Employing Green Roofs to Support Endangered Plant Species: The Eastern Suburbs Banksia Scrub in Australia

The purpose and context for the study relates to urban growth. Australian cities are experiencing particularly rapid urbanization, taking the form of land clearing to accommodate outward expansion as well as developing to higher densities in existing urban areas. Both forms of development degrade native biodiversity, resulting in loss of vegetation with the possibility that the remnant indigenous plants will become locally extinct. One endangered ecological community in Sydney, the Eastern Suburbs Banksia Scrub (ESBS), still survives along some sections of Sydney’s heavily urbanized coastline. At the time of European settlement, the ESBS covered approximately 5300 ha, but it is now a highly fragmented 146 ha across 24 sites with some sites under im-minent threat of development. Conservation legislation enacted by the state of New South Wales (NSW), Australia has declared the ESBS as critically endangered. Despite recovery plans, in 2016 the NSW Threatened Species Scientific Committee indicated that the community faces an extremely high risk of extinction in Australia in the immediate future. A practical option in the face of declining open space in our cities is to examine the potential of urban rooftops for conserving and propagating threatened or endangered flora. While there is a limited amount of international research on using green roofs for endangered plant protection, there is no information from Australia about how green roofs perform in this geographic region. The approach taken in this research has been firstly, to review the current academic and “grey” literature from a global perspective to identify options for conserving endangered flora on green roofs. We derive an evidence-based research protocol to be used to test the green roof environment in Sydney for propagating the endangered ESBS. We establish the general applicability of green roofs monitoring growth and germination performance over the ESBS community’s development cycle, with the longer-term objective of establishing a viable rooftop seed orchard.


Introduction
Rapid urbanisation has been identified as a so-called "megatrend", with an estimated 1.5 million people being added to the global urban population every week. While most of this growth is occurring in Africa and Asia, Australia's rate of urban growth is among the highest in the developed world. The statutory government agency, Infrastructure Australia [1] has projected an additional 11.8 million Australian city dwellers between 2017 and 2046, equivalent to building a new Canberra-sized city every year for the next 30 years.
From a systems perspective, key environmental impacts of urban growth include an upsurge in consumption of resource inputs (energy, water, materials) and in production of detrimental outputs (air and water pollutants, waste) [2]. Further, land clearing to accommodate both outward expansion and internal densification degrades native biodiversity and supplants urban and peri-urban agriculture [3]. However, it should be noted that removal of vegetation is by no means restricted to urban development, and some authors [4] point out that land clearing remains the single greatest threat to terrestrial biodiversity in Australia. In the State of New South Wales (NSW) for example, clearing of native vegetation, predominantly for agriculture and mining, jumped 800% between 2013 and 2016. As at December 2018, 692 plant species were listed as "threatened", representing about 14% of the total number of native plants in NSW [5].
The above provides the context for a pragmatic approach to the conservation of one particular ecological community, the Eastern Suburbs Banksia Scrub (ESBS), which still survives along some sections of Sydney's heavily human-modified coastline. The ESBS was the first ecological community to be listed as endangered under the NSW Threatened Species Conservation Act 1995; this Act has been superseded by the NSW Biodiversity Conservation Act 2016, under which the ESBS is now listed as critically endangered. The [6].
Species Scientific Committee has indicated that the community faces an extremely high risk of extinction in Australia in the immediate future [7] and, as noted above, the threat to the ESBS is not unique.
Protecting endangered species in their existing habitat in conditions of rapid development is inevitably going to be difficult. One pragmatic but innovative option is to expand our horizons to consider the untapped resource of our urban rooftops. Thus, the framework for this research is conservation biology and the main objective of this proposal is to examine the little remaining vacant space in our metropolises and specifically to evaluate the capacity of green roofs (GRs) to grow and sustain key ESBS species. An additional aim is to determine the ability of GRs more generally to help conserve endangered plant communities and spe-cies, acknowledging that there is a limit to the size of plants suitable for rooftop planting. In this way, it may be possible to shrink the human ecological footprint to a small degree.

Green Roofs and Their Benefits
Multiple environmental, social and economic benefits connected with GRs are widely reported in the literature, as summarized in Table 1. Several landmark studies have been conducted in the past including those by [8] who modelled the monetary benefits of installing green roofs in Toronto; the Centre for Neighbourhood Technology and American Rivers (CNTAR) which also focused on quantifying benefits [9]; the researchers [10] who compiled a professional guide to the design, installation and maintenance of green roofs; and the United States' General Services Administration (USGSA) [11]. The USGSA study was a meta-analysis of 200 research studies on the costs, benefits, challenges and opportunities of green roofs, and was accompanied by an original cost-benefit analysis and discussion of best practice. It is important to check the assumptions behind the studies that have attempted to monetize benefits. For example, [8] assumed that 100% of available roof space and all roofs over 350 m 2 would be vegetated.
Other research includes a study on stormwater quality from the UK [12]; a "state of the art" analysis of environmental benefits [13] and a white paper on the  Biodiversity conservation and protection including endangered and threatened flora and fauna [32] [54]- [60]  CO 2 sequestration (though minor compared to street trees) For example, [9] [43] and [61] Open Journal of Ecology potential unintended consequences of reflective cool roofs, increasingly proposed as an alternative to more costly green roofs to mitigate urban overheating [14].
Collectively, the GR benefits cited in the international literature are wide-ranging. Reduced stormwater runoff together with improved water quality are common findings, although [12] as a rare exception, found high concentrations of heavy metals in runoff from an old GR in Manchester, England. Microclimatic effects include a contribution to reducing the urban heat island [13] [15] [16] and [17]. Using advanced simulation techniques, [15]  These are not dramatic reductions and they lessen at lower wind speeds. On the other hand, researchers [18] using a heat index comprising temperature and relative humidity, found that a combination of green roof and green wall retrofits offered a distinctly helpful role in attenuating interior heat stress in residential buildings.
Green roofs reduce cooling and heating loads for the floors immediately beneath the roof [19] and [20] and some particulate atmospheric pollutants are adsorbed, while greenhouse gases as well as other gaseous pollutants are absorbed [21], CO 2 being removed through photosynthesis. Biodiversity improvements are indisputable compared with conventional roofs. However, biodiversity protection and conservation of endangered flora have rarely been addressed in the GR literature compared with the research carried out on stormwater quality and detention, on building energy savings, or indeed on faunal biodiversity [22].
With the continued pace of research on GRs, however, there are recent signs of researchers investigating the capacity of the artificial environment of green roofs for biodiversity protection and conservation of endangered flora, as noted in Table 1.
Other somewhat less conspicuous benefits of GRs include noise attenuation [23], the potential for urban agriculture [24] and enhanced roof membrane durability. In the last case, the USGSA [11] suggests that a conservative estimate puts the average life of a GR membrane at 40 years compared with 17 years for a conventional roof. Green roofs' role in providing additional passive recreational space in dense urban settings is also valuable [25] and [19]. Site visits to several GRs in Sydney demonstrate that some are used as pleasant settings for passive relaxation.   There are likely to be significant differences in outcomes from extensive as distinct from intensive roofs 1 for the benefits listed [13] [26]. Moreover, GRs need to be implemented on a large scale for bio-physical benefits to be appreciable and some of those benefits could be obtained by other means [27]. For ex- 1 For the purpose of this paper, intensive green roofs have a thicker substrate (typically > 200 mm), generally support a much greater variety of plants including shrubs and small trees, require more maintenance and are designed for people to use. Extensive roofs are constructed on a substrate < 150 -200 mm deep and comprise a shallow layer of vegetation such as sedums, grasses and other groundcover species. In practice, green roof depth varies even within the same site and the extensive -intensive categorization is a continuum, as seen in Figure 3.
ample, while the CO 2 sequestration capacity of a 500 m 2 extensive Sedum spp.
green roof is measurable, it is negligible compared with planting a single medium size tree [9]. Furthermore, reducing roof temperatures through planting could be achieved equally effectively by using a high albedo surface or "cool roof" technique like reflective paint [28]. Recent research from Lawrence Berkeley National Laboratory using a 50-year life cycle costing framework, suggests they could cool our urban areas three times more effectively than green roofs per unit area [29]. On the other hand, [14] point to a series of unintended consequences associated with reflective roofs and pavements such as glare, the health impacts of higher levels of UV radiation and possibly reduced local precipitation.
Regarding the societal goal of reducing atmospheric CO 2 , GRs can reduce energy consumption in a low rise building as well as sequester carbon in plants and substrate, but there is a distinct carbon cost involved in installing GR components and maintaining the installation during their lifecycle. Japanese research [61]  In Australia, GRs have been less popular perhaps due to the perceived expansiveness of the physical environment, with its generous suburban landscape [62].
In the 1960s, a limited number of GRs were installed in Sydney, mainly on new apartment buildings, designed to extend domestic living space. A rare commercial example from that era is the 1967 Reader's Digest Building, featuring an intensive GR with exotic and Australian native plants, included as an outdoor 'retreat' for employees of the company (Figure 2, above).
Research on the bio-physical benefits of GRs has been conducted at several green-roofs-as-part-of-planning-changes-).
One of the smaller municipalities in the Perth metropolitan area has a water sensitive urban design policy in which green roofs are listed as a contributory element [77] and Darwin has a fleeting reference to green roofs in general discussion in its master plan [78]. However, a recent report commissioned by the city [15] presents convincing evidence that counterbalancing high ambient temperatures and the impact of urban heat islands is entirely feasible by adopting mitigation strategies like cool roofs and pavements, green roofs and urban greenery, shading and the use systems like water sprinklers and fountains [15].
Lastly, the local government of Australia's national capital of Canberra also refers to the value of green infrastructure and green roofs [79].

Conservation Biology
Natural ecosystems provide a critical range of services such as food, fuel, and  [85]. Urbanization is a particularly significant threat but at the same time, urban green space, offers many opportunities for biodiversity conservation if it is managed with this objective in mind [57]. Ecological restoration [86] aims to slow the rate of species extinction and ecosystem service decline with two methods in particular. They are the conservation of currently viable habitat, and the restoration of degraded habitat, both forming part of the four-tier urban ecological hierarchy established by [87] for New South Wales. Actions might include erosion control, reforestation, removal of non-native species and weeds, revegetation of disturbed areas, and the reintroduction of native species. Also, part of the hierarchy is a third technique, the creation of new habitat, for example by using spaces previously uncolonized by the target species or indeed, any plant species at all, for example the roofs of buildings, the subject of this research proposal.
One researcher [88], noting a lack of basic information on how green roofs contribute to biodiversity, investigated the diversity of beetle communities, finding that it is the nature of the vegetation employed, in this case meadow grass species rather than forbs, that promotes faunal biodiversity, not so much the placement, age or height of the roof. However, [89] inventoried 51 GRs in Open Journal of Ecology Helsinki, Finland and found that substrate depth was a critical factor in structuring plant communities and vegetation abundance. It also was apparent that roof age was highly influential in structuring vegetation [89]. The plant communities changed from young sedum-moss dominated roofs and meadow-species communities chiefly characterized by the presence of sedums, into moss-dominated or almost pure meadow-species communities on older roofs.
Meanwhile in northern France, [90] installed green roofs consisting of 176 vascular plant species, 86% of which were indigenous, across 115 roofs. They tested several variables, also finding that plant diversity was strongly related to substrate depth as well as green roof age, its surface area and height above grade, and even maintenance intensity at building scale [90].
While not formally endangered, bees are critical to human food security and their population is declining rapidly in some parts of the world. Field research in Illinois [32], indicated in the USA that urban green roofs may enhance popula- Like [93], [94] have argued for a more disciplined conservation planning to ensure the representation of a region's biodiversity by separating it from threatening impacts. Separation in the city is difficult because habitat is increasingly fragmented into smaller, numerous remnant patches [95] within a hostile matrix of urbanization. Species richness often declines too, as fragment area decreases [96]. As available land at grade diminishes, the significance of alternative types of urban green space for biodiversity conservation simultaneously grows [97], between green roofs [101]. However, the habitat and corridor potential of a green roof needs to be evaluated in the context of its physical characteristics, microclimate and its relation to the urban landscape [31]. Other [102] accept that establishing a viable network of green roofs would support biodiversity by serving to shorten links between existing habitats but see this as an ideal. They caution that confounding issues of roof age, size and height above ground level, substrate depth and roof load bearing capacity as well as identifying roof tops in strategic locations to connect the fragmented networks of threatened fauna, may be difficult in practice.
Despite some scepticism about the value of corridors, the evidence from well-designed studies suggests that they are valuable conservation tools according to a detailed meta-analysis by [103]. Figure 4 shows schematically how green spaces such as backyards and green roofs could be planted with indigenous species to act as links between more significant fragments of native biodiversity.
The diagram represents a progression from a pre-development condition through traditional development with parks and open space (light areas, centre image) to a situation in which backyards and green roofs (the light green areas in RH image) act more as ecological stepping stones than corridors between the parks and reserves. They are valuable in functioning as resting or foraging points for birds and invertebrates [38] and provide an opportunity-albeit small-for seeds to disperse, settle and germinate.
A citywide green roof strategy to support biodiversity conservation would also be supported by the numerous other ecosystem services offered by GRs, noted earlier in this paper. Such a strategy represents an important principle, using the ecosystem synergies provided by GRs to build on the positive aspects of city living and mitigate in a small way some of its negative aspects, such as air pollution and stormwater flows. Several other principles operate too:  Sites which might be individually unimportant might become significant if they can be linked together into a web of habitat conservation sites;  The greater the number of ecosystem connections, the greater the chance of robustness and resilience [105] thus strengthening the goal to improve current linkages as well as create new ones; Figure 4. A possible progression for enhancing biodiversity in our cities [104]. From left: Pre-development; developed landscape with public open space; stepping stone spaces using green roofs and domestic gardens. Open Journal of Ecology  Full restoration of the pre-development plant communities is ecologically unrealistic in the city (particularly on rooftops with limited load-bearing capacity), so partial depiction or indication is a key goal [106];  Implementing a meaningful GR strategy across the roofs of private residential, business, industrial premises and government and other public buildings is likely to be slow and incremental. While government agencies may wish to be seen in supporting a formal strategy, there will be two particular obstacles to overcome regarding non-government buildings. First, there is a lack of flat roofs in the large residential matrix of our cities outside the central business districts; second, business and industry will need to be willing to undertake both retrofits and new installations. It will depend heavily on the principle of reconciliation ecology, the last principle enunciated below [107]; and  An important principle behind reconciliation ecology is enlisting community support by householders and businesses in ecological care, with the numerous bush care groups operating in the Sydney Region offering ample evidence of community interest in maintaining our native species. Green roof installation and plant maintenance would also entail reaching a compromise between human and non-human use of urban space to support biological conservation [107]. However, a green roof conservation initiative above grade and on private property would entail significantly higher levels of collaborative management than the typical ground level spaces in public ownership tended by residents.

International Experience
There has been much international research in relation to fauna and green roofs, for example work by [108] regarding invertebrates; the colonization of GRs by beetles [55]); the positive effect of GRs on populations of both native and exotic bees [32]; the use of GRs as nesting sites for birds [109] and [110]; a year-long comparison of avian use of green roofs versus nearby conventional roofs by [111], and a study in Sydney, Australia by [59] which showed that green rooftops host up to three times the number of invertebrates and twice as many invertebrate taxa compared with bare roofs. However, recent studies from several countries have found that local species will flourish on extensive green roofs as well as, if not better than a low-diversity mix of cosmopolitan succulents like Sedum spp [108] [112]. Researchers like [113] for example, evaluated four different types of vegetation for their stormwater attenuation capacity and found that sedum spp. showed the largest amount of water runoff and was the only species group with more water runoff than the bare ground. However, [114], found that many North American prairie and grassland species, subject to harsh growing conditions in their natural habitats, will flourish on extensive green roofs and there can be significant additional maintenance costs of maintaining non-native vegetation types [115]. These might include the fertilizer, pesticides and significant irrigation needs of using temperate climate species for planting in the hot desert environment of the United Arab Emirates. The researchers also noted that native plants also help to restore wildlife by providing food and shelter for local fauna.
In Japan, researchers successfully propagated 13 plant species, four of which were classed as threatened [54], from the Jogasaki seacoast in Japan and planted them in three kinds of substrate on a newly constructed green roof at a nearby coastal location. The researchers pointed out that a complete understanding of the natural habitat of the native plants and replicating it carefully was critical to successfully simulating a local landscape on the green roof. Some years earlier, [116] in arguing for natural habitats, found the key to maintain populations of some rare species was to preserve the top layer of natural substrate, seed bank and soil organisms, for subsequent installation on the roofs of new developments.
Urban habitats can harbour self-sustaining populations of threatened or endangered native species. They are not likely to be a complete substitute for the functionality of the original ecosystem [100] although recent work by [117] highlights the relative importance of small, isolated habitat patches for biodiversity protection and show that they often have unique ecological and environmental characteristics. Functional diversity is strongly associated with the provision of ecosystem services and is a useful concept for designing the type of ecosystem that might be found on GRs, including providing opportunities to incorporate threatened local species [118]. Functional traits are defined, for example, by plant height, longevity, leaf area, succulence and flowering time. Inclusion of threatened or endangered indigenous plants with desirable traits on a green roof could meet the dual objectives of maximising ecosystem services and conserving flora. Preservation of threatened species thus becomes one way in which the principle introduced above of indication [106] together with skilful design may be implemented on urban rooftops.
The biodiversity research in relation to fauna conducted on green roofs can offer a degree of guidance on how to manage endangered flora, although most of the research has focused on locally abundant species. Researchers [113] in earlier Open Journal of Ecology research point out that the harsh growing environment of extensive GRs tends to restrict the range of plant species used but suggest from their research on forbs, sedum and grasses, that a greater functional diversity of vegetation provided more resilience to drought than a monoculture and was also rated higher from an aesthetic viewpoint.
Surprising performance improvements in growth rates and vegetation abundance in the local species has been found by [108], using species such as Carex argyrathra and C. nigra as opposed to industry standard green roof succulents, Sedum acre and Sedum spurium. Perhaps Sedum vegetation is popular because it is easier to install, easily modularised and relatively cheap. Quoting [108]: "the problem is that Sedum plants aren't really performing on green roofs… They're just there." Apart from not absorbing water as efficiently as other species, at certain times of the year Sedum actually absorbs heat instead of reflecting it [119].
Thus, a broad conclusion from the literature is that indigenous species which thrive in shallow soils, tolerate drought and are adapted to high winds, extreme temperatures and intense sunlight, particularly dry grassland, coastal, and alpine floras [120] are well suited to green roof installation though [121] point out that flora which does not fall into this category may be best conserved at grade until further knowledge is accumulated.
Another frequent theme is the need to match substrate depth and nutrient quality with the desired plant species [108]. Some research [122] has found that found that response to higher levels of organic matter was different for different species, and that species from a nitrogen-rich habitat tended to be encouraged by a high nitrogen content. Similarly, [109] working on green roofs in Zurich, Switzerland, has shown that use of natural soils can benefit biodiversity with useful implications for sustaining endangered species. Hence roof substrate nutrient status needs to be carefully considered in using local indigenous species, whether natural or manufactured growing media are selected. However, Australia has a uniquely high proportion of nutrient-poor soils to which much of the continent's native flora is adapted, which augurs well for the transfer of flora to GRs [123].
Equally critical is the need to consider the micro-fauna associated with green roofs. [124] quantifying the total microbial biomass and fungal levels in roof substrates and park soils, finding that park soils had greater microbial biomass and bacterial to fungal ratios than green roof substrates in New York. Microbial levels may influence plant functionality and the authors suggest that microbes and fungi on green roofs may be a functionally underestimated component of these systems. The same principle was emphasized by [125] in research on the immensely positive role of microbes in GR installations, noting that plants in natural habitats benefitted from interactions with the fungi and bacteria of the local microbiome and demonstrated improved means of survival and productivity. Clearly, careful investigation of this issue is needed so as to achieve the best possible outcomes for the ESBS on green roofs.

The Endangered Community of the Eastern Suburbs Banksia Scrub
The ESBS is confined to the coastal suburbs of Sydney [126] and is a shrub-dominated and largely sclerophyllous heath community on nutrient-poor Aeolian dune sand. The ESBS consists of a minimum of 63 plant species and is near-extinct as an ecological community [127]. Common species in the ESBS community include over-storey 4 -5 metre trees like the Heath-leaved Banksia (Banksia ericifolia), Old Man Banksia (B. serrata)-see Figure 5 and Coast Teatree (Leptospermum laevigatum); shrubs and ground cover species like Epacris spp., Pink Wax Flower (Eriostemon australasius), Variable Sword Sedge (Lepidosperma laterale, Tree-Broom heath (Monotoca elliptica) and Grass Trees (Xanthorrhoea resinifera). The ESBS is a predominantly fire-adapted community, highly dynamic and readily regenerates from re-sprouting and germination from the soil seed bank [6] [128] and [129].
The NSW Scientific Committee is currently reviewing a proposal to expand the definition of the ESBS to include certain vegetation communities near Bundeena, a southern suburb in Sydney surrounded by the Royal National Park [55] [130]. Local ecologists [131] suggest the endangered ecological community is regarded as particularly vulnerable to climate change due to its very limited north-south range and that ESBS remnants could cease being viable at remaining sites. Rather than focus on impacts per se, the authors applied a climate adaptation approach to conservation assessment through a case study at Queen's Park in inner-eastern Sydney. The authors concluded that new options for managing the site were needed, although using roof tops was not considered by the researchers or local stakeholders. update also indicates that six priority sites have been declared for active management, which will include prescribed burns.
Dealing with the ESBS's status involved a recovery plan which was approved in 2004, followed by best-practice guidelines [6] and management plans for a limited number of sites. The degree of regeneration from the persistent seed bank following restoration at one of the medium size sites was remarkable according to [133]. The total number of native species increased only slightly over the six years (2001-2007) from 31 to 35 species, but a dramatic increase in abundance occurred across the site [127] and [134].
The removal of weeds or heavy shade was sufficient to trigger germination of a range of species in most of the managed quadrats, although germination still occurred in untreated plots without the addition of seeds. Practitioners also found that discarding the thick layer of leaf litter was important in that it revealed the more natural Aeolian sandy soil and simultaneously removed weedy and nutrient-rich top soils. This is an important finding given the intention to raise species indicative of the ESBS on trial green roofs in coastal Sydney, and given that plant resilience in the ESBS's case has been unusually good [133]. It is a characteristic that raises confidence of achieving conservation objectives in the relatively testing environment of green roofs.

Proposed Research
The research should be long term to allow plant conditions and characteristics to be monitored, and especially to observe a full cycle of seed planting and germination, growth into young seedlings, maturation, further seed generation by the maturing plants and natural germination. The research in this case would have the overall goal of supporting Objective 3 of the ESBS Recovery Plan which is to "To restore, and where practical, connect and enlarge remnants of ESBS through appropriate management" [135] and help to combat the apparent on-going loss of species and diminishing gene pool in some fragments of the ESBS [127]. Such research is also applicable broadly to green roofs and biodiversity conservation as well as offering more general ecosystem benefits and will necessitate cross-disciplinary contact among UNSW researchers as well as collaboration with external industry, government and community partners.

Detailed Research Objectives
The

Cultivation Principles
The research involves designing a biotope or habitat of small-scale spaces on the GRs [137]. The researchers note that the key to success lies in examining site conditions carefully, possessing detailed knowledge of the species to be transferred to the GRs, acknowledging the small-scale characteristics that occur in the natural environment which provide niches for various forms of wildlife and paying attention to soil and microbial composition. The guiding principle would be to match conditions on the green roof closely with those enjoyed by the natural community [31] [138] and [120]. If this is achieved, it may become possible to replicate "unique ecological processes and ecological linkages in the space" [137]. Other cultivation principles that need to be borne in mind for the preparatory period for the research are summarized in Table 2.

Installation Protocol
The proposed site for the research consists of two 72 m 2 roofs at first floor level, 4 metres above ground, on the University of New South Wales Kensington campus in Sydney. In addition to installing a safe external access to both roofs and providing GR infrastructure by way of waterproof membranes, drainage layers, root barriers and substrate, a number of other measures, chiefly physical, will need to be implemented as part of the GR experimental design. They are summarized below in Table 3.

Monitoring Plant Performance
Both plant and faunal characteristics must be monitored to gauge achievement This relationship was pointed out by [90] and [109] but may not apply to the ESBS since lush vegetation can exist on infertile and thin sands. Nevertheless, the key should be to match conditions on the GR closely with those enjoyed by the natural community, a factor likely to be critical to success [31] [138] and [120].
Test ground level soil mantle before installation on the GR The principle of matching substrate depth and characteristics, noted immediately above, should also be applied to the soil micro-organisms and fungal populations of the ESBS Aeolian sand mantle to ensure representative communities are included in the GR substrate [124].
Test landscape materials for toxins Landscape materials should be checked for toxins (see Installation Protocol below and Table 3).

Fine landscaping details
Employ features on the GR found at grade such as rocks and tree branches to provide suitable microclimates as well as shade and sunny areas to encourage plant (and faunal) diversity [60]. Features like tree branches may need to be fabricated from lightweight materials to minimise roof loads and large rocks will need to be avoided.
Substrate depths Vary substrate depths in order to assess the ESBS community's response to this variable.

Maximise ecosystem services
The planting design should attempt to increase the functional diversity of species to amplify ecosystem output, while being aware of GR limitations regarding substrate depth.

Irrigation
One researcher's findings [138] regarding improved plant survival with supplementary watering do not apply to Sydney's rainfall or the drought-adapted nature of the ESBS. However, one test cell may be irrigated for experimentation purposes.

Select plant species
Select plant species to reflect concerns about the lack of fire regimes, species which have difficulty forming soil seed banks and those plants which are no longer present on remnant sites. Selection to be conducted in association with Randwick Community Nursery and the School of Biological, Earth and Environmental Sciences at UNSW. There is an overarching limit on substrate depth and plant height dictated by the roof load-bearing capacity.
Cultivation guidelines Prepare cultivation guidelines including the need for hand weeding and for collecting and germinating seed to test the viability of green roofs as seed orchards.
Implementation and outcomes Tracking progress in achieving objectives will require a detailed monitoring plan (see Section 5.4 below). Table 3. Installation process.

Installation element Comments
Establish the test modules Following infrastructure installation: 1) Subdivide the roofs into test modules (see Figure 6); 2) Select suitable species from the ESBS community for planting; Select substrates 1) Select both natural and artificial substrates, e.g. one directly from an ESBS site, an artificial aggregate version of the natural substrate and an industry manufactured substrate, each at 3 different depths e.g. 50 mm, 100 mm and 150 mm; 2) Discard rich leaf litter; 3) Note comment in 5.1 regarding avoidance of materials which may contain hazardous or toxic substances.

Test soils for lead content
Many inner suburban Sydney soils still contain lead particles from the era of lead in petrol. This may be a factor in selecting substrate options for the test beds (https://www.smh.com.au/national/nsw/lead-levels-in-sydney-soil-dangerously-high-20170119-gtuea6.html).

Test landscape materials for toxins
Landscape materials should be checked for toxins, eg test cell edging, whether timber or plastic. Non-natural substrate materials should similarly be checked since not all substrate will be obtained from in-situ ESBS remnants (https://www.betterhealth.vic.gov.au/health/HealthyLiving/copper-chrome-arsenic-cca-treated-timber). of objectives. Monitoring is a major undertaking in the research although the location of the GRs at UNSW is a major logistical advantage for the researchers.  Seed will need to be collected and germinated to assess ESBS species' reproductive viability, and indices developed to assess survival, viability and growth and checked against performance benchmarks such as leaf area index, plant coverage and plant abundance. Volunteer plants appearing on the two experimental roofs will also be identified and monitored and will either be left in situ or removed if they appear to be crowding out the ESBS. Performance comparisons will be Open Journal of Ecology conducted with natural remnant populations and benchmarked against at least two of the remnant species locations and an appropriate frequency of field inspection undertaken. Equipment will also need to be installed for monitoring soil moisture and insolation levels while soil microflora will need periodic sampling [124]. Key performance indicators will also be constructed to measure the achievement of objectives, such as germination rates/percentage, viability and growth rates of seedlings.
The GRs attractiveness to fauna will also be monitored through data collection and a set of suitable diversity indices developed to evaluate faunal biodiversity (e.g. avian, reptilian, invertebrate) in collaboration with research students from the School of BEES at UNSW. These parameters will also be compared against

Conclusions
Planetary ecosystems provide a wide range of services both to humanity and to life in general, estimated to be worth trillions of dollars each year. Green roofs contribute multiple benefits to these services, including reducing stormwater run-off, moderating heating and cooling loads, carbon sequestration and storage, aesthetic and passive recreation opportunities and biodiversity conservation. It is the last named area that this paper is concerned about and there is growing realization that GRs may offer many opportunities for biodiversity conservation if they are managed with this objective in mind. Thus, researchers like [54] [109] and [138] are raising their objectives from simply obtaining bio-physical benefits from plants on GRs to selecting those which favour flora and fauna conservation, thus extending the function of rooftops as a vehicle for greening. At the same time, such research is showing how the difficulty of safeguarding threatened or endangered flora and fauna at ground level in our urban areas can be countered by harnessing currently vacant spaces on GRs.
In the absence of any Australian material focused on using GRs to rescue threatened or endangered flora, our research proposes to manage a currently threatened community of native Australian plants and support them using GRs.