The Axial Granulite Complex

The Axial Granulite Complex mainly consists of granulitic orthogneisses and charnockites. Its prominent ductile fabrics (D_M2) oscillate in trend between NW–SE and NE–SW (Pinna et al. 1993). The peak granulite facies metamorphism is dated between 571 and 549 Ma by U–Pb sensitive high-resolution ion micro-probe (SHRIMP) and Pb/Pb evaporation analyses of individual metamorphic zircon grains (Kröner et al. 2001). Subsequently, isobaric cooling caused a static retrograde amphibolite facies overprint (Pinna et al. 1993; Kröner et al. 2001). Amphibolite facies deformations of likely Pan-African age (D_P1–D_P3), as exemplified in the Geci Group (Lulin 1985), are mainly confined to transpressive thrust zones that trend between ENE–WSW and NW–SE (e.g. Costa et al. 1992).

The Lurio Belt foreland

The southern basement region is bound to the north by the ENE–WSE-trending Lurio Belt (Cadoppi et al. 1987; Pinna et al. 1993; Kröner et al. 1997). This belt is a discontinuous high-strain zone of dismembered layers, pods and lenses of granulite facies rocks (Norconsult Consortium 2007), and extends from the Indian Ocean into southern Malawi (Fig. 2). The southern basement dominantly consists of biotite–hornblende and migmatic, leucoratic gneisses (Nampula Supergroup). Supracrustal, mainly meta-sedimentary units of allochthonous (Chiure Supergroup) and autochthonous origin (Mecuburi Group), cover this basement. Remenant, granulite facies ortho- and paragneisses (Granulite Klippen) rest as unrooted klippen on the Nampula basement. Single-grain Pb/Pb evaporation and U–Pb SHRIMP ages of c. 615 Ma of metamorphic zircons from the granulites are interpreted to date the peak granulite facies metamorphism in the southern basement and in the Lurio Belt (Kröner et al. 1997). This event is post-dated by an unconstrained retrograde amphibolite facies overprint (Pinna et al. 1993). One Rb–Sr biotite cooling age of 449 ± 7 Ma (Costa et al. 1992) is reported from a meta-granitoid in the Nampula basement (Fig. 2).

In the southern basement the principle ductile fabrics (D_M2) display easterly trends (NE–SW to NW–SE) that broadly parallel the Lurio Belt orientation (Pinna et al. 1993). The NNE–SSW-trending Namama Thrust Belt is located in the central southern basement (Fig. 2). According to Cadoppi et al. (1987) the eastward thrusting altered the basement's ductile fabric (D_M2 of Pinna et al. 1993), at all scales, to strongly aligned east–west trends. The Namama Thrust Belt is considered to be younger than the Lurio Belt (Cadoppi et al. 1987) and is probably of Pan-African age (Sacchi et al. 2000).

Gabbroic–granitic intrusives of late Pan-African age (c. 500 Ma) cross-cut all units in the southern basement. Some intrusions exhibit weak, linear fabrics and elliptic shapes that parallel the Lurio Belt trend. These features are interpreted to indicate their syn- to late-kinematic emplacement (Pinna et al. 1993). Rb–Sr biotite cooling ages obtained from these bodies (Fig. 2) range between 420 and 434 Ma (Sacchi et al. 1984; Costa et al. 1992). Undeformed pegmatite occurrences in the southern basement are largely regarded as Pan-African in age (Afonso 1976; Araujo 1976; Costa et al. 1992). For one them a Rb–Sr muscovite age of 454±7 Ma is reported by Costa et al. (1992) (Fig. 2).

In this paper we follow the tectonostratigraphic scheme of Pinna et al. (1993). We would like to make the reader aware that recent mapping efforts in Mozambique will soon result in a significantly revised tectonostratigraphic framework (Bjerkgård et al. 2006; Hollick et al. 2006; Thomas et al. 2006; Viola et al. 2006).

Late Palaeozoic–Early Mesozoic intracontinental rift basin

In the Late Carboniferous–Late Triassic/Early Jurassic, the regional tectonic regime of NE Gondwana was governed by tensional stresses that propagated from the diverging southern Tethyan margin southwards into the supercontinent (Wopfner 2002). Within the East African sector this tectonic regime resulted in the formation of graben and extended intracratonic rift structures. These rifts are filled with thick and mainly terrigenous sedimentary deposits (Karoo Group equivalents) that are of remarkable similarity across eastern Africa (Wopfner 1993, 1994; Visser & Praekelt 1996; Reeves et al. 2002). In the entire region, Late Triassic deposits are terminated by a major unconformity that reflects the incipient Gondwana break-up and terminates the intracratonic rift stage (Catuneanu et al. 2005 and references therein).

Rift basins fringing the north Mozambican basement to the west (Shire Valley, Cabora Bassa Basin), north (Metangula Basin) and NE (Selous Basin) uniformly preserved Karoo age, terrestrial infillings of Early/Late Permian–Triassic and/or Jurassic age (Fig. 2). Within these basins, syn-sedimentary faulting observed in the oldest Karoo age strata indicates a common occurrence of incipient sedimentation and tectonic activity in Early–Late Permian times (Habgood 1963; Verniers et al. 1989; Castaing 1991; Catuneanu et al. 2005 and references therein). The Cabora Bassa Basin and the Shire Valley (Fig. 2) formation was linked to the evolution of the sinistral Zambesi pre-transform system, and was governed by a prominent NW–SE and a subordinate NE–SW tensional stress fields (Castaing 1991). Instead, the evolution of the Metangula and Selous basins was guided by the development of the Malagasy Chasm rift system between Africa and Madagascar (Verniers et al. 1989; Wopfner & Kaaya 1991; Wopfner 2002). In SW Madagascar, intracontinental rift basins contain Late Carboniferous/Early Permian–Late Triassic strata (Besairie & Collignon 1972), whereby earliest pull-apart basins initiated along NE–SW-trending sinistral strike-slip zones in the Early–Late Permian (Schandelmeier et al. 2004).

Mesozoic marginal rift basins

The Southern Rovuma Basin, located along the NE coast of Mozambique, represents the southern termination of the pericratonic East African marginal basin. In the south the north Mozambican basement is bordered by the Mozambique Basin (Figs 1 & 2). Both basins mainly contain marine sedimentary deposits of Early/Middle Jurassic–Cenozoic age (Salman & Abdula 1995). In Rovuma Basin Karoo age strata are inferred below Jurassic deposits based on seismic correlations with the Tanzanian Selous Basin. In the Mozambique Basin their existence is only supposed (Salman & Abdula 1995).

⁴⁰Ar/³⁹Ar analysis

Hornblende and biotite grains were extracted following standard mineral separation procedures. Approximately 300 µm-sized grains were selected from the 150–315 µm sieve fraction. The mineral separates were individually wrapped in aluminium foil packets, and all the samples were inserted into an aluminium irradiation package. Biotite age standard Tinto B (K–Ar age of 409.2 ± 0.7 Ma: Rex & Guise 1995) was placed at 5 mm intervals along the package to monitor the neutron flux gradient. The package was Cd-shielded and irradiated in the 5C position of the McMaster University Nuclear Reactor, Hamilton, Canada, for 90 h. Upon return, the samples were loaded into an ultra-high vacuum laser chamber fitted with a Kovar viewport and baked to 120 °C overnight to remove adsorbed atmospheric argon from the samples and chamber walls.

A 110 W Spectron Laser Systems continuous-wave neodymium–yttrium–aluminium–garnet (CW–Nd–YAG) (λ=1064 nm) infra-red laser, fitted with a TEM00 aperture, was used to slowly laser step-heat the mineral samples. The laser was fired through a Merchantek computer-controlled X–Y–Z sample chamber stage and microscope system, fitted with a high-resolution CCD (charged-coupled device) camera, 6× computer-controlled zoom, high-magnification objective lens and two light sources for sample illumination. The gases released by laser heating were ‘gettered’ using three SAES AP10 getter pumps to remove all active gases. The remaining noble gases were equilibrated into a high-sensitivity mass spectrometer (MAP 215-50), operated at a resolution of 570 and fitted with a Balzers SEV 217 multiplier. The automated extraction and data acquisition system was computer controlled, using a LabView program. The mean 5 min extraction system blank Ar isotope measurements obtained during the experiments were 1.18×10⁻¹², 1.32×10⁻¹⁴, 4.65×10⁻¹⁵, 6.70×10⁻¹⁴ and 1.43×10⁻¹⁴ cm³ STP (standard temperature and pressure) for ⁴⁰Ar, ³⁹Ar, ³⁸Ar, ³⁷Ar and ³⁶Ar, respectively. The Ar isotope analyses were corrected for system blanks, mass discrimination (⁴⁰Ar/³⁶Ar=281.0), radioactive decay of ³⁷Ar, and minor interference reactions from Ca and K (³⁹Ar/³⁷Ar_Ca=0.00065, ³⁶Ar/³⁷Ar_Ca=0.000255 and ⁴⁰Ar/³⁹Ar_K=0.0015). Errors quoted on the ages (Tables 1 & 2) are 1σ and include the J value error (Mitchell 1968). The ⁴⁰Ar/³⁹Ar ages were calculated using the decay constant of Steiger & Jäger (1977). J values and errors are noted on the sample ⁴⁰Ar/³⁹Ar data tables (Table 2). The ⁴⁰Ar/³⁹Ar data presented in this study were undertaken at the Western Australian Argon Isotope Facility (Curtin University and the University of Western Australia).

Titanite fission-track analysis

Titanite separates with grain sizes of 150–315 µm were extracted using conventional preparation techniques including crushing, sieving, Wilfley table, heavy liquid and magnetic separation. Batches of titanite grains were embedded in epoxy resin, then ground and polished to expose internal crystal surfaces. The titanite fission tracks were revealed by etching the polished crystal mounts in an acid solution of 1 part concentrated HF, 2 parts concentrated HNO₃, 3 parts concentrated HCl and 6 parts H₂O. Samples were etched individually at room temperature for 17–27 min (Naeser & McKee 1970). Distinctly recognizable terminations of confined tracks were used as an evaluation criterion for sufficient fission-track etching. All the samples were loaded into aluminium capsules and irradiated at the FRM II research reactor facility in Garching, Germany. Corning dosimeter glasses (CN 2) were used to monitor a neutron fluence gradient. An irradiation time of 60 s was applied to obtain a total thermal neutron fluence of 0.5×10¹⁶ neutrons cm⁻². Induced tracks were recorded in white micas following the external detector approach (Gleadow 1981) and revealed after irradiation by etching the micas for 15 min in 40% HF at 21±1 °C. All track density measurements in mounts and micas were performed using a Zeiss® Axiophot microscope. The fission-track ages were calculated according to the zeta calibration method (Hurford 1990; Galbraith & Laslett 1993) using a weighted mean zeta factor (ζ) of 134.5±8.2 obtained from Fish Canyon Tuff and Mt Dromedary TFT age standards. Errors are quoted at the 2σ level and were derived according to conventional methods (Green 1981).

⁴⁰Ar/³⁹Ar hornblende analysis

The ⁴⁰Ar/³⁹Ar results of six hornblende samples (BZ 216, GZ 90, PZ 37, RMZ 11, RMZ 13 and RMZ 45) are presented in Tables 1 and 2, and Figure 3. All hornblende samples exhibit disturbed age spectra and yield no plateau ages. In all samples, old ages are preserved in the low-temperature steps of the sample's age spectra (Fig. 3a–f). Some of these ages are anomalously old of up to 3.8 Ga (e.g. sample BZ 216, Fig. 3a). All samples contained very high ⁴⁰Ar* concentrations of 95–99%, with only the occasional presence of 11–36% atmospheric ⁴⁰Ar in sample PZ 37 (Table 2). No statistically reliable inverse isochron ages were obtained. Pseudo-plateau and weighted mean ages for all hornblende samples range from 462.5±2.7 to 550.7±3.2 Ma, and are statically unreliable (Table 2). They either comprise less than 45% of the cumulative ³⁹Ar (BZ 216 and PZ 37) or yield mean squared weighted deviate (MSWD) and probability values higher than 2.5 and lower than 0.05 (GZ 90, RMZ 11, RMZ 13 and RMZ 45).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

³⁷Ar/³⁹Ar and apparent age spectra plots of incremental heating steps from hornblende sample analyses. Insets are presented for samples exhibiting strongly discordant spectra and depict entire age range of incremental heating steps. Errors for each step are ±1σ and do not include the error of the J value.

⁴⁰Ar/³⁹Ar biotite analysis

The ⁴⁰Ar/³⁹Ar results of three biotite samples GZ 39, RMZ 13 and RMZ 18 are given in Tables 1 and 2, and Figure 4a–c. The biotite samples contain high concentrations of ⁴⁰Ar* (92–100%), with only the occasional presence of 11–28% atmospheric ⁴⁰Ar in samples GZ 39 and RMZ 13 (Table 2). Only sample GZ 39 yielded a statistically reliable inverse isochron age of 440.4 ± 4.1 Ma. Biotite samples GZ 39 and RMZ 18 yielded plateau ages of 443.8 ± 0.4 and 428.4 ± 0.3 Ma. The highest-temperature steps (i.e. > 80% cumulative ³⁹Ar) of both samples revealed discordant ⁴⁰Ar/³⁹Ar age spectra (Fig. 4a, c). Sample RMZ 13 yielded a discordant ⁴⁰Ar/³⁹Ar age spectrum (Fig. 4b) with no plateau age. A calculated weighted mean age of 448.2 ± 3.7 Ma (steps 2–6) is consistent with the sample's total fusion age of 445.9 ± 10.4 Ma (Table 1). The first low-temperature steps of GZ 39 and RMZ 13 yielded ages that are younger than the plateau and weighted mean ages (Table 2 and Fig. 4b, c). RMZ 18 exhibits low-temperature step ages that are slightly older than the obtained plateau age (Fig. 4a and Table 2).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

³⁷Ar/³⁹Ar and apparent age spectra plots of incremental heating steps from biotite samples analyses. Insets are presented for samples exhibiting strongly discordant spectra and depict entire age range of incremental heating steps. Errors for each step are ±1σ and do not include the error of the J value.

Titanite fission-track analysis

The ages and analytical details of 25 TFT analyses are presented in Table 3. Their spatial distribution pattern is depicted in Figures 5 and 6. The TFT ages range from 384 ± 40 to 219 ± 24 Ma and do not show a distinct trend with elevation (Fig. 5a, b). The associated errors are quoted at 2σ confidence levels. The χ² probability values of all samples are larger than 5%. All TFT ages are younger than the ⁴⁰Ar/³⁹Ar hornblende and biotite ages of this study.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.

Diagrams depicting the results of the TFT analyses; weighted mean ages of Y and O populations are depicted as black horizontal or vertical lines, and the grey bars indicate the associated ±2σ confidence intervals. (a) Latitude v. age plot. (b) Age v. elevation plot. (c) Age v. elevation plot of traverses A, B and C. (d) Latitude v. age plot of traverses A, B and C.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 6.

Topographic map of northern Mozambique with superimposed spatial distribution pattern of TFT ages. In general, the TFT ages remain constant in the east–west direction and young from north to south. Contour lines were calculated using Generic Mapping Tool (GMT) 4.0, module surface. Sample locations are given as black squares and black circles for the O and Y groups, respectively. The open grey boxes indicate ungrouped samples. TFT age trend traverses A, B and C are plotted in Figure 5 and are drawn here as blue lines.

On the age v. latitude plot (Fig. 5c) two TFT age clusters are apparent, although partly overlapping within their 2σ confidence intervals. In the north of the study area (Figs 5c & 6) eight TFT ages (termed the O group) range from 378±40 to 291±30 Ma and group around a weighted mean age of 329±24 Ma. A younger TFT age population of 17 samples with ages spanning between 306 ± 34 and 219±24 Ma (termed the Y group) clusters around a weighted mean age of 255±20 Ma. The Y group occupies the central–southern part of the study area. The grouping is confirmed by a discriminant analysis (Bahrenberg et al. 1992), using the TFT ages, latitudes, longitudes and elevations (Table 3) as class parameters. A Wilks' Λ value of 0.1612 indicates a distinct class separation, and a probability value of P(χ²) > 92% shows that the estimated discriminant functions are significant.

TFT samples from three north–south traverses, A, B and C, delineate the age v. latitude trends across the region from west to east (Fig. 6). Profile B covers the largest section, whereas profile C represents a composite of two locations (Figs 5c, d & 6). Generally, the TFT ages decrease from north to south. The ages remain constant in the NE–SW to ENE–WSW directions between traverses A and B (35.5°E–37°E), and in the east–west direction between traverse B and C (37°E–40.5°E) (Fig. 6). Traverses A and B exhibit an abrupt change in the TFT ages from the O to the Y group over a range of approximately 0.5° longitude (c. 55 km). The location of sudden TFT age change in profile A appears to be offset by approximately 1° towards the south against the one in profile B (Fig. 5c, d).

⁴⁰Ar/³⁹Ar hornblende data

The observed discordant hornblende age spectra (Fig. 3a–f) are interpreted to result from the degassing of inclusions or impurities incorporated in the analysed crystals (Lee et al. 1991; Lee 1993). A degassing of non-hornblende phases is corroborated as the discordant hornblende age spectra are usually accompanied by similar variations in their corresponding ³⁷Ar/³⁹Ar ratios (Fig. 3a–f) (Di Vincenzo et al. 2003).

We interpret the old ages of the first, low-temperature age spectra steps in all hornblende samples (Fig. 3a–f) to reflect the release of excess ⁴⁰Ar (Damon 1968) from extended crystal defects by short-circuit diffusion (Lee 1993, 1995; Lo et al. 2000). The very high radiogenic ⁴⁰Ar* and very low atmospheric ⁴⁰Ar concentrations in all hornblende samples (Table 1) suggest that their argon budgets were not significantly altered by mixing with meteoric fluids. As optical thin-section inspection revealed fresh, unaltered mineral phases, a large-scale, external excess ⁴⁰Ar input by non-meteoric fluids is considered less likely. We suggest that the observed excess ⁴⁰Ar may either have accumulated during original mineral formation or may be derived from internal build-up above the isotopic system closure (Baxter et al. 2002; Baxter 2003). As a consequence, the weighted mean and pseudo-plateau ages of all hornblende samples might also be affected by excess ⁴⁰Ar. Some of the low-temperature excess ⁴⁰Ar in PZ37 also appears to be associated with a Ca-rich phase (Fig. 3c).

We infer that the discrepant total fusion and weighted mean ages of BZ 216 and PZ 37 reflect a significant alteration by excess ⁴⁰Ar and degassing of non-hornblende phases. The ⁴⁰Ar/³⁹Ar ages of these two samples are interpreted as geologically meaningless. Because of the good agreement between the total fusion (466±30–569± 46 Ma) and weighted mean ages (455±6–550±3 Ma) in samples RMZ 11, RMZ 13, RMZ 45 and GZ 90 (Fig. 3d–f, and Tables 1 & 2), we suggest that these samples are least affected by excess ⁴⁰Ar. Their weighted mean ages of approximately 455–550 Ma are considered as the best estimations for hornblende ⁴⁰Ar/³⁹Ar cooling ages in the region.

⁴⁰Ar/³⁹Ar biotite data

The discordant ⁴⁰Ar/³⁹Ar age spectra pattern of RMZ 13 and the high-temperature age spectra steps of GZ 39 and RMZ 18 (cumulative ³⁹Ar > 80%) are accompanied by similar variations in their corresponding ³⁷Ar/³⁹Ar ratios (Fig. 4a–c and Table 1). These patterns are interpreted as resulting from the degassing of non-pristine biotite phases, e.g. inclusions (Lo et al. 2000; Kuiper 2002; Di Vincenzo et al. 2003).

The low ages in the first, low-temperature spectra steps of GZ 39 and RMZ 13 are not accompanied by significantly lowered ³⁷Ar/³⁹Ar ratios (Fig. 4b, c and Table 1). Both observations indicate either loss of ⁴⁰Ar or the degassing of non-biotite phases (Lo et al. 2000). The slightly older ages in the first, low-temperature age-spectra step of sample RMZ 18 may reflect the degassing of minor amounts of excess ⁴⁰Ar from extended defects by short-circuit diffusion (Lo et al. 2000).

Because of minor interferences from ⁴⁰Ar loss in RMZ 13 and excess ⁴⁰Ar in RMZ 18, we suggest that both their weighted mean and plateau ages of 448.2±3.7 and 428.4±0.3 Ma provide good biotite ⁴⁰Ar/³⁹Ar cooling age estimates in the region. Owing to a possible influence of ⁴⁰Ar loss, the plateau age of sample GZ 39, 443.8±0.4 Ma, is regarded as a minimum biotite ⁴⁰Ar/³⁹Ar cooling age estimate.

Cooling rates for ⁴⁰Ar/³⁹Ar hornblende and biotite results

The ⁴⁰Ar/³⁹Ar analyses of sample RMZ 13 yield hornblende and biotite cooling ages of 475.8 ± 2.8 Ma and 448.2±3.7 Ma, respectively (Δt = 21–34 Ma: Fig. 1 and Table 2). Iteratively derived cooling rates range from c. 10.5 to 6.5 °C Ma⁻¹ and correspond to closure temperatures of c. 524–513 and 311–305 °C for hornblende and biotite, respectively. Hornblende sample RMZ 11 has a proximate location to, and yields a similar hornblende ⁴⁰Ar/³⁹Ar cooling age of 473.8±2.4 Ma as, sample RMZ 13 (Table 2 and Fig. 2). Therefore, similar cooling rate and closure temperature values are inferred for RMZ 11. Identical biotite ⁴⁰Ar/³⁹Ar cooling ages of sample RMZ 13 (448.2±3.7 Ma) and sample GZ 39 (443.8±0.4 Ma) imply similar minimum cooling rates of 6.5–10.5 °C Ma⁻¹ and closure temperatures of 311–305 °C for GZ 39. Hornblende sample RMZ 45 (455.5 ± 6.2 Ma) and biotite sample RMZ 18 (428.4±0.3 Ma) yielded younger ages than the hornblende and biotite cooling ages of sample RMZ 13. The approximated cooling rates of 5–50 °C Ma⁻¹ as well as the associated closure temperatures of 556–511 and 334–302 °C, are also applied to RMZ 45 and RMZ 18, respectively.

Titanite fission-track data

The χ² probability values of all TFT samples (>5%) demonstrate that in each sample the dispersion of the single-grain ages is explained by a Poissonian distribution and reflects a single population. As no track-length information was obtained, all TFT ages are regarded as minimum cooling ages, with a TFT closure temperature of 275±25 °C (Kohn et al. 1993).

Annealing characteristics of titanite fission-tracks are fairly complex and rather poorly constrained (Jonckheere & Wagner 2000). We therefore interpret the broad TFT age scatters in both age groups (c. 90 Ma, Table 3) and the partial age overlaps between them within their 2σ errors (Fig. 5) to result from varying annealing characteristics. Age scatter due to variable annealing behaviour is a known feature in other mineral fission-track chronometers such as apatite (Barbarand et al. 2003).

The absence of a TFT age–elevation trend (Fig. 5c) suggests a non-uniform age–elevation relationship in the region, representing diverse and complex cooling path records (e.g. Braun 2002). We interpret this as reflecting two different cooling path records for the O and Y group samples.

Only samples RMZ 11 and RMZ 31 yield TFT ages of c. 300 Ma (Table 3). Hence, we suggest that the spatial separation between the O and Y group TFT cooling history records is broadly approximated by the interpolated 300 Ma age contour (Fig. 6).

As the present-day altitude (c. 1200 m) of O group samples GZ 87 and GZ 90 (Figs 5a & 6, and Table 3) is a Neogene feature due to elevation of the western Malawi rift flank over its eastern counterpart by about 500 m (Chorowicz 2005 and references therein), we infer a common cooling history for all O group samples.

Relating the ⁴⁰Ar/³⁹Ar biotite results for RMZ 13 (448.2±3.7 Ma) to TFT samples RMZ 13 (348±40 Ma) and RMZ 16 (330 ± 40 Ma) yields cooling rates of less than 0.6 °C Ma⁻¹. A similarly low cooling rate of less than 0.9 °C Ma⁻¹ is obtained for ⁴⁰Ar/³⁹Ar biotite sample RMZ 18 (428.4 ± 0.3 Ma) and the closest TFT sample is RMZ 14 (338±44 Ma) (Tables 1, 3, and Figs 2, 3, 6 & 7). These very low cooling rates indicate that the O group samples cooled very slowly through the TFT partial annealing zone (PAZ: 310–265±10 °C: Coyle & Wagner 1998) in the Late Carboniferous–Early Permian times (c. 320–290 Ma).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.

Diagram illustrating the t–T cooling paths of the southern basement in northern Mozambique. Black rectangles represent t–T spaces constrained by the ⁴⁰Ar/³⁹Ar hornblende (Hbl) and biotite (Bt) analyses. Rb–Sr biotite (Bt) cooling ages of Costa et al. (1992) are represented by the blue dashed outlined box. Older (O) and younger (Y) TFT age groups are depicted by green squares and orange circles, respectively. Constrained and inferred cooling paths are shown in solid and dashed brown lines, respectively. The light grey area denotes a general t–T trend envelope, inferred from error intervals of thermochronological data and cooling rates. An inferred temperature range, where Y group TFT samples resided prior to their uplift and cooling is indicated by the orange dotted outlined box. Numbers 1, 2 and 3 indicate periods of the Pan-African metamorphism, granite and pegmatite intrusions and incipient rifting in the Early–Late Permian, respectively. The question mark denotes an exemplary cooling path for the post-metamorphic thermally influenced basement.

One Y group cooling rate estimate of less than 0.3 °C Ma⁻¹ is derived from the GZ 39 ⁴⁰Ar/³⁹Ar biotite (443.8±0.4 Ma) and TFT (254±32 Ma) age results (Tables 1 & 3, and Fig. 2). Relating the Y group TFT ages to Rb–Sr biotite cooling ages (450–420 Ma) of Costa et al. (1992) (Table 3, and Figs 2 & 6) yields even lower cooling rate estimates of much less than 0.3 °C Ma⁻¹. Based on the abrupt north–south change in TFT ages (Fig. 5c, d), we argue that these extremely low rates do not represent a southwards decrease in cooling since post-late Ordovician times. We suggest, instead, that these rates result from a complex, multistage cooling history of Y group TFT samples. The absence of an age–elevation trend in the Y group TFT ages (Fig. 5a, b) may indicate a period of more rapid cooling (cf. Braun 2002), whereby the youngest TFT ages, located furthest south, cooled last.

We suggest that initially all TFT age samples cooled at similar rates of less than 1 °C Ma⁻¹ from about 450–420 Ma onwards. The Y group samples were located at a deeper crustal level (c. 1 km) than the O group samples. During, presumably, Late Carboniferous–Early Permian times (c. 300–290 Ma) the O group samples had cooled through the TFT PAZ, whereas the Y group samples still resided within and/or above the high-temperature threshold of the TFT PAZ (c. >310 °C). Then, the Y group samples cooled more rapidly to below c. 275±25 °C at more recent times (c. <280 Ma) (Fig. 7). Alternatively, all TFT sample groups shared a common crustal level position and cooling history through the TFT PAZ, similar to the O group. At younger times (c. <280 Ma) the Y group samples experienced a thermal overprint (causing partial to entire TFT age resetting) and subsequently renewed cooling.

Proterozoic–Early Palaeozoic cooling in the SW Axial Granulite Complex

The hornblende ⁴⁰Ar/³⁹Ar cooling age of GZ 90 (551±3 Ma) coincides with the youngest age constraint of the granulite facies metamorphism, dated between about 571 and 549 Ma (Kröner et al. 2001), in the southern Axial Granulite Complex (Fig. 2). We suggest that hornblende sample GZ 90 records the cooling to below about 550–500 °C at c. 550 Ma, subsequent to the peak granulite facies metamorphism. Furthermore, we suggest that ⁴⁰Ar/³⁹Ar cooling age of sample GZ 90 dates the cooling from the static, retrograde amphibolite facies overprint (Pinna et al. 1993; Kröner et al. 2001) in the SW Axial Granulite Complex.

Early Palaeozoic cooling in the southern basement

⁴⁰Ar/³⁹Ar hornblende cooling ages of samples RMZ 13 and RMZ 11 date cooling to below about 525 °C at 476–474 Ma in the eastern part of the southern basement. Both cooling ages are younger than the syn-tectonic (= D_M2) granulite facies metamorphism, dated at approximately 615 Ma by Kröner et al. (1997) and recently at about 551 Ma by Grantham et al. (2005a) using U–Pb SHRIMP single zircon dating. A time difference of around 30–80 Ma exists between the peak granulite facies metamorphism and the ⁴⁰Ar/³⁹Ar hornblende cooling ages. Both ⁴⁰Ar/³⁹Ar hornblende ages are also younger than the inferred timing (c. 500 Ma) for the syn- to late-kinematic granitic intrusions (Pinna et al. 1993). The ⁴⁰Ar/³⁹Ar hornblende cooling ages could, therefore, either point towards a period of very slow cooling or a protracted, yet not completely resolved, multiphase evolution of the basement during Pan-African times.

The formation of the Namama Thrust Belt (Fig. 2) is inferred to post-date the peak granulite facies metamorphism (Sacchi et al. 2000). The related thrust tectonics account for the distribution of some late-phase, high-grade metamorphic assemblages (Sacchi et al. 2000) and, presumably, reworked a large region of the southern basement including some granitoids emplaced at about 500 Ma (Pinna et al. 1993). We suggest that the ⁴⁰Ar/³⁹Ar hornblende cooling ages of RMZ 11 and RMZ 13 date the cooling from the Namama Thrust Belt thermo-tectonic event at 476–474 Ma with subsequent cooling rates of 6.5–10.5 °C Ma⁻¹ (Fig. 7). These cooling ages also delimit the age of the amphibolite–granulite facies Namama Thrust Belt metamorphism to ≥ c. 500 Ma. An unconstrained but widely recognized amphibolite facies overprint in the southern basement, described by Pinna et al. (1993), post-dates the peak granulite facies metamorphism. The cooling path traced by the ⁴⁰Ar/³⁹Ar hornblende ages appears to be in line with a cooling from such an overprint. We therefore consider it reasonable that cooling to below about 525 °C at 476–474 Ma was not solely limited to the eastern part, but occurred throughout large parts of the southern basement.

An ⁴⁰Ar/³⁹Ar hornblende cooling age of 455.5±6.2 Ma (RMZ 45: Fig. 2) coincides with a pegmatite Rb–Sr muscovite cooling age of 454±7 Ma (Costa et al. 1992) (Fig. 2). Recent monazite U–Pb CHIME (chemical Th–U-total Pb isochrone method) analyses on pegmatites and U–Pb SHRIMP zircon analyses on two undeformed granites from the southern basement yielded intrusion ages of between c. 500 and 450 Ma (Grantham et al. 2005a). These data indicate localized thermal influences by igneous activity in the central southern basement between about 500 and 450 Ma. The preserved younger ⁴⁰Ar/³⁹Ar hornblende age of RMZ 45 is interpreted to record cooling to below 560–510 °C at around 455 Ma (Fig. 7) from this a late- to post-metamorphic reheating.

⁴⁰Ar/³⁹Ar biotite sample RMZ 13 and GZ 39 recorded the cooling to below about 310–300 °C at 448.2 ± 3.7 and 443.8±0.4 Ma, at rates of 7–11 °C Ma⁻¹ in the eastern part of the southern basement, respectively (Figs 2 & 7). Both cooling ages coincide with a Rb–Sr biotite cooling age of 449±7 Ma (Costa et al. 1992) from a basement gneiss in the Namama Thrust Belt hinterland (Fig. 2). These results reflect widespread cooling of the southern basement to below c. 350–300 °C at around 450 Ma (Fig. 7).

Younger Rb–Sr biotite cooling ages of 434–420 Ma from mainly undeformed granitoids (Sacchi et al. 1984; Costa et al. 1992) coincide with the ⁴⁰Ar/³⁹Ar biotite cooling age of RMZ 18 (428.4±0.3 Ma). We suggest that biotite sample RMZ 18 records cooling to below 335–300 °C from the basement reheating event between c. 500 and 450 (Fig. 7). The pegmatite and granite emplacements locally delayed the final basement cooling to below about 350 °C to around 434–420 Ma (Fig. 7). Conclusively, the ⁴⁰Ar/³⁹Ar biotite and Rb–Sr biotite cooling ages (Table 2 and Fig. 2) indicate that the southern basement experienced ultimate, widespread cooling to below c. 350–300 °C after the latest Pan-African thermo-tectonic events, which occurred between >615–551 Ma and ≤450 Ma, at around 450–420 Ma, i.e. in Late Ordovician–Early Silurian times (Fig. 7).

Late Palaeozoic cooling and denudation history

Further cooling to below c. 275±25 °C is recorded by the O group TFT ages (Figs 6 & 7: c. 380–290 Ma) and occurred in Late Devonian–Early Carboniferous times. With respect to the ⁴⁰Ar/³⁹Ar analyses (c. 7–11 °C Ma⁻¹), the cooling rates dropped by one order of magnitude to values of less than 1 °C Ma⁻¹ (Fig. 7). We infer that this cooling rate decrease corresponds to reduced denudation rates within the southern basement. Such a decrease in denudation is in line with the establishment of pre-Karoo peneplains (before Late Carboniferous), subsequent to a prolonged period of erosion in central Gondwana (Wopfner & Kaaya 1991; Catuneanu et al. 2005).

Preliminary apatite fission-track results (Daszinnies et al. 2004) and remnants of Karoo age lavas (Jaritz et al. 1977; Grantham et al. 2005b) along the continental margin (Figs 1, 6 & 7) indicate that a second, younger cooling event to below c. 275 ± 25 °C, recorded by the Y group TFT ages, must have occurred prior to Late Triassic–Early Jurassic times (c. >200–180 Ma).

Permo-Triassic rift basins adjacent to the north Mozambican basement provide evidence for incipient rifting in Early–Late Permian times at about 280–260 Ma (Castaing 1991; Catuneanu et al. 2005). The onset of rifting post-dates the O group TFT ages, but pre-dates and overlaps with the Y group TFT ages (Fig. 8). We infer that the cooling recorded by the Y group TFT samples is related to regional incipient rifting activity. Basement cooling due to rift-related denudation and exhumation could result from:

• crustal extension, i.e. rift basin formation;

• erosional base level lowering due to incipient rifting;

• erosion of an uplifted rift flank adjacent to a rift basin.

In addition, basement cooling could also reflect the fading of a thermal event accompanying the rifting process (e.g. Bott 1995; Ziegler & Cloething 2004).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 8.

Diagram depicting Latitude v. TFT age plot of Y and O populations. Dashed sigmoid lines represent the inferred north–south age decrease trends along the TFT age traverses A, B and C. These sigmoid trends reflect the thermochronological imprint of a denuding, uplifted rift flank. The horizontal grey bar shows the timing of incipient Permian rifting (c. 260–280 Ma) in central Gondwana.

In the southern basement regional-scale brittle and ductile structures, except the remnant basal granulites klippen and Namama Thrust Belt ductile shear zones (Fig. 2), are absent (Pinna 1995; Norconsult Consortium 2007). Based on the present-day exposure of mid-crustal rocks, the removal of formerly existing upper-crustal structures can not be strictly discarded. However, owing to the apparent lack of regional-scale extensional structures, we regard crustal extension on the southern basement as a less likely cause for the Y group cooling pattern (Figs 6 & 7).

TFT dating (PAZ: 310–265±10 °C) is underpinned by similar basic physical principles as FT dating on apatites (PAZ: 110–60 °C ± 10 °C), but is sensitive to recording cooling processes at higher temperatures, i.e. at deeper crustal levels (Wagner & Van den Haute 1992). We infer that thermo-tectonic processes affecting the entire crust would, for both thermo-chronometers, yield FT results that differ in age (although still cooling rate dependent) but bear resemblance within their spatial FT age distribution patterns.

Models of rift-margin evolution involving erosional base-level lowering triggered by incipient rifting produce synthetic patterns of AFT ages that decrease continuously from interior towards rift basin (Van der Beek 1995). As the TFT age pattern in the southern basement displays a prominent drop in age from north to south (Figs 5 & 6), we discard base-level lowering as a suitable explanation for the observed TFT age pattern (Figs 6, 7, 8).

All post Pan-African igneous activity in northern Mozambique is younger than approximately 180 Ma (e.g. Eby et al. 1995; Grantham et al. 2005b). We hence consider a rift-related thermal overprinting in the southern basement as a less likely cause for the observed TFT age distribution. Moreover, the absence of a rift-related Permian magmatism argues for passive rifting conditions (Bott 1995; Ziegler & Cloething 2004).

All three TFT age–latitude profiles (Fig. 5d) show a rapid age decrease over a distance of about 50 km from north to south. In the absence of regional-scale faults between both TFT age groups, a sharp age change over a discrete zone is regarded less likely (Figs 2 & 5, 6, 7). We suggest that the observed age-decrease pattern between the O and Y TFT age groups can be approximated by a sigmoid trend line (Fig. 8). Such a sigmoid age-decrease pattern, including the distance over which the TFT ages decrease, strongly resembles sigmoid AFT age patterns that are:

• observed across rifted continental margins displaying prominent escarpments (Gallagher et al. 1994);

• predicted by numerical models combining lithosphere extension and surface processes (Van der Beek et al. 1995).

The youngest TFT ages (c. <250 Ma) are located furthest south and predate the regional timing of incipient rifting (c. 280–260 Ma: Fig. 8). This finding is also in good agreement with sigmoid AFT age patterns found across rifted continental margins. There, the youngest AFT ages occur on the uplifted rift flanks close to the basin edge and commonly predate the timing of rifting as they were exhumed from below the AFT PAZ (Van der Beek et al. 1995; Gallagher et al. 1998).

We conclude that the Y group TFT samples were also exhumed from below the TFT PAZ; T≥310 °C. The Y group TFT ages, therefore, directly record the cooling from the denuding uplifted rift flank, whereby cooling results from interwoven processes of tectonic uplift, denudation and local isostatic compensation (Van der Beek et al. 1995 and references therein). Consequently, the youngest TFT age samples furthest south (Figs 6, 7, 8) represent the zone of highest denudation, and highest tectonic and isostatic uplift on the rift flank, located proximate to the edge of a rift basin (Fig. 9). As passive rifting conditions are considered, rift-flank uplift occurred synchronous with incipient rifting (Braun & Beaumont 1989) in Early–Late Permian times (c. 280–260 Ma) and the Y group TFT samples recorded the subsequent cooling. We interpret the north–south decreasing TFT age pattern (Figs 5, 6, 7) in the southern basement as the thermo-chronological imprint of an east- to ENE-trending rift flank that was uplifted and denuded during Karoo times. Thereby, the O group TFT samples represent the unaffected hinterland, inland of the uplifted region. Rift-flank uplift is flexural in nature (e.g. Bott 1995) and this fact is in line with the lack of regional extensional structures on the southern basement.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 9.

Illustration (a) depicts the spatial TFT age pattern in northern Mozambique and the inferred rift flank. The black squares and dots show the sample locations of the older and younger TFT age populations, respectively. The ductile basement fabrics are derived from Pinna et al. (1991) and the orientation of the Permian tensional stress field is obtained from Castaing (1991). Illustration (b) shows the inferred rift geometry. Thick dashed lines trace the approximated boundary between the hinterland and the rift flank. The grey circle segments are graphical representations of the acute angles formed by trends of ductile fabrics and the rift structure with the tension stress field, respectively. Orientation measurements of joints obtained from the light grey area shown in the Rose and Schmidt diagram. Modes of extension after Morley et al. (2006).

A corresponding rift basin was probably located in the vicinity of the southern continental margin (Fig. 9), suggestive that the uplifted rift flank could be up to approximately 200 km wide. This is quite a large distance, but is in line with numerical modelling predictions (Van der Beek et al. 1994, 1995) that indicate that the extent of the uplifted section scales with the elevation of the pre-existing surface and the presence of a regional uplift. However, the elevation of the Permo-Carboniferous prerift surface is unknown, as is the existence of a synchronous regional uplift.

Between longitudes of approximately 37–40.5°E the regional ductile basement fabrics display easterly trends (Cadoppi et al. 1987; Pinna et al. 1993; Norconsult Consortium 2007) (Fig. 9a). They broadly parallel the easterly contour trends of the TFT age pattern and, hence, the east to ENE orientation of the uplifted rift flank (Fig. 9). Within this region the NW–SE-orientated Early Permian tensional stress regime (Castaing 1991) (Fig. 9b) encloses acute angles of approximately 40°–85° with the regional basement fabric trends, and an acute angle of about 50° with the general trend of the rift flank and its associated southern rift structure (Fig. 9b). This suggests that pre-existing basement fabrics trending at high angles of 45°–50° to the tensional stress direction were activated by oblique extensional faulting during rifting (Morley et al. 2004) (Fig. 9b). West of longitude 37°E, near the Liciro lineament, the TFT age contours approach an NE–SW orientation (Figs 6 & 9) and coincides with the NE-trending ductile basement fabrics (Norconsult 2006). Here also, steeply dipping joints parallel the ductile basement fabrics in trend (Fig. 9b). We infer that this parallelism of brittle and ductile structures indicates brittle, extensional reactivation of ductile basement fabrics. Crustal extension, therefore, could have taken place in a more orthogonal mode to the west of longitude 37°E.

The east–west-orientated extensional fault trends in the Cabora Bassa Basin (Fig. 2) are attributed to a reactivation of easterly-trending crystalline basement fabrics (Castaing 1991). We suppose that the Cabora Bassa Basin and the Permian rift basin to the south of the southern basement were linked via the Zambesi pre-transform system during incipient rifting in the Early–Late Permian (Fig. 2). Harrowfield et al. (2005) envisioned the formation of an extensive Permo-Triassic rift basin bordering Australia/Antarctica and India. If this is true, a branch could extend from the Australo-Antarctic rift basin into central Africa via northern Mozambique (Fig. 2).

At rifted margins, geothermal gradients are fairly constant as the upper crustal isotherms are not significantly disturbed (Stüwe et al. 1994; Brown & Summerfield 1997) by the sufficiently low (<100 m Ma⁻¹) exhumation rates (Leeder 1991; Van der Beek 1995). Estimating the amount of denudation is generally difficult as direct paleao-heat flows are unknown. A geothermal gradient of 25–30 °C km⁻¹, based on present-day heat flow records and vitrinite reflectance data from the Karoo basins of southern Tanzania, has been used in the reconstruction of the Late Palaeozoic–Cenozoic denudation histories on the Malawi and Rukwa rift flanks (Van der Beek et al. 1998). These values are adopted here. As the youngest TFT samples of the Y population were exhumed from below the TFT PAZ (T>310 °C), approximately 10–12 km of crust have been removed by denudation from the rift flank since its uplift in the Early–Late Permian times. Further inland, the TFT ages of the O group indicate an amount of around 9–11 km of denudation since the Early Carboniferous (c. 330 Ma, T_c=275 °C). Our denudation estimates are regarded as extreme upper values as they assume a constant geothermal gradient over very long periods. According to Jacobs & Thomas (2004), the thickened continental crust in the central part of East African–Antarctic orogen was thinned by an orogenic collapse at c. 530–490 Ma. This could suggest that subsequently average crustal thicknesses (c. 30–50 km) prevailed in the area of northern Mozambique. If high denudation of up to 12 km (c. one-third of the average thickness of continental crust) has amounted since the Early Carboniferous, the question arises to what extent it can be evidenced by other observations. One could expect that the present-day crustal thickness in the southern basement should be distinctly thinner compared to formerly adjacent regions in Africa and East Antarctica. In relation, the present-day Mohorovičić discontinuity underneath northern Mozambique should be located at a shallower depth.

An alternative interpretation to the rift-margin uplift model (preferred by one of us, G. H. Grantham) is as follows. The data clearly show that the TFT ages young southwards, implying greater degrees of exhumation towards the south. Early workers have interpreted the Lurio Belt as a significant high-strain zone and the overthrust klippen of granulite grade lithologies as derived from the Lurio Belt (e.g. Pinna et al. 1993; Sacchi et al. 2000). Grantham et al. (2003) have suggested that the Lurio Belt is correlatable with the shear zone that separates the Central Highlands Complex from the Vijayan Complex in Sri Lanka (see the tight-fit reconstruction of Gondwana after Lawyer et al. 1998 and others including Reeves used in Grantham et al. 2003) upon which similar klippen structures of Highland Complex rocks rest (Kröner 1991; Kriegsman 1995). Recent pressure–temperature (P–T) estimates from the granulite structure in the southern basement (Mugeba Klippe) show that the syn-tectonic peak metamorphic assemblages indicate conditions of approximately 10 kb and 900 °C, after which post-tectonic static cooling to c. 8 kb and c. 700 °C occurred (Roberts et al. 2005). U–Pb zircon SHRIMP data indicate that the high-grade metamorphism occurred at about 556 Ma. Pb/Pb zircon data from the same Mugeba Klippe yielded marginally older ages (Kröner et al. 1997). However, the evaporation method used by Kröner et al. (1997) lacks the high spatial resolution of the SHRIMP. The data of Kröner et al. (1997) are considered unreliable because the U–Pb zircon SHRIMP data show complex zircons with concordant ages of about 556 Ma and discordant zircons suggesting an upper intercept of around 1120 Ma.

During the period from c. 550 Ma, at about 8 kb and 700 °C, rapid exhumation of the southern basement to c. 3.5–2.8 kb and c. 280 °C at 245 Ma followed (assuming a geothermal gradient of 25–30 °C km⁻¹). This exhumation implies an uplift rate of about 15 km (4.5 kb) over approximately 300 Ma or 50 m Ma⁻¹. The southern basement contains a significant volume of largely undeformed granites, with most being c. 500 Ma in age (c. 453–530 Ma: Grantham et al. 2005a). The hornblende ⁴⁰Ar/³⁹Ar ages are broadly consistent with these conditions, with the granite minimum melt temperatures being around 650–700 °C. The large-scale genesis of the granites from c. 550 to 500 Ma is consistent with the southern basement being the footwall of the overthrust granulite klippen remnants. The temperature increase due to burial in the footwall would promote dehydration reactions at depth contributing to an extensive partial melting of the c. 1100 Ma gneisses to form the c. 450–530 Ma granites. Rapid exhumation and inversion of the southern basement would have followed the Lurian front deformation. On a larger scale, considering a cross-section at approximately 550–600 Ma ago from Cobue-Geci region to the southern Kirwanveggan (Antarctica), the central portion of granulite klippen region experienced depths at 6–8 kb about 500 Ma ago. The orogenic belt between these two areas has been exhumed. A simple cooling at a rate of 2.5 °C Ma⁻¹ could be assumed for the cooling history from 550 Ma, 900 °C and 10 kb until Karoo times, i.e. the surface at Gondwana break-up at 180–200 Ma.