Annual rings of a cross-dated teak tree core from Muna Island, Sulawesi, Indonesia in the southern equatorial tropics were analysed for radiocarbon from 1951–1979.14C levels at Muna started rising in 1956 and reached a maximum value of 667‰ in early 1965. The Muna Δ14C levels are significantly higher than those derived from the other Southern Hemisphere (SH) 14C records (including tree rings and atmospheric CO2sampling) for 1959 and 1963–1965. During the growing season of teak tree rings at this location (Nov–Apr) the Inter-tropical Convergence Zone (ITCZ) moves southward of Muna. Our results indicate that the island is strongly influenced by Northern Hemisphere (NH) air masses carried by the winter Asian monsoon, while the other more southerly SH sites remain covered by SH air masses. This monsoonal effect on atmospheric14C at Muna is evident during the periods of rapidly rising atmospheric 14C (1959 and 1963–1965), when there is an enhanced 14C contrast between northern and southern air masses.
 Above-ground nuclear weapon tests, mostly in the late 1950s and early 1960s, released a considerable amount of14C into the upper atmosphere, leading to an enormous increase in the concentration of atmospheric 14CO2 (or atmospheric 14C in short) and large 14C differences between the atmosphere and other carbon reservoirs. As a consequence, excess 14C was injected mainly into the stratosphere, leaked into the troposphere and transferred to the oceans and biosphere. This anthropogenic perturbation enables researchers to use 14C as a powerful tracer for the study of atmospheric transport, exchanges between the carbon reservoirs and the global carbon cycle [Oeschger et al., 1975; Broecker et al., 1980; Nydal and Lövseth, 1983; Randerson et al., 2002; Hua and Barbetti, 2007; Levin et al., 2010].
 Because most of the atmospheric nuclear tests were conducted in the NH, there were large 14C disequilibria in the troposphere (north versus south and high-latitudes versus low-latitudes) during the heavily bomb-influenced period 1963–1967. This significantly increased the contrast between regional air masses, which helps to improve our knowledge of the influence of atmospheric circulation on regional14C differences. Hua and Barbetti [2004, 2007] demonstrated that the spatial distribution of bomb 14C during 1963–1967 depended on atmospheric circulation and did not have a simple latitudinal gradient. Atmospheric circulation created 3 different zones in the NH separated more or less by the Ferrel cell–Hadley cell boundaries between zones 1 and 2, and the summer ITCZ between zones 2 and 3 (Figure 1). In particular, the summer Asian/African monsoon carried SH air masses, containing much less 14C than those from the NH during the bomb peak period 1963–1967, to Mandla (India, 23°N, 81°E), Doi Inthanon (Thailand, 19°N, 99°E), Saigon (Vietnam, 11°N, 107°E) and Debre Zeit (Ethiopia, 9°N, 39°E) in NH zone 3 located south of the summer ITCZ but north of the equator. This monsoonal influence resulted in significantly lower atmospheric 14C levels for those sites during boreal summers compared to other sites at similar latitudes in northwestern Africa in NH zone 2, located north of the summer ITCZ [Hua and Barbetti, 2007]. Note that herein, unless specified, summers and winters refer to these boreal seasons. However, there have been no comparable data sets of atmospheric 14C during the bomb period for sites located between the equator and the winter ITCZ [Hua et al., 1999] to see if the winter monsoon influences atmospheric 14C in the southern tropics.
 In this paper, we report bomb radiocarbon derived from teak tree rings from Muna Island, Sulawesi, Indonesia, located in the southern tropics but north of the winter ITCZ position, and discuss possible monsoonal influences on the SH 14C record.
2. Materials and Methods
 Teak (Tectona grandis) is one of very few tropical tree species producing distinct layers of annual growth [Jacoby and D'Arrigo, 1990; Buckley et al., 2007]. For this study we used a cross-dated tree-ring core (Muna-TG14B) from Muna Island off southeastern Sulawesi, Indonesia at 5°S, 122°E (Figure 1). This core, 5 mm in diameter, was collected in Feb 1996, spanning 117 years starting from AD 1879. Muna-TG14B was chosen because of its large ring widths with most of the rings being >2 mm. This made it easier to extract samples from the central portions of individual rings – well-separated from the ring boundaries – with enough material for accelerator mass spectrometry (AMS) radiocarbon analysis. Moisture is the key parameter controlling the growth of teak in Indonesia [D'Arrigo et al., 2006]. During the growing season (the rainy months from Nov to Apr [Jacoby and D'Arrigo, 1990]), Muna Island is influenced by the winter Asian monsoon [Aldrian and Susanto, 2003] with the air mass over the sampling site mainly originating from the NH.
 Twenty-nine samples of single annual tree rings for the period 1951 to 1979 were selected for AMS14C analysis. The woody material from the central portion of each ring, well-separated from the ring boundaries, was carefully sampled, sliced and milled to 0.5–1 mm, and pre-treated to alpha-cellulose [Hua et al., 2004a]. The pre-treated material was combusted to CO2using the sealed-tube techniques and then converted to graphite using the H2/Fe method [Hua et al., 2001]. AMS 14C measurements were carried out using the STAR facility at ANSTO [Fink et al., 2004], with a typical precision of 0.30–0.35%.
3. Results and Discussion
 The AMS 14C results for Muna-TG14B are reported as Δ14C values, after corrections for isotopic fractionation using measured δ13C, and radioactive decay [Stuiver and Polach, 1977]. They are reported in Data Set S1 in the auxiliary material.
 Teak growth is dependent on moisture availability not only during the current monsoon but also during the dry season of the previous year [D'Arrigo et al., 2006; Buckley et al., 2007; Ram et al., 2008]. Thus it is possible that teak trees might store and use some carbohydrate reserves from the prior year to form the current year ring. If this were the case, the effect would be most noticeable during the period of strong influence of bomb 14C in the 1960s and early 1970s when atmospheric Δ14C levels between consecutive years were very different [Hua et al., 1999]. To test for this possibility, we therefore compared Δ14C values in teak tree rings from Mandla, India [Murphy et al., 1997] with those of corresponding pine tree rings from Doi Inthanon [Hua et al., 2000] as well as with atmospheric CO2 sampling results from Debre Zeit during the growing season of NH tree rings (May–Aug) [Nydal and Lövseth, 1983]. All three sites are in NH zone 3. The Mandla teak 14C data agreed very well with the other data during the period 1960–1970 (five out of the six data pairs – Mandla vs Doi Inthanon or Debre Zeit – agreed within 1σ uncertainties and the other pair agreed within 2σ uncertainties). This indicates that, for our protocols and for 14C, any carry-over of carbohydrates from the prior year is negligible for teak tree rings. We therefore infer that the Δ14C values of the Muna teak rings accurately reflect atmospheric 14C values for each growing season.
 The Muna 14C data are illustrated in Figure 2 together with previously published 14C data from tree rings representing the four zones defined by Hua and Barbetti , which include northeastern Hungary and Obrigheim, Germany (NH zone 1), Agematsu, Japan (NH zone 2), Doi Inthanon, Thailand (NH zone 3), and Armidale and Tasmania, Australia from mid-latitudes in the SH. Tree-ring data are plotted as points in the middle of the growing period (Jun–Jul for the NH, Dec–Jan for Armidale and Tasmania, and Jan–Feb for Muna).
 Atmospheric Δ14C at Muna starts rising in 1956 due to above-ground nuclear bomb tests. For the period 1957–1959, the rate of Δ14C increase at Muna is higher than that of the SH but lower than that of the NH. In the absence of atmospheric nuclear detonations from Jan 1959 to mid-1961, Δ14C at Obrigheim, Agematsu and in northeastern Hungary (NH zones 1–2) decreases in 1960 and 1961. Meanwhile, Δ14C at Doi Inthanon (NH zone 3) increases in 1961 after a small decrease in 1960, and Δ14C at Muna and the SH increase steadily during 1960–1961, but the rate is lower than that for 1957–1959. That is because excess tropospheric 14C from NH zones 1–2 was still being transported southwards to the tropics and SH at that time.
 A large number of nuclear bomb tests in 1961–1962 lead to a dramatic increase in atmospheric Δ14C from 1962–1966 (Figure 2). The rate of Δ14C rise at Muna is similar to that of Doi Inthanon and significantly higher than that of the SH while lower than for NH zones 1–2. Bomb 14C at Muna reaches a maximum level in early 1965 then starts falling at a rate similar to that of the other records, due to the absence of major atmospheric nuclear explosions and rapid exchange between the atmosphere and other carbon reservoirs. By 1968 there are no significant differences between the tree-ring Δ14C records because atmospheric bomb 14C more or less reached a global equilibrium at that time [Telegadas, 1971; Manning et al., 1990].
 The levels of the tree-ring14C bomb peaks decrease from north to south with the highest value in northeastern Hungary of 939 ± 16‰ in the middle of 1964 (Figure 2). The peak value is lower for Agematsu at 794.1 ± 3.9‰ in the middle of 1964. The bomb values are even lower for the tropics with peaks at 694.3 ± 7.0‰ in the middle of 1965 for Doi Inthanon and at 667.0 ± 4.5‰ for Muna at the beginning of 1965. The lowest bomb peak value is observed for the southern temperate regions of 630.8 ± 4.7‰ (average value for Armidale and Tasmania) at the beginning of 1966. This indicates a fast transport of bomb 14C within 0.5–1 yr from the northern temperate regions to the tropics and a further 0.5–1 yr to the southern temperate regions.
 Although Muna is located in the SH, its Δ14C levels are significantly higher than those from the other tree-ring records in the SH for early 1959 and 1963–1965, the periods of rapidly rising atmospheric14C (Figure 2). In addition, as mentioned above, the rates of the Muna Δ14C rise during these periods are closer to those of Doi Inthanon in NH zone 3 than those for the SH. When the Muna data are compared with 14C records from atmospheric sampling at Debre Zeit in NH zone 3 and at Suva, Fianarantsoa, Pretoria and Wellington in the SH, similar 14C offsets are observed, with Muna levels being much higher than the SH records for early 1959 and 1963–1965, but significantly lower than Debre Zeit for 1964–1965 (Figure 3). It is worth noting that an atmospheric 14C record from Funatufi, Tuvalu (8.5°S, 179.2°E) [Manning et al., 1990] was not plotted in Figure 3 for comparison with our Muna data because this record is quite short (from Aug 1966 to Mar 1972), covering mostly the falling arm of bomb radiocarbon (outside the period of interest) with sparse data.
 There is a large 14C gradient between the troposphere and other carbon reservoirs during the bomb peak period (as mentioned above), resulting in large fluxes of excess 14C from the atmosphere to the oceans and to the terrestrial biosphere [Naegler and Levin, 2006]. Is it possible that these fluxes might have contributed to the observed atmospheric 14C gradient between Muna and the SH?
 The air-sea exchange flux results in atmospheric14C differences between land and sea for the period immediately following nuclear detonations. Similarly, the atmosphere-terrestrial biosphere exchange causes atmospheric14C offsets not only between land and ocean, but also between lands having different residence times of carbon in terrestrial systems during the bomb peak. Randerson et al.  estimated atmospheric Δ14C offsets arising from these fluxes for the year 1965. According to their modelling work the atmosphere-terrestrial biosphere exchange flux could result in a14C difference of 6–12‰ between Muna and the SH stations shown in Figure 3 (with Muna being lower). Yet, there was no significant 14C offset between these sites due to the air-sea exchange flux. However, a small negative north-south14C offset in total (6–12‰ with Muna being lower than the SH) due to these fluxes cannot be responsible for the large positive north-south14C differences during 1959 and 1963–1965 shown in Figures 2 and 3 (where Muna data are actually higher than other SH records).
 Thus, the most likely explanation for such large 14C differences between Muna and other SH records (including tree-ring and atmospheric data) is that Muna is located well north of the mean winter position of the ITCZ (Figure 1) and that the air mass over Muna during the growing season of teak tree rings comes mostly from the NH, being carried by the winter Asian monsoon. This monsoonal effect on Muna 14C is evident during periods of rapidly rising tropospheric 14C, as in 1959 and 1963–1965, when there is a large 14C gradient across NH and SH air masses. Outside these time intervals, there is no significant offset between Muna and the other SH records because the contrast of different air masses in terms of 14C is small.
 According to Hua and Barbetti , influences of the summer Asian/African monsoon on Debre Zeit in northeastern Africa during the bomb peak period 1963–1967 resulted in much larger 14C offsets between this site and atmospheric stations in northwestern Spain and Africa, which were not influenced by the summer monsoon, during summers compared to those during winters. Unfortunately, we cannot use the same approach for Muna and the SH stations shown in Figure 3 (which are located well south of the winter ITCZ) to further support our argument on the monsoonal effect on Muna 14C. This is because the Muna record only represents atmospheric 14C during winters (austral summers) when its tree rings grow and no 14C data are available during summers (austral winters) at Muna. However, if Muna is influenced by the winter Asian monsoon one might expect the magnitude of the monsoonal influence for Muna during the winters of 1963–1967 to be similar to that for Debre Zeit during the summers.
 In order to test the above hypothesis, 14C differences between sites for two different seasons, summers (May–Aug) and winters (Nov–Feb), were calculated for 1963–1970 and are plotted in Figure 4. It is worth noting that inter-laboratory14C offsets for the data sets used in Figure 4 are small and much less than the differences shown in this figure (see auxiliary material). The largest differences between Muna (located north of the winter ITCZ position) and atmospheric stations in the SH (located south of the winter ITCZ) are in 1963 and 1964 (98 ± 11‰ and 137 ± 53‰, respectively; Figure 4d) when there is a large north–south 14C gradient. The Muna-SH difference then becomes smaller (26 ± 10‰) in 1965 and from 1966 onwards there are no significant differences between Muna and the SH records. Similarly, the largest summer differences between northwestern Spain and Africa in NH zone 2 (Santiago de Compostela, Mas Palomas, Izaña and Dakar located north of the summer ITCZ) and Debre Zeit (influenced by the summer African/Asian monsoon) are in 1963 and 1964 (188 ± 47‰ and 125 ± 21‰, respectively;Figure 4b). This NH2-Debre Zeit difference becomes smaller in 1965 (33 ± 37‰) and negligible (10 ± 19‰) in 1966.
 The magnitude of summer NH2-Debre Zeit offsets of 33–187‰ in 1963–1965 is slightly larger than that of winter (austral summer) Muna-SH offsets of 26–137‰ during that period. A large14C gradient in summers over the NH during the bomb peak, caused by large injection of excess 14C from the stratosphere to the troposphere in northern high latitudes during springs and summers, could contribute to the above summer NH2-Debre Zeit offsets [Hua and Barbetti, 2007]. The maximum magnitude of this injection effect should be equal to the summer-winter differences between Fruholmen, Norway in northern high latitudes (NH zone 1) and stations in NH zone 2 of ∼30‰ during mid-1964 to early 1966 (Figure 4a). If this injection effect is taken into account, the winter Muna-SH offset (Figure 4d) is similar to that of summer NH zone 2-Debre Zeit (Figure 4b) not only in pattern but also in magnitude during the bomb peak period. This further supports our reasoning that the winter monsoon airflow is responsible for the data observed at Muna.
 Significant 14C differences between Debre Zeit in NH zone 3 and Muna (Figure 4c) and between Muna and the SH sites located south of the winter ITCZ (Figure 4d) during 1963–1967 suggest that Muna does not belong to these zones. To investigate this, we compared winter NH zone 3-Muna differences (Figure 4c) with winter NH zone 1-NH zone 2 offsets (Figure 4a) and winter NH zone 2-NH zone 3 differences (Figure 4b) during the bomb peak period. The comparison of the 14C differences was only carried out for winters because the injection effect for NH zones 1–2 and the monsoonal influences for NH zone 3 during winters were negligible. The magnitude of the winter differences during 1964–1967 in Figure 4c (74 ± 10‰, 48 ± 7‰, 43 ± 20‰ and 10 ± 7‰, respectively) is similar to that in Figure 4a (92 ± 23‰, 41 ± 9‰, 22 ± 12‰ and 19 ± 13‰, respectively) and that in Figure 4b (100 ± 23‰, 39 ± 9‰, 0 ± 12‰ and 12 ± 13‰, respectively). This indicates that Muna and sites in the SH located north of the mean winter position of the ITCZ can be attributed to another atmospheric zone in the SH in regards to bomb radiocarbon distribution during the bomb peak period.
 Our results show a rapid rise in bomb 14C at Muna in early 1959 and 1963–1965, at a rate much higher than that of other, previously published, SH records. These results, together with the position of the winter ITCZ south of Muna during the teak growing season, indicate that Muna is influenced by NH air masses carried by the winter Asian monsoon containing much more 14C than those from the SH during the periods of rapidly increasing atmospheric 14C. In addition, during the bomb peak period, winter 14C differences between Muna and SH sites (located south of the winter ITCZ) are similar to summer 14C offsets between NH zone 2 (located north of the summer ITCZ) and Debre Zeit (influenced by the summer African/Asian monsoon) in both pattern and magnitude. This gives further support to the concept of a monsoonal influence on atmospheric 14C at Muna.
 The magnitudes between winter NH zone 3-Muna differences in14C, winter NH zone 1-NH zone 2 offsets, and winter NH zone 2-NH zone 3 differences are similar from 1964–1967. This together with the remarkable differences between Muna and SH sites (located south the winter ITCZ) during that period suggest that Muna and the SH region located north of the winter ITCZ should form a previously unidentified atmospheric zone in the SH. This implies a revision of zonal distribution of bomb radiocarbon during the bomb peak defined byHua and Barbetti  to include two SH zones that are separated by the mean winter position of the ITCZ.
 We thank Paul Krusic for collecting tree-ring samples from Muna Island and Stuart Hankin for the preparation ofFigure 1. We also thank two anonymous reviewers for constructive comments, which improved the manuscript. AMS 14C measurements were supported by ANSTO's Isotopes in Climate Change and Atmospheric Systems (ICCAS) Project.
 The Editor thanks two anonymous reviewers for their assistance in evaluating this paper.