Introduction

Integrated microstructural and microprobe investigations are essential tools to obtain information about progressive foliation development, porphyroblast growth-timing and chemical zoning patterns, in turn necessary to constrain through geothermobarometry the different stages of pressure-temperature-deformation (P-T-d) evolution of metamorphic rocks.

Significant advances in P-T path reconstructions have been made with the development of the techniques of thermodynamic modeling of representative samples via computation of P-T pseudosections (e.g. Holland and Powell 1998; White et al. 2000; Connolly and Petrini 2002; Connolly 2005, 2009). Nevertheless, high reliability of achieved constraints may be ensured only if the thermodynamically derived data closely match the actual structural and minerochemical features of studied rocks, with particular reference to the correct sequence of mineral equilibria (e.g. Zeh 2001; Gaidies et al. 2006; Cirrincione et al. 2008; Angì et al. 2010; Groppo and Castelli 2010).

For this reason, the most detailed P-T paths have been usually obtained on medium-grade porphyroblastic metapelites, known to be particularly well suited to reconstruct reliable sequences of blasto-deformational events. Indeed, lack of porphyroblasts in lower grade metamorphics often prevents to obtain meaningful blasto-deformational relationships while, in high-grade rocks, both earlier foliations and paragenetic equilibria are frequently completely obliterated and/or chemically re-homogenized by the latest pervasive metamorphic stages.

In this paper thermodynamic modeling of a medium-grade metapelite has been carried out by integrating thermobarometric, microstructural, mineral association and minerochemical data. Besides, in an attempt to minimize the risk of personal bias in the determination of reliable mineral equilibria, a new image processing technique developed by means of multivariate image statistical data handling via GIS based tools, has been applied to X-ray maps of representative micro-domains, to help identifying effective equilibrium assemblages and reaction volumes. Computation of P-T pseudosections before and after fractionation effects constrained by means of this technique proved to be very useful to quantify the composition and relative percentage of effective equilibrium assemblages.

In more detail, a sample of porphyroblastic ilmenite-garnet bearing schist cropping out locally in the Mandanici Unit of the Peloritani Mountains has been selected for the purpose of the present study. Indeed, no detailed P-T path has been up to now elaborated for the basement rocks of this segment of the original southern European Hercynian belt, likely also because of the absence of suitable parageneses in the low- and very-low-grade metamorphic rocks of the Peloritani Mountains (Cirrincione et al. 2012) and of the pervasive late-Hercynian thermal re-equilibration in the medium- to high-grade metamorphic rocks (Appel et al. 2011). In particular, the selected metapelite sample is characterized by i) well-preserved porphyroblast-matrix relationships, ii) fine grained matrix able to record a multistage deformational sequence and, iii) apparent absence of compositional porphyroblasts re-equilibration by diffusion processes.

This sample was studied integrating microstructural and microprobe investigations with thermodynamic modelling (Connolly and Petrini 2002; Cirrincione et al. 2008 and reference therein) assisted by image processing of X-ray maps, with the aim of reliably estimating the sequence of P-T conditions attained during its orogenic cycle. This approach proved to be extremely useful and provided precious novel information about the tectono-metamorphic evolution of a segment of the Hercynian belt exposed in the Peloritani Mountains.

Geological setting

Peloritani Mountains (NE Sicily) are the southernmost part of the Calabria-Peloritani Orogen (CPO), an Alpine nappe-pile edifice located between the NW-SE-trending Southern Apennine and EW-trending Siculo-Maghrebide chains (Fig. 1a). They constitute a very complex continental crust sector (Ferla 2000), structured in a collisional belt and subdivided into the Lower Complex, outcropping in the southern part of the Peloritani Mountains, and the Upper Complex, exposed in the northern part of the chain (Cirrincione et al. 1999, 2012) (Fig. 1b).

Fig. 1
figure 1

Regional geological framework of the study area: a) Southern Italy with location of Calabrian Peloritani Orogen; b) Geological sketch map of the Peloritani Mountains and Aspromonte Massif crystalline basement complexes with the subdivision of the tectono-metamorphic units (from Cirrincione et al., 2012, modified); c) Geological sketch map of the study area with sample location (from Lentini et al., 2000, modified)

Full size image

The Lower Complex comprises three tectonic units composed by very-low to low-grade Hercynian metamorphic rocks (below the chlorite-isograd, T ≤ 350 °C, P ≥ 2 Kb; Atzori and Ferla 1992) and sedimentary covers, both unaffected by Alpine metamorphic re-equilibration.

The Upper Complex comprises the most internal nappes of the Peloritani Mountains, and it is constituted by the low- to medium-grade Mandanici Unit (MU) and the overlying medium-high grade Aspromonte-Peloritani Unit (APU), which crops out also in the Aspromonte Massif (southern Calabria). The basement rocks of the latter unit are mainly composed of paragneisses and augen gneisses and minor micaschists, amphibolites and marbles, produced by a polyphase HT-LP retrograde Hercynian metamorphism (Pezzino et al. 1990; Puglisi and Pezzino 1994) and later intruded by late Hercynian granitoids (Fiannacca et al. 2008; and reference therein).

The MU consists of relatively low pressure, low to medium temperature rocks with paragenetic equilibria consistent with upper greenschist to lower amphibolite-facies conditions (Atzori and Sassi 1973). Very few, and mostly only qualitative, data are available about the P-T metamorphic conditions of the MU rocks. Atzori and Ferla (1992; and reference therein) describe the MU as a unit mainly composed of epizonal to mesozonal metamorphic rocks and report similar P-T clockwise evolution of the metamorphic rocks of MU and APU, with thermal and baric peaks achieved at T < 620 °C and P < 5.5 Kbar, respectively, for the highest grade rocks, as deduced from mineral assemblages and conventional geothermobarometry. Phyllites are the dominant MU lithotypes, while quarzites, marbles, metabasites and micaschists are subordinate (Cirrincione and Pezzino 1991; Ferla 2000). The metamorphic grade increases towards the present-day upper levels of the unit (Ferla 1972), suggesting an overturned setting, further confirmed by the presence of an anchimetamorphic Mesozoic sedimentary cover (i.e. Alì Series) at the bottom of the present-day structural framework, interpreted as the original cover of the crystalline basement involved in the Alpine metamorphic cycle (Cirrincione and Pezzino 1991; Cirrincione et al. 2012).

Although an Alpine subgreenschist- to greenschist-facies metamorphic overprint at relatively high pressure has been locally documented in the Mandanici and Aspromonte-Peloritani Units (Atzori et al. 1994; Pezzino et al. 2008 and references therein), the Hercynian orogeny has long been considered responsible for the main metamorphic events in the Peloritani Mountains (Atzori et al. 1994; De Gregorio et al. 2003; Appel et al. 2011). Dubois and Truillet (1971) suggested that the Mandanici Unit was the sedimentary cover of a pre-Hercynian basament metamorphosed during the Hercynian orogeny, whereas Ferla (2000; and reference therein) proposed a pre-Hercynian origin of the whole Upper Complex basement. According to the latter author the basement of the Upper Complex would have been pervasively reworked by the Hercynian metamorphism together with its Paleozoic sedimentary cover, presently represented by the basement units of the Lower Complex. Recent SHRIMP U-Pb zircon studies carried out on paragneiss and augen gneiss samples of the Aspromonte-Peloritani Unit (Williams et al. 2012; Fiannacca et al., 2012) have demonstrated that large portions of the unit formed during the latest stages of the Avalonian-Cadomian orogeny (~550 Ma). As far as the Mandanici Unit is concerned, most authors still consider it as composed of Early Paleozoic volcano-sedimentary sequences turned into low-grade metamorphic rocks during the Hercynian orogeny (e.g., Russo et al. 2006; Ferla and Meli 2007).

Petrographic and microstructural features

Three thin sections of an ilmenite-garnet schist sample cropping out in the Vernà Mountain area (Fig. 1c) have been examined; one of them cut parallel and the other two perpendicular to the mesoscopic lineation, given by microfold axial culmination, in order to ensure identification of all the structural features produced by the succession of deformational stages, as well as to provide precise information on the 3D distribution of the fabric and on the modal abundances of the mineral constituents.

The selected schist is characterized by a dominant lepidoblastic microstructure (Fig. 2a), with poikilitic garnet and ilmenite porphyroblasts (Fig. 2b–d) wrapped by alternating muscovite+chlorite and minor quartz+plagioclase layers, with apatite, monazite and zircon as accessory phases. Three schistosity surfaces were recognized: S1, recognizable only at the thin section scale, consists of an isoclinal folding schistosity, largely obliterated and only preserved as microfold hinge relics within the granoblastic domains between S2 layers (Fig. 2a–e); the second one (S2), which represents the main foliation to be observed in the field, is a penetrative and pervasive crenulation schistosity generated by a non-coaxial deformational event (D2). It also produced rotated porphyroblasts (Fig. 2b); the last one (S3), clearly recognizable only in the porphyroclast pressure shadows as well as between fragments of pinch-and-swell garnet porphyroclasts (Fig. 2f), was generated by a retrograde non-coaxial shearing event (D3) also responsible for the formation of mica-fish.

Fig. 2
figure 2

Petrographical features: a) general microstructure given by a fine-grained matrix depicting a dominant lepidoblastic texture (parallel polars); b) poikiloblastic garnet with folded inclusion trails surrounded by dominant muscovite+chlorite lepidoblastic layers and subordinate quartz+plagioclase granoblastic layers (crossed polars); c) evidence of late- to post-kinematic garnet overgrowth (crossed polars); d) ilmenite porphyroblasts mechanically re-oriented along the crenulation schistosity (parallel polars); e) symmetric microfold hinge relics preserved within the granoblastic domains (crossed polars); f) pinch-and-swell garnet porphyroblasts (crossed polars)

Full size image

Rock-forming minerals, in decreasing order of abundance (Fig. 3), are:

Fig. 3
figure 3

Histogram of the point counting analysis (1,000 points per thin section)

Full size image

White mica (36.5 vol.%) characterized by three different crystallization stages. The first one is only recognizable by the presence of structural relics in the microfold hinges. The second generation comprises aggregates of small flakes parallel to the main schistosity S2, while the third generation crystallizes in the pressure shadows of former porphyroblasts during D3 shearing stage.

Quartz (35.6 vol.%) constitutes equidimensional blasts, with undulose extinction and S2-parallel elongation. It is concentrated as granoblastic layers in the fold hinges, as large crystals around garnets in the pressure-shadows, and as σ-shaped inclusion trails in garnet cores.

Chlorite (20.1 vol.%) is localized along the S1 relic surfaces, in the S2 planes in association with white mica and in garnet pressure-shadows. Total to partial polycrystalline chlorite pseudomorphs after garnet are locally observed.

Plagioclase (4.8 vol.%) occurs both in the matrix and as minor porphyroblasts mostly elongated parallel to S2.

Elongated, or roundish-shaped, poikiloblastic garnet (1.7 vol.%) exhibits quartz and rarer ilmenite inclusion trails, well preserved in garnet core, defining a σ-shaped internal schistosity (S2) (Fig. 2b, c). This observation suggests that garnet crystallized during the second metamorphic event (M2), incorporating quartz and ilmenite that produced an internal foliation continuous with the external schistosity represented by the S2 surface (Fig. 2b). Syn-kinematic garnet was locally overgrown by unfolded rims, masking the original crenulated garnet morphology (Fig. 2c). The D3 shearing event induced porphyroblast rotation, which evolved locally to the formation of mica-fish and stretched and pinch-and-swell garnet porphyroclasts suggesting extensional shearing conditions; growth of retrograde chlorite and muscovite occurred during this stage (Fig. 2f).

Ilmenite accounts for 1.4 vol.% in the sample: euhedral crystals are parallel to S1 or mechanically re-oriented throughout S2. It shows occasionally reaction rims and recurrent leucoxene alteration interpreted as retrograde reaction (e.g. Raase et al. 1986) (Fig. 2d).

Accessory minerals are represented by apatite, usually in association with monazite or included in garnet, monazite in foliation-aligned mostly anhedral grains, and zircon, mostly occurring as partly corroded grains.

Microstructural analysis permitted to reconstruct the metamorphic-deformational relationships, highlighting a polyphase deformation evolution. The attention was focused especially on garnet porphyroblastic domains useful for interpreting the sequence of the tectono-metamorphic stages. During M1 the stable mineralogical assemblage, partly preserved in microlithons, comprised quartz, white mica I generation, ilmenite plus, probably, chlorite I and plagioclase I generations. The synkinematic mineral phases with respect to M2, representing the dominant event recorded in the rock, are quartz, white mica II, chlorite II, plagioclase II, ilmenite reaction rims and rotated garnet porphyroblasts, overgrown by inclusion-free rims. Although M3 was a dominantly mechanical event, blastesis of quartz, chlorite III, and white mica III can be locally observed.

Whole-rock chemistry

Whole-rock composition was obtained through XRF analysis on pressed powder pellets, performed on a Philips PW 2404 spectrometer equipped with a Rh anticathode at the Department of Biological, Geological and Environmental Sciences, Catania University, corrected for matrix effects (Franzini et al. 1972); FeO content was determined by titration with potassium permanganate, H2O was measured as weight loss on ignition (L.O.I.) and was determined with standard gravimetric procedures, after heating the powder at ~900 °C for about 6 h. Whole-rock chemistry was used to derive information about the rock protolith, assuming a dominantly isochemical metamorphism and to obtain the chemical composition of the equilibrium volume useful to calculate P-T pseudosections.

The studied sample plots as shale in classification diagrams (e.g. De La Roche 1978; Herron 1988) and within the field representative of Mandanici Unit phyllites compositions. Normalization of the chemical composition to PAAS (Post-Archean Australian Shale; Taylor and McLennan 1985) also shows a composition similar to average shales, except for lower CaO contents (0.38 wt.% in the studied sample compared to 1.3 wt.% in PAAS), possibly because of a low concentration of CaCO3, as sediment component or as authigenic product, in the original basin.

Mineral-chemistry and X-ray map image elaboration

Analytical procedure

Analyses and observations at micrometer scale were performed in three distinct domains hosting garnet porphyroblasts which were chosen to acquire Si-Ti-Al-Fe-Mg-Mn-Ca-Na-K X-ray maps useful to highlight compositional variations associable to specific metamorphic reactions.

Spot chemical data were then used to compute mineralogical formulae by means of Minpet 2.02 software (Richard 1995) on the basis of 6 oxygens for ilmenite, 8 oxygens for feldspars, 22 oxygens for white mica and 36 oxygens for chlorite. Garnet formulae were calculated on the basis of 12 oxygens and 8 cations, considering all iron as FeO, in order to allow comparison of the obtained spot chemical analyses with the garnet compositions modeled through the available solid solution model used for P-T pseudosection computation, that does not include the andradite end-member (Powell and Holland 2008). Used mineral abbreviations are after Kretz (1983), reviewed by Siivola and Schmid (2007) (see Appendix A for device characteristics and representative analyses).

Following the methodological approach of Launeau et al. (1994), X-ray maps of the three selected domains were elaborated to obtain high contrast pseudocolor images by means of multichannel classification procedure, here implemented with the image processing tools contained in the open source software with public license GRASS in its last release (i.e. 6.4.1). This operative procedure has the purpose of extracting quantitative information from classified images by means of a stepwise-controlled procedure, as reported in the flowchart of Fig. 4 (see Appendix B for further details).

Fig. 4
figure 4

Flowchart of the stepwise procedure used for image processing of X-ray maps via GIS based tools (GRASS 6.4.1)

Full size image

During the first step, obtained pseudocolor maps were elaborated in order to: i) identify mineral phases up to few micrometers in size, such as inclusion trail phases and very fine-grained retrograde aggregates (e.g. leucoxene after ilmenite), ii) define modal abundances of all rock-forming minerals constituents, iii) better define the textural relationships between the mineral constituents. In the second step, image processing was performed exclusively on the garnet crystals of the three sites with the purpose of detecting mineral zoning developed during growth at variable P-T conditions.

Results

Garnet shows a prevalent almandine composition, in the range Alm64Sps17Grs14Prp5 (inner-core) and Alm69Sps15Grs11Prp5 (outer-core) to Alm75Sps13Grs6Prp6 (rim), with higher almandine and slightly lower spessartine towards rims (Fig. 5), suggesting weak temperature increase during crystal growth.

Fig. 5
figure 5

Minerochemical features of the major mineral phases: a) composition profile and chemical compositions of garnet; b) ternary classification diagram for Ti-bearing phases; c) distribution of phengite content in white mica; d) classification diagram for chlorite; (e) ternary feldspar classification diagram

Full size image

Ilmenite is characterized by a limited solid solution towards pyrophanite end-member with Mn content (Mn/(Mn+Fe+Mg)) ranging from 0.050 to 0.075 (Fig. 5). Analyses carried out on leucoxene, formed as a widespread retrograde phase after ilmenite, give almost pure rutile composition (Fig. 5).

White mica is a muscovite with moderate phengite content, with Si a.p.f.u. ranging from 3.08 to 3.28 (Fig. 5).

Chlorite compositions plot in the ripidolite field of the (Fe2++Fe3+) vs Si diagram with a Fe/(Fe+Mg) ratio ranging from 0.64 to 0.67 (Fig. 5).

Pure albite forms the bulk of the plagioclase grains; the rare porphyroblasts show an oligoclase rim, with average composition An = 17.2 % (Fig. 5).

BSE imaging and X-ray map analysis of the three selected domains (Fig. 6) confirm that garnet porphyroblasts started to grow during M2 event, responsible for the formation of the S2 surface, continuing to grow up to the later stages of this metamorphic event. Mn and Fe X-ray maps qualitatively confirm chemical variation during garnet growth as well as the appreciable compositional shift of ilmenite toward pyrophanite end-member. Ca X-ray maps illustrate the distribution of fluoroapatite, while the Mg ones well depict microfolded chlorite domains. K and Al X-ray maps reveal white mica distribution, while Na X-ray maps emphasize the location of albite crystals.

Fig. 6
figure 6

Back scattered electron (BSE) images of the three selected garnet porphyroblastic domains, with location of spot analyses, and related elemental X-ray maps

Full size image

Following the proposed procedure for the image processing of X-ray maps of the selected micro-domains, multivariate statistical analysis via principal component computation (Launeau et al. 1994; Carr 1998) was first used to minimize the redundancy of data channels, permitting to better separate the contribution of the constituent mineral phases. After this step, maximum likelihood classification of all the principal components by means of supervised classification procedure (Flesche et al. 2000; Friel and Lyman 2006) allowed to pinpoint the modal percentage of the single phases in each analyzed domain, as well as to better define porphyroblast-matrix relationships in the studied rock (Fig. 7) (see Appendix B for further details).

Fig. 7
figure 7

Maximum likelihood classification obtained via principal component analysis of X-ray maps of the three selected garnet porphyroblastic domains with bar diagram of the computed volume percentage of detected mineral phases. Each image is composed of 512 columns×400 rows for an average pixel dimension of 2μm

Full size image

Furhermore this procedure helped to confirm that inclusions of quartz, fluoroapatite and ilmenite depict an internal foliation within garnet porphyroblasts consistent with a syn-kinematic crystallization of these three mineral phases, in equilibrium with external albite. Chlorite is often in decussate aggregates mimicking the original S1 surface. This may suggest former presence of prograde biotite, now entirely replaced by chlorite. White mica is instead preferentially located in the garnet pressure shadows crystallizing along the incipient formation of the S3 surface.

Extraction of a new set of X-ray maps of the sole garnet via map algebra operation obtained with GRASS, allowed to obtain detailed growth zoning patterns (Fig. 8) highlighting the existence of a progressive chemical change during continuous garnet growth with changing of P-T conditions. This new visualization technique further permitted to emphasize the relative amounts of reactant garnet during the retrograde metamorphic evolutionary stages, and therefore to quantify the potential percentage of fractionated garnet after the attainment of the peak metamorphic stage (Appendix B).

Fig. 8
figure 8

Maximum likelihood classification obtained via principal component analysis of filtered garnet X-ray maps with bar diagram of volume percentage of fractionated and reactant garnet

Full size image

Reconstruction of the P-T evolution

Methodological approach

The P-T trajectory of the selected upper greenschist facies ilmenite-garnet-bearing metapelite of the Mandanici Unit has been reconstructed by calculation of P-T pseudosections, that represent total phase diagrams sections for specific bulk-rock compositions (Will 1988). They are based on the composition of the “equilibrium volume”: volume that contains those minerals that are interpreted to have once been in equilibrium with each other (i.e. paragenesis) (Powell and Holland 2008). According to Stüwe (1997), mineral parageneses can be directly derived using the bulk rock chemistry, or an effective bulk rock composition obtained excluding from the calculation those inner portions of porphyroblastic minerals affected by chemical fractionation from the syn-reacting mineral assemblage (Stüwe 1997; Evans 2004). The use of fixed bulk compositions in the pseudosection calculations implies indeed that the chemical systems in question are considered in overall equilibrium. This condition is fulfilled as long as no chemical fractionation occurs, as a consequence of the multi-stage growth of some porphyroblastic minerals, since that results in the activation of an effective bulk composition (Stüwe 1997) operating during metamorphic evolution. Occurrence of mineral fractionation can be assessed through inconsistency between calculated and observed mineral equilibria, as for instance evidenced by the lack of intersection at least of three garnet compositional isopleths (e.g., almandine, grossular, and spessartine) within a relatively small area in the pseudosection P-T space (Evans 2004).

In these cases, the volumes of fractionated phases can be estimated, by means of quantitative compositional maps analysis, capable to constrain the real equilibrium volume and then the correct chemical analysis to be used for inferring the P-T constraints.

Taking into account the above mentioned methodology limits (e.g. Hetherington and Le Bayon 2005; Le Bayon et al. 2006; Caddick and Thompson 2008; Cirrincione et al. 2008; Angì et al. 2010; Groppo and Castelli 2010), pseudosections allow P-T path segments to be determined, by interpreting mineral proportions and compositional evolving trends. Perple_X is a software package that permits pseudosections to be calculated, on the basis of free energy minimization approach (Connolly 1990; Connolly and Petrini 2002). P-T conditions estimate depends on field position corresponding to observed assemblage in P-T space and on isopleths (i.e. chemical iso-compositional curves) intersections. The obtained constraints may be considered reliable only if a good match between observed and calculated mineralogical assemblages is verified (e.g. Connolly and Petrini 2002; Cirrincione et al. 2008).

P-T pseudosections

A first P-T pseudosection for the selected sample was calculated using the July 2011 upgraded version of Perple_X (Connolly http://www.perplex.ethz.ch/) in the range T = 400–550 °C and P = 0.20–1.00 GPa, considering: 1) a closed-system behavior in the model chemical system TiO2–MnO-Na2O–CaO-K2O–FeO–MgO-Al2O3-SiO2-H2O (TiMnNCKFMASH), suitable to exemplify the real system, which contains Ti in widespread ilmenite and Mn in garnet and ilmenite. 2) SiO2 in excess, as well as H2O which also represents the only fluid phase, in agreement with the observed blastesis of hydrous mineral phases and the lack of carbonate ones; 3) the internally consistent thermodynamic database of Holland and Powell (1998, their upgrade 2002; 4) the Compensated Redlich Kwong (CORK) fluid equation of state of Holland and Powell (1991; 1998); 5) the XRF composition as the effective bulk rock chemistry; 6) the following solid solution models: i) the Ti-bearing biotite of White et al. (2007) with DQF corrections to annite; ii) a quaternary garnet model (Holland and Powell 1998) with a restricted subdivision range on Mn 0 %<X<20 %; iii) the chlorite model of Holland et al. (1998); iv) the white mica model of Coggon and Holland (2002) revised by Auzanneau et al. (2010); Ti-phengite was excluded from the calculation because not considered a reliable substitution at the relatively low pressure conditions of the given system; v) a binary plagioclase solid solution (Newton et al. 1980); vi) the Fe-Mg-Mn staurolite after Holland and Powell (1998); vii) an ideal Mg-Fe-Mn chloritoid model with restricted range of X(Mn) up to 20 % (White et al. 2000); viii) an ideal ilmenite-geikielite-pyrophanite solid solution; see the Perple_X web site for further details (http://www.perplex.ethz.ch/PerpleX_solution_model_glossary.html )

The first calculated pseudosection reveals that almandine increment is mainly controlled by increasing temperature, spessartine component increases with decreasing pressure and temperature, grossular component increases with decreasing temperature and increasing pressure. The pyrophanite component increase is mainly related to decreasing temperature, contrary to the pure ilmenite component. Finally, the increase of phengite content in white mica is directly related to pressure increase while Fe/(Fe+Mg) ratio in chlorite decreases with temperature increase (Fig. 9).

Fig. 9
figure 9

P-T pseudosection calculated in the TiMnNCKFMASH model system using equilibrium rock composition obtained via XRF bulk-rock analysis

Full size image

The modeled isopleths closer to garnet inner core compositions (Alm64Sps17Grs14Prp5) generate intersections in a small region, characterized by T = 515 ± 5 °C and P = 0.93 ± 0.03 GPa. This area also comprises pyrophanite isopleths (Mn/(Mn+Mg+Fe) = 0.05 to 0.75), related to chemical re-equilibration of ilmenite porphyroblasts during syn-M2 garnet core growth. Outer core garnet isopleth intersections (Alm69Sps15Grs11Prp5) provided comparable P-T estimates of T~525 °C and P~0.93 GPa, connected to a later phase of the same metamorphic event. The small area depicted by garnet rim isopleths intersections (Alm75Sps13Grs6Prp6; Fig. 9) indicates that the final garnet growth stages took place in equilibrium with the bulk rock composition and provides further P-T constraints (T~540 and P~0.803 GPa), highlighting a prograde evolution, during which crenulation schistosity obliterated most of the previous foliation. All isopleth intersections fall into the Bt+Chl+Wmca+Ilm+Grt pseudosection P-T field, in agreement with the observed equilibrium mineralogical assemblage, except for biotite, which was likely replaced by chlorite during retrogression. Obtained P-T estimates indicate greenschist-amphibolite facies boundary conditions. Retrograde evolution produced blastesis of oligoclase in textural equilibrium with chlorite, formed at expense of biotite and/or garnet as pseudomorphs and reaction rims, leucoxene replacing ilmenite and white mica III with phengite intermediate content. This assemblage did not provide any useful intersection in the pseudosection P-T space (Fig. 9) highlighting the fact that the equilibrium volume represented by this mineral paragenesis resulted in disequilibrium with the used XRF bulk rock chemistry.

In this view, according to the procedure developed by Zuluaga et al. (2005), the molar amount of oxides fractionation responsible for the effective bulk rock chemistry shift was calculated, considering garnet and ilmenite as the only mineral phases involved in the fractionation process. The amount of ilmenite fractionation was computed considering the relative percentage of unaltered ilmenite (0.61 % vol; Appendix B) and retrograde leucoxene (0.72 % vol; Appendix B) quantified by means of the image classification of the three selected porphyroblastic domains (Fig. 7). This implies that about 46 % of Ti-bearing phases was likely subtracted from the retrograde reacting assemblage. Since the total amount of these phases accounts for the 1.4 % average volume in the whole sample (Fig. 3), it follows that the total volume of fractionated ilmenite is equal to 0.64 %.

In order to obtain the volume of fractionating garnet, a second set of X-ray maps, portraying only the three garnet porphyroblasts, was elaborated by means of multivariate statistical analysis (Fig. 8) allowing to detect and quantify the intra-garnet chemical variations not identifiable throughout the first image processing step. Taking into account that the total modal percentage of garnet is of 1.7 % (Fig. 3), this value was corrected subtracting the calculated relative percentage of fractionated garnet not involved in the retrograde reactions, estimated by means of the image processing procedure (light-gray regions in Fig. 8). Being the relative average percentage of retrograde reactant garnet of about 46 %, this implies that the fractionated garnet accounts for a total modal abundance of about 0.9 %.

The chemical fractionation effects to be subtracted from the XRF bulk rock chemistry (Table 1) were therefore calculated, taking into account the contributions of ilmenite and garnet by using two different density correction factors, both calculated at the same P-T values estimated for peak metamorphic conditions. The density ratio between ilmenite and whole rock allowed to extract a correction factor of 1.65, providing the reactant retrograde bulk rock chemistry due to ilmenite fractionation effects (Table 1). The second density ratio correction factor, between garnet and whole rock, allowed to extract a correction factor of 1.45, which permitted to add also the contribution due to garnet fractionation, providing a realistic molar correction of the reactant retrograde bulk rock chemistry (Table 1). The corrected composition shows that MnO disappears, TiO2 content is significantly reduced in the new effective bulk rock chemistry, while some decrease in Al2O3 and FeO as well as an increase in SiO2 and MgO can be observed.

Table 1 Calculation of amount of oxides to be fractionated from ilmenite+garnet porphyroblasts
Full size table

The newly obtained equilibrium chemical system TiO2-Na2O-CaO-K2O-FeO-MgO-Al2O3-SiO2-H2O (TiNCKFMASH) is considered to represent the observed retrograde chemical equlibria. Also in this case SiO2 and H2O were considered to be in excess and the same fluid equation of state and the internally consistent thermodynamic database of Holland and Powell (1991; 1998), as well as the already mentioned solid solution models were used for calculation.

The pseudosection, calculated taking into account the fractionation effects, revealed a simpler topology. Chemical isopleths in this pseudosection P-T space permitted to ascertain a significant increment of the plagioclase anorthite content, mainly controlled by decompressional effects, and an increment of white mica phengite content with pressure increase (Fig. 10). The observed compositional ranges of anorthite (An15-19) and phengite (Phg12-18), consistent with the retrograde compositional equilibrium, also characterized by the presence of retrograde rutile (i.e leucoxene) and chlorite, permitted to constrain a relatively large pseudosection P-T space with temperature ranging between 420 and 460 °C and pressure of 0.30–0.60 GPa, suggesting a low greenschist facies retrograde stage. The computed assemblage (i.e. Bt+Chl+Wmca+Pl+Ab+Rt) fits well the observed paragenetic equilibria except for the presence of Bt, not observed in our sample.

Fig. 10
figure 10

P-T pseudosection calculated in the TiNCKFMASH model system using equilibrium rock composition obtained via fractionation method (Zuluaga et al. 2005; modified) assisted by image processing results as illustrated in Figs. 7 and 8 (see text for explanation)

Full size image

Discussion and conclusion

Thermodynamic modeling of metapelitic systems has been affirmed as a routinely performed method for unraveling the P-T evolution of metamorphic complexes. Nevertheless, a correct definition of the system parameters is not always straightforward due, for instance, to the difficulty in defining the real reacting equilibrium volume. Therefore, in this study, multispectral statistical image analysis has been applied to X-ray maps in order to minimize the operator subjectivity in the identification of equilibrium assemblages.

This paper provides the first detailed tectono-metamorphic reconstruction for the greenschist facies basement of the Peloritani Mountains, obtained through geothermobarometric modeling assisted by image data handling, via principal component analysis, of a fine-grained porphyroblastic metapelite from the Mandanici Unit. The used methodological approach is well suited to reliably reconstruct the porphyroblast-matrix relationships and therefore to better define both the succession and P-T conditions of the deformational and metamorphic events. Furthermore this method has proven to be a particularly powerful tool for quantifying the volumetric amounts of minerals that continue to participate in metamorphic reactions after fractionation of their inner portions.

The selected rock sample, whose mineralogical assemblage comprises quartz+muscovite+chlorite+ilmenite+leucoxene/rutile+garnet±fluorapatite±monazite±zircon, shows mainly the paragenesis stable at peak or near-peak conditions, depicting a crenulation schistosity S2 that largely obliterates the former schistosity S1. Retrograde evolution is mostly represented by the breakdown of garnet and ilmenite, replaced by chlorite and leucoxene/rutile, respectively, and by development of a S3 mylonitic schistosity locally preserved between fragments of pinch and swells garnet porphyroblasts.

Through a microstructural analysis especially focused on garnet porphyroblast microstructures, metamorphic-deformational relationships were reconstructed. Garnet recorded the second and third deformational event: cores with quartz and ilmenite inclusions exhibit a folding pattern linked to D2, indicating nucleation of the porphyroblast cores during the early stages of the second metamorphic event. The later stages of M2 is also responsible for unfolded rims over growing the crenulated cores. D3 produced stretching, testified by local pinch-and-swell structures and porphyroclasts rotation.

Garnet porphyroblasts gave essential information not only about the sequence of the deformational stages, but also about the changing metamorphic conditions. X-ray maps and compositional transects crossing rim, outer-core and inner-core of the investigated pophyroblasts revealed a growth-zoning pattern, from Alm64Sps17Grs14Prp5 (inner-core) and Alm69Sps15Grs11Prp5 (outer-core) to Alm75Sps13Grs6Prp6 (rim), with spessartine decreasing towards rims in favor of almandine. Furthermore, image analysis of X-ray maps, allowed to better define location and mutual textural relationships between garnet and ilmenite porphyroblasts, inclusion trails and matrix minerals, as well as to emphasize the geometry of the elemental composition distribution within garnet porphyroblasts, useful for the correct interpretation of the minerochemical data.

A recent version of the Perple_X phase diagram computation software package allowed to calculate a P-T pseudosection in the TiMnNCKFMASH system, able to define also the thermodynamic behavior of ilmenite that played a significant role in Mn-exchange reactions and resulted therefore very useful to further refine the P-T estimates. As a whole, garnet and ilmenite inclusion isopleths intersections allowed peak conditions (T~540 °C and P~0.8 GPa) to be determined.

In more detail, isopleths referred to garnet inner-core and pyrophanite isopleths indicate T~515 °C and P~0.93 GPa (baric peak), confirming that ilmenite was likely re-equilibrated during nucleation of garnet.

Outer core garnet isopleths intersections gave T~525 °C and P~0.93 GPa. By integrating microstructural and chemical data, these P-T estimates can be referred to the first metamorphic stage associated to isoclinalic folding. Rim garnet isopleths intersections provided T~540 °C and P~0.80 GPa, representing metamorphic peak conditions, associated to the development of the crenulation schistosity. Retrograde P-T trajectory was finally determined after quantification of the chemical fractionation effects, caused by progressive depletion of specific chemical constituents sealed in the inner portions of the crystals during early garnet growth stages, as well as in ilmenite portions not affected by retrogression to leucoxene. This permitted to calculate the new effective bulk rock chemistry operating during the retrograde metamorphic evolution and to obtain P-T pseudosection stability fields consistent with the observed retrograde parageneses.

Chemical isopleths in this P-T pseudosection permitted to ascertain a significant increment of the anorthite content in plagioclase, mainly controlled by decompressional effects, and an increment of the phengite content in white mica with pressure increase (Fig. 10). Observed compositional ranges of anorthite (An15-19) and phengite content (Phg12-18), coherent with the retrograde compositional equilibrium, allowed to constrain a relatively large pseudosection P-T space with temperature ranging between 420 and 460 °C and pressure of 0.30–0.60 GPa, coherent with a lower PT, greenschist-facies, retrograde stage.

The reconstructed pressure-temperature path (Fig. 11) depicts a typical clockwise P-T evolution. The prograde stage, peaking at boundary P-T conditions between greenschist and amphibolite facies, involved burial at middle-lower levels during crustal thickening, possibly connected to Hercynian continental collision processes. The following retrograde evolution took place under lower greenschist facies conditions, during a late- to post-orogenic shearing stage likely related to late Hercynian exhumation.

Fig. 11
figure 11

Comprehensive P-T path including all the reconstructed tectono-metamorphic events

Full size image

The obtained results introduce an element of great novelty to the geological knowledge of the Peloritani Mountains. Indeed, the basement rocks forming the Mandanici and Aspromonte-Peloritani Units have been up to now considered integral parts of the same chain sector developed at upper and middle crust conditions, respectively. The relatively high pressure values of ~0.93 GPa recorded by the studied metapelite sample highlight, on the contrary, that lower crustal conditions were attained by the rocks of the Mandanici Unit during their orogenic evolution.

It is demostrated in this study that new integrated techniques as the ones used are very powerful in reconstructing longer segments of a polyphase evolution in more detail, including reliable identification of both baric and thermal peak conditions, and have therefore provide much stronger constraints for the reconstruction of the tectono-metamorphic and geodynamic history of the mountain belts. These results open the way for a general review of previous interpretations on the crustal evolution of the chain sectors cropping out in the southern CPO, in the light of the highest potentialities offered by the new petrological techniques.