Orbital or Thermal Causes of Glaciation in New Zealand

 

View of the Waimakariri River at the Poulter moraine

 

Personnel

Dr Jamie Shulmeister, Dr Maureen Marra, Dr Phil Tonkin, Henrik Rother, Craig Woodward, Emeritus Prof. Jane Soons.

 

Collaborators

Dr David Fink (Australian Nuclear Science and Technology Organisation), Dr Rich Leschen (Landcare Research), Dr Jim Renwick (NIWA), Dr Uwe Rieser (Victoria University of Wellington), Dr Geoff Seltzer (Syracuse University), Jacquie Smith (Syracuse University), Dr Glenn Thackray (Idaho State University)

 

Project Support

Marsden Grant 2003-2006

 

Project Description and Background

This project will determine the causes of glacial advances in New Zealand over the Late Quaternary (last 750,000 years, but focussing on the last 100,000 years). We propose to test a new model of westerly wind forced glaciation against the existing NH thermal forcing model of glaciation. The two models would create dramatically different glacial worlds (wetter and milder versus drier and colder) and the results from this work will allow us to resolve important questions on the links between glaciations in the two hemispheres and the role of the Southern Hemisphere westerlies in determining longer term global climate change.
Despite many years of work (e.g. Dobson, 1872; Gage, 1958; Suggate, 1965; 1990), the glacial geology of New Zealand is still inadequately understood. While the general pattern of glaciation during the last (Otiran) ice age is defined, two new ice masses have been added recently (Bacon et al., 2001; Shulmeister et al., 2003) and the timing of advances remains a serious problem. In fact, the keystone glacial stratigraphic sequence in North Westland (Suggate and Waight, 1999) is correlated to the marine isotope record (MIS stages) via raised marine benches which themselves have no secure age control prior to the Holocene (last 10,000 years). In addition, there are no well dated glaciated valley sequences in New Zealand extending to times pre-dating the last glacial maximum (LGM, c.21,000 years ago) though it has been apparent that some pre-LGM advances were significantly larger than LGM events (e.g. Porter, 1975). Of these earlier events, the only reliable ages come from the speleothem sequences in Aurora Cave, Fiordland (Williams, 1996; Williams and Fink, 2002), where larger than LGM advances occurred at 49-47 ka and about 65 ka.
The existing paradigm is that New Zealand glaciation mirrors the timing and pattern of glaciation in the northern mid-latitudes. This is based on marine records (i.e. Nelson et al., 1985) which clearly display 100,000 year glacial-interglacial amplitude. These records are, in the main, of too low resolution to infer glacial patterns within MIS 2-4 in New Zealand. The apparent synchrony of ice advances since the LGM in New Zealand (e.g. Denton and Hendy, 1994) with the global record is consequently used as proof of the synchrony of the entire record (Broecker,1997). In order to explain these non-lagged responses to NH forcing, thermal drivers via atmospheric or oceanic transfers are invoked to explain glacial advances (e.g. Denton et al., 1999).

 

Recent work has demonstrated that the strength and pattern of westerly wind flow and concomitant issues of precipitation and cloud cover are critical to modern glacier mass balance in New Zealand (Fitzharris et al., 1992; Tyson et al., 1997; Hooker and Fitzharris, 1999). Specifically, the modern West Coast glaciers respond with a 5-7 year lag to averaged seasonal synoptic conditions, with advances occurring after positive SW flow anomaly years and retreats associated with years of meridional (northerly) anomalies. East coast glaciers have longer response times and respond to longer period (decadal) forcing. Changes in westerly flow can be directly correlated with El Niño, and other climate phenomena, and at least for the West Coast glaciers, a prima facie relationship between a westerly index and glacier fluctuation has been established. Millennial scale changes in the westerlies will be controlled by the pole-equator temperature gradient, which is most likely driven by orbital forcing through modification of seasonality (e.g. Dodson, 1998; Shulmeister, 1999; Shulmeister et al., accepted). We propose that longer term fluctuations in westerly flow, in association with modest regional cooling during stadial periods, may be the primary drivers of New Zealand glaciation, rather then transmission of large thermal signals from the NH.
Berger (1992) and Berger and Loutre (1991) have calculated solar insolation at key latitudes for the last ten million years at 1000 year intervals. Using changes in the Southern Hemisphere insolation gradient, a long-term proxy for westerly flow can be constructed (Shulmeister et al., accepted). Because this is largely precessionally forced (see data in Berger, 1992) the model predicts the largest positive westerly anomaly at c.47 ka. with subsidiary maxima at c. 72 ka and 21 ka.
If the westerly forcing hypothesis is correct, glacial advances will be in synchrony with precessional maxima, of short duration and characterized by rapid glacier growth and decay under relatively mild conditions. In contrast, if Northern Hemisphere thermal forcing dominates, glacier growth should occur at precessional minima, be gradual and overall conditions should be colder. We are developing a novel consensus method that involves three independent lines of research and climate modelling to resolve these questions.
Firstly, the thermal model predicts a last glacial maximum in New Zealand at c. 21ka while the westerlies model predicts c. 47 ka as the maximum. In order to test these hypotheses we propose to document the extent, timing and nature of a limited number of glacial systems in detail the Rakaia, Waimakariri, and Hope/Waiau valleys. We have selected three adjacent East Coast river valleys for the work because they were large glacial systems with similar lithologies and climatologies. Reproducibility is critical and the size and close proximity of the three valleys means that any glacial advances should have been synchronous. We need to partially remap the glacial systems because much of the original work (i.e. Clayton, 1968; Gage, 1958; Soons, 1963) was done without air photographs, or the ability to precisely constrain elevations, which are critical when terrace remnants are being examined. We will use differential GPS to establish spot heights on terraces and digital elevation models, satellite imagery and aerial photographs to remap the valleys, concentrating on the LGM and older parts of the systems. Having identified the major moraines and their associated outwash systems and proglacial lakes, we will undertake a high resolution dating campaign based on cosmogenic dating (dominantly 26Al, 10Be: Fink, Smith) of the glacial moraines and luminescence (IRSL) dating of glacifluvial and lacustrine deposits. Cosmogenic dating targets bedrock outcrops and large boulders on moraines and cosmogenic 26Al and 10Be require target rocks with high quartz contents (e.g. Gosse and Phillips, 2001). Preliminary fieldwork suggests that cosmogenics will be the primary tool in the Waimakariri whereas the profusion of glacio-lacustrine sediments in the Rakaia system will require IRSL ages. The Hope/Waiau system falls between the two and will facilitate cross-referencing of the techniques.
We will also use modern beetle data, in the form of molecular phylogenies of extant taxa to determine colonisation rates into areas affected by past glaciations, as a validation of the geochronology. Phylogenetic evolution among populations can be converted into numerical ages by determining the rate of nucleotide changes along branches of a phylogenetic tree. The technique has recently been used with arthropods to successfully test ‘beech gap’ hypotheses in New Zealand (Trewick and Wallis, 2001). We will sample populations of two ubiquitous beetle species (Scaphisoma hanseni, Staphylinidae, Scaphidiinae and the flightless Pristoderus bakewelli, Zopheridae) along transects passing through glaciated terrains into non-glacial areas in the Hope/Waiau and Rakaia systems. We will sequence 2 mitochondrial genes (in 2 directions) for a total of about 1.5kb of gene sequence and use these data to reconstruct phylogenetic trees (Swofford et al 1996) at the population level (see Taberlet et al 1998, Flanagan et al 1999, Howitt 2001) and date them using the Bayesian “two-tip” method proposed by Edwards and Beerli (2000). The technique will provide an independent cross-check on the ages of older glacial deposits.
Secondly, the thermal transfer model requires a significant reduction in Mean Annual Temperatures (MAT) to trigger the glacial advances. We will focus on reconstructing the thermal decline at the LGM, which is inferred to be in the range of 4-7°C for New Zealand (e.g. Soons, 1979; McGlone et al., 1993; Shulmeister et al., 2001). There are clear anomalies in these estimates. For example, macrofossils from a glacial maximum site in a tributary of the Rakaia (Soons and Burrows, 1978) suggest virtually modern conditions even though the Rakaia glacier was proximal to the site. While the vegetation records do indicate widespread disturbance of forest at the LGM, increased landscape instability may play as significant a role as thermal decline in creating this landscape. New Zealand floral records are widely regarded as poor discriminants of thermal change (e.g. McGlone et al.,1993, p299) and we have pioneered fossil beetle work in New Zealand to provide a reliable paleothermometer. We have initiated trial work with chironomids (midges) for paleothermometry also, and are continuing with pollen work for moisture balance and general paleoecological support of the other techniques. We will develop a detailed regional scale paleoclimatic reconstruction based primarily on beetles for the LGM and, if we can identify appropriate sites, for the 47 ka period. The quantitative paleoclimatology will be derived from transfer functions (CANOCO: ter Braak and Smilauer, 2002) for chironomids and from a mix of bioclimatic modelling (e.g. Kershaw and Nix, 1988) and Mutual Climatic Range (MCR) (e.g. Elias, 1997) determinations for the fossil beetle work.

 

Thirdly, while papers frequently cite increased windiness in New Zealand at the LGM (e.g. Fenner et al., 1992; McGlone, 2001), detailed evaluation of the records shows strong evidence for increased dust emissivity in the landscape and for increased persistence of westerlies, but no conclusive evidence of increased wind speed (Shulmeister et al., accepted). Changes in the intensity of westerly wind through time can be directly tested by measuring changes in grain size of loess downwind of the glacial systems. Banks Peninsula rises up to 800 m above the Canterbury plains, about 50 km east of the Rakaia and Waimakariri Gorges and is mantled with thick loesses (e.g. Raeside, 1964; Goh et al., 1978; Berger et al., 2001). Since the loesses are greywacke derived and the Peninsula is a basaltic caldera complex, determining aeolian source is straightforward. Grain size analyses and provenance analyses underpin most paleo-aeolian studies (e.g. Thiede, 1979; Alloway et al., 1992) but no systematic grain size studies have been undertaken on the peninsula to determine paleo-wind parameters, though numerous small scale sediment studies for engineering purposes exist. We (Shulmeister/Tonkin) will carefully examine outcrop to select only primary loess deposits and avoid sites with visible pedogenesis. Age control will be established using luminescence dating (Rieser) and an index of westerly flow based on changes in modal grain size will be constructed for as much of the last glacial cycle as feasible.

We will run global and regional climate simulation using the UKMO Global Circulation Model (Cullen, 1993; Hewitt et al., 2003) with glacial limits for the 21 and 47 ka periods derived from this study (and other sources) as an input and using thermometry for the 21 ka period to force the model run. If we find sites to reconstruct temperatures at 47 ka we will also use these data to constrain parameters at 47 ka. The reconfiguration of the regional model for the 21 ka (and if available 47 ka) interval/s will then be used to retrodict habitat modification at the LGM (and 47 ka). Habitat modification under different glacial models will be quite different and we will examine the biodiversity implications of the climatic modelling.

If the westerly forcing hypothesis is correct, similar controls on glaciation would be expected in Northern Hemisphere regions subject to dominant westerly flow. The ENSO/westerly link in New Zealand glaciation also means that patterns visible in New Zealand data should be reflected in conditions in tropical and mid-latitude west coast South America. This proposal links with researchers in the Western USA (Thackray) and tropical South America (Seltzer) who are establishing glacial chronologies in these regions. The aim is to ultimately build global models of climate forcing, and the New Zealand work is a pilot for the larger project.

The outcome of all the work will be an improved understanding of the specific driver(s) of glaciation in New Zealand. Deciphering the climatic teleconnections between the Northern Hemisphere and the southern mid-latitudes is critical for determining the actual processes of global climate change (e.g. Denton et al., 1999). Understanding the mechanisms is also necessary for understanding the impacts of climate change on New Zealand. The rival models create quite different glacial environments in New Zealand and we use the GCM output to evaluate the ability of New Zealand fauna and flora to respond to past climate changes. We will show the implications and options for future biodiversity, which is particularly important for New Zealand because it tends to be highly susceptible to extinction driven ecological change (e.g. Leschen and Rhode, 2002).

 

References