Detrital Tourmalines in the Cretaceous–Eocene Julian and Brkini Flysch Basins (SE Alps, Italy and Slovenia)

: In the SE Alps, two Cretaceous–Eocene flysch basins, Julian and Brkini, filled with turbidite sediments, are present. This study novelly reports heavy mineral assemblage counts and detrital tourmaline characterization for 11 samples. It is possible to define three different groups, characterized by the presence of (1) a clinopyroxene–epidote–low-ZTR (zircon+tourmaline+rutile; 5%) sample association, (2) a high-ZTR (>48%)–garnet–apatite association and (3) a low-ZTR (<40%)–Cr-spinel– garnet association. Detrital tourmalines from both the Julian and Brkini flysch basins are rather similar in composition, associated with metapelites and metapsammites coexisting or not coexisting with an Al-saturating phase, ferric-iron-rich quartz–tourmaline rocks and calc–silicate rocks; however, their number is drastically different. In fact, even if the percentage of heavy minerals is very low and similar in both basins (0.17–1.34% in weight), in the Julian basin, the number of tourmaline crystals is much lower than that in Brkini (1–14 vs. 30–100), suggesting an important change in the provenance area. Interestingly, the presence of a high amount of tourmaline derived from ferric-iron-rich quartz–tourmaline rocks and calc–silicate rocks makes these two basins different from all the Cretaceous flysch basins of Bosnia and the Northern Dinaric zone, where these supplies are missing or very limited.


Introduction
The Julian and Brkini sedimentary basins in the SE Alps are two nearby Cretaceous-Tertiary basins filled with turbidite sediments.In the past, several studies dealt with their mineralogy in terms of their main constituents, either clay or heavy minerals.As far as concerns the Julian basin, Cr-spinel [1] and HP-LT minerals such as amphibole and omphacitic pyroxene [2] have been found and studied.In the Brkini basin, Refs.[3,4] studied the heavy mineral paragenesis.A more detailed study on Cr-spinel in the Brkini basin was performed by [5], while [6,7] studied garnet, zircon and rutile in both the Julian and Brkini basins.

Geological Settings
The Apulian plate formed during the Early Mesozoic, following the break-up of Pangea and the consequent separation of Africa from Eurasia.Later on, during the Jurassic, extensive spreading of the seafloor occurred to the west of and north of the Adria plate.
extensive spreading of the seafloor occurred to the west of and north of the Adria plate.This caused the Tethys subduction below the eastern border of the Adria promontory.The Vardar Ocean formed by the end of the Jurassic, and in the Early and Middle Cretaceous, the collision of the microplate resulted in two suture zones, from which the Dinarides originated.Finally, in the Late Cretaceous, subduction culminated in continental collision of the Alpine orogenic belts, causing the creation of several foreland basins [26,27].In the studied area, it is possible to recognize the following basins: Claut and Clauzetto (Upper Paleocene-Lower Eocene) [28]; Julian (Upper Cretaceous-Middle Eocene) [29,30]; Vipava and Pivka (Early Eocene) [3]; Brkini (Lower-Middle Eocene) [31,32]; Istrian and Kvarner (Krk, Pag and Rab Islands) (Middle-Upper Eocene) [33] (Figure 1).Progressive emersion of the land caused the shifting to the SE of the basins so that they were consequently filled, with possible episodes of recycling [32].[34] indicating the main geological units and the main ophiolitic bodies.Flysch deposits of the Southeastern Alps and sample positions within the stratigraphic column.JB1 cannot be considered part of the stratigraphic succession of the Julian basin (see text for details), but it has been located at its correspondent age [35].No stratigraphic column is available for Brkini basin (see text for details).
The Julian basin (JB; sometimes described as the Slovenian Trough) is a sedimentary basin that has been active from the Maastrichtian (Upper Cretaceous) to the Middle Eocene; it is elongated in the NW-SE direction.According to [30], deposition of the turbiditic sequence was initiated during the Senonian and continued until the early Eocene [29].The turbiditic sequence is characterized by mixed siliciclastic-carbonate material with a total  [34] indicating the main geological units and the main ophiolitic bodies.Flysch deposits of the Southeastern Alps and sample positions within the stratigraphic column.JB1 cannot be considered part of the stratigraphic succession of the Julian basin (see text for details), but it has been located at its correspondent age [35].No stratigraphic column is available for Brkini basin (see text for details).The Julian basin (JB; sometimes described as the Slovenian Trough) is a sedimentary basin that has been active from the Maastrichtian (Upper Cretaceous) to the Middle Eocene; it is elongated in the NW-SE direction.According to [30], deposition of the turbiditic sequence was initiated during the Senonian and continued until the early Eocene [29].The turbiditic sequence is characterized by mixed siliciclastic-carbonate material with a total thickness of about 4000 m.According to [36], siliciclastic turbidites alternated with carbonate strata from the Dinaric carbonate platform during regression so that several carbonate megabeds, such as those of Ioanaz Mt., Staipa Mt., Topli Uorch, Vernasso and Porzus, are present [37][38][39].Venturini and Tunis (1992) [28] recognized quartz and calcite as the main constituents; plagioclases, clay minerals and dolomites are minor components, whereas microcline and micas are very rare.Among the lithic fragments, dolostones, limestones, radiolarites, cherts, diabase, sandstones, quartzites, gneisses and low-grade schist are present [28] (Figure 2).These authors classified them as lithic greywacke.Later on, Refs.[40,41] analyzed some of the samples present in this study, confirming the previous classification by [28].In the JB, Ref. [40] found in all samples chert, limestone and undetermined lithics.No significant changes in the lithics were noticed by [40].In the BK, Ref. [41] found chert, limestone, marl, schist and undetermined igneous rock fragments.thickness of about 4000 m.According to [36], siliciclastic turbidites alternated with carbonate strata from the Dinaric carbonate platform during regression so that several carbonate megabeds, such as those of Ioanaz Mt., Staipa Mt., Topli Uorch, Vernasso and Porzus, are present [37][38][39].Venturini and Tunis (1992) [28] recognized quartz and calcite as the main constituents; plagioclases, clay minerals and dolomites are minor components, whereas microcline and micas are very rare.Among the lithic fragments, dolostones, limestones, radiolarites, cherts, diabase, sandstones, quartzites, gneisses and low-grade schist are present [28] (Figure 2).These authors classified them as lithic greywacke.Later on, Refs.[40,41] analyzed some of the samples present in this study, confirming the previous classification by [28].In the JB, Ref.
[40] found in all samples chert, limestone and undetermined lithics.No significant changes in the lithics were noticed by [40].In the BK, Ref.
[41] found chert, limestone, marl, schist and undetermined igneous rock fragments.Tunis and Venturini (1992) [35] suggested that a possible source of the siliciclastic could be located in the nearby north and northeast areas of the basin, probably in the South Alpine area, implying longitudinal paleocurrents.In the lower Maastrichtian conglomerate at the base of the sequence in Bovec (Slovenia) (JB1), Ref. [42] found and studied about 60 well-sorted volcanic clasts (average diameter of 0.8 cm), with tholeiitic affinities showing similarities with the metabasites from the ophiolitic complexes of the Rhodopes and the Vardar zone.According to [43,44], the Bovec basin was a narrow basin belonging to the Julian nappe.Since the structure is complex and with scarce and laterally discontinuous exposure, it is difficult to locate the exact position of this basin within the regional paleogeography.The Bovec basin was presumably part of the Tolmin basin to the south but may be connected to the Belluno basin in the Southern Alps [43,44].In the proper Julian basin, by studying the Cr-spinel chemistry, Ref. [1] recognized both peridotitic and volcanic supplies, i.e., mantle (ophiolitic) peridotites and mantle-derived volcanic rocks sensu [45].Peridotitic spinels are more abundant and are characteristic of suprasubduction-zone harzburgites.The chemical diversities in volcanic spinels suggest multiple sources, including MORB and back-arc-type spinels, as well as oceanic island basalts and island arc tholeiite supplies [1].A change in the relative percentage between volcanic and peridotitic spinels appears at about 56 Ma, when volcanic sources increased.Such supplies suggest a provenance from the Inner Dinarides.Towards the top of the sequence, at about 52 Ma, Ref. [2] found actinolite, Mg-hornblende, barroisite and glaucophane associated with omphacite.These minerals are related to the erosion of HP-LT terranes.The paucity Tunis and Venturini (1992) [35] suggested that a possible source of the siliciclastic could be located in the nearby north and northeast areas of the basin, probably in the South Alpine area, implying longitudinal paleocurrents.In the lower Maastrichtian conglomerate at the base of the sequence in Bovec (Slovenia) (JB1), Ref. [42] found and studied about 60 well-sorted volcanic clasts (average diameter of 0.8 cm), with tholeiitic affinities showing similarities with the metabasites from the ophiolitic complexes of the Rhodopes and the Vardar zone.According to [43,44], the Bovec basin was a narrow basin belonging to the Julian nappe.Since the structure is complex and with scarce and laterally discontinuous exposure, it is difficult to locate the exact position of this basin within the regional paleogeography.The Bovec basin was presumably part of the Tolmin basin to the south but may be connected to the Belluno basin in the Southern Alps [43,44].In the proper Julian basin, by studying the Cr-spinel chemistry, Ref. [1] recognized both peridotitic and volcanic supplies, i.e., mantle (ophiolitic) peridotites and mantle-derived volcanic rocks sensu [45].Peridotitic spinels are more abundant and are characteristic of suprasubduction-zone harzburgites.The chemical diversities in volcanic spinels suggest multiple sources, including MORB and back-arc-type spinels, as well as oceanic island basalts and island arc tholeiite supplies [1].A change in the relative percentage between volcanic and peridotitic spinels appears at Minerals 2024, 14, 598 4 of 19 about 56 Ma, when volcanic sources increased.Such supplies suggest a provenance from the Inner Dinarides.Towards the top of the sequence, at about 52 Ma, Ref. [2] found actinolite, Mg-hornblende, barroisite and glaucophane associated with omphacite.These minerals are related to the erosion of HP-LT terranes.The paucity of these minerals and the limited extension in time suggest that they derived from probably limited outcrops, exhumed at about 56 Ma during a phase of Dinarides uplift [46].According to [6], supplies from amphibolite facies rocks and mafic and ultramafic metamorphic rocks are the most important for the garnets in the Julian basin.Moreover, skarns, very low-grade metabasites and ultra-high-temperature metamorphosed calc-silicate granulites are present.Recently, Refs.[47,48] studied the OH defects in detrital quartz crystals from the Julian basin and suggested that the supplies changed from a mixed metamorphic/magmatic (50%-50%) origin in samples older than 56 Ma to an almost exclusive metamorphic supply at the top of the sequence (less than 20% magmatic).
The Brkini basin (BK) occupies a syncline located in SW Slovenia, with an elongated NW-SE shape.Sedimentation was active from the Late Paleocene, and in particular, flyschoid and molassic material filled the basin during the Eocene, at a thickness of about 1000 m [32].The presence of strong folding and extensive vegetation cover prevent perfect recognition of the stratigraphic sequence, so it has never been wholly described.At the base, clastic deposits overlie Alveolina-bearing limestones with a 78 m thick olistostrome (Lower Eocene) [31].According to [32], the top of the succession can be attributed to the Middle Eocene.A 100 m thick sequence of siltstones, marls and sandstones in molasse facies closes the succession (Lutetian or post-Lutetian?) [32].The paleocurrent data are controversial.Orehek (1972Orehek ( , 1991) ) [3,49] suggested a sediment supply from the SE, while [32] observed transport from the NW.According to [3], within the heavy mineral paragenesis, it is possible to recognize rutile, zircon, tourmaline, garnet, pyrite and other opaque minerals.Among the opaque minerals, Ref. [4] described Cr-spinel and ilmenite, while in the transparent heavy minerals, they also noticed the presence of orthopyroxene.According to [47], the clastic succession is constituted by lithic greywacke.Lenaz et al. (2003) [5] showed that in the Brkini flysch, the Cr-spinels are mainly peridotitic (80%), and their chemistry suggests a provenance from the same areas of those in the Julian basin.Garnets are represented by those originating from amphibolite facies rocks, whereas those from high-grade metabasic rocks are very few.It is interesting to notice that concerning the Bi-type garnets (sensu [50]), there are, according to their trace element content, two different types of garnets in the BK.The garnets from the sample BK41 show an Eu/Eu* ratio higher than 0.1, similar to those present in the JB, while the garnets from the other samples show an Eu/Eu* ratio lower than 0.1.According to this difference, Ref. [6] ascribed the sample BK41 to the bottom of the sequence and the others to the top.
De Min et al. (2014) [51] studied the whole rock geochemistry of several sandstone samples in the Julian basin, showing that the first strata are chemically influenced by the disaggregation of metamorphic and not-metamorphic rock types related to the ancient Vardar sea closure, where island-arc-and MORB-like-related rock types were generated.Successively, at about 56 Ma, there was strong involvement of continental upper crustal source rocks.Later on, a new upwelling involved the Julian sediments, which, possibly, partly contributed, with recycled materials, to the creation of the Brkini basin.In the latter basin, other protolith rock types begin to be significant.

Materials and Methods
Seven samples from the Julian basin were selected according to their stratigraphic age: JB5 is Maastrichtian, JB10 is Danian, JB14 is Thanetian and JB17, JB21, JB24 and JB25 are Ypresian.Three samples from the Brkini basin were selected considering their Cr-spinel and garnet content has been already studied.BK41 is from the base of the sedimentation cycle in Brkini, BK35 is from the top and BKNV represents the molasse facies closing the sedimentation in the basin.JB1 is a Maastrichtian sample that cannot be ascribed to the sequence of the Julian basin, but it is rather close.Major details on it can be found in the "Discussion" section.The rocks were crushed in a mortar; then, the heavy minerals were concentrated using sieving and heavy liquids (sodium polytungstate Na 6 (H 2 W 12 O 40 )) at the Department of Earth and Environmental Sciences of Bicocca University in Milan, according to the procedure in [52].We analyzed heavy minerals with a broad spectrum of grain sizes between 75 and 300 microns because this was the only type of material available for heavy mineral analysis.About 200 transparent granules were recognized.The analyzed samples are very rich in heavy grains that are opaque and semi-opaque in transmitted light, often weathered, overwhelming the signal of transparent heavy minerals.While recognizing dimensional variety in the observed granules, it was preferred to use the area method to obtain the suite of minerals, applying a procedure that was not timeconsuming [53].In samples in which transparent heavy minerals prevail, the point counting method has often been applied [54], but even this does not completely solve the problem of 2D vision under an optical microscope, and therefore the transformation of frequencies into percentages remains an approximation [54,55].To minimize the impact of the abundance of opaque and semi-opaque grains, all grains spread in a randomly drawn area on the slide were counted, reconstructing the relative frequency ratios of all the varieties of grains and minerals encountered.Then, for the subsequent areas, only transparent heavy minerals were considered to create a robust dataset, and only at the end was an estimate of the percentages of opaque and semi-opaque granules calculated following the previous relative proportions.
Raman spectra for olivine crystals were collected in the 140-1900 cm −1 spectral range with a Renishaw inVia TM (Wotton-under-Edge, UK) spectrometer equipped with a Leica DM2500 (Wetzlar, Germany) polarizing microscope, a 50 × LWD (long working distance) objective, a solid-state 532 nm laser and a grating of 1800 lines/mm.The acquisition time was ~15 s, and the spectral resolution was ±1 cm −1 .Calibration was carried out before each session with the internal standard of a silicon wafer at 520.6 cm −1 .
Then, all the tourmalines from every studied separate were hand-picked under the microscope (Table 1).Microprobe analyses were performed at the Milano Statale University using a Superprobe JEOL Jxa-8200 (Tokyo, Japan), a 1 µm beam spot, 15 kV and 5 nA emission.The counting time was 30 s for the peak and 10 s for the background.The following standards were used for the analyses: omphacite USNM 110-697-154 for Na; Martire grossular for Al, Si and Ca; fayalite USNM 85276-143 for Fe; rhodonite for Mn; K-feldspar PSU-Or-1A-113 for K; Olivine (Fo83) USNM 2566-153 for Mg; ilmenite USNM 96189-149 for Ti and pure V, Zn, Cu and Cr.In the presence of zoned crystals, spot analyses were performed on areas with different chemistry.The generalized tourmaline structural formula is XY 3 Z 6 (T 6 O 18 )(BO 3 ) 3 V 3 W, where the most common ions (or vacancy) at each site are X = Na 1+ , Ca 2+ , K 1+ and vacancy; Y = Fe 2+ , Mg 2+ , Mn 2+ , Al 3+ , Li 1+ , Fe 3+ and Cr 3+ ; Z = Al 3+ , Fe 3+ , Mg 2+ and Cr 3+ ; T = Si 4+ , Al 3+ and B 3+ ; B = B 3+ ; V = OH 1− and O 2− and W = OH 1− , F 1− and O 2− [56].The most compositional variability occurs at the X, Y, Z, W and V sites.Tourmaline species are defined in accordance with the dominant valency rule, such that at a relevant site, the dominant ion of the dominant valence state is used for the basis of the nomenclature.To calculate the structural formulae for the tourmaline analyses based on the current International Mineralogical Association nomenclature scheme (IMA-2011; [56]), we used the Microsoft Visual Basic program, WinTcac, by [57].This software calculates tourmaline analyses both based on 31 O atom normalization as the default and 15 cations and 6 silicons optionally, shares out the recalculated cations at different sites (T, Z, Y and X) and estimates the OH −1 and O 2− contents at the V site and the OH −1 , F −1 , Cl −1 and O −2 contents at the W site, which is a recommended approach because it is appropriate for tourmalines in almost all metamorphic and most igneous rocks [56].The latter is the approach we adopted.For the chemical analyses, see the Supplementary Data.

Results
After separation, we collected an amount of heavy minerals corresponding to about 0.17-1.34wt.% of the rocks (Table 1).
The heavy mineral assemblage of the studied samples shows that the JB1 sample is very different from all the others, being dominated by a clinopyroxene-epidote-titanite assemblage and with a prevalence of hydrogrossular over other garnets (Table 2; Figure 3).The ZTR value is very low, about 5 wt.% (Figure 3).The other older samples from JB, JB5 and JB10, as well as BK35, show ZTR values between 48 and 55% in association with garnet and apatite (Figure 3).It is worth noticing that in the JB samples, there are certain amounts of clinopyroxene (augite and diopside), olivine and Cr-spinel that are missing in the BK one.The occurrence of olivine in sedimentary detritus is usually rare due to its lowest stability among all detrital heavy minerals [58].In order to understand the possible forsterite content of these olivines, Raman analyses were performed on the crystals found.Olivines were analyzed at both the cores and rims, and no compositional changes were noted.According to [59], the dominant feature of the olivine Raman spectrum is a doublet that occurs in the spectral region of 815-825 cm −1 and 838-857 cm −1 , showing monotonic shifts following the cation substitution between forsterite and fayalite (Figure 4).This fact allows us to estimate the chemical composition from the Raman spectra.Interestingly, the rare olivine crystals display a different forsterite content.In fact, in JB5, the unique olivine crystal presents a double peak at 820 and 851 cm −1 , suggesting a Fo65 composition.In JB10, there are two olivine crystals: one is similar to the one found in JB5, while the second shows a double peak at 823 and 855 cm −1 (Fo90).The remaining olivines (JB21 and JB24) are both Fo90.All the other samples are characterized by a ZTR value slightly decreasing with age (from about 40 to 24) and an association mainly comprising Cr-spinel and garnet.The other older samples from JB, JB5 and JB10, as well as BK35, show Z between 48 and 55% in association with garnet and apatite (Figure 3).It is wor that in the JB samples, there are certain amounts of clinopyroxene (augite and olivine and Cr-spinel that are missing in the BK one.The occurrence of olivi mentary detritus is usually rare due to its lowest stability among all detrital he als [58].In order to understand the possible forsterite content of these olivin analyses were performed on the crystals found.Olivines were analyzed at bot and rims, and no compositional changes were noted.According to [59], the dom ture of the olivine Raman spectrum is a doublet that occurs in the spectral reg 825 cm −1 and 838-857 cm −1 , showing monotonic shifts following the cation su  Considering the heavy mineral percentages, it is possible to see that they are slightly greater in the BK compared to the JB (avg.0.8 ± 0.3 vs. 0.5 ± 0.4 wt.%).In the JB, there is no trend apart from a slight increase (0.3 to 0.5 wt.%) in its youngest sample (Table 1).The amount of tourmaline is low, 1 to 14 crystals, and there is no trend (Table 1).In the BK, there is a clear inverse correlation between the percentage of heavy minerals and the number of tourmalines; in fact, in BK35, there is the lowest amount of heavy minerals (0.4 wt.%) and the highest number of tourmalines (100), while in the molasse sample BKNV, there is the highest amount of heavy minerals (1.1 wt.%) and the lowest number of tourmalines from the BK (30) (Table 1).Zoned tourmalines are present in all the Ypresian samples from the JB and in the JB5 sample.In the BK, they are always present (Figure 5).Considering the heavy mineral percentages, it is possible to see that they are slightly greater in the BK compared to the JB (avg.0.8 ± 0.3 vs. 0.5 ± 0.4 wt.%).In the JB, there is no trend apart from a slight increase (0.3 to 0.5 wt.%) in its youngest sample (Table 1).The amount of tourmaline is low, 1 to 14 crystals, and there is no trend (Table 1).In the BK, there is a clear inverse correlation between the percentage of heavy minerals and the number of tourmalines; in fact, in BK35, there is the lowest amount of heavy minerals (0.4 wt.%) and the highest number of tourmalines (100), while in the molasse sample BKNV, there is the highest amount of heavy minerals (1.1 wt.%) and the lowest number of tourmalines from the BK (30) (Table 1).Zoned tourmalines are present in all the Ypresian samples from the JB and in the JB5 sample.In the BK, they are always present (Figure 5).
According to [56], for a first classification, the diagram X □/( X □+Na+K) vs. (Mg/(Mg+Fe), where □ is a vacancy, in Figure 6 has been proposed to illustrate the generalized tourmaline species found in our samples.It is possible to see that almost all fall into the dravitic tourmaline field.
In the JB samples, about 9%-10% of the tourmalines fall into the field of schörl, while in the BK, the schörl tourmalines range between 8 (BK35) and 32% (BKNV) (Figure 7).There is only one foititic tourmaline in BKNV.If we consider the rims/overgrowth, almost all are dravitic tourmalines in composition, with minor schörl types present.According to [56], for a first classification, the diagram X □/( X □+Na+K) vs. (Mg/(Mg+Fe), where □ is a vacancy, in Figure 6 has been proposed to illustrate the generalized tourmaline species found in our samples.It is possible to see that almost all fall into the dravitic tourmaline field.In the JB samples, about 9%-10% of the tourmalines fall into the field of schörl, while in the BK, the schörl tourmalines range between 8 (BK35) and 32% (BKNV) (Figure 7).There is only one foititic tourmaline in BKNV.If we consider the rims/overgrowth, almost all are dravitic tourmalines in composition, with minor schörl types present.According to [56], for a first classification, the diagram X □/( X □+Na+K) vs. (Mg/(Mg+Fe), where □ is a vacancy, in Figure 6 has been proposed to illustrate the generalized tourmaline species found in our samples.It is possible to see that almost all fall into the dravitic tourmaline field.In the JB samples, about 9%-10% of the tourmalines fall into the field of schörl, while in the BK, the schörl tourmalines range between 8 (BK35) and 32% (BKNV) (Figure 7).There is only one foititic tourmaline in BKNV.If we consider the rims/overgrowth, almost all are dravitic tourmalines in composition, with minor schörl types present.As can be seen above, there is a strong difference between the content of tourmalines within the basins.In the eight JB samples, we found a total amount of 54 tourmalines, while in three samples from the BK, there are 184 tourmalines, even if the percentage of heavy minerals is not so different (0.5 vs. 0.8 wt.%).This fact clearly indicates a substantial change in the provenance of the tourmalines.Considering their compositions, these tourmalines are almost the same; in fact, considering all the tourmalines from the JB and those from BK41 and BK35, about 80%-90% are dravitic tourmalines, with the remnants being schörlitic ones.Only the youngest sample, BKNV, shows some differences.It is enriched in schörl (30%), and there is one foititic tourmaline (Figure 7).As can be seen above, there is a strong difference between the content of tourmal within the basins.In the eight JB samples, we found a total amount of 54 tourmali while in three samples from the BK, there are 184 tourmalines, even if the percentag heavy minerals is not so different (0.5 vs. 0.8 wt.%).This fact clearly indicates a substan change in the provenance of the tourmalines.Considering their compositions, these to malines are almost the same; in fact, considering all the tourmalines from the JB and th from BK41 and BK35, about 80%-90% are dravitic tourmalines, with the remnants be schörlitic ones.Only the youngest sample, BKNV, shows some differences.It is enric in schörl (30%), and there is one foititic tourmaline (Figure 7).
Considering such diagrams, it is easy to see that the main tourmaline supplies are from ferric-iron-rich quartz-tourmaline rocks, calc-silicate rocks and metapelites, followed by metapelites and metapsammites not coexisting with an Al-saturating phase.In the JB, there are two tourmalines from low-Ca metaultramafics and Cr, V-rich metasediments and metacarbonates and meta-pyroxenites.In BK41 and BKNV, there are tourmalines from Li-poor granitoids and their associated pegmatites and aplites, while in BK35 and BKNV, there are a couple of tourmalines from ferric-iron-rich quartz-tourmaline rocks.Similarly to the JB, there are some tourmalines from low-Ca metaultramafics and Cr, V-rich metasediments and metacarbonates and meta-pyroxenites.
As far as concerns the zoned tourmalines, it is possible to recognize two different behaviors.As an example, it is possible to see that almost all of the rims of the tourmalines from BK35 show an increase in iron moving towards the Al50Fe50 vertex; on the contrary, the rims of the crystals from BK41 and BKNV move towards the Al50Mg50 vertex (Figure 9).Considering such diagrams, it is easy to see that the main tourmaline supplies are from ferric-iron-rich quartz-tourmaline rocks, calc-silicate rocks and metapelites, followed by metapelites and metapsammites not coexisting with an Al-saturating phase.In the JB, there are two tourmalines from low-Ca metaultramafics and Cr, V-rich metasediments and metacarbonates and meta-pyroxenites.In BK41 and BKNV, there are tourmalines from Li-poor granitoids and their associated pegmatites and aplites, while in BK35 and BKNV, there are a couple of tourmalines from ferric-iron-rich quartz-tourmaline rocks.Similarly to the JB, there are some tourmalines from low-Ca metaultramafics and Cr, V-rich metasediments and metacarbonates and meta-pyroxenites.
As far as concerns the zoned tourmalines, it is possible to recognize two different behaviors.As an example, it is possible to see that almost all of the rims of the tourmalines from BK35 show an increase in iron moving towards the Al50Fe50 vertex; on the contrary, the rims of the crystals from BK41 and BKNV move towards the Al50Mg50 vertex (Figure 9).Considering such diagrams, it is easy to see that the main tourmaline supplies are from ferric-iron-rich quartz-tourmaline rocks, calc-silicate rocks and metapelites, followed by metapelites and metapsammites not coexisting with an Al-saturating phase.In the JB, there are two tourmalines from low-Ca metaultramafics and Cr, V-rich metasediments and metacarbonates and meta-pyroxenites.In BK41 and BKNV, there are tourmalines from Li-poor granitoids and their associated pegmatites and aplites, while in BK35 and BKNV, there are a couple of tourmalines from ferric-iron-rich quartz-tourmaline rocks.Similarly to the JB, there are some tourmalines from low-Ca metaultramafics and Cr, V-rich metasediments and metacarbonates and meta-pyroxenites.
As far as concerns the zoned tourmalines, it is possible to recognize two different behaviors.As an example, it is possible to see that almost all of the rims of the tourmalines from BK35 show an increase in iron moving towards the Al50Fe50 vertex; on the contrary, the rims of the crystals from BK41 and BKNV move towards the Al50Mg50 vertex (Figure 9).

Discussion
The difference between the heavy mineral assemblage of JB1 and the other samples, as well as the rare presence of olivines, is worthy of an attempt at an explanation, even if it is not the main purpose of this study.In the area close to Bovec (Slovenia), ascribed to the lower Maastrichtian, conglomerate and arenite outcrop.In the conglomerate, [42] found and analyzed sixty well-sorted volcanic clasts.The clasts were variably spilitized.They are classified as tholeiitic, andesitic and transitional basalts, with pheno-and microphenocrystals of plagioclase, augite (Wo 39-42 En 41-48 Fs 11-20 ), pigeonite and opaques.Olivine is usually altered into iddingsite.Their chemistry suggested similarities with metabasites from the ophiolitic complexes of the Rhodopes and the Vardar zone and a provenance from quite a delimited and close protolithic area (1-3 hundred kilometers) that did not suffer from sub solidus recrystallization.
In the Dinarides areas, several different Mesozoic flysch basins can be found.The Bosnian flysch forms a ca.3000 m thick intensely folded deposit of Upper Jurassic to Cretaceous mixed carbonate and siliciclastic sediments [61].The trench sedimentation was controlled by a dual sediment supply from the sub-ophiolitic high-grade metamorphic soles and from the distal continental margin of the Adriatic plate.The heavy mineral spectra of the Bosnian flysch are overall dominated by Cr-spinel, followed by ZTR, garnet and lesser amounts of apatite, titanite and monazite.Epidote was found in the lower Cretaceous Vranduk formation [61].No clinopyroxene is present in the heavy mineral assemblage.More northward, the Late Cretaceous consisted of more or less coeval basins, receiving mixed ophiolitic-continental siliciclastic detritus [62].
There, the Maastrichtian Glog formation represents the youngest part of a transgressive Upper Cretaceous alluvial to deep-water succession overlying the rocks of the Paleozoic-Triassic basement wedge, which underwent low-grade Cretaceous metamorphism.In this formation, Ref. [63] found high ZTR, with only minor proportions of heavy minerals such as epidote and garnet.The deep-water turbiditic deposits of the Kravlijak formation (Albian-Cenomanian age) are dominated by ZTR but also contain significant Cr-spinel and minor amounts of garnet and apatite, while the Vivodina formation (Maastrichtian) is a deep-water deposit containing ZTR and considerable amounts of Cr-spinel, as well as minor amounts of garnet and epidote/zoisite [64].Dismembered ophiolites occur in the Dinaride ophiolite zone, related to the open-ocean Tethyan realm, whereas highly dismembered ophiolites occur in the Vardar zone, related to a back-arc basin.According to [65], the ophiolites are predominantly fertile spinel lherzolite (olivine Fo 89.9 ) in the western and central parts of the Dinaride ophiolite and Vardar zones, with depleted harzburgites (olivine Fo 91.1 ) in their southeastern parts.Cumulate ultramafics (with the same mineralogy as the peridotites but enriched in iron; olivine Fo 85 ) and gabbros are subordinate and are in some places overlain by massive or sheeted dyke complexes, capped by metabasaltic pillow lava.Diabases and dolerites are composed of plagioclase, augite, actinolite and olivine Fo 70 .These bodies are cross-cut by veins and veinlets filled with epidote, among others.The metamorphic soles of ophiolites are represented by varieties of amphibolites with subordinate pyroxenite schists and scarce eclogites with ultramafic interlayers, which were progressively metamorphosed under the P-T conditions of eclogite, granulite, amphibolite and greenschist facies.Amphibolites related to diabases-dolerites are usually bimineralic and composed of amphibole and plagioclase, with subordinate epidote.
Within the Alpine domain, since the Oxfordian, several flysch basins have been present in the area, including the Belluno, Bovec, Bled and Tolmin basins [43,66].As seen above, Refs.[43,44] suggested the Bovec basin was presumably part of the Tolmin basin but may have belonged to the Belluno one.Stefani et al. (2007a) [67] suggested a connection between these two basins as well.However, Ref. [67] showed that in the Belluno basin, the heavy mineral assemblage is rather different from that found in the JB1 sample, being enriched in ZTR (from 10% up to 40%), Cr-spinel, epidote and garnet but completely missing pyroxene and titanite.
Given this evidence, it seems that in our samples, the peridotite ophiolitic source is closer to the NW Dinarides formations, while metamorphic sole supplies are limited.On the contrary, the epidote-bearing metamorphic sole and all the sensu lato basaltic (diabase and dolerite) source rocks that could be related to ophiolites are possibly located to the north of this area and closer to the Bovec and Julian basins.Only later does the presence of ophiolite detritus increase in the heavy mineral assemblage of the Julian and Brkini basins.This can be also detected from the change in the forsteritic content of the rare olivine grains.At the beginning of sedimentation, they are Fo65, with a forsteritic content similar to that found in fresh olivine crystals found in the Triassic magmatism of the Alps [68], while it changes to Fo90 in the younger rocks, a value resembling that of the olivine in the peridotite rocks of the Dinarides (and Cr-spinel is associated).
There are few studies reporting the presence of tourmaline in the surroundings of the studied basins.There are detrital tourmalines in the Paleozoic Slavonian Mountains [69] and in the Late Miocene basins of Northern Croatia [70], but no chemical analyses of these are present.In the Paleozoic Hochwipfel (the Karawanken Mountains between Austria and Slovenia), Ref. [71] reported the presence of detrital tourmaline in both the Eastern and Western Karawanks (Figure 10).The Eastern Karawanken tourmalines show compositions similar to metapelites and metapsammites coexisting with an Al-saturating phase and metapelites and metapsammites not coexisting with an Al-saturating phase (fields 4 and 5 in the diagram by [60]).Those from the Western Karawanks show compositions similar to those from ferric-iron-rich quartz-tourmaline rocks, calc-silicate rocks and metapelites (Field 6 by [60]) but with an Al content lower than that of the here-studied tourmalines (Figure 9).Lužar-Oberiter et al. (2012) [62] studied detrital tourmalines from the Cretaceous basins of the NW Dinarides in Croatia (Oštrc, Bistra, Kravljak, Vivodina and Glog formations).Tourmalines from granitoid sources (Field 2 by [60]) are slightly more common in the Early Cretaceous and Cenomanian than in the Uppermost Cretaceous (Figure 10).Considering this, it is hard to define whether the rims originate during their magmatic/metamorphic origin or in a diagenetic environment.However, it is possible that, at least for some samples, these effects could be due to specific diagenetic conditions (Figure 11).Tourmaline can be stable under very diverse pressure-temperature conditions, from diagenetic to ultra-high pressure and high-temperature conditions, with the latter showing an increase in Al at the T site and an increase in the Na, Ca and K contents and a decrease in the vacancies at the X site [90][91][92].In the Lower Cretaceous, tourmaline analyses fall within the field of metapelites coexisting with an Al-saturating phase (Oštrc and Bistra), while those from the Cenomanian and Uppermost Cretaceous (Kravljak, Vivodina and Glog) are metapelites not coexisting with an Al-saturating phase (Field 5 by [62]) (Figure 10).In the Cretaceous Bosnian flysch, Ref. [61] found tourmalines falling in fields 4 and 5 in the diagram by [60] (Figure 10).Consequently, it is possible to say that in all these cases, apart from the one from the Western Karawanks, the here-studied tourmalines are different.Balen and Petrinec (2011) [72][73][74][75] analyzed tourmalines from Cretaceous Moslavačka granite (Croatia), Permian-Triassic Sopron orthogneiss (Hungary), Cretaceous Pohorje marble (Slovenia) and Permian Velence granite (Hungary), but none of these tourmalines are similar to our detrital ones, so a more precise determination of their provenance is rather impossible at present.
Zoned tourmalines can occur in metapelites, gneisses and pegmatites [76][77][78], with tourmaline grains displaying distinct asymmetric chemical zoning indicating a diagenetic and/or low-grade metamorphic origin [79].The formation of rims in tourmaline can be due to the change in the fluid circulating during its crystallization within the source rock or during the diagenetic phase.Pirajino and Smithies (1992) [80] showed that within granite, there is a correlation between the distance from the core to the distal part and the content of MgO, with MgO increasing from endogranitic to proximal, from proximal to intermediate and from intermediate to distal.According to [81,82], tourmaline crystallizes in weakly to moderately acidic solutions but not in alkaline ones.Authigenic tourmaline shows significant X-site vacancies (as a result of low Na content in fluids) and a low Ca content [82][83][84][85][86].The magnesium content can be very variable, as significant content can be the result of the circulation of marine brines [87,88], while a distinct decrease in Mg and an increase in Fe in the growth direction could be considered an effect of the fluid dilution with respect to Na and Ca over time [88], a decrease in the fluid temperature or, in a metamorphic environment, a decreasing temperature of crystal growth [89].Considering this, it is hard to define whether the rims originate during their magmatic/metamorphic origin or in a diagenetic environment.However, it is possible that, at least for some samples, these effects could be due to specific diagenetic conditions (Figure 11).Tourmaline can be stable under very diverse pressure-temperature conditions, from diagenetic to ultra-high pressure and high-temperature conditions, with the latter showing an increase in Al at the T site and an increase in the Na, Ca and K contents and a decrease in the vacancies at the X site [90][91][92].Tourmaline can be stable under very diverse pressure-temperature conditions, from diagenetic to ultra-high pressure and high-temperature conditions, with the latter showing an increase in Al at the T site and an increase in the Na, Ca and K contents and a decrease in the vacancies at the X site [90][91][92].
Detrital tourmalines from the Cretaceous to Eocene flysch basins of the Julian and Brkini flysch basins are rather similar in composition, being associated with metapelites and metapsammites coexisting or not coexisting with an Al-saturating phase, ferric-iron-rich quartz-tourmaline rocks and calc-silicate rocks; however, their concentration is drastically different.In fact, in the Julian basin, the number of tourmaline crystals (1-14 vs. 30-100) within a heavy mineral concentrate that is equal in percentage in the two basins is much smaller than in the BK, suggesting an important change in the provenance area of such detritus, similarly to the increase in non-igneous quartz recorded by [48] between the JB and BK basins.The presence of zonation in some grains suggests the possibility of changes in the fluid composition during their crystallization in the source rock, as well as in the diagenetic phase.

Figure 1 .
Figure 1.Simplified geological map modified after[34] indicating the main geological units and the main ophiolitic bodies.Flysch deposits of the Southeastern Alps and sample positions within the stratigraphic column.JB1 cannot be considered part of the stratigraphic succession of the Julian basin (see text for details), but it has been located at its correspondent age[35].No stratigraphic column is available for Brkini basin (see text for details).

Figure 1 .
Figure 1.Simplified geological map modified after[34] indicating the main geological units and the main ophiolitic bodies.Flysch deposits of the Southeastern Alps and sample positions within the stratigraphic column.JB1 cannot be considered part of the stratigraphic succession of the Julian basin (see text for details), but it has been located at its correspondent age[35].No stratigraphic column is available for Brkini basin (see text for details).

Figure 2 .
Figure 2. QFL diagram according to [40,41].Numbers in the figure correspond to samples analyzed in this study too.

Figure 2 .
Figure 2. QFL diagram according to [40,41].Numbers in the figure correspond to samples analyzed in this study too.

Minerals 2024 ,Figure 3 .
Figure 3. Percentage of heavy minerals in the studied samples according to their stratig tion.Labels as in Table2.
Figure 3. Percentage of heavy minerals in the studied samples according to their stratig tion.Labels as in Table2.

Figure 3 .
Figure 3. Percentage of heavy minerals in the studied samples according to their stratigraphic position.Labels as in Table2.

Figure 4 .
Figure 4. Raman spectra of the olivines found in the JB samples and fayalite and forsterite spectra [59] for comparison.

Figure 4 .
Figure 4. Raman spectra of the olivines found in the JB samples and fayalite and forsterite spectra [59] for comparison.

Figure 5 .
Figure 5.Some images of the tourmalines found in BKNV, BK41, JB25 (top to bottom).To the right, BSE imaging of the zoned tourmalines from JB17, JB24, BK41, BKNV (from the top left moving clockwise).

Figure 5 .
Figure 5.Some images of the tourmalines found in BKNV, BK41, JB25 (top to bottom).To the right, BSE imaging of the zoned tourmalines from JB17, JB24, BK41, BKNV (from the top left moving clockwise).

Figure 5 .
Figure 5.Some images of the tourmalines found in BKNV, BK41, JB25 (top to bottom).To the right, BSE imaging of the zoned tourmalines from JB17, JB24, BK41, BKNV (from the top left moving clockwise).

Minerals 2024 ,Figure 7 .
Figure 7. Frequency of the different generalized tourmaline species in the studied samples.H the samples of JB are grouped according to their ages.The group JB1-14 includes the samples r ing from Maastrichtian to Thanetian; the group JB17-25 is for Ypresian samples.

Figure 7 .
Figure 7.Frequency of the different generalized tourmaline species in the studied samples.Here, the samples of JB are grouped according to their ages.The group JB1-14 includes the samples ranging from Maastrichtian to Thanetian; the group JB17-25 is for Ypresian samples.

Figure 11 .
Figure 11.Ca versus X-site vacancy diagram of [84] used for the discrimination of tourmaline from low-Al schists, calcareous metasediment and those from a diagenetic environment.

Table 1 .
Percentage of heavy minerals and number of tourmalines and zoned tourmalines in the selected samples.Samples are in stratigraphic sequence (bottom to top is oldest to youngest).

Table 2 .
Number of mineral grains found in the studied samples.