Lithofacies characteristics of diatreme deposits: Examples from a basaltic volcanic field of SW Sardinia (Italy)

Journal of Volcanology and Geothermal Research 255 (2013) 1–14

Contents lists available at SciVerse ScienceDirect

Journal of Volcanology and Geothermal Research

j ourna l homepage: www.e lsev ie r .com/ locate / jvo lgeores

Lithofacies characteristics of diatreme deposits: Examples from a basaltic volcanicfield of SW Sardinia (Italy)

F. Mundula a,⁎, R. Cioni a,b, A. Funedda a, F. Leone a

a Dip.to di Scienze Chimiche e Geologiche, Universita' di Cagliari, Via Trentino 51, 09127, Italyb Istituto Nazionale di Geofisica e Vulcanologia, Pisa, Via della Faggiola 32, 56126, Italy

⁎ Corresponding author. Tel.: +39 070 6757711; FaxE-mail address: [email protected] (F. Mun

0377-0273/$ – see front matter © 2013 Elsevier B.V. Allhttp://dx.doi.org/10.1016/j.jvolgeores.2013.01.014

a b s t r a c t

Article history:Received 31 August 2012Accepted 23 January 2013Available online 31 January 2013

Keywords:DiatremeLithofaciesAlkali basaltsSW Sardinia

A deeply eroded diatreme field, consisting in several, decametric-sized, vertical, mainly clastic volcanic bodiesof basaltic composition is described for the first time in the Variscan basement of SW Sardinia. The recognitionand description of four different lithofacies in these diatremes allowed discussion of the role of the differentprocesses which control magma eruption and conduit infilling, and making general inferences about dia-tremes. The studied diatremes have a cross-sectional shape from elliptical to sub-triangular, and are slightlyelongated nearly parallel to the main foliation of the intruded meta-sedimentary rocks. Foliation of hostrocks is locally reoriented or folded close to the contact with the diatremes, suggesting that magma possiblyrose to the surface through fissures oriented nearly parallel to host rock foliation. Textural features of thevolcanic bodies show many analogies with kimberlitic diatremes, despite the difference in petrography andcomposition. Juvenile lapilli are mainly made by ghosts of mafic phenocrysts (olivine and clinopyroxene)set in a groundmass formed by plagioclase microlites immersed in a cryptocrystalline, chlorite-rich matrix.The four lithofacies were described mainly based on the shape and physical features of the clasts andtextural anisotropy: a globular, juvenile-rich, lapilli tuff facies (GJLt); an angular, juvenile-rich, lapilli tuff facies(AJLt); a lithic-rich, lapilli tuff facies LiRLt), and a coherent, lava-like facies (COH). All the clastic lithofacies aregenerally well sorted and typically lack a fine-grained matrix. Juvenile fragments are lapilli sized and fromequant to oblate in axial ratio, and from rounded-globular to very angular in shape. Conversely, lithic clasts arelargely variable in shape and size, and are mainly represented by basement-derived clasts. The absence ofbedding, the scarcity of the coherent facies and the dominance of clast supported, structureless, volcaniclasticfacies suggest that the outcropping portion of these volcanic bodies represents the lower diatreme zone. Thepresence of diffuse welding and the globular shapes of some juvenile fragments, together with their vesicularity,suggest that magma fragmentation was mainly driven by magmatic gas exsolution occurring at a deeper levelrespect to classical, basaltic explosive activity. Textural features, facies association and facies architecture of thestudied deposits are suggestive of an important affinity with kimberlitic and other ultramafic diatremes.

© 2013 Elsevier B.V. All rights reserved.

1. Introduction

The deep structure of volcanic systems represents an importantsource of information for understanding eruptive dynamics and pro-cesses (e.g. Kano et al., 1997; Soriano et al., 2006; Lorenz andKurszlaukis, 2007; Keating et al., 2008; White and Ross, 2011). Theknowledge of the large geological and shape variability presentedby volcanic conduits is of utmost importance for the definition ofthe processes and boundary conditions which control magma ascentto the surface. While the geology of the crater zone is generally wellexposed in numerous volcanoes worldwide, the lithological andtextural features which characterize the products filling the conduitregion can only be studied in deeply eroded terrains, where they are

: +39 070 282236.dula).

rights reserved.

often badly preserved due to superimposed tectonic deformation andmetamorphism. The present knowledge about their characteristics ismainly based on drillings (e.g. Goto et al., 2008) or on the direct obser-vation of conduit-filling deposits especially in diatreme structures(e.g. Németh et al., 2007; Brown et al., 2009). The geology of thesebodies is variously well described for kimberlitic diatremes, whereseveral papers investigated the deepest part of the structure usingboth surface data and observations made on excavations and drillcores (Downes et al., 2007; Hetman, 2008; Seghedi et al., 2009). Anincreasing amount of data is now available concerning maar–diatremestructures of non-kimberlite composition allowing a generalizationof the knowledge previously confined to kimberlite diatremes (Whiteand McClintock, 2001; Ross and White, 2006; Ross et al., 2008; Whiteand Ross, 2011; Ross and White, 2012). In particular, the fieldobservation of diatremes, and the spatial relationships betweenthe different lithofacies, can give important information about the

2 F. Mundula et al. / Journal of Volcanology and Geothermal Research 255 (2013) 1–14

general geometry of the conduit system and the original depth ofemplacement, and about the definition of syn-eruptive processesdriving magma ascent and fragmentation.

A field of volcanic bodies with a circular to elongated shape inmap view and very steep contacts was recognized in the Sulcis area(SW Sardinia, Italy). They crop out on an erosive surface cutting theVariscan basement and are interpreted as diatremes filled by pyro-clastic material, formed by the explosive eruption of alkali basalt.

Each diatreme is few tens of metres across, cuts nearly verticallythrough Upper Ordovician metasediments, and are unconformablycovered by Lower Trias and Paleocene–Eocene sediments (Fig. 1a, c).We describe and discuss here the geological features of these dia-tremes, mainly focussing on the macroscopic and microscopic charac-terization of the different lithofacies recognized in the infillings and ontheir lateral relationships. They were investigated in order to drawlocal and general inferences on: 1) the facies association and architec-ture within the diatremes; 2) the nature and emplacement dynamics;and 3) the structural relationships with the host rocks.

1.1. Terminology

A volcanic conduit is defined as “a pipe-like pathway for the transportof magma that evolves by preferential channelling of flowwithin a dike”(Carrigan, 2000). More generally a conduit can be considered a pathwaythat connects a magma reservoir to the volcanic edifice during aneruption (Goto et al., 2008 and references therein).

Most of the currently exposed or drilled volcanic conduits havebeen named diatremes (e.g. Németh and Martin, 2007; Goto et al.,2008) but the term diatreme has been alternatively used with atleast two different meanings: a) a pipe-like volcanic conduit filledwith volcanic debris constituting the substructure of a maar volcano

Fig. 1. Geology of the studied diatremes, NE of Carbonia, Sardinia, Italy. (a) Geological map of(b) Simplified geological map of South Sardinia and location of the study areas; (c) Schema

(Cas and Wright, 1987; Martin et al., 2007; White and Ross, 2011);b) or more specifically, in the study of kimberlitic volcanism, theintermediate zone of class 1 kimberlitite volcanoes (Field and ScottSmith, 1999; Skinner and Marsh, 2004). Throughout the paper, wewill use the term diatreme according to the first definition, as aremnant of volcanic conduit filled by volcanic debris independentlyby composition, prevailing eruptive style and textural feature of thefilling material.

Here we describe and group the different rock types on the basis ofthe widely accepted terminology of Fisher and Schmincke (1984) andCas and Wright (1987) for pyroclastic rocks, recently reviewed inWhite and Houghton (2006) and in Cas et al. (2008). The rocks are de-scribed in terms of components, structure, and texture. In the texturaldescription of the particles forming the pyroclastic rocks, the mor-phological features comprise terms descriptive of the relative axial di-mensions (equant, oblate, etc.) and of the general shape (e.g. rounded,globular, and angular). Terms describing clast shape are used withoutany specific genetic significance. Vesicularity is described according tothe classification proposed by Houghton and Wilson (1989).

2. Geological setting

2.1. The host rocks

The study area is located in the Sulcis region, SW zone of Sardinia,Italy (Fig. 1b). The geology of this area is mainly formed by an UpperOrdovician terrigenous succession formed by metasandstone, meta-siltites, metargillites and minor metaconglomerates, from the baseto the top: M.te Argentu Formation (Leone et al., 1991); M.te OrriFormation, Portixeddu Formation (AGU, MRI and PTX respectively inFig. 1a, c,). These formations unconformably lay on the Cambrian

M.te Pertunto area (SW Sardinia) and location of the different diatremes for area 1 in (b).tic geological section of the study area.

3F. Mundula et al. / Journal of Volcanology and Geothermal Research 255 (2013) 1–14

and Lower-Ordovician sedimentary succession, constituted by pre-vailing metasiltites (Cabitza Group — CAB in Fig 1c) and metacar-bonates (Gonnesa Group — GNN in Fig. 1c). All these were involvedin the Variscan orogenesis during the Middle Carboniferous, undergo-ing an anchizonal to very low grademetamorphism (Carmignani et al.,2001). The Variscan structural framework consists of very close toisoclinal upright folds (Funedda, 2009; Casini et al., 2010), with N–Strending axis and with a well developed axial plane foliation (generallya spaced cleavage in original sandstones and a slaty cleavage in finegrain metasiltstones) (Fig. 1c). Cleavage is the main anisotropy visiblein the field; it strikes between N010° to N030° with a nearly verticalplunge (Fig. 2a).

2.2. The diatremes

The folded metasedimentary succession is cut by at least ten,vertically oriented, volcanic bodies, mainly formed by clastic volcanicfacies and minor coherent facies. These bodies have been, for longtime, interpreted as basaltic dikes of Upper Carboniferous–LowerPermian age (Beccaluva et al., 1981). The general trend of the foliationof the hosting metasedimentary rocks rotates close to the diatremes

Fig. 2. (a) Rose diagrams showing dispersion of strike measures on Variscan cleavage atmore than 100 m from the diatremes (left) and close to the contact with the diatremes(right); (b) geological maps of the diatremes D, G, H and L; diatreme L is located in studyarea 2 (Fig. 1b).

(within a 100 m distance) where it shows a higher dispersion, withstrikes between N010° to N060° (Fig. 2a).

In map view, these bodies vary between 200 and 1200 m2

(Fig. 2b); they generally correspond to small hills with a maximumelevation of about 15 m above the surrounding surface. They are notaffected by the Variscan tectonics (foliation and folds), have a verticalpipe-like shape, and, in map view, have shapes from elliptical tosub-triangular with NNE–SSW maximum elongation, sub-parallel tothe main trend of Variscan foliation (Fig. 2b). The outer sectors ofthese bodies show a clear anisotropy related to the 3D oblate shape(a=b>c) of the fragments in the clastic facies, with the ab planesubparallel to the contact with the host rocks. Where these marginsare made up by coherent facies, the major axis of phenocrysts andvesicles is vertical and lays on a plane subparallel to the contact. At theoutcrop scale, clast, crystal and vesicle anisotropy closely simulates apseudo-cleavage.

The most evident anisotropic feature of the diatreme infillings isa flattening surface, underlined in some cases by the shape of thejuvenile clasts and by the local occurrence of a largely spaced foliation.The bodies are always fractured; fractures are mainly open, withmm-sized spaces between walls, in some cases filled by secondaryminerals. Fracture distribution is heterogeneous in space and orienta-tion at the outcrop scale, and it is characterized by multiple systemsof sub-parallel, dm-spaced joints intersecting without substantial dis-placement along the fractures planes. There is no evident correlationbetween these fractures and the cleavage observed in the host rocks.The contact between the diatremes and the host rocks is often concor-dant with the Variscan cleavage, whilst in several cases cross cuts thecleavage at low angle (Fig. 3a). Drag folds deforming the Variscancleavage are present in the host rock close to the contact with thediatremes.

The flat erosional surface cutting the Variscan basement is, in gen-eral, unconformably covered by undeformed Paleocene–Eocene con-tinental and marine sediments with sub-horizontal bedding. Rarelens-shaped outcrops of conglomerates with Palaeozoic clasts cropout below the Paleocene cover and have been referred to the baseof Triassic succession (Buntsandstein), suggesting that the erosionalsurface developed during the Late Palaeozoic (Pasci et al., 2012).Diatremes or related deposits do not cut through the Post-Variscansuccession in the study area. Other volcanic bodies, with similarfeatures and the same overprinting relationships with the Variscanbasement, but not adequately well exposed to be studied, crop outin the Sulcis area, both few km westward from the study area andfew km east from the Villamassargia village (Fig. 1b).

3. Lithofacies characteristics

The lithology of diatremes is largely variable, showing three mainclastic facies, dominated by lapilli and bomb-sized fragments, and acoherent facies, apparently not clastic. Rocks from all pyroclasticfacies are generally unbedded, clast supported, well sorted lapilli tuff(up to lapillistone), typically characterized by a scarcity of a fine-grained matrix. The general colour of the diatremes filling rocksvaries from dark grey to green to yellow or red, mainly related to thedifferent extent of alteration. The juvenile material, largely prevalentover lithic clasts (generally lower than 25%), is characterized by a largehomogeneity of lithology, shape, vesicularity and colour (Figs. 5–8).

The juvenile material is aphanitic to weakly phaneritic, and theprimary mineral paragenesis is commonly obliterated by alteration.Only few mm-sized phenocrysts of unaltered olivine (Fig. 8d) andclinopyroxene (Fig. 8e), opaques, and abundant pseudomorphs ofmafic phenocrysts substituted by chlorite, carbonate and clay min-erals are present, set in a cryptocrystalline mass of chlorite, mica, epi-dote and opaques with abundant albitized plagioclase microliths.Inferred former glass in the groundmass is completely obliterated bysubsequent alteration. The juvenile fraction of clastic facies is always

Fig. 3. (a) Angular unconformity at the contact between the host rock and COH at theSE margin of diatreme D (the white line highlights the cleavage of the host, the blackline highlights the contact); (b) field image of the GJLt in the diatreme G; (c) bouldersized Cambrian carbonate lithic clast in LiRLt.

4 F. Mundula et al. / Journal of Volcanology and Geothermal Research 255 (2013) 1–14

poorly to moderately vesicular and, in many cases, vesicles are filledby secondary calcite and are partially collapsed or deformed. Lithicscomprise metamorphic siliciclastic and carbonate basement material.Rare gabbroic xenoliths made by few aggregated crystals of plagio-clase and pyroxene are present; no mantle derived xenoliths havebeen observed.

Chemical analyses were performed on selected samples from thecoherent facies, due to the difficulty of separating juvenile materialfrom lithic material or inter-clast matrix in the clastic facies. Foursamples were analyzed for major and trace elements (major elementsby ICP-OS; trace elements by ICP-MS; Table 1) in order to characterizethe volcanic products and define the geodynamic context responsiblefor the emplacement of the volcanic bodies (Table 1). Mainly basingon the still recognizable mineral paragenesis of the juvenile material,characterized by a large abundance of plagioclase microlites andminor mafic phenocrysts, we classify the diatreme-filling rocks asalkaline basalt. We interpret the very low SiO2 content reported inthe analyses (about 41–42 wt.%; Table 1) as mainly related to thedeep alteration of the groundmass, in which pristine glass is largelysubstituted by hydroxyl-bearing minerals (mainly chlorite) andcalcite, also resulting in a very high LOI (Loss on Ignition) value, up toabout 9 wt.% (Table 1). Indeed, when recalculated on an anhydrousbase, rocks show a classical basaltic composition. Trace element dataconfirm this interpretation (Winchester and Floyd, 1977; Fig. 4a), alsosuggesting a tectonic setting typical of within-plate alkaline basalts,as evidenced by the classical diagrams of Meschede (1986; Fig. 4b) orby the OI-normalized spidergram of Fig. 4c.

Four different lithofacies have been distinguished in the diatremes,representing the end-members of a continuous variation in terms ofmacroscopic texture, shape, surface features and lithology of the clasts,abundance of lithic fragments, nature of the matrix and of the vesiclesfilling. The following facies have been distinguished: 1) globular,juvenile-rich, lapilli tuff facies (GJLt); 2) angular, juvenile-rich, lapillituff facies (AJLt); 3) lithic rich, lapilli tuff facies (LiRLt); and 4) coherentfacies (COH).

3.1. The two juvenile-rich lapilli tuff facies

These two facies are characterized by an unbedded, generally clas-supported structure. On average, the amount of thematrix poses thesetwo facies at the limit between the fields of lapillistone with lapilli tuff(Fisher, 1966). According to the new classification scheme proposedby White and Houghton (2006), we prefer the term lapilli tuff. Thejuvenile fragments vary from cm- to mm-sized and their shape variesfrom equant-spherical to oblate, to subspherical and rarely polygonal,and from very angular to well rounded. Juvenile lapilli vary frompoorly to moderately vesicular, with ellipsoidal vesicles mainly filledby secondary calcite and minor microcrystalline quartz; collapsedvesicles are largely subordinate. Lapilli are always porphyritic forthe presence of mafic crystals up to few millimetres in size; in somecases, mm- to cm-sized lithic clasts in the core are also present. Lithicfragments are largely variable in shape and size, and are mainly repre-sented by clasts from the siliciclastic host rocks or from the underlyingCambrian carbonates. The cement ismainly constituted by carbonates,quartz and chlorite.

The distinction into two different facies is mainly based on theexternal shape of the juvenile fragments: globular (GJLt) vs. angular(AJLt).

3.1.1. Globular, juvenile-rich, lapilli tuff facies (GJLt)Juvenile fragments in the GJLt (Fig. 3b) are characterized by a

sub-spherical (Fig. 5a) to ellipsoidal shape (Fig. 5c), a smooth externalsurface, and vary from 2 to 3 cm to few mm in size. Packing featuresvary from an open-framework, where clasts are in contact and the20–40% of the space is represented by the cement (Fig. 5b) or byinter-clast voids, to a very compact packing, where clasts form clusters

Table 1Bulk chemistry analyses (major elements by ICP-OS; trace elements by ICP-MS) per-formed on samples from COH.

Oxide wt.% PP96a PP202 PP200 PP96b

SiO2 41.27 42.95 42.26 42.36Al2O3 14.44 14.51 14.81 15.01Fe2O3(tot) 9.67 9.18 9.33 9.21MnO 0.162 0.152 0.154 0.15MgO 6.36 6.9 6.69 6.55CaO 12.61 11.4 11.05 11.12Na2O 3.24 3.71 3.5 3.65K2O 0.24 0.15 0.24 0.2TiO2 2.145 2.118 2.104 2.164P2O5 0.56 0.5 0.51 0.54LOl 7.69 8.25 8.94 8.4Total 98.39 99.83 99.57 99.36

Trace ppmSc 23 22 22 23Be 2 2 2 2V 220 217 216 221Cr 180 160 160 160Co 36 34 34 35Ni 50 140 140 140Cu 60 60 60 70Zn 220 260 290 260Ga 17 15 15 16Ge 1.2 1.5 1.8 1.6As 8 8 6 7Rb 7 b1 2 b1Sr 666 447 452 454Y 28.2 25 25.8 25.8Zr 207 216 224 228Nb 53.9 62.3 68.4 68.9Mo 2 b2 b2 b2Ag b0.5 b0.5 0.5 0.6In b0.1 b0.1 b0.1 b0.1Sn 3 2 2 2Sb 0.3 1.1 1 0.7Cs 1.4 0.5 0.7 0.6Ba 1121 855 600 511La 44 40.4 40.6 40.9Ce 80.7 74.9 75.5 76.7Pr 8.51 9 9.16 9.27Nd 29.9 32.8 32.9 34.4Sm 6.55 6.16 6.65 6.7Eu 2.11 1.93 1.93 1.93Gd 6.38 5.89 5.88 6Tb 1.04 0.87 0.89 0.92Dy 5.72 4.87 4.88 5.08Ho 1.04 0.88 0.91 0.93Er 2.84 2.43 2.53 2.5Tm 0.397 0.344 0.357 0.346Yb 2.39 2.25 2.32 2.26Lu 0.347 0.334 0.342 0.342Hf 4.5 4.2 4.3 4.4Ta 3.75 3.97 4.15 4.15W 1.2 b 0.5 b 0.5 b 0.5Tl 0.09 b0.05 b0.05 b0.05Pb 33 74 60 66Bi b0.1 b0.1 b0.1 b0.1Th 5.54 5.53 5.59 5.66U 1.55 1.52 1.49 1.54

Fig. 4. (a) Classification scheme for altered volcanic rocks (Winchester and Floyd, 1977);(b) Nb–Zr–Y abundances and fields for different petrologic affinity (Meschede, 1986);(c) OIB-normalised REE spidergram.

5F. Mundula et al. / Journal of Volcanology and Geothermal Research 255 (2013) 1–14

resulting from partial sintering (the surface of each clast is no morerecognizable at the interface between different clasts), and inter-clastcement constitutes less than 20% of the volume. The cement is mainlymade of microcrystalline carbonate minerals, while quartz and chlo-rite are largely subordinate. Pinching inwards contacts between juve-nile fragments are observed at the meso- and micro-scale (Fig. 5band d). The lithic content is generally low in GJLt, varying between2% and 15%. The globular, smooth shape of the clasts suggests that,after fragmentation, surface tension of the melt was still able toreshape the clasts.

3.1.2. Angular, juvenile-rich, lapilli tuff facies (AJLt)AJLf differs from GJLf mainly for the shape and surface features of

the juvenile clasts, for the nature of the cement and for the degree oftextural anisotropy. Juvenile fragments are weakly oriented and canbe approximated to an oblate ellipsoid on the basis of the ratiobetween the three main axes. The external surface of the clasts isrough and jagged, with angular cuspate outlines (Fig. 6b and d). Pack-ing varies between open framework and very compact packed (Fig. 6aand c). At the micro-scale, contacts between different clasts are oftensintered, and microliths and vesicles are arranged along a preferential

Fig. 5. Representative images of GJLt. (a), (b) and (c) are polished slabs. (a) Lapilli-supported isotropic texture in GJLt (sample PP73 from diatreme G). (b) Detail of image(a) evidencing, partial sintering between lapilli-sized juvenile fragments and pinching inward contacts. Some juvenile fragments show a core constituted by carbonate lithics orby mafic phenocrysts. (c) Lapilli-supported anisotropic texture in GJLt. The grey particles in the upper half of the slab represent metapelite lithics (sample PP46 from diatreme G).(d) Thin section image showing features of juvenile fragments. The inter-clast space is filled by calcite. On the upper left cuspate outlines merging inward the juvenile fragmentcould represent the original contact between two fragments (sample PP59 from diatreme E).

6 F. Mundula et al. / Journal of Volcanology and Geothermal Research 255 (2013) 1–14

direction, with their long axes sub-parallel to clast elongation. Theinter-clast space is completely filled by microcrystalline quartz, andsubordinate carbonate and chlorite. The lithic content of AJLt variesbetween few percent to 25%.

3.2. Lithic rich lapilli tuff (LiRLt)

LiRLt is a lapilli tuff (up to limit with the tuff breccia) characterizedby the presence of abundant (25 to 45 vol.%), large lithic fragmentsfrom the Paleozoic substrate. Space between lithic fragments is filledby juvenile fragments and secondary minerals. Juvenile fragmentsare generally deformed (Fig. 7c) and sometimes squeezed (Fig. 7b)in relation to the amount and size of nearby lithics. Features of juvenilefragments vary between those typical of AJLt (irregular or oblateshape and angular-rough surface) and those of GJLt (from sphericalto ellipsoidal in shape and rounded and smooth surface). The lithicfragments vary in size from mm-sized lapilli to dm–sized boulders(Fig. 3c), whereas juvenile fragments are always in the fine to coarselapilli range. Dimensional relationships between lithic and juvenileclasts allow for distinguishing between a well sorted LiRLt, character-ized by the presence of pebble-sized xenoliths from the siliciclasticcountry rocks (Fig. 7a, c), and a poorly sorted LiRLt (Fig. 7b) character-ized by the presence of pebble and boulder-sized carbonatic clasts,andminor pebble-sized siliciclastic clasts. The latter shows a completealteration and replacement of the juvenile fraction by chlorite andcarbonates, and only fiamme-shaped pyroclasts are still recognizable,squeezed up between gravel-sized lithics (Fig. 7b and d).

3.3. Coherent facies (COH)

COH is macroscopically characterized by the absence of a clearlydefined clastic structure. Lithic fragments from cm to mm-sized arenevertheless present (0–10% by volume; Fig. 8a). The texture isporphyritic with dark, mm-sized, mafic crystals, and varies from iso-tropic (Fig. 8a) to strongly anisotropic, with oriented ellipsoidal tostrongly attenuated vesicles and elongated phenocrysts and lithics(Fig. 8b).

At the micro-scale, the arrangement of plagioclase microlithsand the spatial distribution of the micro-cryptocrystalline matrix de-pict in some cases ghosts of fragments (Fig. 8c), suggesting that COHpossibly results from dense welding/sintering of an originally clasticfacies. The shape of the original clasts is in some cases evidencedby concentrically or radially arranged plagioclase microliths, whichmimic the original shape of the pseudo-clasts. The inter-clast spaceis defined by the absence of plagioclasemicroliths and by the presenceof a cryptocrystalline groundmass. Fractures and vesicles are distrib-uted heterogeneously in COH, and are alternatively filled by carbonateand silicaminerals, or by chlorite, talc andmica; in some cases vesiclesare empty, possibly due to epigenetic dissolution.

3.4. Lateral and vertical facies relationships inside the diatremes

Four out of at least ten diatremes were mapped in detail (Fig. 2b),in order to discuss the mutual spatial relationships between the de-scribed lithofacies. Facies transitions inside the exposed diatremes

Fig. 6. Representative images of the AJLt. (a), (b) and (c) are polished slabs. (a) Lapilli-supported weakly anisotropic texture. Packing is highly heterogeneous at the scale of handsamples (sample PP76 from diatreme H). (b) Detail of image (a) showing fringed outlines of juvenile fragments and the presence of a lapillus cored by a xenoliths. (c) Example ofAJLt showing heterogeneity in the packing degree and the presence, in the upper half, of a cm-sized polygonal xenolith (sample PP74 from diatreme H). (d) Thin section imageshowing the vesicularity of juvenile fragments and the presence of a microcrystalline matrix partially filling the inter-clast space (sample PP76 from diatreme H).

7F. Mundula et al. / Journal of Volcanology and Geothermal Research 255 (2013) 1–14

are generally gradual, and abrupt changes of facies are subordinate.The transition from the juvenile rich facies to LiRLt is evidenced bya progressive increase of lithic content and of the deformation ofjuvenile clasts, which show incipient squeezing and deformationstarting at a lithic content higher than 30–40%. Transition betweenGJLt and AJLt is often marked by a progressive change in the shapeand surface roughness of juvenile clasts. The transition betweenthe two juvenile-rich lapilli tuff facies and COH generally occursthrough a progressive increase of clast packing, up to obliteration ofthe inter-clast space. Finally, the transition between LiRLt and theCOH is characterized by an increase of packing of the juvenile clasts,associated with a decrease of the lithic content.

Diatreme D is ellipsoidal in shape (in map view, Fig. 2b), and has aNE–SW elongation. The length is about 65 m and the width about35 m. Facies distribution is complex, but it can be schematicallydescribed as concentric. In the outer zone AJLt dominates, with themain anisotropy (marked by oblate juvenile fragments) sub-parallelto the margins of the diatreme; GJLt dominates the inner zone withthe same elongation of the body; the central zone of the pipe is char-acterized by the presence of COH and LiRLt, which are also recogniz-able along the east margin. In the central zone of diatreme D, LiRLt ischaracterized by the sole occurrence of cobble-sized and boulder-sized carbonates (up to 40 cm in size; Fig. 3c) from the Cambrianformations, whereas, at the diatreme boundaries, it is characterizedby pebble-sized, foliated and randomly oriented, siliciclastic materialmainly from the Upper Ordovician host rocks. Despite the transition

between COH and all the other clastic facies being generally progres-sive, abrupt changes are sporadically observed.

Diatreme G presents a sub-triangular shape with maximum elon-gation in the NE–SW direction, a maximum length of 40 m and awidth of 30 m. COH dominates the central sector of the diatreme,and it is surrounded by rocks of GJLt and, locally, AJLt. LiRLt is limitedto a very small area at the contact with the Paleozoic basement(Fig. 2b). Rocks from this diatreme are strongly altered, and the juve-nile fraction varies in colour from red to green to yellow.

Diatreme H presents an elongate shape in NNE–SSW direction,maximum length of 25 m and width of 8 m. It is entirely representedby AJLt, with a pseudo-foliation sub parallel to the direction of elonga-tion of the pipe. Clast-packing progressively increases from the mar-gin to the central zone of the pipe, where textural features aretransitional between AJLt and COH.

Margins of diatreme L cut at low angle the host rock foliation(San Marco Fm., Upper Ordovician; Leone et al., 1991) locally E–Woriented. In map view, the diatreme has maximum length of 25 mand width of 10 m. It is mainly represented by GJLt and LiRLt, andthe transition between the two facies is marked by a progressiveincrease in lithic content and dimensions, with lithics being largelyvariable at the meso-scale. Some features differentiate GJLt of diatremeL from the same facies in the other bodies. In particular, in diatreme L:1) the juvenile fragments have a higher vesicularity and are moreporphyritic, and 2) the matrix between clasts is represented by alteredfine material rather than secondary minerals.

Fig. 7. Representative images of LiRLt. (a), (b) and (c) are polished slabs. (a) Clast-supported weakly anisotropic texture. Marble fragments represent the main lithic constituents ofthe rock (sample PP19 from diatreme D). (b) The sample is mainly made of marble fragments and the juvenile fragments are squeezed up between lithic clasts (sample PP71 fromconduit D). (c) Clast-supported anisotropic texture. Metapelite fragments represent the main lithic constituent of the rock. The fiamme-like shape of the juvenile fragments could bedue to plastic deformation between rigid lithic clasts (sample PP64 from diatreme B). (d) Thin section showing chloritized juvenile fragments squeezed up between marblexenoliths (sample PP71 from diatreme D).

8 F. Mundula et al. / Journal of Volcanology and Geothermal Research 255 (2013) 1–14

4. Quantitative textural analysis

4.1. Methods

In order to get an objective description of textural variability be-tween the different facies, and to understand the relationships betweenthe host rock cleavage and the anisotropy observed in the diatremes,image analysis was performed on eight representative samples ofthe clastic facies. Samples were cut and polished, and digital imagescollected. Juvenile clasts on the images were outlined with a graphicaltool of Adobe Photoshop© CS2 software, in order to obtain binarizedimages (Fig. 9). The binarized images were analysed with the ImageJsoftware (Rasband, 1997), and size, circularity (circularity=4π (area/perimeter2)), aspect ratio of the juvenile fragments (long/short axis),and relative amount of juvenile clasts, lithics and inter-clasts matrixwere measured. Lithics were distinguished in included lithics, repre-senting the core of cored lapilli, and free lithics, loose in the matrix. Inorder to quantify the median diameter and sorting (MdΦ and σΦ;Inman, 1952) relative to the juvenile fragments plus free lithics, 3Dsize distribution was derived from 2D size data with CSD correction1.38 software (Higgins, 2000) and results were plotted with SFT soft-ware (Wohletz et al., 1989).

On two oriented samples from diatreme H, analyses were performedcutting samples along twoplanes orthogonal to the near vertical pseudo-foliation (A-side vertical, B-side sub-horizontal) in order to evaluate dif-ferences in clasts orientation and flattening. To estimate the dispersion of

the orientation of elongated pyroclasts, we used the circular variance(Vc) parameter (Tran, 2007; Mundula et al., 2009) for circular statisticson orientation data.

4.2. Results of image analysis

Results of image analysis are listed in Table 2. Circularity varies be-tween 0.3 and 0.7, out of a possible range of 0 (theoretical value for aninfinitely elongated shape) to 1 (perfect circle) and shows distinctfields for GJLt (higher values), LiRLt (intermediate values) and AJLt(lower values). The average aspect ratio varies between 1.7 and 2.8and is unrelated to the facies type and to the circularity parameter.The size distribution of clasts shows MdΦ values between −4.4 and−3.0 ϕ (between about 21 and 8 mm) and σΦ values between 0.9and 1.1 ϕ. MdΦ and σΦ testify that the deposit is mainly constitutedby well sorted cm-sized clasts. In the limit of the low statistics ofpresented data, no clear relationships were observed between theMdΦ and σΦ parameters or between the grain size parameters andthe facies type. Juvenile fragment abundance varies between 34%(in LiRLt) and 78% (in GJLt) of the samples. Lithic content varies be-tween 43% of LiRLt and 2% of GJLt. In AJLt, lithic content varies between6% and 16%.Vc of the oriented AJLt samples varies between 0.3 and 0.6,while aspect ratio (AR) is poorly variable, between 2.1 and 2.3. Thesevalues testify that pyroclasts are generally oblate and weakly orientedon the vertical plane.

Fig. 8. Representative images (a, b: polished slabs; c, d and e: thin sections) of COH. (a) Isotropic texture evidenced by randomly oriented mm–sized crystals, lithics and vesicles.Cm-sized lithics of heterogeneous nature (sample PP72 from diatreme G). (b) Anisotropic texture evidenced by sub parallel orientation of elongated mm-sized lithics, crystals andflattened vesicles (sample PP14 from diatreme I). (c) Porphyritic texture evidenced by a microphenocryst of olivine (Ol) pseudomorph in a microcrystalline matrix of albitizedplagioclase (Pl). The distribution of Pl microliths depicts in some cases ghosts of fragments (outlined with a white dash line). The grey-white ovoidal-shape particles, inside thedepicted ghosts of fragments and dispersed through the thin section, could represent vesicles amygdales (sample PP70 from diatreme D). (d) Unaltered Ol phenocryst (samplePP200 from diatreme D). (e) Unaltered clinopiroxeno (Cpx) microphenocryst (sample PP200 from c diatreme D).

9F. Mundula et al. / Journal of Volcanology and Geothermal Research 255 (2013) 1–14

5. Discussion

The internal structure, shape and size of the observed volcanicbodies allow drawing some inferences about their nature and processesof formation. In particular, the facies characterization revealed a veryuseful tool for the interpretation of themain processeswhich controlledthe emplacement of the diatremes, and their relative timing.

5.1. Nature and age of the studied diatremes

The observed volcanic bodies have been interpreted in the pastas remnants of dykes or tabular intrusive bodies (Beccaluva et al.,1981), and their clastic nature had never been described before. Thedetailed description of the different lithofacies which characterizethese bodies, and their shape, size and vertical geometry, supportan alternative interpretation as diatremes. The juvenile fraction ofthe described deposits is characterized by a porphyritic texture and afinely microcrystalline matrix, possibly completely substituting a for-merly glassy groundmass. The homogeneous petrographic featuresof juvenile fragments from all the clastic facies as well as from COH,and their close association inside a single body, suggest a commonmagmatic origin, and the primary pyroclastic nature of the clasticfacies. The lack of evident bedding in the studied deposits, the nearlyvertical contacts with the metasedimentary host rocks, in most casescross cutting at a low angle the Variscan foliation, and their limitedlateral extent, suggest that they represent portions of the lower

diatreme/root zone. Also the association of several of these bodiesover a restricted area represents another typical feature in commonwith diatremes. The absence of deeply-derived (mantle) xenolithsrepresents an important difference from kimberlite diatremes, whichalways present this type of xenoliths (White and Ross, 2011).

Geological and structural features of the diatremes and of theirhost rocks are not resolutive for pinpointing the age of the relatedvolcanic activity. The absence of Variscan deformation superimposedon these bodies testifies for the post-Variscan age of this volcanicactivity. The geometry of the volcanic bodies and their interpretationas deposits of the diatreme/root zone suggest that an importanterosion (possibly hundreds of metres), occurred before the deposi-tion of the Lower Triassic unconformable conglomeratic sequencewhich covers the Upper Ordovician rocks. All these data are sugges-tive of an age for the volcanic event between the end of the Variscandeformation (Middle–Upper Carboniferous) and well before the be-ginning of Middle Triassic sedimentation. This event well correlatesin space, time and compositional affinitywith a volcanic event recordedin northern and central Sardinia by alkaline basalt dykes (Traversa andVaccaro, 1992; Cortesogno et al., 1998).

5.2. Inferences on the fragmentation mechanisms

The characteristics of the deposits and fragments of the clastic faciescan give some indications around the possible mechanisms which con-trolled magma fragmentation. Deposits are generally rich in juvenile

Fig. 9. Selected thresholded images used for image analysis. Black particles represent juvenile fragments while grey particles represent lithic fragments. (a) AJLt (sample PP209 Aside from diatreme H) e (b) AJLt (sample PP76 B side from diatreme H), (c) LiRLt (sample PP64 from diatreme B), d) GJLt (sample PP73 from diatreme G).

10 F. Mundula et al. / Journal of Volcanology and Geothermal Research 255 (2013) 1–14

fragments, with lithic rich facies confined to some (minor) portionsof the diatremes. Juvenile fragments from GJLt and AJLt are evidentlyvesicular, and clasts boundaries are generally smooth or jagged. InGJLt, the abundance of globular lapilli suggests the occurrence of pro-cesses of magma relaxation mainly driven by the dominating effectof surface tension over viscosity, and indicating high temperature ofemplacement. Similarly, the sintering observed between fragmentsboth in AJLt and GJLt implies temperatures of the clasts above thecharacteristic temperature of glass transition (Riehle et al., 1995;Giordano et al., 2008). By contrast, clasts with blocky-polygonal shapes,typical of phreatomagmatic activity, are rare, and we never observedany clasts with chilled margins or with externally fractured breadcrust-like structures. Without entering in a detailed discussion of fragmenta-tion processes, the textural and morphological features of the juvenile

Table 2Image analysis results. MdΦ and σΦ are median and sorting parameter (Inman, 1952); circularea %; lithic clasts are subdivided into clasts included in the juvenile material (incl. lithics

Sample Facies Md(ϕ)

Σ(ϕ)

Circularity St. dev.Circ.

AR

PP46 GJLt −3.8 1.0 0.6 0.2 2.2PP73 GJLt −4.0 1.0 0.7 0.2 1.7PP68 GJLt −3.0 1.1 0.6 0.1 1.8PP76 A side AJLt −4.2 1.1 0.3 0.1 2.3PP76 B side AJLt −3.8 1.1 0.5 0.2 2.1PP209 A side AJLt −4.3 0.9 0.4 0.1 2.1PP209 B side AJLt −4.4 1.0 0.4 0.1 2.2PP19 LiRLt −3.5 1.1 0.6 0.1 2.0PP64 LiRLt −3.8 1.1 0.5 0.2 2.8PP25 transitional −4.3 1.0 0.6 0.1 2.0

material are suggestive of an important role played by juvenile gasexsolution at fragmentation. In addition, the largely variable lithiccontent (varying from 2 vol.% in GJLt, up to 45 vol.% in LiRLt), mainlyderived from the basement, suggests a deeply-rooted explosive activity,different from that typical of explosive basaltic eruptions (hawaiian lavafountaining, strombolian explosions) generally characterized by shallowlevel magma fragmentation (e.g. Keating et al., 2008), and more similarto that discussed for diatreme-like volcanism. In this latter type of activ-ity, fragmentation of magma is on the contrary normally assigned toexplosive magma–water interaction, with external water derived froman aquifer or surface water. Although fragmentation can occur at anylevel in the first few 100 s of metres of the crust or more (Valentineand White, 2012), it is often deep, explaining the general abundance oflithics in the deposits. In basaltic diatremes, the proportions of lithics

arity=4π (area/perimeter2); AR= aspect ratio; Juvenile and lithic content are given in) and loose in the matrix (free lithics); Vc = circular variance (Tran, 2007).

St. dev.AR

Juvenile(%)

Incl. Lithics(%)

Free Lithics(%)

Matrix Vc

0.9 77.2 5.1 3.5 19.30.5 73.9 7.3 5.5 20.60.6 77.7 1.3 0.3 22.00.8 73.2 2.1 13.7 13.1 0.51.3 67.0 4.0 2.3 30.7 0.41.6 67.2 5.7 3.4 29.4 0.60.9 77.3 4.9 1.7 21.0 0.30.9 33.6 2.2 40.3 26.11.4 49.5 1.4 32.5 18.01.1 75.0 2.2 5.7 19.3

11F. Mundula et al. / Journal of Volcanology and Geothermal Research 255 (2013) 1–14

vary from a few % to 80% (Ross andWhite, 2006;White and Ross, 2011),and the juvenile fragments can have any vesicularity, depending onthe moment of magma–water interaction with respect to the vesicula-tion history of the magma (Houghton and Wilson, 1989; White andRoss, 2011). In addition, globular lapilli have been also observed insome phreatomagmatic maar deposits (Sottili et al., 2009), and the gen-eral scarcity of textures unequivocally indicating phreatomagmaticmagma fragmentation could be also related to the fact that in manycases, maar volcanism is associated to an alternation of magmaticand hydromagmatic phases, in which the hydromagmatic phases areoften related to the initial activity. In this case, the lack of a clearhydromagmatic signature does not rule out the possibility that suchactivity did occur earlier in the volcano activity, as the studied depositsclearly represent the final stages of the diatreme formation. In conclu-sion, we suggest that syn-eruptive degassing and vesiculation certainlyoccurred during magma (possibly rapid) ascent, and that external fluidsdid interact at depth with magma playing an important role in control-ling the moment of magma fragmentation and in propelling the ejectionof the magmatic mixture.

5.3. Geometry of the diatremes: relationship with the host rocks

The collected structural data concur to define the relationshipsbetween host rocks and the diatremes, and to define the role of thehost rocks anisotropy on determining the final path of magma ascentand the diatreme margins stability. The shape and direction of themaximum elongation of the diatremes margins, the observed mutualrelationships between diatremes and host rocks cleavage, and theoccurrence of drag folds in the host rocks close to the diatrememargin,indicate that magma rose to the surface through fissures orientednearly parallel to the pre-existing foliation, suggesting that diatremespossibly formed under conditions of a low differential stress (Barnettand Lorig, 2007). The sigmoid shape of some of the diatremes(e.g. diatremes D and H), similar to jogs or large tension gashes, sug-gests that the diatreme system formed by reactivation of the Variscanstructures in a right lateral strike-slip regime. Anisotropy of the stressellipsoid and the difference between magma pressure and the mainstress are other factors that strongly influence diatreme geometry.During their development and growth, diatremes possibly experiencedmultiple events of over-pressure followed by strong under-pressureconditions, in relation to the balance between the rate of magma supplyand magma discharge during the eruption. Delaney et al. (1986) dem-onstrated, analysing the relationships between dikes and paleostressdirection, that in conditions of lowdifferential stress, if themagmapres-sure is the driving emplacement force, dikeswill utilize any pre-existingstructure oriented at a high angle with the tensile component, in agree-ment with the observed paraconconformity between diatrememarginsand the pre-existing anisotropy of host rock.

As the diatremes did not suffer important tectonic deformationafter emplacement, the orientation of pyroclasts and vesicles caninstead be used to define the strain ellipsoid during infilling. Imageanalysis of all the lapilli-bearing facies reveals a preferred oblateshape and sub-vertical orientation of the clasts, suggesting that defor-mation was related to a horizontal principal stress (σ1), possiblyresulting from the progressive infilling of the diatreme and/or thelate relaxation of the diatreme walls.

5.4. The juvenile-rich and lithic-rich facies; fluidization vs. ash alteration

Textural observations and quantitative analysis of the juvenile-richfacies highlighted the very good sorting (compared with the typicalrange of variation for pyroclastic deposits) and the coarse, nearly uni-form, average grain size of these deposits. The observed absence of afine-grainedmatrix in the deposits suggests that during or after depo-sition of the pyroclastic material some selective process operated inthe diatreme. This feature has been described in many other diatreme

massive deposits (e.g. Kurszlaukis and Lorenz, 2008; Porritt and Cas,2009), and it has been interpreted as the result of a strong fluidizationrelated to forced gas escape through the fragmentary, permeableinfilling, or as an effect of the intense alteration/substitution of thematrix by the late hydrothermal fluids. In the studied deposits, it isnot easy to clearly distinguish between the relative role of the twoprocesses; however, the presence in some of GJLt of empty inter-granular spaces, the general good sorting of the deposits and thevery fine grained mass of chlorite and calcite locally present in theintergranular spaces can be interpreted as an evidence of the circula-tion and partial filling by late-stage hydrothermal fluids on a depositwhich had already undergone an important fine-depletion possiblyrelated to the effect of a massive fluidization.

Although there is a general agreement for the occurrence of impor-tant processes of gas escape through the deposits inside a diatreme,the role of fluidization in generating and modifying these depositsis still not clearly understood. Fluidization conditions are reachedwhen gas flow velocity supports the weight of the bed of particles(e.g. Sparks et al., 2006; Walters et al., 2006; Gernon et al., 2009). Flu-idization generally produces well sorted and fines depleted deposits,following elutriation of fine particles and sinking of larger particlesinside the fluidized bed (Wilson, 1984; Walters et al., 2006). Thismeans that only a small range of particle sizes can be in mechanicalequilibrium with the gas phase for given conditions of gas densityand velocity. Conversely, fluidization has been invoked by someauthors also to explain poorly sorted-mixed facies (Gernon et al.,2009; White and Ross, 2011). Sparks et al. (2006) and Walters et al.(2006) assessed that highly sorted pelletal lapilli deposits in thediatreme zone of kimberlitic volcanoes are the result of a fluidizationprocess, and grain-size features and heterogeneities inside the diatremecan be related to the diatreme shape and gas velocity instability. Basedon these considerations, we suggest that elutriation related to gasfluidization was the main cause of the well sorted characteristic of thestudied deposits. Basing on the Ergun equation (Sparks et al., 2006),and varying the gas medium density in a wide range of H2O–CO2 ratiosat high T (400 °C) and P (10–100 MPa; Seitz and Blencoe, 1999),we cal-culated that gas velocity able to sustain particles in the observed rangeof diameters (3×10−3 to 3×10−2 m) falls in the range 0.1–3 m s−1,not far from the values of 1–10 m s−1 calculated by Sparks et al.(2006) for kimberlitic diatremes. Additional important informationabout the role of fluidization derives from the mutual relationshipsobserved in the two subfacies of LiRLt in diatreme D. Here, the wellsorted LiRLt generally occurs along the margins of the diatreme,while the poorly sorted LiRLt occurs in its central zone. While thewell sorted facies shows textural features consistent with a fluidiza-tion process, compatibility between the poorly sorted LiRLt andfluidization process needs more explanations. Walters et al. (2006)demonstrated, through analogue experiments, that the gas velocityprofile inside a conduit with an inverted cone geometry has a gauss-ian shape. They suggested that large or denser particles, sinking atthe margins of the pipe where the gas velocities are low, can bere-entrained at lower levels along the centreline where gas velocityis high, favouring the mixing-up of particles with a large size range.This capability of mixing together particles of different size and den-sity is made easier by the heterogeneous nature of the bubbly fluid-ized regime (Gernon et al., 2008). These considerations, mainlybased on experimental data and field observations, suggest that thepresence of poorly sorted deposits in the central zone of diatremeD could be consistent with bubbly fluidization conditions in an up-ward diverging pipe, possibly at the passage between the root zoneand the diatreme zone.

5.5. GJLt and COH: the role of welding

Welding is a term applied to the sintering process of juvenile frag-ments occurring at temperature above the glass transition, during

12 F. Mundula et al. / Journal of Volcanology and Geothermal Research 255 (2013) 1–14

(mainly by agglutination) or following deposition (for the effect ofcompaction at high temperature conditions). Incipient welding up tocomplete sintering characterizes some of the observed facies.Weldingin volcanology has been mainly studied on ignimbrites (e.g. Smith,1960; Sheridan and Ragan, 1976; Branney and Kokelaar, 1992;Grunder et al., 2005), proximal fall deposits (e.g. Sparks and Wright,1979;Wolff andWright, 1981; Carey et al., 2008) and spatter deposits(e.g. Sumner et al., 2005). Magma temperature, composition andwater content, together with the rate of deposition, the cooling rateand the loading pressure are the main parameters controlling thewelding of pyroclastic deposits. Although welding generally occursunder conditions of high particle concentration, experimental andnumerical modelling of the transport and emplacement process ofpyroclastic flows (Freundt, 1998; Freundt, 1999) demonstrated thatalso the transport of particles in a fluidized state is consistent withthe development of welding.

Physical conditions in the diatremes during their formation are dif-ficult to constrain but we can reasonably suppose, on the basis of theobserved facies, that the studied deposits were emplaced at a depth inthe order of hundreds of metres (possibly between 5 and 30 MPa).Minimum welding pressures around 5 MPa were also experimentallyfound by Grunder et al. (2005) for producing welding at temperaturesclose to the temperature of glass transition.

Sparks et al. (2006) invoked a high grade welding process as apossible explanation of some coherent facies in kimberlitic diatremes,suggesting that the presence of lithics and of ghosts of juvenile frag-ments testify to the pyroclastic origin of this facies. Textural observa-tions on COH revealed the presence of mm to cm-sized lithics, and thepresence of lapilli ghosts evidenced by a concentric or radial arrange-ment of microliths, and pinching inward contacts. Analogue texturalfeatures were recognized in some transitional facies between boththe juvenile-rich lapilli tuff facies and COH, and in some clusters oflapilli of GJLt. We interpret these textural features as the product ofwelding processes developed inside the diatreme, and responsible,in extreme cases, of clasts agglutination to form coherent material.

5.6. The nature of AJLt

AJLt represents a special case of lapilli-supported facies, mainlyarranged along the margins of the studied diatremes, and character-ized by an anisotropic texture, given by deformed clasts with jaggedoutlines. Differently from clasts from GJLt, those from AJLt have anirregular, elongated shape with outward cuspate and inward convexoutlines. Clasts of similar shape are described in some peperite-likefacies developed at the interface between diatreme-filling depositsand wet host sediments (Junqueira-Brod et al., 2005a, 2005b). De-spite this similarity, while peperite textures are typically observedas a gradual passage between the intrusive body and the host, AJLtalways shows a sharp subvertical contact with host rocks.

The time scale (Τref) needed to a magma fragment to regain, underthe action of surface forces, a smooth, nearly spherical shape of radiusR can be taken as

Τref ¼ R�μ �σ–1

(with μ= viscosity and σ= suface tension). Assuming values for vis-cosity (10–103 Pa s) and surface tension (3∗10−1 N m−1) typical ofbasaltic magma (McBirney and Murase, 1973), and particles in therange between 10−3 and 10−2 m, it follows that the time needed toassume a completely relaxed shape is in the order of tens of seconds.This suggest that the clasts from AJLt, with the same composition ofthose of GJLt, in order to maintain an unrelaxed shape must have un-dergone a very rapid quenching, which largely increased their viscos-ity immediately after deposition. While in the inner zone of thediatreme the juvenile fragments of GJLt had sufficient time to reshapetheir surface in response to surface tension, at the diatreme margins,

juvenile fragments of AJLt rapidly cooled without substantial modifi-cation of their surface during the first phases ofmagma ascent througha cold environment. The average oblate shape observed on the clastsof both facies is suggestive of a general process of flattening ratherthan extension of the clasts, possibly occurring over longer timesand related to the concomitance of the differential rheological behav-iour and cooling of the inner vs. the outer portions of the pipe, and thecontinuous infilling of fragmented material of the pipe during itsformation.

5.7. A possible model of diatreme formation

Basing on field observation of the studied diatremes, we proposehere a multistage model for the interpretation of the observed faciesvariability inside these bodies, which could also represent a generalmodel for other similar deposits.

a) Diatremes formed in the presence of a small differential stress,mainly following pre-existing anisotropies (in this case the Variscancleavage) in the host rocks. During diatreme formation, magma riselocally deformed the pre-existing rocks. The abundance of basementfragments within the lithics suggests that fragmentation occurreddeep in the conduit, as is typical in diatreme-type volcanism. Theincorporation of large, deeply-derived lithic blocks suggests a veryfast ascent for the magma–gas mixture. The presence of diffusewelding and the globular shape of some fragments, together withthe vesicularity of the juvenile material, suggest that magmadegassing had a prominent role in the evolution of the diatreme, atleast during the late stages, responsible for its final infilling.We can-not completely exclude thatmagma–water interaction, even thoughnot clearly evident in the products, could have had also a role in theevolution of the diatreme.

b) After the main explosive phase, strong fluidization resulted in anefficient winnowing out of the fine material and in the formationof well sorted, open textured lithofacies in the pipe filling materialal the level of the passage between the root and the diatremezones. The geometry of gas fluidization inside the diatreme locallyconcentrated the largest lithic clasts in the axial zone of the diatreme.Portions of pyroclastics materials along the axial portion of thediatreme were locally able to agglutinate and weld together.

c) The infilling of the axial portions of the diatreme, together with thepossible final intrusion of unfragmented magma tongues and thefinal relaxation of the diatreme walls, contributed to the develop-ment of a local flattening of the clasts in the marginal zone of thediatremes.

5.8. Comparisons with other diatremes

A systematic comparison of the studied diatremes with othersubsurface unbedded volcaniclastic deposits of similar basaltic andultrabasic diatremic structures is useful to highlight similarities anddifferences in the textural features, facies association and facies archi-tecture. The absence of layered facies, the very limited presence ofCOH and the dominance of unbedded, clast supported volcaniclasticfacies containing a variable amount of lithics (GJLt, AJLt and LiRLt)represent the distinctive features of the studied diatremes and are de-scribed both for non-kimberlitic (e.g. the Hopi Buttes volcanic fieldand the Coombs Hills; White and Ross, 2011) and kimberlitic dia-tremes (e.g. Sparks et al., 2006 and references therein). The observedfacies association, together with the facies architecture characterizedby nearly vertical, mainly progressive lateral facies transitions, andwith the approximately concentric distribution of lithofacies, suggestthat in the deep part of the diatreme the process of infilling was pro-gressive and rapid in time. Despite this general similarity (see Whiteand Ross, 2011), basaltic and kimberlite diatremes have somewhatdifferent textural features. The main component of the unbedded

13F. Mundula et al. / Journal of Volcanology and Geothermal Research 255 (2013) 1–14

volcaniclastic facies of basaltic diatremes consists of angular to ame-boid juvenile fragments and lithics. Deposits are poorly sorted andare typically interpreted as the result of phreatomagmatic frag-mentation (e.g. Ross et al., 2005; Ross and White, 2006; Némethand Martin, 2007). Conversely, the main components of “massivevolcaniclastic kimberlites” (Sparks et al., 2006), of “tuffisitic kimberlite”(e.g. Field and Scott Smith, 1999) and of other ultrabasic diatreme(e.g. Junqueira-Brod et al., 2005a, 2005b) are represented by moder-ately to well sorted, globular juvenile fragments commonly cored byphenocrysts (often described as pelletal lapilli; e.g. Skinner andMarsh, 2004; Hetman, 2008; Brown et al., 2009), while deep-derivedxenoliths are generally very poorly sorted. The occurrence of pelletallapilli has been associated to fragmentation processes driven byjuvenile gas exsolution (Field and Scott Smith, 1999; Sparks et al.,2006; Cas et al., 2008). The lithofacies described for the studieddiatremes suggest an origin from processes more similar to thosedescribed for kimberlitic and ultrabasic volcanism rather than moreclassical, maar-type, basaltic volcanism.

6. Conclusions

The study of diatremes increased in recent years, especially con-cerning kimberlite volcanism and basaltic maar–diatreme structures.Despite the increasingly large amount of available geological data, theprocesses which dominate the formation of diatremes still remain ob-ject of debate, as testified by the multiple and sometimes conflictingproposed models. The alkali basaltic diatreme deposits of SW Sardiniadescribed here share many textural features with classical kimberlitediatremes and other ultramafic diatremes. Facies analysis remainsan important descriptive tool for collecting data needed to discussthe main processes responsible of their genesis, and to highlightdifferences and similarities among the deposits of different volcanicsystems and composition. Four main lithofacies have been distin-guished, in which the main features allow to make inferences aboutgeneral processes like fluidization, welding, and cooling modalitiesand timing. Differently frommostmaar-type volcanoes, inwhich the de-posits have clear hydromagmatic signatures, collected data suggest thatdeep magma degassing had a prominent role in the explosivity of thesediatremes, at least during their final stages. Fragmentation of the alkalibasaltic magma occurred at deep levels (possibly hundreds of metresbelow the ground surface), differently from typical explosive basalticsystems, dominated by hawaiian and strombolian activity commonlycharacterized by shallow level fragmentation.We conclude that texturalfeatures, facies association and facies architecture of the studied depositsare more similar to kimberlitic and other ultramafic diatremes, provid-ing a good basaltic analogous for kimberlite eruptions.

Acknowledgements

Researchwas done thanks to theUniversity of Cagliari funding to RC.FM acknowledges the Fondazione Banco di Sardegna for funding hisresearch fellowship. The critical revision of P.-S. Ross and J. Taddeucciis greatly acknowledged. The authors are grateful to P. Comida forsupporting the field activity.

References

Barnett, W.P., Lorig, L., 2007. A model for stress-controlled pipe growth. Journal ofVolcanology and Geothermal Research 159, 108–125.

Beccaluva, L., Leone, F., Maccioni, L., Macciotta, G., 1981. Petrology and tectonic settingof the Paleozoic basic rocks from Iglesiente-Sulcis (Sardinia, Italy). Neues Jahrbuchfür Mineralogie 140, 184–201.

Branney, M.J., Kokelaar, P., 1992. A reappraisal of ignimbrite emplacement: progressiveaggradation and changes fromparticulate to non-particulateflow during emplacementof high-grade ignimbrite. Bulletin of Volcanology 54, 504–520.

Brown, R.J., Tait, M., Field, M., Sparks, R.S.J., 2009. Geology of a complex kimberlite pipe(K2 pipe, VenetiaMine, South Africa): insights into diatremeprocesses during explosive

ultrabasic eruptions. Bulletin of Volcanology 71, 95–112. http://dx.doi.org/10.1007/s00445-008-0211-4.

Carey, R.J., Houghton, B.F., Thordarson, T., 2008. Contrasting styles of welding observed intheproximal Askja 1875 eruption deposits I: Regionalwelding. Journal of Volcanologyand Geothermal Research 171, 1–19.

Carmignani, L., Barca, S., Oggiano, G., Pertusati, P.C., Salvadori, I., Conti, P., Eltrudis, A.,Funedda, A., Pasci, S., 2001. Note illustrative della Carta Geologica della Sardegnaa scala 1:200.000. Memorie descrittive Carta Geolologica Italiana 60, pp. 283.

Carrigan, C.R., 2000. Plumbing systems. In: Sigurdsson, H., Houghton, B., McNutt, S.R.,Rymer, H., Stix, J. (Eds.), Encyclopedia of Volcanoes. Academic Press, San Diego,pp. 219–235.

Cas, R., Wright, J.W., 1987. Volcanic Successions: Modern and Ancient. Allen and Unwin,London (528 pp.).

Cas, R.A.F., Porritt, L., Pittari, A., Hayman, P., 2008. A new approach to kimberlite faciesterminology using a revised general approach to the nomenclature of all volcanicrocks and deposits: descriptive to genetic. Journal of Volcanology and GeothermalResearch 174, 226–240.

Casini, L., Funedda, A., Oggiano, G., 2010. A balanced foreland–hinterland deformationmodel for the SouthernVariscan belt of Sardinia, Italy. Geological Journal 45, 634–649.http://dx.doi.org/10.1002/gj.1208.

Cortesogno, L., Cassinis, G., Dallagiovanna, G., Gaggero, L., Oggiano, G., Ronchi, A., Seno,S., Vanossi, M., 1998. The Variscan post-collisional volcanism in Late Carboniferous–Permian sequences of Ligurian Alps, Southern Alps and Sardinia (Italy): a synthesis.Lithos 45, 305–328.

Delaney, P.T., Pollard, D.D., Ziony, J.I., McKee, E.H., 1986. Field relationships between dikesand joints: emplacement process and paleostress analysis. Journal of GeophysicalResearch 91, 4920–4938.

Downes, P.J., Ferguson, D., Griffin, B.J., 2007. Volcanology of the Aries micaceouskimberlite, central Kimberley Basin, Western Australia. Journal of Volcanologyand Geothermal Research 159, 85–107.

Field, M., Scott Smith, B.H., 1999. Contrasting geology and near-surface emplacementof kimberlite pipes in southern Africa and Canada. In: Gurney, J.J., Gurney, J.L.,Pascoe, M.D., Richardson, S.H. (Eds.), Proceedings of the VIIth InternationalKimberlite Conference. Red Roof Design, Cape Town, South Africa, pp. 214–237.

Fisher, R.V., 1966. Rocks composed of volcanic fragments and their classification. Earth-Science Reviews 1, 287–298. http://dx.doi.org/10.1016/0012-8252(66), 90010-9.

Fisher, R.V., Schmincke, H.-U., 1984. Pyroclastic Rocks. Springer–Verlag, Berlin (472 pp.).Freundt, A., 1998. The formation of high-grade ignimbrites, I: Experiments on high-

and low-concentration transport systems containing sticky particles. Bulletin ofVolcanology 59, 414–435.

Freundt, A., 1999. Formation of high-grade ignimbrites Part II. A pyroclastic suspen-sion current model with implications also for low-grade ignimbrites. Bulletin ofVolcanology 60, 545–567.

Funedda, A., 2009. Foreland- and hinterland-verging structures in fold-and-thrust belt:an example from the Variscan foreland of Sardinia. International Journal of EarthSciences 98, 1625–1642. http://dx.doi.org/10.1007/s00531-008-0327-y.

Gernon, T.M., Gilbertson, M.A., Sparks, R.S.J., Field, M., 2008. Gas-fluidisation in anexperimental tapered bed: insights into processes in diverging volcanic conduits.Journal of Volcanology and Geothermal Research 174, 49–56.

Gernon, T.M., Gilbertson, M.A., Sparks, R.S.J., Field, M., 2009. The role of gas-fluidisationin the formation of massive volcaniclastic kimberlite. Lithos 112, 439–451.

Giordano, D., Russell, J.K., Dingwell, D.B., 2008. Viscosity of magmatic liquids: a model.Earth and Planetary Science Letters 271, 123–134.

Goto, Y., Nakada, S., Kurokawa, M., Shimano, T., Sugimoto, T., Sakuma, S., Hoshizumi, H.,Yoshimoto, M., Uto, K., 2008. Character and origin of lithofacies in the conduit ofUnzen volcano, Japan. Journal of Volcanology and Geothermal Research 175, 45–59.

Grunder, A.L., Laporte, D., Druitt, T.H., 2005. Experimental and textural investigation ofwelding: effect of compaction, sintering, and vapour phase crystallization in therhyolitic Rattlesnake Tuff. Journal of Volcanology and Geothermal Research 142,89–104.

Hetman, C.M., 2008. Tuffisitic Kimberlite (TK): a Canadian perspective on a distinctivetextural variety of kimberlite. Journal of Volcanology and Geothermal Research174, 57–67.

Higgins, M.D., 2000. Measurement of crystal size distributions. American Mineralogist85, 1105–1116.

Houghton, B.F., Wilson, C.J.N., 1989. A vesicularity index for pyroclastic deposits. Bulletinof Volcanology 51, 451–462.

Inman, D.L., 1952. Measures for describing the size distribution of sediments. Journal ofSedimentary Petrology 22, 125–145.

Junqueira-Brod, T.C., Brod, J.A., Gaspar, J.C., Jost, H., 2005a. Kamafugitic diatremes: faciescharacterization and genesis — examples from the Goia's alkaline province, Brazil.Lithos 76, 261–282.

Junqueira-Brod, T.C., Gaspar, J.C., Brod, J.A., Kafino, C.V., 2005b. Kamafugitic diatremes:their textures and field relationshipswith examples from theGoia´s alkaline province,Brazil. Journal of South American Earth Sciences 18, 337–353.

Kano, K., Matsuura, H., Yamauchi, S., 1997. Miocene rhyolitic welded tuff infilling afunnel-shaped eruption conduit Shiotani, southeast of Matsue, SW Japan. Bulletinof Volcanology 59, 125–135.

Keating, G.N., Valentine, G.A., Krier, D.J., Perry, F.V., 2008. Shallow plumbing systemsfor small-volume basaltic volcanoes. Bulletin of Volcanology 70, 563–582.http://dx.doi.org/10.1007/s00445-007-0154-1.

Kurszlaukis, S., Lorenz, V., 2008. Formation of “Tuffisitic Kimberlites” by phreatomagmaticprocesses. Journal of Volcanology and Geothermal Research 174, 68–80.

Leone, F., Hamman, W., Laske, R., Serpagli, E., Villas, E., 1991. Lithostratigaphic unitsand biostratigaphy of the post-sardic Ordovician Sequence in south-west Sardinia.Bollettino della Societa Paleontologica Italiana 30, 201–235.

14 F. Mundula et al. / Journal of Volcanology and Geothermal Research 255 (2013) 1–14

Lorenz, V., Kurszlaukis, S., 2007. Root zone processes in the phreatomagmatic pipe em-placement model and consequences for the evolution of maar–diatreme volcanoes.Journal of Volcanology and Geothermal Research 159, 4–33.

Martin, U., Németh, K., Lorenz, V., White, J.D.L., 2007. Introduction: maar–diatremevolcanism. Journal of Volcanology and Geothermal Research 159, 1–3.

McBirney, T., Murase, A., 1973. Properties of some common igneous rocks and theirmelts at high temperatures. GSA Bulletin 84 (11), 3563–3592.

Meschede, M., 1986. A method of discriminating between different types of mid-oceanridge basalts and continental tholeiiteswith the Nb–Zr–Y diagram. Chemical Geology56, 207–218.

Mundula, F., Cioni, R., Rizzo, R., 2009. A simplified scheme for the description of texturalfeatures in welded ignimbrites: the example of San Pietro Island (Sardinia; Italy).Italian Journal of Geosciences 128, 615–627.

Németh, K., Martin, U., 2007. Shallow sill and dyke complex in western Hungary as apossible feeding system of phreatomagmatic volcanoes in “soft-rock” environment.Journal of Volcanology and Geothermal Research 159, 138–152.

Németh, K., Martin, U., Haller, M.J., Alric, V.L., 2007. Cenozoic diatreme field in Chubut(Argentina) as evidence of phreatomagmatic volcanism accompanied with extensivePatagonian plateau basalt volcanism. Episodes 30, 217–223.

Pasci, S., Carmignani, L., Cincotti, F., Pisanu, G., Sale, V., 2012. Note illustrative del F°564 –

Carbonia:Memorie Descrittive della Carta Geologica d'Italia. ISPRA–Servizio Geologicod'Italia.

Porritt, L.A., Cas, R.A.F., 2009. Reconstruction of a kimberlite eruption, using an integrated vol-canological, geochemical and numerical approach: a case study of the Fox Kimberlite,NWT, Canada. Journal of Volcanology and Geothermal Research 179, 241–264.

Rasband, W.S., 1997. ImageJ. U. S. National Institutes of Health, Bethesda, Maryland,USA (http://rsb.info.nih.gov/ij/).

Riehle, J.R., Miller, T.F., Bailey, R.A., 1995. Cooling, degassing and compaction of rhyoliticash flow tuffs; a computational model. Bulletin of Volcanology 57, 319–336.

Ross, P.-S., White, J.D.L., 2006. Debris jets in continental phreatomagmatic volcanoes:a field study of their subterranean deposits in the Coombs Hills vent complex,Antarctica. Journal of Volcanology and Geothermal Research 149, 62–84.

Ross, P.-S., White, J.D.L., 2012. Quantification of vesicle characteristics in some diatreme-filling deposits, and the explosivity levels of magma–water interactions withindiatremes. Journal of Volcanology and Geothermal Research 245–246, 55–67.

Ross, P.-S., Ukstins Peate, I., McClintock, M.K., Xu, Y.G., Skilling, I.P., White, J.D.L.,Houghton, B.F., 2005. Mafic volcaniclastic deposits in flood basalt provinces: areview. Journal of Volcanology and Geothermal Research 145, 281–314.

Ross, P.-S., McClintock, M.K., White, J.D.L., 2008. Geological evolution of the Coombs-Allan Hills area, Ferrar large igneous province, Antarctica: debris avalanches,mafic pyroclastic density currents, phreatocauldrons. Journal of Volcanology andGeothermal Research 172, 38–60.

Seghedi, I., Maicher, D., Kurszlaukis, S., 2009. Volcanology of Tuzo pipe (Gahcho Kuécluster) — root-diatreme processes re-interpreted. Lithos 112, 553–565.

Seitz, J.C., Blencoe, J.G., 1999. The CO2–H2O system. I. Experimental determination ofvolumetric properties at 400 °C, 10–100 MPa. Geochimica et Cosmochimica Acta63, 1559–1569.

Sheridan,M.F., Ragan, D.M., 1976. Compaction of ash-flow tuffs. In: Chilingarian, G.V.,Wolf,K.H. (Eds.), Compaction of Coarse-Grained Sediments, vol. II. Elsevier, Amsterdam,Netherlands, pp. 677–717.

Skinner, E.M.W., Marsh, J.S., 2004. Distinct kimberlite pipe classes with contrastingeruption processes. In: Mitchell, et al. (Ed.), 8th International Kimberlite Conference:Selected Papers, vol. 1, pp. 183–200.

Smith, R.L., 1960. Zones and zonal variations inwelded ash flows. United States GeologicalSurvey Professional Paper 354–F, 149–159.

Soriano, C., Galindo, I., Martí, J., Wolff, J., 2006. Conduit-vent structures and relatedproximal deposits in the Las Cañadas caldera, Tenerife, Canary Islands. Bulletin ofVolcanology 69, 217–231. http://dx.doi.org/10.1007/s00445-006-0069-2.

Sottili, G., Taddeucci, J., Palladino, D.M., Gaeta, M., Scarlato, P., Ventura, G., 2009. Sub-surface dynamics and eruptive styles of maars in the Colli Albani Volcanic District,Central Italy. Journal of Volcanology and Geothermal Research 180, 189–202.

Sparks, R.S.J., Wright, J.V., 1979. Welded air-fall tuffs. Geological Society of AmericaSpecial Papers 180, 155–166.

Sparks, R.S.J., Baker, L., Brown, R.J., Field, M., Schumacher, J., Stripp, G., Walters, A., 2006.Dynamical constraints on kimberlite volcanism. Journal of Volcanology andGeothermalResearch 155, 18–48.

Sumner, J.M., Blake, S., Matela, R.J., Wolff, J.A., 2005. Spatter. Journal of Volcanology andGeothermal Research 142, 49–65.

Tran, N.H., 2007. Fracture orientation characterization: minimizing statistical modellingerrors. Computational Statistics and Data Analysis 51, 3187–3196.

Traversa, G., Vaccaro, C., 1992. REE distribution in the Late Hercynian dykes fromSardinia. IGCP N.276. Newsletter 5, 215–226.

Valentine, G.A., White, J.D.L., 2012. Revised conceptual model for maar–diatremes:subsurface processes, energetics, and eruptive products. Geology 40, 1111–1114.http://dx.doi.org/10.1130/G33411.1.

Walters, A., Phillips, J., Brown, R., Field, M., Sparks, R.S.J., 2006. The role of fluidisationprocesses in the formation of volcaniclastic kimberlite: grain size observationsand experimental investigation. Journal of Volcanology and Geothermal Research155, 119–137.

White, J.D.L., Houghton, B.F., 2006. Primary volcaniclastic rocks. Geological Society ofAmerica 34, 677–680. http://dx.doi.org/10.1130/G22346.1.

White, J.D.L., McClintock, M.K., 2001. Immense vent complex marks flood-basalt eruptionin a wet, failed rift: Coombs Hills, Antarctica. Geology 29, 935–938.

White, J.D.L., Ross, P.S., 2011. Maar–diatreme volcanoes: a review. Journal of Volcanologyand Geothermal Research 201, 1–29.

Wilson, C.J.N., 1984. The role of fluidization in the emplacement of pyroclastic flows: 2.Experimental results and their interpretation. Journal of Volcanology and GeothermalResearch 20, 55–84.

Winchester, J.A., Floyd, P.A., 1977. Geochemical discrimination of different magmaseries and their differentiation products using immobile elements. Chemical Geology20, 325–343.

Wohletz, K.H., Sheridan, M.F., Brown, W.K., 1989. Particle size-distributions and theSequential Fragmentation/Transport theory applied to volcanic ash. Journal ofVolcanology and Geothermal Research 94, 15703–15721.

Wolff, J.A., Wright, J.V., 1981. Formation of the Green Tuff, Pantelleria. Bulletin ofVolcanology 44, 681–690.