This thesis demonstrates some of the potential of Phyllocladus trichomanoides D. Don (tanekaha) for dendrochronological research, especially dendroclimatology. The type of vegetation associated with sites producing a chronology showed no obvious pattern. The resulting non-specific range of suitable sites for dendrochronological sampling was thought to be a favorable species characteristic. Standardization of tree-ring series using a 50-year Gaussian filter resulted in 28% improvement of retained common variance above that obtained from using conventional polynomial filters. The problem of autocorrelation was investigated. Tanekaha had a consistent autocorrelation pattern through both space (i.e. latitudinal and altitudinal ranges) and time (preserved forest and contemporary stands). P. glaucus (toatoa) also showed a spatially consistent yet distinctive pattern. These different diagnostic patterns implied a non-climatic cause (i.e. physiological), illustrating the need to ensure their removal before species chronologies are compared or used for climate modelling. Low order autoregressive models (ARMA(p,0)) were used to filter out the significant levels of autocorrelation. Subsequent comparison of the chronologies showed a consistent and highly significantly correlated pattern between both sites and species. The paucity of information about the physiology of Phyllocladus spp. led to the testing of several climatic variables in response function analyses. The "best" variables (based on only statistical evidence) were minimum monthly temperature, monthly total raindays and an optimal 12 month span of August (t) to July (t+1). Preliminary attempts with response functions using different combinations of predictor (climate variables) and predictand (chronologies) data sets generally failed to verify. Neither the extent of time from which the response functions were based (no analogue situation) nor a supposed "physiological shock" period (poor growth period) were the causes for non-verification. Further investigation showed that, for both the individual and combined chronologies, the temperature response was similar but rainfall varied. The instability through time of the rainfall response was particularly disconcerting since it broke one of the fundamental assumptions used in dendroclimatology. It was therefore concluded that climatic reconstruction could be attempted only on the temperature data. The temperature data of three seasons (spring, summer and autumn) were tried with transfer function models. Of these the summer season explained the most variance and had the highest reduction of error(RE) values. The selected model used the period 1918 to 1982 for calibration and 1853 to 1917 for verification. Summer temperatures from 1982 to 1750 were reconstructed. The reconstruction modeled only c.30% of the variance but was highly correlated to the summer temperature series developed by Norton (1983). This was interpreted as further independent verification of the reconstructed series. The summer temperatures reconstructed from 1982 to 1750 were concluded to have been similar to the recorded pattern from 1853. The same regressors were applied to the buried forest chronology and the summer temperatures reconstructed from 105 BC to AD 175. Because of the non-continuous tree-ring record no comment can be made of actual ambient conditions in comparison to those of today. However, the reconstructed pre-Taupo summer series did show increased variation towards the time of the eruption and more pronounced cool summers than hot ones. Another applied use of a tanekaha chronology was demonstrated with 14C dating. A statistically significant fit of 14C ages was obtained between twelve contiguous decades from the buried forest tanekaha chronology with that from a Northern hemisphere decadal calibration curve based on Sequioadendron giganteum(Stuiver & Becker 1986). This "wiggle-matching" to the calibration curve places the year of the Taupo eruption as AD 177±18 10. An Independent check of the "new" date was conducted using the most reliable data of Healy et al. (1964) with the same calibration series as described above. A date of AD177±4 44 was obtained which supports the wiggle-matched date. The confidence interval of the wiggle-matched date also coincides with the date proposed by Wilson et al. (1980) of AD186 (based purely on ancient written records).
| Identifer | oai:union.ndltd.org:ADTP/276980 |
| Date | January 1989 |
| Creators | Palmer, Jonathan Gray |
| Publisher | ResearchSpace@Auckland |
| Source Sets | Australiasian Digital Theses Program |
| Language | English |
| Detected Language | English |
| Rights | Items in ResearchSpace are protected by copyright, with all rights reserved, unless otherwise indicated., http://researchspace.auckland.ac.nz/docs/uoa-docs/rights.htm, Copyright: The author |
Page generated in 0.0024 seconds