The Antarctic Peninsula hosts one of the major Mesozoic-Cenozoic continental magmatic arcs of the circum-Pacific, which extends almost continuously for c. 1350 km along the length of the peninsula (Pankhurst 1982;Leat et al. 1995;Millar et al. 2002). Arc rocks were emplaced from the Late Triassic (Leat et al. 1995;Bastias et al. 2020) to the Miocene (e.g. Leat et al. 1995;Jordan et al. 2014), with a surge in magmatic volume during the Cretaceous (Leat et al. 1995;Riley et al. 2018;Jordan et al. 2020). These arcrelated rocks intrude late Paleozoic-Triassic sedimentary sequences in Graham Land (e.g. Pankhurst 1983;Castillo et al. 2016) and Triassic metamorphic orthogneisses (Millar et al. 2002;Flowerdew et al. 2006) associated with an active margin in Palmer Land (Bastias et al. 2020;Riley et al. 2020b).

Globally, Cretaceous arc magmatism is considered to represent a significant Phanerozoic phase of growth of the continental crust (e.g. Kemp et al. 2009;Ducea et al. 2015). Regionally, the Cretaceous is also characterized by significant deformation along much of the western margin of Gondwana (e.g. Vaughan and Livermore 2005;Bryan and Ferrari 2013), which was accompanied by a global plate reorganization event (Matthews et al. 2012). Arc magmatism along the Antarctic Peninsula formed during eastdipping subduction of Pacific oceanic lithosphere beneath the Antarctic plate in southwestern Gondwana (Fig. 1a; e.g. Pankhurst 1990;Burton-Johnson and Riley 2015;Bastias et al. 2021a).

Evidence of arc magmatism includes individual plutons, composite intrusions and extensive batholith-like units (e.g. Leat et al. 1995). Previous studies of the Cretaceous peak of arc magmatism in the Antarctic Peninsula have focused on their episodicity (Riley et al. 2018) and magmatic-tectonic relationships (Burton-Johnson et al. in press), although the regional processes that controlled and triggered a higher rate of magma addition to the crust remain unclear.

The aim of this study is to further constrain the magmatic and tectonic evolution of the Cretaceous arc of the Antarctic Peninsula by providing new geochronological, geochemical and isotopic data, and integrating our data with the timing of magmatism, subduction architecture and tectonic history presented in previous studies. We present 36 new zircon U-Pb (crystallization) dates (obtained using laser ablation inductively coupled plasma mass spectrometry; LA-ICP-MS), and geochemical (whole-rock) and isotopic (whole-rock Nd, Sr, Pb; zircon Hf ) data acquired from Cretaceous igneous units exposed in the Antarctic Peninsula (Graham and Palmer Land; Fig. 1b). These are complemented by mid-temperature (>350°C) apatite U-Pb thermochronology data obtained by LA-ICP-MS. The igneous rocks crop out in remote locations that were sampled during two field seasons as part of this study, or samples were provided by the British Antarctic Survey archive and the Byrd Polar and Climate Research Center, USA. We combine our data with previous results from the Antarctic Peninsula (Riley et al. 2001(Riley et al. , 2016(Riley et al. , 2018(Riley et al. , 2020a;;Ryan 2007;Leat et al. 2009;Haase et al. 2012;Bastias 2014Bastias , 2020;;Bastias et al. 2019Bastias et al. , 2020Bastias et al. , 2021a, b;, b;Burton-Johnson et al. in press).

The continental crust of the Antarctic Peninsula was interpreted by Suárez (1976) to represent an autochthonous segment of an extensive continental arc that spanned Mesozoic western Gondwana. Alternatively, other researchers have suggested that the Antarctic Peninsula crust formed by collision and accretion of an allochthonous arc with a block of Gondwanan affinity either during the middle of the Cretaceous (Vaughan and Storey 2000) or earlier, close to the Jurassic-Triassic boundary (Vaughan et al. 2012). More recently, Burton-Johnson and Riley (2015) and Bastias et al. (2020) provided substantial geochronological and isotopic evidence for an autochthonous to parautochthonous Mesozoic evolution, with subduction initiation during the Late Paleozoic, supporting the initial interpretation of Suárez (1976).

Arc magmatism along the Antarctic Peninsula occurred from at least the Late Triassic (Pankhurst 1982;Bastias et al. 2020;Riley et al. 2020b), when it formed a segment of the Terra Australis margin of western Gondwana (e.g. Cawood 2005). Subsequently, a Late Triassic active margin was associated with the Rymill Granite Complex (Bastias et al. 2020), which is mostly composed of orthogneisses and is widely exposed in the central Antarctic Peninsula and formed during an extensional tectonic regime that modified the entire Pacific margin of west Gondwana (Spikings et al. 2016). Early Jurassic arc magmatism was concentrated in Palmer Land (Riley et al. 2017;Bastias et al. 2021a) and shifted to Graham Land during the Middle-Late Jurassic (e.g. Bastias et al. 2021a). Most of the Jurassic magmatic rocks in the Antarctic Peninsula have been linked to the influence of an active margin and perhaps the migration of the Karoo mantle plume from southern Africa as well (Pankhurst et al. 2000;Riley et al. 2001). However, Bastias et al. (2021a) recently suggested that most Jurassic magmatism probably formed within an active margin that was characterized by a flat-slab setting.

Cretaceous magmatic rocks are exposed along the west coast of Graham Land and in eastern Palmer Land (Fig. 2a). Magma addition rates to the Antarctic Peninsula peaked during the Early Cretaceous (e.g. Leat et al. 1995Leat et al. , 2009;;Riley et al. 2018), which is considered to be the most voluminous episode of Phanerozoic plutonism in the Antarctic Peninsula (e.g. Leat et al. 1995Leat et al. , 2009;;Vaughan et al. 2012;Riley et al. 2018;Burton-Johnson et al. in press). These rocks range from mafic to felsic compositions, but are predominantly intermediate, and the published radiometric dates (K-Ar, Rb-Sr, 40 Ar/ 39 Ar and U-Pb methods) range between c. 141 and c. 67 Ma (Leat et al. 1995).

Early Cretaceous volcanic rocks are abundant along the west coast of the Antarctic Peninsula and the South Shetland Islands (Fig. 2a; e.g. Leat et al. 1995). Most of these are basalts and andesites with a calc-alkaline affinity (e.g. Haase et al. 2012). The Early Cretaceous plutonic record includes a few exposures on the South Shetland Islands (c. 137-109 Ma;Hervé et al. 2006;Bastias et al. 2019) and in SE Palmer Land, where the Lassiter Coast Intrusive Suite is exposed (Fig. 2a; Rowley et al. 1983;Riley et al. 2018;Burton-Johnson et al. in press). These rocks are mainly tonalite, quartz diorite and granodiorite and crop out over an area of c. 80 000 km 2 (Fig. 2a; Burton-Johnson et al. in press). Zircon U-Pb concordia dates suggest that the Lassiter Coast Intrusive Suite was emplaced in three pulses at 130-126, 118-113 and 108-102 Ma (Riley et al. 2018). Burton-Johnson et al. (in press) further resolved the central episode of magmatism with secondary pulses at c. 118-116, c. 114-112 and c. 110-109 Ma, which are consistent with the model of Paterson and Ducea (2015). Pankhurst et al. (1991) and Leat et al. (2009) obtained whole-rock 87 Sr/ 86 Sr i and εNd i values from the Early Cretaceous volcanic rocks of the Antarctic Peninsula, which span 0.7080-0.7056 and -0.4 to -1.8, respectively, from rocks that yielded Rb-Sr ages of c. 132-116 Ma.

Late Cretaceous igneous rocks are less widely exposed (Fig. 2a; e.g. Leat et al. 1995), with isolated calc-alkaline granodioritic plutons associated with Late Cretaceous mafic dykes in the central Antarctic Peninsula (Wever et al. 1994;Leat et al. 1995). Most of the mafic dykes yield calc-alkaline compositions, although some ocean island basalt-like mafic dykes occur locally (Leat and Riley 2021), which Leat et al. (1995) utilized to infer an extensional setting. Ryan (2007) obtained zircon LA-ICP-MS U-Pb concordia dates that range between c. 92 and 89 Ma, and 87 Sr/ 86 Sr i and εNd i values that span 0.7047-0.7045 and 2.5-2.2, respectively, for plutons exposed in west-central Graham Land (Fig. 2a). Leat et al. (2009) reported Rb-Sr isochron ages between c. 96 and c. 71 Ma for volcanic rocks located within the central-western Antarctic Peninsula, which yield whole-rock 87 Sr/ 86 Sr i and εNd i values of 0.7112-0.7056, and 0.0 to -1.8, respectively. Recently, Riley et al. (2020a) published U-Pb zircon dates ranging between c. 94 and 64 Ma from felsic volcanic rocks located in northwestern and central Palmer Land and suggested that arc magmatism migrated trenchward during the Late Cretaceous along the southern Antarctic Peninsula, which is consistent with previous work (e.g. Thomson and Pankhurst 1983;Leat et al. 1995). In contrast, the locus of magmatism in Graham Land does not appear to change with respect to the trench throughout the Cretaceous.

An extensional regime dominated the Antarctic Peninsula during the Cretaceous, as evidenced by the emplacement of large volumes of magmatic rocks as plutons and dykes, accompanied by normal faults (Meneilly et al. 1987;Wever et al. 1994;Leat et al. 1995). However, the Cretaceous period was also associated with brief compressional events at c. 138, c. 113 and c. 107-100 Ma in the Antarctic Peninsula (e.g. Meneilly 1988;Leat et al. 1995;Vaughan and Storey 2000;Riley et al. 2020a). These compressional events have been linked either to increases in the subduction convergence rate (at c. 138 and c. 113 Ma; e.g. Riley et al. 2020a) or to the collision of an allochthonous terrane (c. 107-100 Ma Vaughan and Storey 2000). The latter collisional event has been disputed in more recent studies (Burton-Johnson and Riley 2015;Bastias et al. 2020). A more extensive compressional episode along western Gondwana has been proposed (e.g. Vaughan 1995;Mpodozis et al. 2005;Vaughan and Livermore 2005;Burton-Johnson and Riley 2015;Spikings et al. 2015;Boyce et al. 2020).

We present new zircon U-Pb geochronological, geochemical and isotopic data from 36 igneous rocks of the Antarctic Peninsula (Fig. 1). The rocks were taken from the Lassiter Coast Intrusive Suite in eastern Palmer Land, and from intrusions that are scattered along the west coast of central Palmer Land at the latitude of the Black Coast. Additional plutonic rocks were collected from the west coast of Graham Land and the South Shetland Islands. The methods used in this work are described in detail in the Supplementary material.

We present U-Pb zircon concordia ages from 36 igneous rocks that vary in age between 139 ± 1 and 79 ± 1 Ma (Table 1; Fig. 2a). Early Cretaceous ages were obtained from 33 samples, and range from 139 ± 1 to 101 ± 1 Ma, with no particular geographical trend (Fig. 2a). Three samples yield Late Cretaceous ages from 92 ± 1 to 79 ± 1 Ma and are located along the west coast of the Antarctic Peninsula. The kernel density estimate (KDE) of the 206 Pb/ 238 U concordia ages yields several peaks during the Cretaceous, and these have been separated into three groups to facilitate the presentation of the data. These groups are (1) a cluster of older ages at c. 140-132 Ma (Berriasian-Valanginian-Hauterivian), (2) a younger group that includes several KDE peaks from 126 to 100 Ma (Barremian-Albian) and ( 3) the youngest group with Late Cretaceous ages spanning c. 92-79 Ma (Turonian-Campanian; using the International Chronostratigraphic Chart time scale of Cohen et al. 2013;Fig. 2b). Cathodoluminescence images (see examples in Fig. 2c) show that most zircons have patchy or oscillatory zonation, mostly with no clear rim-core relationships; all are typical of igneous zircon (e.g. Chelle-Michou et al. 2014).

Six granodiorites and monzogranites of the Lassiter Coast Intrusive Suite yield 206 Pb/ 238 U concordia dates of c. 126-112 Ma (Fig. 2a), consistent with previous studies (c. 130-102 Ma; Pankhurst et al. 1991;Riley et al. 2018;Burton-Johnson et al. in press). Further north, seven samples from the west coast of Palmer Land yield zircon 206 Pb/ 238 U concordia dates of c. 140-79 Ma (Fig. 2a), whereas five monzogranites, syenogranites, tonalites and quartz-monzonites are Early Cretaceous in age (c. 139-114 Ma) and two syenogranites have Late Cretaceous ages (92 ± 1 and 79 ± 1 Ma). At the latitude of Adelaide Island (c. 67°S), eight alkali granites, granodiorites and quartz-monzonites yield Early Cretaceous dates that span c. 118-101 Ma (Fig. 2a). Further north, seven alkali granites, syenogranites, monzogranites and monzodiorites from the west coast of southern Graham Land also yield Early Cretaceous ages that span c. 135-101 Ma, whereas a gabbro yields a Late Cretaceous date of 81 ± 1 Ma (Fig. 2a). The northernmost Cretaceous 206 Pb/ 238 U concordia dates were obtained from three monzogranites and a granodiorite from the South Shetland Islands, which span c. 138-101 Ma, and are consistent with previous age estimates of volcanic and intrusive rocks from that region (Hervé et al. 2006;Bastias 2014;Israel 2015;Bastias et al. 2019). These data show that Early Cretaceous rocks occur extensively along the Antarctic Peninsula from the SE in Palmer Land to the NW in Graham Land, whereas Late Cretaceous plutons are less abundant, and crop out along the west coast of the central and northern Antarctic Peninsula (Fig. 2a). The complete dataset is presented in Supplementary material Table 1.

Apatite U-Pb LA-ICP-MS dates have been obtained from the Cretaceous rocks that were used for U-Pb zircon LA-ICP-MS geochronology in this study, and also from Late Triassic (Bastias et al. 2020) and Jurassic igneous rocks (Bastias et al. 2021a; Table 2; Fig. 3) previously dated by the U-Pb zircon LA-ICP-MS method. For each sample, single spot analyses are plotted in Tera-Wasserburg space where they represent a mixture of radiogenic ( 206 Pb) and initial (common) lead ( 207 Pb/ 206 Pb). In the case of rapid cooling and no subsequent perturbation of the U-Pb system, a sample containing a suite of cogenetic crystals with a large spread in radiogenic Pb/common Pb ratios (Fig. 4) can be used to define a well-constrained linear array in Tera-Wasserburg space (e.g. Kirkland et al. 2018). However, this assumes U-Pb concordance in the sample (e.g. Petrus and Kamber 2012) and requires a significant spread in the radiogenic Pb/common Pb ratios to ensure a well-constrained linear array (Kirkland et al. 2018). Linear regressions through the data from each sample yield 28 238 U/ 206 Pb (Tera-Wasserburg concordia lower intercept) dates that range between 147 ± 15 and 76 ± 25 Ma, whereas the All the ages were obtained by LA-ICP-MS U-Pb analysis.

207 Pb/ 206 Pb initial ratio was constrained by the Tera-Wasserburg concordia upper intercept. Several results have large uncertainties (up to ±26 Ma) owing to the high common to radiogenic Pb ratios in the apatites and a resultant small spread in U/Pb ratio with some analyses clustering close to the y-axis of the Tera-Wasserburg concordia. Nevertheless, most apatites have relatively high U concentrations (>20 U ppm) and yield useful uncertainties that are lower than ±7 Ma (Table 1).

A comparison of the zircon U-Pb crystallization age and the apatite lower intercept 206 Pb/ 238 U age (Fig. 4) reveals an approximately linear 1:1 correlation for rocks with zircon U-Pb crystallization ages younger than c. 156 Ma, suggesting that these apatite U-Pb ages record rapid cooling (thermal relaxation) following magmatic crystallization. On the other hand, rocks that crystallized before c. 156 Ma yield apatite 238 U/ 206 Pb lower intercept ages of c. 147-128 Ma (Fig. 4). Significantly, four of the five samples that yield apatite Tera-Wasserburg lower intercepts with MSWD > 2 come from rocks that crystallized at c. 153 Ma and older. The origin of this elevated dispersion is discussed in the section 'Thermal histories of Mesozoic arc crust through 550-380°C'.

Whole-rock major oxides, trace element and REE concentrations have been determined in the same 36 rocks that were dated using the zircon U-Pb method (Supplementary material Table 1). The majority of the Cretaceous plutonic rocks are classified as alkali granite to granodiorite in the cationic scheme of de La Roche et al.

(1980; Fig. 5a), although a few diorites, gabbros and olivine gabbros were also sampled. The Early Cretaceous intrusions span the calcic and alkali-calcic differentiation trends on the modified alkali-lime index of Peacock (1931;Fig. 5b). The two Late Cretaceous rocks have calc-alkaline to alkali-calcic compositions (Fig. 5b). The Cretaceous rocks yield aluminium saturation indices (ASI) (Maniar and Piccoli 1989) that straddle the metaluminous-peraluminous fields, and range between 0.74 and 1.61 (Fig. 5c), with no relationship to crystallization age. Normal mid-ocean ridge basalt (N-MORB) normalized trace element abundances (Fig. 5d) reveal no distinct changes through the Cretaceous, with an enrichment in large ion lithophile elements (LILE), and negative Nb, Ta and Ti anomalies, suggesting a subduction-derived component in the magma source regions, and that they may have formed within a continental arc (Fig. 5d). Minor negative Ba, Eu and Sr anomalies, combined with a strong negative Ti anomaly, suggest that plagioclase and Fe-Ti oxides have fractionated, and the positive Pb anomaly is probably derived from an upper crustal source. Trace element concentrations of the Cretaceous rocks normalized to average upper continental crust scatter close to unity (Fig. 5e), supporting a significant crustal origin for the Cretaceous rocks. Tectonic discrimination using (Y + Nb) v. Nb/Y (Whalen and Hildebrand 2019) supports an arc setting for the Cretaceous rocks (Fig. 5f ), which is consistent with direct comparisons of Y and Nb (Fig. 5g; Pearce et al. 1984). A comparison of Sr/Y v. Y (Fig. 5h) shows that the rocks that formed at c. 140-132 and c. 92-73 Ma plot in the fields of volcanic arc and adakite, respectively, whereas the rocks that formed during c. 126-100 Ma straddle these two fields.

A lack of temporal trends in La n /Yb n , Sr/Y and Eu/Eu* (Fig. 6) through the Cretaceous suggests that the crustal thickness of the arc did not significantly change during this period (e.g. Hildreth and Moorbath 1988;Mantle and Collins 2008;Chiaradia 2015;Profeta et al. 2015). However, we acknowledge that estimates of the thickness of continental arc crust using geochemical indices are problematic owing to the multiple petrogenetic processes that occur in such settings (Ducea et al. 2015), and these should perhaps only be used as qualitative indicators (e.g. Kay and Mpodozis 2001;Best et al. 2009;Oliveros et al. 2019).

The 87 Sr/ 86 Sr i and εNd i values of the Cretaceous intrusions (139 ± 1 to 79 ± 1 Ma) range between 0.7100 and 0.7040 and +4.1 and -9.7, respectively (Fig. 7). These values reveal no significant trends with time, although εNd i and 87 Sr/ 86 Sr i values span a wider range between c. 139 and 112 Ma, compared with the rocks that crystallized between c. 112 and 79 Ma (Fig. 7a-c). These data are consistent with the results of Leat et al. (2009), who reported from a smaller dataset: (1) 87 Sr/ 86 Sr i ratios of 0.70634-0.70563, and εNd i values of -0.4 and -0.5 from a rock that yields a Rb-Sr isochron date of 132 Ma ± 9 Ma, (2) 87 Sr/ 86 Sr i ratios of 0.7080-0.7061 and a εNd i value of -1.8 from a rock that yields a Rb-Sr isochron date of 116 ± 2 Ma, and (3) 87 Sr/ 86 Sr i ratios of 0.7112-0.7044 and εNd i values ranging between 2.6 and 0.0, from rocks that yield K/Ar dates spanning c. 96-71 Ma (Fig. 7a-c). These results yield similar age and Sr isotopic data to previous work (Pankhurst 1982;Pankhurst et al. 1991). In addition, Ryan (2007) reported 87 Sr/ 86 Sr i ratios between 0.7047 and 0.7045 and εNd i values ranging between 2.5 and 2.3 from rocks that yield U-Pb zircon dates of 92 ± 1 and 89 ± 1 Ma (Fig. 7a-c).

Whole-rock Pb isotopic compositions of the granitoids that intruded during 139 ± 1 to 79 ± 1 Ma are 7d ande), and plot between the upper crust and orogenic curves of Zartman and Doe (1981). There are no significant variations throughout the Cretaceous, although the granitoids that formed during c. 126-101 Ma are slightly depleted in 208 Pb relative to 206 Pb, compared with the older and younger Cretaceous intrusions, and thus they plot slightly closer to upper crustal compositions (Fig. 7e).

In situ Hf isotopic compositions of zircon have been determined from a suite of the same Cretaceous zircons that were dated using the U-Pb method. Similar to the Nd, Sr and Pb isotopes, there is no systematic variation of εHf t through the Cretaceous, which varies between +9.3 and -6.4 (Fig. 8). Two Early Cretaceous granitoids (c. 139-136 Ma) from the northwestern coast of Palmer Land yield εHf t values that range from 7.9 to 6.4 and -2.4 to -4.8, respectively. Eleven Aptian-Albian intrusions (126 ± 1 to 101 ± 1 Ma), located in southeastern Palmer Land to northwestern Graham Land and the South Shetland Islands (Fig. 8), yield εHf t values that range from 9.3 to -6.4. Finally, three Late Cretaceous intrusions (92 ± 1 to 79 ± 1 Ma), yielded εHf t values that range between 7.9 and -2.2.

The zircons that crystallized during c. 139-132 Ma yield a large range of Lu-Hf model ages (TDM Hf ), which range between 1.07 and 0.36 Ga. Similarly, younger intrusions that formed during 126 ± 1 to 101 ± 1 Ma yield TDM Hf ages that range between 1.14 and 0.27 Ga, whereas the model ages of the Late Cretaceous intrusions lie between 0.89 and 0.32 Ga. The full dataset is presented in Supplementary material Table 4.

Early Cretaceous igneous rocks crystallized between 139 ± 1 and 101 ± 1 Ma, and crop out in eastern Palmer Land, the west coast of northern Palmer Land and along the west coast of Graham Land, and within the South Shetland Islands (Fig. 2a). Late Cretaceous intrusions are more spatially restricted and have been identified along the west coast of the central and northern Antarctic Peninsula, and crystallized between 92 ± 1 and 79 ± 1 Ma. Our compilation (Supplementary material Table 3) of 85 well-constrained crystallization ages (79 U-Pb concordant zircon and six 40 Ar/ 39 Ar wholerock plateau dates) shows that Cretaceous arc magmatism peaked at c. 118-110 Ma, with minor peaks at c. 137-136, c. 103-102, c. 94-93, c. 81-80 and c. 71-70 Ma (Fig. 2b), which are consistent with the results of Riley et al. (2020a) and Jordan et al. (2020). However, peak magmatism at c. 118-110 Ma revises previous suggestions that Cretaceous magmatism peaked at c. 142 Ma (Leat et al. 1995) or c. 141-129 Ma (Vaughan et al. 1998).

Our new geochemical data are consistent with a continental arc setting for the Cretaceous igneous rocks of the Antarctic Peninsula. An arc interpretation is supported by (1) the geographical distribution of intrusions that are generally parallel to the margin (Fig. 2a), ( 2) enriched N-MORB normalized LILE and light REE (LREE), with negative Nb, Ta and Ti anomalies (Fig. 5d), which are typically associated with slab-dehydration reactions at active margins, and (3) whole-rock Nd, Sr and Pb (Fig. 7a, b, d ande), and zircon Hf isotopic compositions (Fig. 8) of Cretaceous igneous rocks that show that the magmas formed from mixed sources within the continental crust, which is common in continental arc settings (e.g. Stern 2002). This is consistent with previous interpretations (e.g. Leat et al. 1995;et al. 2018, 2020a) of the Cretaceous intrusions in the Antarctic Peninsula. Our results also reveal no temporal geochemical or isotopic (Nd, Sr, Pb and Hf) trends through the Cretaceous.

Igneous rocks that yield zircon U-Pb concordia ages between c. 156 ± 1 and c. 81 ± 1 Ma yield lower intercept apatite concordia dates ( 206 Pb/ 238 U) on Tera-Wasserburg plots that are indistinguishable from their zircon 238 U-206 Pb concordia ages (Fig. 4), suggesting that they cooled to below c. 380°C rapidly after magmatic intrusion. In contrast, apatites from older intrusions that yield zircon U-Pb concordia dates between c. 217 and c. 184 Ma yield significantly younger apatite lower intercept concordia dates ( 206 Pb/ 238 U), consistently c. 147 to c. 128 Ma. These apatite dates are interpreted to record cooling through the Pb-in-apatite partial retention zone (c. 550-380°C; e.g. Cochrane et al. 2014; following the Pb-in-apatite diffusion parameters of Cherniak et al. 1991) during c. 147-128 Ma. The older plutonic samples (zircon dates between c. 217 and c. 184 Ma) yield apatite lower intercept ages with slightly elevated MSWD values compared with the apatite intercept ages from the younger igneous rocks (zircon dates between c. 156 ± 1 and c. 81 ± 1 Ma). This slight dispersion is considered to be a consequence of (i) partial diffusive loss of Pb during cooling through the Pb-in-apatite partial retention zone, and ( 2) laser sampling of intra-grain regions that were more retentive (e.g. cores of large grains) or less retentive (e.g. small grains or rims of large grains) of Pb.

The older rocks that yield U-Pb zircon ages of c. 217 to c. 184 Ma and apatite U-Pb ages of c. 147-128 Ma form parts of intrusions that are dispersed over a distance of c. 400 km (Fig. 3), and thus reveal a regional magmatic emplacement event and subsequent cooling that affected most of the Antarctic Peninsula. Poblete et al. (2011) reported significant remagnetization of pre-Jurassic igneous and sedimentary rocks on the Antarctic Peninsula, which would require temperatures of at least 500°C (Hunt et al. 1995). This same event was probably responsible for heating the intrusions that formed during c. 217 and c. 184 Ma to temperatures higher than the apatite Pb partial retention zone, which then subsequently cooled through c. 550-380°C at c. 147-128 Ma. The cause(s) of the heating event is unclear, although we hypothesize that it may be due to burial and high heat flow. Jurassic turbidite-like deposits exposed along the west coast of the Antarctic Peninsula in Adelaide and Alexander Island (Riley et al. 2012) and South Shetland Islands (Bastias et al. 2019) may be evidence for the formation of depocentres and burial during this period.

The narrow spread in U-Pb apatite dates (147-128 Ma) from plutons that crystallized during 217-185 Ma suggests that these record a general phase of Early Cretaceous cooling through the apatite Pb partial retention zone (e.g. Cochrane et al. 2014), which may be a consequence of tectonic exhumation. However, the driving mechanism for Early Cretaceous tectonic exhumation of the Antarctic Peninsula is unclear and our interpretation is speculative. The Early Cretaceous Andean margin experienced rock uplift and erosion during the Early Cretaceous during westward migration of South America, induced by the opening of the South Atlantic (e.g. Mpodozis and Ramos 1989). Additionally, ridge formation is also recorded in the Weddell Sea (e.g. König and Jokat 2006) and Rocas Verdes (e.g. Calderón et al. 2007), located in the South Atlantic and Patagonia, respectively. Tectonic exhumation related to the opening of the Atlantic Ocean has been already suggested for the western margin of Patagonia during the Mesozoic (Homovc and Constantini 2001). Furthermore, Gianni et al. (2018Gianni et al. ( , 2020) ) reported Early Cretaceous compression in Patagonia, and Sarmiento and Rangel (2004), Martin-Gombojav and Winkler (2008), Villagómez and Spikings (2013) and Spikings et al. (2015) reported compression and exhumation of the Colombian and Ecuadorian margin during c. 120-110 Ma, leading to the emplacement of HP-LT complexes (e.g. Raspas Complex; Spikings et al. 2015). It is noteworthy that HP-LT rocks are also exposed along the South Shetland Islands (Smith Island; e.g. Grunow et al. 1992) within the Antarctic Peninsula, and they reside in the same structural position to the west of the Mesozoic intrusions as they do elsewhere in the Andes. Therefore, it is feasible to suggest, pending future work, that these HP-LT rocks were exhumed during Early Cretaceous compression.

Summarizing, we suggest that the small variation in apatite U-Pb dates in plutons that crystallized during c. 217 to c. 184 Ma reflects Late Jurassic-Early Cretaceous followed by tectonic exhumation caused by the early opening of the Atlantic Ocean. This event is not recorded by the younger plutons (c. 156-81 ± 1 Ma) because these were probably cooler than the apatite Pb partial retention zone prior to the exhumation event.

The general uniformity in whole-rock geochemical compositions (Figs 5 and6) suggests that there were no significant changes in tectonic setting during c. 140 and c. 79 Ma. The exhumation phase identified by the apatite U-Pb data may have been slow, and did not result in significant changes in crustal thickness. The intrusions are mostly calcic and alkali-calcic alkali-granite and granodiorites that formed in a continental subduction-zone setting (Fig. 5). Exceptions to the general geochemical uniformity of the igneous rocks throughout the Cretaceous consist of the identification of adakite-like magmas in the interval c. 126-100 Ma, which immediately followed a period of elevated compression and higher exhumation rates (Fig. 5f), and higher magma addition rates from c. 118 to 100 Ma (Fig. 2b).

A comparison of the Hf-isotopic compositions reveals significant differences in the source regions of Cretaceous (this study; Zheng et al. 2018;Riley et al. 2020a) and pre-Cretaceous zircons (Bastias et al. 2020(Bastias et al. , 2021a)). Zircons from Ordovician to Triassic plutons show a steady trend towards less radiogenic εHf t values with time (Fig. 9a; Bastias et al. 2020). With the exception of a Late Jurassic Zartman and Doe (1981) for upper crust and orogen. Approximate composition of EM1 (enriched mantle with recycled lower continental crust), EM2 (enriched mantle with upper continental crust and continental derived sediments) and DMM (depleted MORB-mantle) are from Hanan and Graham (1996) and Stracke et al. (2003). HIMU, high U/Pb mantle composition. NHRL, Northern Hemisphere Reference Line (Dupre and Alle gre 1980).

granite (sample R.5957.3;Bastias et al. 2021a, b) that yielded εHf t values between 7.8 and 5.6, Triassic and Jurassic zircons yield broadly similar εHf t values that range between 0.6 and -9.2 (Fig. 9a; Bastias et al. 2020Bastias et al. , 2021a, b), b). These trends contrast with the Cretaceous zircons, which yield more radiogenic εHf t values that range between 12.5 and -6.5, revealing the involvement of more radiogenic source regions in the Early Cretaceous. With the exception of granite R.5957.3, Ordovician to Jurassic intrusions yield a consistent range of Lu-Hf model ages that span 1.37 Ga (at c. 212 Ma; Bastias et al. 2020) to 0.78 Ga (at c. 440 Ma; Bastias et al. 2020), with the majority ranging between 1.2 and 0.8 Ga (Fig. 9c). This suggests that the Ordovician-Jurassic arc magmas mainly incorporated juvenile Sunsas-aged crust (1.19-0.92 Ga; Cordani and Sato 2000) that is exposed in South America and crust of similar age that is exposed in East Antarctica (1.1-1.0 Ga; Goodge and Fanning 2016). Cretaceous zircons yield Lu-Hf model ages that range between 1.14 Ga (at c. 112 Ma) and 0.28 Ga (at c. 106 Ma), whereas the majority of Cretaceous zircons yield Lu-Hf model ages of <0.8 Ga (Fig. 9b), suggesting that the structure of the crust may have been modified after the Jurassic, reducing the volume proportion of material derived from Precambrian basement (Fig. 9d).

The combination of U-Pb zircon dates, geochemical and isotopic data, and mid-temperature thermochronological constraints is consistent with an extensional regime during the Late Jurassic-Early Cretaceous, which may have been interrupted by a mild compressional phase that exhumed the pre-c. 184 Ma intrusions during c. 147-128 Ma. Evidence for compressive pulses at c. 107 and c. 103 Ma (Vaughan and Pankhurst 2008;Vaughan et al. 2012;Riley et al. 2020a) has been accounted for by an increase in plate convergence rates. First, Hf-isotopic compositions of zircons show that more isotopically juvenile crust was incorporated into Cretaceous magmas compared with older intrusions, which is in agreement with the interpretation of Pankhurst (1982) based on Sr isotopes. Extension may have driven decompression of the underlying mantle, promoting the incorporation of mantle melts into the arc magmas (e.g. Cochrane et al. 2014). Second, remagnetization of pre-Jurassic igneous and sedimentary rocks (Poblete et al. 2011), along with resetting of apatite U-Pb dates (via diffusive Pb loss) was probably caused by increased heat flow and burial during extension (e.g. Lachenbruch et al. 1994). Finally, synchronous extension has been recorded in adjacent regions of the Antarctic Peninsula throughout most of the Cretaceous, which was related to the break-up of Gondwana. Oceanic lithosphere formed in the Weddell Sea at c. 147 Ma, outboard of the northeastern Antarctic Peninsula (Fig. 1a; König and Jokat 2006), which led to the opening of the Southern Atlantic. An extensional setting is also documented along the western margin of Patagonia, where it formed the Rocas Verdes Basin, a marginal back-arc basin that developed oceanic lithosphere during the Late Jurassic to Cretaceous (e.g. Dalziel et al. 1974;Dalziel 1981;Calderón et al. 2007). Moreover, most of the South American western margin was dominated by an extensional setting during the Late Jurassic-Early Cretaceous (e.g. Atherton and Aguirre 1992;Mpodozis and Allmendinger 1993;Morata and Aguirre 2003;Spikings et al. 2015).

The period between c. 148 and 140 Ma is characterized by magmatic quiescence along the entire length of the Antarctic Peninsula. Most Late Paleozoic-Jurassic plate reconstructions show subduction of Pacific oceanic lithosphere beneath the western margin of Gondwana at this time (e.g. Meert and Lieberman 2008;Nelson andCottle 2017, 2018) ), although the margin was continuously active throughout the Early Cretaceous, corroborating previous studies (e.g. Leat et al. 1995;Riley et al. 2018). Early Cretaceous intrusions include the Lassiter Coast Intrusive Suite in eastern Palmer Land, and the west coast of southern Graham Land in the Black Coast sector. Further north, arc magmatism from c. 140 to 100 Ma occurs along the west coast of Graham Land up to its northernmost exposure in the South Shetland Islands (Fig. 2a). This geographical distribution of Early Cretaceous igneous rocks from c. 140 to 100 Ma could be accounted for by either flat-slab, east-dipping subduction of the Phoenix Plate, westdipping subduction of the lithosphere of the Weddell Sea, or an allochthonous origin for the rocks of Alexander Island, each of which is discussed below.

Flat slab. Most Early Cretaceous palaeogeographical reconstructions show east-dipping subduction of oceanic lithosphere of the Phoenix Plate beneath the Antarctic Peninsula (e.g. Barker 1982;Larter and Barker 1991;Sutherland and Hollis 2001;Jordan et al. 2020). Assuming that the rocks of Alexander Island are autochthonous to the Antarctic Peninsula and there was no significant strike-slip displacement along the margin, this implies a distance of c. 700 km between the trench and arc rocks of the Lassiter Coast Intrusive Suite (Fig. 10a), which have a trench-parallel extent of at least c. 300 km. Consequently, in this scenario the rocks of the Lassiter Coast Intrusive Suite (c. 130-102 Ma; Pankhurst et al. 1991;Riley et al. 2018) formed above a flattened slab (Fig. 10a). To the north of the Black Coast, igneous rocks were emplaced along the west coast of the Antarctic Peninsula, suggesting that the extent of the flat slab was constrained to a segment of Palmer Land (Fig. 10a).

West-dipping subduction. Arc rocks of the Lassiter Coast Intrusive Suite of eastern Palmer Land may have formed within an active margin associated with west-dipping subduction of lithosphere of the Weddell Sea beneath eastern Palmer Land (Fig. 10b). This hypothesis was proposed by Grunow (1993), who suggested that counterclockwise rotation of the Antarctic Peninsula during the Jurassic in conjunction with the general southward motion of East Antarctica may have driven west-dipping subduction. However, this hypothesis is inconsistent with palaeogeographical reconstructions based on seafloor magnetic anomalies in the Weddell Sea region (e.g. Ghidella et al. 2002Ghidella et al. , 2007;;Jokat et al. 2003;König and Jokat 2006), and the accommodation of lithosphere that may have been subducted during c. 140-100 Ma. Extension in the South Atlantic leading to the development of seafloor spreading and the formation of the Weddell Sea occurred during the Late Jurassic (e.g. Ghidella et al. 2002); most of this seafloor is still present to the east of the Scotia Plate (Fig. 2a; Ghidella et al. 2002Ghidella et al. , 2007;;König and Jokat 2006), and thus poses a problem when trying to account for subduction east of Palmer Land. However, this argument is not conclusive, considering that the age of significant portions (e.g. the southern sector) of the Weddell Sea remains unknown; therefore west-dipping subduction remains a possibility.

An allochthonous origin for Alexander Island. The present sinuous spatial trend of Early Cretaceous magmatism (c. 140-100 Ma) may be a consequence of post c. 100 Ma bending and rotation of the peninsula, implying that the original Early Cretaceous arc was parallel to the ocean-continent interface, with an approximately similar, trench-parallel arc-trench gap. This hypothesis also requires that Alexander Island is part of an allochthonous or parautochthonous crustal block that was located elsewhere prior to c. 100 Ma, and that the Early Cretaceous trench would have been located close to the present position of George VI Sound (Fig. 10c). In this model, Alexander Island arrived close to its present position after c. 100 Ma. However, Alexander Island hosts Mesozoic forearc sequences (Fossil Bluff Group; Butterworth et al. 1988), which have been chronologically and lithostratigraphically correlated with Mesozoic forearc sequences further north in Adelaide Island (Riley et al. 2012) and in the South Shetland Islands (Bastias et al. 2019). Thus, an allochthonous origin for Alexander Island suggests that it may have been accreted along the west Antarctic margin via displacement along the Eastern Palmer Land Shear Zone (Fig. 2a), which is a major ductile to brittle-ductile shear zone with a lateral extent of at least 1500 km (Vaughan and Storey 2000;Vaughan et al. 2012;Fig. 2). Translation of Alexander Island and emplacement close to its current location would have occurred after c. 100 Ma, although U-Pb zircon and 40 Ar/ 39 Ar biotite dates of syntectonic intrusions within this shear zone suggest that it was active during c. 106-102 Ma (Vaughan et al. 2002a, b), and there is no evidence for displacement since c. 102 Ma.

Exposures of Late Cretaceous igneous rocks are less voluminous than the Early Cretaceous units (Fig. 2a) and occur along the west coast of the central and northern Antarctic Peninsula (Fig. 11). Late Cretaceous igneous rocks in Graham Land formed with similar trench-arc distances at its west coast. However, arc magmas migrated westward during the Late Cretaceous-Paleogene at the latitude of northern Alexander Island in Palmer Land (Pankhurst 1982;Storey et al. 1996;McCarron and Millar 1997;Riley et al. 2020a). The absence of Late Cretaceous or Paleogene intrusive rocks at the latitudes of the Lassiter Coast Intrusive Suite suggests that subduction had ceased at c. 112 Ma in more southerly latitudes, which probably signals subduction of the last remaining oceanic lithosphere of the Phoenix Plate under the margin of the Antarctic Peninsula (e.g. Barker 1982;Larter and Barker 1991).

(1) Arc magmatism resumed in the Antarctic Peninsula at c. 140 Ma following a magmatic hiatus during the interval c. 148-140 Ma, forming abundant intrusions that are exposed along Graham and Palmer Land. Magmatism was continuous until c. 79 Ma, with the main peak of activity at c. 118-110 Ma, which represents one of the main periods of Mesozoic magmatism in the Antarctic Peninsula. This magmatism was formed within a continental active margin setting, which is supported by (a) the trench-parallel distribution of the igneous rocks, (b) chemical compositions that reveal an enrichment in LILE and LREE, with negative Nb, Ta and Ti anomalies, which are typical of slabdehydration reactions and thus active margins, and (c) whole-rock Nd and Sr and zircon Hf isotopic compositions revealing mixed sources that resided within the continental crust.

(2) Apatite U-Pb dates show that intrusions that crystallized during c. 217 and c. 184 Ma within the Antarctic Peninsula cooled through the Pb-in-apatite partial retention zone (c. 550-380°C) during c. 147-128 Ma. First, these intrusions were probably heated to temperatures hotter than c. 550°C via increased flow and burial during Late Jurassic-Early Cretaceous extension, which also remagnetized the pre-Jurassic igneous and sedimentary rocks of the Antarctic Peninsula. Subsequent cooling may have been a consequence of exhumation driven by compression and rock uplift, caused by the westward migration of South America during the early opening of the South Atlantic.

(3) An overall extensional setting during the Cretaceous is supported by (a) progressively more radiogenic εHf compositions of Cretaceous zircons revealing the incorporation of mantle melts during attenuation of the crust and (b) evidence for high Late Jurassic to Cretaceous heat flow that magnetized pre-Jurassic rocks and reset apatite U-Pb ages of pre-Middle Jurassic rocks. However, this extensional period was punctuated by compressive events, the most pronounced of which may have been at the beginning of the Early Cretaceous (see conclusion (2)).

(4) Early Cretaceous exposures of arc rocks crop out from the east to the west coast at the latitude of the Black Coast in Palmer Land. This spatial trend may be due to either (a) continuous subduction beneath the western margin of the Antarctic Peninsula with a flat-slab episode in Palmer Land, or (b) east-dipping subduction of the Phoenix Plate in Graham Land and west-dipping subduction in Palmer Land, or (c) an active western margin with east-dipping subduction of the Phoenix Plate, along with an allochthonous or parautochthonous origin for Alexander Island. Our current dataset is unable to distinguish between these possibilities, and testing these hypotheses would require a better understanding of the geological history of the pre-Cretaceous rocks of Alexander Island and the inaccessible areas of the Weddell Sea in the Filchner-Ronne Ice Shelf sector.

(5) The geochemical and isotopic compositions of the Late Cretaceous arc rocks are extremely similar to their Early Cretaceous counterparts. Therefore, we suggest that active margin magmatism was continuous throughout the Cretaceous in the Antarctic Peninsula, which is consistent with previous studies (e.g. Leat et al. 1995;Riley et al. 2020a). Late Cretaceous arc magmatism occurred during the closing stages of subduction of the oceanic lithosphere of the Phoenix Plate in Palmer Land, which gradually ceased moving northward throughout the Cenozoic (e.g. Barker 1982;Larter and Barker 1991).