40Ar/39Ar ages of alkaline and tholeiitic rocks from the northern Deccan Traps: implications for magmatic processes and the K–Pg boundary

The Deccan large igneous province in India was emplaced temporally close to the Cretaceous–Palaeogene (K–Pg) boundary and is formed by tholeiitic flood basalts and less abundant alkaline rocks. Definition of the origin of Deccan magmatism and of its environmental impact relies on precise and accurate geochronological analyses. We present new 40Ar/39Ar ages from the northern sector of the province. In this area, tholeiitic and alkaline rocks were contemporaneously emplaced at 66.60 ± 0.35 to 65.25 ± 0.29 Ma in the Phenai Mata area, whereas rocks from Rajpipla and Mount Pavagadh yielded ages ranging from 66.40 ± 2.80 to 64.90 ± 0.80 Ma. The indistinguishable ages for alkaline and tholeiitic magmatism suggest that distinct mantle sources were synchronously active. The new ages are compared with previous ages, which were carefully screened and filtered and then recalculated to be comparable. The entire dataset of geochronological data does not support a time-related migration of the magmatism related to the northward Indian plate movement relative to the Reunion mantle plume. The main phase of magmatism, including the newly dated rocks from the northern Deccan, occurred at the K–Pg boundary. This suggests a causal link between the emplacement of the province and the K–Pg mass extinction. Supplementary material: Whole-rock and mineral compositions and the complete 40Ar/39Ar dataset are available at https://doi.org/10.6084/m9.figshare.c.2674441.


2015)
. However, many of the published ages are of poor precision or cannot be mutually compared. This latter problem affects in particular 40 Ar/ 39 Ar ages calculated relative to poorly intercalibrated standards or standards that have been demonstrated to be heterogeneous and not suitable for high-precision dating (e.g. MMhb-1, SB-3, LP-6; see Onstott et al. 1991;Spell & McDougall 2003;, and references therein).

Data filtering
To compare the 40 Ar/ 39 Ar ages obtained so far on Deccan samples, they have been recalculated relative to the age of Fish Canyon sanidine (FCs) of 28.294 ± 0.037 Ma and using the decay constants of Renne et al. (2011) calculated following the approach of . It should be noted that recalculated ages would become about 0.2 myr younger if the age (28.1986 ± 0.038 Ma) for the Fish Canyon sanidine standard recently proposed by Wotzlaw et al. (2013) was considered.
A rigorous screen of the available ages has been undertaken and ages satisfying the following criteria have been considered further: (1) age data need to be calibrated with reliable standards (see ); (2) 40 Ar/ 39 Ar plateau ages must be defined by at least 70% of the released 39 Ar and by at least three consecutive steps yielding the same apparent age within 95% confidence level and a probability of fit (P-value) of at least 0.05 (Baksi 2007;; (3) if excess Ar is present, robust 39 Ar/ 40 Ar-36 Ar/ 40 Ar isochron ages (P > 0.05) that take into account the measured isotopic composition of the trapped argon will be considered instead of the plateau model age, which assumes that the initial trapped argon has an atmospheric composition (Harrison et al. 1985). Ages fulfilling these criteria are listed in Table 1, and all other available age  Table 1); U/Pb ages from Schoene et al. (2015) are also shown. The new 40 Ar/ 39 Ar ages from the Phenai Mata, Pavagadh and Rajpipla areas are shown in bold type. (b) Simplified geological map of the Phenai Mata area (after Gwalani et al. 1993), showing sampling localities (squares, alkaline samples; circles, tholeiitic samples) and 40 Ar/ 39 Ar ages. data are not discussed further. All age uncertainties are given at the 2σ level (Table 1). The K/Ar data for plagioclase from basaltic lavas of the Western Ghats sequence suggested apparent ages ranging from 64.5 ± 0.6 to 64.8 ± 0.6 Ma (Chenet et al. 2007). However, the K/Ar method yields ages that are significantly different from 40 Ar/ 39 Ar or U/Pb zircon ages on the same lava flow formations (see below) and it furthermore does not allow a test of the goodness of the results, in terms of detection of excess Ar, or determining in which measure the alteration affected the apparent age. Therefore, we will not consider such K/Ar data further.

Retained ages
The geochronological data suggest that the total Deccan activity lasted some 4 myr. The oldest activity occurred in the north. Two intrusive alkaline complexes from the Cambay graben (Sarnu Dandali and Mundwara) yielded 40 Ar/ 39 Ar ages of 69.62 ± 0.08 Ma and 69.58 ± 0.16 Ma respectively on biotite, making them the likely first continental phase of Deccan magmatism (Basu et al. 1993). Basu et al. obtained an age of 66.04 ± 0.16 Ma on biotite for an olivine gabbro from the Phenai Mata complex, Narmada valley (Gujarat). Further evidence for an early Deccan activity was obtained for alkali-basaltic lava flows from the Anjar Traps (Kutch region), dated at 67.47 ± 0.30-67.67 ± 0.60 Ma on whole-rock and plagioclase separates by Courtillot et al. (2000), and for diamondiferous Mainpur field kimberlites, Bastar Craton (central India), dated at 67.37 ± 0.80 and 62.77 ± 1.40 Ma ( 40 Ar/ 39 Ar on whole-rock; Lehmann et al. 2010).
Most geochronological investigations have focused on the Western Ghats lava pile, the most voluminous and complete sequence in the Deccan Traps. 40 Ar/ 39 Ar analyses have revealed that at least 1700 m of the lava sequence were erupted in a short interval of c. 1 myr (Baksi 1994;Hofmann et al. 2000), with the ages of the bottom (mean age 66.07 ± 0.70 Ma) being indistinguishable from those at the top (65.87 ± 0.4 Ma; Hofmann et al. 2000). Evidence for a relatively short eruption history has been recently confirmed by Schoene et al. (2015), who obtained U/Pb ages on single zircon (2-8 data per sample) ranging from 66.29 ± 0.03 Ma (lowest lava flow unit, Jawhar formation) to 65.53 ± 0.03 Ma (topmost lava flow unit, Mahabaleshwar formation). New 40 Ar/ 39 Ar ages by Renne et al. (2015) on plagioclase are virtually indistinguishable from the U/Pb ages of Schoene et al. (2015) and thus support the same eruptive history for the Western Ghats sequence, lasting from 66.38 ± 0.10 to 65.62 ± 0.08 Ma. In particular, both the U/Pb ages of Schoene et al. (2015) and the 40 Ar/ 39 Ar ages of Renne et al. (2015) suggest that the last two flow units (Ambenali and Mahabaleshwar) were erupted after the K-Pg boundary, dated at 66.04 ± 0.09 Ma by Renne et al. (2013).
Lava flows cropping out along the eastern coast of India in the Rajahmundry traps yielded ages (65.33 ± 0.50 Ma, 40 Ar/ 39 Ar on plagioclase; Knight et al. 2003) similar to those of the upper Western Ghats formations (Ambenali and Mahabaleshwar). Rajahmundry and Western Ghats lavas are also correlated on the basis of their geochemical characteristics and remanent magnetization (Knight et al. 2003). Basaltic lava flows from the Mandla lobe, located on the eastern margin of the main Deccan volcanic province, have been dated by Shrivastava et al. (2015). Those researchers provided a weighted mean 40 Ar/ 39 Ar age for the section (64.42 ± 0.33 Ma) and a geochemical correlation of the Mandla lobe lavas with the uppermost units of the Western Ghats succession (Poladpur-Ambenali-Mahabaleshwar formations). This suggests that the post-K-Pg phase of flood basalt activity erupted over much of the province (Shrivastava et al. 2015).
In general, these geochronological data suggest that the bulk of the Deccan was emplaced between c. 67 and 65 Ma, with the exception of some early alkaline activity in the far north (Cambay graben) at c. 69 Ma. It should be noted that 40 Ar/ 39 Ar ages on plagioclase (typically for tholeiites) yield relatively large errors, which makes it difficult to confirm or exclude any time-related southward migration of the magmatism. A better precision can be achieved for phases included in alkaline rocks, but those are rare and very localized. Palaeomagnetic and biostratigraphic correlations provide additional constraints and indicate that the general evolution of the Deccan volcanism occurred in three distinct phases (Chenet et al. 2009;Keller et al. 2011): a first phase at the boundary between magnetostratigraphic chrons C30r and C30n and covering the northern half of the Deccan; a second phase starting in chron C29r, straddling the K-Pg boundary and constituting about the 80% of the total volume of the province; and a third, waning phase lasting until chron C29n.

Sampling
For the present study, we sampled alkaline and tholeiitic rocks from the northern Deccan. Sampling was focused in the western Narmada rift region, in an area north and west of the Amba Dongar carbonatite complex (Fig. 1b). The Narmada rift is characterized by an east-westtrending tholeiitic dyke swarm cross-cutting the flood basalt sequence and extending across Peninsular India, and by alkaline dykes with various directions (north-south, east-west, NNE-SSW). Intrusive bodies are composed of both alkaline and tholeiitic rocks (essentially gabbros to syenites), such as the Phenai Mata intrusion (Gwalani et al. 1993;Fig. 1b). The Phenai Mata complex shows the association of alkaline rocks and a layered tholeiitic intrusion (Sukheswala & Sethna 1973;Gwalani et al. 1993), and the surrounding areas are mainly constituted by phonolite, lamprophyre and nepheline-syenite, which form plugs and dykes, with ENE-WSW and WNW-ESE trend. Samples of alkaline (PL3) and tholeiitic gabbro (PL9, PL20), as well as a nepheline-syenite (PL2) and a lamprophyric dyke (PL36), have been analysed from this region.
Samples were also collected from the Rajpipla area, where some of the oldest Deccan lavas should be expected (e.g. Chenet et al. 2007). This area shows a succession of early tholeiitic lava flows overlain by K-rich alkaline flows, which form the main exposed sequence, in turn cut by late tholeiitic dykes (Krishnamurty & Cox 1980). The Rajpipla basalt sample PL54 that has been dated here is a tholeiitic lava that belongs to the early phase of this region.
The Pavagadh hill is an outlier cropping out to the north of the Narmada-Tapti rift. The mafic lavas have a peculiar geochemical and isotopic composition having no equivalent in the bulk of the Deccan sequence but resembling some Reunion Island lavas (see Melluso et al. 2006). The Pavagadh section (Melluso et al. 1995;Sheth & Melluso 2008) consists of a 550 m-thick sequence of igneous rocks ranging from alkali olivine basalt (PL61) to rhyolite lavas often with glassy textures (i.e. sample PL63).

40
Ar/ 39 Ar dating was carried out on eight mineral separate samples: four biotites, one amphibole and three plagioclase separates (Table 2). Each sample was carefully hand-picked under the binocular microscope and washed with distilled water and methanol.
Samples were loaded into eight large wells of one aluminium disc of 1.9 cm diameter and 0.3 cm depth (one for plagioclase and one for biotite). These wells were bracketed by small wells that included Fish Canyon sanidine (FCs) used as a neutron fluence monitor, for which an age of 28.294 ± 0.036 Ma (1σ) was adopted (Renne et al. 2011). The discs were Cd-shielded (to minimize undesirable nuclear interference reactions) and irradiated for 40 h in the US Geological Survey nuclear reactor (Denver, CO) in central position. However, it should be noted that, according to Chafe et al. (2014), unshielded irradiation would provide better information on the Cl/K ratio.
The mean J-values computed from standard grains within the small pits range from 0.00337800 ± 0.00000405 (0.12%) to 0.00357500 ± 0.00000501 (0.14%) for the two discs, determined as the average and standard deviation of J-values of the small wells for each irradiation disc. Mass discrimination was monitored using an automated air pipette and provided values from 1.00630 (±0.34%) to 1.00633 (±0.34%) per atomic mass unit relative to an air ratio of 298.56 ± 0.31 (Lee et al. 2006). The correction factors for interfering isotopes were ( 39 Ar/ 37 Ar) Ca = 7.30 × 10 −4 (±11%), ( 36 Ar/ 37 Ar) Ca = 2.82 × 10 −4 (±1%) and ( 40 Ar/ 39 Ar) K = 6.76 × 10 −4 (±32%). The 40 Ar/ 39 Ar analyses were performed at the Western Australian Argon Isotope Facility at Curtin University using the step-heating technique. Amphibole and biotite samples were stepheated using a 110 W Spectron Laser System, with a continuous Nd-YAG (IR; 1064 nm) laser rastered over the sample for 1 min to ensure a homogeneously distributed temperature. Plagioclase samples were loaded in Cu-foil packages (Ar below detection limit) and step-heated using a Pond Engineering® double vacuum resistance furnace.
The gas was purified in a stainless steel extraction line using a GP50 and two AP10 SAES getters and a liquid nitrogen condensation trap. Ar isotopes were measured in static mode using a MAP 215-50 mass spectrometer (resolution of c. 450; sensitivity of 4 × 10 −14 mol V −1 ) with a Balzers SEV 217 electron multiplier. The data acquisition was performed with the Argus program written by M. O. McWilliams and ran under a LabView environment. The raw data were processed using the ArArCALC software (Koppers 2002). Blanks were monitored every three samples. Age data are given with 2σ error and were calculated for an age of 28.294 ± 0.037 Ma for the Fish Canyon sanidine monitor and the 40 K decay constant values recommended by Renne et al. (2011). All uncertainties are included in the calculation following the Monte Carlo simulation error calculation of . Summary data of 40 Ar/ 39 Ar analyses are reported in Table 2.
A control on the composition of the separated crystals was provided by electron microprobe analysis (EMPA). Mineralogical compositions were measured at the IGG-CNR of Padova, by means of a CAMECA SX50 electron microprobe, equipped with four wavelength-dispersive spectrometers. For all analysed phases an acceleration voltage of 15 kV was used, and the beam current was set at 10 nA.

Results
All eight analysed samples yielded robust plateau ages, defined by over 90% of the released gas ( 39 Ar) and by seven or more heating steps. Analytical uncertainties (2σ) are small (<0.4 Ma) for biotite samples, whereas plagioclase samples yielded clearly larger errors (0.8-2.8 Ma) owing to the scarcity of fresh crystals and to their low K/Ca ratio (0.022-0135, measured by electron microprobe). For all samples, plateau ages are indistinguishable from those of the 39 Ar/ 40 Ar-36 Ar/ 40 Ar isochron (where the intercept with the 36 Ar/ 40 Ar axis gives the trapped argon composition) (Fig. 2). The 39 Ar/ 40 Ar-36 Ar/ 40 Ar isochron plots yield 40 Ar/ 36 Ar intercepts that are generally consistent with initial argon being of atmospheric origin, thus indicating that no excess Ar was present. Only the amphibole of sample PL2 has excess Ar ( 40 Ar/ 36 Ar intercept = 324 ± 13, significantly higher than the air value of 295.5), suggesting that for this sample the 39 Ar/ 40 Ar-36 Ar/ 40 Ar isochron age (66.60 ± 0.35 Ma) is more reliable than the plateau age (66.87 ± 0.32 Ma). In general, K/Ca ratios as calculated from 39 Ar/ 37 Ar are consistent with the chemical compositions of the analysed phases as determined by electron microprobe measurements, and the presence of secondary phases such as sericite or adularia (with higher K/Ca) can be ruled out. The four samples from Phenai Mata provided robust plateau (PL3, PL9, PL20; biotite) or 39 Ar/ 40 Ar-36 Ar/ 40 Ar isochron (PL2; amphibole) ages, which are indistinguishable at the 2σ level (from 66.60 ± 0.35 to 66.24 ± 0.37 Ma; Table 2) and yielded a mean age of 66.42 ± 0.17 Ma (MSWD = 0.73; P = 0.53) for the emplacement of the Phenai Mata intrusion. Notably, these ages are indistinguishable from those (66.01 ± 0.11 Ma) obtained by Basu et al. (1993) on a sample equivalent to sample PL9 dated in the present study.

Age of magmatism in the Narmada valley
The new 40 Ar/ 39 Ar data indicate that magmatic activity in the northern Deccan (Narmada valley) continued from 66.60 ± 0.35 to 64.9 ± 0.8 Ma and is generally synchronous with the main phase or with a late phase of Deccan volcanism as well as with some of the Rajahmundry lava flows (Figs 2 and 3). In particular, alkaline and tholeiitic rocks of the Phenai Mata intrusion were emplaced in a short time span at c. 66.4 Ma. Globally, these ages show that the Phenai Mata magmatism was contemporaneous with the onset of Deccan activity from the Western Ghats. In contrast, a slightly younger magmatic phase characterizes alkaline dykes (such as the lamprophyre PL36; 65.25 ± 0.29 Ma) and the Pavagadh complex (64.9 ± 0.8 Ma), which are indistinguishable in age from the late Western Ghats flows (Schoene et al. 2015).
Because our ages for the northern Deccan are indistinguishable from those of the Western Ghats, they are not consistent with a southward migration of the volcanism owing to northward movement of the Indian plate above a fixed Reunion hotspot (Figs 1 and 4). In particular, our ages for the Pavagadh complex and for alkaline magmatism east of Phenai Mata show that late-phase magmatism also occurred in the northern regions of the Deccan province (with alkaline igneous rocks). Therefore, it is suggested that the evolution of the magmatism in this large area reflects a pulsating mantle melting regime with magmatism active for well over 1 myr, rather than a linear evolution of a progressively southward migrating magmatism.
Based on our new data and on the recalculated and filtered 40 Ar/ 39 Ar ages from previous studies, we can provide further constraints for the duration of the Deccan volcanism and its belonging to the magnetic chrons C31-C29. In particular, the recalculated ages have shown that the first phase of alkaline Deccan magmatism (Sarnu and Mundwara complexes in the northern Deccan; Basu et al. 1993) can be placed at the boundary between the magnetic chrons C31r and C31n. We have provided the first age for the Rajpipla magmatism (PL54: 65.86 ± 1.68 Ma), and its maximum age (67.54 Ma) suggests that the activity in the northern Deccan may have started within or after chron C30n (c. 67.5 Ma) and not at the boundary between chrons C30r and C30n (>68 Ma) as previously suggested (Chenet et al. 2008(Chenet et al. , 2009).  Renne et al. (2013) are also shown for comparison. Geomagnetic polarity timescale data are from Cande & Kent (1995), with their ages recalculated after Renne et al. ( , 2011. The analysed Rajpipla basalt (PL54) and Pavagadh rhyolite (PL63; 64.9 ± 0.8 Ma) indicate a maximum and minimum duration of the tholeiitic Deccan magmatism of c. 3 myr and <1 myr, respectively. Conversely, alkaline magmatism lasted c. 4 myr, from 69.62 ± 0.08 Ma (Cambai Graben; Basu et al. 1993) to 65.25 ± 0.29 Ma (PL36), but it should be noted that high-quality age data for the alkaline magmatism are limited to the northern Deccan alkaline complexes.

Implications for the genesis of alkaline and tholeiitic Deccan magmatism
Alkaline samples are a distinctive feature of the Deccan Traps and are often thought to constitute the early and late phase of the volcanism (Basu et al. 1993). In general, alkaline and tholeiitic magmas require different degrees of partial melting and mantle sources (e.g. Simonetti et al. 1998). In a mantle plume scenario, such variations are expected to be time-related, with alkaline magmatism preceding and possibly following the main phase of tholeiitic magmatism such as observed for the Hawai'i hotspot (e. g. Wyllie 1988). However, the geochronological data do not support such a model for the Deccan. Although we cannot discriminate between the age of, for example, Phenai Mata alkaline and tholeiitic rocks, our data show that alkaline and tholeiitic rocks of early (c. 66.4 Ma) and late phase (c. 65 Ma) both occur in the northern Deccan. As both alkaline and tholeiitic rocks have been produced during both early and late phases in a relatively small region, the possibility for them to have formed from the same mantle source is unlikely and the existence of different mantle sources for synchronous alkaline and tholeiitic magmas is required, as already suggested by Simonetti et al. (1998) and Melluso et al. (2002Melluso et al. ( , 2006. Those researchers invoked, besides the Reunion plume, the significant involvement of the subcontinental lithospheric mantle for the generation of Deccan magmatism. Implications for the age of the Deccan magmatism v. the K-Pg boundary One of the most important mass extinctions in Earth's history has been identified at the K-Pg boundary, but its causes are still a matter of debate. Two events are invoked as the principal causes for the extinction, owing to their synchronicity with the K-Pg boundary: the Deccan volcanism and the Chicxulub bolide impact in the Yucatan Peninsula (Alvarez et al. 1980;Schulte et al. 2010a). As discussed above, the main phase of Deccan volcanism occurred at c. 66 Ma with main activity starting some 0.5 myr before and ending after the K-Pg boundary and the Chicxulub impact (dated at 66.04 ± 0.09 Ma and 66.04 ± 0.05 Ma, respectively; Renne et al. 2013;Sprain et al. 2015). As pointed out by Renne et al. (2015) and Richards et al. (2015), a change in eruption style and rate of Deccan volcanism occurred at the K-Pg boundary and was possibly caused by the Chicxulub impact.
The debate about the main causes of the end-Cretaceous extinction is still very lively (Archibald et al. 2010;Courtillot & Fluteau 2010;Keller et al. 2010;Schulte et al. 2010a,b). Stratigraphic and geochronological studies constrain the synchronicity of the Chicxulub impact, marked worldwide by a razor-sharp iridium spike (Alvarez et al. 1980;Smit & Hertogen 1980;Alroy 2008;Schulte et al. 2010a;Renne et al. 2013; but see also contrasting results of Keller et al. 2011). Degassing of the impacted sedimentary strata released 100-500 Gt of SO 2 (Pierazzo et al. 2003), which, combined with the effects of dust release, led to a sudden global cooling of possibly up to 10°C (Pope et al. 1997;Schulte et al. 2010a). These and other effects of the Chicxulub impact (e.g. tsunamis and large-magnitude earthquakes) have been proposed as main devastating causes of the sudden and widespread extinction of the latest Mesozoic fauna and flora (e.g. Vajda et al. 2001;Bown 2005;MacLeod et al. 2007). Other studies have noted, however, a progressively more stressed global environment during the Maastrichtian that heralded the K-Pg turnover before the Chicxulub impact (e.g. Keller et al. 2008Keller et al. , 2009). The Late Maastrichtian events are characterized by global climate instability (abrupt cooling and sea-level drop; Li & Keller 1998;Scheffer et al. 2009), and a decrease of vertebrates (e.g. Barrett et al. 2009) and planktic foraminifera species (e.g. Globotruncanidae) started in chron C30n (Keller et al. 2008(Keller et al. , 2011. It is this early phase of global climate instability and biotic reduction events as well as a delayed recovery during the early Palaeogene that may well be explained by the Deccan volcanism and cannot be attributed to the Chicxulub impact. The recalculated ages and our new data allow us to identify two early magmatic pulses (at c. 69.5 and 67.5 Ma) in the Deccan Traps before the K-Pg boundary, followed by a peak activity straddling the boundary at c. 66 Ma (comprising the Phenai Mata intrusion) and followed by a prolonged late activity (Fig. 3). The two early magmatic pulses at c. 69 and c. 67 Ma should correspond to the magnetic chrons C31n and C30n. This early Deccan magmatism was recorded by a negative 187 Os/ 188 Os excursion in Maastrichtian ocean sediments (Ravizza & Peucker-Ehrenbrink 2003). Moreover, the late phase of Deccan volcanism, defined by our analyses for the northern Deccan (samples such as PL61 and PL36) as well as by the ages for the Mahabaleshwar flows (66.55 ± 0.03 Ma; Schoene et al. 2015) and the Rajahmundry Traps (65.33 ± 0.50 Ma; Knight et al. 2003), could be responsible for the delayed biotic recovery after the K-Pg boundary, as observed in the Danian foraminifera assemblages in the intertrap sediments at Jhilmili and Rajahmundry (Keller et al. 2008(Keller et al. , 2009 and in the proliferation of the disaster opportunist survivor species such as Guembelitria cretacea (Keller 2003;Keller & Pardo 2004). Therefore, the observations that the onset of the end-Cretaceous mass extinction can be placed within chron C30n and thus coincides with the early Deccan volcanism and that a slow recovery after the K-Pg mass extinction may be associated with a slow waning of late Deccan volcanism strongly underline the role of Deccan volcanism in controlling biotic and climatic changes from the end of the Cretaceous into the early Palaeogene. In general, comparison between mass extinctions and flood basalt emplacements in the geological record back to the Cambrian showed that this association is recurrent and is statistically unlikely to be due to random chance (Jourdan et al. 2014). Furthermore, no other impact-extinction pairs have been yet identified in the geological record (Jourdan 2012).

Conclusions
The Deccan Traps are formed by volumetrically dominant tholeiitic rocks and by less abundant, but important alkaline rocks. New 40 Ar/ 39 Ar ages on such rocks cropping out in the northern portion of the province (Narmada valley) provide new constraints on the relationship between the two magma series and on the evolution of Deccan magmatism in general. We have provided a new filtered dataset with the most reliable ages available for the Deccan Traps.
The recalculated 40 Ar/ 39 Ar ages show that the magmatic activity started at the boundary between magnetic chrons C31r and C31n with the emplacement of the Sarnu Dandali and Mundwara complexes, and that Deccan magmatism lasted at least 4 myr. The new data along with previous data show that alkaline magmatism is not confined to the early and late phase of the evolution of the province, but occurred also within the main phase, when the most voluminous lava sequence (the Western Ghats) was erupted, thus pointing towards two distinct mantle sources responsible for synchronous production of alkaline and tholeiitic magmas. Moreover, the distribution of the new data is not consistent with a simple southward migration of the volcanism, as younger rocks (such as Mount Pavagadh) are still well preserved in the northern Deccan. This shows that Deccan magmatism from the Narmada rift to the Western Ghats and the Rajahmundry Traps was essentially synchronous, straddling the K-Pg boundary. In contrast, the most precise data on alkaline samples indicate that ages significantly older than 67 Ma are limited to the northwestern sector of the Deccan (Fig. 4).
In addition, the age of the Deccan Traps suggests a causal link between the emplacement of the province and the K-Pg mass extinction. The onset of the mass extinction can be placed at the boundary between chrons C30n and C29r, together with Deccan volcanism. The Deccan is also compatible with the biotic evolution after the K-Pg boundary and after the Chicxulub impact.