense University of Madrid (Madrid, Spain), and a CAMECA SX100 at the Scientific-Technical Services of the University of Oviedo (Oviedo, Spain). The operating conditions for the first laboratory were a voltage of 15 kV and beam current of 20 nA, and the calibrations standards used were lithium niobate and tantalite for Nb and Ta, kaersutite for Mg, almandine for Mn and pure metals for W, Ti, Fe and Sn. The operating conditions for the second were a voltage of 20 kV and beam current of 20 nA, and the calibrations standards used were apatite for Ca, modified titanite for Ti, MnS for Mn, magnetite for Fe, pure metals for Nb, Sn and Ta, modified MnWO4 for W and LiF for F. Data were reduced using the procedure of Pouchou and Pichoir [54], and analytical errors were estimated to be of the order of ±1-2% for major elements and ±10% for minor elements. The structural formulae of CT oxides and cassiterite were calculated on the basis of 6 and 4 oxygens, respectively.
Backscattered electron (BSE) images of oxides were obtained in the SGIker facility of the Basque Country University (UPV/EHU). The samples were previously polished and then covered with a carbon layer (20 nm) in order to obtain a conductive surface. The measurements were carried out by a JEOL JSM-6400 scanning electron microscope with a detector EDX Oxford Instruments Inca Energy 350. The operating conditions were a voltage of 20 kV and beam current of 10 nA. All images were taken with BSE to obtain compositional variations.
Fourteen bulk-rock samples of approximately 2-3 kg were collected and were jaw crushed to a nominal grain size of 5 mm. The crushed sample material was split using a riffle splitter to obtain a representative sub-sample, then milled to a grain size of approximately 100 microns (µm) using a vibratory micro mill (Pulverisette 0, Fritsch, Idar-Oberstein, Germany)). Ten representative bulk samples of aplite-pegmatites were analyzed for major-and trace-elements at the Activation Laboratories Ltd. (Actlabs, ON, Canada). Major elements and Zr values were provided by fusion ICP, whereas trace elements data were obtained by ICP-MS. Fluorine and Li values were obtained by fusion specific ion electrode (ISE) and by sodium peroxide fusion, respectively.
Measurements for the geochronological study were realized by laser-ablation quadrupole inductively-coupled plasma mass-spectrometry (LA-Q-ICP-MS) in the SGIker facility of the Basque Country University (UPV/EHU), using the UP213 laser system and the XSeries 2 ICP-MS. The laser beam was fixed to a 55 µm-wide square section. The spot was pre-ablated for 25 s using a laser repetition rate of 10 Hz and 40% output energy, followed by an ablation of 45 s at 10 Hz with a laser output energy of 75%. Zircon GJ_1 [55] was used for the calibration and correction every five measurements of the analyzed samples. Data reduction was carried out with Iolite 3.32 software (version, manufacturer, city, state abbreviation, country) [56] and VizualAge_UcomPbine (version, manufacturer, city, state abbreviation, country) [57]. See García de Madinabeitia et al. [58] for more details.
In the Tres Arroyos field, cassiterite occurs also as an accessory phase and shows different textural features depending on the aplite-pegmatite type ( Table 1). It appears as fine to medium-grained, subhedral, dark crystals (<2 mm) in barren and intermediate aplite-pegmatites, whereas in the most evolved aplite-pegmatites, cassiterite is anhedral to subhedral with a coarser grain size (<5 mm) ( Figure 4a). Cassiterite usually displays a typical strong pleochroism and chromatic zoning, with alternating dark (dark-brown to reddish-brown) and pale bands (light-brown to yellow; Figure 4b). In the Tres Arroyos field, cassiterite occurs also as an accessory phase and shows different textural features depending on the aplite-pegmatite type ( Table 1). It appears as fine to mediumgrained, subhedral, dark crystals (<2 mm) in barren and intermediate aplite-pegmatites, whereas in the most evolved aplite-pegmatites, cassiterite is anhedral to subhedral with a coarser grain size (<5 mm) (Figure 4a). Cassiterite usually displays a typical strong pleochroism and chromatic zoning, with alternating dark (dark-brown to reddish-brown) and pale bands (light-brown to yellow; Figure 4b).  In the Tres Arroyos field, cassiterite occurs also as an accessory phase and shows different textural features depending on the aplite-pegmatite type (Table 1). It appears as fine to mediumgrained, subhedral, dark crystals (<2 mm) in barren and intermediate aplite-pegmatites, whereas in the most evolved aplite-pegmatites, cassiterite is anhedral to subhedral with a coarser grain size (<5 mm) (Figure 4a). Cassiterite usually displays a typical strong pleochroism and chromatic zoning, with alternating dark (dark-brown to reddish-brown) and pale bands (light-brown to yellow; Figure 4b).
The sum of positive charges of the analyzed samples from Tres Arroyos is in the range 11.994-12.000 (Table 2; Table S2). Titanium, W, and Sn are the main trace elements in the CT oxides from Tres Arroyos. The Ti contents are higher in CT from barren aplite-pegmatites (up to 5.83 wt% TiO 2 ) than in the other types (<1.49 wt% TiO 2 ) ( Table 2; Table S2). The highest W values are also mainly associated with CT from barren aplite-pegmatites (2.67-4.71 wt% WO 3 ) (with the exception of one analysis related to irregular patches with WO 3 = 1.34 wt%). The lowest W values are observed in CT from Li-richest aplite-pegmatites (0.34-0.60 wt% WO 3 ), and mid-range values correspond to Nb-Ta oxides from the intermediate bodies (0.36-2.55 wt% WO 3 ) (with the exception of one analyses, that shows a WO 3 value of 5.45 wt%) ( Table 2; Table S2).
Like the Nb-Ta oxides, cassiterite also presents compositional variations depending on the different types of aplite-pegmatites. Some analyzed cassiterite crystals are close to the ideal SnO 2 composition, whereas others deviate significantly with SnO 2 contents as low as 90.01 wt% (Table 3; Table S3). Cassiterite associated with the Li-rich bodies is overall the purest one, with average wt% SnO 2 values close to 98%. Niobium, Ta, Fe, Mn, Ti, and W are the main elements substituting for Sn (Table 3; Table S3). The highest contents belong to Ta and Nb, which range from 0.00 to 5.80 wt% Ta 2 O 5 and from 0.00 to 3.45 wt% Nb 2 O 5 . The rest of the elements occur in general in lower proportions. Iron contents are higher in the barren and intermediate aplite-pegmatites, where it may reach up to 0.86 and 1.55 wt% FeO, respectively, whereas in the Li-rich aplite-pegmatites Fe contents are much lower, usually <0.1 wt% FeO (Table 3; Table S3). Only in this case does Mn predominate over Fe, with values up to 0.6 wt% MnO (Table 3; Table S3). In general, Mn contents are very low, only exceptionally >0.1 wt% MnO (Table 3; Table S3). The sum of cationic charges for all the analyzed cassiterite crystals is in the range of 7.95-8.00 (Table 3; Table S3). Titanium values for cassiterite are higher in the barren aplite-pegmatites (0.61-0.85 wt% TiO 2 ) than in the cassiterite from the intermediate and Li-rich aplite-pegmatites, where TiO 2 contents are in general <0.1 wt% (Figure 6a). The opposite behavior is observed for W, which presents the highest values in the cassiterite crystals from the Li-rich aplite-pegmatites, with values up to 0.98 wt% WO 3 (Table 3; Table S3).
Like the Nb-Ta oxides, cassiterite also presents compositional variations depending on the different types of aplite-pegmatites. Some analyzed cassiterite crystals are close to the ideal SnO2 composition, whereas others deviate significantly with SnO2 contents as low as 90.01 wt% (Table 3;  Table S3). Cassiterite associated with the Li-rich bodies is overall the purest one, with average wt% SnO2 values close to 98%. Niobium, Ta, Fe, Mn, Ti, and W are the main elements substituting for Sn (Table 3; Table S3). The highest contents belong to Ta and Nb, which range from 0.00 to 5.80 wt% Ta2O5 and from 0.00 to 3.45 wt% Nb2O5. The rest of the elements occur in general in lower proportions. Iron contents are higher in the barren and intermediate aplite-pegmatites, where it may reach up to 0.86 and 1.55 wt% FeO, respectively, whereas in the Li-rich aplite-pegmatites Fe contents are much lower, usually <0.1 wt% FeO (Table 3; Table S3). Only in this case does Mn predominate over Fe, with values up to 0.6 wt% MnO (Table 3; Table S3). In general, Mn contents are very low, only exceptionally >0.1 wt% MnO (Table 3; Table S3). The sum of cationic charges for all the analyzed cassiterite crystals is in the range of 7.95-8.00 (Table 3; Table S3). Titanium values for cassiterite are higher in the barren aplite-pegmatites (0.61-0.85 wt% TiO2) than in the cassiterite from the intermediate and Li-rich aplite-pegmatites, where TiO2 contents are in general <0.1 wt% (Figure 6a). The opposite behavior is observed for W, which presents the highest values in the cassiterite crystals from the Li-rich aplite-pegmatites, with values up to 0.98 wt% WO3 (Table 3; Table S3).

Geochronology
Columbite-tantalite crystals of two intermediate bodies were analyzed by the LA-Q-ICP-MS method and U-Th-Pb systematics for the age determination. Seventeen discordant analyses from two samples were selected, which display variable common Pb. After discarding a possible 207 Pb correction (cf. [57]), the regression of the data allowed us to provide a lower intercept age of 305 ± 9 Ma for the studied aplite-pegmatite bodies (Figure 7; Table 4).

Whole-Rock Geochemistry
A collection of 14 rock samples were selected for the whole-rock geochemical characterization of the aplite-pegmatites from Tres Arroyos (Table 5). Overall, the aplite-pegmatites show lower SiO 2 (67.17-73.74 wt%) and K 2 O (0.75-4.00 wt%) contents than the NA monzogranite, with a significantly wider range of K 2 O (   Table 5). The Sr shows a more complex behavior because it decreases from the monzogranite to the marginal granite and barren aplite-pegmatites, and then increases from barren, through intermediate, up to the Li-rich dykes (Table 5).

Geochronology
Columbite-tantalite crystals of two intermediate bodies were analyzed by the LA-Q-ICP-MS method and U-Th-Pb systematics for the age determination. Seventeen discordant analyses from two samples were selected, which display variable common Pb. After discarding a possible 207 Pb correction (cf. [57]), the regression of the data allowed us to provide a lower intercept age of 305 ± 9 Ma for the studied aplite-pegmatite bodies (Figure 7; Table 4).

Substitution Mechanisms in Nb-Ta-Sn Oxides
Chemical compositions of the members of the columbite-tantalite series can be expressed with the general formula AB 2 O 6 , where A and B positions are mainly occupied by Fe 2+ and/or Mn 2+ and Nb 5+ and/or Ta 5+ , respectively, although Ca 2+ , Mg 2+ , Fe 3+ , Sc 3+ , Ti 4+ , Sn 4+ , and W 6+ elements may also enter (e.g., [66]). Taking into account the sum of positive charges of the analyzed samples from Tres Arroyos ( Table 2; Table S2), the amount of Fe 3+ in the studied CT can be considered negligible. Plotting Mn 2+ vs. Fe 2+ (Figure 9a), the clear deviation of the data from the main trend for the CT oxides of barren aplite-pegmatites reflects the presence of other elements in the A site, in addition to Fe and Mn. Contrarily, Nb and Ta seem to exclusively fill the B site in most of the analyzed samples (Figure 9b). The plot 2(Sn+Ti) + W + vacancy vs. 2(Nb+Ta) + 2(Fe+Mn) (Figure 9c) shows that the substitutional scheme that appears to exert the dominant control over the incorporation of Ti and W into these oxides is the (Ti 4+ ,Sn 4+ ) 2 W 6+ 1vacancy1(Nb 5+ ,Ta 5+ ) exchange vector, particularly for CT associated with barren aplite-pegmatites, with a minor influence in the intermediate and Li-rich dykes.

Substitution Mechanisms in Nb-Ta-Sn Oxides
Chemical compositions of the members of the columbite-tantalite series can be expressed with the general formula AB2O6, where A and B positions are mainly occupied by Fe 2+ and/or Mn 2+ and Nb 5+ and/or Ta 5+ , respectively, although Ca 2+ , Mg 2+ , Fe 3+ , Sc 3+ , Ti 4+ , Sn 4+ , and W 6+ elements may also enter (e.g., [66]). Taking into account the sum of positive charges of the analyzed samples from Tres Arroyos ( Table 2; Table S2), the amount of Fe 3+ in the studied CT can be considered negligible. Plotting Mn 2+ vs. Fe 2+ (Figure 9a), the clear deviation of the data from the main trend for the CT oxides of barren aplite-pegmatites reflects the presence of other elements in the A site, in addition to Fe and Mn. Contrarily, Nb and Ta seem to exclusively fill the B site in most of the analyzed samples ( Figure  9b). The plot 2(Sn+Ti) + W + vacancy vs. 2(Nb+Ta) + 2(Fe+Mn) (Figure 9c) shows that the substitutional scheme that appears to exert the dominant control over the incorporation of Ti and W into these oxides is the (Ti 4+ ,Sn 4+ )2W 6+ 1vacancy1(Nb 5+ ,Ta 5+ ) exchange vector, particularly for CT associated with barren aplite-pegmatites, with a minor influence in the intermediate and Li-rich dykes. Similarly to Nb-Ta oxides, the sum of cationic charges for all the analyzed cassiterite crystals indicates that the amount of Fe 3+ is likely to be negligible. This is characteristic of pegmatitic cassiterite, since in hydrothermal cassiterites the presence of Fe 3+ becomes important (e.g., [60,67]). Moreover, the Similarly to Nb-Ta oxides, the sum of cationic charges for all the analyzed cassiterite crystals indicates that the amount of Fe 3+ is likely to be negligible. This is characteristic of pegmatitic cassiterite, since in hydrothermal cassiterites the presence of Fe 3+ becomes important (e.g., [60,67]). Moreover, the predominance of Mn over Fe seems to be a feature of extremely fractionated pegmatites [60]. The decrease in Ti in cassiterite from the less evolved bodies to the highly evolved ones has been previously observed in other aplite-pegmatite fields from the CIZ, and related to an increase in the fractionation degree (e.g., [12]). The incorporation of Nb, Ta, Fe, and Mn into cassiterite seems to be controlled by the substitution (Nb 5+ ,Ta 5+ ) 2 (Fe 2+ ,Mn 2+ ) 1 Sn 4+ −3 (Figure 6b). However, some cassiterite crystals from the Li-rich bodies deviate from this vector, indicating that other substitutions may have also been operative. For example, taking into account the small difference in cationic radii between Li + (0.82Å) and Sn 4+ (0.77Å), a possible incorporation of Li, via the substitution (Nb 5+ , Ta 5+ ) 3 Li + Sn 4+ −4 could balance the excess of charge in cassiterite generated by the presence of Nb and/or Ta (Figure 6c) [68]. Another substitution mechanism proposed for cassiterite is Sn 4+ −2 Fe 3+ Ta 5+ [67], but the negligible amount of Fe 3+ inferred from the chemical formula calculation excludes this mechanism.

Fractionation of Nb-Ta and Fe-Mn in Columbite-Tantalite Group Minerals
Fractionation of Nb-Ta and Fe-Mn in columbite-tantalite group minerals can be controlled by several factors, including the crystallization of different mineral phases during magma differentiation (e.g., [2,3,17]). The Ta/Nb ratio of CT oxides is expected to increase with fractionation (e.g., [15][16][17]). The solubility of HFSE (Sc, Y, Th, U, Pb, Zr, Hf, Nb, Ta, Ti) seems to be controlled by the temperature and abundance of fluxing elements, such as H 2 O, Li, B, P, and F [17]. It is experimentally proved that the solubility degree of MnTa 2 O 6 is higher than that of MnNb 2 O 6 in H 2 O-saturated peraluminous and metaluminous granitic melts [69], and hence columbite-(Mn) should crystallize before tantalite-(Mn). Fluorine, together with Li and P to a lesser extent, may play an important role in the solubility of Ta relative to Nb, with an important increase in the Nb and Ta solubilities when the content in F±Li±P increases (e.g., [3,[70][71][72][73][74]). In a "normal" fractionation sequence, the gradual increase in the F±Li±P content would lead to a higher solubility of both elements, but mostly in Ta [73], increasing the Ta/Nb ratio in the melt with fractionation. However, in the case of the CT oxides from Tres Arroyos, the Ta/Nb ratio firstly increases from the barren to the intermediate aplite-pegmatites, but later decreases from the intermediate to the Li-rich dykes. This lack of a progressive Ta/Nb increase could be related to the high F and P availability in the melts from the beginning of the pegmatite crystallization. Evidence for high F and P contents is provided by the presence of topaz and phosphates in the three aplite-pegmatite types, with a minimum of 2.5-3.0 wt% F in the melts (e.g., [49,75,76]). This would forestall the differential solubility of Nb and Ta during fractionation. The Ta/Nb decrease observed in the Li-rich aplite-pegmatites could be caused by the depletion of F, Li, and P in the melt by the crystallization of Li-F-bearing minerals, and subsequent reduction in the solubility of both Nb and Ta. Nevertheless, this decrease could be enough for Nb saturation, but not for the saturation of the (more soluble) Ta, which would remain mainly in the melt. Late metasomatic processes could be responsible for the development of Ta-rich irregular "patches" cross-cutting and masking the primary zoning of many CT crystals from barren and intermediate aplite-pegmatites, as it is found in other pegmatitic belts (e.g., [77][78][79]). Recent experimental works, nevertheless, question the role of fluorine in the partition of Ta/Nb ( [80,81]). Some researchers, such as Stepanov et al. [2] or López-Moro et al. [13], suggested that the peraluminosity and F-content have limited consequences in the partition of Ta/Nb, with the crystallization of muscovite being determinant for the fractionation of these elements. Thus, CT oxides could reflect certain control of micas on Ta and Nb fractionation in the parental melt, as suggested by Raimbault and Burnol [82] and Stepanov et al. [2] for some granitic melts and Llorens González et al. [83] for some Nb-Ta mineralizations from the CIZ. Moreover, average values of the Ta/Nb ratios in Al-micas of barren (0.19), intermediate (0.69) and Li-rich aplite-pegmatites (0.41) from Tres Arroyos [47] would correspond to the trend observed for the Ta/Nb ratios in CT oxides.
The controlling factors that lead to the fractionation of Fe-Mn are not consensual (e.g., [9] and references therein). The solubilities of the Fe-bearing CT members are more complex due to their dependence on redox conditions [17]. Iron-rich end-members are an order of magnitude more soluble than Mn-rich ones and, therefore, an enrichment in Fe with magma evolution should be expected (e.g., [71]). However, usually Mn values increase in pegmatitic bodies as fractionation proceeds, parallel to an increase in the X Ta ratio (e.g., [14][15][16]). Some authors related the enrichment of Mn in the melt to the high activity of fluorine (e.g., [15,79,84]). In contrast, other authors suggested that the Fe-Mn contents in the CT oxides are controlled by other phases (e.g., [12,16,17,[85][86][87]). Tourmaline, for example, is reported to show Mn-enrichment with fractionation (e.g., [88,89]). However, in Tres Arroyos, tourmaline occurs only in the granitic facies (with no CT oxides) and in the barren aplite-pegmatites. Consequently, early crystallization of tourmaline and phosphates would deplete the melt in Fe, contributing to the observed increase in the Mn/Fe ratio from those lithotypes to the more evolved ones. Nevertheless, tourmaline and phosphates crystallization could not explain the Mn-enrichment in the CT oxides from the most evolved aplite-pegmatites from Tres Arroyos. In this case, Fe enrichment should be expected rather than a Mn increase [87]. Micas were also proposed to control the Mn/Fe in the melt. Based on the muscovite-matrix partition coefficients reported by Raimbault and Burnol [82], a melt will become strongly depleted in Fe relative to Mn with muscovite crystallization. Additionally, the progressive crystallization of zinnwaldite may contribute to an increase in the Mn/Fe ratio, as described for the Beauvoir granite [90]. Similarly to the Ta/Nb ratios, the Mn/Fe ratio in white micas of Tres Arroyos increases progressively from the barren (0.00-0.06) through intermediate (0.00-5.2) to most evolved aplite-pegmatites (0.04-103.44) [47], reflecting the same pattern of CT oxides. It is conceivable, therefore, that the progressive depletion in Fe content could be caused by tourmaline, phosphates and mica crystallization.
As observed in other pegmatitic fields, such as Fregeneda-Almendra [26,29], the predominant chemical variation of CT oxides in Tres Arroyos may be capitalized on by the X Mn ratio, which clearly increases with fractionation, whereas the variation in X Ta is subordinated and not gradual. The progressive increase in the X Mn ratio from barren, through intermediate to the Li-rich aplite-pegmatites is interpreted to reflect a continuum of magmatic evolution. The crystallization sequence will depend on several factors (e.g., [29]), such as: (1) abundance of Fe, Mn, Nb, Ta, and Sn in the system; (2) temperature and undercooling degree; (3) degree of solubility of Nb-Ta oxides, affected by the presence of H 2 O, F, Li, B, and/or P; and (4) the coexisting mineral assemblage.
Assuming a crystallization temperature for the barren bodies >500 • C [11], the saturation of tantalite could be estimated over 1300 ppm of Ta for a system with a high content of fluxing elements (2 wt% F, 2 wt% Li) and over 120 ppm of Nb [17]. These values are higher than the data of bulk-rock for the Tres Arroyos aplite-pegmatites (54 ppm of Nb and 23 ppm of Ta; 40-118 ppm of Nb and 51-152 ppm of Ta; and 72-93 ppm of Nb and 64-91 ppm of Ta for barren, intermediate and Li-rich aplite-pegmatites, respectively). Contents needed for the saturation in Li-F-free environments are noticeably lower: 70 ppm for Nb and 510 ppm for Ta [17]. This may explain the late crystallization of CT oxides, mainly those richer in Ta, once the F-Li-bearing minerals have crystallized, hence reducing the Nb and Ta values needed for CT oxides saturation. The formation of some primary columbite-(Fe) and columbite-(Mn) could be related to punctual fluctuations in the composition of mineral-forming environment (local melt pools relatively rich in Nb-Ta) during centripetal aplite-pegmatites crystallization, probably with the local development of boundary layers where some incompatible elements are concentrated.

Formation of Primary Cassiterite
The stability of cassiterite depends on different chemical and physical conditions of the melts (e.g., [91] and references therein). According to some authors (e.g., [72,73,82]), the solubility of SnO 2 increases considerably if the melt is enriched in elements, such as F or Li. In contrast, Bhalla et al. [91] suggested that the F's effect in the solubility of SnO 2 in granitic melts is almost negligible, especially in reducing environments where Sn is divalent and acts as a HFSE, with the solubility being highly dependent on the f O 2 , Cl, Al and temperature. In the case of Tres Arroyos, the F content may have been a capital factor in controlling the crystallization of Sn-oxides. Tin would be transported as a fluoride (SnF 4 ) during the magmatic crystallization, and with increasing f (H 2 O) may precipitate via the reaction SnF 4 + 2H 2 O ↔ SnO 2 + 4HF (e.g., [92]). Furthermore, the release of volatiles, including HCl and HF, during the boiling of the system could also lead to the cassiterite saturation (e.g., [93]). This boiling can be caused by decompression of the system (first boiling) or by crystallization of non-hydrous minerals under isobaric conditions (resurgent or second boiling; [94]). In the case of Tres Arroyos, a significant pressure change could be related to the injection of the pegmatitic melts in open fractures, at low-pressure conditions (≈2-3 kbar; [43,46]), once the magmatic system opened and the most fractionated melts, accumulated in the apical part of the magma chamber, were expelled [20,31]. In this sense, Linnen et al. [95] and London [96] suggested that the crystallization of cassiterite in pegmatites may be associated with an undercooling of the system, exceeding the magmatic saturation temperature. Undercooling may be prompted by a sudden P or T change, or by a chemical variation implying an important decrease in fluxing components in the melt (e.g., [19,97]). In the case of Tres Arroyos, the undercooling of the system could be produced by a combination of these three factors. In addition to the just described sudden drop in pressure with the opening of the system (pressure quenching), the heat transfers from the pegmatitic melt to the colder host rocks would imply a thermal undercooling of the melt [48]. Moreover, the possible exsolution of a H 2 O-rich fluid from the pegmatitic melt associated with the boiling of this fluid could suppose a dramatic T decrease in the remaining melt, as this is an endothermic reaction. A loss of fluxing components from the pegmatitic melt, such as H 2 O and B (no tourmaline is found inside the intermediate and Li-rich aplite-pegmatite dykes, but it is common in their hosting metasediments as a result of a metasomatic reaction) would provoke a chemical quenching of the melt and the subsequent rise in the solidus [47].

Petrogenesis of the Aplite-Pegmatites from Tres Arroyos
As Simmons and Webber [97] postulated, there is no a universally accepted model for the granitic pegmatite genesis. However, pegmatitic melt-forming mechanisms can be grouped into two major processes: (1) fractional crystallization of a granitic magma and (2) direct anatexis. In the case of the Tres Arroyos aplite-pegmatite field, the low metamorphic grade of the host metasedimentary rocks [43,46], together with the absence of migmatitic terrains near the studied aplite-pegmatites, and compositional differences between whole-rock analyses of the studied aplite-pegmatites and those of minimum melt fractions from similar country rocks [20] preclude their origin by direct anatexis. The spatial distribution of the distinguished five lithotypes, from the monzogranite of the NA batholith, through the marginal granitic facies, barren aplite-pegmatites, and intermediate aplite-pegmatites, up to the Li-rich aplite-pegmatites, is consistent with a coherent genetic relation between the granitic massif and the aplite-pegmatites. The occurrence of mafic minerals (Fe-tourmaline and/or Fe-Mg-Mn-phosphates) in the barren aplite-pegmatites, as well as that of amblygonite-montebrasite in the intermediate and Li-rich dykes, indicate lower fractionation degrees for the barren and higher ones for the intermediate and Li-rich dykes (e.g., [49,98]). Compositional variations of the main minerals are also coherent with the observed regional zonation of the lithotypes (e.g., [11,47]). The decrease in Ti contents in quartz, as well as the increase and strong correlation of Al/Ti and Ge/Ti ratios, the trends followed by the Fe/(Fe+Mg) in tourmaline and K/Rb ratios in micas and K-feldspar, are consistent with the increase in the fractionation degree, from the NA monzogranite, through the marginal granitic facies, barren and intermediate aplite-pegmatites to Li-rich aplite-pegmatites [11,47]. Likewise, the progressive increase in the Mn/(Mn+Fe) ratio in columbite-tantalite group minerals from barren aplite-pegmatites up to the Li-rich bodies (Figure 5), suggests a single evolutionary sequence for the three types of studied aplite-pegmatites (e.g., [15,26]). Similarly, the higher Ti contents in cassiterites from the barren aplite-pegmatites, compared to those from the intermediate and Li-rich bodies (Figure 6), suggests that the parental melt of the former was less evolved (e.g., [12]). Whole-rock major-and trace-element compositions of the studied samples show a progressive evolution in the same sense of NA monzogranite → marginal granitic facies → barren aplite-pegmatites → intermediate aplite-pegmatites → Li-rich aplite-pegmatites (e.g., K/Rb ratio, Nb, Ta, Sn, Li, REE contents; Figure 8). In addition, geochemical modelling carried out for these aplite-pegmatites indicates that the evolution of the system could be explained by different fractionation degrees (up to~99%) of quartz + K-feldspar + plagioclase (±biotite, ±muscovite) [11]. Thus, all the above mentioned evidence supports an origin by fractional crystallization for the aplite-pegmatites from Tres Arroyos.
In the case of an origin by fractional crystallization of a granitic magma, the relationship between the pegmatitic melts and their parental granitic magma could be evaluated by means of a geochronological study, since this relationship is frequently obscured by the lack of a spatial and/or chemical continuity between them (e.g., [20,27,28]). The use of CT oxides with this purpose is supported by different facts: (1) columbite-tantalite group minerals cannot be inherited from the source of a granitic magma, (2) they show high U and low common Pb contents, and (3) they occur in pegmatites where zircon and monazite could be absent or inappropriate for dating (e.g., [1,6,32,99,100]). Moreover, the development of techniques, such as LA-ICP-MS, has allowed the measurement of unaltered and inclusion-free crystal zones in these oxides, becoming suitable mineral phases for the dating of pegmatitic bodies (e.g., [6,32,99]). Existing geochronological data on CIZ aplite-pegmatites manifest a diachronic character of these pegmatites based on U-Pb dating of minerals from the CT group, showing a southern propagation of ages in the range of 310-301 Ma ( [101]). Based on U-Pb zircon dating, granitic magmatism in the Nisa-Alburquerque-Los Pedroches Magmatic Alignment of the southern CIZ occurred from 314 to 304 Ma, (e.g., [40] and references therein), and the emplacement of the NA batholith was considered to be in the range of 305-309 Ma [44,45]. Thus, the obtained U-Pb age (305 ± 9 Ma) for the Tres Arroyos aplite-pegmatites would be in agreement with that of the granitic magmatism in the southern CIZ and would coincide with that of the latest pulses of the NA batholith and the CIZ pegmatites.
The NA batholith belongs to the P-rich, Ca-poor, highly peraluminous S 2 granitic suite defined by Villaseca [102] for the southern realm of the CIZ. Granites belonging to this suite are considered to be derived from Neoproterozoic to Early Cambrian psammopelitic metasediments of the SGC that were Ca-poor and P-(F, Li) rich [20]. The low Ca proportions in the starting melts derived from these metasediments would prevent the crystallization of apatite, allowing P and F to remain in a residual melt together with other fluxing elements. A high concentration of fluxing components such as P, F, B, Li and H 2 O in the granitic melts is believed to reduce their viscosity and polymerization degree [103] as well as their liquidus temperature. Thereby, internal diffusion, mobility of melts, and their ability to segregate from the source area are increased [104]. The lack of internal zonation in the aplite-pegmatite dykes without appreciable geochemical differences between the distinguished layers constituting the dykes, together with the crystallization of lepidolite across the whole width of the Li-rich dykes, suggest that the fractionation of the pegmatitic melt was insignificant during the crystallization of the aplite-pegmatites (e.g., [11]). Consequently, the fractionation process should be developed before the crystallization of the dykes, either (1) during lateral propagation of the pegmatitic melts or "en route" to the surface as proposed by Tartèse and Boulvais [105], or (2) by fractional crystallization + vertical movement of fluxing elements in the magmatic chamber (NA batholith). Although the lateral fractionation cannot be excluded, the strongly evolved compositions of the Li-rich dykes, together with the short distances from the granitic batholith and fast crystallization rates [11], point to the second hypothesis as the most likely scenario. Assuming this, a vertical zonation of the magmatic chamber should have been developed in Tres Arroyos, as it has been described for other pegmatitic fields in the CIZ [20,106], comprising different zones constituted by the parental magmas of each type of the distinguished aplite-pegmatites. Therefore, at the low emplacement levels of the NA batholith, once the vertical zoning of the magma chamber was developed, fluid pressure could be high enough for the system to open it at the apical part. In this way, the different melt fractions (± some minor mineral phases) would scape and reach up to distances that would vary depending on their location in the chamber, degree of fractionation and mobility. This would result in the zonation observed in the granite-aplite-pegmatite system of Tres Arroyos, with the Li-F-richest melt portions moving the furthest from the magma source.

Conclusions
The main conclusions of this contribution can be summarized as follows: (1) The main substitution mechanism for the CT group minerals seems to be (Ti 4+ ,Sn 4+ ) 2 W 6+ 1 vacancy 1 (Nb 5+ ,Ta 5+ ), whereas that for cassiterite would be (Nb 5+ ,Ta 5+ ) 2 (Fe 2+ ,Mn 2+ ) 1 Sn 4+ -3 . (2) The Mn/(Mn+Fe) ratio in CT oxides can be used as a petrogenetic indicator in the Tres Arroyos system, suggesting a single sequence of magmatic evolution from the less evolved to the most evolved bodies. The high F and P availability in the parental pegmatitic melts may have caused the lack of an expected progressive Ta/Nb increase with the fractionation. (3) The fractionation of Nb-Ta and Fe-Mn in the pegmatitic melts reflected in CT oxides seems to be controlled by tourmaline, phosphates and mica crystallization. (4) Some of the columbite-(Fe) and columbite-(Mn) crystals would have been crystallized by punctual chemical variations in the boundary layer, whereas the crystallization of other CT oxides, mainly those richer in Ta, would be controlled by Ta saturation in the pegmatitic melt. On the other hand, the formation of primary cassiterite may have been related to a high undercooling of the system. (5) Whole-rock geochemical data of the studied aplite-pegmatites and granitic units mirror the trends defined by mineral chemistry and strengthen the observed geochemical evolution from the NA monzogranite up to the Li-rich bodies. (6) The newly obtained U-Pb age of 305 ± 9 Ma on CT minerals from the Tres Arroyos aplite-pegmatites reinforces the genetic relationship between the studied aplite-pegmatites and the NA monzogranite, which would represent the parental granitic magma.