Climate change since the Last Interglacial in northern New Zealand inferred from pollen and chironomid records of the Auckland Maars

In light of contemporary climate change it is more important than ever to understand past shifts in climate, especially past warm phases, and their effects on ecosystems and societies. From compilations of global climate reconstructions, several periods have been identified that might have been warmer than today, the two most recent of which are the Holocene Thermal Maximum (~11 – 5 kyr BP) and the Last Interglacial (~129 – 116 kyr BP). However, spatio-temporal complexities are typically smoothed out in global climate reconstructions and we do not have a good understanding of the regional differences in past climate. The southern mid-latitudes especially are underrepresented in palaeoclimate research.  For this thesis I analyse the sediments from two maars within the Auckland Volcanic Field: Orakei Basin, which erupted ~126.0 kyr BP and accumulated sediments until ~9 – 8.5 kyr BP; and Lake Pupuke, which still contains a lake today and therefore covers the Holocene. Quantitative climate reconstructions are necessary to put the Orakei Basin and Lake Pupuke records in a broad context and to enable comparisons of past and future climates. For this study I focus on biological proxies preserved by lake sediments, namely pollen, which primarily responds to mean annual air temperatures (MAAT), and chironomids, a surrogate for summer air temperatures (SmT). Together, MAAT and SmT reconstructions from the same site can provide insight into changing seasonality over time, an underexplored dimension of proxy-based reconstructions. The chironomid record covers just the last ~14 cal kyr BP however, because of low head capsule abundances in older sediment sections.  The Orakei Basin pollen record and associated MAAT reconstruction cover ~85 to 9 cal kyr BP and show five distinct phases comparable to Marine Isotope Stages (MIS) 5 to 1. This association is confirmed by the preliminary tephrochronology of the core. The broad similarity of the Orakei MAAT trend to the MIS and other records from New Zealand implies all were driven by northern high-latitude summer insolation, consistent with the Milankovitch orbital forcing hypothesis. Several patterns superimposed on the general trend stand out: first, MIS 4 is a brief cool period, which is inconsistent with the observation that glacier advances equivalent to those of the late last glacial maximum occurred ~65 kyr BP in the Southern Alps, possibly due to the seasonal distribution of energy from solar insolation. Second, MIS 3 displays an earlier warm phase followed by a progressive cooling trend which might be correlated to decreasing local summer insolation intensity. Third, glacial conditions of MIS 2 appear consistent with the early onset of the last glacial maximum in the southern mid latitudes, which was likely driven by regional insolation intensity.  The Lake Pupuke pollen and chironomid records, covering the last ~14 cal kyr BP, show no evidence of a past warm period equivalent to the Holocene Thermal Maximum. MAAT is stable throughout the Holocene, whereas SmT increases between 10 and 3 cal kyr BP. The latter shows a strong relationship with integrated local summer insolation. The temperature reconstructions lead to the conclusion, first, that seasonality was low during the Early Holocene (12 to 9.3 cal kyr BP), and second, that during mid-to-late Holocene (after ~7 cal kyr BP) summers were hot and dry, allowing the tall conifer kauri to expand throughout northern New Zealand.  The Lake Pupuke chironomid-SmT reconstruction highlighted an issue with the transfer function model, namely, that it was not able to reconstruct values close to modern day (18.9°C). Therefore, I explore an extended training set which encompasses a longer temperature gradient. New models are fitted using both traditional techniques and modern machine learning methods. The new model improves the SmT reconstruction from Lake Pupuke, in the sense that reconstructed temperatures now reach modern day values. However, the SmT trend is the same as the original trend, substantiating the previously drawn conclusions.  During the course of this research, I discovered that density separation during pollen preparation can lead to varying relative abundances, depending on the specific gravity used. After some experimentation I found that using a low specific gravity (2.0; recommended value in the literature) can result in the overrepresentation of buoyant pollen grains, leading to erroneous interpretations.  Together, these results point out the importance of considering regional-to-local drivers of climate changes superimposed on global reconstructions. Multi-proxy records can help disentangle the different aspects of the climate system, where especially chironomids can be helpful to elucidate the role of SmT and local summer insolation. Finally, this thesis shows the importance of questioning the appropriateness of conventional methodologies and where possible, addressing their limitations.