An interpretation of the 1883 cataclysmic eruption of Krakatau from geochemical studies on the partial melting of granite

Pumice flow from the 1883 Krakatau eruption significantly differs in both mineral and chemical compositions from any other volcanic rocks or ejecta of the Krakatau group, which belong to the tholeiitic series. Lithic fragments of granitic Rock, discovered in the pumice flow, are similar to West Malayan granitic rocks. No other granitic rock occurs throughout the Krakatau group, therefore, we consider that the granitic fragments came from the underlying complex at depths, where they were captured as foreign materials by the magma.It is possible that sialic crustal materials plunged into depths along a peculiar tectonic structure located at the Sunda Strait, which appears to be a sheared portion caused by deformation of the Sunda arc due to differential movement between the Indo-Australian oceanic plate and the Eurasian continental crust. The crustal materials were partially melted and produced a magma of granitic composition. The magma was mixed with or assimilated by an ascending basaltic magma originating probably from the upper mantle. This resulted in a dacitic magma distinctly dominant in silica, alkalis and volatile components, and the 1883 Krakatau eruption, characterized by the pumice flow of dacitic composition, took place.     

Emplacement of a Debris Avalanche during the 1883 eruption of Krakatau (Sunda Straits,Indonesia)

Conclusion The data collected during the Mentawai cruise help to clarify understanding of the 1883 eruption of Krakatau. We have previously discussed the weaknesses of the interpretation of Williams (1941) and others (Self and Rampino 1981) and emphasized that only a Mount St. Helens-type collapse during the course of the eruption could account for all the characteristics of the eruption and of the related deposits.The discovery on land of deposits attributable to a debris-avalanche, in the stratigraphic position where they were expected, is a strong argument for the validity of our scenario.Marine surveys confirm that the sea bottom around Krakatau is covered by a thick ignimbritic deposit. But the presence of this deposit does not invalidate the presence of a debris-avalanche deposit under the ignimbrites. The hummocky morphology favours this hypothesis.Flank-failure of volcanoes is generally considered as a very efficient mechanism for triggering tsunamis (Kienle et al. 1987; Siebert et al. 1987). However, the majority of the volcanoes where flank-failure has been described are tall and bulky and the collapse of a broad edifice like Krakatau may be surprising. However the geological evidence shows that such a mechanism can act at various scales; for example the flank collapse of Mayu Yama volcano (height 700 m, volume 0,3 km³), a parasitic cone of Unzen volcano (Japan), triggered a debris-avalanche into the sea that was 1 km long, with a characteristic hummocky surface; the resulting tsunami killed 9528 people (Katayama 1974). In the same way, a partial collapse of Iliwerung volcano, Indonesia (50 × 10⁶ m³) in July 1979, triggered a tsunami which killed several hundred people (McClelland et al. 1989). At Krakatau, the main summit was 822 m asl; the collapse took place along the edge of the prehistoric caldera and this structural unconformity probably facilitated the triggering of the process.     

The Krakatau islands: The geotectonic setting

The Sunda Strait is located in a transitional zone between two different modes of subduction, the Java frontal and Sumatra oblique subductions. Western Java and Sumatra are, however, geologically continuous.The Krakatau complex lies at the intersection of two graben zones and a north-south active, shallow seismic belt, which coincides with a fracture zone along this seismic belt with fissure extrusion of alkali basaltic rocks commencing at Sukadana and continuing southward as far as the Panaitan island through Rajabasa, Sebuku and Krakatau.Paleomagnetic studies suggest that the island of Sumatra has been rotating clockwise relative to Java from at least 2.0 Ma to the present at a rate of 5–10h/Ma, and therefore the opening of the Sunda Strait might have started before 2 Ma (Nishimura et al. 1986).From geomorphological and seismological studies, it is estimated that the western part of Sumatra has been moving northward along the Semangko fault and the southern part of Sunda Strait has been pulled apart.Assuming that the perpendicular component (58 mm/yr; Fitch 1972) of the oblique subduction has not changed, we can estimate that the subduction started at 7–10 Ma. Huchon and LePichon (1984) also estimated that the subduction started at 13 Ma.Recent crustal earthquakes in the Sunda Strait area are clustered into three groups: (1) beneath the Krakatau complex where they are typically of tectonic origin, (2) inside a graben in the western part of the strait, and (3) in a more diffuse zone south of Sumatra. The individual and composite focal mechanisms of the events inside the strait show an extensional regime. A stress tensor, deduced from the individual focal mechanisms of the Krakatau group shows that the tensional axis has a N 130°E orientation (Harjono et al. 1988).These studies confirm that the Sunda Strait is under a tensional tectonic regime as a result of clockwise rotation along the continental margin and northward movement of the Sumatra sliver plate along the Semangko fault zone.     

The birds of Anak Krakatau: the assembly of an avian community

The avian colonization of Anak Krakatau, an emergent volcanic island, is described for the period 1952–1991 with emphasis on the resident land birds. The few species that had colonized the island by 1952 were destroyed in a volcanic eruption. Recolonization has been rapid and extensive but only documented by regular surveys that began in the 1980s. The waves of avian colonists associated with the following stages of plant succession are described: pre-vegetation, beach and grassland pioneers, casuarina woodland and mixed forest. Rainforest frugivores first reached the island in 1985 coinciding with the fruiting of the first fig trees. The three stages of plant succession and the bird communities associated with them co-exit on three separate lowland sites with about a 10–15 year delay between them. The guilds of shorebirds, predators, nectarivores, frugivores and insectivores are described. Densities of insectivores are very high; however between 1986 and 1990, some generalist species have fallen in abundance whereas some specialists have increased. The establishment species have fallen in abundance whereas some specialists have increased. The establishment of new colonists, especially pigeons, has been limited by the arrival of two species of falcons.     

Historic eruptions of Tambora (1815), Krakatau (1883), and Agung (1963), their stratospheric aerosols,and climatic impact

Decreases in surface temperatures after the eruptions of Tambora (1815), Krakatau (1883), and Agung (1963) were of similar magnitude, even though the amount of material (dust and volatiles) injected into the stratosphere by these three events differed greatly. Large amounts of fine ash and volatiles were dispersed into the upper atmosphere by Krakatau and Tambora; the Agung eruption in 1963 was a much smaller, vulcanian-type eruption which injected dust and volatiles into the stratospheric aerosol layer more directly. Analyses of magmatic volatiles indicate that the Agung eruption was proportionately richer in SO₂ and Cl than either Tambora or Krakatau. Relative amounts of fine ash produced by the Tambora, Krakatau, and Agung eruptions are estimated at about 150:20:1, whereas the masses of atmospheric sulfate aerosols produced were on the order of 7.5:3:1.Decreases in surface temperature of a few tenths of a degree C for several years following volcanic eruptions are primarily a result of the sulfate aerosols, rather than of the silicate dust. The similarity in the atmospheric response after these three eruptions supports the idea of limiting mechanisms on volcanic stratospheric-aerosol loading, which is suggested by microphysical processes of aerosol particles. Fluctuations in stratospheric aerosol optical depth seem to be controlled to a large degree by high-intensity sulfur-rich eruptions (e.g., Agung, 1963), which may however be relatively small in total ejecta volume. Such eruptions leave little geologic record, but appear as acidity peaks in polar ice cores.     

The varve-related ice-dammed lake events in Glen Roy and vicinity: a new interpretation

This paper presents a new interpretation of the sequence of events in Glen Roy and vicinity during the Loch Lomond Stadial that can be inferred from a detailed varve record constructed by Palmer et al. (2010). 300 years of Younger Dryas glacier advance in the Scottish Highlands are recorded by very thin varves formed in an ice-dammed lake up to 35 km long. At a varve site now occupied by Loch Laggan the lake stood permanently at 260 m, but in Glen Roy varves were also laid down in a lake at 325 m and, later, 350 m caused by glacier advance. Initial ice retreat recorded by a gradual increase in varve thickness was soon followed by much thicker varves. The varve sequences are interrupted by a sand bed caused by sudden drainage of the 350 m lake. The major varves of the Glen Roy sequence show that storminess was still increasing in intensity at least 160 years after glacier retreat had begun. At the Loch Laggan site 15 cm of deformed sediments register an earthquake that produced 3 m faulted uplift of all three Glen Roy shorelines, a response to the abrupt removal of 5 km³ of water when the 260 m lake was catastrophically drained by jökulhlaup. The deformed sediments are immediately followed by varves deposited in a local lake, ice-dammed lake sedimentation now having ceased, having lasted more than 460 years.     

The great 1950 Assam Earthquake revisited: Field evidences of liquefaction and search for paleoseismic events

Extensive field investigations were carried out for the first time in the meizoseismal area of the great 1950 Assam Earthquake aimed at exploring the paleoseismic history of the NE Indian region through documentation of liquefaction features and radiocarbon (¹⁴C) dating. Trenching at more than a dozen locations along the Burhi Dihing River valley and within the alluvial fans adjoining the Brahmaputra and Dibang Rivers resulted in the identification of more than a dozen very prominent liquefaction features (sand dykes, sills, sand blows etc.) as evidences of large to great earthquakes. ¹⁴C dating of the organic material associated with some of the features indicates a paleoseismic record of about 500 yrs archived by the sediments in this region. Compelling geological evidence(s) of the great 1950 earthquake are well constrained by ¹⁴C dating. Out of the two historically reported seismic events (1548 AD and 1697 AD) from this region, ¹⁴C dating could constrain the 1548 AD event though not distinctly. Further studies using combined ¹⁴C and OSL dating may better constrain the seismo-chronology of the study region.     

Colonization of the Krakatau islands by vertebrates

80 species of non-marine, non-migrant vertebrates, excluding also shore birds, have been recorded from the Krakatau archipelago since the explosive eruption of 1883.73 of these are believed to have established breeding populations and successfully colonized the islands (47 resident land birds species, 13 bats, 11 reptiles and 2 rats). A large proportion of these colonizers consists of species with wide distributions, broad ecological tolerances and/or association with urban or rural situations. 11 of the 73 colonizing species (8 resident land birds, 2 reptiles and a bat) have become extinct on the islands. No amphibian species has yet been recorded and successful non-volant mammal colonists consist only of rats.Species turnover is considered to be largely associated with successional change of habitat. Progressive loss of coastal habitat through marine erosion and of open inland habitats through forest formation and canopy closure has been important and volcanic activity is indicated in the case of one or two extinctions.Active flight has been the predominant dispersal method (63 of the recored species, 79%, are birds or bats). Of the 17 recorded non-volant vertebrates, 6 species are assessed as having dispersed by swimming, 4 by means of natural rafts and 7 most likely through human agency (on boats). Human-assisted dispersal is seen to be of increasing importance with the increasing number of tourist visits and this may prove to be significant in determining the nature and rate of future colonization by vertebrates.     

Colonisation of the Krakatau islands by invertebrates

Arrival of invertebrates on the Krakatau islands, and the development of invertebrate assemblages there since 1883, are reviewed. Many data are imprecise, and difficulties of interpretation are outlined. A few well-known groups, such as butterflies, are used to suggest that much colonisation so far has been by relatively generalist species,and that the rate of addition of new species may become slower as the less vagile specialists characteristic of later successional vegetation may find natural colonisation harder. Studies of the assemblages developing in association with particular plant species are likely to be helpful in assessing future trends of invertebrate colonisation.     

Crustal volatile release at Merapi volcano; the 2006 earthquake and eruption events

High‐temperature gas in volcanic island arcs is widely considered to originate predominantly from the mantle wedge and from subducted sediments of the down‐going slab. Over the decade (1994–2005) prior to the 2006 eruption of Merapi volcano, summit fumarole CO₂ gas δ¹³C ratios are relatively constant at ?4.1 ± 0.3‰. In contrast, CO₂ samples taken during the 2006 eruption and after the May 26th 2006 Yogyakarta earthquake (M6.4) show a dramatic increase in carbon isotope ratios to ?2.4 ± 0.2‰. Directly following the earthquake (hypocentre depth 10–15 km), a 3–5‐fold increase in eruptive intensity was observed. The elevated carbon isotope gas data and the mid‐crustal depth of the earthquake source are consistent with crustal volatile components having been added during the 2006 events, most probably by the thick local limestone basement beneath Merapi. This ‘extra’ crustal gas likely played an important role in modifying the 2006 eruptive behaviour at Merapi and it appears that crustal volatiles are able to intensify and maintain eruptions independently of traditional magmatic recharge and fractionation processes.     

A cascade of magmatic events during the assembly and eruption of a super-sized magma body

We use comprehensive geochemical and petrological records from whole-rock samples, crystals, matrix glasses and melt inclusions to derive an integrated picture of the generation, accumulation and evacuation of 530 km³ of crystal-poor rhyolite in the 25.4 ka Oruanui supereruption (New Zealand). New data from plagioclase, orthopyroxene, amphibole, quartz, Fe–Ti oxides, matrix glasses, and plagioclase- and quartz-hosted melt inclusions, in samples spanning different phases of the eruption, are integrated with existing data to build a history of the magma system prior to and during eruption. A thermally and compositionally zoned, parental crystal-rich (mush) body was developed during two periods of intensive crystallisation, 70 and 10–15 kyr before the eruption. The mush top was quartz-bearing and as shallow as ~3.5 km deep, and the roots quartz-free and extending to >10 km depth. Less than 600 year prior to the eruption, extraction of large volumes of ~840 °C low-silica rhyolite melt with some crystal cargo (between 1 and 10%), began from this mush to form a melt-dominant (eruptible) body that eventually extended from 3.5 to 6 km depth. Crystals from all levels of the mush were entrained into the eruptible magma, as seen in mineral zonation and amphibole model pressures. Rapid translation of crystals from the mush to the eruptible magma is reflected in textural and compositional diversity in crystal cores and melt inclusion compositions, versus uniformity in the outermost rims. Prior to eruption the assembled eruptible magma body was not thermally or compositionally zoned and at temperatures of ~790 °C, reflecting rapid cooling from the ~840 °C low-silica rhyolite feedstock magma. A subordinate but significant volume (3–5 km³) of contrasting tholeiitic and calc-alkaline mafic material was co-erupted with the dominant rhyolite. These mafic clasts host crystals with compositions which demonstrate that there was some limited pre-eruptive physical interaction of mafic magmas with the mush and melt-dominant body. However, the mafic magmas do not appear to have triggered the eruption or controlled magmatic temperatures in the erupted rhyolite. Integration of textural and compositional data from all available crystal types, across all dominant and subordinate magmatic components, allow the history of the Oruanui magma body to be reconstructed over a wide range of temporal scales using multiple techniques. This history spans the tens of millennia required to grow the parental magma system (U–Th disequilibrium dating in zircon), through the centuries and decades required to assemble the eruptible magma body (textural and diffusion modelling in orthopyroxene), to the months, days, hours and minutes over which individual phases of the eruption occurred, identified through field observations tied to diffusion modelling in magnetite, olivine, quartz and feldspar. Tectonic processes, rather than any inherent characteristics of the magmatic system, were a principal factor acting to drive the rapid accumulation of magma and control its release episodically during the eruption. This work highlights the richness of information that can be gained by integrating multiple lines of petrologic evidence into a holistic timeline of field-verifiable processes.     

Beloc, Haiti, revisited: multiple events across the KT boundary in the Caribbean

Examination of new expanded K/T boundary sections near Beloc, Haiti, reveals deposition of a glass spherule-rich deposit (SRD) and two (PGE) anomalies (one Ir-dominated and one Pd-dominated) during the early Danian Parvularugoglobigerina eugubina Zone [Pla(l)]. The presence of the Haiti SRD within the early Danian is interpreted as being due to reworking. Ir is only slightly elevated within the SRD but forms an anomaly at the top of the SRD extending into the overlying pelagic limestones. It is unclear at present whether this Ir anomaly results from mechanical reworking of an impact at the K/T boundary, or an additional impact event in the early Danian. The second PGE anomaly upsection is dominated by Pd and Pt and is more compatible with a magmatic origin. This suggests a multi-event scenario consistent with one (and possibly two) impact(s), followed by a PGE-enriched volcanic event in the Caribbean.     

A new interpretation of the sedimentary environment before and during eruption of the Emeishan LIP,Southwest China

The environment where the Permian Emeishan large igneous province (LIP) of Southwest China erupted remains controversial, especially regarding whether it was terrestrial, involving a 1 km scale domal uplift, or submarine. Slightly younger Daqiao conglomerate and Binchuan pillow lavas suggest that the Emeishan LIP erupted in a submarine environment. We show that at Binchuan, sandstone and rhyolite lie beneath the pillow lavas. In the Daqiao cross-section, there is an eastwards-verging syncline that reverses the succession of basalt and conglomerate. The conglomerate is not a basal conglomerate, and it does not contain any magmatic hydrovolcanic deposits. The basalt underlying the conglomerate is not the first of the LIP eruptions; that first eruption is found ~420 m below, on top of the Permian Maokou limestone. All together, these observations show that the deposits, including the conglomerate and pillow lava, do not represent the environment at the very start of the LIP volcanic eruptions, but represent conditions that existed before and possibly during the Emeishan LIP eruptions. Based on field investigations, the petrology of the rocks, and structural features, we conclude that submarine sedimentation and subaerial basalt eruptions coexisted in time and space in the region during or prior to the Emeishan LIP basalt eruptions.     

Ecological aspects of plant colonisation of the Krakatau Islands

Species assemblage data from the Krakatau Islands, Indonesia, are presented for the period 1883 to 1989 (including previously unpublished data from the 1989 survey). Since 1934, 16 additional families of higher plants have colonised. Recent arrivals at the family level are mostly of zoochorous species of forest tree, indicating (subject to the effects of disturbance) a continuing increase in potential niche space within the island interiors. The data for Rakata (an uninterrupted prisere) conform to a successional explanation in which identifiable ecological groups of plants exhibit differing colonization and turnover patterns. Animal-dispersed canopy tree species and species which are widespread within the group, exhibit a very low probability of extinction once they have colonized successfully. There are, however, several constraints on the rapid spread of species within the group, in particular those connected to local dispersal (eg lack of large terrestrial mammals). In respect of dispersal to the group, partial survey data for the island of Sebesi from 1921 (revised) and 1989 provides the basis for comment as to the changing biogeographical circumstances of the Sunda Straits and the role of Sebesi as a stepping stone island.The varied data discussed in the paper indicates that with the exception of the strand-line, no component of the Krakatau flora or vegetation has yet approached a stable composition. Both floral and faunal diversification are argued to be proximally controlled not only by dispersal opportunities but also by the dynamics of the dominant life-forms of the system, ie, the forest trees. Such hierarchical links, across trophic boundaries, should receive greater recognition in the construction of island biogeographical theory.     

Development of the ecosystem on surtsey with references to Anak Krakatau

There is a strong similarity in the formation and size of the two new volcanic islands: Anak Krakatau, Indonesia (emerging 1930) and Surtsey, Iceland (emerging 1963). They are both piled up from an ocean floor of over 100 m depth, forming islands of cinder and lava covering more than 2 km² in area. They are both extremely valuable for various ecological studies, ie of investigation of means of dispersal and colonization. However, due to their tropical and subarctic locations, respectively, there is an extreme difference in climate of the two islands and thus in the kind of available life forms, their means of colonization and rate of growth.The first vascular plant was found on Surtsey in 1965, bryophytes became established in 1967 and lichens were first detected in 1970. The advancement of the vascular plant colonization has been of greatest importance in the establishment of vegetation on the island. Twenty-five species of vascular plants, all herbaceous, had been recorded on Surtsey in 1990, the dominant species being Honkenya peploides. Less than 1% ofthe island has vegetation cover, root biomass is low and the organic carbon and nitrogen status of the soil is poorly developed. Birds started nesting on Surtsey in 1970 and six species, all depending on marine life for food, had nested on the island in 1990. The land-invertebrate fauna has developed slowly on Surtsey and the island is devoid of vertebrates other than birds.     

The 1983 Mount Etna eruption: thermochemical and dynamical inferences

Thermochemical calculations and laboratory phase equilibration experiments on lavas of the 131 day 1983 Mt. Etna flank eruption of 0.1 km³ were undertaken to investigate possible systematic variations in inferred melt-phenocryst equilibration conditions as a function of time. The 1983 Mt. Etna lavas are multiply saturated; plagioclase, clinopyroxene and olivine, the dominant phenocrysts, occur in the ratio 1:1/2:1/4. Melts (glasses) plot close to the plagioclase saturated olivine-clinopyroxene low pressure cotectic on a Walker-O'Hara diopside-forsterite-silica diagram suggesting equilibration of melt and phenocrysts in a high level magma reservoir. Total pressures, temperatures and dissolved H₂O concentrations were calculated using the isoactivity method of Carmichael and coworkers based on about 300 elelctron microprobe analyses of coexisting olivine, clinopyroxene and plagioclase phenocrysts, microphenocrysts and groundmass microlites for samples collected 6, 46 and 125 days after the start of the eruption. Total pressures (P _t), temperatures and H₂O contents based on representative olivine-clinopyroxene pairs are 140 MPa, 1105°C, 2.4 wt% H₂O; 255 MPA, 1112°C, 1.0 wt% H₂O and 85 MPa, 1096°C, 1.8 wt% H₂O respectively for the early (283), middle (I83) and late (L83) samples. Corresponding equilibration depths are in the range 3 to 10 kilometers. Plagioclase feldspar phenocrysts, while showing more evidence of disequilibrium, provide compatible estimates of P _t and T when analysis is restricted to the low anorthite mode of the plagioclase frequency-composition histograms: 133 MPa and 1115°C; 260 MPa and 1117°C and 103 MPa and 1104°C, repectively for 283, I83 and L83. The pre-eruptive (i.e., in situ) temperature-pressure gradient calculated from olivine-clinopyroxene equilibria is 10.6 K/kbar. This compares well with independent estimates of the temperature-pressure derivative of the (pseudo) invariant point composition (10 to 12 K/kbar) in both model (e.g., diopside-forsterite-anorthite, Presnall et al. 1978) and natural (e.g., Walker et al. 1979; Grove et al. 1982) systems. Apparently, magma within the Etna reservoir was in a quasiequilibrium state buffered by its multiply-saturated character immediately preceding eruption. The temporal variation of computed P _t, T and H₂O concentrations for melt-phenocryst equilibrium agrees well with predictions based on simulations of the withdrawal of magma from a body zoned with respect to dissolved H₂O provided the temporal record of magma discharge is taken into account. Discharge varied by a factor of about 100 during the sample collection interval. The intermediate P _t but high H₂O content inferred for sample 283 reflects the withdrawal of H₂O enriched magma during an early phase of high average discharge of about (350 m³/s) before evaculation isochrons became quasistationary. The high P _t and relatively dry I83 magma reflects the deepening of the evacuation isochrons after 50 days of intermediate discharge with the development of quasi-stationary isochrons in time and space. Sample L83 from day 125 near the end of the eruption reflects the shoaling of evacuation isochrons (hence low P _t and relatively high H₂O content) associated with the observed low (0.5 m³/s) discharge. Our results show that thermochemical modeling efforts provide important opportunities for testing the predictions of magma with-drawal simulations.     

塔里木溢流玄武岩的喷发特征

通过对柯坪地区二叠系野外火山岩露头剖面和英买力、哈拉哈塘井区二叠系火山岩钻井剖面的对比,将塔里木早二叠世溢流玄武岩划分为三个旋回,从老到新依次是:库普库兹满溢流玄武岩旋回(KP),长英质火山碎屑岩旋回(FP)和开派兹雷克溢流玄武岩旋回(KZ)。KP旋回以巨厚溢流玄武岩夹凝灰岩为特征,在柯坪露头区和英买力井区均可划分出三层巨厚玄武质熔岩流,至哈拉哈塘井区减少为一层玄武岩流,但长英质火山碎屑岩和熔岩厚度增加。FP旋回在柯坪露头区自下而上包括空落相凝灰岩,熔结凝灰岩,再沉积火山碎屑岩和正常碎屑岩夹火山灰层,该层可与英买力及哈拉哈塘井区的凝灰岩层对比,表明在塔北存在一期面积广泛的长英质火山喷发。KZ旋回以溢流玄武岩为主,在开派兹雷克剖面识别出四期喷发共8层溢流玄武岩和一期安山质玄武岩,每期喷发之间夹少量碎屑岩,但未见长英质火山碎屑岩夹层,该特征与英买力和哈拉哈塘井区的火山层序组合不同,而与塔中溢流玄武岩类似。三个火山旋回的划分表明塔里木大火成岩省经历了"基性溢流玄武岩-酸性火山碎屑岩-基性溢流玄武岩"的演变过程,与Afro-Arabian溢流玄武岩省相似,可进行对比研究。     

Recent plumbing system of the Krakatau volcano revealed by teleseismic earthquake distribution

Spatial and temporal analysis of global seismological data 1964–2005 reveals a distinct teleseismic earthquake activity producing a columnar-like formation in the continental wedge between the Krakatau volcano at the surface and the subducting slab of the Indo-Australian plate. These earthquakes occur continuously in time, are in the body-wave (m _b) magnitude range 4.5–5.3 and in the depth range 1–100 km. The Krakatau earthquake cluster is vertical and elongated in the azimuth N30°E, suggesting existence of a deep-rooted fault zone cutting the Sunda Strait in the SSW-NNE direction. Possible continuation of the fault zone in the SW direction was activated by an intensive 2002/2003 aftershock sequence, elongated in the azimuth of N55°E. Beneath the Krakatau earthquake cluster, an aseismic gap exists in the Wadati-Benioff zone of the subducting plate at the depths 100–120 km. We interpret this aseismic gap as a consequence of partial melting inhibiting stress concentration necessary to generate stronger earthquakes, whereas the numerous earthquakes observed in the overlying lithospheric wedge beneath the volcano probably reflect magma ascent in the recent plumbing system of the Krakatau volcano. Focal depth of the deepest events (~100 km) of the Krakatau cluster constrains the location of the primary magma generation to greater depths. The ascending magmatic fluids stress fault segments within the Sunda Strait fault zone and change their friction parameters inducing the observed tectonic earthquakes beneath Krakatau.