High-resolution palaeoclimatology of the last millennium: a review of current status and future prospects P.D. Jones,1* K.R. Briffa,1 T.J. Osborn,1 J.M. Lough,2 T.D. van Ommen,3 B.M. Vinther,4 J. Luterbacher,5 E.R. Wahl,6 F.W. Zwiers,7 M.E. Mann,8 G.A. Schmidt,9 C.M. Ammann,10 B.M. Buckley,11 K.M. Cobb,12 J. Esper,13 H. Goosse,14 N. Graham,15 E. Jansen,16 T. Kiefer,17 C. Kull,18 M. K��ttel,5 E. Mosley-Thompson,19 J.T. Overpeck,20 N. Riedwyl,5 M. Schulz,21 A.W. Tudhope,22 R. Villalba,23 H. Wanner,5 E. Wolff24 and E. Xoplaki5 ( 1Climatic Research Unit, School of Environmental Sciences, University of East Anglia, Norwich NR4 7TJ, UK 2Australian Institute of Marine Science, Townsville MC, QLD 4810, Australia 3Australian Antarctic Division & ACE CRC, Private Bag 80, Hobart Tasmania 7001, Australia 4Centre for Ice and Climate, Niels Bohr Institute, University of Copenhagen, Juliane Maries Vej 30, DK-2100 Copenhagen ��, Denmark 5Oeschger Centre for Climate Change Research (OCCR) and NCCR Climate and Institute of Geography, Climatology and Meteorology, University of Bern, Hallerstrasse 12, CH-3012 Bern, Switzerland 6Division of Environmental Studies and Geology, Alfred University, NOAA-Paleoclimatology, Boulder CO 80305, USA 7Climate Research Division, Environment Canada, 4905 Dufferin Street, Toronto Ont. M3H 5T4, Canada 8Earth System Science Center, Department of Meteorology, Pennsylvania State University, 523 Walker Building, University Park PA 16802, USA 9NASA Goddard Institute for Space Studies, 2880 Broadway, New York NY 10025, USA 10Climate & Global Dynamics Division, NCAR, Boulder CO 80307-3000, USA 11Tree-Ring Laboratory, Lamont- Doherty Earth Observatory, Palisades, New York NY 10964, USA 12School of Earth and Atmospheric Sciences, Georgia Institute of Technology, 311 Ferst Drive, Atlanta GA 30332-0340, USA 13Swiss Federal Research Institute WSL, Z��rcherstrasse 111,CH-8903 Birmensdorf, Switzerland 14Institut dAstronomie et de G��ophysique G. Lema��tre, Universit�� Catholique de Louvain, Chemin du cyclotron 2, 1348 Louvain-la-Neuve, Belgium 15Hydrologic Research Center, 12780 High Bluff Drive, ��, La Jolla CA 92130-3017, USA 16Department of Geology, University of Bergen, Bjerknes Centre for Climate Research, Allegaten 55, NO-5007 Bergen, Norway 17PAGES International Project Office, Sulgeneckstrasse 38, 3007 Bern, Switzerland 18Advisory Body on Climate Change (OcCC), Schwarztorstrasse 9, CH-3007 Bern, Switzerland 19Department of Geography and Byrd Polar Research Center, Ohio State University, 108 Scott Hall, 1090 Carmack Road, Columbus OH 43210, USA 20Institute for the Study of Planet Earth, University of Arizona, 715 N. Park Avenue, 2nd Floor, Tucson AZ 85721, USA 21MARUM ��� Center for Marine Environmental Sciences and Faculty of The Holocene 19,1 (2009) pp. 3���49 �� 2009 SAGE Publications 10.1177/0959683608098952

4 The Holocene 19,1 (2009) Geosciences, Universit��t Bremen, Postfach 330 440, D-28334 Bremen, Germany 22School of Geosciences, University of Edinburgh, West Mains Road, Edinburgh EH9 3JW, UK 23Argentinean Institute for Snow, Ice and Environmental Sciences, IANIGLA-CRICYT, Casilla de Correo 330, Mendoza 5500, Argentina 24Physical Sciences Division, British Antarctic Survey, High Cross, Madingley Road, Cambridge CB3 OET, UK) Received 20 March 2008 revised manuscript accepted 29 August 2008 *Author for correspondence (e-mail: p.jones@uea.ac.uk) Abstract: This review of late-Holocene palaeoclimatology represents the results from a PAGES/CLIVAR Intersection Panel meeting that took place in June 2006. The review is in three parts: the principal high-resolu- tion proxy disciplines (trees, corals, ice cores and documentary evidence), emphasizing current issues in their use for climate reconstruction the various approaches that have been adopted to combine multiple climate proxy records to provide estimates of past annual-to-decadal timescale Northern Hemisphere surface tempera- tures and other climate variables, such as large-scale circulation indices and the forcing histories used in cli- mate model simulations of the past millennium. We discuss the need to develop a framework through which current and new approaches to interpreting these proxy data may be rigorously assessed using pseudo-proxies derived from climate model runs, where the ���answer��� is known. The article concludes with a list of recommen- dations. First, more raw proxy data are required from the diverse disciplines and from more locations, as well as replication, for all proxy sources, of the basic raw measurements to improve absolute dating, and to better distinguish the proxy climate signal from noise. Second, more effort is required to improve the understanding of what individual proxies respond to, supported by more site measurements and process studies. These activ- ities should also be mindful of the correlation structure of instrumental data, indicating which adjacent proxy records ought to be in agreement and which not. Third, large-scale climate reconstructions should be attempted using a wide variety of techniques, emphasizing those for which quantified errors can be estimated at specified timescales. Fourth, a greater use of climate model simulations is needed to guide the choice of reconstruction techniques (the pseudo-proxy concept) and possibly help determine where, given limited resources, future sam- pling should be concentrated. Key words: Palaeoclimatology, high-resolution, last millennium, tree rings, dendroclimatology, chronology, uncertainty, corals, ice-cores, speleothems, documentary evidence, instrumental records, varves, borehole temperature, marine sediments, composite plus scaling, CPS, climate field reconstruc- tion, CFR, pseudo-proxy approach, time series, climate forcing. Introduction and rationale In its Fourth Assessment Report (AR4), Working Group 1 of the Intergovernmental Panel on Climate Change (IPCC, 2007) con- cluded, with respect to the palaeoclimate record of the last two mil- lennia (Jansen et al., 2007), that: ���Average Northern Hemisphere temperatures during the second half of the 20th century were very likely ( 90% certainty according to IPCC���s definition) higher than during any other 50-year period in the last 500 years and likely ( 66% certainty) the highest in at least the past 1300 years���. A similar conclusion was also reached by the US National Academy of Sciences (National Research Council (NRC), 2006). Study of palaeoclimate of the last 1���2 millennia (late Holocene) has under- gone dramatic developments in the last 15 years. Up to the early 1990s, there was little cross-disciplinary work and few studies attempted to bring together reconstructions from diverse proxies except at the subcontinental scale. Although there had been earlier qualitative attempts to compare different reconstructions (eg, Williams and Wigley, 1983), the first quantitative extension of the hemispheric-scale instrumental record was produced by Bradley and Jones (1993). This decadal-average curve showed that since about 1930 onwards, summer (June to August) Northern Hemisphere (NH) temperatures were warmer than they had been for any time since 1400. This work was reported in the Second Assessment Report (SAR, IPCC, 1995), replacing the schematic, conceptual temperature history that had been presented in the First Assessment Report (IPCC, 1990: figure 7.1c ��� see Appendix A for a discussion of the validity and likely source of this figure). Although the early results (Bradley and Jones, 1993) were received with considerable scientific interest, they were not allotted much prominence by the SAR, and did not receive much public attention. The availability of several independent or partially inde- pendent annually resolved NH average temperature reconstructions (Jones et al., 1998 Mann et al., 1998, 1999 ��� hereafter MBH98, MBH99 Briffa, 2000 Crowley and Lowery, 2000), along with the explicit representation of quantitative uncertainty estimates led to the reconstructions having greater prominence in the Third Assessment Report (TAR, IPCC, 2001). The MBH98 reconstruc- tion was also prominently featured in the associated Summary for Policymakers (SPM). The lack of awareness of the 1993 paper and the SAR is evident in many studies which often only contrast the IPCC position in 1990 with that in 2001 (see also Appendix A), though this may not be for purely scientific reasons. Subsequent analyses since the TAR have not substantiallychanged the interpre- tation of recent palaeoclimatic data and, if anything, the earlier con- clusions have been strengthened (SPM of IPCC, 2007). While the visibility of large-scale (average and spatially detailed) reconstruc- tions stems from their ability to contextualize ���unprecedented��� cli- mate change in the twentieth century against a multicentury backdrop, such multiproxy reconstructions are critical to a variety of climate science studies. They provide a large-scale context with which to compare regional climate variability as reconstructed by single proxy records, which may ultimately help resolve the large- scale mechanisms of past low-frequency climate change. They also provide much-needed tests of the response of large-scale climate to a variety of climate forcings which occurred during the last millen- nium (most notably solar and volcanic forcing). The purpose of this review is to consider what direction under- lying scientific investigations might most profitably take in the

immediate future, to reduce existing uncertainties, eg, those inherent in the basic proxy data and the reconstruction of past temperature variability, and to provide greater insight into the factors governing past climate change. More work is needed to better understand and more comprehensively quantify the sources of uncertainty in various climate proxy archives and errors in reconstructions, and then to improve them. However, we first need to understand the reasons for the differences among existing climate reconstructions which make use of different types or combinations of climate proxy data and different statistical meth- ods to combine these data within a climate reconstruction. There is, as always, a clear need for more local and regional recon- structions from diverse proxies in as many parts of the world as possible. Climate model simulations can also play an important role here, acting as a surrogate climate, whose variability through time is perfectly known, and with which we may test the per- formance of alternative reconstruction approaches. Such tests, however, are tied to the underlying assumptions regarding the error structure of the climate proxy data. To make significant progress there is a parallel need to improve our understanding of the nature of the processes, both climatic and non-climatic, that influence climate proxy data, and the need to recognize and account for these intrinsic limitations. It is also widely recognized that future work ought to extend beyond the reconstruction of simple hemispheric average temperature series and important large-scale circulation indices. Instead, further investigations should also attempt to resolve seasonally specific variations and identify the associated patterns of temperature, pre- cipitation and circulation variability, perhaps examining specific anomalous periods in the past (selected from analysis of climate proxy observations, estimated forcing histories or model simula- tion results). This paper represents the outcomes of discussions at Wengen, Switzerland in June 2006 under the auspices of PAGES/CLIVAR held in an attempt to define how progress could be made on these issues. Background The purpose of the Wengen workshop was to synthesize the cur- rent state of late-Holocene climate reconstruction efforts, to assess the approaches to data���model comparisons and to elaborate on what possible or likely advances are expected in the coming years. The workshop discussions were organized into three principal subject areas: (1) Proxy data availability and reliability (2) Large-scale/regional reconstruction approaches and their uncertainties, and how these can be informed by climate mod- elling (from simple Energy Balance Models (EBMs) to fully coupled Atmosphere/Ocean General Circulation models (A/OGCMs), even including atmospheric chemistry) (3) Factors that influence/force the climate system (both natural and anthropogenic external factors as well as internal vari- ability) and how these are treated within climate models. This review is an extensive and critical examination based on these discussions and is organized in a similar framework. A brief summary from the meeting has already been published (Mann et al., 2006). The section ���Proxy data uncertainty��� below discusses aspects of the use of various types of proxy data, with a specific focus on their current availability and uncertainty. Although struc- tured by discipline, each subsection addresses the most important issues: reducing uncertainties (improving both reconstruction reli- ability and, for less-than-annually resolved proxies, improving dating accuracy and resolution) and making best and full use of instrumental records for calibration and verification (particularly important for decadally to annually resolved proxies) of recon- structed climatic parameters. The issues discussed in this section differ in focus and detail, in part reflecting the different maturity of each discipline. Some weight and a corresponding amount of text is allotted to the dendroclimatic issues. The justification for this is the generally large proportion of tree-ring based proxies used in many of the large-scale reconstructions, but also because it is felt that much of the discussion on interpretational limitations of these data has relevance for the increasing use of other forms of high-resolution proxy data, particularly as they become available over increasingly longer periods of time. The section ���Combining proxies to reconstruct large-scale pat- terns, continental and hemispheric averages��� discusses the various approaches that have been developed to combine currently avail- able proxy series into large-scale (continental to hemispheric) averages and to reconstruct internally consistent climate fields, representative of seasonal or annual conditions during the past. This section also describes the application of experiments using pseudo-proxies (derived from GCM output) to provide benchmark tests for assessing the performance of statistical reconstruction techniques. The following section, ���Climate forcing and histories��� reviews the development of past climate forcing histories, dis- cussing which are the most important for millennial-scale climate integrations, how different models implement the forcings, and discusses uncertainties in forcing histories. The final section con- cludes with a summary of major findings and a comprehensive set of recommendations for future work. Proxy data uncertainty Tree rings and the need for improved regional and temporal coverage Tree-ring-derived records have played a prominent role in attempts to establish how climate has varied in the recent past. Networks of climatically sensitive tree-ring chronologies have long been used to reconstruct detailed spatial patterns of interan- nual climate variability on regional and near-hemispheric scales, typically extending observed climate records by several centuries (Fritts, 1991 Schweingruber et al., 1991 Briffa et al., 1994, 2002b Cook et al., 2004b, 2007). Several chronologies extend- ing over a longer time span, with variability displaying a strong and direct association with changing local temperatures, have been utilized in virtually all published studies aimed at recon- structing Northern Hemisphere (NH) or global average surface temperature changes during the millennium leading up to the present (Jansen et al., 2007). We do not attempt here to discuss in detail the well-known pos- itive attributes of dendroclimatology. Reviews of the scope and general strengths of different types of tree-ring data (that they are continuous, precisely defined with annual resolution or better, accurately dated on a calendar timescale, widely distributed and rigorously calibrated against observed climate data) are already well described in a number of books and general articles (Fritts, 1976b Cook and Kairiukstis, 1990 Briffa, 1995 Schweingruber, 1996 Treydte et al., 2001, 2006 Hughes, 2002 McCarroll and Loader, 2004 Luckman, 2007) and the published proceedings of international conferences (eg, Hughes et al., 1982 Fritts and Swetnam, 1989 Bartholin et al., 1992 Dean et al., 1996). However, the continuing advancement of dendroclimatology as a discipline is not based solely on the exploitation of these strengths. It also involves an explicit appreciation of limitations or weak- nesses. Hence, our purposes in this review article are best served by drawing attention to some of the general lessons learned in tree-ring research, with a focus on the shortcomings of this proxy: P.D. Jones et al.: High-resolution palaeoclimatology of the last millennium 5

those that are innate characteristics of tree-ring data themselves, but also those that may be identified in current dendroclimatic practise (Hughes, 2002 Esper et al., 2007b). The issues discussed do not relate solely to the most commonly used tree-ring ���width��� data per se, but apply to all forms of tree-ring-derived proxy data, including densitometric (eg, Schweingruber, 1996), chemical or stable isotope (McCarroll and Loader, 2004 Treydte et al., 2007) data. The following discussion draws attention to some selected aspects of tree-ring data and dendroclimatology that are relevant to the continuing efforts to better understand late-Holocene climate variability. We first discuss the aspects of temporal and spatial coverage of tree-ring data, with a focus on the status of dendroclimatic studies in the tropics and the Southern Hemisphere. We then describe a number of issues relating to potential improvements in statistical methods used to assemble long tree-ring chronologies and how their statistical quality might be better represented. Finally, we draw attention to problems in the way chronologies are generally interpreted in terms of specific climate parameters. The points we raise about the need for improved ways to measure chronology confidence and possible limitations in the reconstruction of longer-timescale climate vari- ability are described at some length here because we believe them to be equally relevant to the analysis and interpretation of other proxy records discussed in the following sections of this review. Knowledge of the expanding geographic coverage and recent regional developments in dendroclimatology may be gleaned from various reviews (Hughes et al., 1982 Dean et al., 1996 Hughes, 2002 Luckman, 2007). Here we focus on the most recent developments in those regions highlighted in the AR4, as virtually devoid of tree-ring data, specifically the tropics and the Southern Hemisphere (SH), though we include discussion of the state of ���long-chronology��� development in these and other regions of the world. Prospects for tropical dendroclimatology Tropical dendrochronologywas long considered impractical because the growth periodicit y of most tropicaltree species is seldom clearly and unambiguously defined (eg, Jacoby, 1989 Gourlay, 1995 Vetter and Wimmer,1999 Worbes,1995). However,this has turned out not to be entirely true, as seasonally dry regions have produced tree-ring records from tropical Asia (eg, Berlage, 1931 Buckley et al., 1995, 2005, 2007a, b D���Arrigo et al., 1994, 1997, 2006a Pumijumnonget al., 1995 Sano et al., 2008), from Africa (eg, Stahle et al., 1999 Tarhule and Hughes, 2002 Therrell et al., 2006) and from the American tropics (eg, Biondi, 2001 Sch��ngart et al., 2004a, b Therrel et al., 2004 Biondi et al., 2005 Brienen and Zuidema, 2005, 2006 D���Arrigo and Smerdon, 2008). Furthermore, the notion that tropical regions that lack clear seasonality pose an insurmountableproblemhas turned out not to be the case. Evans and Schrag (2004),Poussartet al. (2004)and Poussart and Schrag (2005) are among the first to have applied improved methods of stable iso- tope geochemistry that show the possibility of using many appar- ently ���ringless��� species for dendroclimatic studies in tropical environments. Evans and Schrag (2004) demonstrated the applica- tion of these methods on species from Costa Rica, while Poussart et al. (2004) and Poussart and Schrag (2005) applied them to Thai and Indonesian trees. In all cases clear periodicityin ��18O and ��13C was demonstratedand determinedto be annual,and relationships to rain- fall were established.Poussartet al. (2006) also appliedx-ray micro- probe synchrotronanalysis to a ringlessspeciesfrom Thailand,in an attempt to more easily define annual periodicity through chemical cycles of calcium, and this too holds promise for wider application. In spite of the successes, however, formidable obstacles still restrict the development of tropical dendroclimatology, including the scarcity of suitable tree species with easily identifiable and measurable annual rings, and continued pressure on forest resources due to logging and disturbance that limits the availability of old- growth trees. A veritable absence of information regarding the eco- physiology and phenology of tree species further exacerbates problems of working in the tropics (Borchert, 1995). Even with the use of isotopic time series the problem of crossdating is still severe, given the nature of locally absent banding that is often severe in many tropical trees. Furthermore, the time and costs associated with isotopic geochemistry currently inhibit widespread application. Temporal control remains an issue with tropical tree-ring analy- ses under many situations, and radiocarbon (14C) measurements have been used to assess the annual nature of growth rings through detection of the radiocarbon ���bomb spike��� (eg, Biondi and Fessenden,1999 Hua et al., 1999) and to estimate the age and aver- age growth rates of some tropical trees (eg, Poussart and Schrag, 2005). However, species-specific effects can limit the applicationof 14C dating because of uncertainties associated with soil respiration, internal carbohydrate transfer and other ecophysiological factors (Worbes and Junk, 1989). Dendrometer studies have been employed (eg, Buckley et al., 2001 DaSilva et al., 2002), and these provide a useful alternative, along with cambium-wounding or ���pin- ning��� methods that give a reference for growth from a time of known scarring of the cambium (eg, Mariaux, 1967 Nobuchi et al., 1995). Analysis of tropical wood chemistry may also reveal sea- sonal signatures of cambium activity, as exemplified by Poussart et al. (2004) in Indonesia, and Poussart and Schrag (2005) in Thailand, both of whom demonstrate that the generation of replicated suban- nual ��18O and ��13C records from ringless tropical trees is possible over several decades. More traditional approaches to dendroclimatology in the tropics still have their place. Buckley et al. (2007b) produced an inferred reconstruction of Palmer Drought Severity Index (PDSI), a meas- ure of soil moisture availability (Palmer, 1965), based on ring- widths of teak from northwestern Thailand that extended back nearly 500 years. This record illuminated periods of decadal-scale drought, the most significant of which persisted from 1690 to 1720 and 1735 to 1765, respectively, and coincided with periods of extreme social unrest. Sano et al. (2008) produced a 535-yr ring-width record from living Fokienia hodginsii (Dunn, A. Henry and H.H. Thomas) of the family Cupressaceae. The authors used this record to reconstruct PDSI for the pre-monsoon period of March to May over the past 500 years for northern Vietnam. Significantly this is the first record from the region that success- fully calibrated and verified the reconstruction statistically, using the available instrumental records. When compared with the Buckley et al. (2007b) teak record from Thailand, a similar drought from 1750 to 1780 is revealed, suggestive of a ���megadrought��� that may have extended from Burma to Vietnam. This period coincides with the outright collapse of all the major kingdoms in Southeast Asia (Leiberman, 2003), pointing to a possible direct climatic impact on the societies of the region. In some instances it has been shown that the response of tropi- cal tree species to climate is not as straightforward as for temper- ate regions. Rivera et al. (2002) showed that for a number of species the influence of subtle changes in photoperiodicity in the American and Asian tropics is more important than rainfall for producing annual flushing of leaves in shade-sensitive species. Buckley et al. (2007a) illustrated an apparent inverse response to drought in Pinus merkusii across southeast Asia, possibly linked to reduction of photosynthetically active radiation during times of intense convective rainfall. Saleska et al. (2007) found that Amazon forests significantly increased their biomass in response to a prolonged drought in 2005, counter to model predictions of forest collapse in response to drought (eg, Betts et al., 2004). At the peak of the 2005 Amazon drought, the authors found a signif- icant increase in canopy ���greenness���, indicating an ecological and physiological vegetation response that is opposite to a presump- 6 The Holocene 19,1 (2009)

tion that anomalously low rainfall would negatively influence these forests. Clark et al. (2003) found a negative relationship between minimum temperature and tropical tree growth at La Selva in Costa Rica, but no relationship with interannual varia- tions in precipitation (Clark and Clark, 1994). These findings have implications for the interpretation of isotopic time series from tropical trees, illustrating the need for research into the physiolog- ical response of tropical forest species to climate, and a greater understanding of their anatomy. Furthermore, the research of both Feely et al. (2007) and Clark et al. (2003) suggest a ���deceleration��� in growth rates of tropical forests in Costa Rica, Panama and Malaysia, in opposition to some predictions that pan-tropical forests could experience increased growth induced by CO2 fertilization or increases in water use effi- ciency (eg, Lewis et al., 2006). There continues to be much debate about the response of tropical forests to climate change and funda- mental questions remain about whether or not tropical forests serve as carbon sink or source, and the extent to which any changes in biomass can be attributed to anthropogenic change (eg, Wright, 2006). Dendrochronology can help to answer many of these ques- tions if it can be more widely applied, and the developments so far are encouraging. As more species are analysed for their eco-phys- iological characteristics and their response to climate, more old- growth species are discovered and improvements are made to geochemical methodologies, the prospects for widespread applica- tion of dendrochronology in the tropics look promising. Recent developments in South America and southern Africa For the interval 1890���2000, tree-ring width chronologies (from Swietenia macrophylla and Cedrela odorata) were developed by D��nisch et al. (2003) in Mato Grosso, Brazil (10��09���S, 59��26���W). Correlation analyses revealed a significant relationship between seasonal precipitation and the growth of both species. Several month-long inundation indices influence the formation of annual rings in trees growing in the seasonally flooded Amazon plains (3��� 4��S, 65��W). Ring widths are inversely related to duration of the flood. A 200-yr long Euphorbiaceae chronology (Piranhea trifoli- ata) has been used to estimate the length of the vegetation season, which is, in turn, related to ENSO events (Sch��ngart et al., 2004a, b). The results indicate that during the last two centuries, the sever- ity of ENSO events in the Amazon basin has significantly increased. In a related study, Sch��ngart et al. (2004b) developed tree-ring chronologies from Macrolobium acaciifolium in two different floodplains in Central Amazonia. Maximum tree age in the nutri- ent-poor ���black-water��� was more than 500 years, contrary to the nutrient-rich ���white-water��� floodplain, where ages are not older than 200 years. Ring-width variations in both floodplain forests were significantly correlated with the length of the vegetation period derived from the daily recorded water level at the port of Manaus since 1903. Both chronologies showed increased wood growth during El Ni��o events associated with negative precipita- tion anomalies and lower water discharge in Amazonian rivers. Exploratory work has established the dendrochronological potential of several tropical lowland species in the Bolivian sector of the Amazon basin (11���15��S, 66���68��W). Brienen and Zuidema (2005) crossdated six rainforest species and established the influ- ence of annual and seasonal rainfall on radial growth. Ongoing work in Argentina and Brazil has also identified several species typical of the dry tropical forest of southern Bolivia���northern Argentina (16���24��S) and southeastern Brazil (22���25��S) that show clear rings and potential for dendroclimatic studies. A major advance in the effort to expand the spatial coverage of tree-ring records across the Americas has been the recent devel- opment of Polylepis tarapacana chronologies in the Bolivian Altiplano (Argollo et al., 2004). These records located between 16 and 22��S and above 4500 m elevation, are the closest to the Equator in the Andes and the highest-elevation chronologies in the world. Most Polylepis records cover the past three to four cen- turies, but some extend over seven centuries (Argollo et al., 2004 Soliz et al., 2008). Examination of interannual variations in ring width and climate in the Altiplano indicate that the growth of Polylepis is associated with summer water balance (Morales et al., 2004). In addition, P. tarapacana chronologies from the south- central tropical Andes provide high-resolution records that are extremely sensitive to ENSO in the tropical Pacific, and represent an important component to be considered in future multiproxy ENSO reconstructions (Christie et al., 2008). A major requirement in Southern African dendroclimatology is the need to examine the many tropical and subtropical species to assess their dendrochronological potential (see the pioneer work of Lilly, 1977 and February, 1996). Stahle et al. (1997) have begun surveying the diverse indigenous forests in tropical Africa for species suitable for dendroclimatology. Two tropical chronolo- gies of Pterocarpus angolensis from Zimbabwe are strongly cor- related with total rainfall amounts during the wet season. Both chronologies reach back only to 1870 at present, but P.angolensis is the most important timber species in south tropical Africa, and old samples survive in buildings and other diverse sources that may permit the eventual development of 200���300 yr chronologies in southeastern Africa (Stahle et al., 1997). Developing more ���long��� tree-ring-based chronologies There are very few tree-ring chronologies from around the globe that extend back 1000 years. This is very apparent in figure 6.11 of Jansen et al. (2007), which shows only 16 locations globally from which tree-ring data have been used in large-scale tempera- ture reconstructions to date. Of these, three are in the SH. The remainder are virtually all restricted to the western edge of North America or the high latitudes of Eurasia, with the exceptions being two sites adjacent to the Mediterranean and one in Mongolia. The contribution of dendroclimatology to improved late-Holocene cli- mate reconstruction must involve a geographic expansion of work developing long composite chronologies. Many areas, some with a proven history of (admittedly short-timescale) tree-ring research and demonstrably climatically sensitive trees, have yet to be sys- tematically investigated for their potential to produce subfossil wood and hence much longer chronologies. More long records are needed in the northern mid latitudes though work is ongoing in Europe (Nicolussi and Schiessling, 2001 Grabner et al., 2001 Helama et al., 2005 B��ntgen et al., 2006, 2008 Popa and Kern, 2008), N Africa (Esper et al., 2007a), N America (Barclay et al., 1999 Buckley et al., 2004 Luckman and Wilson, 2005) and Asia (Sidorova et al., 2006 Esper et al., 2007c). Many more long chronologies are needed in the SH. Significant recent achievements in South America include the developmentof a 5666-yr-long composite chronology from Fitzroya cupressoides, currently the longestcontinuouschronology in the SH (Wolodarsky- Franke, 2002). Chronologies of this species (from Argentina and Chile) were used in the 1990s to develop the first millennium-long temperature reconstructions for South America. However, recent work suggests a lack of stabilityin the relationship between the low- frequency component of climate and F. cupressoides tree growth, particularly during recent decades. Studies using Austrocedrus chilensis have developed several new records. A 1864-yr-long chronology from northern Argentinean Patagoniais a major compo- nent of a preliminary SH ENSO reconstruction for the last 1300 years and collaborative studies between tree-ring laboratories in Chile, Argentinaand USA have developed800-yr-long precipitation reconstructions for central Chile (LeQuesneet al., 2006, 2008). Development of millennia-long tree-ring chronologies from Australia and New Zealand has been based on three tree species: P.D. Jones et al.: High-resolution palaeoclimatology of the last millennium 7