A 250-year periodicity in Southern Hemisphere westerly winds over the last 2600 years

. Southern Hemisphere westerly airﬂow has a signiﬁcant inﬂuence on the ocean–atmosphere system of the mid- to high latitudes with potentially global climate implications. Unfortunately, historic observations only extend back to the late 19th century, limiting our understanding of multi-decadal to centennial change. Here we present a highly resolved (30-year) record of past westerly wind strength from a Falkland Islands peat sequence spanning the last 2600 years. Situated within the core latitude of Southern Hemisphere westerly airﬂow (the so-called furious ﬁfties ), we identify highly variable changes in exotic pollen and charcoal derived from South America which can be used to inform on past westerly air strength. We ﬁnd a period of high charcoal content between 2000 and 1000 cal. years BP, associated with increased burning in Patagonia, most probably as a result of higher temperatures and stronger westerly airﬂow. Spectral analysis of the charcoal record identiﬁes a pervasive ca. 250-year periodicity that is coherent with radiocarbon production rates, suggesting that solar variability has a modulating inﬂuence on Southern Hemisphere westerly airﬂow. Our results have important implications for understanding global climate change through the late Holocene.


Introduction
A major limitation for quantifying the magnitude and impact of change across the Southern Ocean is the relatively short duration or low resolution of ocean-atmosphere records. This is particularly significant with regards to the Southern Hemisphere westerly storm belt, which since the mid-1970s, has undergone a significant intensification and southward shift (Gillett et al., 2008;Messié and Chavez, 2011). One measure of this change in atmospheric circulation is the Southern Annular Mode (SAM), described as the pressure difference between Antarctica (65 • S) and the latitude band at around 40 • S (Karpechko et al., 2009;Marshall, 2003). Since the mid-1970s, SAM appears to have undergone a positive shift in the troposphere, which has been associated with hemispheric-wide changes in the atmosphere-ocean-ice domains, including precipitation patterns and significant surface and subsurface ocean warming (Cook et al., 2010;Delworth and Zeng, 2014;Domack et al., 2005;Gille, 2008Gille, , 2014Thompson et al., 2011). This trend is projected to continue during the 21st century as a result of both ongoing greenhouse gas emissions and a persistence of the Antarctic ozone hole (Liu and Curry, 2010;Thompson et al., 2011;Yin, 2005), potentially resulting in reduced Southern Ocean uptake of anthropogenic CO 2 (Ito et al., 2010;Le Quére et al., 2009;Lenton et al., 2013;Marshall, 2003;Marshall and Speer, 2012).
While no observational records for SAM extend beyond the late 19th century (Fogt et al., 2009;Marshall, 2003;Visbeck, 2009), proxy records of past westerly airflow have been generated on annual-to-centennial timescales through the Holocene (Abram et al., 2014;Björck et al., 2012;Lamy et al., 2010;Lisé-Pronovost et al., 2015;McGlone et al., 2010;Strother et al., 2015;Villalba et al., 2012). Crucially, the association between proxies and changes in westerly wind strength and/or latitude is often implied but few provide a direct measure of past airflow or directly test their interpretation through time. One possibility is the identification of exotic airborne particles preserved in sedimentary sequences. Ideally, the peat or lake record should be close enough to the source to have a relatively high input of material (e.g. pollen, charcoal) but not so close that the influx is constant over time. Whilst numerous studies have been undertaken in the Arctic (Fredskild, 1984;Jessen et al., 2011) and the high latitudes of the Indian and Pacific oceans (Mc-Glone et al., 2000;Scott and van Zinderen Barker, 1985), few have been reported from the South Atlantic. Recent work on a lake core taken from Annekov Island, South Georgia (Strother et al., 2015) demonstrates the considerable potential of this approach but the relatively large distance from the nearest source in South America ( Fig. 1) (approximately 2100 km) limits the delivery of pollen, with no charcoal reported.
Here we report a new high-resolution record of westerly airflow over the past 2600 years from the Falkland Islands. The Falkland Islands (52 • S) lie within the main latitudinal belt of Southern Hemisphere westerly airflow, 500 to 730 km east of Argentina and 1410 km west of Annekov Island. The close proximity to South America means that these islands receive a relatively high input of particles from the continental mainland (Barrow, 1978;Rose et al., 2012), making them an ideal location to investigate past changes in westerly airflow.

Methods
The Falkland Islands are a low-lying archipelago in the South Atlantic Ocean, situated in the furious fifties wind belt on the southeast South American continental shelf at 51-52 • S, 58-61 • W (Fig. 1). The Falkland Islands experience a cool temperate but relatively dry oceanic climate, dominated by westerly winds (Otley et al., 2008). Across the year, the temperature ranges from 2.2 • C (July) to 9 • C (February), with the islands receiving a relatively low but variable precipitation (typically ranging between 500 and 800 mm year −1 ) lying in the lee of the Andes. Modern climate records show that the prevailing wind direction across the Falkland Islands is predominantly from the west with strong winds throughout the year, with no significant seasonal variation (Upton and Shaw, 2002).
Climate amelioration following the Last Glacial Maximum led to the establishment of blanket peat across large parts of the islands from 16 500 cal. years BP (Wilson et al., 2002). To investigate past westerly airflow in the late Holocene, an exposed ericaceous-grass peatland was cored on Canopus Hill, above Port Stanley Airport (51.691 • S, 57.785 • W, approximately 30 m above sea level) (Fig. 1). The one-metre sequence reported here comprises a uniform darkbrown peat from which the uppermost 90 cm was contiguously sampled for pollen, charcoal and comprehensive dating.
Pollen samples were prepared using standard palynological techniques (Faegri and Iverson, 1975). Volumetric samples were taken every 1 cm along the core and Lycopodium spores were added as a "spike". The samples were deflocculated with hot 10 % NaOH and then sieved through a 106 µm mesh. The samples then underwent acetolysis, to remove extraneous organic matter before the samples were mounted in silicon oil. Pollen types/palynomorphs were counted at 400× magnification until a minimum of 300 target grains were identified. The pollen counts were expressed as percentages, with only terrestrial land pollen (TLP) contributing to the final pollen sum. Pollen/palynomorphs were identified using standard pollen keys (Barrow, 1978;Macphail and Cantrill, 2006) and the pollen type slide collection at Exeter University. Past fire activity was assessed using microcharcoal counts of fragments (< 106 µm) identified on the pollen slides (Whitlock and Larsen, 2001). Counts were undertaken at each level until a fixed total of 50 lycopodium spores were counted and the total expressed as a concentration (fragments per cm 3 ). More than 99 % of charcoal fragments were less than 50 µm in size, with negligible amounts identified in the 50-106 µm and > 106 µm fractions.
Terrestrial plant macrofossils (fruits and leaves) were extracted from the peat sequence and given an acid-base-acid (ABA) pretreatment and then combusted and graphitised in the University of Waikato AMS laboratory, with 14 C / 12 C measurement by the University of California at Irvine (UCI) on a NEC compact (1.5SDH) AMS system. The pretreated samples were converted to CO 2 by combustion in sealed prebaked quartz tubes, containing Cu and Ag wire. The CO 2 was then converted to graphite using H 2 and an Fe catalyst, and loaded into aluminium target holders for measurement at UCI. This was supplemented by 137 Cs measurements down the profile to detect the onset of nuclear tests. 137 Cs analysis was undertaken following standard techniques with measurements made using an ORTEC high-resolution, lowbackground coaxial germanium detector. Detectable measurements were obtained down to 8.5 and 9.5 cm and the lowest depth assigned an age of CE 1963, the time of early radionuclide fallout at these latitudes (Hancock et al., 2011).
The radiocarbon and 137 Cs ages were used to develop an age model using a P_sequence deposition model in OxCal   Bronk Ramsey, 2008) with General Outlier analysis detection (probability = 0.05) (Bronk Ramsey, 2011). The 14 C ages were calibrated against the Southern Hemisphere calibration (SHCal13) data set. Using Bayes' theorem, the algorithms employed sample possible solutions with a probability that is the product of the prior and likelihood probabilities. Taking into account the deposition model and the actual age measurements, the posterior probability densities quantify the most likely age distributions; the outlier option was used to detect ages that fall outside the calibration model for each group, and if necessary, down-weight their contribution to the final age estimates. Modelled ages are reported here as thousands of calendar years BP or cal. BP (Table 1 and Fig. 2). The pollen sequence reported here spans the last 2600 years with an average 30-year resolution (Fig. 3).
To investigate the periodicities preserved in the palaeoenvironmental proxies utilised herein, we undertook Multi-Taper Method (MTM) analysis using a narrowband signal, red noise significance and robust noise background estima-tion (with a resolution of 2 and 3 tapers) (Thomson, 1982). We also applied single spectrum analysis (SSA), which applies an empirical orthogonal function (EOF) analysis to the autovariance matrix on the chronologies. Here we undertook a Monte Carlo significance test (95 % significance), using a window of 9, a Burg covariance, and eight components. Both analyses used the software kSpectra version 3.4.3 (3.4.5). Wavelet analysis and coherence was undertaken on the 30-year averaged data using the wt() and wtc() functions respectively in the R package "Biwavelet" (Gouhier, 2013). The Morlet continuous wavelet transform was applied, and the data were padded with zeros at each end to reduce wraparound effects (Torrence and Webster, 1999). To test the robustness of the obtained periodicities, the Lomb-Scargle algorithm was employed, a spectral decomposition method that computes the spectral properties of time series with irregular sampling intervals (Ruf, 1999) -in this instance, the "raw" charcoal values. This method minimises bias and induced periodicities that may arise from interpolating missing or unevenly spaced data. The technique was undertaken using the lsp() function within the "lomb" R package. Periodicities were extracted from data sets using Analyseries (Paillard et al., 1996).
A measure of solar variability was derived by calculating the 14 C production rate using the IntCal13 atmospheric radiocarbon data set  and an oceanatmosphere box diffusion model (Oeschger et al., 1975); the same as that reported in previous studies (Bond et al., 2001;Turney et al., 2005). The model consists of one box for the atmosphere, one for the ocean mixed layer, 37 boxes for the thermocline, five boxes for the deep ocean and two for the biosphere (short and long residence time) (Stuiver and Braziunas, 1993a). The climate-influenced mixing parameters (air-gas sea exchange, eddy diffusivity, and biospheric    uptake and release) were held constant through the run using the same setup as Marine04 (Table 2) (Hughen et al., 2004). The model was parametrised to produce a pre-industrial ma-rine mixed layer 14 C of −46.5 ‰ and a deep ocean value of −190 ‰ at CE 1830 for the 2013 marine calibration data set Marine13 .

Results and discussion
Only a limited number of Holocene pollen records have been reported from the Falkland Islands (Barrow, 1978). The pollen record in the uppermost 90 cm at Canopus Hill is dominated by Poaceae and Empetrum, consistent with previous work and today's vegetation (Barrow, 1978;Broughton and McAdam, 2003;Clark et al., 1998). The most significant change in the pollen taxa is a pronounced shift to increased representation of Asteroideae (accompanied by a relative decline in Poaceae) centred on 47 cm (equivalent to 1100 cal. BP) (Fig. 2). Although undifferentiated in the counts, the Asteroideae are most likely Chilliotrichum diffusum, common on the island across a range of habitats including Empetrum heath (Broughton and McAdam, 2003). The shift in the pollen diagram therefore most likely reflects the replacement of upland grasslands by Empetrum heath. Highly variable charcoal counts were obtained through the sequence (< 106 µm) (Fig. 2), with negligible macrocharcoal fragments (> 106 µm) identified, suggesting there was little or no fire on the site.
The exotic pollen taxa were expressed as concentration values to explore their changing input onto the site over the last 2600 years (Fig. 2). Although these data could be reexpressed as a pollen influx, the interpretation of flux data in non-annually laminated sequences can be strongly influenced by the choice of age model and the density of dated points down the core (Davis, 1969;Hicks and Hyvärinen, 1999). Consideration of the radiocarbon and 137 Cs ages (Table 1) suggests that the depth-age relationship can be de-  (Bond et al., 2001;Turney et al., 2005) vs. Marine04 (Hughen et al., 2004).

Parameter
Marine98 Marine04 Air  scribed by a linear relationship (r 2 = 0.98) below a depth of 18 cm (Fig. 3). This means that the pollen (and charcoal) concentration data below this depth are equivalent to influx. In the uppermost section of the core (above 18 cm) a faster rate of sediment accumulation (or less compaction) means that the deposition time is reduced. Importantly, the sequence preserves a record of exotic pollen delivery into the site, with Nothofagus dominating the input but with trace amounts of Podocarp, Ephedra fragilis, and Anacardium-type pollen (< 0.5 % total land pollen), all originating from South America. Whilst the low levels of most exotic pollen precludes meaningful interpretation, all samples contain Nothofagus (< 5 % total land pollen), a taxa not known to have grown on the Falkland Islands since the Middle Miocene/Early Pliocene (Macphail and Cantrill, 2006) but which has been detected in Lateglacial (Clark et al., 1998) and Holocene (Barrow, 1978) sequences. Producing relatively small pollen grains (20-40 µm in diameter) (Wang et al., 2000), the nearest source of contemporary Nothofagus is South America which extends from 33 • S in central Chile to 56 • S on Tierra del Fuego (Veblen et al., 1996). The youngest arboreal macrofossils of the other exotic taxa are dated to late Tertiary deposits on West Point Island, West Falkland (Birnie and Roberts, 1986).
Whilst exotic pollen values are relatively low, peaks in Nothofagus coincide with increased amounts of charcoal in the Canopus Hill sequence. Importantly, negligible amounts of macro-charcoal (> 106 µm) were identified, suggesting that the charcoal has been blown to the site from Patagonia. The aerial delivery of the charcoal to the Falkland Islands is supported by the close correspondence with charcoal in Laguna Guanaco in Southwest Patagonia (51 • S) (Moreno et al., 2009). Importantly, Nothofagus dominates lowland Patagonian vegetation and, in areas away from human activity, was established by 5000 cal. years BP (Iglesias et al., 2014;Kilian and Lamy, 2012), with a stepped expansion in Nothofagus at Laguna Guanaco centred on 570 cal. BP (Moreno et al., 2009) and evidence for temporary forest fragmentation during periods of stronger westerly airflow (Moreno et al., 2014). In marked contrast to Patagonia, the Falklands Nothofagus pollen record is highly variable and of sufficient concentration to recognise similar changes to those in the charcoal record, with periods of high fire frequency associated with high input of exotic pollen.
Although charcoal fragments less than 106 µm might reflect fire in the local environment, charcoal of this size can be transported long distances (Clark, 1988). The vast majority of the charcoal fragments are less than 50 µm, comparable in size to exotic Nothofagus (20-40 µm) and Podocarpus (40-50 µm in diameter) pollen (Wang et al., 2000;Wilson and Owens, 1999). The close correspondence between the Nothofagus pollen record and charcoal fragments in the Canopus Hill sequence on the Falkland Islands strongly suggests similar sources, indicating that the higher charcoal counts provide a more robust measure of the westerly airflow. A sustained period of charcoal delivery to the Falk- land Islands is observed between 2000 and 1000 cal. BP, with prominent peaks in Nothofagus and charcoal recognised at approximately 2400, 2100, 1800-1300, 1000, 550, and 250 cal. BP (Fig. 2) which we interpret here as stronger westerly wind flow. Our results suggest that reports of pre-European human activity on the Falkland Islands as inferred by the presence of charcoal in peat sequences (Buckland and Edwards, 1998) may be premature.
In contrast to work at Annenkov Island which reported enhanced westerly airflow is associated with wetter conditions (Strother et al., 2015), we observe the reverse on the Falkland Islands. Modern comparisons between the SAM (as a measure of westerly airflow) (Marshall, 2003) and air temperature suggest a positive correlation (Abram et al., 2014). Comparing historic observations of SAM with ERA79 Interim reanalysis (Dee et al., 2011), we observe a highly significant relationship with more positive phases of SAM associated with warmer 2-10 m height air temperatures and wind speeds across much of South America, the Antarctic Peninsula, and the Falkland Islands (Fig. 4), supporting our interpretation. The contrasting moisture interpretation to that in South Georgia may be a result of the rain shadow effect of the Andes on the Falklands. It should be noted, however, that the reanalysis product used here is only for the period commencing CE 1979 (the satellite era) and that different atmospheric dynamics may have been involved in the delivery of exotic pollen and charcoal to the Falkland Islands on centennial timescales. The MTM analysis identifies two different periodicities in the charcoal record (< 106 µm) from Canopus Hill significant above 95 % -242 and 95 years, with the former exhibiting a broad multi-decadal peak (Fig. 5a). To test whether the MTM spectral peak is robust, we undertook SSA on the sequence chronologies. A Monte Carlo significance test identified a significant periodicity (above 95 %) at 231 years (Fig. 5b). Furthermore, the Lomb-Scargle algorithm identified a 268year peak (Fig. 5c), indicating that this periodicity is pervasive through the record regardless of the sampling method, and therefore robust.
The existence of a 200-250-year periodicity has been identified in numerous Holocene records globally (Galloway et al., 2013;Poore et al., 2004), including Southern Ocean productivity as recorded in Palmer Deep (Domack et al., 2001;Leventer et al., 1996) and dust deposition over Antarctica (Delmonte et al., 2005). Furthermore, whilst no spectral analysis was undertaken, a series of recurring 200-year long dry/warm periods have recently been reported from Patagonia over the last three millennia and linked to positive SAM- like conditions (Moreno et al., 2014). The origin of the ∼ 250 years periodicity may be linked to postulated centennialscale changes in climate modes of variability including the El Niño Southern Oscillation (ENSO) (Ault et al., 2013) or Southern Ocean convection (Martin et al., 2013). Importantly, a 200-250-year periodicity has also been observed in records of atmospheric 14 C and 10 Be (Adolphi et al., 2014;Steinhilber et al., 2012;Stuiver and Braziunas, 1993b;Tur-ney et al., 2005), suggesting that the so-called de Vries solar cycle may play a role (Leventer et al., 1996). The detection of solar forcing in palaeo records is highly sensitive to the chronological framework being investigated (Gray et al., 2010). To explore the possible role of solar variability on Southern Hemisphere westerly airflow we first analysed the modelled production rate of 14 C derived from 5-year resolved tree-ring data , a cosmogenic radionuclide that is produced in the upper atmosphere (with 14 C increasing with reduced solar activity) (Bond et al., 2001;Turney et al., 2005). We resampled the 14 C data set at 30-year resolution to mimic the resolution of the Canopus Hill sequence and compared these data to the Total Solar Irradiance (TSI) generated from the polar ice core 10 Be which is reported at a 20-30-year resolution (Steinhilber et al., 2009) (Fig. 6). Regardless of the data set used, the same pattern is observed with large amplitude changes in solar irradiance between 2600 and 2300 years ago and from 1300 cal. years BP to present day, but with sustained high irradiance between 2300 and 1300 cal. years BP (Fig. 6a, c and e). We find that the 5-year resolved IntCal13 data set produces a periodicity comparable to the Falkland Islands record (225 years at 99 % confidence; Fig. 6a and b). Importantly, when we look at the downscaled records of solar irradiance, the statistical significance decreases in the lower-resolved 14 C data set (230 years at 90 %; Fig. 6c and d) or shifts to a lower frequency in the 10 Be record (202 years at 99 %; Fig. 6e and f).
Our results imply that the central Southern Hemisphere westerlies were particularly strong between 2000 and 1000 cal. BP and/or lay close to the latitude of the Falkland Islands -at least within the South American sector and possibly hemispheric-wide (Turney et al., 2016) (Fig. 7). Records of comparable latitude and age from South America are Laguna Guanaco (51 • S) (Moreno et al., 2014) and Palm2 (53 • S) (Lamy et al., 2010). The Laguna Guanaco record captures a remarkably similar fire history as preserved in the Canopus Hill with a pronounced peak in charcoal over the same period (Fig. 7d). In Palm2, accumulation rates of biogenic carbonate provide a proxy for salinity changes in surface fjord waters off the west coast of Chile with lower salinities associated with strong winds and relatively high precipitation, limiting the influence of the open ocean water and reducing biogenic carbonate production. While the data set from Palm2 does not have the resolution of the other records, a similar trend with pervasive lower salinities (stronger westerly winds) is recorded between 2000 and 1000 cal. years BP (Fig. 7e). Whilst the change in the trend may be interpreted as reflecting either a change in the latitude and/or strength of the winds, the parallel peaks and troughs in Nothofagus and charcoal from Canopus Hill (in contrast to constant Nothofagus levels at Laguna Guanaco - Moreno et al., 2009) imply that the core latitude of the westerly winds has not changed and instead was particularly strong between 2000 and 1000 cal. years BP, resulting in increased fire frequency in Patagonia . This is supported by a study on Patagonian Fitzroya cupressoides from 40-42 • S (Roig et al., 2001). Whilst a living series spanning 1229 years did not identify a 200-250-year periodicity, a 245-year cycle was identified in a floating 50 000-year-old tree ring series of comparable length, consistent with our record suggesting a suppression of this periodicity across a large latitudinal range over the last 1000 years. Importantly, the ∼ 250-year periodicity identified in the charcoal record varies in amplitude over the last 2600 years (Fig. 7a-c). A Gaussian filtered curve and wavelet plot shows the ∼ 250 year periodicity is expressed between 2600 and 1000 cal. BP, and spans the prominent (sustained) peak in charcoal, with an implied reduction in the expression of the ∼ 250-year periodicity over the last millennium.
The role that changing solar output may have on westerly airflow is not immediately apparent. The strongest inferred winds fall within a millennial-duration period of high solar irradiance (Fig. 6). In spite of the relatively muted amplitude of the 225-year periodicity in the 14 C record, wavelet coherence with the charcoal data sampled at 30-year resolution shows coherency centred on 1500 cal. years BP (Fig. 7g), with the proxy of solar irradiance leading westerly wind strength (arrows up). Furthermore, we observe peaks in solar irradiance leading charcoal by of the order of 20-40 years (Fig. 7f), suggesting that Southern Hemisphere westerly winds may be particularly sensitive to the de Vries cycle during periods of high solar irradiance and less sensitive with reduced solar output. How solar periodicity may influence the strength of Southern Hemisphere westerly airflow is not precisely known. One possibility is that the ∼ 250-year periodicity may change salinity in the North Atlantic (Stuiver and Braziunas, 1993b), driving changes in the Meridional Overturning Circulation that are transmitted globally. However, the existence of the same periodicity in the delivery of dust onto the East Antarctic Ice Sheet (Delmonte et al., 2005) does imply a direct atmospheric link, either through changing sea ice extent or sea surface temperatures, or via the westerlies themselves (Shindell et al., 1999). Recent work has highlighted the role of high solar irradiance in increasing tropospherestratosphere coupling, extending the seasonal length during which stronger Southern Hemisphere westerly winds are experienced at the surface (Kuroda and Yamazaki, 2010), similar to that observed in the Northern Hemisphere (Ineson et al., 2011). Alternatively, recent modelling work suggests that insolation changes can lead to increased "baroclinicity" (Fogwill et al., 2015) or a "Split Jet" (Chiang et al., 2014), strengthening westerly winds. Further work is required to understand the driving mechanism(s) behind the ∼ 250-year periodicity on global climate.

Conclusions
Southern Hemisphere westerly airflow is believed to play a significant role in precipitation, sea ice extent, sea surface temperatures and the carbon cycle across the mid-to high latitudes. Unfortunately, the observational record only extends back to the late 19th century, limiting our understanding of what drives past changes in westerly winds. Although proxies of westerly airflow can provide long-term perspectives on past change, few provide a direct (passive) measure of westerly winds. Exotic pollen and charcoal fragments sourced upwind of sedimentary sequences can potentially provide a valuable insight into past variability. Here we report a new, comprehensively-dated high-resolution pollen record from a peat sequence on the Falkland Islands which lies under the present core of Southern Hemisphere westerly airflow and spanning the last 2600 years. We observe peaks in taxa from South America (particularly Nothofagus) and charcoal fragments (< 106 µm) that appear to be linked to warm and windy conditions. Spectral analysis identifies a robust ∼ 250-year periodicity, with evidence of stronger westerly airflow between 2000 and 1000 cal. years BP. In comparison with other Southern Hemisphere records, the 250-year periodicity suggests that solar forcing plays a role in modulating the strength of the Southern Hemisphere westerlies, something hitherto not recognised. This will form the focus of future research.