The late Holocene kauri chronology: assessing the potential of a 4500-year record for palaeoclimate reconstruction

• The updated and revised late Holocene kauri chronology spans 4491 years (2488 BCE–2002 CE).
• The chronology includes data from modern, archaeological, and sub-fossil sites.
• The character of the three data sets is different.
• There is a potential difference in amplitude of the signal in sub-fossil kauri.
• Complex changes in chronology composition may influence centennial scale trends.

Millennial and multi-millennial tree-ring chronologies can provide useful proxy records of past climate, giving insight into a more complete range of natural climate variability prior to the 20th century. Since the 1980s a multi-millennial tree-ring chronology has been developed from kauri (Agathis australis) from the upper North Island, New Zealand. Previous work has demonstrated the sensitivity of kauri to the El Niño-Southern Oscillation (ENSO). Here we present recent additions and extensions to the late Holocene kauri chronology (LHKC), and assess the potential of a composite master chronology, AGAUc13, for palaeoclimate reconstruction. The updated composite kauri chronology now spans 4491 years (2488 BCE–2002 CE) and includes data from 18 modern sites, 25 archaeological sites, and 18 sub-fossil (swamp) kauri sites. Consideration of the composition and statistical quality of AGAUc13 suggests the LHKC has utility for palaeoclimate reconstruction but there are caveats. These include: (a) differences in character between the three assemblages including growth rate and sensitivity; (b) low sample depth and low statistical quality in the 10th–13th century CE, when the record transitions from modern and archaeological material to the swamp kauri; (c) a potential difference in amplitude of the signal in the swamp kauri; (d) a westerly bias in site distribution prior to 911 CE; (e) variable statistical quality across the entire record associated with variable replication; and (f) complex changes in sample depth and tree age and size which may influence centennial scale trends in the data. Further tree ring data are required to improve statistical quality, particularly in the first half of the second millennium CE.