Ashes to ashes, and dust to dust: Is scattering garden the sustainable destination for cremated ashes?

Cremation is commonly practiced around the world because it requires small space for the disposal of ashes. Among various options for ash disposal, many people choose to scatter the ashes of their loved ones in a scattering garden. What are the impacts of ash scattering on the vegetation of the garden? Is scattering garden a sustainable solution to the disposal of cremated ashes? This study aimed at answering these questions by characterizing and assessing the vegetation performance of a scattering garden using remote sensing techniques and field measurements. The results indicated that, overall, approximately half of vegetation was degraded to either unhealthy or bare soil. The area of bare soil in the lawns of high scattering level was larger than that of low scattering level. Furthermore, the belowground biomass of vegetation in the lawns of high scattering level was significantly lower than that of low scattering level. It is concluded that the current practice of ash scattering in Hong Kong was not sustainable and the intensity of impacts was dependent upon the level of ash scattering. The findings of this study may provide a reference for the policy and management of ash scattering in Hong Kong and other cities around the world.


Introduction
Death is an essential aspect of our existence that we cannot avoid. The disposal of the dead is an inevitable activity of human society that is of personal, emotional, social, and environmental significance (Canning and Szmigin 2010). Although different cultures and traditions have their death rites and rituals, land burial had been the commonest practice since pre-history (Decker et al. 2018). However, we see a dramatic increase in the cremation rate in western countries in the last half-century (Davies and Mates 2005). Take the USA as an example, the cremation rate increased from 5.7% in 1975 to 48.7% in 2015 and since then, cremation has taken over land burial to be the commonest way of handling the dead. The rate was forecasted to further increase to 79.1% in 2035 (Statista 2020).
Around the world, the cremation rate is often higher than 70% in those areas with a high density of population but limited land space (National Funeral Directors Association 2015). For example, Hong Kong sustains a population of approximately 7.5 million people on 1107 km 2 of land (Census and Statistics Department 2019), and approximately 90% of the deceased bodies are cremated (Food and Environmental Hygiene Department 2018).
The popularity of cremation reflects the modern society holding a more secularized stance on cremation (Heessels 2012; Nordh et al. 2021). From the viewpoint of public administration, local authorities promote cremation to address the constraints of cemetery capacity because cremation turns the body into ashes that greatly saves space for disposal (Canning and Szmigin 2010). However, cremation is not the final solution to the disposal of the dead. It is just a treatment that makes the disposal more flexible (Hupková 2014). There are many options for ash disposals, such as ash scattering (either on land or at sea), urn storage (either keeping at home, at a columbarium), burying in a cemetery, etc. (Mathijssen 2017). Among these options, many people choose to scatter the ashes of their loved ones in a scattering garden (Nordh et al. 2021). For example, more than 70% of the total number of cremation cases in the Netherlands were disposed of by scattering (Dijk and Mengen 2002). In the Czech Republic, more and more people inter their deceased relatives' ashes in a collective site of placement, such as scattering gardens or meadows, where the ash is dispersed in an anonymous way (Palánová and Juračka 2018). In Asia, ash scattering is not as popular as in western countries, but the number of ash scattering has greatly increased in recent years. For example, in Hong Kong, there were 5,573 cases of scattering in scattering gardens, respectively, accounting for about 12.1 % of the total number of 45,883 deaths in 2017 (Legislative Council 2020), increased from 4.6% in 2010 (Legislative Council 2018).
Although ash scattering is one of the most common forms of disposal of the dead, to the best of my knowledge, no scientific studies had ever attempted to evaluate the environmental sustainability of ash scattering using empirical data. There is an urgent need for scientific data and guidelines that inform the policy and plan for the disposal of the dead (Basmajian and Coutts 2010). Because ashes are an alternative form of human body, research ethics requires researchers to seek consent from the studied subjects before measurements or experiments are conducted. However, such consent is impossible to obtain, implying the existence of insurmountable difficulty. Furthermore, researchers tended to avoid having physical contact with cremated ashes because of sensitive implications, such as cultural beliefs and issues linked to the phenomenon of death history (Decker et al. 2018). Although a few studies had discussed the environmental implications of ash scattering, their views were diverse or even contradictory. For example, Niziolomski et al. (2016) indicated that cremated ashes would be toxic to plants because the high sodium content of cremated ashes exceeded the tolerable limit of plants. On the other hand, Strand et al. (2008) indicated the potential of cremation ashes as sources of phosphorous for soil additive or fertilizer, based on a laboratory experiment that used chemical solvents to extract phosphorus from cremated ashes.
With this in mind, this study aimed at filling the research gap by characterizing and assessing the vegetation health of a scattering garden that had been open for ash scattering for 6 years. Specifically, the vegetation quality and quantity of six lawns were assessed by analyzing the aerial photos of the lawns; then, grass samples were harvested and the weight of aboveground and belowground biomass was determined to examine the effects of ash scattering on the growth of the plant. The findings of this study may contribute to the body of literature by providing a portrait of the vegetation performance of a scattering garden after receiving cremated ashes. Furthermore, the findings provide reference to the policy and management of ash scattering in Hong Kong and other cities around the world.

Vegetation health assessment
The concept of vegetation health is the analogy to human health (Wicklum and Davies 1995). Whilst the human body involves the coordination of internal organs and metabolic activities, vegetation is a complex system that involves the interaction between plant community and the physical environment (Xu and Guo 2015). The concept, framework, and methodology of vegetation health assessment provide tools for this study.
There are mainly two approaches to the study of vegetation health. The first approach examines the vegetation cover that provides a general sense of the lawn's health condition (Xu and Guo 2015). Remote sensing has been intensively used for the estimation of vegetation quality and quantity (Soubry et al. 2021). Various vegetation indices have been proposed for different management and assessment purposes (Karnielia et al. 2013;Al-Ali et al. 2020). Remote sensing applications are cost-effective as they can cover a large area and extract the required information on the phenomenon on a minimal budget (Luna and Lobo 2016;Al-Ali et al. 2020). Dependent on the subject matter, data of different wavelength bands, such as red-green-blue (RGB), near-infrared, and Lidar, can be taken by aircrafts or flying devices at different heights, ranging from near-ground unmanned aerial vehicle (UAV) to space-borne satellite (Soubry et al. 2021). However, remote sensing data represents a visual inspection of the aboveground part of grass but is not a direct measurement of vegetation (Xu and Guo 2015). Furthermore, there is a lack of a consistent approach to the study of grassland ecosystems (Joshi et al. 2016;Soubry et al. 2021).
The second approach aims at determining productivity (Ni 2004). The concept of productivity is related to the idea of energy flow in the ecosystem (Kent and Coker 1992;Cao et al. 2020). Plants absorbed solar energy through the activity of photosynthesis and stored it in form of organic substances. Plants use the stored energy in these organic compounds fixed in photosynthesis for their respiration. The balance between carbon fixation in photosynthesis and carbon loss in respiration is net primary productivity (Yazaki et al. 2016). Therefore, a synthesis of productivity data is best based on biomass measurements (Ni 2004). For most terrestrial ecosystems, harvests of aboveground biomass have been the most frequently used technique (Clark et al. 2001). However, for grassland, it may be necessary to determine the belowground biomass because the root part of grass often exceeds its shoot by factors of two to thirty (Gao et al. 2008). Precise estimates of both aboveground and belowground biomass are important for the understanding of the productivity turnover in grassland (Gao et al. 2008). Different from remote sensing which inspects every inch of the study area, harvesting biomass samples would be possible only at selected sites because it is expensive and time-consuming (Muukkonen et al. 2006). There is always a balance between the budget available and the accepted error (Flombaum and Sala 2007).
From the above observations, it is clear that these two strands are not mutually exclusive but complementary to each other. Thus, this study adopted both remote sensing techniques and field measurements to study the performance of lawns in a scattering garden.

Studied area
This study selected the Garden of Remembrance (GoR) at Diamond Hill Crematorium and Columbarium as the studied area because it was the most popular venue for ash scattering in Hong Kong. In 2017, the GoR handled 2,365 cases of ash scattering, representing 42.45% of the total number of 5,573 scattering cases in Hong Kong (Legislative Council 2020).
The GoR is carefully crafted to be an oasis among concrete graves (Fig. 1). The GoR is situated at the terrace deposits in a small river valley. The center-piece of the GoR is six lawns, accounting for a total area of 502.76 m 2 for ash scattering, located on the eastern side of the valley. A small scenic garden, for resting and medication, is the extension of the GoR located on the western side. There is only one entrance and exit at the south end of the GoR.
Although these lawns were varied in size, they were identical in terms of design, construction, and management. The lawns were defined by local topography and decorative bushes. Users should enter the lawn via the footpath and be free to scatter the ashes along the journey. However, they were not allowed to step on the vegetation for the reason of respecting the dead. The grass on the lawn was Zoysia grass (Zoysia tenuifolia) and the soil was a mixture of peaty soil and sand (1:1), respectively. The lawns were watered daily and clipped once a month.
These lawns received different loadings of cremated ashes that allowed a comparison of the conditions under different levels of ash scattering. Using the overall mean scattering rate (4.41 cases/m 2 ) as the reference, these lawns were categorized into two groups: high level of scattering (>4.41 cases/m 2 year), and low level of scattering (<4.41 cases/m 2 year). Table 1 provides the basic information on these lawns.

Pilot study
In the initial stage, the author visited the GoR and discussed with the management team the feasibility of this study. One key issue was the seasonality, especially the weather and number of users, that could affect the vegetation performance. The manager indicated that the vegetation condition was not varied in different seasons as the lawns were carefully maintained by experienced artisans. On the other hand, the number of users remained fairly stable throughout the year because the death rate was fairly constant over time. Based on these observations, the sampling was conducted in December 2018 because the weather was stable and suitable for fieldwork.

Vegetation cover
The performance of vegetation was assessed in two ways. First, the quality and quantity of vegetation cover, indicating plant abundance (Kent and Coker 1992), was determined by analyzing the aerial photos of the lawns acquired by an UAV (Dji model: Mavic Pro) equipped with a professional-grade camera with a 35 mm focal length and a sensor (78.8 mm × 26 mm) of 12.71 megapixels. The red-green-blue (RGB) images were acquired at the heights of 25 m and 110 m above the GoR. These images were taken around noon time to avoid the formation of shadows. The advantages of remote sensing and image analysis are the non-destructive nature of these techniques and at the same time, allowing an accurate estimation of vegetation types and areas (Luna and Lobo 2016).
ArcGIS ver. 10.5 was used to correct and provide the georeference using the images acquired from the flight at 110 m above ground. The simple first-degree polynomial transformation was used because the terrain was flat and a few pairs of ground control points were available to be the reference. Images acquired from the flight at 25 m above ground were used to produce a mosaic image with a pixel resolution of 0.83 cm. The resultant mosaic approximately had a geometric error of 5 cm estimated by the visual identification of the grasses and other features of the GoR.
Envi ver. 5.3 was used to classify three classes of vegetation using the red-green-blue bands of the mosaic image. They were as follows: (1) healthy grass, (2) unhealthy grass, and (3) bare soil (i.e., no grass). A sufficient number of representative pixels of each class were selected for the training set (Richards and Jia 2005). To test the quality of training samples, spectral separability between classes was accessed by calculating the transformed divergence (TD) and Jeffries-Matusita (J-M) distance of class pairs. The range of TD and J-M is 0 to 2. The value of 1.9 indicated good statistical separability. Training samples would be modified when the separability is low. Table 2 and Table 3 show the number of training pixels and the spectral separability report, respectively.
Maximum likelihood classifier (MLC) was used, with the pavements being masked out in the process. MLC is a parametric classifier making use of Bayes' theorem of decision making, with the assumption of normally distributed classes in each band. It assigns the class to a pixel based on the highest probability of membership to a class, derived from the mean vector and the covariance matrix (Zheng et al. 2009). The discriminant function of each pixel was calculated using Equation (1) (Richards and Jia 2005).

Vegetation biomass
In this study, 90 sampling points were systematically selected from the GoR for the determination of the biomass of vegetation, based on the following criteria. First, a random start was set and samples were collected in an interval of 1.5-2.0 m. Second, the distance between the sampling point and the edge was as far as possible not less than 1 m.
(2) Vegetation performance index = (area of healthy vegetation × 1) + (area of unhealthy vegetation × 0.5) + (area of bare soil × 0) Third, the inner part of the lawn was not sampled because stepping on the lawn was prohibited. There were 64 samples in lawns of high scattering level and 26 samples in lawns of low scattering level, respectively.
At each sampling point, plant samples were harvested by a metal core sampler with a diameter of 4.8 cm. After the attached soils and ashes were carefully washed and returned to the lawn, the grass samples were oven-dried for 24 h and the weight of aboveground and belowground biomass was determined.
Data were input to SPSS 25.0 for statistical analyses. The normality of the data was checked, and skewed data were log-transformed. The homoscedasticity of the data was tested by Levene's test of equality of error variances. Oneway analysis of variance (ANOVA) was employed to test the biomass difference between high and low rates of use.
Because of the sensitive nature of sampling human remains, a relatively small number of sample points were selected so the statistical basis of the sample size was compromised (Catchpole and Wheeler 1992).

Vegetation cover
The aerial photos taken at the height of 25 m above ground were integrated into a mosaic photo (Fig. 2a); then, the photo was used for the classification of vegetation cover (Fig. 2b). All lawns showed different degrees of degradation as unhealth vegetation and bare soil were found in all lawns. Unhealthy grasses were usually found along the footpath but relatively healthy vegetation was established at inner locations. Figure 3a summarizes the results of vegetation classification. Overall, approximately half of the vegetation was degraded to either unhealthy (39.6%) or bare soil (13.9%). The area of bare soil in the lawns of high scattering level was larger than that of low scattering level. For the lawns of high scattering level, up to 26.8% of vegetation cover was eliminated. The results of vegetation composition were used to calculate the vegetation indexes using Equation (2) (Fig. 3b). Overall, the vegetation index was 0.68 for the whole GoR. The index of the lawns of high scattering level was 0.60, and that of low scattering level was 0.69, respectively. Because the vegetation indexes were lower than 0.75, the performance of the GoR was unsatisfactory; but the indexes were higher than 0.50 which would not be considered unacceptable.

Vegetation biomass
The results of the aboveground biomass and belowground biomass are shown in Fig. 4, respectively. Overall, the amount of aboveground biomass (0.5141 ± 0.2772 kg/ m 2 ) was similar to that of belowground biomass (0.5283 ± 0.2697 kg/m 2 ). There was no significant difference (F= 0.067, p>0.05) in aboveground biomass between the lawns of high scattering level (0.5260 ± 0.3301 kg/m 2 ) and low scattering level (0.5093 ± 0.2551 kg/m 2 ). However, these figures might not truly reflect the aboveground productivity of the lawns because the lawns were clipped once a month and aboveground biomass was regularly removed. On the other hand, the results of belowground biomass were more reliable and deserved more attention (Fiala 2010). The difference of belowground biomass between lawns of high scattering level (0.4131 ± 0.1823 kg/m 2 ) and low scattering level (0.5752 ± 0.2861 kg/m 2 ) was significant (F= 7.14, p<0.01). The results of belowground biomass were consistent with the vegetation performance index, both indicating that the level of ash scattering significantly affected the growth of grass.

Environmental impacts of ash scattering
The results indicated that the current practice of ash scattering generated significant environmental impacts on the GoR. When the scattering level is higher, the impacts are larger.
Overall, more than half of the lawn area was degraded, indicating that the performance of the GoR was not satisfactory. The vegetation index was 0.68 for the whole GoR (4.41 cases/m 2 year), which was lower than the satisfactory value of 0.75. For the lawns under low use (3.50 cases/ Ash scattering also resulted in the reduction of grass biomass in the GoR, indicating the negative effects of cremated ashes on the plant production. The underground biomass of the lawns of high scattering level was significantly lower than that of low scattering level. Both the results of underground biomass and vegetation cover indicated that the level of ash scattering generated significant impacts on the plant growth. While ash scattering is considered a form of green burial, its environmental and ecological impacts should not be underestimated and deserve further research. However, this study did not quantify the relations between the use rate (in terms of scattering cases) and the impacts on vegetation. Consequently, this study was not able to recommend standards for the rate of ash scattering in the GoR.

Landscape considerations and design of scattering garden
Unhealthy grasses along the footpath but healthy vegetation at inner locations indicated that impacts were associated with pedestrian traffic. However, the specific spots which users selected for ash scattering may be affected by the characteristics of the landscape. There are a few factors for consideration. First, the accessibility of the lawn may affect whether or not the users select it for ash scattering. Whilst higher levels of use were found at the lawns near the entrance of the GoR, lower levels were at those lawns at the peripheral locations. For example, lawns 1 and 2 enjoyed the highest levels of scattering as they were located at the convenient locations, but lawns 5 and 6 had low levels of scattering because of the far distance to the entrance. Second, favorable features may enhance the popularity of the lawn for scattering. For example, lawn 4 was a relatively sensible choice for scattering because of the nearby stream, bridge, and scenic garden; but lawn 6 was not popular because of the adjacent power cable tower located at a distance of approximately 50m to the lawn. Third, users prefer enclosed spots to open sites for ash disposal probably because of the privacy and/or security perception. It was obvious that bare soil was commonly found at the corners of the lawns.
The above observations may prompt the importance of landscape considerations for the design of scattering garden. Because the disposal of the dead is not merely disposing of the body, its meaning always involves the remembrance and extended love to the beloved. To strengthen the multifaceted meaning of death and to link it more with the GoR, attention to potential outcomes for improving landscape design accommodating better facilities and services for the disposal of the dead should be promoted and developed (Długozima 2020).
Globally, the demand for venues for ash disposal is in increases. Many countries, e.g., Singapore (The National Environmental Agency 2021), Taiwan (Ong, 2020), and the USA (Detroit Free Press, 2019) have developed their scattering gardens for ash disposal. If there is proper management, scattering gardens are green urban spaces with cultural and natural values. They can provide a public service and be integrated into the green infrastructure planning system (Nordh and Evensen 2018). These findings may cast light on the design of scattering gardens in the future.

Environmental sustainability of ash scattering and policy implications
As Hong Kong, like many Asian cities, sustains a large population in a relatively small area, the high population density inevitably translates into high demands for infrastructures for the disposal of the dead. Because of lacking land space for land burial, currently over 90% of the bodies of the deceased are cremated (Food and Environmental Hygiene Department 2018). The problem of space shortage is so acute that there are not sufficient columbaria to meet the demand for ash disposal. Consequently, there are around 200,000 sets of ashes waiting for a niche space, with many stored at funeral parlors (Keegan 2019).
In recent years, more and more people have chosen to scatter ashes, either on terrestrial land or at the sea, accounting for about 12.9% of the total number of deaths in 2017, relative to 4.6% in 2010 (Legislative Council 2018). Currently, 11 public GoRs are available for ash scattering. From 2001 to 2017, a total of about 23,400 bereaved families applied for scattering the ashes of their family members in these GoRs (Legislative Council 2018). The trend is still on the increase.
The unsatisfactory performance of vegetation prompts room for improvement in the current practice of ash scattering in Hong Kong. Because of inadequate provision of venues for ash scattering, the GoR is overloaded by an excessive number of scattering cases. The situation would become worse as more people would choose scattering garden as the final destiny for their beloved in the future. Therefore, this study recommended that the government should open more GoRs to meet the increasing demand for ash scattering. Of course, it is possible to promote other forms of ash scattering, e.g., scattering at sea, that help reduce the demand for scattering gardens. In the long run, the government should improve the services and facilities for ash scattering and at the same time, the environmental impact of ash scattering should be further explored but it is under-researched.

Limitations and recommendations
This paper has four limitations that future studies should address. First, although this study demonstrated that ash scattering affected both vegetation cover and productivity, the mechanism is still unknown. Soil is believed to play an important role in mediating the effects because it is the substrate for the growth of plants (Tan 2014). Future studies should examine the relations between plants and soil by measuring the physical and chemical properties of the soil. Understanding the interaction between ashes, soil, and vegetation can facilitate a better scheme for the management of scattering garden. Second, the landscape and spatial preference of the users deserves further research. This study identified the associations between landscape, and use and impact patterns primarily based on observations. Therefore, both qualitative (e.g., in-depth interview) and quantitative (e.g., questionnaire survey) techniques are recommended to verify these associations. Understanding the landscape and spatial preferences not only facilitates a better design of scattering garden but also helps promote scattering in the society. Third, it is arbitrary to use the yardsticks set up by the management team of the GoR to assess the vegetation performance of the GoR. Although these yardsticks were reasonable and of common sense, different conclusions may be drawn if different yardsticks were used for specific targets and practical needs. Fourth, the relations between ash scattering and impacts could not be quantified in this study because a few variables were not controlled. For example, the amount of cremated ashes was varied in scattering cases; also, the ashes were not evenly applied to the lawns but up to the preference of the persons who scattered the ashes. However, it is extremely difficult, if not impossible, to conduct a controlled laboratory-typed experiment because cremated ashes are another form of human body. There is a long list of difficulties and restrictions when controlled experiments involving human subjects are commenced. Overall, although the scope and depth of discussion in this study were compromised with the limitations, some findings are timely and meaningful to the policy and management of ash scattering in Hong Kong.

Conclusions
This study aimed at assessing the environmental impacts of ash scattering on a scattering garden in Hong Kong. Two complementary methods were employed to study the vegetation performance of a scattering garden that had been open for ash scattering for 6 years. Overall, approximately half of the vegetation was degraded to either unhealthy or bare soil due to ash scattering. The unsatisfactory performance of the GoR indicated that the current practice of ash scattering is not environmentally sustainable. When the scattering level is higher, the impacts are larger; the trend is reflected by the vegetation indexes of lawns. On the other hand, the belowground biomass of lawns under high use was significantly lower than those of lawns under low use, indicating the negative effects of cremated ashes on plant biomass. However, the difference in the aboveground biomass between lawns was not significant. The findings of this study may provide a reference for the policy and management of ash scattering in Hong Kong and other cities around the world.