DISTRIBUTION OF TRACE ELEMENTS IN ALTERED PYROCLASTITES FROM MONTE VULTURE VOLCANO ( SOUTHERN ITALY )

Three pyroclastic deposits from Monte Yulture volcanic area (Potenza, southern Italy) looking like paleosols in the field were investigated in a previous study for mineralogy and major elements to estimate the stage of the weathering. Here is dealt with the behaviour of sorne trace elements (Ce, La, Ba, Ni, Cr, Y, Rb, Sr, Y, Zr and Nb) in the same deposits to give a comprehensive geochemical picture. The distribution of the chemical elements within the whole rock and after its attack with Na-pyrosulfate (residue + solute) has been considered. Ba and Sr, as well as their distribution, appear to be controlled by the residual crystals in each of the deposits; La, Ce, Y and Nb are more concentrated in the solute that once was represented by vitric component, allophane, and Fe-Si-Al gels, biotite, carbonates and analcite; Y, Cr, and Ni show similar trends in whole rock and in solute. In particular La, Ce, Y, Y, Cr and Ni in the lowermost unit increase with the depth, as well as the contents of gels and allophane. Probably this behaviour was superimposed by the fluctuation of the water tables, as documented by the occurrence of a carbonate level upon the unit lies. It is concluded that the earliest stage of weathering did not affect the trace element distribution and that interpretations about chemical changes in deeply altered pyroclastic rocks should be always the outcome of careful accurate analyses.

Se concluye que los estados precoces de meteorización no afectan prácticamente a la distribución de los elementos traza, y que las interpretaciones sobre los cambios químico sufridos por las rocas piroclásticas deben hacerse sólo después de un detallado análisis.alkalis fram the vitric components and the increase of allophane and Si-Al gel contents in the higher part of the levels.Therefore, the eventual chemical modification in the content of major elements during weathering was masked.In the deposits studied by Fiore et al. (1992), although mineralogical transformation and mineral dissolution by weathering were lacking, eomponents characterised by high chemical reactivity (glass, allophane, Si-Al gels, phlogopite, analcite) were abundanl.This circumstance induced us to complete the investigation eonsidering also the behaviour of sorne trace element during the early stage of weathering in the same samples analysed for major elements and mineral contents.
Pyroclastic deposits consist of crystals, glass, lithic fragments having a highly variable density.The proportions of these components depend on the origin of magma, the mechanism of transport and emplacement, the topography of the area, etc. (e.g.Fischer and Schmincke, 1984).Such factors may cause granulometric, chemical and mineralogical vertical variations within a given pyroclastic deposil.So, geochemical and mineralogical variations shown by weathered and non-weathered pyroclastic rocks will reflect inherited characteristics and eventual modifications induced by alteration might be masked.These aspects were focused in a study by Fiore et al. (1992) on three buried pyraclastie flow units from Monte Vulture volcanic area (southern Ita1y).In this study was documented the existence of: (i) mineral alteration predating the eruptive events; (ii) persistence of grain-size distribution along the studied levels and low clay fraction content; (iii) alteration due to the cireulating water and (iv) incipient weathering effects.The authors attributed to weathering the hydration and the loss of  Baldassarre et al., 1989, simplified) and locality of the sampling.l.Pre-pleistocenic terrains.2. Ignimbrites A and B and phonolitic lava domes.3. Masseria Boccaglie Tephra.4. Rionero Barile Tephra. 5. Mt.Vulture-San Michele Tephra.6. Case López-Masseria Granata Tephra.7. Fluvio-Iacustrine, travertine and present day deposits.a. Crater and caldera rims.b.Faults.c. Cinder cones.Star: sampling site.

Geology andfield sampling
The Monte Vulture volcanic area (fig. 1) is located on the eastern margin of the Apennine chain, in Ba O   V oe Cr ..  southern Italy.It is a composite volcano, 1,327 m above sea level, lying on a sedimentary substratum mostly formed by fIyschoid sediments.According to La Volpe and Principe (1994), the volcanic complex was built up in the middle Pleistocene and its climax occurred between 740 ka and 590 ka.The volcanic sequence started with phonotrachytic ignimbrites (Fara d'Olivo Unit) and followed with: Masseria Boccaglie Unit, made up of phonolotic tephritic and tephritic tephra; Rionero-Barile Unit, consisting of thiny stratified pyroclastites; Monte Vulture-San Michele Tephra and Case Lopes-Masseria Granata Unit, made up of tephritic and phoiditic pyroclastites and lavas; Laghi di Monticchio Unit consisting of pyroclastic deposits having carbonatitic-melilititic composition.The pyroclastic levels described in this paper are from the Masseria Boccaglie unit.
The three paleoweathered pyroclastic units outcrop along a wall of the S. Antonio quarry (Rionero in Vulture, Potenza, Italy).The uppermost part of each level consists of brown and fine grained centri-metric material.The higher level (SUOV3) and the intermediate (SUOV2) level lay one upon another whereas the intermediate level is separated, by sorne pyroclastic fIows, from the lower one (fig.2).The stratigraphic interval considered is 3.2 m thick.

Experimental methods
The samples were subdivided into two grain-size fractions « 2,000 ¡.¡m and > 2,000 ¡.¡m), according to soil granulometric classification, which were subjected to a selective chemical dissolution with Na-pyrosulfate (Kiely and Jackson, 1965) to separate vitreous, amorphous or easily soluble minerals from those more resistant ones.X-ray powder diffraction analyses revealed that this chemical compound dissolved glass, phlogopite, analcite, calcite, allophane, Al-Si gels, Fe-Mn hydroxides and did not attack pyroxenes, feldspars, quartz, gamets, amphiboles, magnetite (hereafter called «residuum»), present in the studied materials.Quartz is sedimentary in ori~in and tom up from the flyschoid basement during the exploslve activity (the products of Monte Vulture are undersatured in Si0 2 ).The chemical composition of such dissolved material (hereafter called «soluble») was calculated subtracting the eomposition of the «residuum» from that of the untreated samples, taking into account the mass balance.Trace element contents were obtai-   (1972; 1975) and Leoni and Saitta (1976).The analytieal preeision, eheeked on international rack standards, was better than 5 % for Ni, Cr, V, Rb, Sr, Y, Zr, Nb and than 10 % for Ce, La and Ba.

Results
The results of the chemical analyses carried out on the untreated samples for both grain-size fractions are given in table 1 and graphicaHy represented in figure 3a.A variability is evident along a given pyroclastic level and among the levels.The data-points relative to SUOV1 and SUOV2 show clear trends; on the contrary SUOV3, formed only by few sublevels, doesn't exhibit tendency.
Ba, Sr, La, Ce, Y abundance increases towards the bottom of SUOV1 and SUOV2 levels and this tendency is decidedly more evident in SUOVl.The abundance of V, Cr, and Ni arranges in a similar way only in the SUOV 1 leve!.
As for the two grain-size fractions, sorne splitting of the values is noticed: Rb contents are distinctly lower in the < 2,000 /..lm fraction of aH the deposits; Zr and Nb are depleted in the < 2,000 /..lm fraction of SUOV2 whereas they are, or tend to be, enriched in the same fraction of SUOV 1; Ce and Y contents tend to be higher in the finer fraction.Interestingly, in the finer fraction Rb, Zr and Nb increase downward in SUOV2.
Again in figure 3a, we observe that sorne elements, such as Ba, Y, Ce, La, Sr, progressively decrease in their abundance upward fram the bottom (SUOVIR) of the lower level to the upper part (SUOV3a) of the higher leve!.
Figure 3b and figure 3c illustrate, respectively, the distributions of the elements determined in the «residuum» and those calculated for the «soluble» after the attack with Na-pyrosulfate.Numerical values are given in tables 1b-c.Ba and Sr in the «residuum», composed by px, Kf, pI, qtz and accessory phases, mimic the behaviour of the untreated samples both in a single level and in the entire sequence.Cr and Ni were enriched in the «residuum» of SUOV2 and did not vary in the SUOV1 and SUOV3.The other elements here considered are concentrated mostly in the «soluble» of the three pyraclastic levels.

Data analysis and conclusions
As discussed in Fiore et al. (1992), the vertical variations along the examined pyroclastic levels Factor analysis shows that three factors account for fraction (s), > 2,000 Jlm fraction (g) and of basallevel (R).
88 % of the total variance (table 3) and that: (i) the Data from Fiore et al. (1992) first factor (FI) is controlled by Y, V, Zr, Cr, Ce, Ni Furthermore, it is relevant that the deepest level (SUOVI) is characterised by the highest positive Px = pyroxene; Kf = feldspar; PI = plagioclase; Qz = quartz; value for F3 that, in turo, gradually decreases going PhI = phlogopite; Anl = analcite; Cal:ealcite; tr.= trace; upwards in the sequence.In addition, the ability of L. =summation.
depend on: (i) syndepositional mineralogical variations along a level and among the diverse levels; (ii) effect of weathering in a given level before the next volcanic products was deposited; (iii) influence of magmatic fluids or of the water table after the deposition of the sequence.
In the study case, the increase of sorne elements from the upper sublevel to the lower one gives the idea that the alteration effects are induced by water freely moving downward through the sequence.Indeed, the cause of such a variation cannot be ascribed to water alteration.
To test the influence of minerals on abundance and distribution of trace elements in the pyroclastic flow deposits under study, R-mode factor analysis with VARIMAX rotation was performed between trace elements and mineral contents of the «residuum» (table 2; data fram Fiore et al., 1992).Obviously, the factors will join to the minerals considered in the statistical analysis also the elements entering the crystallattice of coexisting accessories.   1 Fl to discriminate the grain-size fractions suggests the coexistence of two pyraxenes, different in chemical composition, within the deposits.This interpretation is supported by literature data (e.g.De Fino et al., 1986;Caggianelli et al., 1990) indicatíng the presence of chemically different phenocrysts of pyroxenes in the lowermost Monte Vulture volcanics, just under the deposits here studied.The mineralogical variations are not confined to a given sublevel.A chemical variation related to the stratigraphic position of the sublevels within the pyraclastic sequence is evident.Simple linear correlation between depth and abundance of Ce, La, Ba, Sr and Y indicate, with a level of significance ex < 0.001, that the elemental decrease within the three levels is not casual.This variation, because of its regularity, seems to suggest sorne primary features of the deposit.
To point out any relation between the behaviour of trace elements and the products of alteration, another factor analysis was performed.This analysis was necessaríly restricted to the < 2,000 11m fraction in which the weathering products are concentrated.A first analysis (results not given) showed that both calcite and phlogopite don't give any contribution to the variance; therefore, to constrain the system, these two variables were excluded and the statistical analysis recalculated.The results (table 4) show that the four factors account for 89 % of the total variance and that there are no relationships between chemical elements and gels and allophane.Among the four factors, Fl calls for attention because of the negative weight of Rb and glass.This feature might be explained in two ways; (i) Rb contents reflects original feature and this element may be concentrated in the dissolved minerals amI/or in glass or (ií) Rb is the only trace element mobilised fram glass during the earliest stage of weathering.The first hypothesis is suggested by the similarity of the Rb contents in the products here studied and those of the trachytic-phonolitic volcanics of the Monte Vulture (De Fino et al., 1986 and references therein).The second hypothesis is also possible, as suggested by the linear correlation between H20 and Rb in the whole rack (r = 0.775; ex < 0.001).The geochemical affinity of Rb for K, which was mobilised fram glass (Fiore et al., 1992), supports this interpretatíon.The available data, however, do not remove the ambiguity.
The second factor (F2) refers to the elements that were associated with the pyroxene in the previous statistical analysis; F3 clusters Nb, Cr, Ba, Ni, Zr and H20+; F4 is determined by Sr and, with a lower statistical weight, by Kf.The available data do not permit to establish the causes that determined the grauping of the F3.Nevertheless, the con- tribution of the primary chemical signature in the pyroclastics here studied seems to be of importance.Indeed, in the SUOVl Rb, as well as La, Ce, Y, and Nb are more concentrated in the amorphous fraction that consists of vitric components and its derivative (allophane, Si-Al gels).Instead, in the SUOV2, Rb is concentrated in the minerals of the «residuum» that are more abundant than in SUOVl and in SUOV3.On the other hand the overall values of Rb contents are quite similar to those of chemically comparable products of Monte Vulture (De Fino et  al., 1982).
The data obtained clearly imply that the minerals of the «residuum», and sometimes the glass, playa dominant role in controlling the distribution of the trace elements both in distinct levels and in the two grain-size fractions of the same level.Therefore, it appears that, not withstanding the important loss of sodium from glass (Fiore et al., 1992), the distribution of the trace elements in the pyroclastic sediments, affected by an early stage of weathering, essentially reflects inherited characteristics.This conclusion is apparently in disagreement with the results of Price et al. (1991) who point out and enrichment in Ba-content for basalts affected by an «early stage of weathering».The rocks here studied are decidedly less altered than those studied by Price et al. (1991), who reported mineralogical transformation (e.g.olivine ~«iddingsite»), indicating a more advanced stage of alteration.The problem them becomes of what to mean as «early stage».In this respect, it is suggested that the first stage of weathering should be considered as the one characterised by partial dissolution of glass (if present), local dissolution of primary minerals and without their transformation, precipitation of alumina and silica gels, possible neoformation of protominerals (e.g., allophane, imogolite).
The data presented in this paper show that the earliest stage of alteration of the pyroclastic deposits examined did not substantially influence the distribution of Ba, Ce, La, Ni, Cr, V, Rb, Sr, Y, Zr, Nb.This absence of variation of trace element abundance coexists with the presence of neoformed phases (allophane, gels) originated by weathering.This implies that neoformation occurred with very small modifications in trace element contents, undetectable by this approach.The gradual variations of the abundance of the trace elements here determined reflects original characteristics of the pyroclastic rocks.A direct consequence of the geochemical variability evidenced in a few meters thick sequence is that interpretation of behaviour of elements during intense weathering of pyroclastic flow deposits should be done with a particular attention, because original chemical variation and modification induced by weathering might overlaps.

Fig. 2 .
Fig.2.-Schematic stratigraphic section of the sequence collected at San Antonio quarry (after LaVolpe and Principe, 1989,  modified)  and position of the samples in study.

Fig. 4 .
Fig. 4.-Plots of Fl and F2 versus F3 showing the discrimination induced by the F3-factor.Field-line is arbitrary.Symbols as in figure 3.

Table 3 .
-R-mode factor analysis between minerals of the «residuum» and trace elements