The New Zealand Environment: Influences of Bluespace on Health Rosie Cooper GEOG420 Research Project ABSTRACT In the field of human geography, the relationship between water and human health and wellbeing has been relatively understudied. Interest in the health effects of the outdoors, such as greenspace, has grown rapidly in recent years and this is the first New Zealand study to focus on the salutogenic influences of exposure to bluespace. This dissertation aimed to investigate whether a relationship was present between proximity to bluespace and health throughout New Zealand. All-cause mortality rates were used as a broad, reliable data-set for overall health, and were compared against the straight-line proximity to any area of general bluespace and three bluespace types; coastline, lakes, and wide rivers. Results indicated that a statistically significant positive relationship exists between mortality and coastline proximity, suggesting, somewhat surprisingly, that mortality rates are higher closer to the coast. A statistically significant negative relationship exists between mortality and wide river proximity, indicating that as the distance from wide rivers is decreased the mortality rate is reduced. Similarly, a statistically significant negative relationship existed for general bluespace. No statistically significant relationship was found to exist between mortality and lake proximity, signifying that proximity to lakes does not have an effect on mortality. Some of these findings were contrary to international research, suggesting that bluespace and health relationships may differ according to national and environmental context. The varied findings highlight the strong need for future research in this field, which could ultimately help to shape public health initiatives and improve health. 2 Table of Contents Introduction ............................................................................................................................. 3 Bluespace and Health............................................................................................................. 5 Bluespace ........................................................................................................................................... 5 Perception and Preference ............................................................................................................. 5 Emotional Benefits........................................................................................................................... 7 Recreational Benefits ...................................................................................................................... 8 Direct Health Benefits..................................................................................................................... 9 Greenspace ........................................................................................................................................ 9 Research Question ............................................................................................................... 11 Null Hypotheses..............................................................................................................................11 Methods .................................................................................................................................. 12 Results .................................................................................................................................... 16 Lake Proximity ...............................................................................................................................17 Wide River Proximity ...................................................................................................................18 Coastal Proximity ..........................................................................................................................19 Bluespace Proximity ......................................................................................................................20 Discussion .............................................................................................................................. 23 Conclusion ............................................................................................................................. 30 Acknowledgements .............................................................................................................. 31 References .............................................................................................................................. 32 Appendices ........................................................................................................................... 38 Appendix A .....................................................................................................................................38 Appendix B .....................................................................................................................................39 Appendix C .....................................................................................................................................40 3 Introduction There is ongoing and increasing interest in health inequalities (Ministry of Health, 2002; United Nations, 2009; World Health Organisation, 1998). Some of geography’s earliest engagement with wellbeing was thirty years ago, with the work of Antoine Bailly (1981), who expressed many of the problems that are still significant today (Fleuret & Atkinson, 2007). The concept of wellbeing was somewhat understudied in geography, however recent interest has grown, particularly through the fields of public health, epidemiology and psychology, on the influential salutogenic properties of social and physical environments (Mitchell & Popham, 2008). The increased prominence of wellbeing in policy agendas and public health debates has begun to reemphasise the health benefits of nature, and academic disciplines such as landscape architecture, environmental psychology and health geography have had an extended interest in the remedial effects and conceptualisations of nature (Park et al., 2010). This research aims to identify the relationship between bluespace and health, within a New Zealand context. The influence of landscape on health has been studied fairly extensively (Fumkin, 2001; Maller et al., 2006), therapeutic landscapes provide social, physical and spiritual environments (Volker & Kistemann, 2011). Water is one of the most important aesthetic landscape elements (Kaplan and Kaplan, 1989) and an attractive landscape offers health and wellbeing for humans (Abraham et al., 2010). The relationship between water and health has been the subject of some interest in recent decades; for example it was addressed at an international healthcare conference in 1978 in Alma Ata (WHO, 1978). Subsequently, the Ottawa Charter for Health Promotion of 1986 identified that health encompassed social and personal resources, as well as physical capacities. The Charter was developed under the influence of Antonosvsky’s concept of salutogenesis, which “considers individual and corporate resources for health, wellbeing and quality of life as central requirements to prevent health risks and potential illnesses” (Volker & Kistemann, 2011: 450). The salutogenetic concept proposed by Antonosvsky was supported in the European Charter on Environment and 4 Health (1989) and the United Nations Conference for Environment and Development (1992). Many broader studies have shown that exposure to natural environments, such as greenspace, has influential effects on health and health-related behaviours (Mitchell & Popham, 2008). However, the relationship between bluespace and health is an area of science that has sparked some recent interest (Volker & Kistemann, 2011; Wheeler et al., 2012; White et al., 2010). Limited previous research has been carried out in this field, resulting in a restricted and undeveloped depth of knowledge at present. As a result of this, greenspace literature focusing on wellbeing has been used as a framework to supplement and develop this research project. Greenspace is an area of human geography that has been extensively studied and a wide variety of literature is available in this field. The psychological and physical benefits of greenspace have been proven, leading to curiosity about the potential similar effects that bluespace may possess. A renewed interest on place as a core theoretical and policy relevant determinant of health has occurred, and area level intervention has become a key component of strategies to reduce health inequalities (Pearce et al., 2006). As a result of recent interest and limited information in this field, this dissertation aims to investigate the effect of bluespace proximity on mortality. Previous bluespace and health literature will be reviewed, with a specialised focus on perception and preference of bluespace, emotional, recreational and direct health benefits of bluespace, and the parallel relationships between greenspace and bluespace literature. For the purpose of this research the classification of health is broad, it includes mortality through to a broadened definition of wellbeing. Wellbeing is a multi dimensional, dynamic term that varies over time, shaped by the discoursive construction of policy makers and society (Volker & Kistemann, 2011). It is a “complex measurable state of consciousness, which contains components like the habitual, actual, individual and social wellbeing” (Volker & Kistemann, 2011: 450). This research will refer to health as “a state of complete physical, mental and social wellbeing and not merely the absence of disease or infirmity” in following with the World Health Organisation (1948). 5 Bluespace and Health Bluespace For the purpose of this dissertation, bluespace is defined as significant areas of natural water, including lakes, wide-river sections and coastline. Water is considered to be one of the most important aesthetic elements of a landscape (Kaplan & Kaplan, 1989) and the ocean is a fundamental element of the Earth, covering more than seventy percent of its surface (Brand, 2007). Globally there is a phenomenal relationship between human beings and the ocean, as twenty-three of the worlds largest cities are located on the coast and over a third of the world’s population live along a slim fringe of coastal land (Wheeler et al., 2012). In New Zealand the effects of bluespace is of particular importance as nearly two thirds of the population live within five kilometres of the ocean (Richardson et al., 2010). Perception and Preference Perceptions towards bluespace are expressed using one’s senses to produce feelings and attitudes; usually water represents a natural resource vital for consumption by human beings (Volker & Kistemann, 2011). The perceptions of bluespace vary from this general perception, instead focusing on positivity, fascination and attractiveness (Burmil et al., 1999). Important aspects of the sensual perception of bluespace are the colour, clarity, sound, motion and context of water (Volker & Kistemann, 2011). Aesthetic perceptions of water colour and clarity in New Zealand were researched by Smith et al. (1995), who concluded that blue waters are preferred to yellow1 waters, but yellow waters may be acceptable if they were perceived as ‘natural’. It was also concluded that water appearance and bathing activity are closely linked, and preference towards a particular site of water “was strongly related to their perception of visual clarity, but less strongly related to actual clarity” (Smith et al., 1995: 43). People admire the motion of water and resultant sounds, and the special nature of these sounds. Sound produced by water ranges from calm low velocity flows, to vigorous, roaring sounds, and the variety of these flows is of great importance (Burmil et al., 1999; White et al., 2010). From an evolutionary perspective, humans attracted to 1 Water with a yellow/brown colouration, this can be a result of natural effects such as staining from algae and humic staining from native forests in river catchments. 6 aquatic environments were more likely to survive because of the fresh water availability, food, migration corridors and omega 3 fatty acids which aid brain development and evolution (White et al., 2010). As a result of this humans have evolved towards the reverence of water and preferences for aquatic environments. Investigation into the effects of bluespace is an important area of research as these natural areas of bluespace are associated with higher preferences and positive subjective reactions (White et al., 2010). The context of bluespace is also important in determining human perception. Visual rating of the water is increased if the area is connected with naturalness (Smith et al., 1995), nature and scenery enhance the experience gained and creates a more positive perception. Cleanliness and refreshment are often associated with water, and these adjectives conjure emotions that relate to regained energy, youth and health (Herzog, 1985; Burmil et al., 1999). The combined experience of bluespace and surrounding natural scenery is considered to be “peaceful, traditional, worth-preserving and preferable” (Yamashita, 2002: 9). Perception factors are closely correlated with preference for living by bluespace (Asakawa et al., 2004; Volker & Kistemann, 2011), and the presence of water features in residential areas were perceived as the most influential environmental attribute in eight towns in the Netherlands (Luttik, 2000). The value of a residential property is influenced by both its physical and locational characteristics, “the analysis of views should be placed in the wider context of the analysis of the impact of externalities on property values“ (Bourassa, 2004: 1427). Historically, views were desirable for strategic reasons; a dominating location allowed for the owner to be aware of possible intruders. Today, views are sought for completely different reasons, with the majority being aesthetic. Appleton (1975) explores the ‘prospect-refuge’ theory, suggesting that humans are biologically programmed to have a preference for vantage points where it is possible to see a large amount, but without necessarily being seen. Views are often associated with an ease of access to nature, but the two effects need to be clearly separated. Views are difficult to measure, however proximity of access to nature is easier to identify (Bourassa, 2004). When considering a view, the type of view, scope of the view, and distance to the water in the case of a water view should be taken into account. Since the 1980s urban planning has produced a visible trend for the preference of waterfront revitalisation, allowing access to bluespace, often at an expensive price (Volker & Kistemann, 2011). Bourassa conducted a study of house 7 prices and views, for the Auckland region in 1996, concluding that although water views have a strong positive impact on house values for houses near the coast, such views are not the only type of aesthetic externality that is priced in the residential property market. “Particularly attractive surrounding improvements add an average premium of 27% to values, and structures of superior quality in the neighborhood contribute an average premium of 37%” (Bourassa, 2004: 22). Increased house prices bordering coastlines, rivers and lakes highlight the preference for living near areas of bluespace. In the city of Emmen, in the Netherlands, house prices are ten percent higher when water views are present (Luttik, 2000). Emotional Benefits Environment-behaviour research “examines the relationship between social and behavioural patterns and the physical environment” (Ataov, 1998: 240) and the field of environmental aesthetics explores the human reactions to the visual qualities of an environment. The aesthetics of an environment “evoke feelings such as pleasure, relaxation, excitement, and fear” (Ataov, 1998: 239). Appearances of neighbourhoods and the surrounding built environments influence our emotional reactions and affective responses. Each person creates unique evaluative responses to an environment because factors such as personality, affective state, intentions and socio-cultural experiences influence individual responses (Ataov, 1998). People produce emotions about an environment based on their past experiences with the area, enabling them to identify specific visual features or attributes in an environment that create pleasant feelings and distinguish these from features that produce negative feelings (Ataov, 1998). Water can provide a strong sense of place, emotional feelings and attachment that can in turn influence wellbeing. Water symbolises purity and is expressed through human mental and spiritual life (Volker & Kistemann, 2011). The mental association and immersion with waterscapes can be explained using several aspects such as tranquillity, attention, interest, fascination or compatibility (Herzog & Bosley, 1992; Ulrich, 1981; White et al., 2010). In particular, studies have highlighted the strong emotional influence of water, providing greater positive influences on emotional states in comparison to other environments (Felsten, 2009). Individuals in a relaxed or happy emotional state prefer water environments, compared to those that are stressed (Regan & Horn, 2005), signifying that water is a favourite place for recreational activities and to expend spare 8 time, in addition to the benefits gained from reducing stress and providing restorative effects (Volker & Kistemann, 2011). Recreational Benefits Nature helps to stimulate us to be more active, and water can provide the basis for recreation. Coastal areas, beaches and inland water bodies are particularly important in inspiring recreational activity (Depledge, 2009). Regular contact with these environments can diminish health inequalities by offering three major health benefits; reduced stress levels, increased physical activity and a greater sense of community (Depledge, 2009). Stress is a process by which “an individual responds psychologically, physiologically, and often with behaviours, to a situation that challenges or threatens wellbeing” (Ulrich et al., 1999:202). Human recuperation has been suggested to be faster and more complete when individuals are exposed to natural environments, rather than urban environments, and stress levels are also reduced when exposed to natural environments compared to urban environments (Ulrich et al., 1999). Recreational activity around bluespace is more common when waterfront activities are aided by access routes to encourage movement, and other facilities are present, such as benches to provide small respites (Volker & Kistemann, 2011). Recreational experiences can be separated into four catergories; “kinetic recreational experiences, situation-based recreational experiences, harvest experiences, and substitution or aesthetic experiences” (Volker & Kistemann, 2011: 455). Kinetic recreational experiences include activities with a high degree of motion on the water, such as sailing, canoeing and jet–skiing, or activities on the water edge including jogging and cycling (Yamashita, 2002). Situation based recreational experiences “refer to one location at the water, which is visited several times for experiences such as swimming, playing in the water, social interactions, or walking” (Volker & Kistemann, 2011: 455). Harvest recreational experiences describe activities such as fishing. Contemplative or aesthetic recreational experiences include the perception of the views and sounds of the bluespace, and passive exploration of the water (Smith et al., 1995; White et al. 2010). Kinetic and situation-based recreational experiences are well known for their positive health benefits, including the prevention of cardiovascular illnesses, obesity and cancer (Bell et al., 2008), as well as anxiety and depression. Recreational benefits of harvest and contemplative experiences are often immeasurable, as they are experienced rather than scientifically measurable. 9 Direct Health Benefits Bluespace has been recognised as providing direct health benefits (Volker & Kistemann, 2011), the majority of this has been through research of therapeutic landscapes. The concept of therapeutic landscapes provides a framework to assess the health benefits of blue space. It identifies the salutogenic health effects of a landscape, covering all elements of the landscape. However it has been noted that a therapeutic landscape is not necessarily beneficial to all individuals, and should be considered instead as a ‘potentially therapeutic landscape’ (Conradson, 2005; Volker & Kistemann, 2011). The restorative scenery of Lourdes, with its flowing spring, is described as a remedial core, and water is perceived to be responsible for cures and is dedicated to healing (Gesler, 1996). The holy wells in Ireland are described as “sites of indigenous health” and “a piece of micro-landscape of healing and wellbeing” (Foley, 2011:477), children are bathed in the wells as a representation for health promotion in early life (Volker & Kistemann, 2011). Other research has stated the relationship between water and health, with views of water being potentially beneficial for health (Burmil, 1999) and water bodies being correlated with a high quality of life (Ogunseitan, 2005). Greenspace The influence of greenspace on health has been studied extensively and there may be parallel relationships between greenspace and health, and the influence of bluespace on health. Previous literature has suggested that socioeconomic status may be a determinant of greenspace proximity. Greenspace literature has shown a relationship between proximity to greenspace and health (Groenewegen et al., 2006; Richardson et al., 2010). Groenewegen et al. (2006) examined the effects of greenspace in the Netherlands and concluded that individuals of a lower socioeconomic status are less able to move to areas that have higher proportions of greenspace. This was similar to the studies of Richardson et al. (2010: 2) who were concerned that “locational access to health promoting community resources, such as greenspace, is lower in socioeconomically deprived areas”, which could be a contributing factor to lower levels of physical activity in deprived communities. Richardson et al. (2010) found that in New Zealand total greenspace availability fell with increasing socioeconomic deprivation. 10 Many studies found that there was a positive association between the size of the greenspace and health (Guite et al., 2006; Maas et al., 2006; Mitchell et al., 2007). Mitchell et al. (2007: 681) concluded that a “higher proportion of greenspace in an area is associated with better health” although they identified that the association depends on the degree of urbanity and level of social deprivation in an area. They also noted that the quality and quantity of greenspace was significant in determining health benefits. The strength of this relationship was tested in the Netherlands by Maas et al. (2006), their findings show that the percentage of greenspace in an area is correlated with a higher level of general health for individuals in the region. Numerous other studies have found a link between the size of greenspace and health, such as Guite et al. (2006: 1118) who concluded that there are “modest cross sectional associations between elements of the physical environment and psychological health”. After classifying the England population into groups according to income deprivation and exposure to greenspace, Mitchell and Popham (2008) found that those living in the greenest regions had an overall higher level of health. A contrasting article by Richardson et al. (2010: 1) found that, contrary to expectations, there was “no evidence that greenspace influenced cardiovascular disease mortality in New Zealand, suggesting that greenspace and health relationships may vary according to national, societal or environmental context”. They also concluded that greenspace variation might have lesser relevance for the health of individuals in New Zealand as greenspace is abundant and there is less social and spatial variation in its availability, compared to other contexts. The information extracted from previous literature can form the basis of research on bluespace and health. 11 Research Question 1. Within a New Zealand context, does the proximity of bluespace influence all cause mortality? Null Hypotheses 1) A statistically significant relationship between proximity to lakes and all cause mortality does not exist. 2) A statistically significant relationship between proximity to wide rivers and all cause mortality does not exist. 3) A statistically significant relationship between proximity to the coastline and all cause mortality does not exist. 4) A statistically significant relationship between proximity to any type of bluespace and all cause mortality does not exist. 12 Methods A qualitative analysis of existing studies relevant to bluespace and health was carried out. In following with the methods used by Volker and Kistemann (2011), a critical realist approach was used to integrate qualitative and quantitative studies. The approach “states that a mechanism cannot be disproved by the identification of a missing recognition” (Volker & Kistemann, 2011: 450). To identify relevant articles, electronic databases (including PubMed, Web of Science and Science Direct), relevant journals, reference lists of earlier research papers, and relevant organisations were searched. Keyword and phrase searches related to health, wellbeing, bluespace and landscapes. Different words for bluespace were used in order to find maximum results; water, rivers, lakes, aquatic environments, coast and bluespace. After gathering selected studies, the articles and papers were filtered to exclude papers covering irrelevant topics, such as those covering the pathogenetic impact of water on individual human health, and those dealing with water policy and nature conservation. The relevant articles were read, and important information was extracted and used to build this research. Analysis with a Geographic Information System (GIS) was carried out to find the proximity of bluespace in relation to place of residence. The adoption of a GIS framework enabled the integration of data that had been collected in different ways, and the use of a GIS allowed bluespace proximity for different areas to be calculated (Pearce et al. 2006). To prepare a dataset for analysis, age-sex-standardised mortality rates2 were calculated for each census area unit3 (CAU) in New Zealand, resulting in 2 Age-sex standardised mortality rates are a method of adjusting the crude rate to eliminate the effect of differences in population age and gender structures when comparing crude rates for different geographic regions and different population sub-groups. The adjustments were undertaken for the comparison of populations against a standard population. 3 Census area units are aggregations of meshblocks, they are non-administrative areas that define, or aggregate to define, regional councils, territorial authorities, urban areas and statistical areas. Census area units normally contain a population of about 3,000-5,000 people and usually coincide with suburbs (Statistics New Zealand, 2006a). 13 36 categories4 per CAU. New Zealand 2005-2007 mortality data and 2006 Census population counts (Statistics New Zealand, 2006a) were used to achieve this. Areas of New Zealand coastline, wide rivers and lakes were retrieved online5 (koordinates.com, 2010), and imported into ArcMap10. Coastline was defined as the ‘mean high water mark around North Island, South Island and Stewart Island’, wide rivers were classified as ‘a natural, flowing body of water emptying into an ocean, lake or other body of water and usually fed along it's course by converging tributaries’, and lakes were defined as ‘any standing body of fresh inland water” (koordinates.com, 2010). The ‘wide river’ data set was selected to represent significant areas of bluespace that are highly visible and more likely to be usable, compared to the ‘all river and stream centrelines’ data set which included many minute rivers and streams that are barely visible and insignificant for this research. CAUs (koordinates.com, 2010) were the identifying scale of mortalities, and in accordance, CAU centroids were used to provide an average data point location for each CAU. Using the GIS Software ArcMap10, CAU centroid points were laid over bluespace shapefiles and the nearest distance to each type of bluespace (coastline, wide rivers and lakes) was calculated using the ‘near’ proximity tool. The closest distance from each centroid to any type of bluespace was also calculated. The data was exported to an Excel workbook and imported into Statistics Software SPSS, allowing for the relationships between mortality and proximity to bluespace to be investigated. Graphs were produced in Microsoft Excel to identify the functions of the age and deprivation data, as a linear regression model is not always appropriate to represent the spread of data. Scatter graphs of age and sex data versus mortality rates were formed, allowing the best model (linear, functional, exponential or quadratic) for each variable to be found. For both models the quadratic model was the strongest model to represent the data. The quadratic models for age versus log-mortality and deprivation versus logmortality both had a statistically significant p-value (p<0.001), indicating that they were the correct models to describe and explain the data relationships. This was 4 Each category represented a five-year age group and gender. For example, category 1 was males aged 0-4 years, category 2 was females aged 0-4 years, category 3 was males aged 5-9 years, category 4 was females aged 5-9 years, and so on. 5 The koordinates.com shapefiles were derived from the Land Information New Zealand 1:50,000 New Zealand Topographic database. 14 predictable, as mortality rates are higher in newborns and significantly higher for the elderly. Mortality rates are also higher amongst deprived communities due to a lack of expenditure on healthcare, less emphasis on a healthy lifestyle and reduced living conditions, and mortality rates are also higher in the most affluent areas as a result of conditions such as obesity leading to cardiovascular disease. To account for the quadratic curve, the age and deprivation data was squared to become quadratic, and produced two new columns in the dataset. Diagnostic tests were performed to test if the quadratic data models chosen were correct. To test for model misspecification Cook’s distance test, the residual test and the predicted value of response test were performed. The Cook’s distance tests did not indicate any unusual values as the values ranged from between 0 and 0.002 (Appendix A). The values are all less than one signifying that none of the data points have an undue influence on the fit of the quadratic model, and therefore the quadratic model is appropriate. To test for residual autocorrelation and normality, the residuals were plotted using Q-Q Plots in SPSS. They formed a near straight line (Appendix B), indicating that my choice of quadratic model was appropriate to represent the data. Residual values were then plotted against mean predicted values, using curve estimation in SPSS. This produced a clustered cloud of data points, symmetrical around the zero line (Appendix C), indicating that the quadratic model chosen was appropriate and that the variance of the residuals is constant. The mortality rate was then logged, using the logit function in SPSS in order to transform the variable so it can fit to a straight line; the new variable was inserted as a new column in the dataset. Once the quadratic models were fitted, a generalized linear regression model was used to find out if the data was significant, and socioeconomic status was controlled for with New Zealand Deprivation Index data6. The response factor was mortality rate, and covariate factors included age, deprivation, gender and distance to bluespace. The 6 The New Zealand Deprivation index is based on ten categories and ranges on a scale from 1 – 10, where 1 represents the areas with the least deprived scores and 10 the areas with the most deprived scores. The scores include nine variables; people on the benefit, people living below an income threshold, people not living in their own homes, people in a single parent family, people that are unemployed, people without qualifications, people living in households below a bedroom occupancy threshold, people with no access to a telephone and people with no access to a car. It divides New Zealand into tenths of the distribution of the first principal component scores and is based on areas rather than individual people (Ministry of Health, 2007). 15 regression was carried out four times, using proximity to rivers, lakes, coastline, and any bluespace, as the four differing factors. The confidence levels used were 95%, so if the p-value was less than 0.05, the null hypothesis is rejected and the findings are statistically significant. 16 Results The results showed that 73% of New Zealanders live within 10 km of the coastline, highlighting the close proximity and abundance of bluespace in New Zealand, with the mean distance to bluespace 2.4 km (Table 1). The minimum distances of all bluespace types was zero, or extremely close to zero. This is because some CAUs had lakes, wide rivers or coastline running through the centroid, resulting in a distance of zero. The maximum distance from a census area centroid to the coastline is Dunstan, Central Otago, with a distance of 116.4 km. The greatest distance from a census area centroid to a lake is White Island, with a distance of 50.0 km. However, the population of White Island is zero, so the greatest distance of a mainland census area unit centroid to a lake is Te Kaha, in the Opotiki District, with a distance of 18.4 km. The greatest distance from a census area centroid to a wide river is again White Island, with a distance of 49.8 km. However, the greatest mainland distance to a wide river is the Kaikoura Township, with a distance of 28.9 km7. The standard deviations of all bluespace types, with the exception of coastline, were low (Table 1), indicating that the data points were all relatively close to the mean. The high standard deviation of coastline proximity indicates that the data points are spread out over a large range of values. This signifies that the coastal proximity data had much greater variation than other bluespace measures as a result of the coastline bordering New Zealand. Comparatively, lakes and wide rivers have a more even distribution across New Zealand. This is not surprising, as coastline is only on the coast but rivers and lakes are located everywhere. 7 Even though Kaikoura has several small rivers located in the region, these were not classified in my study as ‘wide rivers’ and were not included in the study. 17 Table 1: Descriptive statistics of bluespace proximity data by bluespace type. The minimum mortality rate was 0.00, and was a result of CAUs with zero population (inlets, bays, islands, harbours, marinas, ports and estuaries). The highest mortality rate was in Strowan, Christchurch, for males aged 85-89. There was a population count of 20 in this category, and the number of deaths for this age-sex group from 2005 – 2007 was 59, resulting in an average deaths per year of 19.67 (Statistics New Zealand, 2006a). The mortality rate was therefore extremely high for this group, 98%, when calculated with these data sets. Lake Proximity The results showed that nearly one third (32%) of New Zealanders live within one kilometre of a lake. This is a very close proximity for many New Zealanders, and the data set representing lakes from koordinates.com can be partly responsible for this. Any standing body of fresh inland water was mapped, resulting in a large amount of minor lakes throughout the country being represented, and in turn influencing the lake proximity data. The results of the model fitting (Table 2) indicate that the proximity to lakes was associated with lower mortality (log-odds ratio of -0.003), but the effect was not statistically significant (p=0.389). Therefore the null hypothesis for the relationship between lake proximity and wellbeing cannot be rejected, and it can be concluded that proximity to lakes does not have a significant association with mortality. This suggests that no relationship does exist between residential proximity to lakes and mortality. 18 Table 2: Generalised Linear Regression Model of lake proximity. Parameter Estimates Parameter B Std. 95% Wald Confidence Error Interval Lower Upper Hypothesis Test Wald Chi- df Sig. Square (Intercept) -4.81 0.0241 -4.858 -4.763 39759.585 1 0.000 age -0.088 0.0038 -0.095 -0.08 525.616 1 0.000 nzdep 0.131 0.0073 0.117 0.145 323.914 1 0.000 sex 0.76 0.0093 0.742 0.778 6664.175 1 0.000 age2 0.01 0.0002 0.01 0.011 2718.734 1 0.000 nzdep2 -0.011 0.0006 -0.012 -0.01 300.352 1 0.000 lake -0.003 0.0032 -0.009 0.004 0.742 1 0.389 (Scale) 1.257a 0.0074 1.242 1.271 Wide River Proximity The results showed that approximately one sixth (16%) of New Zealanders live less than one kilometre away from a wide river. The generalised linear model for river proximity produced a statistically significant result (Table 3). Proximity to rivers was associated with lower mortality (log-odds ratio of -0.027), and the effect was statistically significant (p<0.001). The null hypothesis can be rejected, and it can be concluded that an increased proximity to rivers has a significant negative association with mortality. By reducing the distance lived from a wide river, there is an effect of a lowered mortality rate; this indicates that a relationship does exist between residential proximity to wide rivers and mortality. Therefore, for every additional kilometre towards a wide river, the odds of mortality are decreased by 2.7%. 19 Table 3: Generalised Linear Regression Model of wide river proximity. Parameter Estimates Parameter B Std. 95% Wald Error Confidence Interval Lower Upper Hypothesis Test Wald Chi- df Sig. Square (Intercept) -4.672 0.0244 -4.72 -4.624 36732.715 1 0.000 age -0.088 0.0038 -0.095 -0.08 531.051 1 0.000 nzdep 0.125 0.0073 0.111 0.14 298.161 1 0.000 sex 0.76 0.0093 0.742 0.778 6714.009 1 0.000 age2 0.01 0.0002 0.01 0.011 2742.647 1 0.000 nzdep2 -0.011 0.0006 -0.012 -0.01 300.018 1 0.000 wide river -0.027 0.0013 -0.029 -0.024 440.697 1 0.000 (Scale) 1.247a 0.0073 1.233 1.262 Coastal Proximity Just over a quarter (28%) of New Zealanders live within one kilometre of the coast. The generalised linear model for coastal proximity produced a statistically significant effect (Table 4). Proximity to the coast was positively associated with higher mortality (log-odds ratio of 0.002), and the effect was statistically significant (p<0.001). The null hypothesis can be rejected, and we can therefore conclude that an increased proximity to the coast has a significant positive association with mortality. By reducing the distance lived from the coastline, there is an effect of an increased mortality rate; this signifies that there is an evident relationship between coastal proximity and mortality. Therefore, for every additional kilometre towards the coast, the odds of mortality are increased by 0.2%. 20 Table 4: Generalised Linear Regression Model of coastal proximity. Parameter Estimates Parameter B Std. 95% Wald Error Confidence Interval Lower Upper Hypothesis Test Wald Chi- df Sig. Square (Intercept) -4.836 0.0236 -4.883 -4.79 41845.109 1 0.000 age -0.088 0.0038 -0.096 -0.081 529.797 1 0.000 nzdep 0.129 0.0073 0.115 0.144 314.167 1 0.000 sex 0.76 0.0093 0.742 0.778 6673.763 1 0.000 age2 0.01 0.0002 0.01 0.011 2733.169 1 0.000 nzdep2 -0.011 0.0006 -0.012 -0.01 288.99 1 0.000 coast 0.002 0.0002 0.001 0.002 62.731 1 0.000 (Scale) 1.255a 0.0074 1.241 1.27 Bluespace Proximity The generalised linear regression model produced a statistically significant output (Table 5) for overall bluespace proximity. Proximity to overall bluespace was associated with lower mortality (log-odds ratio of -0.015), and the effect was statistically significant (p<0.001). The null hypothesis can be rejected and it can be concluded that increased proximity to overall bluespace has a statistically significant negative association with mortality. By reducing the distance lived from general bluespace, there is an effect of a lowered mortality rate; this indicates that a relationship does exist between general bluespace proximity and mortality. Therefore, for every additional kilometre towards general bluespace, the odds of mortality are decreased by 1.5%. 21 Table 5: Generalised Linear Regression Model of overall bluespace proximity. Parameter Estimates Parameter B Std. 95% Wald Error Confidence Interval Lower Upper Hypothesis Test Wald Chi- df Sig. Square (Intercept) -4.776 0.0238 -4.823 -4.729 40167.556 1 0.000 ageid -0.088 0.0038 -0.095 -0.08 523.893 1 0.000 nzdep 0.131 0.0073 0.117 0.146 325.375 1 0.000 22 sex2 0.76 0.0093 0.742 0.778 6673.595 1 0.000 age2 0.01 0.0002 0.01 0.011 2714.079 1 0.000 Nzdep -0.011 0.0006 -0.013 -0.01 308.137 1 0.000 Bluespace -0.015 0.0016 -0.019 -0.012 92.272 1 0.000 (Scale) 1.255a 0.0074 1.24 1.269 Overall, the results showed that the influence of general bluespace, wide rivers and coastline proximity has a statistically significant association with health. The coastal association was positive, indicating that people who live closer to the coast, on average, have an increased effect of mortality than those who live further away. The general bluespace and river proximity was negative, indicating that people who live closer to general bluespace and wide rivers, on average, have a decreased effect of mortality than those who live further away. The effects were all relatively small, but a significant relationship was evident. The proximity of lakes had no significant association with health, and no effect was present. 23 Discussion Bluespace plays an important role in the environment, influencing landscape perception, preference and design. It is a major challenge to understand the diverse individual perspectives of experiencing bluespace. This study adds to a small, but growing research area, and largely agrees with previous research by confirming that characteristics of the natural environment are related to wellbeing, regardless of age, sex and deprivation. The findings showed that living nearer to the coastline is associated with a very small increase in mortality, despite controlling for socioeconomic status. This was surprising, but after reflecting on other New Zealand research this seemed feasible. New Zealand differs from other countries, in that the waterfront properties are not always the most affluent areas. This unusual pattern between socioeconomic measures and location is evident through local knowledge and supported by examining the location based census information (Figure 1). In Christchurch the annual median household income is NZ$$48,200, with a maximum median household income of $98,500 in Holmwood, and a minimum of NZ$31,200 in Linwood8. Households in the eastern and southeastern areas of Christchurch, and west along major arterial road corridors, generally have lower incomes. In contrast households around the Port Hills and to the northwest region of Christchurch have higher incomes. The waterfront areas9 have a wide distribution of median incomes (Figure 1); for example, New Brighton has a low median household income of NZ$34,200 and Waimairi Beach has a high median household income of NZ$68,200. 8 The lowest median household income was NZ$30,600 in Riccarton West, although this area is dominated by a high percentage of low income students. For the purpose of this study student dominated areas were considered anomalies, and as a result of this Linwood was regarded as the lowest median household income area in Christchurch. 9 Statistical areas where majority of the census area unit lies within 1km of the coastline. 24 Figure 1: Choropleth Map Showing Median Household Income per CAU, Christchurch, New Zealand 2006. 25 The varied socioeconomic pattern in Christchurch is mirrored further north in Wellington City. Wellington City has an average median household income of NZ$74,200, notably higher than Christchurch. Waterfront areas have a wide range of median household incomes, with a minimum median household income of $31,700 in Haitaitai, to a maximum of $100,000 in Eastbourne. These income patterns are also evident throughout Auckland. Overseas, in countries such as Australia, waterfront areas are considered to be highly desirable locations. In the area of Bellingham, Western Australia, a full view of the ocean could increase house prices by almost 60% (Benson et al., 1998). In Sydney, Australia, the median weekly household income is A$1,447 (Australian Bureau of Statistics, 2011). The waterfront areas10 all have relatively high median weekly household incomes; Balmain has a median household weekly income of A$2,544, Lindfield has a median household weekly income of A$2,558, and Rose Bay has a median weekly income of A$2,524. These median weekly incomes are all much higher than the Sydney average, indicating that waterfront areas are affluent and sought-after locations. Only 24.6% of Sydney’s coastal suburbs have a median household income less than the Greater Sydney median (Australian Bureau of Statistics, 2011), the majority of these lower income coastal suburbs are located towards Botany Bay11. In New Zealand this percent is significantly greater, with 42.9% of Christchurch coastal CAUs having a lower median household income than the Christchurch City median (Statistics New Zealand, 2006b). This is also observed in Wellington City, with 40.0% of coastal CAUs below the city’s median, emphasizing the diverse socioeconomic coastal land patterns in New Zealand. There may be a possible number of reasons why the results from this research differ from international studies of bluespace. A lack of variation in exposure to bluespace for New Zealanders may be present, as bluespace is extremely abundant with 73% of New Zealanders living within ten kilometres of the ocean. One kilometre has been considered as a relatively easy walking or cycling distance (Richardson et al., 2010), and in New Zealand approximately one third of people live within this distance from a lake, one quarter from the coast and one sixth from a wide river section. The average 10 Statistical areas adjacent to the coastline. 11 An industrial area hosting an airport and container port. 26 New Zealander lives just 2.4 km from any bluespace type. These proximity figures contrast with other countries, for example, only 6.4% of the United States population live within one kilometre of the coast (Inter-American Institute, 2010). Another factor influencing the results could be the coastline to land-area ratio of New Zealand. The coastline to land-area ratio for New Zealand is 56.5m/km2, this ratio is comparatively larger than most other non-island countries. The United States has a coastline to land area ratio of 2.2m/km2, China has a ratio of 1.5m/km2, Australia has a ratio of 3.4m/km2 and all land-locked countries have a ratio of zero. The high ratio in New Zealand could be a contributing factor towards the low influence of coastline proximity as the coastline is highly accessible and therefore may not be a significant health determinant. Viewing New Zealand as an exception to international findings has also been suggested in New Zealand greenspace literature. Richardson et al. (2010) investigated the association between green space and cause-specific mortality in urban New Zealand. Their results found that contrary to expectations, there was no evidence that greenspace influenced cardiovascular mortality in New Zealand. As a result of this they concluded that greenspace and health relationships might vary according to national, societal or environmental contexts, and a lack of variation in exposure to greenspace was present in New Zealand. They suggested the idea that greenspace variation may have lesser relevance for New Zealand health issues, as greenspace is more abundant in New Zealand compared to other countries, and social and spatial variation is less varied compared to other contexts. Another national study by Witten at al. (2008) also drew upon the conclusion that greenspace and health relationships in New Zealand may differ from those found in other countries, as they found no relationship between greenspace and Body Mass Index. The descriptive findings for overall bluespace and wide rivers lends support to previous claims (Volkner & Kistemann, 2011; Wheeler et al., 2012) that bluespace in the physical environment may be an important influence upon individual wellbeing levels and has many positive influences on human health. The results indicated that living closer to general bluespace and wide rivers is associated with a decreased mortality rate. This study and others (Maas et al., 2006; Mitchell & Popham, 2008; Wheeler et al., 2012) indicate that access to natural blue and green environments may 27 play a part in reducing health inequalities. Evidence towards this concept is still relatively limited although the issue is starting to become acknowledged in health inequalities policy. Suggestions relating to greenspace and health have occurred at the 2010 Marmot Review for the UK Department of Health (Marmot, 2010) and greenspace and bluespace importance were discussed as part of the Christchurch City Council’s ‘Open Space Strategy’ as an area that can be used for recreation or public health benefit (Christchurch City Council, 2011). The results also indicated that living near lakes was not associated with an increased or decreased mortality. It was surprising that this result was not significant for health, as the proximity for wide rivers was statistically significant. The lake data set (koordinates.com) mapped all New Zealand lakes to a high level of detail, including many insignificant lakes with a small surface area. This resulted in most CAUs being extremely close to a lake, and 99.8% of New Zealanders living within 10 km of a lake. A possible consequence of the high detail mapping of small lakes could be that lake proximity had no effect on mortality. By only including significant lakes in future studies, for example over 1km2 surface area, lake proximity may have an affect on mortality rates. Many previous experimental studies in this discipline have used abstract images of reality, portraying contrasting environments to investigate the preference for and benefits of natural environments (Dramstad et al., 2006; Kaltenborn & Bjerke, 2002; Van den Berg et al., 2003; White et al., 2010). These studies had varied results, for example Van den Berg et al. (2003) found that the presence of water did not have a reliable influence on environmental preference or restoration, although in contrast with these findings, White et al. (2010) found that scenes containing water were associated with higher preferences, greater positive effect and higher perceived restorativeness ratings. An underlying factor for these mixed results could be due to the high variance of diversity and individual perceptions and preferences (Volkner & Kistemann, 2011). Experimental studies such as these only concentrate on photographs or videos that contain a special focus on the environment, so the individual perspective may disappear (Volkner & Kistemann, 2011). By experiencing an area of actual bluespace and the surrounding environment, an individual receives a three-dimensional view, and contributing sensory components of sound, smell and taste, whereas images only provide a two-dimensional view of the bluespace and lack additional sensory 28 constituents. Understanding a landscape “cannot be known or understood simply from publications, from maps, diagrams, photographs and descriptions, because these are only representations, there can be no substitute for the human experience of place – of being there” (Tilley, 2004: 218). The variety of research methods in this field is extremely diverse, leading to mixed results. For the effects of bluespace proximity to be understood at a greater depth further investigation is necessary. Despite some significant results showing that bluespace has positive influences on human health, research in this field is far from advanced. This study is subject to limitations. The cross sectional nature of the analysis means that migration is not accounted for. However, the findings from this research can still be used to build a hypothesis about the bluespace-health relationship. As identified by Wheeler et al. (2012), a ‘healthy migrant effect’ may have occurred, whereby the “healthiest (and wealthiest) proportion of the population are more mobile and potentially more able (physically and financially) to move towards the desirable environments including the coast” (Wheeler et al. 2012:3). An alternative hypothesis may also be drawn from this study; the coastal and river locations may be the preferred living environment for people who place a higher value on physical activity. It is also possible that the found associations between bluespace proximity and health may be as a result of residual confounding by one or more unaccounted population/area characteristic that is associated with bluespace and health. Components of the methodology may have influenced the relationship found between living closer to the coastline and the associated increased mortality rate. Coastline proximity was calculated by measuring the straight-line distance, although visibility and accessibility of the ocean was not taken into account. Many people live within one kilometre of the coastline, but this does not necessarily mean that they have a view of the ocean or can easily access the ocean. Views may be veiled as a result of elements of the natural and built environment, such as the topography of the land, urban developments, vegetation, and residential housing. Accessibility may be limited or complicated due to features of the land and network systems. For example, residents on a cliff may have a close proximity to the ocean but ocean accessibility is not direct, or alternatively residents may live within a close proximity to the ocean but accessibility is complicated as a result of the network systems available in the area. Chen & Jim (2010) found that the visibility of a bay view was perceived as an extremely desirable 29 attribute, and valued higher than accessibility and availability for use. A residential view of Shenzhen Bay, China, increased house prices by 11.2%, and house prices dropped by 0.7% with each increasing kilometre of distance away from the bay. Bluespace visibility could be analysed by using the viewshed analysis ArcMap tool, which determines the surface locations visible to an observer. The network-analyst ArcMap tool can dynamically model realistic network conditions whilst providing options for routing, travel directions, closest facility, service area, and locationallocation (ESRI, 2012). Incorporating the use of these tools into the methodology would provide valuable information, and may help to explain further the positive relationship between coastline proximity and mortality. If this study were continued, it would be valuable to use the New Zealand 2006/2007 Health Survey data12 to supplement the all cause mortality data. The Health Survey information would have provided data at all levels of impact, not just the extreme outcome of death. This would have allowed for preventative measures to be studied at a greater depth and would have stratified all impact levels. This would be useful in influencing a wide array of mental and physical public health policies, and providing information before the individual had passed away. It would have also been constructive to separate the all-cause mortality data into different categories, such as cardiovascular disease, suicide and cancerous diseases, to see if there were any significant relationships. This field of study would benefit from increased research into emotional and experiential response to bluespace, as this would assist the researchers and practitioners analysing health effects. 12 The Health Survey included questions such as whether the individual had received treatment for depression, bipolar, anxiety disorder, alcohol disorder and drug disorder. It also asks the individual to rate their self-reported general health, vitality, social functioning, role-emotional, and mental health. The survey also includes a New Zealand Index of Socioeconomic Deprivation for Individuals category (New Zealand 2006/07 Health Survey, 2012). 30 Conclusion In spite of the fact that water covers over two-thirds of the earth’s surface, solid conclusions regarding the relationship between bluespace and human health and wellbeing are still largely unknown. Initial research suggests that bluespace has a salutogenic effect on human health and wellbeing, although, due to a variety of research methodologies and diverse study locations, no conclusive conclusions can be drawn. The findings showed that being within close proximity to areas of general bluespace had a negative influence on mortality, for every additional kilometre towards general bluespace, the odds of mortality are decreased by 1.5%. A similar relationship was also found for wide rivers and mortality, as every additional kilometre towards a wide river, decreased the odds of mortality by 2.7%. The variation in the nature of the findings for the health effects of living within close proximity to the coastline and lakes further suggest the research evidence is inconclusive and that that further investigation is necessary. Being within close proximity to the coastline was found to have a positive influence on mortality, for every additional kilometre towards the coast the odds of mortality are increased by 0.2%. Lake proximity had no effect on mortality rates, however by using a data set that only included lakes over 1km2 , a different outcome may have been found. The results highlight the idea that bluespace in New Zealand may not be as important a determinant of health as has been demonstrated in other countries. As this field of research is built upon, it could provide valuable information for public health and urban planning policy initiatives. New initiatives would need to balance potential benefits of the close proximity to bluespace against threats from overdeveloping waterfront locations, extreme events and climate change impacts. By adhering to rational precautions, health inequalities may be reduced and there is potential to achieve major health improvements. 31 Acknowledgements I am sincerely and extremely grateful to my supervisor, Simon Kingham, for the endless encouragement and direction throughout the writing of my dissertation. I would also like to thank Christopher Bowie and Elena Moltchanova, for their guidance with the statistical analysis of my data. Furthermore, I would like to show my gratitude towards my parents, as this dissertation would also not have been possible without their ongoing support and enthusiasm. 32 References Abraham, A., Sommerhalder, K., Abel, T. (2010). Landscape and well-being: a scoping study on the health-promoting impact of outdoor environments. International Journal of Public Health, 55 (1), 59-69. Appleton, J. (1975). The experience of Landscape. London, UK: Wiley. Asakawa, S., Yoshida, K., Yabe, K. (2004). Perceptions of urban stream corridors within the greenway system of Sapporo. Japan Landscape Urban Planning. 68, 167–182. Ataov, A. (1998) Environmental Aesthetics. Journal of Planning Literature, 13 (2), 239-257. Australian Bureau of Statistics. (2011). 2011 Census Quick Stats. Retrieved from http://www.abs.gov.au/websitedbs/censushome.nsf/home/quickstats?opendocument &navpos=220 Bell, J., Wilson, J., Liu, G. (2008). Neighborhood greenness and 2-year changes in body mass index of children and youth. American Journal of Preventative Medicine. 35, 547-553. Benson, E., Hansen, J., Arthur, J., Schwartz, L., Smersh, G. (1998). Pricing residential amenities: the value of a view. Journal of Real Estate Finance and Economics. 16, 55-73. Bourassa, C., Hoesli, M., Sun, J. (2004). What's in a View? Environment and Planning A, 36 (8), 1427-50. Brand, D. (2007). Bluespace: a typological matrix for port cities. Urban Design International. 12, 69-85. Burmil, S., Daniel, T., Hetherington, J. (1999). Human values and perceptions of water in arid landscapes. Landscape Urban Planning. 44, 99–109. 33 Chen, W., Jim, C. (2010). Amenities and disamenities: a hedonic analysis of the heterogeneous urban landscape in Shenzhen (China). The Geographic Journal. 176, 227-240. Christchurch City Council. (2011) Open Space Strategy. Retrieved from http://www.ccc.govt.nz/thecouncil/policiesreportsstrategies/strategies/healthyenviron mentstrategies/openspacestrategy.aspx Conradson, D. (2005). Landscape, care and the relational self: therapeutic encounters in rural England. Health and Place. 11, 337–348. Depledge, M., Bird, W. (2009). The Blue Gym: Health and wellbeing from our coasts. Marine Pollution Bulletin. 8, 947–948. Dramstad, w., Tveit, M., Fjellstad, W., Fry, G. (2006). Relationships between visual landscape preferences and map-based indicators of landscape structure. Landscape Urban Planning. 78, 465-474. ESRI. (2012). ArcGIS Network Analyst. Retrieved from http://www.esri.com/software/arcgis/extensions/networkanalyst. Felsten, G. (2009). Where to take a study break on the college campus: an attention restoration theory perspective. Journal of Environmental Psychology. 29, 160–167. Foley, R. (2011). Performing health in place: the holy well as a therapeutic assemblage. Health and Place. 17, 470-479. Fleuret, S. & Atkinson, S. (2007). Wellbeing, health and geography: A critical review and research agenda. New Zealand Geographer. 63, 106-118. Frumkin, H. (2001). Beyond Toxicity: Human Health and the Natural Environment. American Journal of Preventative Medicine. 20, 234–240. Gesler, W. (1996). Lourdes: healing in a place of pilgrimage. Health and Place. 2, 95105. Groenewegen, P., van den Berg, A., de Vries, S., Verheij, R. (2006). ‘Vitamin G: effects of green space on health, well-being and social safety’. BioMed Central Public Health, 6 (149), 1-9. 34 Guite, H., Clark, C., Ackrill, G. (2006). The impact of the physical and urban environment on mental well-being. Public Health, 120 (12), 1117-1126. Herzog, T. (1985) A cognitive analysis of preference for waterscapes. Journal of Environmental Psychology. 5, 225-241. Herzog, T., Bosley, P. (1992) Tranquility and preference as affective qualities of natural environments. Journal of Environmental Psychology. 12, 115–127. Inter American Institute. (2010). Heads above water. How many people live in vulnerable coastal areas of the U.S.A.? Science Snapshots 3. Retrieved from http://www.iai.int/files/snapshots/Snapshot3_heads_above_water.pdf Kaltenborn, B., Bjerke, T. (2002). Associations between environmental value orientations and landscape preferences. Landscape Urban Planning. 59, 1-11. Kaplan, R., Kaplan, S. (1989). The Experience of Nature: A Psychological Perspective. New York, NY: Cambridge University Press. Koordinates.com. (2010). All topographic. Retrieved from http://koordinates.com/#/search/category/topographical/ Luttik, J. (2000) The value of trees, water and open space as reflected by house prices in the Netherlands. Landscape Urban Planning. 48, 161–167. Maas, J., Verheij, R., Groenewegen, P., Vries, S., Spreeuwenberg, P. (2006). Green space, urbanity, and health: how strong is the relation? Journal of Epidemiology and Community Health. 60, 587-592. Maller, C., Townsend, M., Pryor, A., Brown, P., Leger, L. (2006). Healthy nature healthy people: ‘contact with nature’ as an upstream health promotion intervention for populations. Health Promotion International. 21, 45–54. Marmot, M. (2010). Fair Society, Health Lives: The Marmot Review. Retrieved from www.marmotreview.org/ Ministry of Health. (2002). Reducing Inequalities in Health. Retrieved from http://www.moh.govt.nz/moh.nsf/49ba80c00757b8804c256673001d47d0/523077dd deed012dcc256c550003938b?OpenDocument 35 Ministry of Health. (2007). NZDep2006 Index of Deprivation. Retrieved from http://www.health.govt.nz/publication/nzdep2006-index-deprivation Mitchell, R., Popham, F. (2007). Greenspace, urbanity and health: relationships in England. Journal of Epidemiology and Community Health. 61, 681-683. Mitchell, R., Popham, F. (2008). Effect of exposure to natural environment on health inequalitites: an observational population study. The Lancet, 372 (9650), 1655-1660. Murray, C., Gakidou, E, Frenk, J. (1999). Health inequalities and social group differences: what should we measure? Bulletin of the World Health Organization, 77 (7). New Zealand 2006/07 Health Survey. (2012). A Portrait of Health: Online data tables of 2006/07 New Zealand Health Survey results. Retrieved from http://www.health.govt.nz/publication/portrait-health-online-data-tables-2006-07new-zealand-health-survey-results Ogunseitan, O. (2005). Topophilia and the quality of life. Environmental Health Perspectives. 113, 143-148. Park, J., O’Brien, L., Roe, J., Ward Thompson, C., Mitchell, R. (2010). The natural outdoors and health: Assessing the value and potential contribution of secondary public data sets in the UK to current and future knowledge. Health and Place, 17 (1), 269-279. Pearce, J., Witten, K., Bartie, P. (2006). Neighbourhoods and health: a GIS approach to measuring community resource availability. Journal of Epidemiology and Community Health. 60, 389-395.Richardson, E., Pearce, J., Mitchell, R., Day, P., Kingham, S. (2010). The association between green space and cause-specific mortality in urban New Zealand: an ecological analysis of green space utility. BioMed Central, 10 (240), 1-14. Smith, D., Croker, G., McFarlane, K. (1995). Human perception of water appearance. New Zealand Journal of Marine and Freshwater Research. 29, 29-50. 36 Statistics New Zealand. (2006a). Census Definitions and Questionnaires. Retrieved from http://www.stats.govt.nz/census/about-2006-census/2006-census-definitionsquestionnaires/definitions/geographic.aspx Statistics New Zealand. (2006b). Interactive Boundary Maps. Retrieved from http://apps.nowwhere.com.au/StatsNZ/Maps/default.aspx Tilley, C. (2004). The Materiality of Stone: Explorations in Landscape Phenomenology 1. Oxford: Berg. Ulrich, R. (1984). View through a window may influence recovery from surgery. Science. 224, 420-421. United Nations. (2009). Implementing the Millennium Development Goals: Health Inequality and the Role of Global Health Partnerships. Retrieved from http://www.unicef.org/health/files/MDG_and_Health_Inequalities_UN_2009.pdf Van den Berg, A., Koole, S., van der Wulp, N. (2003). Environmental preference and restoration: (how) are they related? Journal of Environmental Psychology. 23, 135146. Volker, S., Kistemann, T. (2011). The impact of blue space on human health and wellbeing – Salutogenic health effects of inland surface waters: A review. International Journal of Hygiene and Environmental Health. 214, 4449-460. Wheeler, B., White, M,. Stahl-Timmins, W., Depledge, M. (2012). Does living by the coast improve health and wellbeing? Health and Place, 18 (5), 1198-1201. White, M., Smith, A., Humphryes, K., Pahl, S., Snelling, D., Depledge, M. (2010). Blue space: The importance of water for preference, affect, and restorativeness ratings of natural and built scenes. Journal of Environmental Psychology. 30, 482-493. Witten, K., Hiscock, R., Pearce, J. & Blakely, T. (2008). Neighbourhood access to open spaces and the physical activity of residents: A national study. Preventive Medicine, 47, 299-303. World Health Organisation. (1948). Preamble to the Constitution of the World Health Organization as Adopted by the International Health Conference. New York: World Health Organisation. 37 World Health Organisation. (1978). Declaration of Alma-Ata: International Conference on Primary Health Care, Alma-Ata, USSR, 6-12 September. Retrieved from http://www.who.int/publications/almaata_declaration_en.pdf World Health Organisation. (1998). Fifty-first World Health Assembly, Health-for-all policy for the twenty-first century. Retrieved from http://www.who.int/archives/hfa/ear7.pdf Yamashita, S. (2002) Perception and evaluation of water in landscape: use of PhotoProjective Method to compare child and adult residents’ perceptions of a Japanese river environment. Landscape Urban Planning. 62, 3-17. 38 Appendices Appendix A Cook’s distance output, showing maximum value of 0.002 indicating there are no points with undue influence on the fit of the quadratic model. N Cook's Distance 64080 Valid N (listwise) 64080 Minimum .000 Maximum .002 Mean .00002 Std. Deviation .000046 39 Appendix B Q-Q Plot of residuals showing a slight deviation from normality. 40 Appendix C Predicted value versus residuals, showing a constant cloud formation of data points along the zero line, indicating that the variance of the residuals is constant.