The volcano-tectonic structural framework obtained in this study is derived from the combined interpretation of several data collected in the CARG project ¹ (Servizio Geologico d’Italia, 2018, Geological map of Italy, F464 Isola d’Ischia, in a scale of 1:25,000, ISPRA; Carta Geologica della Regione Campania, scala 1:10,000, Foglio 464 Isola di Ischia, Note Illustrative, Regione Campania, Assessorato Difesa del Suolo. LAC Firenze; Sbrana et al., 2011), including on-land (subsurface geology reconstruction using thermal water borehole stratigraphies) and offshore information (bathymetry and seismic line acquired in CARG project in correspondence of the Lacco Ameno debris avalanche and reinterpreted in this paper) (Figures 2 and 3).

In particular, the stratigraphies used in this work (Figure 3) were derived from the water wells drilled for thermal use for hotels and spa-wellness center in the north sector of the island.

The offshore area is investigated using seismic line 21 (Figure 4), acquired in 2005 during geophysical CARG surveys, finalized to the geological and volcanological reconstruction of the marine areas of this active insular volcanic field (Sbrana et al., 2011).

High-resolution sparker seismic surveys were performed by Parthenope University staff by means of a single-channel reflection seismic system composed by the energizing one (EG&G Trigger Capacitor Bank mod. 231 and power supply mod. 232), the firing or seismic source (SAM96 sparker, at 96 electrodes), and the measuring receiver. The frequency range of the signal extends from 100 to 3,000 Hz, and the sampling rate is 12,000 samples per second. The measuring apparatus was composed of a hydrophonic chain, an amplifier, an anti-aliasing filter with a bandwidth from 150 to 3,000 Hz and by the D-Seismic acquisition system. The maximum vertical resolution is about 0.25 m, while the horizontal is about 3 m. Depths are expressed in milliseconds (two-way travel time) and converted in meters (for the calculation of the thicknesses of the various seismic units) considering the conventional speed of acoustic waves in water equal to 1,500 m/s. Multibeam surveys were carried out from the coastline to the depth of −200 m, in order to obtain detailed bathymetric map of the study area (Passaro et al., 2016). The positioning of the boat was detected with an Omnistar DGPS system with an error of about 1 m. The geophysical and geological interpretation of the high-resolution line is based on the good knowledge of the on-land geology and volcanology. This allows the team to perform an integrated interpretation of the main volcanic units on the island and of the seismic units, interpreted in terms of seismic facies and lithological characters of volcanic and sedimentary and epiclastic geological bodies, applying the principle of continuity of on-land units, coastal units, and offshore marine units.

In general, melt inclusion (MI) investigations can provide significant information on the pre-eruptive settings (composition, temperature, and depth) of feeding system of volcanoes (Blundy and Cashman, 2005; Marianelli et al., 2006). Here MI analyses allowed us to obtain constrains about the depth and temperatures of magma feeding the northern sector of the island, during the final stage of volcanism (10 ka–1302 A.D.). Sample collection of juvenile fraction was restricted to trachytic pumices from fallout deposits of Cretaio Plinian eruption, to latitic scoriae from Vateliero scoria cone, and to latitic–trachytic banded pomiceous scoriae from Arso explosive/effusive eruption. The juvenile fractions show variable phenocryst content in a glassy matrix. Phenocrysts were handpicked, mounted on slides, and double polished. After petrographic inspection, selected MIs were prepared for further analyses. All the selected samples host melt inclusions having a glassy appearance. SEM-EDS microanalysis was carried out on MI and also on minerals and glasses using a Philips XL30 EDAX Genesis. Operating conditions were 20 kV and about 0.1 nA beam current. A raster area of about 100 μm² was employed for glass analysis to reduce the light element loss. The analyses were normalized to 100 wt% by using the EDAX software. A set of reference standard of natural trachytic (CFA47), basaltic (ALV981R23), and pantelleritic (KE12) glasses were analyzed before every session. Analytical results, errors, reproducibility, and detection limits using international standards for SEM-EDS technique are reported in Marianelli and Sbrana (1998).

Microthermometric homogenization experiments on MIs were carried out on doubly polished wafers of crystals. Only glassy MIs, with a clear shrinkage bubble and no evidence of heterogeneous trapping (Roedder, 1984), were chosen. For the high-temperature experimental study of MIs, a modified Leitz 1,350 heating stage was used. The temperature was measured with a Pt-Pt90Rh10 thermocouple. The accuracy of measurement was around ±10°C, controlled by the melting point of gold and silver. Experiments were performed in a He atmosphere. The rate of heating was varied as a function of the rate of transformation in the inclusions and ranged from 2°C to 40°C/min to the point of complete homogenization of MI, i.e., disappearance of bubbles. Pilot runs were carried out in order to define the correct heating/step procedure before the starting of thermometric data collection. Vapor loss can be excluded due to the reproducibility of experiments on the same MI.

H₂O and CO₂ in MI were analyzed by transmission IR spectroscopy (FTIR) using a Nicolet iN10 equipped with a high-intensity EverGlo IR source and MCT-A detector cooled with liquid nitrogen. At least three spectra for different areas were collected for each MI (minimum area of 10 × 10 µm). The concentrations, C, were calculated according to the Beer–Lambert law: C = 100∙A∙M/(ε∙ρ∙d), where A (absorbance) is the peak height of the absorption in dimensionless absorbance units, M is the molar mass (g/mol), ε is the molar absorptivity (l∙mol⁻¹∙cm⁻¹), ρ is the density (g/cm³) of glass (estimated using the iterative method of the Church and Johnson, 1980, and Gladstone–Dale rules), and d is the thickness measured by visual determination under a calibrated microscope, with an error of 2–3 µm depending on the proximity of a given inclusion to the edge of the wafer. The quantitative procedure and the absorption band assignments described in Silver et al. (1990) and Wallace et al. (1999) were followed in this work. The molar absorptivity values chosen were 62 for the 3,570 cm⁻¹ according to Di Matteo et al. (2004). No carbon peak is observable above the background, indicating that the CO₂ is below detection limit (about 50 ppm, depending on MI thickness).

The map in Figure 2 highlights the geological, geomorphological, and structural patterns of the Casamicciola area and its northern offshore sector. On land, north of Ischia, we observe an interdigitation of 1) thick volcanic units linked to the caldera-forming phase of the volcanic field, 2) clay-rich marine sedimentary bodies from Campanian plain, and 3) thick epiclastic units formed during the caldera filling and the successive resurgence.

The structural framework of the studied area is characterized by faults with vertical offsets that reach tens of meters, well evident in the geological section displayed in Figure 4, which includes the stratigraphy of the on-land wells (A–A′) and the interpretation of the seismic line (B–B′) offshore. The north portion of the seismic line (B–B′) appears undeformed, while several faults bordering marine terraces evidence a progressive uplift of the volcanic and sedimentary units toward the inland. The acoustic basement is represented by a unit having a characteristic reflection-free facies typical of ignimbrites (Aiello et al., 2020). This is relatable to the caldera-forming phase (60–56 ka, Sbrana et al., 2018), in which thick ignimbrites are emplaced, including the Mt. Epomeo Green Tuff unit. The acoustic basement unit is topped by reflection rich seismic units, interpreted as corresponding to the sequence of epiclastics and clay-rich marine sediments covering the Ischia ignimbrite deposits, outcropping on the northeast slopes of Mt. Epomeo resurgent block (Campomanno epiclastics and Cava Leccie clays and sands, marine units).

These units and the underlying tuffs are arranged in several step faults accommodating the Epomeo uplift. On the other hand, in the marine area, they appear affected by a bending linked to a normal fault system that connects to the Grande Sentinella horst (Figures 2 and 4). In the on-land part, on this structure, the marine sediments are uplifted up to about 80 m a.s.l. Wells stratigraphy (Drilling 1 to 4, Figure 3) and morphostructural counterslope show the existence of a small graben placed between the Grande Sentinella horst and the Mt. Epomeo normal master fault (Figure 3) bordering the resurgent block.

Two deformation systems having different amounts of deformation are shown in the geological map (Figure 2) and in the land–sea section (Figure 4). The most deformed area, experiencing almost 900 m of uplift, coincides with the Mt. Epomeo block resurgence (56–33 ka post caldera activity); this is delimited by the north dipping normal fault of the Piazza Bagni–La Rita graben (Figures 2 and 4). Northward deformations are less developed (tens of meters) and the maximum uplift (about 100 m) coincides with the Grande Sentinella horst (Figure 4). This structure is part of the uplifted domed belt formed after 5 ka (during phases 5 and 6 of Sbrana et al., 2018), between Cafieri area, Casamicciola, and Mezzavia in the northern coastal sector of the volcanic field and highlighted by marine fossiliferous epiclastics units uplifted up to 50–80 m above sea level.

The presence of marine terraces, both on land and offshore (Figure 4), records several stages of deformation linked to fault activity and sea level variations in the northern sector of the island: in particular, four order of marine terraces are mapped offshore at −10, −20, −50, and −75/80 m b.s.l. (partially covered by the morphology of Lacco Ameno debris avalanche), while on land, terrace rims are located at 10, 20, 50, 80, and 110 m a.s.l., uplifted by the recent uplift in the Grande Sentinella, Casamicciola, Mezzavia, and Fundera areas.

Volcanism of the last 10 ka of activity (phases 5 and 6) is mainly concentrated on northern sectors of the island and is represented by lava domes, lava flows, and tuff cones. The volcanic products of Ischia are characterized by a strong alkalinity (Na₂O + K₂O up to 14 wt.%) and the bulk rock composition ranges from latite to trachyte (Supplementary Figure S1A), with rare shoshonite and phonolite, as already reported by Rittman and Gottini (1980), Crisci et al. (1989), Civetta et al. (1991), Sbrana et al. (2018). Most of the erupted magmas are highly differentiated and consists of lava flows, lava domes, and pyroclastics trachytic in composition, with crystalline felsic xenoliths present in the pyroclastic units. However, petrochemical data (Piochi et al., 1999; Pelullo et al., 2020) highlight that mingling/mixing processes affect volcanic products of phase 5 (e.g., Zaro) and phase 6 (e.g., Fiaiano, 530–960 A.D., Arso, 1302 A.D. among the others) (Figure 1). Several studies agree with a relatively shallow magmatic system present in the island subsoil (Rittmann, 1930; Poli et al., 1987; Crisci et al., 1989; Civetta et al., 1991; Sbrana et al., 2009). The evidence of a shallow magmatic system is supported by the very high geothermal gradient measured in the island, by gravimetric, potential field, and magnetotelluric data (Nunziata and Rapolla. 1987; Paoletti et al., 2009; Sbrana et al., 2009; Di Giuseppe et al., 2017; Carlino, 2018); however, this system is still weakly defined.

In order to obtain reliable data from MI investigation, the juvenile fractions of Ischia volcanic deposits were selected, on the basis of cooling rate (Cioni et al., 1998). The selection was restricted to samples of pumice deposits. In particular, for the phase 6 volcanic activity, well-quenched primary two-phase glassy MIs (20–60 µm in size) are found in clinopyroxene of pumices and scoriae of the juvenile fraction of the Cretaio, Arso, and Vateliero eruptions. The mineralogical assemblage of their hosting rock is formed by K-feldspar, plagioclase, clinopyroxene, biotite, oxides, apatite, and titanite. The analyzed MIs have trachytic composition (Supplementary Table S1 and Supplementary Figure S1); as the major element composition of most MIs is similar to that of the glassy matrix, this could indicate that only minor differentiation of melts occurred between entrapment and eruption. Microthermometric experiments provided a homogenization temperature between 956°C and 1,043°C, with a mode of 1,005°C (Supplementary Figure S2).

FT-IR investigation revealed that the dissolved water content ranges from 1.0 to 2.9 wt.% with a median value of 2.2 wt.%, (Supplementary Table S2 and Supplementary Figure S3) similar to the range of values reported in Sbrana et al. (2009) for older eruptions (phase 4 and phase 2) of the southern sector of the Ischia volcanic field.

The dissolved water content of MIs can be assumed as representative of the pre-eruptive volatile content of melts. According to the hypothesis of water saturation conditions for the trapped melts, the H ₂ O solubility model of Di Matteo et al. (2004) for trachytic melts was applied to calculate saturation pressure for these MIs and a pressure range of 17–52 MPa was obtained (Supplementary Table S2 and Supplementary Figure S4). This pressure range, throughout a model based on measured densities of the rocks (from 1.5 to 2.0 gcm⁻³ for hydrothermally altered tufaceous rocks, 2.4 to 2.5 gcm⁻³ for fresh trachytes and syenites, Sbrana et al., 2009), is then converted to depth. Accordingly, MIs were trapped within a trachytic storage region located between 800 and 3,500 m from actual ground surface (<2.000 m, for the median value of saturation pressure; Supplementary Figure S4 and Supplementary Table S2), at least in its shallower apophyses. Based on compositional (major elements and H₂O content) and temperature data recorded by MIs, very low viscosity values (about 10³ Pa s, analogous to values known for basaltic melts) have been estimated for trachitic melts following the procedure of Giordano et al. (2008). Therefore, trachitic melt having relative high mobility can be hypothesized in the shallower portions of the Ischia magmatic feeding system, although the high crystal content of some volcanic products (e.g., Zaro) determines a significant change in magma rheology, such as increase in viscosity (Costa, 2005). The occurrence of felsic crystalline xenoliths in pyroclastic deposits testifies that also crystalline portions of magma chamber (solidification front of Marsh, 2000) and/or related intrusive or subvolcanic bodies are present in the shallower portion of the feeding system, together with low-viscosity melts. Furthermore, rare MIs in Arso and Vateliero samples have less differentiated composition and higher homogenization temperature, well in agreement with shoshonitic to basaltic melts rising from deeper storage zone (5–8 km of depth) and mixed with resident low-pressure trachytes (Piochi et al., 1999; Pelullo et al., 2020).

The interpretation of seismic and the borehole data, joined with the CARG geological data show that the process of resurgence was characterized by different phases, which involved in a different fashion the outer and the inner part (Sbrana et al., 2011, 2018) of the present resurgent block. In the northern offshore, the presence of normal faults dipping northward, with small vertical throw (tens of meters) gradually increasing southward and cutting the deposits of eruptions following the caldera formation, indicates a differential uplift of the island (Figure 4). Assuming an about radial symmetry for the magmatic intrusion of Ischia (Carlino, 2012, and references therein), the location of the offshore faults related to the resurgence indicates a magmatic sill intrusion of about 5 km in diameter.

The faults characterized by larger vertical throw delimit the main uplifted structure of Mt. Epomeo, whose dislocated blocks appear slightly tilted southward, with evidence of uplift starting from Casamicciola–Grande Sentinella up to about 3.5 km north of the coastline (Figure 4). This structural pattern has been recognized in other resurgent calderas (i.e., Long Valley caldera: Bailey, 1989; Hildreth et al., 2017. Phlegrean caldera: Sbrana et al., 2021; Steinmann et al., 2018; Sacchi et al., 2014) as well as in experimental analog modeling of shallow magma intrusion (Walter and Troll, 2001; Acocella et al., 2004). A localized uplift appears to have occurred in the final stage of the resurgence, post 5 ka, involving the Casamicciola-Grande Sentinella sector, uplifted by about 100 m with respect to the sea level. This provides a relatively high average uplift rate of at least 1–2 cm∙year⁻¹.

The interpretation of borehole data highlights the dislocation of the oldest marine deposits (caldera filling sediments) encountered in the Hotel Tusculum (Drilling 1) and Pantano (Drilling 2) wells (Figures 3, 4), being characterized by an offset of about 80 m between the top of the deposits (Cava Leccie fossiliferous clay and sands) in the two wells. These sediments are covered by quaternary Lacco Ameno debris avalanche deposits and debris flows (Figure 4).

According to the above observations, the southern sector of the Grande Sentinella uplifted block would have been cut by a roughly E-W normal fault (or faults system), for which there are also morpho-structural evidences (Sbrana et al., 2018). The formation of a graben-like structure—favored by the doming occurred in phases 5 and 6 and likely by the regional extensional tectonics (Hippolyte et al., 1994)—accommodates the dislocation occurring along a southward dipping fault (Figure 4). This structure appears not to be related (spatially nor chronologically) to the main resurgence phase of the Mt. Epomeo. This hypothesis is supported by the roughly E-W alignments of most recent eruptive centers at Casamicciola (Grande Sentinella-Mezzavia-Fundera) and by the MI data indicating a very shallow depth (mean value <2,000 m) of the magma bodies along the northern sector and outside the resurgent block. We suggest that the graben-like structure was originated when magma migrated at a shallower depth (as indicated by MI data) in the northern sector about 10–5 ka, in the final stage of the resurgence. The different patterns of ground deformation and the related structural pattern might be associated with the different evolution of the intrusion dynamics. At Ischia, the change in pattern of deformation possibly reflects a complex shape of the underlying magma intrusions. For instance, steeper edges of the uplifted zones are generally located above the shallower parts of the intrusion (Galland, 2012), as confirmed at Ischia by geophysical investigations (Di Giuseppe et al., 2017). During the most recent geological history of the island (10–5 ka), the pattern of magma migration in the northern sectors might have been controlled by the presence of roughly vertically dipping faults related to the resurgence, along which dike migration up to the surface is favored by extensional tectonics (Piochi et al., 2005; Torrente et al., 2010; Gudmundsson, 2020). On the other side, the onset of the subsidence process may have squeezed out a fraction of the resident magma in the shallow source (Gudmundsson, 2020). This process, possibly starting after 5 ka, enhanced magma displacement toward the surface, generating localized uplifts. Furthermore, the petrological data show a contemporary presence of low and high crystalline magma batches in the shallow source of Ischia. These data suggest that the higher viscosity of the resident magma may have not favored the magma hybridization with the mostly basaltic, less viscous, magma that raised from the depth of about 5–8 km without residing in the shallow source (e.g., Arso and Vateliero eruptions); indeed, the interaction between magmas with different viscosity results only in the observed sporadic mingled products.

The horst-graben structure and the related fault dislocating the northern block of the Grande Sentinella represent crucial volcano-tectonic features, since they correspond to the seismic zone of the island generating the destructive earthquakes in historical and recent time (Carlino et al., 2021). The lateral and vertical extension of this seismogenic zone is controlled by the dimension of the resurgent block of Mt. Epomeo and the depth of brittle–ductile transition, respectively (Carlino et al., 2021). Thus, the fault appears enclosed into a complex kinematics, where the resurgence in a former phase and the subsequent subsidence have generated an intricated volcano-tectonic system, on a relatively small area. Reconstructing the past deformations and measuring the present ground movements are thus crucial to understand the island dynamics and the associated seismicity.

The 2017 earthquake was the first considerable event in the island recorded instrumentally according to modern standards. It provided the possibility to obtain a quantitative model of the seismic fault from the analysis of geodetic and seismic data (De Novellis et al., 2018). The solution is characterized by an almost purely normal fault plane, with most dislocation concentrated on an area of about 1.5 × 1 km², slightly dipping southward (70° ± 7°), with strike 86° ± 5° and rake −80° ± 5°, located at the northern base of Mt. Epomeo and with its top at about 400 m below free surface (De Novellis et al., 2018). The higher accuracy of the location of the fault retrieved by De Novellis et al. (2018) with respect to other proposed solutions (see for instance Braun et al., 2018, and Calderoni et al., 2019) is demonstrated by the result of the high-precision leveling survey performed by INGV immediately after the 2017 earthquake (Ricco et al., 2019). In fact, the surface projection of the fault runs through the 97B and 98A leveling benchmarks, where a 3.2-cm offset associated to the coseismic displacement of the 2017 fault was detected by the survey (Figure 5). According to Carlino et al. (2021), this is the fault responsible for the historical and recent earthquakes of the island. Based on its location, geometry, and kinematics, this seismogenic structure is consistent with the southward dipping Grande Sentinella (GS) fault, bordering the graben structure described above.

In this framework, the GS fault would act in response to the local stress field associated with the subsidence of the Mt. Epomeo block, which represents the stress load mechanism (De Novellis et al., 2018; Trasatti et al., 2019; Carlino et al., 2021) and appears to be active since historical time. In fact, the ground deformation records show that at least the northern coastline experienced subsidence since about 2 ka, with rate of the order of a few millimeters per year (Buchner et al., 1996). Besides, the present measurements–as recorded since at least 30 years from the INGV GPS and leveling surveillance networks and from DInSAR data (Manzo et al., 2006; De Martino et al., 2011; De Novellis et al., 2018)—indicate a general subsidence of the island with a similar rate, between 1 and 11 mm yr⁻¹.

Focusing on the most recent ground deformation patterns, the horizontal velocity field derived from the continuous GPS stations operating on the island (Figure 5) evidences that in the eastern part the horizontal displacements are mainly westward, with rates of about 3 mm year⁻¹ (De Martino et al., 2021). The FORI station, located in western part of the island, outside the caldera structure, shows negligible horizontal trend, indicating a condition of greater stability in this part of the island (De Martino et al., 2021). The horizontal velocity vector at MEPO (2.6 mm year⁻¹), located at the top of Mt. Epomeo, shows velocity pointing toward NNW, while OSCM, in the north part of the island, has a rate of about 3 mm year⁻¹ in direction SW (De Martino et al., 2021). On the other hand, vertical deformations recorded by GPS, DInSAR, and leveling data highlight an increasing subsidence rate going toward the center of the island, with the maximum rate recorded at MEPO (11 mm year⁻¹) (Figure 5 and Table 3).

The GS fault represents the boundary between the main resurgent zone, to the south, presently subsiding, and a relatively less subsiding area along the coast of Casamicciola, to the north. This subsidence trend could have activated extensional structures at the border of the resurgent dome (Walter and Troll., 2001). Furthermore, the overall GPS measurements indicate the sinking of Mt. Epomeo with a slight northward tilt of the block, consistently with Trasatti et al. (2019). This pattern suggests that the maximum shear stress related to the concurrent effect of tilting and differential subsidence must be located at the northern base of the central block, somewhere in between MEPO and OSCM. The GS fault is located between these stations. These observations suggest that during the present subsidence phase the activation of the faults located at the northern base of the Mt. Epomeo block can be favored by the increase of shear stress (Figure 6). Moreover, the main hot springs—located in the Casamicciola area (Bagni-La Rita-Maio) (Figure 6)—are aligned along the GS fault, which likely operates as a fluids discharge zone of the shallow geothermal system (Chiodini et al., 2004; Carlino et al., 2014; Trasatti et al., 2019). This could facilitate fault slipping due to the significant reduction of the effective stress (Sibson, 1981), favoring dislocation on this structure rather than on the southern side of the graben.

In the light of the new data and according to previous models of resurgence at Ischia (Carlino et al., 2006; Sbrana et al., 2009; Carlino, 2012), we suggest the following conceptual scheme of island resurgence (Figure 7): the onset of a magmatic intrusion (occurred after caldera formation) at a few (2–3) kilometers of depth (Sbrana et al., 2009; Paoletti et al., 2013; Di Giuseppe et al., 2017) produced the initial uplift of a large portion of the island, about 10 km in diameter, and the onset of faulting related to the resurgence. As magma injected proceeded and the volume has grown, the intrusion did not extend anymore, due to crystallization of the periphery (Pagie, 1913), and active pushing was exerted by the central part of the intrusion (the present Mt. Epomeo block), as a piston stage (Carlino, 2012). This process generated a pervasive fracturing, while the displacement of the central block occurred along the faulting at its boundary, where volcanic activity took place. Moreover, at Ischia Island, it is observed that the available magmatic energy was largely dissipated through uplift, rather than eruptions (Acocella and Funiciello, 1999). Assuming a minimum magma volume of at least 10 km³ to generate the uplift of the Mt. Epomeo block (Carlino, 2012; Galetto et al., 2017) and giving the erupted volume emitted after the caldera collapse, which is less than 0.1 km³ (Carlino, 2012 and references therein), a very high intrusive/eruptive ratio is found. This observation supports the hypotheses of low mobility of the resident magma and possibly its high viscosity acted as a barrier for new magma to erupt (Galetto et al., 2017). In this case, as suggested by Galetto et al. (2017), magma could be injected at the boundary of the more viscous zone. We suggest that this process occurred in the final stage of resurgence, when the arrival of new magma (Civetta et al., 1991) has migrated along the eastern and northern boundary of the resurgent structure (Sbrana et al., 2018). We also suggest that the onset of subsidence (<5 ka) may have further squeezed out a fraction of the resident magma facilitating its movement toward the surface (Gudmundsson, 2020) and possibly triggering a part of the most recent volcanic activity (phase 6) in the Island. In the northern sector, between the base of the resurgent block and the Grande Sentinella, magma bulging dislocated two blocks, which formed the presently active seismogenic structure of the island. The northern block was uplifted (about 100 m) at a rate of 2 cm year⁻¹ at least, while the southern one subsided by about 50 m at about half rate (∼1 cm year⁻¹). At the end of the resurgence and with the inversion of ground movement of the Mt. Epomeo block, the slow subsidence (a few mm year⁻¹) progressively accumulated vertical shear stress along the border of a sinking zone, between the two GPS stations MEPO and ECSM, which show a differential vertical deformation of about 7.6 mm y⁻¹. The 2017 earthquake was not preceded, neither followed, by any geophysical signals (e.g., deformation, gravity variations, and foreshocks) indicating possible movement of magma at shallow depth (De Novellis et al., 2018; Berrino et al., 2021). Furthermore, the similarity of this event with historical ones at Ischia in terms of distribution of felt reports, and source location, extension, and kinematics (Carlino et al., 2021) points to the same process as origin of the earthquakes, suggesting that no magma movement is involved in the generation of these earthquakes. We conclude that the faster long-term subsidence of the Mt. Epomeo block, resulting in differential vertical deformation, originates recurrent seismic reactivation of the inward dipping GS normal fault. The differential subsidence rate has been explained in terms of different response of a viscous layer to the loading of the overburden (Castaldo et al., 2017) or to the deflation and contraction of a silt-like source below Mt. Epomeo (Galvani et al., 2021).