Geomorphology
Petite-Terre is a 16 km2 island located in Mayotte lagoon (Indian Ocean) and mainly composed by Quaternary volcanic products including mafic scoria cones and phonolitic explosive edifices previously interpreted as maars (Debeuf, 2004; Nehlig et al., 2013).
Figure 2a show a geomorphological map of both Petite-Terre Island and the eastern tip of Grande-Terre Island presenting the repartition of the volcanic edifices. Even though Petite-Terre volcanism is young, the location of the island on the edge of a barrier reef led to intense coastal erosion, which partially destroyed the eastern flanks of the maars.
The numerous mafic scoria cones represent the oldest manifestations of the volcanic activity in the area. In total, 17 Pleistocene scoria cones were identified: three on Grande-Terre, five in the lagoon and nine on Petite-Terre (Fig. 2a.). These cones, basanitic to tephritic in composition (Nougier et al., 1986; Debeuf, 2004; Nehlig et al., 2013; Pelleter et al., 2014), are estimated to be formed during or before the last glacial era (> 10 ka BP) while the lagoon of Mayotte was dried out and above the sea level (Zinke et al., 2003). These monogenetic cones experienced pronounced erosion due to their littoral position and thus lack well-defined structures like crater rim and associated lava flow and are old enough to be covered by a lateritic paleosoil. On Petite-Terre, some cones are partially (Mirandole, a small scoria cone north to Mirandole named Mirandolino in this study, and Totorossa) or totally (Ha Shiwawa and Sandravouangué) covered by the more recent phonolitic pyroclastic deposits. These last, totally covered cones can only be seen thanks to coastal erosion (Ha Shiwawa cone on the Airport beach) or construction sites (Sandravouangué). The currently visible part of these scoria cones thus likely represents a small part of their total extent. Scoria cones deposits contain very few accidental clasts. Average phenocryst content (with the notable exception of La Ferme) is small.
The phonolitic pyroclastic deposits originate from the different explosive edifices located on the east coast of the island of Petite-Terre. Phonolitic pumice clasts found in sedimentary cores sampled in the lagoon (Zinke et al., 2003a) suggest ages between 7 and 4 ka BP based on 14C dating of the sediment cores, while cores from Dziani lake indicate no volcanic activity since at least 6.2 ka BP but maybe even older since drillings from Javovic (2020) didn’t not reach the pyroclastic deposit filling. However, no compositional data are available for the phonolitic pumices from Zinke et al. (2003b), which prevents to tied it with confidence to Petite-Terre explosive activity, especially in the context of the recent unveiling of young pyroclastic phonolitic edifices close to Petite-Terre along the submarine volcanic chain (Berthod et al., 2021; Puzenat et al., 2023; Thivet et al., 2023)
Dziani maar (Fig. 2c), the northernmost, culminates at 101 m a.s.l. and encloses an up to 18 m deep green anoxic lake within its crater (Cadeau et al., 2023). The overall morphology of the edifice is preserved despite some erosion of the east side facing the open ocean with ridges of the crater being sharp suggesting an almost pristine conservation state. The crater has a mean diameter of 1074 m and a depth of 102 m The morphology of the edifice displays smooth external slopes along with a sharp crater ridge and steep internal slopes suggestive of a young age. On the crater floor can be observed two structures (Fig. 2) which could be terraces or non-eruption related collapses as proposed by Nehlig et al.(2013). The crater itself exhibits a lobate rim whose origin could be from multiple explosions during a single or several eruptions and/or crater rim collapses. The current floor of the crater is unlikely to reflect its original depth since syn-eruptive pyroclastic deposits and post-eruptive sedimentation, as well as lacustrine sedimentation or collapses, generally fills the original crater (Lorenz, 1986, 1968; Valentine et al., 2017; White and Ross, 2011), however, considering that Dziani lake is 18 m deep, the true crater depth must at least be 119 m deep.
To the south of the Dziani maar is located a cluster of craters whose characteristics can be found in Table 3. This ensemble is composed of three edifices: Moya crater to the north (Fig. 2e), La Vigie crater to the south and Central crater in the middle (Fig. 2d), nested within the two. The highest of all is La Vigie, culminating at 201 m a.s.l. while Moya is the widest with a mean diameter of 911 m. All those edifices are deeply cut by the coastline erosion and have lost a significant part of their rim but their remaining half allow to have an overall idea of their original morphologies.
Table 3
Geomorphological characteristics of the Petite-Terre explosive edifices. Notice that only Dziani crater is in a pristine state, the other ones being cut by the coastline the measurement were done by extrapolating rather circular crater rims. For parameter definition see Graettinger (2018). For Grande-Terre edifices, Hmin and Hmax were measured on the rim of each which are located on the slopes of the old shield which could lead to an overestimation of their true depth.
|
Petite Terre
|
Grande-Terre
|
|
Dziani
|
Central
|
Vigie
|
Moya
|
Kavani
|
Kavéni
|
Dmin
|
1024
|
520
|
911
|
866
|
923
|
1784
|
Dmax
|
1171
|
552
|
1007
|
1037
|
1012
|
2112
|
Dmean
|
1074
|
498
|
843
|
911
|
928
|
1919
|
A
|
905395
|
194879
|
557495
|
651786
|
675872
|
2892645
|
P
|
3527
|
1605
|
2682
|
2942
|
2959
|
6197
|
Hmax
|
101
|
114
|
201
|
136
|
94
|
101
|
Hmin
|
-19
|
-1
|
46
|
16
|
15
|
-3
|
Htotal
|
120
|
115
|
155
|
120
|
79
|
104
|
A.R.
|
0.87
|
0.94
|
0.90
|
0.84
|
0.91
|
0.84
|
El.
|
0.84
|
0.81
|
0.70
|
0.77
|
0.84
|
0.83
|
I.C.
|
0.91
|
0.95
|
0.97
|
0.95
|
0.97
|
0.95
|
D.D
|
0.10
|
0.21
|
0.15
|
0.12
|
0.09
|
0.05
|
Type
|
Tuff-ring
|
Tuff-cone
|
Tuff-ring/cone
|
Tuff-ring
|
(Maar ?)
|
(Maar ?)
|
Kawéni et Kavani are the two basic (basaltic ?) maars present on Grande Terre (Fig. 2a). Even though their ages remain poorly constrained (> 150 ka, Debeuf, 2004) their morphology, much seems more eroded, suggest an older activity than Petite Terre (Supplementary S2.1). Kawéni is the largest of all Mayotte maars with a mean diameter of 1920 m while Kavani is smaller with a rim diameter of 923 m. Their deposits are welded/indurated but those were only observed on Grande Terre but not on Petite Terre which is the focus here and therefore we excluded them from this study. Crosscutting relationships show that La Vigie is the oldest of all Petite-Terre phonolitic edifices, its rim is cut by Moya crater itself cut by Central crater. At Papani beach the coastal cliff exposes the relationship between the deposits from Moya crater and the overlying Dziani crater rim. These relationships are consistent with the observed morphologies (Fig. 2a) where La Vigie and Moya crater display dull crater ridges and incised external slopes while Central and Dziani crater have very sharp crater ridges and smoother outer slopes. Hence, in agreement with previous work (Nehlig et al., 2013), the following relative chronology, from the older to the younger, is retained for the emplacement of Petite Terre’s explosive edifices:
1. La Vigie
2. Moya
3. Central/Dziani
The MaarVLS 2.0 database (Graettinger, 2018) compiles the geomorphological characteristics of more than 450 well-conserved maars and tuff rings (> 75% of the crater rim remaining) from various geodynamic settings, chemistry and ages. On Petite-Terre, the edifices are incomplete due to coastal erosion and therefore their shapes were extrapolated from their remnants, their geomorphological characteristics are reported in Table 3. A comparison of Mayotte phonolitic edifices with this database shows that, in spite of their unusual composition (phonolite) and the widespread occurrence of mantle xenoliths, they fall in the MaarVLS database fields where most edifices have a dominantly mafic composition and are not exotic endmembers in terms of morphology (Fig. 3).
Despite historically being classified as maars, the morphometric characteristics of Petite-Terre explosive edifices resumed in Table 3 allow to refine their classification as maars, tuff rings or tuff cones.
Maars are defined by a crater floor cut through the pre-eruptive surface. They possess a low-lying, almost flat bedded (< 12°), rim which rarely exceeds 30 m in height (Lorenz, 1973; Ollier, 1967; White and Ross, 2011). Their deposits mostly consist of laminated, fine-grained deposits with numerous bedforms emplaced by turbulent and low concentration base surges of super-heated steam (Vespermann et al., 2000; Wohletz and Sheridan, 1983). These deposits are also very rich in non-juvenile material from the country-rock, ejected during explosions forming the maar, which can represent up to 90 wt.% while the juvenile material is usually dense and low- or non-vesiculated (Lorenz 1985; Vespermann et al. 2000; White et Ross 2011; Ollier 1967).
Tuff rings share a lot of characteristics with maars (Lorenz 1986), except that their crater floor is located at or above the pre-eruptive surface. Their rim can be up to 50 m high with low-angle side and bedding (< 25°) with deposits similar to maars but where juvenile material is dominating (< 25 wt.% of non-juvenile), fallout beds can also be observed (Wohletz et Sheridan 1983; Lorenz 1985; 1986; Vespermann et al. 2000; White et Ross 2011; Brož et Németh 2015).
Tuff cones also have crater floor above the pre-eruptive surface but differ from their cousins by their steep slopes and bedding (> 25°) and high rims (usually 100 m and up to 330 m). Their deposits are massive to weakly stratified with only very minor percent of non-juvenile material (< 5 wt.%) (Sheridan and Wohletz, 1983; Vespermann et al., 2000; J.D.L. White and Ross, 2011; Wohletz and Sheridan, 1983).
Diatremes are more or less developed explosive vent structures underlying maars and tuff rings while tuff cones lack them. They are funnel-shaped conduits filled with pyroclastic deposits where multiple explosions take place during the eruption (Lorenz, 1973; Valentine et al., 2017). Diatremes deepen by successive explosions. The deepest diatremes are associated to maars (up to 2.5 km deep) while the shallowest (≤ 300 m) are associated with tuff rings (Lorenz 1985; 1986).
For Petite-Terre, even though the current floor of the craters is possibly covered by erosion and reworking processes, the relatively fresh morphology of the rims indicate that post-eruptive in-fill is still limited (Fig. 2). This, along with the absence of the underlying terrain on craters walls suggest that these edifices are not, in fact, maars and are either tuff rings or tuff cones because of their crater bottom doesn’t seem to lay under the preexisting topography level. Depth/Diameter ratio is also quite variable. Another argument supporting this idea is the thickness of the rims which are at least 100 m thick and even maybe more since drillings in the area reported pyroclastic deposits until 15 m b.s.l. underlain by basaltic beach sands inside Moya crater which could be interpreted as the crater floor (BSS002PNKN drill-holes, BRGM, 2023). Such thicknesses are also more typical of tuff rings and tuff cones. Finally, the external slopes of the different edifice rims are > 12°, which is the highest limit for maars, and even sometimes > 25° in the case of Central crater. Based on these elements we conclude that the pyroclastic phonolitic cones of Petite-Terre are not maars but tuff rings for Dziani and Moya craters and a tuff cone for Central crater. La Vigie shares characteristics of both type of edifices and can be seen either as a tuff ring or a tuff cone.
Facies descriptions and their interpretations
The four phonolitic pyroclastic cones of Petite-Terre present rather similar types of deposits. In general, the deposits are well preserved and only locally are consolidated. The consolidation is not systematic and can vary within the same unit. It rather reflects post-emplacement induration since all the deposits below the water level are lithified and induration is almost absent in the unit above it.
Based on grain size and sedimentary features we identified six main facies (Table 4). These facies can be subdivided in seven sub-categories depending on their characteristics.
Table 4
– Facies description based on field observation of the 4 maars of Petite Terre
Fallout
|
Lapilli-grained, massive, well sorted, clast-supported
|
Fa
|
Lapilli- grained, massive, well sorted, clast-supported with block and bomb ballistic and associated impact-sag
|
Fb
|
PDC
|
Matrix-supported, well sorted, ash and minor lapilli, more or less pronounced stratification, with coarser lenses
|
Pa
|
Stratified to laminated ash with dune forms and lapilli-grained lenses
|
Pb
|
Lapilli tuff, massive with blocks up to 70 cm, coarse reverse tail grading
|
Ba
|
Lapilli tuff, laminated by finer levels with blocks up to 70 cm and impact structure, matrix supported.
|
Bb
|
Lapilli tuff, with coarse normal grading sometime beginning with tuff breccia grain size.
|
Bc
|
Massive fine ash. With no stratification or lamination
|
Ma
|
Massive, laminated, fine ash
|
Mb
|
Massive, laminated, fine ash with dunes
|
Mc
|
Reworked
|
Stratified alternation of coarse ash to coarse lapilli beds. These beds are fine-depleted and juvenile-poor.
|
Rwk
|
Scoria cones
|
Welded red to black scorias ranging from lapilli to block and bomb up to 3 m. Sometime associated with a lateritic weathering shell.
|
Sc
|
Fa and Fb facies (Fallout layers and ballistic clasts)
The Fa facies corresponds to massive, decimetre-thick beds (ranging in thickness from 5 to 40 cm, only in one section the thickness reaches 100 cm) with little changes in thickness at the outcrop scale. These beds are clasts-supported and made of angular, lapilli-sized, pumice and non-juvenile fragments with well-sorted granulometry (Figs. 4a and 4b). The main grain-size distribution (GSD) mode oscillates between − 2.5 Φ and − 1.5 Φ. The almost constant mantling thickness of the deposits at the outcrop scale, their well-sorted nature, and dominant lapilli size GSD are characteristic of a fallout sedimentation from short-lived (thin thickness), steady (absence of grading or of lamination) plumes. This main GSD mode usually coexists with a second coarse mode and a tail towards the fine fractions. The coarsest mode is attributed to ballistic ejecta, which we define as the Fb facies (Fig. 4c and 4d). This mode is truncated on the grain size histograms because the blocks and bombs are too big (up to 1.5 m in diameter) to be measured. The Fb facies is ubiquitous at the base of all the fallout layers, but it can also be observed, as train of ballistics, in the middle of ash-rich units (facies Pa/Pb). The largest blocks, usually formed by fragments of coral or mafic lava, can sag in the underlying deposits to a maximum depth of 2 m in proximal position. The fine-ash tail is almost ubiquitous in all the facies, and its amount ranges from 0.9 to 17.5 wt.% of the Fa facies. For the fallout deposits, the presence of this fine grain-size population is indicative of the contemporaneous sedimentation of co-PDC ash cloud during simultaneous emplacement of both PDCs and fallouts (Di Muro et al. 2008; Engwell and Eychenne 2016; Eychenne et al. 2012; Rosi et al. 2001)
Pa and Pb facies (Diluted Pyroclastic Density Current)
Ash deposits with minor and scattered lapilli (up to 3 cm in diameter), parallel stratification/lamination and/or lenses, and scarce ballistic blocks and bombs with associated impact-sag define the Pa facies (Fig. 4e). The stratification is formed by the alternation of coarse- and fine-grained ash layers (up to 50 cm thick each bed). The matrix-poor lenses are formed by angular lapilli. The lenses can reach a length of 20 m and thicken up to 1 m alternating with the finer matrix. In proximal position, this facies can be 15 m thick, thinning to 15 cm towards distal outcrops. Topographic highs also contribute to locally thin these units, as observed at the contact with the underlying mafic Ha Shiwawa cone (Airport beach, Fig. 2a) where the overlying ash unit thins from 3 m to only 1 m as it appears plastered against the cone. No grading was observed in these well- to very-poorly sorted (1.67 < σ < 4.73) units with a median GSD ranging from − 2.16 to 3.67 Φ (mostly of coarse ash). This facies can grade vertically or laterally to the cross stratified Pb facies (Fig. 4f), dominated by dunes (up to 1 m in height and 10 m in length). In ultra-proximal position (i.e., in the tuff ring) these units are the only ones observed intercalated with highly disturbed and discontinuous fallout beds and large-scale coarse lenses sometimes 2 m thick and 10 m long containing juvenile elements as pumice clasts up to 30 cm in diameter. All these characteristics indicate that these units were deposited from diluted, turbulent PDCs where traction sedimentation is dominant.
Ba, Bb, Bc facies (Dense Pyroclastic Density Current)
The B facies (Ba, Bb, Bc) are mostly present at the Airport beach site in La Vigie sequence but can be observed in the Dziani sequence and at Totorossa. The contact with the underlying units is sharp suggesting a very erosive PDC front (Fig. 5) with no deposition in proximal position, deposition occurs more distally as it can be seen in sections ARP3 to ARP8.
The Ba facies is a matrix supported coarse ash deposit, with scattered blocks (up to 1 m in diameter, found 3 km away from the vent) and pumice lapilli and bombs with scarce ballistics and their associated impact-sags (Fig. 4g). This facies is massive to crudely stratified, and poorly sorted (1.94 < σ < 3.49) with coarse tail grading of large clasts toward the top. The biggest obsidian bombs (up to 70 cm in diameter) were found in this unit. This facies is up to 3 m thick at Airport beach site and thins away from the vent to be totally absent in the distal sections at Mirandola island. Bb facies (Fig. 4h) is similar to Ba facies but present parallel laminations (tenth of centimeters long), thicker and less defined than in the Pa facies. It is similar to what is observed during aggradation of thick pulses in a granular flow regime (Sulpizio et al., 2014, 2007; Valentine et al., 2022) and is usually found above the Ba facies. The last Bc facies can sometimes be found at the base of these units as a coarse lapilli layer with normal grading (Fig. 4i and 4j). This facies only occurs at the foot of La Vigie tuff ring. There, the break in slope could lead to a change in flow regime, suppressing the turbulence and inducing a sharp transition to a granular flow. The change in slope would also explain the change from a erosional/transport regime to a depositional regime (Giordano, 1998; Macı́as et al., 1998; Sulpizio et al., 2014, 2010) and is similar to facies described by Wohletz and Sheridan (1983) and Valentine et al. (2022) in tuff cones architecture.
Ma, Mb, and Mc facies (Ash cloud)
Ma facies is defined by massive, fine ash (-0.17 Φ > MdΦ > 4.42 Φ), very well- to very poorly-sorted (0.77 < σ < 3.12) beds, ranging in thickness between a few cm to up to 1 m regardless their distance from the vent (Fig. 4k). The Mb facies differs in respect to the Ma facies, due to the presence of laminations a few mm thick (Fig. 4k). The Mc facies (Fig. 4l) is defined by the presence of dunes (60 cm high, 7 m long for the biggest observed). Based on their GSD these facies can be considered as ash cloud deposits (with or without horizontal movement) following the emplacement of PDCs or very dilute PDC especially in the case the Mc facies.
Rwk facies (Reworked deposits)
Rwk facies is defined as a stratified, fine- and juvenile-poor facies, dominated by fine to coarse lapilli with scattered blocks up to 40 cm for the biggest but with no impact sag associated (Fig. 4m). These facies can be found as channel infill. Stratification varies from mm to dm thick beds, some richer in dense non-juvenile clasts (e.g.: mafics fragments, peridotites) than others. Such characteristics, especially the porous texture and the lack of a primary juvenile fraction, suggest that it represents secondary deposits, reworked after initial emplacement. This facies can be formed long after the eruption (months to years) but also between two phases of the same eruption since erosion channels made by water runoff on freshly emplaced material can be filled quickly by reworked deposits.
Sc facies (Scoria cones)
Found at the base of several sections the Sc facies is composed of red to black scoria varying in size from fine lapilli to bombs up to 3 m (Fig. 4n). Fusiform bombs are common and show that magma was very fluid during emission. A lateritic shell indicating a rather pronounced weathering always covers this facies. Sc facies represents therefore old scoria cone fallout deposits present long before the phonolitic activity, sometimes buried under the phonolitic deposits.
Components
The pyroclastic deposits produced by the phonolitic explosive activity of Petite-Terre contains a large range of clast types (Table 5). Three main classes are present based on the classification of White and Houghton (2006), juvenile, non-juvenile and composites, depending on the origin of the clasts, all of which can be separated in subclasses.
Table 5 - Components found in the phonolitic pyroclastic deposits of Petite-Terre. The juvenile component can be split into three classes: pumices (J1), dense phonolites (J2) and obsidians (Obs.) forming a textural continuum. The mixed component is made of two classes: aggregates of smaller clasts (Aggr.) and accretionary lapilli (Accr. Lap.). The lithic component contains three classes: coral chunks from the barrier reef basement (L1), mafic lava (L2), and mantle xenoliths.
Three types of juvenile clasts are identified here. The glassy type fragments of these juvenile clasts have a phonolitic composition with a variability range which can be explained by different degree of micro crystallinity (Fig. 6, Supplementary S5). The main type are highly vesicular pumice clasts (ρ ≤ 1000 kg.m− 3) with a color varying from very light grey to grey. Quite rare, elongated vesicles are present in this pumiceous class. A second type are dense phonolitic clasts, which are much less vesiculated than their pumiceous counterparts (ρ ≥ 1000 kg.m− 3), and have a colour varying from grey, to greenish dark grey or black (Table 5). Rare rough or stretched surfaces can be observed. The third subclass of juveniles is formed by obsidian clasts: black, glassy fragments representing less than 1 wt.% of the juvenile component (Table 5). Obsidian sometimes represents the crust of bread-crusted bombs, but some bombs (up to 70 cm large) are entirely made of obsidian. These three subclasses should be seen as pure poles in between which there is a continuum with clasts that can be classified as one type of juvenile or another. Banded juveniles with pumiceous / dense phonolite / obsidian are common whether at the scale of lapilli or bomb (Fig. 7).
The non-juvenile component consists of three subclasses with very diverse origins. The studied pyroclastic cones are located on the edge of the modern coral reef and contain an important fraction of carbonate clasts. These sedimentary clasts are present as pieces of white to pink or orange limestone and/or sandstone in the deposit as well as recrystallized coral branches (Table 5) sourced from the Petite-Terre coral reef. Detrital sandstone (quartz grains + carbonate matrix) xenoliths already documented on other islands of the archipelago were also found in Petite-Terre pyroclastic deposits, suggesting that some of the non-juvenile classes originate from underlying older sedimentary layers (Masquelet et al., 2023) or from of the crust below the archipelago (Flower and Strong, 1969; Nougier et al., 1986). The non-juvenile component is also represented by an important fraction of mafic fragments. This subclass consists of dark brown to black fragments of basalt or more probably basanite and tephrite (Fig. 16, Table 5) which could originate from older mafic edifices like those present on Petite-Terre and in the lagoon or even from the older volcanic shield of Mayotte. The last type of non-juvenile clast described are mantle xenoliths (Table 5) carried by the magma during it ascent (Berthod et al., 2021b). When collected in-situ and not reworked, they are sub-angular and covered by a very thin layer of glassy phonolite. These mantle xenoliths seem to be more abundant in La Vigie and Moya and almost absent or scarce and very small in size (a few centimeters) in Dziani and Central.
Composite clasts are represented by two subclasses and are only observed in the PDC units. The first group is represented by the accretionary lapilli (Table 5) formed by fine ash accreted around a nucleus in the ash cloud present above the moving PDC during their emplacement in presence of free water (as vapor or liquid) (Schumacher and Schmincke, 1991). The aggregates are made by all the previously described types of components aggregated together (Table 5). The origin is post-eruptive with a cementation created by circulation of seawater in the deposits. For this reason, aggregates were excluded from the grain-size analyses since they reflect secondary processes.
Fine ash characteristics
The pumiceous juvenile component of the fine ash is as highly vesicular as its macroscopic equivalent (coarse ash, lapilli, and bombs). Vesicles on these pumiceous juvenile clasts are round but can be elongated sometimes and with a minimum size down to 10 µm (Fig. 8d) to a few centimeters for the biggest. Dense phonolite component is vesiculated at the microscopic scale with lower vesicles density but similar size (Fig. 8e). Obsidian was not found in the fine ash fraction but we think it is rather due to the fact that it appears textural similar to dense phonolites at a microscopic scale than because of a real lack of obsidian.
Non-juvenile fine ash grains represent, like their macroscopic counterparts, blocky fragments of mafic lavas and sedimentary material. However, no accretionary lapilli or peridotitic xenoliths were found at this scale.
In every sample investigated, very fine ash grains (< 2 µm) were present and aggregated at the surface bigger ash grains regardless of their nature (Fig. 8) These adhering particles are very diverse with pulverised carbonates, sulfides and sea salt (Fig. 8e/f), the latter likely originating from secondary processes since sampling was achieved on coastal outcrops. Primary phreatomagmatic fragmentation can be evidenced by the presence of typical surface features, such as branching cracks and mossy textures with round shapes, on the so-called “active particles” formed during MFCI (Büttner et al., 1999b, 2002b; Zimanowski et al., 2015a). It is the 4.0 Φ (64 µm) grain size fraction which is supposed to experience most of the energy during such fragmentation events (Büttner et al., 1999b). Particles that interacted with water after the primary fragmentation event are called “passive particles” and also bear typical surface features as pitting or adhering particles (Büttner et al., 1999b, 2002b). Though these surface features were identified for mafic composition with low viscosity in the first place, Austin-Erickson et al. (2008) showed that identical classification of particles and surface features is possible for felsic and high viscosity compositions as well.
In our case, no active particles surface features were observed in Petite-Terre deposits. This observation as well as the blocky and sharp morphology of the particles suggest that the fragmentation was brittle, as it is in the case of purely magmatic event (Fig. 8). However, occurrence of passive particles surface features as pitting and adhering particles was reported at multiples occasion in our samples suggesting that the fragmented materials interacted later with water, after its fragmentation (Fig. 8b/c).
Juvenile fraction density/porosity and vesicle connectivity
Density (n = 491) and connectivity (n = 74) measurements were gathered on juvenile phonolitic clasts from three major sites of Petite-Terre (La Vigie and Dziani and Mirandola island) and several mafic scoria cones (Foungounjou, Dzaoudzi, Ha Shiwawa, Totorossa, Mirandola, Sandravouangué, La Ferme). DRE measurements on finely powdered samples (i.e., density of the non-vesiculated rock) yield a value of 2478 ± 0.002 kg m− 3 for the phonolites and 2836 ± 0.002 kg.m− 3 for mafic lavas.
Porosity of the phonolitic juvenile material vary independently of the location but the different types cover a broad range from high to low porosity (Figs. 9a, 9b, 9c). Pumices are the juvenile category with the highest porosities (80.5 ± 7.5%) followed by the dense phonolites (47.7 ± 17.9%) while the obsidian have the lowest porosities (27.8 ± 12.4%). Despite the broad range of porosities, the degree of connectivity of the phonolitic juvenile material remains high through all types (respectively 95.9 ± 8.8%, 92.8 ± 20.0% and 91.5 ± 11.4%, Fig. 9e).
Scoria cones data (Fig. 9d) show low porosity (31.6 ± 23.5%) with a high degree of connectivity (102.2 ± 16.6%), in agreement with a typical strombolian activity (Colombier et al., 2017b; Gurioli et al. 2018; Thivet et al., 2020a, b).
Porosities for the pumice clasts suggest that for Petite-Terre phonolitic edifices, fragmentation was magmatic with values similar to known explosive eruptions (Colombier et al., 2017b, 2017a; Mueller et al., 2011). The lower values for dense phonolites and obsidians could be explained by phreatomagmatic processes in term of porosity (Mueller et al., 2011) but in this case would not display such high connectivity degrees. The banded textures which can be observed in the different juvenile types, along with the high connectivity degrees measured, is consistent with granular densification of initially porous pumices to denser phonolites and obsidians in the conduit after an initial phase of vesiculation (Colombier et al., 2017b; Okumura et al., 2013; Robert et al., 2008; Wadsworth et al., 2020), in a purely magmatic fragmentation scenario.
Edifices stratigraphy
The outcrops selected for a detailed study are distributed at ten different sites (Fig. 10) that are grouped in four main sectors. Detailed stratigraphic logs are available in supplementary S.6.
La Vigie (+ La Ferme + Sandravouangué + Totorossa scoria cones)
Deposits from La Vigie stretch along the airport beach forming a 1.7 km long outcrop which height varies from 50 m in proximal section to the north (within La Vigie rim) to 5 m in distal section to the south (Fig. 10a), although la Vigie rim is up to 202 m a.s.l. A mafic scoria cone (Ha Shiwawa) is buried under the deposits and crops out almost halfway (Fig. 10b). The outcrops are formed by a whitish succession of pyroclastic units but can vary in colour from white to yellow (Fig. 10c zoom of the proximal deposits). This colour variation is visible across the same unit but does not seem here to correlate with any chemical or facies change and might be due to surficial processes. The sequence observed within the northern (and most proximal) part of La Vigie outcrop (Airport beach site, ARP 1, Figs. 2b, 11 and S6.1.1) includes three fallout units with associated ballistics (Fa and Fb facies), at least four dense PDC units (Ba, Bb and Bc facies) and numerous diluted PDC units (Pa and Pb facies). Unfortunately, the contact between La Vigie tuff cone and the bedrock is not visible and the total thickness of the sequence remains unknown (Supplementary S6.1.1). Nevertheless, drilling in the area showed that pyroclastic material can be found at least until 20 m b.s.l. outside the crater in the central part of Petite-Terre (BSS002PNML and BSS002PNKM drill-holes, BRGM, 2023).
The lowermost units of La Vigie are observed at ARP1 section (Figs. 2b, 10c and S6.1.1), located within the tuff ring where the sequence thickens. The base of the sequence includes 5 m of stratified unit (Pa facies) with beds tens of centimeters thick. Based on field observations, some beds and lenses are visibly dominated by non-juvenile clasts (Fig. 11). This unit is also found at the base of ARP8 section (Fig. 2b and S3.1.8) where its base is plastered on the flanks Ha Shiwawa scoria cone (Fig. 10b) and is rich in accretionary lapilli suggesting that during its emplacement the eruption clouds contained liquid water. Above, it is a 20 m juvenile-dominated coarsely stratified unit, containing ballistics fragments (up to 20 cm in diameter) with associated bomb sags (Pa, Pb and Fb facies). The first four meters of this unit are characterized by a high content in obsidian coarse lapilli, which decreases upward before becoming very scarce, while pumice clasts increasingly dominate the unit (Fig. 11). The last 6 meters of the sequence visible at ARP1 section are composed of finely stratified (~ 5 cm thick beds) coarse ash units (Pa unit), starting with a 20 cm thick lithic-rich (42 wt.%) fallout bed with ballistic bombs and blocks (Fa and Fb facies). This fallout unit is later referred as the “Ballistic frigo”, due to the presence of metric ballistic blocks (Fig. 4c).
The continuation of the sequence is visible at the ARP3 section (Figs. 2b, 10d and S6.1.3) with the “Ballistic frigo” at its base. Above is the last unit observed at ARP1 with stratifications 1 to 2 cm thick and dunes (1 m high, 4 m long) (Pa and Pb facies) for a thickness of 2.5 m (Fig. 10f). This unit also contains accretionary lapilli observed at ARP7 section (Fig. 2b and S6.1.7). Above that is a sequence consisting of an alternation of fine laminated ash units, intercalated between coarser, juvenile-rich (between 60 and 98 wt.% of juvenile material) units (facies Fa and Fb), and massive units (Facies Ba). The ash layers are dominated by the Mb Facies. These units vary in thickness from 20 cm to 1 m with scattered accretionary lapilli (Figs. 10e and 11). The clast-supported layers, ranging in thickness from 30 to 20 cm, contain juvenile-rich, coarse lapilli (Fa facies) with scattered bombs and blocks, Fb facies (Figs. 4c/d and 11). The first massive unit, 20 cm to 1 m thick, contains coarse lapilli with small bombs and very discrete stratification (Ba, Bb and Fb facies, Figs. 10e and 11). Finally, the last massive unit is a 1 m thick massive, matrix-supported, lapilli and coarse ash unit with bombs and blocks and a laminated base (Ba and Bb facies). The block and bombs locally define a reverse grading. These last four 4 units contains obsidian as big bombs (tens of centimeters in diameter) and this sequence (Subunit A in Fig. 11) thins to a total of 2.5 m in stoss side position in respect to Ha Shiwawa cone where the fine ash intercalations thin down to 2 cm and sometimes even disappears (See ARP8 and ARP7 sections in S6.1.7). This thinning is due to the interaction between the PDCs and the topographic height represented by the scoria cone (Fig. 10b). After the cone, in lee side position, this sequence thickens to a total thickness of 4.5 m. The missing fine ash units here are interpreted as due to an increased erosive power of the dense PDCs during their emplacement.
The end of the La Vigie eruptive sequence is found at ARP7 section (Fig. 2b, S6.1.7) where, above the Ba facies, we observed a 30 cm of juvenile-dominated coarse ash and lapilli unit with small lenses (10 cm thick) rich in pumiceous bombs (Ba facies). This unit is topped by 60 cm of fine laminated brown ash, whose first 20 cm are a fine alternation (at the centimetre scale) of coarse ash and fine lapilli (Ma and Mb facies). The sequence finally ends with 30 cm of fine-poor clast-supported lapilli unit (Rwk facies) fading off to the overlying soil.
Despite coarse units alternating with fine ash in Sub-sequence A (see Fig. 11), Unit 1 lies directly on top of Unit 2 with no fine ash intercalated. At ARP3 section it is possible to observe an incision at the base of Unit 1 into Unit 2 (ARP3, S.6.1.3, Fig. 4m). This depression is filled with dense centimetric clasts (mafic lavas, coral, peridotite, dense phonolites, and obsidians) while pumice clasts and fine particles are not present (Rwk facies). This small depression and its in-filling, the absence of fine ash intercalation between the two units, and the flat lamination at the base of Unit 1, show that the depression was already filled when Unit 1 was deposited. This suggests that the contact between Unit 1 and Unit 2 marks a pause in the eruption of La Vigie rather than only an erosive effect during Unit 1 emplacement. Such depressions, as small erosive channel, have been observed from historical eruptions or identified in deposits (Nemeth and Cronin, 2007). Their quick (hours to days) formation happens during very short period of calm within an eruption. Such channels are also observed between Unit 2 and Unit 3 at ARP 5 section also suggesting a pause in the eruption between the emplacements of these two units.
Through more and more distal sections Unit 4 and Unit 2 lose their well-defined fallout layers, transitioning to finally being very similar to Unit 1. This transition is rapid since it happens in only a few decametres right to the south of ARP3 section for Unit 4. We interpret this as a remobilization of the fallout unit by dense PDCs.
The juvenile fraction dominates the whole sequence, with a dense phonolite component (in weight) compared to pumice clasts. However, given the density of both pumices and dense phonolite, their volumetric components are similar. The dense phonolite component also increases toward smaller fraction. Some ballistic fallouts are rich in non-juvenile elements up to 42 wt.% (‘‘Ballistic Frigo’’). The whole sequence is synthesized in Fig. 11.
La Ferme scoria cone is located 2 km west of La Vigie (Fig. 2, S6.5.4) and is the most prominent of all Petite-Terre scoria cones peaking at 132 m a.s.l. Despite its size and distance from La Vigie, which is the nearest phonolitic vent, pyroclastic deposits can be found up to 102 m a.s.l. with a thickness of 2 m as a massive ash bed with very thin lamination at the base (Pa facies, Fig. 10g). These PDCs from Petite-Terre eruptions not only were able to cover the immediate surroundings of the vent (as Sandravouangué scoria cone) but also to climb up significant reliefs a few kilometres away from their emission point. However, we did not identify phonolitic deposits on top of the scoria cones of Foungounjou and Dzaoudzi (Fig. 2,).
Sandravouangué is a mafic scoria cone located south-west of the La Vigie culminating at 44 m a.s.l. The cone is nowadays totally covered by houses, its existence only being inferred from DEM study where its shape is visible or from scarce and short-lived outcrop at construction sites. Given its very close position from La Vigie, this scoria cone was entirely buried under the La Vigie pyroclastic deposits during the eruption. At the summit of Sandravouangué we found a pyroclastic sequence at least 5 m thick lying on the weathering shell of the scoria cone dipping toward La Vigie tuff ring. The sequence begins with 3 m of alternation of fine lapilli and ash beds with laminations (Pa facies) followed by 2 m of massive ash with lamination (Pa facies). The top of the sequence is a 30 cm-thick lapilli fallout level with no visible ballistic (Fa facies). No grading is observed in this fallout bed (S.6.5.6).
The TOT2 section (Fig. 2b, S6.5.7), located on the southern flank of the Totorossa scoria cone also exposes a 8 m thick sequence composed of five different units of phonolitic pyroclastic deposits lying on the lateritic paleosoil of the cone. The base is made of 3 m of stratified alternation of fine ash and fine lapilli beds of centimetric thickness plastering the lateritic shell of Totorossa. It is followed by 1.5 m of massive breccia with coarser lenses (Ba facies). Above this unit is 0.7 to 1 m of fine, massive brownish ash (Ma facies) containing small pores and accretionary lapilli suggesting presence of liquid water during the emplacement of this unit. Another stratified unit of fine and coarse centimetric ash beds follows with some lapilli-supported 20 cm thick beds (Pa facies). The orientation of scattered ballistic bombs and blocks and their associated impact sag (up to 30 cm in diameter) (Fb facies) show that they originated from La Vigie or Moya tuff rings. This sequence ends with 1 m of fine and juvenile poor stratified layer-bearing fine and coarse lapilli, likely reworked.
Moya and Central
Moya crater deposits are accessible on the Moya Isthmus (Figs. 10h/i, S.6.3). This outcrop (MOYA1 section, S6.3.4) stretches along 110 m on the northern flank of the isthmus exposing the ultra-proximal units of Moya tuff ring but also the units of Central tuff cone, lying on an erosive surface. There, the units from Central cone climb on the Moya units, the whole sequence representing a total thickness of approximately 35 m.
Moya tuff ring units
The first 20 m of this sequence are matrix supported and rich in pumice ash and lapilli. The sequence is characterized at the base by parallel stratifications (Bb facies), and dunes (Pb facies) 10 m long and 2 m high. Three meters above the base a big complex lens (2 m thick, 10 m long) is rich in ash and contains smaller lens (50 cm thick, 2 m long) of big pumices (up to 15 cm in diameter). Stratification then fades and the unit becomes more massive with scarce bombs and blocks (Ba facies). Some stratifications are present sometimes as well as some coarse lapilli lenses. Peridotitic xenoliths up to 10 cm have been found in this unit. On the beach, the lithified units dip toward the center of Moya crater. Considering the geometry of this unit, it is likely to represent the remains of the eastern part of the Moya crater rim.
Central tuff cone units
After an angular unconformity, another 15 m thick pyroclastic unit can be observed, that we attribute to Central tuff cone. This stratified unit is formed by an alternation of 10 to 20 cm thick beds of fine ash and coarse lapilli beds with intercalations of pumiceous bomb lenses some with large ballistics (up to 1 m) (Pa and Fb facies). This unit, which “climbs” above the isthmus of Moya (Fig. 10n/o), is accessible from the southern flank of the isthmus of Moya but is lithified there and contains very few peridotitic xenoliths.
Central tuff cone is nested inside both La Vigie and Moya tuff rings and cut their deposits (Fig. 10p, S6.3.1/2/3). The deposits of this edifice can be seen at the south of Moya beach where the sea cuts through the Central tuff cone edifice. There, it is possible to observe the 50 m thick ultra-proximal sequence of Central tuff cone. This sequence is very similar to the other ultra-proximal sequences that can be seen at Badamiers or Airport beach site (for Dziani and La Vigie edifices, respectively), dominated by stratified PDC units with coarser lenses impacted by blocks and bombs and their associated bomb sags, but no clearly distinguishable fallout due to the chaotic environment and the height of the outcrop. Near the base of the section is a 20 cm thick ballistic layer overlying a 3.5 m thick breccia facies beginning with stratifications of tens of centimeters before transitioning to a more massive breccia facies toward the top (Bb to Ba facies).
Post-eruptive deposits
On Moya Isthmus a second angular unconformity can be observed above the Central tuff-cone unit (MOYA3 section, S6.3.6). Above this unconformity, the deposits are horizontal and similar to deposits observed inside Moya crater on a small (2 to 5 m high) littoral cliff along the northern Moya beach (MOYA2 section, S6.3.5). Those units (at MOYA2 and MOYA3 sections) are characterized by planar, stratified, and fine-depleted coarse and fine lapilli beds (2 to 3 cm) with some very coarse lapilli lenses (5 cm thick, 2 m long). All these units are juvenile-poor, and the biggest blocks (20 cm) are not associated with any impact sag suggesting none of these are primary but rather remobilized volcanic deposits. The horizontal stratification and excellent bed continuity resemble post-eruption lacustrine deposits. The juvenile-poor character of these last units is compensated by an abundance of dense clasts such as dense mafics and peridotite clasts (Rwk facies).
Near the base of Central tuff-cone northern flank a ravine cut where an angular unconformity can be observed, followed by a 6 m thick sequence. This sequence has horizontal stratifications and likely emplaced after a long pause of eruptive activity. It consists of four finely stratified units with beds around 1 cm and a coarser, 60 cm-thick, unit containing some blocks up to 40 cm wide and with only dense material (Rwk facies). It could be associated with the similar deposits interpreted as post-eruption lacustrine deposits described at Moya Isthmus sequence.
This likely represents reworked lacustrine deposits inside the La Vigie-Moya-Central complex. The thickness and the repartition inside the bottom of Moya crater suggests the former presence of a lake inside the complex, as is the case today for Dziani tuff-ring before marine erosion break through the tuff ring barrier.
Dziani (+ Papani beach + Totorossa)
Deposits of the Dziani activity are exposed at the Badamiers (Figs. 10/e/f/g) and Papani (Fig. 10j) beaches, which cover the northern part of Petite-Terre on its western and eastern coasts, respectively. The deposits are well exposed and the cliffs even cut the Dziani tuff ring where 60 m of the thickest sequence can be observed. This sequence thins toward the north where it barely exceeds 10 m in thickness despite being almost complete at the Northern tip of Petite-Terre. It consists of an alternation of PDC unit (Pa and Pb facies, most likely base surges), fallouts and ballistic trains (Fb facies, 7 units identified) (Fig. 12).
In ultra-proximal position the sequence exceeds 60 m (BAD1 section, S6.2.1), no continuous fallouts can be identified, but discontinuous, coarse grained, lapilli-supported lenses are visible. The lowermost deposits of Dziani can only be observed in ultra-proximal position, although the contact with the substratum was not observed. The two units are characterized by very coarse lenses (with pumice clasts up to 30 cm in diameter, Fig. 10i/k), the lowest dominated by pumice clasts and the overlying by dense phonolite material in a stratified matrix of indurated ash. Above that, the sequence is a succession of laminated units (Pa facies) with no clear distinction.
The sequence rapidly thins within the first 500 m toward the north and the fallout units become more distinguishable from the PDC units. Up to seven fallout layers were individualized within the Dziani sequence, with an individual unit thickness varying from 6 cm to a maximum of 40 cm. The rest of the sequence is characterized by PDC units (Pa and Pb facies). In the middle of the sequence the “breccia-like” unit stands out. This unit consists of a 1 to 1.5 m thick level of breccia facies (Ba facies) coarse lapilli-grained with blocks and bombs surrounded by two 70 cm thick PDC stratified units (Pa facies), the lowest one sometimes being absent (BAD 5, S6.2.5). Above this “breccia-like” unit, a key-layer was observed, consisting in 30 to 40 cm of brown ash (Ma facies), sometimes divided into two 30 cm thick levels surrounding a thin (5 cm) white ash layer (Fig. 12).
At Badamiers beach, it is also possible to observe the interaction between the PDC and the paleo-topography through various field evidence. On the west coast a 100 m-wide depositional fan opens at the end of a paleo-valley. This fan is composed by primary volcanic deposits that connect to units of the BAD4 section, today lithified, channelized by a paleo valley and/or redirected like the ones found at the end of the valleys at Soufriere Hills (Cole et al., 1998). At the same site, but in the cliff, the pyroclastic units overpass a 1 m high crest perpendicular to flow direction. On the lee side of this crest, coarse material likely deposited after the blocks and bombs hit the obstacle and lost their kinetic energy. Some of these blocks are associated with small impact-sags.
Papani beach (Fig. 10l/m, S6.4.1) offers a 300 m long outcrop located between Dziani and Moya tuff-rings. The 40 m high coastal cliff exposes a sequence dominated by stratified units of fine and coarse ash with beds around 10 to 20 cm thick (Pa facies). Near the base of the outcrop cross-stratifications (5 m long, 4 cm high) can be seen. A giant dune (24 m long, 4 m high) (Pb facies) can be observed in the middle of the sequence (Fig. 10m). The orientation of this dune suggests a flow direction originating from Moya tuff ring whereas 2 m above, another dune (2 m high 8 m long) presents a flow direction originating from Dziani. These two sequences with divergent paleo-directions currents are also characterized by different colours: a yellowish matrix in continuity with Moya tuff-ring for the lower layer and a whitish one in continuity with Dziani for the upper level. It may thus suggest that contrary to La Vigie, the change in colour might represent here the limit between both tuff rings. The origin of this colour is however still unknown and wasn’t investigated as part of this study. The absence of paleosoil or any pause indicator such as erosional channel within this sequence also suggest that Moya and Dziani tuff ring were active simultaneously or closely in time with Moya having preceded, even by a small timespan, the formation of Dziani.
Mirandola island
Mirandola Island, located on the western side of Petite-Terre, consists of a mafic edifice covered by the phonolitic pyroclastic deposits (Fig. 10q, S6.5.1/2). The total pyroclastic sequence at this location varies between 80 cm and 2 m. At the base of the sequence MIRA1(S.6.5.1) the contact between the pyroclastic units and the red lateritic weathering shell of the scoria cone (Sc facies) can be observed. The sequence begins massive, very fine grey ash deposit with mode under 4.5 Φ, locally laminated at the base (Ma and Mb facies) that we interpret as dilute PDC deposit. This unit is followed by three fallout units (Fa facies only) with a massive ash layer (Ma facies) intercalated and ending with a laminated ash layer on top (Mb facies).
The three fallout units above this first dilute PDC unit are all separated by 10 cm-thick fine ash beds. These fallout units share most of their characteristic with a first mode around − 2.0 Φ to -2.5 Φ a second one at or under 4.5 Φ (Fig. 13). The weight proportion of this second mode increase trough these fallout toward the top of the sequence with up to 26 wt.% of fine ash in the third fallout suggesting an intensification of the fragmentation or a diminishing plume height. These fallout units are dominated by a population of juvenile clasts, with a majority of pumice clasts which proportion increases toward the top respective to the dense phonolite juvenile clasts. These granulometric and componentry variations could be explained by a more and more effective and dominant magmatic fragmentation. The mixed components are only present in a minor fraction (6 wt. %) in the second fallout indicating a much less important role of water.
The last sampled level (Ma facies) is also very fine grained. In contrast to the three-underlying fallout, the dense phonolite component is dominant. The top of these sections (MIRA1 and MIRA2) is composed by 50 cm thick soil developing on the phonolitic deposits.