Coexisting calc-alkaline and ultrapotassic magmatism at Monti Ernici, Mid Latina Valley (Latium, central Italy

Page 1

Eur. J. Mineral. Fast Track ArticleFast Track DOI: 10.1127/0935-1221/2007/0019-1754

Coexisting calc-alkaline and ultrapotassic magmatism at Monti Ernici, MidLatina Valley (Latium, central Italy)

M L FREZZOTTI1,2, G DE ASTIS3, L DALLAI4 and C GHEZZO1

1 Dipartimento di Scienze Della Terra, Università di Siena, Via Laterina 8, 53100 Siena, Italy*Corresponding author, e-mail: [email protected] IGAG – C.N.R., P.le A. Moro 5, 00185 Roma, Italy

3 Istituto Nazionale di Geofisica e Vulcanologia, Osservatorio Vesuviano, Via Diocleziano, 328 - 80124 Napoli, Italy4 Istituto di Geoscienze e Georisorse – C.N.R, Via G. Moruzzi, 1 - 56124 Pisa, Italy

Abstract: New major and trace element data, and Sr–Nd–Pb-O isotopic ratios for volcanic mafic rocks outcropping at MontiErnici in the Mid Latina Valley (southern Latium) are reported, with the aim of investigating the nature and evolution of Plio-Quaternary K-rich volcanism in Central Italy. Petrographical and geochemical studies allow us to identify mafic rocks rangingfrom ultrapotassic (HKS) to shoshonitic (SHO), and calc-alkaline (CA), these last ones being identified for the first time. The CArocks exhibit the most primitive signatures for Sr, Nd, and Pb isotopes (87Sr/86Sr = 0.706326-0.706654; 143Nd/144Nd = 0.512388–0.512361; 206Pb/204Pb = 18.944-18.940). The δ18O values are variable (δ18Ocpx from +5.75 to +7.08 ‰; and δ18Ool from +5.50 to+6.23 ‰), suggesting interaction with carbonate wall rocks. Radiogenic isotope ratios and incompatible elements distribution haveseveral characteristic in common with equivalent rocks from Pontine Islands (Ventotene), Campania and Aeolian arc volcanoes.Conversely, the HKS rocks closely resemble the ultrapotassic rocks from the Roman Province (87Sr/86Sr = 0.709679–0.711102;δ18Ocpx from +6.27 to +7.08 ‰). The high ratios of LILE (Large Ion Lithophile Elements: Rb, Cs, Th, U, K, LREE) and HFSE(High Field Strength Elements: Ta, Nb, Zr, Hf, Ti), and radiogenic isotope compositions of CA to HKS rocks indicate that allsuites contain subduction-related components, and suggest a N-MORB-type mantle source variably contaminated by hydrous fluidsand/or melts released by undergoing slabs, possibly during two distinct stages of metasomatism. The coexistence of ultra-alkalineand sub-alkaline orogenic magmatism, combined with tectonic, geophysical and geological evidence, support the possibility thatthe mantle beneath central-southern Italy (Ernici-Roccamonfina Province) was vertically zoned and produced different magmasuites during time.

Key-words: Monti Ernici, Latium magmatic Province, calcalkaline rocks, Ultrapotassic Rocks, geochemistry, igneous petrology.

1. Introduction1

The origin and geodynamic significance of Plio-2

Quaternary potassic alkaline volcanism in central-southern3

Italy has been extensively debated (see Peccerillo 2005,4

for a review). Most authors agree that K-rich mafic5

magmas originate in an upper mantle previously modified6

for both incompatible trace elements and radiogenic7

isotopes by metasomatic processes (e.g., Cox et al.,8

1976; Hawkesworth & Vollmer, 1979; Peccerillo, 1985;9

Conticelli & Peccerillo, 1992, Conticelli et al., 2002;10

Peccerillo 2005). There is, however, debate on the nature11

and timing of the processes that generated these mantle12

anomalies, as well as on the geodynamic setting: some13

authors suggest an origin by intracontinental rift environ-14

ment, whereas others propose a subduction-related setting15

(e.g., Vollmer, 1989; Ayuso et al., 1998; Peccerillo, 1985,16

1999).17

One of the most remarkable characteristics of K-rich 18

rocks in central-southern Italy – ranging from high-K calc- 19

alkaline (HKCA) and shoshonitic (SHO), to potassic (KS) 20

and ultrapotassic (HKS) – is their typical arc-type trace el- 21

ement signature (i.e., high LILE/HFSE ratios, low TiO2), 22

which has led to the suggestion of subduction-related man- 23

tle metasomatism and melting (e.g., Di Girolamo, 1978; 24

Peccerillo, 1985; 2003; Rogers et al., 1985). On the other 25

hand, these rocks have radiogenic isotope ratios and, in 26

most cases, a degree of enrichment in LILE, that are dis- 27

tinct from those of typical arc rocks. Moreover, although 28

magmatism in central Italy show variable potassium con- 29

tents, calc-alkaline rocks, which are typical of subduction 30

environments (e.g., Gill, 1981), are extraordinarily rare. 31

The Monti Ernici volcanoes (hereafter Ernici) in the Mid 32

Latina Valley (Central Italy), represent a key locality to 33

study these issues, since the erupted mafic magmas con- 34

tain variable potassium and incompatible element contents, 35

bridging the gap between typical arc-related volcanics 36

0935-1221/07/0019-1754 $ 7.20DOI: 10.1127/0935-1221/2007/0019-1754 © 2007 E. Schweizerbart’sche Verlagsbuchhandlung, D-70176 Stuttgart

Page 2

2 M.L. Frezzotti, G. De Astis, L. Dallai, C. Ghezzo Fast Track Article

and ultrapotassic rocks. Radiogenic isotope compositions1

previously measured in the different rocks are also vari-2

able, with 87Sr/86Sr ∼ 0.7062–0.7112, and 143Nd/144Nd ∼3

0.51237–0.51173 (Civetta et al., 1981; D’Antonio et al.,4

1996; Conticelli et al., 2002; Gasperini et al., 2002). Fur-5

ther, the Ernici volcanoes consists of several small mono-6

genetic centres that erupted basaltic products, suggest-7

ing that evolutionary processes in shallow-level magma8

chambers might have been moderate, or negligible. There-9

fore, Ernici volcanic rocks may represent rather primitive10

mantle-derived melts, that have the potential to provide11

maximum information on mantle processes and geological12

setting of the volcanism.13

In this paper, we report major, trace element, and Sr-Nd-14

Pb-O isotopic data on selected representative rock-samples15

from Ernici to discuss their petrogenesis within the geo-16

dynamic setting responsible for the potassic magmatism17

in central-southern Italy. The investigated samples include18

ultrapotassic, potassic and low-potassium rocks, these last19

ones showing major element chemistry falling in the field20

of calc-alkaline (CA) suites. Present study allows us to rec-21

ognize for the first time a calc-alkaline series in spatial and22

temporal association with the HKS rocks, making the Er-23

nici volcanoes the most compositionally variable district in24

central Italy.25

2. Geological setting and sampling26

Both the geological and geophysical evidence, together27

with the compositional features of the magmas erupted28

in central-southern Italy (e.g., Selvaggi & Amato, 1992;29

Lucente et al., 1999; Savelli, 1988; Peccerillo, 2005, and30

references therein) suggest that Mio-Pliocene subduction31

processes affected the Italian peninsula (e.g., Doglioni32

et al., 1999, and references therein). The convergence33

of Africa with the European plate generated a west-34

ward subducting slab, probably made of both oceanic or35

thinned continental lithosphere (Ionian Sea), and continen-36

tal lithosphere (Adriatic plate). A sub-vertical body iden-37

tified through seismic studies and located beneath central-38

southern Italy (Panza, 1984; Wortel & Spakman, 2000), has39

been interpreted as the relict detached slab, whereas un-40

der the Calabria tear migration and Ionian slab roll-back41

continue (e.g., Gvirtzman & Nur, 2001). From about 7 Ma,42

persistent magmatism developed along the Tyrrhenian mar-43

gin of the Italian Peninsula, through a progressive migra-44

tion from Tuscany and Latium to Campania, which gave45

rise to the volcanoes of the Tuscan, Roman and Campanian46

Magmatic Provinces (Fig. 1a). Starting from the Pliocene,47

the compressional front responsible for the formation of48

the Apenninic chain migrated eastward and co-existed with49

extensional tectonic and rifting processes within the in-50

active thrust belt (i.e., Meletti et al., 2000). The Marsili51

basin opening (from ∼ 1.8 Ma) and the anticlockwise ro-52

tation of the Apennines, together with the SE migration of53

the Calabrian arc, addressed and supported this change to-54

ward the present extensional regime in southern Italy (from55

∼ 0.7 Ma; De Astis et al., 2006, and references therein).56

ADRIAPLATE MARGIN

AEOLIAN Arc

0 200 400 Km

N

EW

S

RomanProvince

Mt.Vulture

Vesuvius

Mt. Etna

Roman Province

Phleg. Fields - Vesuvius

ODP (Site Numbers)

Ernici - Roccamonfina

Ionian Plate subduction

Adria Plate margin

RMERN

ODP655 ODP

651

TYRRHENIAN SEA

SICILY

MARSILIStromboli

CALABRIA

IONIANSEA

LEGEND

CampanianProvince

HybleanPlateau

ODP 650

Sabatini

Albani

Vulsini

Adriatic Sea

& Seamounts

a

b

POFI

Villa Santo Stefano

Tecchiena

Ceccano

Giulianodi Roma

Patrica

Volcanic Centres

HKS Rocks

SHO-KS rocks

Frosinone

Tuscany

Mt.ERNICI

O-RLine

Main Faults

Colle Castellone

Is.-Pro.APULIA

A-ALine

Fig. 1. a) Map of the central-southern Italy and Tyrrhenian Sea,showing the main magmatic provinces, and the investigated area.The ODP sites are indicated by numbers. b) geological sketch mapof the Monti Ernici volcanic area (modified from Civetta et al.,1981), showing the main eruptive centres and the most importanttectonic lines. ERN =Monti Ernici volcanoes; RM = Roccamonfinavolcano; O-R = Ortona – Roccamonfina lithospheric discontinuity;A-A = Ancona-Anzio lithospheric discontinuity; Phleg. Fields =Phlegraean Fields; Is. – Pro. = Ischia and Procida Islands.

The volcanoes of Monti Ernici occur close to the Tyrrhe- 57

nian margin in an area affected by Lower Pliocene NW- 58

SE faulting, which generated graben-horst structures paral- 59

lel to the Apennine chain (Fig. 1b). As observed for the 60

surrounding regions of Campania and northern Latium, 61

NE-SW transverse faults – having a normal to strike-slip 62

component of motion – were associated with the main 63

“Apenninic” fault system. In particular, the Ernici vol- 64

canoes formed within the so-called Ancona-Anzio and 65

Ortona-Roccamonfina lines, that represent two important 66

NE-SW trending tectonic lineaments. The former cuts the 67

Apennine chain and divides the northern Apennines from 68

Page 3

Fast Track Article Calc-alkaline and ultrapotassic magmatism at monti ernici 3

the Lazio and Abruzzi geological domains (Castellarin1

et al., 1982), whereas the latter separates the southern2

Apennines block from the central-northern sectors. Ac-3

cording to Locardi (1988) these lineaments were formed4

during the anticlockwise rotation of the Apennine chain.5

Paleomagnetic data support this hypothesis, and highlight6

different degrees of block rotation for the various Apennine7

sectors (Meloni et al., 1997),8

Most of the Ernici volcanoes consist of small mono-9

genic cones made of pyroclastic deposits, or, subordinately,10

lava flows (e.g., Colle Castellone, Giuliano di Roma, Colle11

Spinazzeta, Villa S. Stefano, Fig. 1b). Some cones (e.g.,12

Pofi volcano, and Patrica, Fig. 1b) have larger dimensions,13

resulting from a more intensive and alternating effusive and14

explosive activity. The bedrocks of the volcanic sequences15

consist of a wide variety of early Mesozoic to Quaternary16

rocks (e.g., Accordi et al., 1986), including Upper Triassic17

to Miocene neritic and pelagic carbonates, Upper-Middle18

Miocene calcareous and arenaceous sediments, and thin19

deposits of Pliocene to Lower Pleistocene terrigenous sed-20

iments. The crustal thickness in the Ernici area is estimated21

about 25–30 km (Piromallo & Morelli, 2003). The pyro-22

clastic deposits often contain sedimentary xenoliths, espe-23

cially carbonate rocks.24

Previous work on Ernici volcanics by Civetta et al. (1981)25

indicated the occurrence of two distinct series of alka-26

line rocks, characterised by different enrichments in potas-27

sium, incompatible elements, and radiogenic Sr: i) potas-28

sic (KS) rocks characterised by moderately potassic basalts29

and trachybasalts with shoshonitic affinity, and ii) ultra-30

potassic (HKS) rocks consisting mostly of leucite phono-31

litic tephrites. A different timing in the eruption of the32

two series was reported, based on K/Ar dating: HKS rocks33

range in age from 0.7 to 0.2 Ma, whereas the KS rocks34

are younger (0.2 and 0.1 Ma) (Basilone & Civetta, 1975;35

Civetta et al., 1981).36

The rock sampling was carried out in an extensively ur-37

banised and poorly exposed area, characterised by a limited38

number outcrops of modest area. However, every sector of39

the volcanic region was sampled, resulting in a collection40

of about 50 representative rock-samples. Only at Pofi vol-41

cano (Fig. 1b), it was possible to collect a continuous se-42

quences of volcanic products. Geochemical data coming43

from Pofi rocks (crossed symbols in some figures) indicate44

that low-potassium rocks (i.e., CA-SHO series, see below)45

alternate with HKS only during the final stages of the vol-46

canic successions, in agreement with Civetta et al. (1981).47

3. Analytical techniques48

More than 40 selected samples were analysed for major49

and trace elements by X-ray fluorescence analysis on fused50

glass disks, with a Philips MagixPro at the Dipartimento di51

Scienze della Terra dell’Università di Siena. Samples were52

prepared by mixing 1 g of homogenised powder and 8 g of53

lithium tetraborate (Merck Spectromelt A 10, Li2B4O7), as54

flux material, and by melting into glass beads. The back-55

ground and mass absorption intensities were calculated56

against the calibrations constructed from 24 international 57

geological reference materials. Loss on ignition was deter- 58

mined by heating samples to 1050 ◦C for 2 h. A selection of 59

representative samples, was further analysed for REE, Ta, 60

Hf, Cs, Pb, Th, and U by ICP-MS, (Table 1), at the Cen- 61

tre Petrographiques et Geochimiques (Vandouvre, France). 62

Precision of trace element data is better than 10 % for all 63

trace elements. 64

WDS analyses of mineral phases were performed with a 65

Cameca SX 50 (IGAG-CNR, Roma), using 15 kV accel- 66

erating voltage, 15 nA beam current, and a beam diameter 67

of 5 µm. Natural and synthetic silicates were used as stan- 68

dards for mineral analyses. 69

Sr, Nd and Pb isotopic ratios were determined on selected 70

whole rocks with TIMS 262 Thermo-Finnigan mass spec- 71

trometer at the Instituut voor Aard- and Levenschappen of 72

the Vrije Universiteit in Amsterdam. Sr and Nd isotope ra- 73

tios were normalized to NBS 987: 87Sr/86Sr = 0.710238 ± 8 74

on > 100 samples, and La Jolla: 143Nd/144Nd = 0.511834 ± 75

9 on > 100 samples, respectively. Total blanks were less 76

than 1.1 ng, which represents < 0.05 % of the element mass 77

in the measured fraction. For Sr-Nd, precision was better 78

than one unit on the fifth decimal place. 79

Pb was separated using 0.15 ml quartz ion-exchange 80

columns filled with AG1X8 200–400 mesh. Pb isotopes 81

were measured on solutions of 100 ppb Pb (or less) in 82

1 % HNO3. The method of standard–sample bracketing 83

was chosen since Pb does not have a pair of invariant iso- 84

topes that can be used to correct for mass bias during analy- 85

ses (cf. Elburg et al., 2005). Machine blanks were analyzed 86

before each standard and sample, and the average of these 87

two measurements was subtracted from each cycle before 88

calculation of the Pb isotopic ratios. The blank levels varied 89

between sessions with intensities of 1.5–3.0 mV on 208Pb. 90

A 100 ppb solution of NIST SRM-981 was used as the stan- 91

dard, and normalization was performed using the values of 92

Baker et al. (2004). Precision of Pb isotope ratios is better 93

than one unit on third decimal place. 94

Sr isotope ratios for some samples (see Table 1) were 95

measured at the Southampton Oceanography Centre on 96

a seven-collector VG sector 54 mass spectrometer with 97

a separable-filament source. Rock powder for Sr analy- 98

sis were leached with 2 N HCl for 1h at 140 ◦C prior to 99

dissolution and isolation of Sr using Sr resin. Isotope ra- 100

tios were normalized to 87Sr/86Sr = 0.710252 ± 15 (2s.d., 101

n = 169). Total blanks were less than 1.1 ng, which repre- 102

sents < 0.05 % of the element mass in the measured frac- 103

tion. 104

Oxygen isotope compositions of clinopyroxene and 105

olivine mineral separates were performed at the CNR- Is- 106

tituto di Geoscienze and Georisorse in Pisa by conven- 107

tional laser fluorination (Sharp, 1990), reacting the sam- 108

ples under an F2 gas atmosphere. Purified oxygen gas was 109

directly transferred into a Finnigan Delta XP Mass Spec- 110

trometer via a 13A zeolite molecular sieve for isotopic ra- 111

tio determinations (Sharp, 1995). All the data are given 112

following the standard δS MOW notation. During the course 113

of analysis, an in-house laboratory QMS quartz standard 114

was used (δ18OSMOW = ±14.05 ‰ yielding an average 115

(1σ) δ18O = +14.08 ‰, σ = 0.12 ‰. NBS28 standard 116

Page 4

4 M.L. Frezzotti, G. De Astis, L. Dallai, C. Ghezzo Fast Track ArticleTa

ble

1.S

elec

ted

maj

or,t

race

elem

ent,

and

isot

opic

com

posi

tion

sof

Ern

iciv

olca

nic

rock

s.

Hig

h-K

rock

sSh

osho

niti

cro

cks

Cal

c-al

kalin

ero

cks

Sam

ple

ER

0304

ER

0307

bE

R03

09E

R03

12-1

bE

R03

12-2

ER

0313

ER

0420

-1SC

3E

R03

14Sa

mpl

eE

R03

03E

R03

05E

R03

06SC

1E

R03

08E

R04

17SC

4E

R04

19R

ock

type

lava

lava

lava

lava

lava

lava

scor

iasc

oria

scor

iaR

ock

type

lava

lava

bom

bsc

oria

lava

lava

scor

iala

vaL

ocal

ity

Giu

lian

odi

R.

Pofi

Pofi

Cas

tell

one

Cas

tell

one

C.V

esco

voPo

fiPo

fiC

.Ven

toL

ocal

ity

Cec

cano

Pofi

Pofi

Cec

cano

Spin

azze

taL

aB

adia

Spin

azze

taC

.Fon

tana

SiO

2w

t%48

.72

49.7

450

.92

51.4

251

.05

50.9

649

.68

51.2

754

.37

SiO

2w

t%49

.71

49.9

850

.38

48.9

750

.50

49.4

249

.68

48.0

6T

iO2

0.82

0.81

0.82

1.09

1.06

0.85

0.83

0.88

0.82

TiO

20.

880.

850.

780.

790.

810.

850.

790.

90A

l2O

315

.72

16.4

116

.03

15.8

715

.58

16.6

316

.13

14.3

915

.59

Al2

O3

14.8

415

.32

14.1

712

.98

15.2

614

.42

13.3

616

.89

Fe2O

37.

697.

747.

877.

796.

897.

917.

788.

998.

37Fe

2O3

9.30

8.69

8.17

8.48

8.52

9.11

8.22

9.45

MnO

0.14

0.14

0.14

0.15

0.15

0.15

0.14

0.15

0.16

MnO

0.16

0.16

0.15

0.14

0.15

0.16

0.15

0.17

MgO

5.82

5.45

6.44

4.61

4.55

3.62

5.50

7.48

5.67

MgO

7.67

7.01

8.30

9.74

8.63

7.08

9.25

7.26

CaO

11.6

510

.19

9.52

7.83

8.77

8.53

10.3

412

.84

9.67

CaO

13.7

813

.33

13.8

514

.27

13.4

413

.63

13.7

812

.93

Na2

O1.

601.

641.

810.

761.

441.

563.

781.

521.

41N

a2O

1.98

2.61

2.30

2.60

1.90

3.24

2.64

1.44

K2O

6.84

6.90

6.09

7.00

8.55

7.86

3.30

2.27

2.17

K2O

0.47

0.76

0.51

0.51

0.56

0.49

1.19

1.00

P2O

50.

360.

400.

360.

870.

850.

530.

360.

200.

19P2

O5

0.20

0.24

0.23

0.23

0.24

0.20

0.20

0.19

LO

I0.

640.

890.

802.

731.

471.

482.

160.

031.

89L

OI

1.02

1.04

1.17

1.58

0.14

1.06

0.72

1.51

Vpp

m21

625

021

523

226

223

423

627

323

3V

ppm

264

213

240

243

248

256

243

273

Cr

132

161

182

110

.62

9425

318

2C

r29

719

857

967

358

722

144

326

7C

o19

30.2

3232

33.3

2925

42.3

39C

o40

.821

4040

.638

.933

4339

Ni

6462

.172

3631

.629

5862

65N

i65

6399

109

100

6298

70R

b31

337

631

742

352

945

033

311

312

7R

b15

331

348

552

898

188

106

36Sr

1459

1631

1255

956

1370

1578

1342

863

644

Sr98

793

296

488

710

2984

683

083

3Y

3334

3340

4535

3422

25Y

23.9

2221

.120

.923

.224

2429

Zr

233

275

236

452

588

285

241

102

142

Zr

111

114

111

106

115

115

126

147

Nb

1212

.211

3137

.116

126.

159

Nb

6.9

87.

36.

87.

37

88

Cs

-26

.6-

-63

.14

-5.

69-

Cs

11.5

-17

22.2

25.4

--

-B

a11

7380

978

847

7243

1815

0860

544

952

6B

a50

364

553

953

451

954

547

379

3L

a64

9065

181

264

7565

28.6

30L

a35

.736

32.9

31.5

36.2

2924

37C

e10

218

711

727

645

616

013

660

.554

Ce

69.3

8766

.763

.470

.965

6285

Nd

-84

--

182

--

30.2

-N

d34

.2-

32.3

30.7

34.7

--

-Sm

-15

.3-

-31

.1-

-6.

27-

Sm6.

8-

6.4

6.2

6.9

--

-E

u-

3.19

--

6.2

--

1.63

-E

u1.

73-

1.64

1.59

1.75

--

-T

b-

1.43

--

2.42

--

0.76

-T

b0.

81-

0.74

0.73

0.81

--

-Y

b-

2.48

--

2.73

--

1.89

-Y

b2

-1.

781.

741.

93-

--

Lu

-0.

37-

-0.

38-

-0.

29-

Lu

0.31

-0.

270.

270.

29-

--

Hf

-6.

5-

-12

.9-

-2.

68-

Hf

2.83

-2.

782.

712.

98-

--

Ta-

1.43

--

2.22

--

0.42

-Ta

0.49

-0.

490.

460.

53-

--

Pb35

26.9

3160

126

6035

14.3

20Pb

17.8

1716

.615

16.9

1616

17T

h30

31.2

2688

104

4530

7.9

13T

h10

.111

9.4

8.9

1110

1113

U4

5.5

46

5.4

75

2.28

1U

2.67

62.

642.

332.

443

00

87Sr/8

6Sr

0.70

9679

--

0.71

1102

--

0.70

6446

0.70

6851

87Sr/8

6Sr

--

0.70

6599

-0.

7066

540.

7065

76*

0.70

6326

*-

143N

d/14

4Nd

0.51

2136

--

0.51

2086

--

0.51

2370

-14

3N

d/14

4Nd

--

0.51

2388

-0.

5123

61-

--

206P

b/20

4Pb

18.8

245

--

18.7

349

--

18.9

063

-20

6Pb/2

04Pb

--

18.9

444

-18

.939

9-

--

207P

b/20

4Pb

15.6

876

--

15.6

798

--

15.6

861

-20

7Pb/2

04Pb

--

15.6

860

-15

.684

9-

--

208P

b/20

4Pb

39.0

507

--

39.0

087

--

39.0

667

-20

8Pb/2

04Pb

--

39.0

710

-39

.062

7-

--

d18O

cpx

6.27

--

--

-6.

276.

09d1

8Ocp

x-

-6.

02-

6.40

5.75

6.07

6.50

d18O

ol-

--

--

-5.

84-

d18O

ol-

-5.

50-

-5.

505.

72-

Ast

eris

ksin

dica

teda

taob

tain

edat

Scho

olof

Oce

anan

dE

arth

Scie

nce,

Nat

iona

lOce

anog

raph

yC

entr

e,So

utha

mpt

on,U

K.

Page 5

Fast Track Article Calc-alkaline and ultrapotassic magmatism at monti ernici 5

(δ18O = +9.60 ‰ gave an average values of δ18O =1

9.52 ‰ (σ = 0.14 ‰).2

4. Classification and petrography3

Figures 2a and b show alkalies and K2O vs. SiO2 plots for4

the Ernici rocks: our rock-samples straddle the boundary5

between sub-alkaline and alkaline fields, and range from6

CA up to HKS compositions. A variable degree of sil-7

ica saturation is also observed in the diagram K2O/Na2O8

vs. degrees of silica saturation indicated by ∆Q notation9

(Fig. 2c; Peccerillo, 2003). Compared to previous studies10

(Civetta et al., 1979; 1981), our samples show a consider-11

ably wider range of potassium content, with the occurrence12

of rocks that fall in the CA field; besides, the number of13

samples falling in the SHO field (the KS of Civetta et al.,14

1981) is much lower. All the samples containing low K2O15

abundances are unaffected by secondary alteration, as in-16

dicated by microscopic observations, and low LOI values17

(Table 1): their depletion in potassium likely reflects pris-18

tine magmatic compositions. We conclude that these mag-19

mas can be considered as belonging to the calcalkaline se-20

ries, on the basis of major element chemistry.21

The HKS rocks plot in the alkaline field of Irvine &22

Baragar (1971), and are variably undersaturated in silica,23

whereas the CA-SHO rocks are subalkaline, from mod-24

erately undersaturated to slightly oversaturated in silica25

(Fig. 2a, b, c). Overall, the HKS rocks fall in the field26

of the Roman magmatic Province, whereas SHO and CA27

rocks plot in the same field of Procida, and Ventotene vol-28

canoes, belonging to the Campanian Province (Peccerillo,29

2005). Ernici CA rocks further show similar K2O contents,30

and K2O/Na2O ratios, but slightly higher degrees of sil-31

ica undersaturation (i.e., slightly lower ∆Q) than CA rocks32

from the Aeolian arc with similar SiO2 and MgO contents33

(Fig. 2c).34

Based on the TAS diagram (not shown), the composition35

of the analysed rocks ranges from basalt to K-trachybasalt36

and leucite tephrite, to basaltic andesite; only one sample37

has andesitic composition. Notably, the rocks from Pofi38

volcano (crossed symbols in Fig. 2a, b), replicate, at a39

smaller scale, the same compositional trend outlined by the40

whole Ernici products.41

CA basalts are hypocrystalline and moderately por-42

phyritic rocks, with phenocryst contents varying between43

20 and 30 % in volume. Clinopyroxene is the dominant44

phenocryst phase together with minor olivine, and has a45

diopsidic-salitic composition (i.e., zoned; mg# 0.92–78;46

mg# =Mg/(Mg + Fe); Table 2). Olivine microphenocrysts47

have high MgO contents (mg# = 0.92–0.84), and are of-48

ten partially transformed to iddingsite. The groundmass49

consists of salitic clinopyroxene (mg# = 0.81–0.78), by-50

townitic plagioclase (An = 87–79), and rare olivine (mg#51

= 0.83–0.69), with minor glass, Fe–Ti oxides, and Fe-52

sulphides (Table 2). Orthopyroxene and K-bearing phases53

were not observed.54

SHO basalts are texturally similar to CA basalts, but by-55

townitic plagioclase may occur as a phenocryst phase (Ta-56

ble 2), whereas olivine phenocrysts are hardly ever ob-57

SubalkalineAlkalies

SiO2

a

48 52 56 60 64 68 72 760

2

4

6

8

10

Legend Low/Medium-K rocks (CA)

Medium-K rocks (HKCA-SHO) High-K rocks (HKS)HKS from C.Castellone

Pofi rocks with overprint

b

Arc Tholeiitic

Calc-alkaline

High-K calcalkaline

Shoshonitic

KS rocks(Literature)

HKS rocks(Literature) Ultra-Potassic

SiO2

K2O

Alkaline

40 50 60 700

5

10HKS rocks(Literature)

KS rocks(Literature)

-60 -40 -20 0 200.01

0.1

1

10

K2O/Na2O

∆Q = Q- (lc+ne+kal+ol)

HKS

SHO-KS

CATH

SilicaUndersaturated

SilicaOversaturated

Tuscany & RomanProvinces

AeolianAeolianCA & HKCACA & HKCAbasaltsbasalts

Procida &VentoteneIslands

c

HKS rocks(Literature)

ErniciKS rocks

(Literature)

Phleg.FieldsVesuvius

Fig. 2. Classification diagrams for Ernici volcanic rocks: a) SiO2vs.Alkalies (K2O + Na2O) diagram (Irvine & Baragar, 1971); b) SiO2

vs. K2O diagram (Peccerillo & Taylor, 1976); c) ∆Q vs. K2O/Na2Odiagram; ∆Q is the algebraic sum of normative quartz minus nor-mative undersaturated minerals (lc+ne+kal+ol) (Peccerillo, 2003).Rocks from Pofi volcano have been highlighted with an overprintedblack cross in Fig. 2a and 2b. In Fig. 2c the Colle Castellone HKSrocks are not distinguished. Oxides are normalised on a water-freebasis.

served. Clinopyroxene is zoned and has a diopside-salite 58

composition (mg# = 0.91–0.82; Table 2). The ground- 59

mass consists of salitic clinopyroxene (mg# 0.77–0.74; Ta- 60

ble 2), labradoritic plagioclase, and minor olivine (mg# 61

0.69–0.50) and K-feldspar (with rare glass, Fe–Ti oxides 62

and Fe-sulphides). 63

Page 6

6 M.L. Frezzotti, G. De Astis, L. Dallai, C. Ghezzo Fast Track ArticleTa

ble

2.se

lect

edan

alys

esof

min

eral

phas

es.

Seri

eC

AC

AC

AC

AC

AC

AC

AC

AC

AC

AC

ASH

OSH

OSH

OSH

OSH

OSH

OL

ocal

ity

C.F

onta

naPo

fiPo

fiC

.Fon

tana

Pofi

C.F

onta

naPo

fiPo

fiPo

fiPo

fiC

.Fon

tana

Pofi

Pofi

Pofi

Pofi

Pofi

Pofi

Sam

ple

ER

0419

ER

0306

ER

0306

ER

0419

ER

0306

ER

0419

ER

0306

ER

0306

ER

0306

ER

0306

ER

0419

ER

0416

-cE

R04

16-c

ER

0416

-cE

R04

16-c

ER

0416

-cE

R04

16-c

Phas

eO

lO

lO

lO

lO

lC

pxC

pxC

pxri

mC

pxC

pxC

pxO

lO

lC

pxC

pxco

reC

pxri

mC

pxph

enoc

r.ph

enoc

r.ph

enoc

r.g.

mas

sg.

mas

sph

enoc

r.ph

enoc

r.ph

enoc

r.ph

enoc

r.ph

enoc

r.g.

mas

sg.

mas

sg.

mas

sph

enoc

r.ph

enoc

r.ph

enoc

r.g.

mas

s

SiO

239

.53

41.3

240

.65

39.6

937

.75

51.8

753

.48

52.1

852

.92

53.3

150

.50

37.3

235

.39

52.8

851

.16

50.5

745

.56

TiO

2n.

d.n.

d.n.

d.n.

d.n.

d.0.

400.

220.

290.

360.

200.

45n.

d.n.

d.0.

180.

350.

521.

41A

l2O

30.

060.

040.

06n.

d.0.

252.

381.

802.

082.

091.

634.

460.

032.

051.

893.

354.

058.

23C

r2O

3n.

d.0.

050.

05n.

d.n.

d.0.

180.

710.

250.

230.

63n.

d.n.

d.n.

d.0.

590.

190.

09n.

d.M

gO44

.53

51.0

151

.07

43.1

833

.70

16.3

117

.31

16.9

517

.13

17.5

014

.92

34.5

821

.64

17.7

016

.16

15.4

612

.77

CaO

0.23

0.33

0.44

0.34

1.12

22.8

623

.70

23.8

823

.15

23.5

522

.72

0.35

1.19

23.6

723

.55

23.3

123

.19

MnO

0.27

0.17

0.20

0.36

0.72

0.15

n.d.

0.06

0.13

0.06

0.14

0.75

1.33

0.06

0.05

0.20

0.14

FeO

14.8

18.

298.

5216

.12

27.0

34.

712.

823.

584.

202.

636.

1227

.82

38.4

62.

964.

455.

717.

72N

iO0.

060.

120.

200.

060.

08n.

d.0.

10n.

d.n.

d.n.

d.n.

d.n.

d.0.

050.

06n.

d.n.

d.0.

08N

a2O

n.d.

n.d.

n.d.

n.d.

n.d.

0.17

0.17

0.19

0.14

0.16

0.28

n.d.

n.d.

0.17

0.18

0.25

0.20

Tota

l:99

.49

101.

3910

1.28

99.7

810

0.77

99.0

510

0.32

99.4

710

0.40

99.6

799

.65

100.

9010

0.27

100.

1499

.44

100.

1699

.32

Wo%

46.3

247

.41

47.4

645

.96

47.1

146

.97

46.7

147

.55

47.1

649

.24

En%

45.9

848

.18

46.8

747

.33

48.7

042

.92

48.6

045

.37

43.5

237

.72

Fs%

7.70

4.41

5.67

6.72

4.19

10.1

14.

707.

089.

3313

.04

mg#

0.84

0.92

0.91

0.83

0.69

0.86

0.92

0.89

0.88

0.92

0.81

0.69

0.50

0.91

0.87

0.83

0.75

Seri

eH

KS

HK

SH

KS

HK

SH

KS

HK

SH

KS

HK

SH

KS

CA

SHO

SHO

HK

SH

KS

HK

SH

KS

Loc

alit

yPo

fiPo

fiPo

fiPo

fiPo

fiC

aste

llon

eC

aste

llon

eC

aste

llon

eC

aste

llon

ePo

fiPo

fiPo

fiPo

fiC

aste

llon

eC

aste

llon

ePo

fiSa

mpl

eE

R03

07E

R03

07E

R03

07E

R03

07E

R03

07E

R03

12-2

ER

0312

-2E

R03

12-2

ER

0307

ER

0306

ER

0416

-cE

R04

16-c

ER

0307

ER

0312

-2E

R03

12-2

ER

0307

Phas

eO

lO

lO

lC

pxC

pxC

pxC

pxC

pxC

pxPl

PlPl

PlL

euL

euL

euph

enoc

r.ph

enoc

r.g.

mas

sph

enoc

r.ph

enoc

r.ph

enoc

r.ph

enoc

r.ph

enoc

r.g.

mas

sg.

mas

sph

enoc

r.g.

mas

sg.

mas

sph

enoc

r.ph

enoc

r.g.

mas

s

SiO

240

.03

39.6

834

.91

50.7

450

.74

51.8

551

.15

51.8

849

.15

47.7

646

.59

54.2

649

.09

54.2

954

.10

54.6

0T

iO2

n.d.

n.d.

n.d.

0.58

0.58

0.56

0.64

0.50

0.84

n.d.

n.d.

n.d.

n.d.

n.d.

n.d.

n.d.

Al2

O3

n.d.

n.d.

n.d.

3.58

3.58

2.89

2.46

2.25

4.40

31.3

932

.03

27.5

231

.27

22.3

622

.44

22.0

2C

r2O

3n.

d.n.

d.n.

d.0.

100.

10n.

d.n.

d.n.

d.n.

d.n.

d.n.

d.n.

d.n.

d.n.

d.n.

d.n.

d.M

gO47

.04

47.1

921

.17

15.5

515

.55

15.8

015

.51

16.2

315

.64

n.d.

n.d.

n.d.

n.d.

n.d.

n.d.

n.d.

CaO

0.50

0.46

0.88

23.9

423

.94

24.7

824

.47

24.7

223

.09

15.7

816

.86

10.9

415

.32

n.d.

n.d.

0.05

MnO

0.32

0.32

1.37

0.11

0.11

0.07

0.16

0.11

0.19

n.d.

n.d.

n.d.

n.d.

n.d.

n.d.

n.d.

FeO

11.5

511

.80

41.3

14.

494.

494.

014.

644.

465.

701.

010.

880.

600.

740.

530.

480.

47N

iO0.

100.

16n.

d.0.

030.

030.

04n.

d.n.

d.0.

06n.

d.n.

d.n.

d.n.

d.n.

d.n.

d.n.

d.N

a2O

n.d.

n.d.

n.d.

0.12

0.12

0.14

0.12

0.13

0.19

2.21

1.91

4.88

2.21

n.d.

n.d.

0.33

K2O

n.d.

n.d.

n.d.

n.d.

n.d.

n.d.

n.d.

n.d.

n.d.

0.16

0.13

0.31

0.31

21.0

721

.06

20.4

5To

tal:

99.5

399

.60

99.6

399

.23

99.2

310

0.12

99.1

410

0.27

99.2

598

.41

98.4

898

.67

99.0

998

.33

98.1

598

.09

Wo%

48.6

948

.69

49.6

249

.14

48.6

047

.50

An%

78.9

782

.33

54.3

077

.84

En%

44.0

044

.00

44.0

243

.34

44.3

843

.05

Ab%

20.0

616

.89

43.8

820

.27

Fs%

7.31

7.31

6.36

7.52

7.02

9.46

Or%

0.97

0.78

1.82

1.89

mg#

0.88

0.88

0.48

0.86

0.86

0.88

0.86

0.87

0.82

mg#=

Mg/

(Mg+

Fe);

n.d.=

notd

etec

ted;

phen

ocr.=

phen

ocry

st;g

.mas

s=

grou

ndm

ass.

Page 7

Fast Track Article Calc-alkaline and ultrapotassic magmatism at monti ernici 7

0

2

4

6

8

K2O

3 5 7 9 110

100

200

300

400

500

600

MgO

0

100

200

300

400

500

600

700

Cr

0

100

200

300

400

500

Ce

3 5 7 9 110

100

200

300

400

500

600

Zr

MgO

Rb

107

9

11

13

15

CaOKS rocks(Literature)

HKS rocks(Literature)

KS rocks(Literature)

HKS rocks(Literature)

HKS rocks(Literature)

KS rocks(Literature)

KS rocks(Literature)

HKS rocks(Literature)

HKS rocks(Literature)

KS rocks(Literature)

HKS rocks(Literature)

KS rocks(Literature)

a

c

e

b

d

f

Fig. 3. Harker variation diagrams for selected major (wt.%) andtrace (ppm) elements vs. MgO of Ernici rocks. Symbols as inFig. 2c. Fields of data from literature (Civetta et al., 1981) are alsoreported.

HKS leucite-tephrites are holocrystalline variably por-1

phyritic rocks. Phenocrysts (20–40 % in vol.) consist of2

clinopyroxene and leucite ± olivine, set in a groundmass3

of leucite clinopyroxene ± minor olivine, plagioclase, and4

ilmenite (Table 2). In a few samples, minor brown mica5

and K-feldspar are also present. Diopsidic-salitic clinopy-6

roxene (mg# 0.90–0.83) is the dominant phenocryst, with7

subordinate olivine (mg# 0.88–0.86) and leucite, set in a8

groundmass of salitic clinopyroxene (mg# 0.82–0.77), and9

leucite, with minor bytownitic-labradoritic plagioclase and10

Fe-rich olivine (mg# 0.52–0.48). At Colle Castellone, phe-11

nocrysts consist of leucite and salitic clinopyroxene (mg#12

0.88–0.86), while olivine is absent. The groundmass con-13

sists of salitic clinopyroxene (mg# 0.82–0.77), and leucite,14

with subordinate bytownitic-labradoritic plagioclase, and15

Fe-Ti oxides.16

5. Geochemistry17

5.1. Major and trace element data18

Variation diagrams of selected major and trace elements vs.19

MgO and K2O are reported in Fig. 3, and 4. The analysed20

samples show intermediate to high MgO contents, with21

HKS rocks displaying lower MgO, than CA and SHO sam-22

ples. Cr and Ni show a wide range of values, HKS rocks23

having lower contents, than SHO and CA rocks, in accor-24

dance with their lower MgO (Fig. 3).25

Large Ion Lithophile Elements (LILE: Rb, Ba, Th, U,26

Light Rare Earth Elements), and High Field Strength El-27

ements (HFSE: Zr, Hf, Nb, Ta, etc.) show an increase from28

calcalkaline to ultrapotassic rocks. A group of samples29

0

1000

2000

3000

4000

5000

Ba

0

100

200

300

La

0 2 4 6 8 100

20

40

60

80

100

Th

K2O

0

10

20

30

40

Nb

8

12

16

20

24

28

Zr/Nb

0 2 4 6 8 100,706

0,707

0,708

0,709

0,710

0,711

0,712

87Sr/86Sr

K2O

KS rocks(Literature)

HKS rocks(Literature)

HKS rocks(Literature)

HKS rocks(Literature)

KS rocks(Literature)

HKS rocks(Literature)

HKS rocks(Literature)KS rocks

(Literature)

KS rocks(Literature)

KS rocks(Literature)

HKS rocks (Literature)

KS rocks(Literature)

a

c

e

b

d

f

Fig. 4. Variation diagrams of selected Incompatible Trace Elements(ITE) contents (ppm), ITE ratios and 87Sr/86Sr vs. K2O for Ernicirocks. Symbols as in Fig. 2a.

from the HKS centre of Colle Castellone displays consid- 30

erably higher concentrations for all incompatible trace ele- 31

ments (ITE), with respect to other HKS rocks with similar 32

MgO contents (grey dots in Fig. 4). A wide range of Rb 33

concentration is observed in the poorly potassic CA rocks 34

(Fig. 3). Notably, this variation is not related to any other 35

geochemical parameters. 36

REE plots for representative samples (Fig. 5a) dis- 37

play variably fractionated patterns and degrees of LREE 38

enrichments for the different series, with a nega- 39

tive Eu anomaly (slight, but stronger for the HKS 40

rocks). Mantle-normalised incompatible element diagrams 41

(spider-diagrams; Fig. 5b, 6) show Ta-Nb troughs, positive 42

spikes for Rb, Cs, Th,U, LREE and Pb, and negative spikes 43

for Hf, Zr and Ba. All these characteristics are typical of 44

arc rocks, with exception of the negative Ba anomaly. The 45

CA rocks display a pattern very similar to the associated 46

shoshonites, but a much lower abundance in potassium, and 47

a relative enrichment in Cs, Th, U, La and Sr. A negative 48

Sr anomaly is present in Colle Castellone HKS rocks. 49

Figure 6 shows incompatible element patterns for rep- 50

resentative Ernici mafic rocks compared with analogous 51

volcanic rocks from other Italian magmatic provinces. 52

Compared with shoshonites from Procida and Ventotene 53

(Eastern Pontine islands, Campanian Province, Fig. 1a) 54

and with calc-alkaline basalt from Alicudi (Aeolian Arc, 55

Fig. 1a), the CA and SHO samples from Ernici have higher 56

enrichments in LILE (especially Cs, Rb, Ba, Th, U), and 57

stronger negative anomalies of HFSE (Fig. 6a). The HKS 58

magmas from Ernici (Fig. 6b) have incompatible element 59

abundances similar, or often higher (e.g., REE, Zr, Hf) 60

than the HKS rocks from Roccamonfina and Mt. Somma- 61

Vesuvius, and resemble leucite-tephrites from the Roman 62

Province (Peccerillo, 2005). The Colle Castellone sample 63

exhibits the highest enrichment in several elements (Cs, Ba, 64

Page 8

8 M.L. Frezzotti, G. De Astis, L. Dallai, C. Ghezzo Fast Track Article

1

10

100

1000

La Ce Pr Nd Pm Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

Roc

k/C

hond

rites

2

10

100

1000

Cs Rb Ba Th U K Ta Nb La Ce Sr Nd P Hf Zr Sm Ti Tb Y

Roc

k/P

rimor

dial

Man

tle

a

b

Fig. 5. REE and ITE patterns of Ernici mafic rocks (MgO > 4 wt.%)normalised to chondrites (Sun & McDonough, 1989) and to primor-dial mantle (Wood et al., 1979).

Th, Pb, LREE) with respect to any other ultrapotassic rocks1

in central-southern Italy (grey dots in Fig. 5a, and 6b). This2

suggests a bimodality among the Ernici HKS members, and3

that the Colle Castellone rocks represent an extremely en-4

riched HKS melt.5

5.2. Oxygen isotope composition6

Since clinopyroxene and olivine are the early crystalliz-7

ing phases in the CA and SHO basalts, their δ18O val-8

ues should represent the O-isotope composition of primi-9

tive magma, if the O-isotope equilibrium between miner-10

als and melt is maintained. The δ18Ocpx and δ18Ool data11

measured on Ernici clinopyroxenes and olivines are pre-12

sented in Fig. 7, and Table 1. They vary from +5.75 and13

+7.08 ‰, and from +5.50 and +6.23 ‰, respectively, and14

are positively correlated (Fig. 7c); they are significantly15

lower than δ18Owr values reported in previous studies (e.g.,16

Turi et al., 1991), but slightly higher than values reported17

for unaltered mantle rocks (e.g., MORB = δ18O ≈ +5.218

to +6.1 ‰). These results are also higher than ratios mea-19

sured in Alicudi CA basalts and andesites (δ18Ool = +5.1020

to +5.29 ‰; Peccerillo et al., 2004), the latter related to21

recent subduction.22

Roc

k/P

rimor

dial

Man

tle

10

100

1000

Cs Rb Ba Th U K Ta Nb La Ce Sr Nd P Hf Zr Sm Ti Tb Y2

HKS mafic rocks (MgO > 4 %)

Legend Ernici (C.C.) HKS

Ernici HKSRoccamonfina HKS

Som-Ves HKSRP HKS

CA - SHO mafic rocks (MgO > 6 %)

Legend Ernici CAErnici SHO

Procida SHOAlicudi CA

Roc

k/P

rimor

dial

Man

tle

5000

1

10

100

1000

Cs Rb Ba Th U K Ta Nb La Ce Sr Nd P Hf Zr Sm Ti Tb YPb

Pb

a

b

Fig. 6. Incompatible elements patterns of Ernici CA - SHO rocks(5a - MgO > 6 wt.%) and Ernici HKS rocks (5b, MgO > 4 wt.%)normalised to primordial mantle (Wood et al., 1979, except Pb fromSun & McDonough, 1989) and compared with rocks from sur-rounding regions and similar petro-chemical affinity. Legend: C.C. =Colle Castellone volcanic centre, Som-Ves = Mt.Somma.Vesuviusvolcano, RP = Roman Province.

Clinopyroxenes from three ultrapotassic samples were 23

also measured, and yielded δ18O values from +6.27 to 24

+6.45 ‰ (Fig. 7a, and d), in the range of typical Roman- 25

type magmas which are proposed to be affected by mod- 26

erate degrees of contamination by sedimentary carbonate 27

(e.g., Dallai et al., 2004). 28

The ∆18Ocpx−ol of CA rocks is close to the 0.4 ‰ expected 29

for oxygen isotope equilibrium at magmatic temperatures 30

(e.g., Mattey et al., 1994). This is different from what ob- 31

served for Roman-type volcanic rocks, where O-isotope 32

equilibrium between olivine and cpx is rarely achieved, 33

likely due to their crystallization conditions, occurring un- 34

der high Ca, CO2 and O2 activities (carbonate host-rock 35

contribution during pre-eruptive stage; Gaeta et al., 2006). 36

5.3. Sr-Nd-Pb isotopes 37

Radiogenic isotope compositions of the investigated sam- 38

ples are variable (Fig. 8, and Table 1). Measured Sr isotopic 39

Page 9

Fast Track Article Calc-alkaline and ultrapotassic magmatism at monti ernici 9∂

∂

∂

∂∂

Fig. 7. Variation diagram of δ18O of mineral phases of Ernici rocks.Symbols as in Fig. 2a.

ratios range from 0.70633 to 0.71110. There is an overall1

increase of 87Sr/86Sr ratios with potassium; however, CA-2

SHO rocks have similar 87Sr/86Sr ratios ranging between3

0.706326 and 0.706654 (Fig. 8a). HKS rocks have much4

more radiogenic compositions (87Sr/86Sr = 0.709679–5

0.709840) with a large gap between these rocks and CA-6

SHO volcanics; another major variation is observed be-7

tween the HKS and Colle Castellone rocks (87Sr/86Sr =8

0.711102; also characterised by the highest incompati-9

ble element abundances). Nd isotopic ratios show an in-10

verse trend, decreasing from CA-SHO to HKS volcanics11

(143Nd/144Nd = –0.512086–0.512388; Fig. 8a). Compared12

with previous studies, our data show similar ranges of Sr-13

Nd isotopic ratios, although our CA-SHO rock-samples14

show narrower 87Sr/86Sr range than those observed by15

Civetta et al. (0.70622–0.70697; 1981). They match those16

reported by D’Antonio et al. (1996), Conticelli et al.17

(2002), and Gasperini et al. (2002), except for a significant18

difference in the 143Nd/144Nd value for the Colle Castellone19

lava (143Nd/144Nd = 0.51173, in Conticelli et al., 2002).20

Note, however, that our value falls within the array defined21

by central Italy volcanic rocks.22

Considering both present and literature data, the Sr-Nd23

isotope compositions of Ernici rocks cover a very wide24

compositional range, similar to those of Roccamonfina,25

and overlap with the Roman and Campanian provinces26

(Fig. 8a). Notably, the Sr isotopic ratios of the Ernici CA-27

SHO rocks are close to those from Ventotene, and Cam-28

panian volcanoes, having similar K2O contents, whereas29143Nd/144Nd values are slightly lower than Campanian30

rocks (Fig. 8a). Sr- and Nd-isotopic compositions of the31

HKS series fall in the field of the Roman Province.32

Pb isotopic ratios measured on our samples display mod-33

erate variations (Fig. 8b and 9a, and Table 1), with the HKS34

rocks displaying less radiogenic compositions than CA and35

SHO rocks. 206Pb/204Pb ratios range from 18.735 to 18.944,36207Pb/204Pb from 15.680 to 15.686, and 208Pb/204Pb from37

39.009 to 39.071 (Fig. 9a). Such variations are in agree-38

CA - HKCA Aeolian Basalts(central-western AVD sectors and Stromboli)

0.703 0.706 0.709 0.712 0.715 0.7180.5116

0.5120

0.5124

0.5128

143Nd/144Nd

87Sr/ 86 Sr

Vesuvius &Phlegr. Fields

Tuscany &Roman Provinces

Roccamonfina &Ventotene Island

Ernici CA-HKCAErnici SHO-KS

Ernici HKSErnici HKS (C.C.)

Ernici KS (Ref.)Ernici HKS (Ref.)

18.4

18.8

19.2

19.6

20.0

206Pb/ 204 Pb

0.703 0.706 0.709 0.712 0.715 0.71887Sr/ 86Sr

CA - HKCA Aeolian Basalts(central-western AVD sectors)

Tuscany &Roman Provinces

Roccamonfina &Ventotene Island

Procida

Procida

b

a

Vesuvius &Phlegraean Fields

CA - HKCAStromboli

Fig. 8. Sr-Nd-Pb isotopic composition of Ernici mafic rocks. Liter-ature data (Ref.) are from Civetta et al. (1981), D’Antonio et al.(1996) and Conticelli et al. (2002). Compositions of other CA-HKCA Italian volcanoes and Roman Province are also shown (Datafrom Peccerillo, 2005 and references therein).

ment with the few data available from previous studies 39

(D’Antonio et al., 1996; Gasperini et al., 2002). 40

The δ18O vs. 87Sr/86Sr diagram (Fig. 9b) shows that 41

Ernici CA-SHO rocks plot along a sub-vertical trend, a fea- 42

ture observed in many Italian volcanoes. 43

6. Discussion 44

The present study allows us to recognize for the first time 45

low-K2O rocks of CA composition at Ernici, and high- 46

lights an important case, where a close association of calc- 47

alkaline and ultrapotassic magmas occurs in central Italy. 48

The CA rocks have ITE abundances similar to the associ- 49

ated shoshonites, and resemble the equivalent rocks from 50

the Aeolian arc, Procida and Pontine Islands (Ventotene, 51

Fig. 6), although some minor but significant differences 52

in ITE abundances and ratios, and radiogenic isotopic sig- 53

natures are present. On the other hand, Ernici HKS rocks 54

closely resemble the HKS rocks from the Roman volcanoes 55

(i.e., Vulsini to Alban Hills), for trace element enrichment 56

and ratios, and for radiogenic isotope signatures (Fig. 6, 57

and 8). Therefore, the Ernici mafic rocks appear to encom- 58

pass the whole range of geochemical and isotopic features 59

shown by the central-southern Italy magmatism. 60

Page 10

10 M.L. Frezzotti, G. De Astis, L. Dallai, C. Ghezzo Fast Track Article

∂

Fig. 9. a) 208Pb/204Pb vs. 206Pb/204Pb, and b) δ18O vs. 87Sr/86Sr dia-grams for Ernici mafic rocks (MgO > 4 wt.%). Literature data (Ref.)are from Civetta et al. (1981), D’Antonio et al. (1996) and Conticelliet al (2002). Fields of central-southern Italy magmatic provincesarea also shown Data source as in Fig. 8.

The variation diagrams of both incompatible and com-1

patible elements reveal an overall decrease in MgO, Ni2

and Cr, with increasing LILE and HFSE contents (Fig. 3).3

These variations are accompanied by an increase of Sr-4

isotope ratios, and a decrease of Nd and Pb-isotope com-5

positions. Whereas the variations in MgO and ferromagne-6

sian trace elements likely reflect magma evolutionary pro-7

cesses, the isotopic variations could be due to open system8

evolutionary processes (e.g., AFC), or could reflect pris-9

tine magma compositional characteristics, as seen in most10

of Roman Province magmas (e.g., Conticelli & Peccerillo,11

1992; Peccerillo, 2002, and references therein).12

Therefore, the following discussion will focus on: i) the13

roles of shallow level evolution vs. source nature and/or14

processes, in determining the observed rock compositions;15

ii) the genetic relationships between CA, SHO and HKS16

magmatism in the area; iii) the genetic relationships with17

other subalkaline to ultrapotassic rocks in Italy; iv) the geo-18

dynamic implications of the compositional characteristics19

observed at Ernici. These themes apply to other volcanoes20

in Italy, and may help in understanding the genesis and geo-21

dynamic significance of magmatism in the Italian peninsula22

and in the southern Tyrrhenian Sea.23

6.1. Shallow level evolution 24

The broadly negative correlations between MgO and in- 25

compatible trace element (ITE) abundances (Fig. 3) could 26

suggest derivation of HKS magmas from less potas- 27

sic magmas by shallow level evolution processes, dom- 28

inated by fractional crystallisation of dominant clinopy- 29

roxene, plus assimilation of crustal rocks. Clinopyroxene- 30

dominated fractional crystallisation can drive liquid com- 31

positions from CA to SHO and HKS, without changing 32

significantly their silica contents, as suggested by Trig- 33

ila & De Benedetti (1993) for Somma-Vesuvius rocks. 34

Such a process, however, requires substantial assimilation 35

of crustal rocks to explain the observed radiogenic isotope 36

variations. Assuming a CA or SHO starting composition, 37

more than 70% crustal material with a composition as the 38

Hercynian Calabrian basement (Rottura et al., 1991) is nec- 39

essary to obtain HKS magmas. This severely contrasts with 40

the strongly silica undersaturated character of HKS rocks 41

(see also discussion in Peccerillo, 2005). Moreover, oxy- 42

gen isotopic data obtained in this study indicate that HKS 43

values fall within the range of CA-SHO rocks, excluding 44

significant assimilation processes by CA magmas to pro- 45

duce HKS. Derivation of Colle Castellone rocks from HKS 46

parental magmas is similarly problematic. 47

Variation diagrams of some major and trace elements 48

against MgO (Fig. 3) show distinct trends for CA, SHO 49

and HKS, suggesting that individual suites represent dis- 50

crete evolutionary series. In CA rocks, the positive trends 51

of MgO vs. Ni and Cr clearly indicate fractional crystallisa- 52

tion of olivine and clinopyroxene as a first order evolution- 53

ary process. However, the variable oxygen isotopic compo- 54

sitions within this series require an open system evolution. 55

The lack of a significant correlation of δ18Ocpx vs. any in- 56

compatible element concentration and radiogenic isotope 57

ratio (Fig. 7b, d), suggests that heavy oxygen was episod- 58

ically added to CA magmas, without significant addition 59

of any other trace element. This suggests an interaction 60

with a material depleted in ITE, but having high-δ18O, such 61

as sedimentary carbonate country rocks (Peccerillo, 1998). 62

The high δ18Ocpx values of the HKS samples also argues 63

for the interaction of ultrapotassic melts with sedimentary 64

carbonate (cf. Dallai et al., 2004). 65

A further interesting aspect of CA rocks is their wide 66

range in Rb content, which spans over more than one order 67

of magnitude, at nearly constant composition of all other 68

geochemical parameters (Fig. 3). Such a spreading cannot 69

be attributed to secondary processes, because of the lack 70

of petrographic, and chemical evidence for deuteric trans- 71

formations in the studied rocks (i.e., low LOI, relatively 72

invariant concentration of other mobile elements such as 73

alkalies). Therefore, these variations are likely to represent 74

pristine geochemical signatures of CA magmas. It is worth 75

of note that similar or higher Rb variations were also de- 76

tected in potassic rocks at the nearby volcano of Rocca- 77

monfina (Giannetti & Ellam, 1994). 78

The HKS rocks have lower ferromagnesian element con- 79

tents than the bulk of CA-SHO rocks, which might repro- 80

duce a higher degree of evolution, and/or a less enriched 81

nature in compatible elements for the parent magmas (see 82

Page 11

Fast Track Article Calc-alkaline and ultrapotassic magmatism at monti ernici 11

Zr/Th

4

6

8

10

12

14

KSLiterature

0.705

0.706

0.707

0.708

0.709

0.710

0.711

0.71287Sr/86Sr

Ni

0 2 4 6 8 100.705

0.706

0.707

0.708

0,709

0.710

0.711

0.712

K2O

87Sr/86Sr

FCAFCMixing

5

10

15

20

25Zr/Nb

0 20 40 60 80 100 120

ThKS Literature

Mixing

0 20 40 60 80 100 120

KSLiterature

HKSLiterature

HKSLiterature

Pofi volcano

Pofi volcano

HKS Literature

ThHKS Literature

KSLiterature

a b

c dODP655

Roman Province

Ventotene Is. & Roccamonfina

Fig. 10. Inter-element and isotopic variation diagrams for Ernicimafic rocks (MgO > 4 wt%). Symbols as in Fig. 2.

Rogers et al., 1985). Variation of Ni, Cr, and MgO in the1

HKS rocks (Fig. 3, and 10a) is explained by fractional crys-2

tallisation of mafic phases internal to this series.3

SHO rocks have intermediate potassium contents be-4

tween CA and HKS compositions. Some incompatible el-5

ement vs. incompatible element ratios (e.g., Zr/Th vs. Th,6

Fig. 10b) define hyperbolic trends between CA, and Colle7

Castellone HKS compositions, with SHO and HKS plot-8

ting along this trend. This may suggest mixing between9

magmas of extreme compositions to give intermediate-K10

magmas as hybrid products. Interaction between magmas11

with different enrichments in potassium was previously12

suggested by Civetta et al. (1981). However, such a pos-13

sibility is not supported by other geochemical parameters,14

such as Th vs. Zr/Nb, and poorly variable 87Sr/86Sr of CA-15

SHO rocks (Fig. 10c, and d).16

In conclusion, fractional crystallisation, assimilation of17

carbonate rocks, and probably mixing played important18

roles during magma evolution at Ernici. These processes,19

however, did not change significantly the first order trace-20

element compositional characteristics of magmas. Thus,21

we propose that at Ernici, at least three distinct magma se-22

ries – CA-SHO, HKS, and HKS-Colle Castellone – with23

distinct degree of silica saturation, enrichment in incom-24

patible elements, and radiogenic isotope signatures were25

generated and erupted.26

6.2. Petrogenesis of CA-SHO and HKS magmas27

Central Italy magmas originated from an anomalous upper28

mantle, which was contaminated by crustal material (e.g.,29

Hawkesworth & Vollmer, 1979; Peccerillo, 1985; Rogers30

et al., 1985). Several authors (e.g., Peccerillo, 2005 and ref-31

erences therein) identify sedimentary rocks, such as pelites32

and marly sediments, as the crustal component introduced33

in variable amounts into a lherzolitic upper mantle, to gen-34

erate a metasomatised source able to produce the vari-35

able but anomalous compositions of Roman magmas. The36

different degrees of silica saturation have been attributed, 37

based on the experimental evidence, to the depth of partial 38

melting, which generates potassic alkaline magmas with 39

increasing degree of silica undersaturation at increasing 40

pressure (e.g., Wendlandt & Eggler 1980a, and b; Melzer & 41

Foley, 2000). Since magmas with various degrees of silica 42

saturation also have distinct trace element and isotopic sig- 43

natures, it has been concluded that magmas were generated 44

at different depths in an upper mantle that was vertically 45

zoned for ITE and isotopes (Peccerillo, 1999). 46

Alternatively, some authors (e.g., Foley, 1992; Perini 47

et al., 2004) suggest that magmas with different enrich- 48

ments in incompatible elements could derive from partial 49

melting of a mantle crossed by phlogopite veins, involv- 50

ing variable contributions of vein material and host rocks 51

during melting processes: low degrees of partial melting 52

of a veined mantle would allow almost pure vein material 53

to melt, generating highly potassic magmas. At higher de- 54

grees of melting, the same mantle-source would allow in- 55

creasing amounts of lherzolite wall-rock to go into the melt, 56

generating magmas with lower enrichments in incompati- 57

ble elements. In this view, CA and SHO magmas would 58

represent “diluted” HKS melts. 59

This last hypothesis could explain some of the petrologi- 60

cal features of the Ernici rocks. As an example, the overall 61

higher ferromagnesian element contents in the CA rocks 62

could well derive from a higher participation of lherzolite 63

into the melt, whereas the lower Ni and Cr of HKS could 64

reflect the higher proportion of metasomatic veins partic- 65

ipating into the melt. Also the variable enrichments in in- 66

compatible elements could support such a theory. However, 67

this hypothesis fails to account quantitatively for some of 68

the major geochemical variations observed, notably Sr-Nd 69

isotopic signatures. If the Colle Castellone lava is assumed 70

as representative of a pure vein melt, dilution by a melt 71

derived from a normal OIB- or MORB-type mantle would 72

require high amounts of these depleted melts (>70–80 %) 73

to drive Sr isotopic compositions to those of CA rocks 74

(Fig. 8). Therefore, the strong compositional variability of 75

the Ernici rocks likely results from a heterogeneous mantle 76

source. If this conclusion is correct, the isotopic variations 77

observed through the eruptive sequence in the Pofi centre 78

(i.e., 87Sr/86Sr from 0.706446 to 0.709679, Fig. 10c) could 79

testify for a vertically heterogeneous upper mantle. 80

Two temporally distinct mantle metasomatic episodes 81

have been suggested by Peccerillo & Panza in the man- 82

tle beneath Ernici (1999): an earlier event, analogous to 83

that affecting the Roman region, and a later one, similar 84

to that affecting the Campanian region (Peccerillo, 2001, 85

2003). The first metasomatic event could account for the 86

HKS magma compositions, which are close to the equiv- 87

alent rocks of the Roman Province, whereas the latter 88

event could have generated the CA and SHO magmas, 89

which share several elemental and isotopic characteristics 90

with Campanian rocks. However, we note that, while ra- 91

diogenic isotope compositions span the values observed 92

for Campania and Roman provinces – thus supporting a 93

two-stage metasomatic modification for mantle sources – 94

trace element abundances and ratios depict a more complex 95

evolution. 96

Page 12

12 M.L. Frezzotti, G. De Astis, L. Dallai, C. Ghezzo Fast Track Article

ODP655

0

50

100

150

200

250

Ba/Th

0

2

4

6

8

10

La/Nb

0

20

40

60

80

100

Th/Ta

0 2 4 6 8 100100

200

300

400

500

600

700

Zr/Ta

K2O

ODP655

ProcidaVentotene

Somma-Vesuvius

ProcidaVentotene

Somma-Vesuvius

ProcidaSomma-Vesuvius

Aeolian IslandsCA - HKCA

Suites

Aeolian IslandsCA - HKCA

Suites

ODP655

Roman Province

Roman Province

Roman Province

ProcidaSomma-Vesuvius

Roman Province

ODP655

Ventotene Is.Roccamonfina

Aeolian Is.CA - HKCASuites

a b

c d

0 2 4 6 8 10K

2O

Fig. 11. ITE ratios vs. K2O for Ernici mafic rocks (MgO � 4 wt.%)compared with fields of other southern Italy volcanoes. Symbols asin Fig. 5, 6. Data source as in Fig. 8.

Figure 11 shows the comparison of selected key ITE ra-1

tios of Ernici vs. Roman, Campanian, and western-central2

Aeolian arc mafic volcanics. The CA-SHO rocks fall away3

from Campania, and cluster in the field of western-central4

Aeolian arc. In particular, Ernici CA-SHO rocks have5

higher LILE/HFSE (e.g., La/Nb), and distinct LILE/LILE6

(e.g., lower Ba/Th) and HFSE/HFSE ratios (e.g., Zr/Ta)7

than Campanian rocks having similar K2O. Therefore,8

these data suggest a mantle source different from that be-9

neath the Campanian Province, and comparable to that of10

Aeolian volcanoes: the pre-metasomatic mantle composi-11

tion and/or the metasomatic events appear to be different12

for the Campania Province and for the Ernici CA-SHO13

source.14

The relatively low LILE/HFSE ratios at Vesuvius and the15

eastern Aeolian arc has been interpreted to indicate an OIB-16

type source prior to metasomatism (Ellam et al., 1989;17

Serri, 1990; Peccerillo, 2001; De Astis et al., 2003, and18

2004, and references therein). The higher values of these19

ratios and the lower abundance of HFSE at Ernici may re-20

veal a MORB-type pre metasomatic source. This issue will21

be further discussed below.22

As for the metasomatic event affecting source mantle23

rocks, the similar isotopic composition of the Ernici CA-24

SHO rocks, and of the Campanian volcanoes support a25

similar metasomatic agent. The observed difference in26

LILE/LILE ratios between Ernici and Campanian volca-27

noes may in fact either depend on the nature of metaso-28

matic agents or be a consequence of source mineralogy29

(e.g., occurrence of different proportions of phlogopite and30

amphibole), that have profound effects on ITE partition co-31

efficients, and abundance in magmas, (see, De Astis et al.,32

2006). Conversely, Ernici HKS rocks consistently plot in33

the field of the Roman Province, suggesting a similar man-34

tle source (i.e., similar pre-metasomatic composition), and35

nature of metasomatism.36

The issue of a genetic relationship among the different37

Ernici magma series requires a characterisation of the pre-38

metasomatic source composition. Since HFSE are poorly39

mobile during fuid-related metasomatism, the observed 40

HFSE abundances and ratios in the Ernici rocks could iden- 41

tify the pre-metasomatic mantle features. Low HFSE con- 42

centrations in the Ernici rocks suggest a MORB-type man- 43

tle source. The Zr/Nb vs. Th diagram (Fig. 10d) shows 44

that CA-HKCA rocks and “extreme” HKS rocks (i.e., 45

Colle Castellone) have comparable Zr/Nb ranges (CA- 46

HKCA = 13.6–18.4). The bulk of HKS has quite differ- 47

ent and higher Zr/Nb ratios (19.4–22.8). This range closely 48

overlaps the average values of MORB (≈ 22.1), and is sim- 49

ilar to Zr/Nb ratio of MORB-type rocks from the Tyrrhe- 50

nian Sea floor (ODP 655, Fig. 9d). Most of the Ernici 51

CA-HKCA rocks are in the field of Ventotene and Roc- 52

camonfina rocks, and not much different from Aeolian CA 53

rocks (Zr/Nb ≈ 11–12), which have been related to a E- 54

MORB-type pre-metasomatic mantle composition (Ellam 55

et al., 1989: Francalanci et al., 1993). 56

6.3. Relations between CA and SHO magmas 57

Whereas the generation of CA-SHO and HKS magmas in 58

distinct mantle sources is strongly supported by our data, 59

the processes that generated CA and SHO magmas are con- 60

troversial. These two rock groups have very comparable 61

incompatible element patterns (Fig. 5b), suggesting they 62

come from a single mantle source, affected by similar com- 63

positional modifications. However, CA rocks have lower 64

K2O than SHO, yet display higher Cs and very variable Rb. 65

At issue is the nature of the process that generated magmas 66

showing similar compositions for radiogenic isotopes and 67

for all ITE contents, except for K, Rb and Cs. The lack of 68

correlation between K and Rb, and between these and other 69

incompatible elements implies distinct processes for the 70

variation of these two elements. Similar features have been 71

observed also at Roccamonfina, and have been explained 72

in terms of the mineralogy of mantle sources by Giannetti 73

& Ellam (1994). The similar ITE (except for Rb) and ra- 74

diogenic isotopes suggest a homogeneous source rock and 75

similar degrees of partial melting. 76

Potassium is a major element, and its abundance depends 77

on the nature of the mineral phases contributing to the melt. 78

Accordingly, magmas with comparable trace element and 79

radiogenic isotope composition and distinct potassium con- 80

tent could derive from geochemically homogeneous mantle 81

sources, but having a different mineralogy. A phlogopite- 82

bearing source – with this mineral going into the melt – 83

could be responsible for the generation of SHO magmas, 84

whereas comparable degrees of partial melting of a geo- 85

chemically similar source containing amphibole could gen- 86

erate CA magmas. The variable mineralogy in the man- 87

tle might result from variation in the metasomatic fluids, 88

and/or of different pressure of partial melting. Such a very 89

general conclusion, does not explain the variation in Rb, 90

whose abundance can be tentatively ascribed to other pro- 91

cesses, including syn-eruptive transfer from altered HKS 92

rocks. 93

Page 13

Fast Track Article Calc-alkaline and ultrapotassic magmatism at monti ernici 13

6.4. Nature of mantle metasomatic processes1

Geochemical data suggest that the mantle sources of Ernici2

magmas might have been variably modified by LILE-rich,3

and HFSE-poor metasomatic agents, such as a subduction-4

related hydrous fluid and/or melt. The high Sr- and low5

Nd-isotope ratios of Ernici rocks require that mantle con-6

tamination was provided by upper crustal material, such7

as pelitic to marly material (Peccerillo et al., 1999). This8

is supported by Sr-Nd-Pb isotopic ratios of Ernici rocks,9

which plot along a curved (mixing) trend between a HIMU-10

or FOZO-like mantle composition and metapelites (e.g.,11

Gasperini et al., 2002). Further, also incompatible ele-12

ment compositions show many characteristics in com-13

mon with metapelites and marl. However, both CA-SHO14

and HKS rocks have much higher LILE/HFSE ratios than15

both metapelites and marls. Also La/Yb ratios are higher,16

whereas HFSE contents are close to MORB values. More-17

over, the lack of a negative Sr anomaly in the mantle-18

normalised diagrams (except for Colle Castellone), which19

is common in pelites, agrees with marls rather than pelites20

as contaminants for the bulk of Ernici magma sources.21

The high LILE/HFSE ratios and REE fractionation re-22

quire that some element fractionation occurred during man-23

tle contamination. Subduction fluids may well fractionate24

incompatible trace elements. The liquid/mineral partition25

coefficients of trace elements, in fact, vary considerably26

depending on the nature of the agent of metasomatism,27

which has been theoretically and experimentally identi-28

fied in these environments either as aqueous solutions,29

or as hydrous-silicate melts, or as supercritical interme-30

diate aqueo-silicic liquids, mainly depending on temper-31

ature and pressure conditions (e.g., Stalder et al., 2000).32

Recently, experiments by Kessel et al. (2005) have out-33

lined the dichotomy of low-temperature dehydration versus34

high-temperature melting processes of subducting slabs,35

based on the contrasting solubilities of many trace elements36

in metasomatic aqueous fluids or hydrous melts at 4 GPa37

(120 km) (e.g., higher Sr, Ba, Th/U, and LREE in meta-38

somatic melts). At higher pressures (> 6 GPa; 180 km),39

melt-like solubilities are observed for trace elements in su-40

percritical liquids, yet at low temperatures.41

Although a precise reconstruction of the metasomatic42

events is not possible at this stage and requires further43

specific investigations, we propose that the metasomatic44

agents responsible for the modification of the mantle be-45

neath the Ernici region were hydrous fluids, and/or melts46

generated by subducted rocks at different depths. Dehydra-47

tion and/or melting at different P-T conditions generated48

the element fractionation reproduced by the magmas erupt-49

ing at the surface.50

6.5. Geodynamic implications and conclusions51

One of the key problems of a subduction-related ori-52

gin for potassic alkaline volcanism in central Italy is the53

scarcity or absence of calc-alkaline rocks, in comparison54

with the large volume of potassic magmas. However, rocks55

with CA to HKCA affinity occur at Procida and eastern56

Pontine islands (De Astis et al., 2004): Similar rocks are 57

also observed in the late stage of Roccamonfina activ- 58

ity (0.2–0.1 Ma; Peccerillo, 2005 and references therein), 59

whereas CA basalts and andesites with an age � 2 Ma have 60

been found by deep borehole drillings beneath the Campi 61

Flegrei caldera, in the Campanian region (Barbieri et al., 62

1979; De Astis et al., 2006; De Astis, unpublished data). 63

The subalkaline rocks from Ernici analysed in this 64

work also show CA compositions, closely resembling the 65

CA-HKCA rocks from the central-western Aeolian arc 66

(Fig. 11), which are subduction-related, for a number of 67

structural, geophysical and petrological data (see De Astis 68

et al., 2003, and references therein). Therefore, based 69

on the compositional affinities reported above, it can be 70

concluded that a subduction-related volcanism developed 71

along the Italian peninsula (peri-Tyrrhenian margin), erupt- 72

ing CA-HKCA magmas at Ernici, Roccamonfina, Ven- 73

totene, and some Campanian volcanoes, over a time span of 74

about 2 Ma (Metrich et al., 1988; Piochi et al., 2004). From 75

that time, the bulk of calc-alkaline magmatism shifted to- 76

ward the southernmost sector of the Tyrrhenian Sea, giv- 77

ing rise to the Aeolian Volcanic District, as a result of 78

the subduction of the thinned continental or oceanic-type 79

Ionian lithosphere beneath Calabria. However, the Ernici, 80

Roccamonfina and Campanian volcanoes testify for persis- 81

tency of CA-HKCA activity in central Italy. Isotopic sig- 82

natures of these latter rocks suggest introduction of upper 83

crustal material into their mantle source. This likely derived 84

from subduction of the Adriatic plate, which has a conti- 85

nental type composition distinct from the Ionian plate (e.g. 86

Panza & Pontevivo, 2004). The diversity of the subducting 87

foreland may thus provide an explanation for the scarcity 88

of CA-HCKA magmatism in Central Italy compared with 89

the Aeolian Volcanic District. 90

In terms of geodynamic evolution during the Plio- 91

Quaternary, there are several and slightly different geo- 92

physical/petrological models of subduction processes in 93

Central-Southern Italy (e.g. Keller, 1982; Serri et al., 1993; 94

Doglioni et al., 1999; Gvirtzman & Nur, 1999; Lucente 95

et al., 1999; Wortel & Spakman, 2000; Peccerillo, 2005 96

and references therein). Basically, these models differ for 97

the slab geometry beneath the Apennines, for the age and 98

nature of the mantle metasomatism, and/or for the pro- 99

cesses that were responsible the opening of the Tyrrhenian 100

back-arc basin. However, most of these models agree that 101

two different mantle-crust settings developed beneath the 102

northern Apennine chain and beneath the Calabrian Arc 103

– Southern Tyrrhenian Sea, and that the active subduc- 104

tion zones and related volcanism progressively migrated 105

south-eastward to its present position beneath the Aeo- 106

lian arc (Anderson & Jackson, 1987; Selvaggi & Am- 107

ato, 1992; De Astis et al., 2003) due to differential con- 108

tinental collision, and to south-eastward roll-back of the 109

NW dipping Ionian slab. Moreover, both the kinematic 110

model and seismotectonic framework by Meletti et al. 111

(2000) for the Southern Apennines, and the geodynamic 112

setting depicted by De Astis et al. (2006) for the Cam- 113

panian Province and Monte Vulture magmatism, indicate 114

that central-southern Italy recorded a dramatic change in 115

the geodynamic regime at about 0.8–0.7 Ma, because of the 116

Page 14

14 M.L. Frezzotti, G. De Astis, L. Dallai, C. Ghezzo Fast Track Article

end of active subduction, slab detachment (break off) of the1

Adriatic slab (Wortel & Spakman, 2000), and a generalised2

uplift of the chain (Hippolyte et al., 1994), up to the fast3

development of rifting both within the inactive thrust belt4

and along the Tyrrhenian slope. Subsequently, the Campa-5

nian Province experienced an increase in the alkalinity of6

magmatism.7

On the basis of the framework depicted above, the late8

(< 0.2 Ma) occurrence of CA magmatism at Ernici, post-9

dating active subduction and mantle modification processes10

by the westward dipping Adriatic plate, poses the prob-11

lem of the relationships between subduction processes and12

CA magmatism in this area. The late onset of the CA-13

HKCA mafic magmatism at Ernici could reflect an in-14

creased degree of partial melting within the upper mantle,15

generated by back-arc extension and associated isotherms16

uprising following slab detachment, which may have in-17

duced melting in the uppermost levels of a vertically18

zoned, and/or isotopically heterogeneous mantle beneath19

the central-southern Italy. In this case, the transition from20

potassic alkaline to CA magmatism at Ernici would testify21

for isotherms upraise and progressively shallower melting22

events in a vertically zoned mantle sector, variably contam-23

inated by slab derived fluids during active subduction of the24

Adriatic plate.25

Finally, it should be noted that some authors (e.g., Lu-26

cente et al., 1999) suggest the presence of a slab window27

beneath Campania and Ernici-Roccamonfina sectors based28

on tomographic studies. However, slab window formation29

is often associated with OIB-type alkali-mafic volcanism30

(Hole et al., 1991). The opening of such a window be-31

low the central Italy would provide an efficient mecha-32

nism for allowing sub-slab asthenospheric mantle to rise33

and melt in a mantle region previously affected by subduc-34

tion (Thorkelson, 1996), likely interacting with subduction-35

related components. This may explain the mixed OIB- and36

arc-type trace element signatures of Campanian and Vul-37

ture volcanoes, which have much lower LILE/HFSE ratios38

than other central Italy magmas (Peccerillo, 2001, 2005;39

De Astis et al., 2006). However, it is unable to explain the40

high LILE/HFSE ratios of Ernici, which seem to exclude41

an OIB-type component. This does not represent a piece42

of evidence against the slab window model, but militates43

against the involvement of deep mantle components in the44

Ernici magma genesis. The same conclusion also applies45

to Roccamonfina, supporting the idea that OIB-type com-46

ponents in central-southern Italy region likely reflects hori-47

zontal mantle inflow from either the Adriatic plate or from48

the Tyrrhenian Sea area, rather than deep mantle compo-49

nents (Peccerillo, 2001; Peccerillo & Lustrino, 2005; De50

Astis et al., 2006).51

Acknowledgements: Authors are indebted to R.W. Nes-52

bitt and R. Taylor of the School of Ocean and Earth Sci-53

ence, National Oceanography Centre, Southampton for Sr54

isotope analyses, and to G. Davies and W. Lustenhouwer of55

the Instituut voor Aardwetenschappen of the Vrije Univer-56

sitieit in Amsterdam for Sr-Nd-Pb isotope analyses. Sug-57

gestions by Rob Ellam and an anonymous referee greatly58

contribute to improve the manuscript. Research is finan- 59

cially supported by MIUR-PRIN 2004 to M.L.F., and by 60

University of Siena (PAR 2004) to C.G. 61

References 62

Accordi, G., Carbone, F., Civitelli, G., Corda, L., De Rita, D., 63

Esu, D., Funiciello, R., Kotsakis, T., Mariotti, G., Sposato, A., 64

1986. Lithofacies map of Latium–Abruzzi and neighbouring ar- 65

eas. Geological map, 1:250,000. C.N.R., Progetto Finalizzato 66

Geodinamica. 67

Anderson, H. & Jackson, J., (1987): The deep seismicity of the 68

Tyrrhenian sea, Geophys. J. R. Astr. Soc., 91, 613-637 69

Ayuso, R.A., De Vivo, B., Rolandi, G., Seal, R.R., Paone, A. (1998): 70

Geochemical and isotopic (Nd - Pb - Sr - O) variations bear- 71

ing on the genesis of volcanic rocks from Vesuvius, Italy. J. 72

Volcanol. Geotherm. Res., 82, 53-78. 73

Baker, J., Peate, D., Waight, T., Meyzen, C. (2004): Pb isotopic anal- 74

ysis of standards and samples using a 207Pb–204Pb double spike 75

and thallium to correct for mass bias with a double focussing 76

MC-ICP-MS. Chem. Geol., 211, 275-303 77

Basilone, P. & Civetta, L. (1975): Datazione K/Ar dell’attività vul- 78

canica dei Monti Ernici (Latina). Rend. Soc. It. Mineral. Petrol., 79

31, 175-179. 80

Barbieri, M., Di Girolamo, P., Locardi, E., Lombardi, G. and 81

Stanzione, D. (1979): Petrology of the calc-alkaline volcanics 82

of the Parete 2 well (Campania, Italy). Period. Mineral., 48, 53- 83

74. 84

Castellarin, A., Colacicchi, R., Praturlon, A., Cantelli, C. (1982): 85

The Jurassic-Lower Pliocene history of the Ancona-Anzio line 86

(Central Italy). Mem. Soc. Geol. It., 47, 431-447 87

Civetta, L., Innocenti, F., Lirer, L., Manetti, P., Munno, R., 88

Peccerillo, A., Poli, G., Serri, G., (1979): Serie potassica ed 89

alta in potassio dei Monti Ernici (Lazio meridionale): consider- 90

azioni petrologiche e geochimiche. Rend. Soc.It. Miner. Petrol., 91

35, 227-249 92

Civetta, L.„ Innocenti F., Manetti P., Peccerillo A., Poli G. (1981): 93

Geochemical characteristics of potassic volcanics from Mt. 94

Ernici (Southern Latium , Italy). Contrib. Mineral. Petrol., 78, 95

37-47. 96

Conticelli, S. & Peccerillo, A. (1992): Petrology and geochemistry 97

of potassic and ultrapotassic volcanism in central Italy: petro- 98

genesis and inferences on the evolution of the mantle source. 99

Lithos, 28, 221-240. 100

Conticelli, S., D’Antonio M., Pinarelli, L., Civetta, L. (2002): 101

Source contamination and mantle heterogeneity in the gene- 102

sis of Italian potassic and ultrapotassic volcanic Rocks: Sr-Nd- 103

Pb Isotope data from Roman Province and Southern Tuscany. 104

Mineral. Petrol., 74, 189-222. 105

Cox, K.G., Hawkesworth, C.J., O’Nions, R.K., Appleton, J.D. 106

(1976): Isotopic evidence for the derivation of some Roman 107

region volcanics from anomalously enriched mantle. Contrib. 108

Mineral. Petrol., 56, 173-180. 109

Dallai, L., Freda, C., Gaeta, M. (2004): Oxygen isotope geochem- 110

istry of pyroclastic clinopyroxene monitors carbonate contribu- 111

tions to Roman-type ultrapotassic magmas. Contrib., Mineral., 112

Petrol., 148, 247-26. 113

D’Antonio, M., Tilton, G.R., Civetta, L. (1996): Petrogenesis 114

of Italian alkaline lavas deduced from Pb-Sr-Nd isotope 115

Page 15

Fast Track Article Calc-alkaline and ultrapotassic magmatism at monti ernici 15

relationships. Earth Processes: reading the isotopic code.1

Geophys. Monogr., 95, 253-267.2

De Astis, G., Ventura, G., Vilardo, G. (2003): Geodynamic sig-3

nificance of the Aeolian volcanism (Southern Tyrrhenian Sea,4

Italy) in light of structural, seismological, and geochemical data.5

Tectonics, 22, 4, 1040, doi:10.1029/2003TC001506.6

De Astis, G., Piochi, M., Pappalardo, L. (2004): Procida Volcanic7

History: new insights in the evolution of the Phlegraean8

Volcanic District (Campania, Italy). Bull. Volcanol., 66, 622-9

641.10

De Astis, G., Kempton, P.D., Peccerillo, A., Wu, T.W. (2006):11

Trace element and isotopic variations from Mt. Vulture to12

Campanian volcanoes: constraints for slab detachment and man-13

tle inflow beneath southern Italy. Contrib. Mineral. Petrol., DOI14

10.1007/s00410-006-0062-y15

Di Girolamo, P. (1978): Geotectonic settings of Miocene Quaternary16

volcanism in and around the eastern Tyrrhenian Sea border17

(Italy) as deduced from major element geochemistry. Bull.18

Volcanol., 41, 229-250.19

Doglioni, C., Harabaglia, P., Merlini, S., Mongelli, F., Peccerillo,20

A., Piromallo, C. (1999): Orogens and slabs vs. their direction21

of subduction. Earth Sci. Review, 45,167-208.22

Elburg, M., Vroon, P., van der Wagt, B., Tchalikian A. (2005): Sr23

and Pb isotopic composition of five USGS glasses (BHVO-2G,24

BIR-1G, BCR-2G, TB-1G, NKT-1G). Chem. Geol., 223, 196-25

207.26

Ellam, R.M., Hawkesworth, C.J., Menzies, M.A., Rogers, W. (1989)27

The volcanism of southern Italy: Role of subduction and re-28

lationship between potassic and sodic alkaline magmatism. J.29

Geophys. Res. 94, 4589-460130

Foley, S.F. (1992): Veins plus wall-rock melting mechanism in the31

lithosphere and the origin of potassic alkaline magmas. Lithos,32

28, 435-453.33

Francalanci, L., Taylor, S.R., Mc Culloch, M.T., Woodhead, J.34

(1993): Geochemical and isotopic variations in the calc-alkaline35

rocks of Aeolian Arc (Southern Italy): constraints on the magma36

genesis. Contrib. Mineral. Petrol. 113, 300-313.37

Gaeta, M., Freda, C., Christensen, J.N., Dallai, L., Marra, F.,38

Karner, D.B., Scarlato, P. (2006): Time-depend geochemistry of39

clinopyroxene from Alban hills (central Italy): Clues to source40

and evolution of ultrapotassic magmas. Lithos, 86, 330-346.41

Gasperini, D., Blichert-Toft, J., Bosch, D., Del Moro, A., Macera,42

P., Albarede, F., (2002): Upwelling of deep mantle material43

through a plate window: evidence from the geochemistry of44

Italian basaltic volcanics. J. Geophys. Res., 107, B12, 2367-45

2385.46

Giannetti, B. & Ellam, R. (1994): The primitive lavas of47

Roccamonfina volcano, Roman Region, Italy: new constraints48

on melting processes and source mineralogy. Contrib. Mineral.49

Petrol., 116, 21-31.50

Gill, J.B. (1981): Orogenic andesites and plate tectonics. Springer,51

Berlin, 358 p.52

Gvirtzman, Z. & Nur, A., (1999): The formation of Mount Etna as53

the consequence of slab rollback. Nature, 401, 782-785.54

Gvirtzman, Z. & Nur, A.,(2001): Residual topography, lithospheric55

thickness, and sunken slabs in the central Mediterranean. Earth56

Planet. Sci. Lett., 187, 117-130.57

Hawkesworth, C.J & Vollmer, R. (1979): Crustal contamina-58

tion versus enriched mantle 143Nd/144Nd and 87Sr/86Sr mantle:59143Nd/144Nd and 87Sr/86Sr evidence from the italian volcanics.60

Contrib. Miner. Petrol., 69, 151-165.61

Hippolyte, J., Angelier, J., Roure, F. (1994): A major change re- 62

vealed by Quaternary stress patterns in the Southern Apennines, 63

Tectonophysics, 230, 199-210. 64

Hole, M.J., Graeme, R., Saunders, A.D., Storey, M. (1991): 65

Relation between alkalic volcanism and slab-window forma- 66

tion. Geology, 19, 657-660. 67

Irvine, T.N. & Baragar, W.R.A. (1971): A guide to chemical classi- 68

fication of common volcanic rocks. Can. J. Earth Sci., 8, 523- 69

548. 70

Keller, J. (1982): Mediterranean island arcs. in: “Andesites”. R.S. 71

Thorpe, ed. Wiley, Chichester, 307-325. 72

Kessel, R., Schmidt, M.W., Ulmer, P., Pettke, T. (2005): Trace ele- 73

ment signature of subduction-zone fluids, melts and supercriti- 74

cal liquids at 120-180 km depth. Nature, 437, 724-727. 75

Locardi, E. (1988): The origin of the Apenninic arc. Tectonophysics, 76

146, 105-123. 77

Lucente, F.P., Chiarabba, C., Cimini, G.B., Giardini, D. (1999): 78

Tomographic constraints on the geodynamic evolution of the 79

Italian region. J. Geophys. Res., 104, 20307-20327. 80

Mattey, D.P, Lowry, D., Macpherson, C.G (1994): Oxygen isotope 81

composition of mantle peridotite. Earth Planet. Sci. Lett., 128, 82

231-241. 83

Meletti, C., Patacca, E, Scandone, P. (2000): Construction of a seis- 84

motectonic model: the case of study. Pure Appl. Geophys., 157, 85

11-35. 86

Meloni, A., Alfonsi, L., Florindo, F., Sagnotti, L., Speranza, F., 87

Winkler, A. (1997): Neogene and Quaternari geodynamic evo- 88

lution of the Italian peninsula: the contribution of paleomagnetic 89

data. Ann. Geophys., 40, 705-727. 90

Melzer, S. & Foley, S.F. (2000): Phase relations and fractionation 91

sequences in potassic magma series modelled in the system 92

CaMgSi2O6-KAlSiO4-Mg2SiO4-F at 1 bar to 18 kbar. Contrib. 93

Mineral. Petrol., 138,186-197. 94

Metrich, N., Santacroce, R., Savelli, C. (1988): Ventotene, a potas- 95

sic Quaternary volcano in central Tyrrhenian Sea. Rend. Soc. It. 96

Mineral. Petrol., 43, 1195-1213. 97

Nakamura, N. (1974): Determinations of REE, Ba, Fe, Mg, Na 98

and K in carbonaceous and ordinary chondrites. Geochim. 99

Cosmochim. Acta, 38, 757-773. 100

Panza, G.F., (1984): Structure of the lithosphere-asthenosphere sys- 101

tem in the Mediterranean region. Ann. Geophys., 2, 137-138. 102

Panza, G.F., & Pontevivo, A. (2004): The Calabrian Arc: a detailed 103

structural model of the lithosphere-asthenosphere system. Rend. 104

Accad. Naz. Sci. dei XL, Mem. Sci. Fis. Nat., 28, 51-88. 105

Peccerillo, A. (1985): Roman Comagmatic Province (Central Italy): 106

evidence for subduction-related magma genesis. Geology, 13, 107

103-106. 108

Peccerillo, A. (1998): Relationships between ultrapotassic and 109

carbonate-rich volcanic rocks in central Italy: petrogenetic im- 110

plications and geodynamic significance. Lithos, 43, 267-279. 111

Peccerillo, A. (1999): Multiple mantle metasomatism in central- 112

southern Italy: geochemical effects, timing and geodynamic im- 113

plications. Geology, 27, 315-318. 114

Peccerillo, A. (2001): Geochemical similarities between the 115

Vesuvius, Phlegraean Fields and Stromboli volcanoes: petro- 116

genetic, geodynamic and volcanological implications. Miner. 117

Petrol., 73, 93-105. 118

Peccerillo, A. (2002): Plio-Quaternary magmatism in central- 119

southern Italy: a new classification scheme for volcanic 120

provinces and its geodynamic implications. in “Geological and 121

Page 16

16 M.L. Frezzotti, G. De Astis, L. Dallai, C. Ghezzo Fast Track Article

geodynamic evolution of the Apennines” R.M. Barchi, S. Cirilli,1

G. Minelli, eds. Boll. Soc. Geol. It., Spec Vol, 1, 113-127.2

Peccerillo, A. (2003): Plio-Quaternary magmatism in Italy.3

Episodes, 26, 222-226.4

Peccerillo, A. (2005): Plio-Quaternary volcanism in Italy. Petrology,5

Geochemistry, Geodynamics. Springer Berlin Heidelberg New6

York. 364 p.7

Peccerillo, A., & Lustrino, M. (2005): Compositional variations8

of the Plio-Quaternary magmatism in the circum-Tyrrhenian9

area: deep- vs. shallow-mantle processes. in “Plates, plumes10

and paradigms” G.R. Foulger, J.H. Natland, D.C. Presnall, D.L.11

Anderson eds. Geol. Soc. Am., Spec. Paper, 388, 421-434.12

Peccerillo, A., & Panza, G.F., (1999): Upper mantle domains be-13

neath Central–Southern Italy: petrological, geochemical and14

geophysical constraints. Pure Appl. Geophys., 156, 421-443.15

Peccerillo, A., & Taylor, S.R. (1976): Geochemistry of Eocene calc-16

alkaline volcanic rocks of the Kastamonu area, northern Turkey.17

Contrib. Mineral. Petrol., 58, 63-81.18

Peccerillo, A., Dallai, L., Frezzotti, M.L., Kempton, P. (2004): Sr–19

Nd–Pb–O isotopic evidence for decreasing crustal contamina-20

tion with ongoing magma evolution at Alicudi volcano (Aeolian21

arc, Italy): implications for style of magma-crust interaction and22

for mantle source compositions. Lithos, 78, 217-233.23

Perini, G., Ponticelli, S., Francalanci, L., & Davidson, J.P. (2004):24

Magmatological evolution and genesis of magmas from Vico25

volcano, Central Italy: multiple differentiation pathways and26

variable parental magmas. J. Petrol., 45, 139-182.27

Piochi, M., Pappalardo, L., De Astis, G. (2004): Geochemical28

and isotopical variation whitin the Campanian Comagmatic29

province: implications on magma source composition. Annales30

of Geophys., 47, 1485-149931

Piromallo, C. & Morelli, A. (2003): P wave tomography of the man-32

tle under the Alpine-Mediterranean area. J. Geophys. Res., 108,33

B2, 2065, doi: 10.1029/ 2002JB001757.34

Rogers, N., W., Hawkesworth, C.J., Parker, R.J., Marsh, J.S. (1985):35

The geochemistry of potassic lavas from Vulsini, central Italy,36

and implications for mantle enrichment processes beneath the37

Roman Region. Contrib. Mineral. Petrol., 90, 244-257.38

Rottura, A., Del Moro, A., Pinarelli, L., Petrini, R., Peccerillo,39

A., Caggianelli, A., Bargossi, G.M., Piccarreta, G. (1991):40

Relationships between intermediate and acidic rocks in oro-41

genic granitoid suites: petrological, geochemical and isotopic42

(Sr, Nd, Pb) data from Capo Vaticano (Southern Calabria, Italy).43

Chemical Geol., 92, 153-176.44

Savelli, C. (1988): Late Oligocene to Recent episodes of magma-45

tism in and around the Tyrrhenian Sea: implications for the pro-46

cesses of opening in a young inter-arc basin of intra-orogenic47

(Mediterranean) type. Tectonophysics, 146, 163-181.48

Selvaggi, G. & Amato, A., (1992): Subcrustal earthquakes in the49

Northern Apennines Italy: evidence for a still active subduction?50

Geophys. Res. Lett., 19, 2127-2130.51

Serri, G. (1990): Neogene-Quaternary magmatism of the Tyrrhenian52

region: characterization of the magma sources and geodynamic53

implications. Mem. Soc. Geol. It., 41, 219-242.54

Serri, G., Innocenti, F., Mannetti, P. (1993): Geochemical and petro- 55

logical evidence of the subuction and delaminated Adriatic con- 56

tinental lithosphere in the genesis of the Neogene-Quaternay 57

magmatism of central Italy. Tectonophysics, 223, 117-147. 58

Sharp, Z.D. (1990): A laser-based microanalytical method for the 59

in situ determination of oxygen isotope ratios of silicates and 60

oxides. Geochim. Cosmochim. Acta, 54, 1353-1357. 61

Sharp, Z.D., (1995): Oxygen isotope geochemistry of the Al2SiO5 62

polymorphs. Am. J. Sci., 295, 1058-1076. 63

Stalder, R., Ulmer, P., Thompson, A.B., Günther, D. (2000): 64

Experimental approach to constrain second critical endpoints in 65

fluid/silicate systems: Near-solidus fluids and melts in the sys- 66

tem albite-water. Amer. Mineral., 85, 68-77. 67

Sun, S.S. & Mc Donough, W.F. (1989): Chemical and isotopic sys- 68

tematics of oceanic basalts: implications for mantle composition 69

and processes. in: “Magmatism in ocean basins” A.D. Saunders 70

& M.J. Norry, eds. Geol. Soc. London., Spec. Publ., 42, 313- 71

345. 72

Thorkelson, D.J. (1996): Subduction of diverging plates and the 73

principles of slab window formation. Tectonophysics, 255, 47- 74

63. 75

Trigila, R. & De Benedetti, A. (1993): Petrogenesis of Vesuvius his- 76

torical lavas constrained by Pearce element ratios analysis and 77

experimental phase equilibria. J. Volcanol. Geotherm. Res., 58, 78

315-343. 79

Turi, B. & Taylor, H.P. (1976): Oxygen isotope studies of potassic 80

volcanic rocks of the Roman Province, central Italy. Contrib. 81

Mineral. Petrol., 55, 1-31. 82

Turi, B., Taylor, H.P. & Ferrara, G. (1991): Comparisons of 18O/16O 83

and 87Sr/86Sr in volcanic rocks from the Pontine Islands, M. 84

Ernici and Campania with other areas in Italy. in “Stable Isotope 85

Geochemistry: A Tribute to Samuel Epstein”. H.P. Taylor, J.R. 86

O’Neil, I.R. Kaplan, eds. Geochem. Soc. Spec. Publ., 3, 307-324 87

Vollmer, R. (1989): On the origin of the Italian potassic magmas. A 88

discussion contribution. Chem. Geol., 74, 229-239. 89

Wendlandt, R.F. & Eggler, D.H. (1980a): The origins of potassic 90

magmas: 1. Melting relations in the systems KAlSiO4-Mg2SiO2 91

and KAlSiO4-MgO-SiO2-CO2 to 30 kilobars. Am. J. Sci., 280, 92

385-420. 93

Wendlandt, R.F. & Eggler, D.H. (1980b): The origins of potassic 94

magmas: 2. Stability of phlogopite in natural spinel lherzolite 95

an in the KAlSiO4-MgO-SiO2-H2O-CO2 at high pressures and 96

high temperatures. Am. J. Sci., 280, 421-458. 97

Wood, D.A., Joron, J.L., Treuil, M., Norry, M., Tarney, J. (1979): 98

Elemental and Sr isotope variations in basic lavas from Iceland 99

and the surrounding ocean floor. Contrib. Mineral. Petrol., 40, 100

319-339. 101

Wortel, M.J.R. & Spakman, W. (2000): Subduction and slab de- 102

tachment in the Mediterranean-Carpathian region. Science, 290, 103

1910-1917. 104

Received 23 October 2006 105

Modified version received 26 March 2007 106

Accepted 18 June 2007 107