The aim of this thesis was to examine the spatial and the temporal patterns of anxiety and chest pain resulting from the Canterbury, New Zealand earthquaeks. Three research objectives were identified: examine any spatial or termporal clusters of anxiety and chest pain; examine the associations between anxiety, chest pain and damage to neighbourhood; and determine any statistically significant difference in counts of anxiety and chest pain after each earthquake or aftershock which resulted in severe damage. Measures of the extent of liquefaction the location of CERA red-zones were used as proxy measures for earthquake damage. Cases of those who presented to Christchurch Public Hospital Emergency Department with either anxiety or chest pain between May 2010 and April 2012 were aggregated to census area unit (CAU) level for analysis. This thesis has taken a unique approach to examining the spatial and spatio-temporal variations of anxiety and chest pain after an earthquake and offers unique results. This is the first study of its kind to use a GIS approach when examining Canterbury specific earthquake damage and health variables at a CAU level after the earthquakes. Through the use of spatio-termporal scan modelling, negative and linear regression modelling and temporal linear modelling with dummy variables this research was able to conclude there are significant spatial and temporal variations in anxiety and chest pain resulting from the earthquakes. The spatio-termporal scan modelling identified a hot cluster of both anxiety and chest pain within Christchurch at the same time the earthquakes occurred. The negative binomial model found liquefaction to be a stronger predictor of anxiety than the Canterbury Earthquake Recovery Authority's (CERA) land zones. The linear regression model foun chest pain to be positively associated with all measures of earthquake damage with the exception of being in the red-zone. The temporal modelling identified a significant increase in anxiety cases one month after a major earthquake, and chest pain cases spiked two weeks after an earthquake and gradually decreased over the following five weeks. This research was limited by lack of control period data, limited measures of earthquake damage, ethical restrictions, and the need for population tracking data. The findings of this research will be useful in the planning and allocation of mental wellbeing resources should another similar event like the Canterbury Earthquakes occur in New Zealand.
Globally, the maximum elevations at which treelines are observed to occur coincide with a 6.4 °C soil isotherm. However, when observed at finer scales, treelines display a considerable degree of spatial complexity in their patterns across the landscape and are often found occurring at lower elevations than expected relative to the global-scale pattern. There is still a
lack of understanding of how the abiotic environment imposes constraints on treeline patterns, the scales at which different effects are acting, and how these effects vary over large spatial extents. In this thesis, I examined abrupt Nothofagus treelines across seven degrees of
latitude in New Zealand in order to investigate two broad questions: (1) What is the nature and extent of spatial variability in Nothofagus treelines across the country? (2) How is this variation associated with abiotic variation at different spatial scales? A range of GIS, statistical, and atmospheric modelling methods were applied to address these two questions.
First, I characterised Nothofagus treeline patterns at a 15x15km scale across New Zealand using a set of seven, GIS-derived, quantitative metrics that describe different aspects of treeline position, shape, spatial configuration, and relationships with adjacent vegetation.
Multivariate clustering of these metrics revealed distinct treeline types that showed strong spatial aggregation across the country. This suggests a strong spatial structuring of the abiotic environment which, in turn, drives treeline patterns. About half of the multivariate treeline
metric variation was explained by patterns of climate, substrate, topographic and disturbance variability; on the whole, climatic and disturbance factors were most influential.
Second, I developed a conceptual model that describes how treeline elevation may
vary at different scales according to three categories of effects: thermal modifying effects, physiological stressors, and disturbance effects. I tested the relevance of this model for Nothofagus treelines by investigating treeline elevation variation at five nested scales (regional to local) using a hierarchical design based on nested river catchments. Hierarchical linear modelling revealed that the majority of the variation in treeline elevation resided at the broadest, regional scale, which was best explained by the thermal modifying effects of solar radiation, mountain mass, and differences in the potential for cold air ponding. Nonetheless, at finer scales, physiological and disturbance effects were important and acted to modify the regional trend at these scales. These results suggest that variation in abrupt treeline elevations
are due to both broad-scale temperature-based growth limitation processes and finer-scale stress- and disturbance-related effects on seedling establishment.
Third, I explored the applicability of a meso-scale atmospheric model, The Air
Pollution Model (TAPM), for generating 200 m resolution, hourly topoclimatic data for
temperature, incoming and outgoing radiation, relative humidity, and wind speeds. Initial assessments of TAPM outputs against data from two climate station locations over seven years showed that the model could generate predictions with a consistent level of accuracy for both sites, and which agreed with other evaluations in the literature. TAPM was then used to generate data at 28, 7x7 km Nothofagus treeline zones across New Zealand for January
(summer) and July (winter) 2002. Using mixed-effects linear models, I determined that both
site-level factors (mean growing season temperature, mountain mass, precipitation,
earthquake intensity) and local-level landform (slope and convexity) and topoclimatic factors (solar radiation, photoinhibition index, frost index, desiccation index) were influential in
explaining variation in treeline elevation within and among these sites. Treelines were
generally closer to their site-level maxima in regions with higher mean growing season
temperatures, larger mountains, and lower levels of precipitation. Within sites, higher
treelines were associated with higher solar radiation, and lower photoinhibition and
desiccation index values, in January, and lower desiccation index values in July. Higher treelines were also significantly associated with steeper, more convex landforms.
Overall, this thesis shows that investigating treelines across extensive areas at multiple study scales enables the development of a more comprehensive understanding of treeline variability and underlying environmental constraints. These results can be used to formulate new hypotheses regarding the mechanisms driving treeline formation and to guide the optimal choice of field sites at which to test these hypotheses.
