Next Article in Journal
Iron Oxide Nanoparticles for Photosynthetic Recovery in Iron-Deficient ‘Micro-Tom’ Tomato Plants
Previous Article in Journal
Electric Field Effects on Microbial Cell Properties: Implications for Detection and Control in Wastewater Systems
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Environments
Volume 12
Issue 10
10.3390/environments12100345
environments-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Evaluating Native Grassland Species for Application in Extensive Green Roofs in Japan †

by
Tsukasa Iwata
,
Ryosuke Shimoda
,
Terumasa Takahashi
and
Kiyoshi Umeki
*
Graduate School of Horticulture, Chiba University, Chiba 271-8510, Japan
*
Author to whom correspondence should be addressed.
†
This paper is an extended version of our paper published in the Proceedings of the International Conference 2025 on Spatial Planning and Sustainable Development (SPSD 2025), Hankyong, South Korea, 7–9 August 2025, originally entitled “Assessing the performance of native grassland species on extensive green roofs under Japanese climatic conditions”.
Environments 2025, 12(10), 345; https://doi.org/10.3390/environments12100345
Submission received: 28 August 2025 / Revised: 21 September 2025 / Accepted: 24 September 2025 / Published: 26 September 2025
(This article belongs to the Topic Climate, Health and Cities: Building Aspects for a Resilient Future)
Browse Figures
Review Reports Versions Notes

Abstract

Extensive green roofs (EGRs) are increasingly recognized as multifunctional components of urban green infrastructure. In recent years, interest is growing in the use of native grassland species as alternatives to conventional green roof plants, both to enhance ecological function and to support biodiversity conservation. This study evaluated the performance of six native grassland species on extensive green roofs by assessing their growth characteristics (cover, survival, and flowering) throughout a single growing season (May–November 2024). We used three different substrates that differed in nutrient level: a nutrient-rich reused substrate, a mixed substrate, and a nutrient-poor perlite-based substrate. The results indicated that most species successfully established across all substrate types, although patterns in growth and mortality varied. Substrate nutrient levels strongly influenced early growth, but their long-term effects may diminish as nutrient dynamics stabilize over time. These findings suggest that native grassland species represent promising alternatives to conventional green roof plants in Japan, with several species showing strong adaptability to EGR conditions. Substrate nutrient management is essential for balancing plant growth, biodiversity, and maintenance requirements. This study contributes to improving the ecological performance and long-term sustainability of green roofs in urban environments.

1. Introduction

Urban green infrastructure has received growing attention as a strategy to mitigate the adverse environmental impacts of urbanization. Among various approaches, green roofs are widely recognized for providing essential ecosystem services, including reducing stormwater runoff [1,2], alleviating urban heat island effects [3,4,5], sequestering carbon [6,7,8], and enhancing urban biodiversity [9,10,11,12]. However, the extent of these benefits largely depends on plant species selection [13].
In Japan, extensive green roofs (EGRs), which have a shallow substrate depth (20 cm or less), are the most common type due to their minimal structural requirements and low maintenance needs [14]. These roofs are typically planted with monocultures of non-native Sedum species, selected for their drought tolerance, shallow root systems, and spreading growth habit. However, under Japan’s hot, humid summers and prolonged rainy seasons, Sedum species often show inconsistent performance [15,16]. Their limited adaptability may reduce the ecosystem services that EGRs can provide in the Japanese context. Therefore, there is a need to explore alternative plant species that are better suited to local climatic conditions. Native herbaceous species, already adapted to these conditions, are promising candidates.
Previous research suggests that increasing plant diversity enhances the multifunctionality of green roof ecosystems [17]. Moreover, using native species can contribute to biodiversity conservation, particularly for species whose habitats have been lost due to land-use change. In Japan, semi-natural grasslands have declined sharply in recent decades, resulting in the loss of habitat for many herbaceous species [18]. These ecosystems are unique in that they are human-maintained and require continuous land-use to persist. This dependency has led to the development of distinctive plant communities with high conservation value. Evaluating the potential of these species for green roofs provides an opportunity for ex situ conservation [19].
Grassland ecosystems are generally nutrient-poor, a trait they share with EGRs. According to the habitat template approach [20], species from natural environments with similar abiotic conditions are likely to be suitable for artificial systems like green roofs. Many previous studies evaluated potential native plant communities for green roofs in other contexts, such as Mediterranean arid regions [21,22], North American prairies [23,24], and Japanese rocky coastal habitats [25]. However, few studies have investigated native plant selection for EGRs in Japan [25,26].
Another promising strategy to enhance the sustainability of green roofs is to incorporate spontaneously colonizing species. These species can form resilient plant communities and reduce maintenance costs, as they are more likely to establish successfully and regenerate naturally. In Japan, many EGRs originally planted with Sedum have transitioned to communities dominated by spontaneous colonizers [16]. One of the most observed spontaneous species is Imperata cylindrica, a dominant grass in semi-natural grasslands [15]. Although its use in green roof vegetation is gradually increasing [27], its performance under typical EGR conditions has not yet been systematically assessed. Since EGRs are characterized by low nutrient availability and limited water retention, it is important to evaluate whether I. cylindrica and other grassland species can tolerate such conditions.
In addition to plant species selection, the choice of substrate materials substantially influences the ecosystem services provided by green roofs [28]. Previous life cycle assessment studies have recommended the use of recycled materials in green roof substrates, as this reduces the consumption of natural resources (e.g., peat moss) and lowers associated carbon emissions [29]. While many studies have evaluated recycled materials as components of green roof substrates [30,31,32,33], research on reusing entire green roof substrates as a material for new substrates remains scarce [34]. To achieve a truly circular economy in green roof construction, it is therefore necessary not only to use recycled materials from other industries but also to assess the potential for reusing green roof substrates themselves.
This study aims to evaluate the performance of selected native grassland species on extensive green roofs, using the habitat template approach as a guiding framework. While these species may be theoretically suited to green roofs, the environmental conditions on EGRs are often harsher than those in natural habitats, and responses may vary among species. We focused on I. cylindrica as a key species due to its frequent spontaneous colonization on green roofs. In addition, identifying companion species that can coexist and thrive alongside I. cylindrica is crucial for forming biodiverse and resilient plant communities. Moreover, we evaluated the feasibility of reusing substrates from existing green roofs for renewing green roofs, which could also enhance ecosystem services.

2. Materials and Methods

2.1. Experimental Site

We conducted a green roof experiment (May to November in 2024) under real-world conditions on the rooftop of Sumida Bunka Junior High School, located approximately 7 km from central Tokyo. The site is situated on a four-story building at a height of 14 m above ground level and receives full sunlight throughout the day. Figure 1a shows the monthly average maximum and minimum temperatures, as well as precipitation data from the past four years (2021–2024), recorded at a nearby weather station (Edogawa seaside, 8.7 km from the experimental site) [35]. The climate in Sumida is generally mild, with hot and humid summers. Figure 1b presents the daily average temperature and precipitation during the high-temperature season (July to September) of the experimental period. During this time, precipitation was relatively consistent, and dry periods were short, typically lasting no longer than two weeks.

2.2. Experimental Design

An abandoned Sedum-planted EGR (established in 2001, 54 m2 in area) was available at the experimental site. We reused the existing greening frame (made of concrete, 20 cm in height) and the underlying substrate. The reuse aimed to reduce disposal costs and evaluate the feasibility of substrate recycling. The reused substrate was considered to have relatively low nutrient content compared to most commercial green roof substrates, due to nutrient loss through runoff, making it suitable for comparative testing. The existing vegetation and substrate layer were removed, and twelve experimental quadrats (1 m × 1 m each) were established using a randomized block design, with four replicates for each of the three substrate types (Figure 2). An additional 24 quadrats were established for other treatments, including I. cylindrica monocultures and unplanted controls (see blank areas in Figure 2). However, data from these treatments were excluded from the present analysis due to experimental difficulties, which are described in a later section. Each quadrat was enclosed with a plastic frame (15 cm high, 2 mm thick). From bottom to top, the layers consisted of: a 20 mm gravel drainage layer (6–11 mm particle size), a 0.8 mm-thick non-woven fabric filter layer, and a 100 mm substrate layer (Figure 3). The three substrate treatments were as follows: 1. reused substrate from the original green roof (R), 2. a 1:1 volume mixture of reused substrate and a commercially available perlite-only substrate [36] (PR), 3. a commercially available perlite-only substrate alone (P). The original composition of the reused substrate was unknown due to the loss of records at the time of installation. However, visual assessment suggested that it was essentially loam soil with some pumice grains, and was similar to typical commercial gardening soils in Japan. The reused substrate was sieved through a 6 mm mesh to remove coarse debris, including weed roots. The perlite-only substrate is commercially distributed for green roofs in Japan and is used in several public and private open green spaces without mixing with other materials. To our knowledge, perlite-only substrates have not been used in previous studies, but we used them as a control treatment because of their practical use in multiple green roof projects, and our pilot study confirmed their suitability as an extensive green roof substrate.
To assess the initial physicochemical properties of the substrates, samples were collected on 10 May and analyzed in a laboratory at the Faculty of Horticulture, Chiba University. The measured parameters included particle size distribution, wet and dry bulk density, maximum water-holding capacity (WHC), pH, electrical conductivity (EC), ammonium nitrogen (NH4+-N), nitrate nitrogen (NO3−-N), available phosphorus (P2O5), and the total carbon-to-nitrogen (C:N) ratio. Particle size distribution was determined by dry sieving through a series of standard sieves (8, 4, 2, 1, 0.5, 0.25, 0.125, 0.063 mm). WHC was calculated by subtracting the dry weight (after oven drying at 60 °C for 72 h) from the weight of water-saturated samples. EC was measured using an EC meter (ES-51, Horiba Ltd., Kyoto, Japan) in a 1:2.5 soil-to-water extract, and pH was measured using a pH meter (D-51, Horiba Ltd., Kyoto, Japan) in the same extract following EC measurement. NH4+-N was quantified using a spectrophotometer (PD-303S, Apel Ltd., Saitama, Japan) based on the indophenol blue reaction, with potassium chloride as the extraction solution. NO3−-N was determined by ultraviolet spectrophotometry (ASUV-1100, As One Corporation, Osaka, Japan) using the same extract as that used for NH4+-N. P2O5 was also measured with the same spectrophotometer, using the molybdenum blue method after extraction with Bray No. 2 solution. Total carbon and nitrogen concentrations were measured using a CN analyzer (MT-700, Yanaco Ltd., Kyoto, Japan).

2.3. Planting Selection and Measurements

The experimental vegetation consisted of six native grassland species: Imperata cylindrica, Patrinia scabiosifolia, Dianthus superbus var. longicalycinus, Sanguisorba officinalis, Prunella vulgaris subsp. asiatica, and Adenophora triphylla var. japonica. These species were selected based on previous studies of I. cylindrica-dominated grasslands [37,38], as well as earlier research on the application of I. cylindrica in green roof greening [39,40] (Table 1). P. scabiosifolia and D. superbus var. longicalycinus are listed as threatened species in the Red Data Book Tokyo 2023 [41], representing native species of conservation concern.
The planting arrangement followed the layout shown in Figure 4. We planted one individual per quadrat for I. cylindrica and three individuals per quadrat for the other five species. Only one individual of I. cylindrica was used because this species spreads sporadically via rhizomes, making it difficult to distinguish between multiple individuals once established. I. cylindrica, D. superbus var. longicalycinus, and S. officinalis were planted using plug seedlings, while P. scabiosifolia, P. vulgaris subsp. asiatica, and A. triphylla var. japonica were planted using potted seedlings (in 9–10.5 cm diameter pots) due to limited availability of plug seedlings. After planting, a mulch layer composed of the same gravel used in the drainage layer was applied to prevent substrate displacement by wind. All setup procedures were completed on May 10 (Figure 5). No irrigation was provided after planting, except during the initial installation, and weeding was not conducted until 5 September.
To evaluate species responses to different substrate types, monthly surveys were conducted from May to November on 16 May, 19 June, 9 September, 15 October, and 12 November. Photographs of each quadrat were taken using a digital camera from a height of 150 cm. Flowering status and the survival of each individual were also recorded. The plant cover area of each individual was measured by processing the photographs with Fiji ImageJ software (version 1.54p). Segmentation was performed manually using the Line tool.
Initially, weed cover was also recorded, but by August, overgrowth of weeds made it difficult to assess the growth of the planted species. Therefore, plant cover data were not collected in August. The most common weed species was Ipomoea hederacea, a climbing non-native species that twined around the planted species and likely suppressed their growth. On 5 September, all weeds were removed, and subsequent surveys focused exclusively on the planted vegetation. The removed weeds were dried and weighed as follows: due to the large sample size, representative subsamples were oven-dried at 60 °C until constant weight, and the dry-to-fresh weight ratio obtained from these subsamples was applied to the total fresh weight to estimate the total dry weight.

2.4. Statistical Analysis

Statistical Model

To quantitatively describe the dynamic process of two important aspects (i.e., size change and survival of plants) of vegetation development, the state of individual plants was modeled using a Bayesian state-space model. The state of each individual plant was expressed by size (i.e., plant cover = vertically projected area of foliage) and survival state (i.e., probability that a target plant is living), and was assumed to change in discrete time steps. The length of each time step was one month.
State-space models consist of system models, which describes the development of the target objects, and observation models, which describe how observation values are obtained from “true” states including the effect of observation errors. The states of plants were estimated in all time steps when the plants are assumed to be alive, but the states do not need to be associated with observed data (i.e., missing observations are allowed and the states corresponding to the missing observations are just estimated). We employed a “Bayesian” state-space model because it can model whole developmental process at a time, appropriately accounting for temporal and spatial autocorrelations in the data and separating measurement error and uncertainty in the system. The very high flexibility of the Bayesian method makes such modeling possible.
In short, the system model of the state-space model mathematically describes the change in the size index of an individual plant (i.e., the logarithm of estimated “true” cover of an individual plant, which is a state variable of the model) as a cumulative summation of the discrete-time monthly relative growth rates of plant cover (plus system uncertainty), which is determined by a temporal growth tendency, the effect of substrate, and spatially autocorrelated random effect. Survival state of an individual plant, another state variable, is assumed to be modulated by monthly mortality rate, which is determined by a temporal growth tendency, plant cover, the effect of substrate, and spatially autocorrelated random effect. The model was applied to data for each species. Detailed descriptions of the state-space model and Bayesian inference are provided in Appendix A. The raw data used for the statistical analyses are available in Supplementary Materials S1 and S2.

3. Results and Discussion

3.1. Substrate Physicochemical Properties

The physicochemical properties of the different substrate types are shown in Table 2. Overall, both physical and chemical values except pH were highest in R, lowest in P, and intermediate in PR. All substrates were acidic, with the pH of R slightly falling outside the FLL guideline range (6.0–8.5) [45]. However, a pH of 5.6 is not uncommon in Japan’s natural grassland soils [46]. Nutrient levels were higher in R and PR, with available phosphorus exceeding the FLL guideline for single-layer extensive green roofs (<200 mg/L) by a factor of four to five. Nitrogen concentrations (NH4+-N + NO3−-N), however, remained within the recommended limit (<80 mg/L) across all substrates. Substrate P showed notably low nutrient levels, with total nitrogen and phosphorus amounting to only 3.5% and 0.2% of those in R, respectively. This is expected, as perlite—the sole component of P—is a heat-expanded mineral (primarily silicon dioxide) that contains virtually no nutrients. Nevertheless, as the soil samples were collected after rainfall, it is possible that rain-soluble nutrients from atmospheric deposition had entered the substrate. C/N ratios were 10.05 for substrate R and 10.93 for substrate PR. No measurement was taken for substrate P, as it contained no organic matter. A C/N ratio of around 10 is considered relatively low [47] and likely reflects the decomposition of organic matter that was initially incorporated into the substrate. All substrate types showed low bulk density in both dry and saturated conditions, while maximum water-holding capacities were high, exceeding the FLL guideline range (≥35% and ≤65%).
In terms of particle size distribution, all substrates mainly consisted of fine particles, with more than 75% of the mass composed of particles smaller than 0.5 mm in diameter. Consequently, all substrates were finer than the FLL guideline range (Figure 6). Among them, substrate R was the finest, P was the coarsest, and PR was intermediate. However, substrate P, the perlite-only substrate, was still much finer than the perlite typically used as a component of mixed substrates, which generally has a particle size of 3–6 mm [48]. This characteristic likely contributes to the capacity of substrate P to support vegetation growth, even though it is composed solely of perlite. These differences are likely reflected in WHC gradients (Table 2), since finer particles generally retain more water [49].

3.2. Plant Performance

This study demonstrated that the selected native grassland species can establish and grow on EGRs, although growth responses varied among species depending on substrate characteristics.

3.2.1. Performance of Imperata cylindrica

No mortality was observed for I. cylindrica. Accordingly, only results relating to size change are presented below.
Individuals of this species showed exponential growth until September on substrates PR and R, followed by a decline or stagnation later in the season (Figure 7a). On substrate P, however, growth was limited and inconsistent throughout the study period. The species’ RGR differed significantly between substrates during the early to mid-season (until September), with higher RGRs in PR and R compared to P (Figure 7b).
These results suggest that I. cylindrica is well adapted to EGR conditions—characterized by drought, low nutrient availability, and shallow soil—but requires a longer establishment period in nutrient-poor substrates. Plant nutrient requirements generally correspond to the environmental conditions of their native habitats [50], and can be interpreted through habitat-based indices such as those developed by Ellenberg [51], where values for soil fertility range from 1 to 9, with higher values indicating preference for nutrient-rich conditions. Among the species studied here, Ellenberg nutrient indices are available only for I. cylindrica and S. officinalis, as most Japanese native species are not included in these primarily European-derived classifications. The Ellenberg index for I. cylindrica is 4.7 [52], suggesting that it prefers moderately fertile substrates. Therefore, lowering substrate nutrient levels may help limit the growth of I. cylindrica.
In Sumida region of Tokyo, I. cylindrica has been observed to spontaneously dominate green roofs and suppress co-occurring species [15]. Using low-nutrient substrates may help support more diverse plant communities by limiting the dominance of this species—a concept consistent with biodiversity-oriented grassland restoration practices that reduce soil fertility through topsoil removal [53].

3.2.2. Performance of Patrinia scabiosifolia

No mortality was observed for P. scabiosifolia. Accordingly, only results relating to size change are presented below.
Individuals of this species displayed growth patterns similar to those in I. cylindrica, but reached its peak canopy cover earlier, particularly on substrates PR and R (Figure 8a). Significant differences in estimated RGR between P and R, and between P and PR, were observed only in the early season (until mid-July) (Figure 8b). This may be partly due to phenological factors, as flowering appeared to coincide with a decline in vegetative expansion (Figure 8c). Trade-offs are a fundamental concept in life-history theory [54] and have been applied to a wide range of organisms, including plants [55]. In plants, vegetative growth (i.e., expansion of ground cover) typically precedes the onset of reproductive growth, which shifts resource allocation toward the production of flowers and seeds and consequently halts further cover expansion.
Although the absolute growth of P. scabiosifolia was lower on substrate P compared to PR and R, all individuals survived on all three substrates, and more than 50% flowered at least once during the study period. These findings suggest that P. scabiosifolia, like I. cylindrica, can be a suitable species for EGRs in Japan’s urban settings.

3.2.3. Performance of Dianthus superbus var. longicalycinus

Only two individuals of D. superbus var. longicalycinus died during the experimental period on substrate R. Although the whole model including size- and survival-state-dynamics was applied for this species, the scarcity of mortality events may make the detailed analysis of mortality practically meaningless. Therefore, only results relating size change are presented below.
Most individuals of this species on substrates PR and R reached peak cover in July and declined thereafter. In contrast, some individuals on substrate P exhibited steady growth even later in the season (Figure 9a). The sustained late-season cover may reflect this species’ semi-evergreen nature. RGR differed significantly between P and the other substrates during the first month after planting, and between P and R during the last month (Figure 9b). While early growth was lower on P, this trend reversed later in the season—though the difference between substrate P and PR was not statistically significant. These results suggest that D. superbus var. longicalycinus performs well under nutrient-poor conditions. This is consistent with its natural habitats—stony riverbeds, coastal sand dunes, and ruderal grasslands—where soils are typically nutrient-poor [56,57].
The growth pattern and RGR shifts observed in this species may also be linked to flowering dynamics. Flowering occurred in about 50% of individuals in PR and R, but less than 25% in P (Figure 9c). As with P. scabiosifolia, more abundant flowering may have contributed to reduced vegetative growth in the latter half of the season. Notably, two individuals died on substrate R, whereas no mortality was observed on P or PR. As a relatively short-lived perennial [58], D. superbus var. longicalycinus may behave as a biennial under suboptimal conditions, particularly where nutrient-rich substrates promote rapid growth and early flowering. Although the sample size was limited, the results suggest that nutrient-poor substrates may provide more sustainable conditions for this species.

3.2.4. Performance of Sanguisorba officinalis

Only one individual of S. officinalis died during the experimental period on substrate R. Although the whole model including size- and survival-state-dynamics was applied for this species, the scarcity of mortality events may make the detailed analysis of mortality practically meaningless. Therefore, only results relating size change are presented below.
For S. officinalis, significant differences in RGR were observed between substrate P and PR, and between P and R in the first month. Additional significant differences between P and R were found in the third and fifth months. However, all treatments exhibited a similar overall RGR trend—declining toward mid-season and recovering toward the end of the season (Figure 10b). This resulted in comparable patterns of plant cover development, although cover on substrate P remained lower than on PR and R throughout the experimental period (Figure 10a). These findings are consistent with S. officinalis’ Ellenberg index for soil fertility (4.0), which is slightly lower than that of I. cylindrica (4.7) [52]. The similarity in their nutrient preferences suggests that S. officinalis, like I. cylindrica, is capable of tolerating a broad range of nutrient conditions but tends to perform better under moderately fertile conditions.
The number of flowering individuals of S. officinalis varied among treatments, with three and four in substrates P and R respectively, and eleven in PR (Figure 10c). Additionally, only one individual died on substrate R between July and August (Figure 10a). These inconsistent results may be attributable to weed competition. Weeds were allowed to grow until August and were only removed in September due to observational difficulties. S. officinalis prefers open, sunny environments and is sensitive to poorly ventilated conditions [58]. Thus, excessive weed growth during summer may have created suboptimal conditions, limiting light and airflow. The subsequent recovery in RGR may be related to the removal of weeds, which likely improved microclimatic conditions. Overall, the results indicate that S. officinalis can tolerate a wide range of nutrient conditions if the growing environment remains open and well-ventilated.

3.2.5. Performance of Prunella vulgaris subsp. asiatica

Prunella vulgaris subsp. asiatica showed higher mortality than the other species examined, particularly on substrate P. Some individuals died in June, shortly after planting, irrespective of substrate type, whereas others died later in the season (Figure 11a). Among the surviving individuals, canopy cover declined in June, recovered in July, and then remained relatively stable for the rest of the season. Correspondingly, RGR estimates were negative across all substrates in the first month, reflecting the initial decline in cover. A significant difference in RGR between substrates P and R was detected only during this period (Figure 11b).
Mortality patterns largely followed these observations; mortality was high in the first month across all substrates and increased again toward the end of the season (Figure 11c). Although mortality tended to be higher on substrates P and R than on PR, the differences were not statistically significant. In addition, most individuals flowered immediately after planting, regardless of substrate type (Figure 11d). These results suggest that some plants failed to regrow after flowering. This is consistent with the species’ natural life history: P. vulgaris subsp. asiatica often dies back after blooming and regenerates via rhizomes [58]. Thus, transplanting flowering individuals may have hindered establishment, leading to the high mortality observed in June and the limited canopy development thereafter. However, the cause of the elevated mortality later in the season remains unclear.
The species P. vulgaris, to which subsp. asiatica belongs, has been reported to be drought-intolerant on green roofs. Previous studies showed that P. vulgaris could not survive without irrigation [59] and exhibited lower survival rates on substrates with lower water-holding capacity (WHC, 59.9%) than on those with higher WHC (116.5%) when water was not supplied [49]. In the present study, the WHC values of the substrates were relatively low (65.3–74.0%), but rainfall was sufficient (approximately 100 mm per month in August). These findings suggest that short-term water availability was unlikely the sole cause of mortality. Longer-term monitoring into a second growing season, as well as additional experiments, are needed to clarify the vegetation dynamics and to determine the optimal conditions for establishing P. vulgaris subsp. asiatica on extensive green roofs (EGRs).

3.2.6. Performance of Adenophora triphylla var. japonica

A. triphylla var. japonica exhibited markedly different survival rates depending on substrate type (Figure 12a,c). On substrate P, nearly all individuals survived (11 out of 12), whereas only two individuals survived on PR, and none on R. The higher mortality observed on PR and R may be associated with elevated phosphorus levels. Yamada, Nemoto et al. [60] reported that soils in semi-natural riverbank grasslands—the natural habitat of A. triphylla var. japonica—contained significantly lower levels of available phosphorus (4.4 ± 0.6 mg/100 g) compared to the PR and R substrates (>120 mg/100 g).
On substrate P, most individuals (10 out of 12) flowered during the experimental period (Figure 12d). Accordingly, temporal changes in plant cover appeared to reflect phenological development: a gradual increase during the vegetative growth phase, a temporary decline following flowering (reproductive phase), and a subsequent recovery later in the season as vegetative growth resumed (Figure 12a).
This study demonstrated the feasibility of establishing selected semi-natural grassland species, Imperata cylindrica and the companion species, on EGRs, and identified substrate characteristics—especially nutrient levels—as a key factor in successful plant establishment. While early growth was more vigorous on nutrient-rich substrates (PR and R), most species adapted well on the nutrient-poor P substrate. A. triphylla var. japonica in particular showed clear preference for low-phosphorus conditions, whereas P. vulgaris var. asiatica performed poorly on the same substrate. In the case of P. vulgaris var. asiatica, factors such as planting timing and substrate moisture may have contributed to the outcome.
In addition to these species-specific performances, the mixture of multiple species with different phenological characters can provide another ecosystem service. Flowers supply food resources for pollinators such as bees and butterflies, thereby contributing to urban biodiversity [11]. Planting multiple species, as in this study, has a greater impact than Sedum monoculture because the selected species showed different peak flowering times (see Figure 8c, Figure 9c, Figure 10c, Figure 11c, and Figure 12c), which resulted in an extended overall flowering period. This phenological complementarity may be influenced by earlier flowering on green roofs compared to natural habitats, possibly due to global warming and the hot, dry-stressed conditions typical of rooftops [21,61]. We observed earlier flowering in P. scabiosifolia and P. vulgaris subsp. asiatica compared to descriptions in the literature (Table 1). Therefore, it is important to evaluate a wider range of species under actual green roof conditions in order to sustain and improve the ecosystem services provided by green roofs.

3.3. Feasibility of Substrate Reuse and Nutrient Dynamics of EGRs

From a short-term plant growth perspective, reusing substrate from long-neglected green roofs appears feasible. However, due to challenges in weed control and the labor-intensive nature of substrate preparation, this approach is not considered practically viable. In this study, substantial weed emergence was observed in the R and PR substrates (Figure 13), likely originating from a persistent seed bank in the original green roof soil. Suppressing such germination typically requires a mulch layer at least 10 cm thick [62], which is impractical in lightweight rooftop systems due to structural load limitations. Additionally, substrate preparation involved sieving, a highly labor-intensive process; in this case, it took five to six people more than one to two weeks to complete.
Despite prolonged exposure to wind and rain—conditions that typically promote nutrient leaching [63,64]—the reused substrate (R) still contained notable levels of inorganic nitrogen (NH4+-N: 31.6 mg/L; NO3−-N: 17.1 mg/L) and an excessive concentration of available phosphorus (1096 mg/L). One possible explanation is that residual organic matter, such as plant debris and microbial biomass, decomposed during substrate preparation (i.e., removal and sieving). Microbial activity, possibly triggered by increased temperatures, may have mineralized this organic matter, releasing inorganic nutrients.
Although nutrient-rich substrates can enhance early plant growth, they are generally unsuitable for green roof applications. For EGRs in particular, minimizing substrate nutrient levels is desirable, as excess nutrients can lead to environmental concerns (e.g., nutrient leaching) [64], promote excessive biomass production, reduce drought resilience [65], and increase maintenance requirements. Given these limitations, the use of fresh, weed-free substrate appears to be a more practical and effective option for green roof applications.
Further studies are necessary to determine the optimal nutrient concentrations, minimum water availability thresholds, and appropriate planting schedules for green roof vegetation. Longer-term research is also needed, considering year-to-year climatic variability—especially in precipitation—and changes in substrate properties over time. During the experimental period, summer precipitation was above average, which may have promoted plant growth on nutrient-rich substrates [66] while potentially reducing drought tolerance. Although nutrients may initially leach from substrates, they can also accumulate through dry and wet deposition [64]. Over time, nutrient levels may stabilize [67], suggesting that initial substrate composition may have limited influence on long-term vegetation dynamics. Ultimately, sustainable green roof design should focus on long-term ecosystem development rather than short-term performance.

4. Conclusions

This study evaluated the growth performance of native grassland species on extensive green roofs (EGRs) under Japanese climatic conditions, with a focus on how substrate characteristics influence plant establishment and survival. Using the habitat template approach, we demonstrated that multiple species native to semi-natural grasslands—including Imperata cylindrica, Patrinia scabiosifolia, Dianthus superbus var. longicalycinus, and Sanguisorba officinalis—can successfully establish and persist under typical EGR constraints such as shallow substrate depth, limited nutrients, and variable water availability. Substrate nutrient content was found to play a critical role in species performance. While nutrient-rich substrates (PR and R) promoted faster initial growth for most species, they were also associated with increased weed emergence and, in some cases, higher mortality—particularly for species adapted to low-fertility environments such as Adenophora triphylla var. japonica. Conversely, nutrient-poor substrate (P) supported the survival of species like D. superbus as well as A. triphylla var. japonica, although species such as Prunella vulgaris subsp. asiatica showed reduced performance, likely due to a combination of early flowering and insufficient establishment. Our findings suggest that low-nutrient substrates may offer long-term benefits for biodiversity-oriented green roof design by reducing maintenance requirements, limiting dominance by aggressive species, and creating conditions more suitable for stress-tolerant natives. However, species-specific responses highlight the importance of matching planting timing to species phenological traits. Given the variability of climatic conditions and the temporal dynamics of substrate nutrient availability, long-term monitoring is needed to assess ecosystem development and resilience, as the results presented in this study were based on a short-term experiment. Overall, this study contributes to expanding the plant palette for EGRs in Japan and supports the integration of native grassland species into urban green infrastructure. Such efforts can simultaneously enhance ecological function, promote ex situ conservation, and increase the resilience and sustainability of green roof ecosystems.

Supplementary Materials

The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/environments12100345/s1, S1: CSV file containing raw data for plant cover and flowering status over time, S2: CSV file containing coordinates of experimental quadrats.

Author Contributions

Conceptualization, T.I.; methodology, T.I., T.T. and K.U.; software, T.I. and K.U.; validation, K.U.; formal analysis, T.I. and K.U.; investigation, T.I.; resources, T.I.; data curation, T.I.; writing—original draft preparation, T.I. and K.U.; writing—review and editing, T.I. and K.U.; visualization, T.I.; supervision, K.U.; project administration, R.S.; funding acquisition, T.I. and R.S. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by the Environmental Conservation Division, Sumida City Office and Japan Science and Technology Agency, grant number JPMJSP2109. The APC was funded by Japan Society for the Promotion of Science (JSPS) KAKENHI, Grant-in-Aid for Scientific Research (C), grant number JP24K08969.

Data Availability Statement

The original contributions presented in this study are included in the article and supplementary material. Further inquiries can be directed to the corresponding author.

Acknowledgments

This article is a revised and expanded version of a paper entitled “Assessing the Performance of Native Grassland Species on Extensive Green Roofs Under Japanese Climatic Conditions”, which was presented at SPSD 2025, Hankyong, South Korea, 8 August 2025 [68].

Conflicts of Interest

The authors declare no conflicts of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results.

Abbreviations

The following abbreviations are used in this manuscript:
EGRsExtensive green roofs
RGRRelative growth rate
ECElectrical conductivity
WHCWater-holding capacity

Appendix A

The system model for size of an individual plant ( c t , i ; logarithm of estimated “true” plant cover of plant individual i at time step t) was assumed as:
  c t , i ~ N r g r t − 1 , i + c t − 1 , i , σ c 2
r g r t , i = γ t + α t , R d R + α t , P R d P R + r g , j
where N() is the normal probability density function; r g r t , i is discrete-time monthly relative growth rate (RGR) of individual plant i at time step t; γ t is the effect of monthly period t expressing a temporal growth tendency, which is common across individual plants; d R and d P R are dummy variables, respectively, indicating the substrate R and PR; α t , R is the effect of substrate R relative to substrate P at time step t; α t , P R is the effect of substrate PR relative to substrate P at time step t; r g , j is a spatially autocorrelated random effect for RGR. We assumed that γ t , α t , R , and α t , P R are random walks as:
  γ t ~ N γ t − 1 , σ γ 2
α t , R ~ N α t − 1 , R , σ α R 2
α t , P R ~ N α t − 1 , P R , σ α P R 2
where σ γ 2 , σ α R 2 , and σ α P R 2 are variances of the random walks. We include a spatially autocorrelated random effect ( r g , j ) in the model to deal with a spatial autocorrelation in r g r t , i between neighboring plants. We divided the study site into 40 blocks (each 1 × 1 m2), including not only the 12 experimental quadrats analyzed in this study but also non-target quadrats and the central aisle, which had the same 1 m width as the experimental plots. Five blocks at the right end of the site (see Figure 2) were not included in the analysis because they contained no experimental quadrats, and we assigned block-specific spatially autocorrelated random effects to the quadrats using an intrinsic conditional autoregressive model as:
  r g , j | r g , − j , τ ~ N ∑ k : j ~ k r g , k n j , σ r g 2 n j
where r g , − j = r g , k , k ≠ j , j ~ k means j and k are neighboring block, n j is the number of neighbors of j, σ r g 2 is a conditional variance. Note that γ t + α t , R d R + α t , P R d P R in Equation (A1) gives period- and substrate-specific RGR. The significance of the period-specific substrate effects can be accessed by the posterior distributions of α t , R and α t , P R .
The system model for the survival state ( s t , i ; probability that plant individual i is living at time step t) was assumed as:
  s t , i = 1 − m t − 1 , i s t − 1 , i
m t , i = i n v _ l o g i t μ t + β c e x p c t , i + β t , R d R + β t , P R d P R + r m , j
where m t , i is monthly mortality rate of individual plant i at time step t; inv_logit() is the inverse logit function; μ t is the effect of monthly period t; β c is the effect of plant size; β t , R is the effect of substrate R relative to substrate P at time step t; β t , P R is the effect of substrate PR relative to substrate P at time step t; r m , j is a spatially autocorrelated random effect with a conditional variance σ r m 2 for mortality rate. We assumed that μ t , β t , R , and β t , P R are random walks as:
  μ t ~ N μ t − 1 , σ μ 2
β t , R ~ N β t − 1 , R , σ β R 2
β t , P R ~ N β t − 1 , P R , σ β P R 2
where σ μ 2 , σ β R 2 , and σ β P R 2 are variances of the random walks. We also include a spatially autocorrelated random effect ( r m , j ) in the model to deal with a spatial autocorrelation in m t , i between neighboring plants. Note that i n v _ l o g i t μ t + β c e x p c t , i + β t , R d R + β t , P R d P R in Equation (A2) gives period-, substrate- and size-specific monthly mortality rate. The significance of the period-specific substrate effects can be accessed by the posterior distributions of β t , R and β t , P R . We assume that s t , i becomes one, when the target plant is observed as alive.
The observation model for plant size is assumed as:
  C t , i ~ N c t , i , σ o b s 2
where C t , i is logarithm of observed plant cover of plant individual i at time step t. We fixed the variance of observation error as: σ o b s = 0.1 for a better identifiability of the model.
The observation model for survival state is assumed as:
  S t , i ~ B e r s t , i
where S t , i is observation of “dead or alive” for plant individual i at time step t ( S t , i = 1 if the plant is alive, S t , i = 0 if it is dead); Ber() is Bernoulli distribution. Likelihood of the whole model included all individuals of a selected species was calculated based on Equations (A3) and (A4).
Note that the two parts (size and survival state) of the state of an individual plant in the above state-space model should be estimated simultaneously because Equation (A2) includes the estimated plant cover ( e x p c t , i ). However, no mortality was observed for two species (I. cylindrica and P. scabiosifolia; see Results). The survival-state part was omitted from the model for these two species.
The above model was applied to data for each species. We assumed that the prior distributions for all estimated parameters weakly informative as: x ~ N 0 , 5 2 where x is one of c 1 , i , γ 1 , α 1 , R , α 1 , P R , μ 1 , β c , β 1 , R , and β 1 , P R , and y ~ H N 5 2 where y is one of σ c , σ γ , σ α R , σ α P R , σ r g , σ μ , σ β R , σ β P R , and σ r m , and H N σ 2 is a half-normal distribution with a scale parameter σ .
All parameters in the model were estimated by Markov chain Monte Carlo (MCMC) method. We used Stan version 2.32.2 software [69] to implement MCMC. We ran four independent chains and obtained 2000 samples after a burn-in of 2000 samples for each run. The chains were thinned every 4 runs, yielding 2000 independent random samples from the joint posterior distribution. R ^ values were used to access convergence. We judged the convergence of the chains using the criterion that R ^ is less than 1.1.

References

  1. Leite, F.R.; Antunes, M.L.P. Green Roof Recent Designs to Runoff Control: A Review of Building Materials and Plant Species Used in Studies. Ecol. Eng. 2023, 189, 106924. [Google Scholar] [CrossRef]
  2. Mugita, J.; Tsuchiya, K.; Matsuoka, T.; Yamada, S.; Okuro, T. Effects of Biochar Usage and Plant Species Selections on the Storm Water Runoff Mitigations by Green Roofs. J. Jpn. Soc. Reveg. Technol. 2018, 44, 69–74. (In Japanese) [Google Scholar] [CrossRef]
  3. MacIvor, J.S.; Margolis, L.; Perotto, M.; Drake, J.A.P. Air Temperature Cooling by Extensive Green Roofs in Toronto Canada. Ecol. Eng. 2016, 95, 36–42. [Google Scholar] [CrossRef]
  4. Yamaguchi, T.; Yokoyama, H.; Ishii, K. Mitigating the Urban Heat Island Effect by Light and Thin Rooftop Greening. J. Jpn. Inst. Landsc. Archit. 2005, 68, 509–512. (In Japanese) [Google Scholar] [CrossRef]
  5. Nasr, Y.; El Zakhem, H.; Hamami, A.E.A.; El Bachawati, M.; Belarbi, R. Comprehensive Assessment of the Impact of Green Roofs and Walls on Building Energy Performance: A Scientific Review. Energies 2024, 17, 5160. [Google Scholar] [CrossRef]
  6. Kuronuma, T.; Watanabe, H.; Ishihara, T.; Kou, D.; Toushima, K.; Ando, M.; Shindo, S. CO2 Payoff of Extensive Green Roofs with Different Vegetation Species. Sustainability 2018, 10, 2256. [Google Scholar] [CrossRef]
  7. Shafique, M.; Xue, X.; Luo, X. An Overview of Carbon Sequestration of Green Roofs in Urban Areas. Urban For. Urban Green. 2020, 47, 126515. [Google Scholar] [CrossRef]
  8. Nicol, E.; Terzano, K. Modelling the Optimal Green Roof Type for Carbon Capture in an Urbanised University Campus. Urban For. Urban Green. 2025, 112, 128946. [Google Scholar] [CrossRef]
  9. McGuire, K.L.; Payne, S.G.; Palmer, M.I.; Gillikin, C.M.; Keefe, D.; Kim, S.J.; Gedallovich, S.M.; Discenza, J.; Rangamannar, R.; Koshner, J.A.; et al. Digging the New York City Skyline: Soil Fungal Communities in Green Roofs and City Parks. PLoS ONE 2013, 8, e58020. [Google Scholar] [CrossRef]
  10. Motegi, N.; Yanai, S. A Study on the Characteristics of Bird Distribution in Rooftop Vegetation in Tokyo Ward. J. Jpn. Inst. Landsc. Archit. 2005, 68, 597–600. (In Japanese) [Google Scholar] [CrossRef]
  11. MacIvor, J.S.; Lundholm, J. Performance Evaluation of Native Plants Suited to Extensive Green Roof Conditions in a Maritime Climate. Ecol. Eng. 2011, 37, 407–417. [Google Scholar] [CrossRef]
  12. Muratet, A.; Barra, M.; Hardion, L.; Chiron, F. Origins and Drivers of Roof Plant Assemblages: Designing Green Roofs for Biodiversity Conservation. Urban For. Urban Green. 2024, 94, 128247. [Google Scholar] [CrossRef]
  13. Ksiazek-Mikenas, K.; Chaudhary, V.B.; Skogen, K.A. Combinations of Plant Species with Complementary Traits Have the Potential to Maximize Ecosystem Services on Green Roofs. Urban Ecosyst. 2023, 26, 1193–1208. [Google Scholar] [CrossRef]
  14. Green Roofs Equivalent to Three Tokyo Domes Created In 2023: Results of the National Survey on Rooftop and Wall Greening Projects. (In Japanese). Available online: https://www.mlit.go.jp/report/press/toshi10_hh_000515.html (accessed on 29 May 2025).
  15. Iwata, T.; Shimoda, R. Spontaneous Vegetation in Mandatory Green Roofs at Public Educational Facilities in Sumida-Ku, Tokyo. J. Jpn. Inst. Landsc. Archit. Online Collect. Pap. 2024, 17, 38–46. (In Japanese) [Google Scholar] [CrossRef]
  16. Matsushita, Y.; Moriyama, M. Study of the Maintenance of Green Roof. AIJ J. Technol. Des. 2010, 16, 221–226. (In Japanese) [Google Scholar] [CrossRef]
  17. Lundholm, J.T. Green Roof Plant Species Diversity Improves Ecosystem Multifunctionality. J. Appl. Ecol. 2015, 52, 726–734. [Google Scholar] [CrossRef]
  18. Koyanagi, T.F.; Furukawa, T. Nation-Wide Agrarian Depopulation Threatens Semi-Natural Grassland Species in Japan: Sub-National Application of the Red List Index. Biol. Conserv. 2013, 167, 1–8. [Google Scholar] [CrossRef]
  19. Pritchard, D.J.; Fa, J.E.; Oldfield, S.; Harrop, S.R. Bring the Captive Closer to the Wild: Redefining the Role of Ex Situ Conservation. Oryx 2012, 46, 18–23. [Google Scholar] [CrossRef]
  20. Lundholm, J.T. Green Roofs and Facades: A Habitat Template Approach. Urban Habitats 2006, 4, 87–101. [Google Scholar]
  21. Bellini, A.; Bartoli, F.; Kumbaric, A.; Casalini, R.; Caneva, G. Evaluation of Mediterranean Perennials for Extensive Green Roofs in Water-Limited Regions: A Two-Year Experiment. Ecol. Eng. 2024, 209, 107399. [Google Scholar] [CrossRef]
  22. Trenta, M.; Quadri, A.; Sambuco, B.; Perez Garcia, C.A.; Barbaresi, A.; Tassinari, P.; Torreggiani, D. Green Roof Management in Mediterranean Climates: Evaluating the Performance of Native Herbaceous Plant Species and Green Manure to Increase Sustainability. Buildings 2025, 15, 640. [Google Scholar] [CrossRef]
  23. Liu, J.; Shrestha, P.; Skabelund, L.R.; Todd, T.; Decker, A.; Kirkham, M.B. Growth of Prairie Plants and Sedums in Different Substrates on an Experimental Green Roof in Mid-Continental USA. Sci. Total Environ. 2019, 697, 134089. [Google Scholar] [CrossRef]
  24. Sutton, R.K.; Harrington, J.A.; Skabelund, L.; MacDonagh, P.; Coffman, R.R.; Koch, G. PRAIRIE-BASED GREEN ROOFS: LITERATURE, TEMPLATES, AND ANALOGS. J. Green Build. 2012, 7, 143–172. [Google Scholar] [CrossRef]
  25. Nagase, A.; Tashiro-Ishii, Y. Habitat Template Approach for Green Roofs Using a Native Rocky Sea Coast Plant Community in Japan. J. Environ. Manage. 2018, 206, 255–265. [Google Scholar] [CrossRef]
  26. Komine, M.; Morimoto, J.; Katsuno, T. Planting Experiment Using the Coastal Plant Glehnia littoralis for Rooftop Greening. J. Jpn. Soc. Reveg. Technol. 2005, 31, 21–26. (In Japanese) [Google Scholar] [CrossRef]
  27. Kimura, Y. Creation of Semi-Natural Grassland on Rooftop. J. Jpn. Inst. Landsc. Archit. 2023, 86, 326–329. (In Japanese) [Google Scholar] [CrossRef]
  28. Lata, J.-C.; Dusza, Y.; Abbadie, L.; Barot, S.; Carmignac, D.; Gendreau, E.; Kraepiel, Y.; Mériguet, J.; Motard, E.; Raynaud, X. Role of Substrate Properties in the Provision of Multifunctional Green Roof Ecosystem Services. Appl. Soil Ecol. 2018, 123, 464–468. [Google Scholar] [CrossRef]
  29. Scolaro, T.P.; Ghisi, E. Life Cycle Assessment of Green Roofs: A Literature Review of Layers Materials and Purposes. Sci. Total Environ. 2022, 829, 154650. [Google Scholar] [CrossRef] [PubMed]
  30. Bates, A.J.; Sadler, J.P.; Greswell, R.B.; Mackay, R. Effects of Recycled Aggregate Growth Substrate on Green Roof Vegetation Development: A Six Year Experiment. Landsc. Urban Plan. 2015, 135, 22–31. [Google Scholar] [CrossRef]
  31. Molineux, C.J.; Gange, A.C.; Connop, S.P.; Newport, D.J. Using Recycled Aggregates in Green Roof Substrates for Plant Diversity. Ecol. Eng. 2015, 82, 596–604. [Google Scholar] [CrossRef]
  32. Eksi, M.; Sevgi, O.; Akburak, S.; Yurtseven, H.; Esin, İ. Assessment of Recycled or Locally Available Materials as Green Roof Substrates. Ecol. Eng. 2020, 156, 105966. [Google Scholar] [CrossRef]
  33. Kader, S.; Gratchev, I.; Michael, R.N. Recycled Waste Substrates: A Systematic Review. Sci. Total Environ. 2024, 953, 176029. [Google Scholar] [CrossRef]
  34. Liu, H.; Kong, F.; Yin, H.; Cook, L.M.; Huang, J.; Lensky, I.M.; Tan, T. Substrate Microorganisms Can Be an Ideal Tool for Improving Green Roof Sustainability. Urban For. Urban Green. 2024, 91, 128179. [Google Scholar] [CrossRef]
  35. Japan Meteorological Agency. Weather, Climate and Earthquake Information. (In Japanese). Available online: https://www.jma.go.jp/jma/index.html (accessed on 14 August 2025).
  36. Mitsui Kinzoku Perlite Co., Ltd. Available online: https://mitsui-perlite.co.jp/ (accessed on 15 September 2025).
  37. Yamato, M.; Hattori, T.; Asami, K. Comparison of Floristic Composition of Imperata cylindrica var. koenigii Communities Growing on Consolidated and on Traditional Levee Slopes in Sanda, Hyogo Pref. Western Japan. J. Weed Sci. Technol. 1999, 44, 170–179. (In Japanese) [Google Scholar] [CrossRef]
  38. Asami, K.; Yamato, M.; Hattori, T. Characteristics of the Imperata cylindrica—Erigeron Annuus Community. Veg. Sci. 1998, 15, 33–45. (In Japanese) [Google Scholar] [CrossRef]
  39. Asami, K.; Hattori, T.; Akamatsu, H. A Study on Management by Cutting of the Embankment Vegetation. J. Jpn. Inst. Landsc. Archit. 1994, 58, 125–128. (In Japanese) [Google Scholar] [CrossRef]
  40. Akamatsu, H.; Asami, K.; Tamura, K.; Hukui, S.; Hattori, T. Change after 16 years of artificial Imperata cylindrica var. koenigii grassland and possibility of creating the grassland. Hum. Nat. 2009, 20, 81–91. (In Japanese) [Google Scholar] [CrossRef]
  41. Tokyo Red Data Book (Mainland) 2023. (In Japanese). Available online: https://www.kankyo.metro.tokyo.lg.jp/nature/animals_plants/red_data_book/400100a20230424184941875 (accessed on 15 August 2025).
  42. Kitamura, S.; Murata, G.; Koyama, T. Illustrated Flora of Japan, Herbaceous Plants; Revised, Ed.; Hoikusha: Osaka, Japan, 1964; Volume 3, ISBN 978-4-586-30017-4. (In Japanese) [Google Scholar]
  43. Flora of Japan Volume 3a; Iwatsuki, K., Yamazaki, T., Boufford, D.E., Ohba, H., Eds.; Kodansha: Tokyo, Japan, 1993; ISBN 978-4-06-153420-9. [Google Scholar]
  44. Kitamura, S.; Murata, G. Illustrated Flora of Japan, Herbaceous Plants; Revised, Ed.; Hoikusha: Osaka, Japan, 1961; Volume 2, ISBN 978-4-586-30016-7. (In Japanese) [Google Scholar]
  45. FLL. (Forschungsgesellschaft Landschaftsentwicklung Landschaftsbau e.V). Green Roof Guidelines: Guidelines for the Planning, Construction and Maintenance of Green Roofs, 2018th ed.; FLL: Bonn, Germany, 2018. [Google Scholar]
  46. Hiradate, S.; Kusumoto, Y.; Yoshitake, H.; Baba, Y. Biodiversity Supported by Soils. In Conservation Ecology of Familiar Nature: Understanding Biodiversity; Nemoto, M., Ed.; Baifukan: Tokyo, Japan, 2010; pp. 131–148. ISBN 978-4-563-07810-2. (In Japanese) [Google Scholar]
  47. Aizawa, S. Chemical Properties of Forest Soil. Tree For. Health 2011, 15, 189–192. [Google Scholar] [CrossRef]
  48. Xu, Y.; Liao, Z.; Liu, J.; Liu, C.; Li, Y. Unirrigated Extensive Green Roofs in Humid Subtropics—Plant Selection and Substrate Design for Low Maintenance and Climate Resilience. Urban For. Urban Green. 2024, 101, 128554. [Google Scholar] [CrossRef]
  49. Küçük, N.; Ekşi, M. Evaluation of Native Herbaceous Plants for Green Roof Applications in Istanbul. Ecol. Eng. 2023, 196, 107094. [Google Scholar] [CrossRef]
  50. Elberse, W.T.; Berendse, F. A Comparative Study of the Growth and Morphology of Eight Grass Species from Habitats with Different Nutrient Availabilities. Funct. Ecol. 1993, 7, 223–229. [Google Scholar] [CrossRef]
  51. Ellenberg, H. Zeigerwerte Der Gefäßpflanzen Mitteleuropas; Goltze: Göttingen, Germany, 1974; Volume 9, pp. 1–97. [Google Scholar]
  52. Tichý, L.; Axmanová, I.; Dengler, J.; Guarino, R.; Jansen, F.; Midolo, G.; Nobis, M.P.; Van Meerbeek, K.; Aćić, S.; Attorre, F.; et al. Ellenberg-Type Indicator Values for European Vascular Plant Species. J. Veg. Sci. 2023, 34, e13168. [Google Scholar] [CrossRef]
  53. Walker, K.J.; Stevens, P.A.; Stevens, D.P.; Mountford, J.O.; Manchester, S.J.; Pywell, R.F. The Restoration and Re-Creation of Species-Rich Lowland Grassland on Land Formerly Managed for Intensive Agriculture in the UK. Biol. Conserv. 2004, 119, 1–18. [Google Scholar] [CrossRef]
  54. Stearns, S.C. Trade-Offs in Life-History Evolution. Funct. Ecol. 1989, 3, 259–268. [Google Scholar] [CrossRef]
  55. Bazzaz, F.A.; Ackerly, D.D.; Reekie, E.G. Reproductive Allocation in Plants. In Seeds: The Ecology of Regeneration in Plant Communities; CABI Books: Oxford, UK, 2000; pp. 1–29. [Google Scholar]
  56. Kuramoto, N. Gravel-Bar in the Tama River as Habitat of Endemic Plants. J. Jpn. Soc. Reveg. Technol. 2019, 44, 502–506. [Google Scholar] [CrossRef]
  57. Osawa, S.; Hirose, F. Seed Density within the Topsoil and Seed Germination Ability of Dianthus Superbus Var. Longicalycinus on a Coastal Dune. J. Jpn. Soc. Reveg. Technol. 2023, 49, 3–8. [Google Scholar] [CrossRef]
  58. Nemoto, M.; Yamada, S.; Tabuchi, S. Handbook of Revegetation with Native Herbaceous Species: Restoration of Familiar Nature Vegetation; Asakura Shoten: Tokyo, Japan, 2020; ISBN 978-4-254-42042-5. (In Japanese) [Google Scholar]
  59. Nagase, A.; Dunnett, N. Drought Tolerance in Different Vegetation Types for Extensive Green Roofs: Effects of Watering and Diversity. Landsc. Urban Plan. 2010, 97, 318–327. [Google Scholar] [CrossRef]
  60. Yamada, S.; Nemoto, M.; Koyanagi, T.; Yamamoto, Y.; Yatsuki, H. “Biodiversity-Conscious Vegetation Management in Riverdikes” Vegetation and Environmental Attributes in Species-Rich Semi-Natural Grassland in Riverbanks. J. Jpn. Soc. Reveg. Technol. 2016, 42, 428–432. (In Japanese) [Google Scholar] [CrossRef]
  61. Guidi, M.; Bousselot, J. Earlier Flowering Phenology and Pollinator Visitation on Urban Green Roofs Compared to Ground-Level Gardens. Land 2024, 13, 178. [Google Scholar] [CrossRef]
  62. Dunnett, N. Naturalistic Planting Design: The Essential Guide; Filbert Pr: Bath, UK, 2019; ISBN 978-0-9933892-6-9. [Google Scholar]
  63. Buffam, I.; Mitchell, M.E.; Durtsche, R.D. Environmental Drivers of Seasonal Variation in Green Roof Runoff Water Quality. Ecol. Eng. 2016, 91, 506–514. [Google Scholar] [CrossRef]
  64. Lim, H.S. What Happens to Nitrogen and Phosphorus Nutrient Contributions from Green Roofs as They Age? A Review. Environ. Adv. 2023, 12, 100366. [Google Scholar] [CrossRef]
  65. Rowe, D.B.; Monterusso, M.A.; Rugh, C.L. Assessment of Heat-Expanded Slate and Fertility Requirements in Green Roof Substrates. HortTechnology 2006, 16, 471–477. [Google Scholar] [CrossRef]
  66. Nagase, A.; Dunnett, N. The Relationship between Percentage of Organic Matter in Substrate and Plant Growth in Extensive Green Roofs. Landsc. Urban Plan. 2011, 103, 230–236. [Google Scholar] [CrossRef]
  67. Mitchell, M.E.; Emilsson, T.; Buffam, I. Carbon, Nitrogen, and Phosphorus Variation along a Green Roof Chronosequence: Implications for Green Roof Ecosystem Development. Ecol. Eng. 2021, 164, 106211. [Google Scholar] [CrossRef]
  68. Iwata, T.; Umeki, K. Assessing the Performance of Native Grassland Species on Extensive Green Roofs Under Japanese Climatic Conditions. In Proceedings of the International Conference 2025 on Spatial Planning and Sustainable Development (SPSD 2025), Hankyong, Republic of Korea, 7–9 August 2025; p. 90. [Google Scholar]
  69. Carpenter, B.; Gelman, A.; Hoffman, M.D.; Lee, D.; Goodrich, B.; Betancourt, M.; Brubaker, M.; Guo, J.; Li, P.; Riddell, A. Stan: A Probabilistic Programming Language. J. Stat. Softw. 2017, 76, 1–32. [Google Scholar] [CrossRef]
Environments 12 00345 g001
Figure 1. Mean air temperature and precipitation at the experimental site from 2021 to 2024 (a) and mean daily air temperature and precipitation during July–September 2024 (b).
Figure 1. Mean air temperature and precipitation at the experimental site from 2021 to 2024 (a) and mean daily air temperature and precipitation during July–September 2024 (b).
Environments 12 00345 g001
Environments 12 00345 g002
Figure 2. Layout of experimental quadrats based on a randomized block design. P, PR, and R indicate substrate types. Different colors indicate different blocks in the randomized block design.
Figure 2. Layout of experimental quadrats based on a randomized block design. P, PR, and R indicate substrate types. Different colors indicate different blocks in the randomized block design.
Environments 12 00345 g002
Environments 12 00345 g003
Figure 3. Construction details of the experimental green roof.
Figure 3. Construction details of the experimental green roof.
Environments 12 00345 g003
Environments 12 00345 g004
Figure 4. Planting arrangement of selected species within each experimental quadrat.
Figure 4. Planting arrangement of selected species within each experimental quadrat.
Environments 12 00345 g004
Environments 12 00345 g005
Figure 5. Overview of the experimental site (photographed on 10 May 2024).
Figure 5. Overview of the experimental site (photographed on 10 May 2024).
Environments 12 00345 g005
Environments 12 00345 g006
Figure 6. Particle size distribution of substrates (P = Perlite-only substrate, PR = Perlite and Reused soil mixture, R = Reused substrate.) compared with the FLL guideline. The sky-blue shaded area indicates the recommended particle size distribution range for multi-layer extensive substrates.
Figure 6. Particle size distribution of substrates (P = Perlite-only substrate, PR = Perlite and Reused soil mixture, R = Reused substrate.) compared with the FLL guideline. The sky-blue shaded area indicates the recommended particle size distribution range for multi-layer extensive substrates.
Environments 12 00345 g006
Environments 12 00345 g007
Figure 7. Changes in plant cover (a) and estimated RGR (b) of Imperata cylindrica over time. In each diagram in (a), a line and a ribbon represent the posterior means of estimated plant cover of an individual and 95% Bayesian credible intervals, respectively. Red points represent observed plant cover. Asterisks in (b) indicate statistically significant differences between substrate P and PR, or P and R. P = Perlite-only substrate, PR = Perlite and Reused soil mixture, R = Reused substrate.
Figure 7. Changes in plant cover (a) and estimated RGR (b) of Imperata cylindrica over time. In each diagram in (a), a line and a ribbon represent the posterior means of estimated plant cover of an individual and 95% Bayesian credible intervals, respectively. Red points represent observed plant cover. Asterisks in (b) indicate statistically significant differences between substrate P and PR, or P and R. P = Perlite-only substrate, PR = Perlite and Reused soil mixture, R = Reused substrate.
Environments 12 00345 g007
Environments 12 00345 g008
Figure 8. Changes in plant cover (a), estimated RGR (b), and number of flowering individuals (c) of Patrinia scabiosifolia over time. Star-shaped red points indicate that the plants were in bloom at that time. P = Perlite-only substrate, PR = Perlite and Reused soil mixture, R = Reused substrate. See the caption of Figure 7 for the explanations for lines, points, ribbons, and asterisks in (a,b).
Figure 8. Changes in plant cover (a), estimated RGR (b), and number of flowering individuals (c) of Patrinia scabiosifolia over time. Star-shaped red points indicate that the plants were in bloom at that time. P = Perlite-only substrate, PR = Perlite and Reused soil mixture, R = Reused substrate. See the caption of Figure 7 for the explanations for lines, points, ribbons, and asterisks in (a,b).
Environments 12 00345 g008
Environments 12 00345 g009
Figure 9. Changes in plant cover (a), estimated RGR (b), and number of flowering individuals (c) of Dianthus superbus var. longicalycinus over time. P = Perlite-only substrate, PR = Perlite and Reused soil mixture, R = Reused substrate. See the caption of Figure 6 and Figure 7 for the explanations for lines, points, ribbons, and asterisks in (a,b).
Figure 9. Changes in plant cover (a), estimated RGR (b), and number of flowering individuals (c) of Dianthus superbus var. longicalycinus over time. P = Perlite-only substrate, PR = Perlite and Reused soil mixture, R = Reused substrate. See the caption of Figure 6 and Figure 7 for the explanations for lines, points, ribbons, and asterisks in (a,b).
Environments 12 00345 g009
Environments 12 00345 g010
Figure 10. Changes in plant cover (a), estimated RGR (b), and number of flowering individuals (c) of Sanguisorba officinalis over time. P = Perlite-only substrate, PR = Perlite and Reused soil mixture, R = Reused substrate. See the caption of Figure 6 and Figure 7 for the explanations for lines, points, ribbons, and asterisks in (a,b).
Figure 10. Changes in plant cover (a), estimated RGR (b), and number of flowering individuals (c) of Sanguisorba officinalis over time. P = Perlite-only substrate, PR = Perlite and Reused soil mixture, R = Reused substrate. See the caption of Figure 6 and Figure 7 for the explanations for lines, points, ribbons, and asterisks in (a,b).
Environments 12 00345 g010
Environments 12 00345 g011
Figure 11. Changes in plant cover (a), estimated RGR (b), number of flowering individuals (c), and estimated mortality (d) of Prunella vulgaris subsp. asiatica over time. P = Perlite-only substrate, PR = Perlite and Reused soil mixture, R = Reused substrate. See the caption of Figure 6 and Figure 7 for the explanations for lines, points, ribbons, and asterisks in (a,b).
Figure 11. Changes in plant cover (a), estimated RGR (b), number of flowering individuals (c), and estimated mortality (d) of Prunella vulgaris subsp. asiatica over time. P = Perlite-only substrate, PR = Perlite and Reused soil mixture, R = Reused substrate. See the caption of Figure 6 and Figure 7 for the explanations for lines, points, ribbons, and asterisks in (a,b).
Environments 12 00345 g011
Environments 12 00345 g012
Figure 12. Changes in plant cover (a), estimated RGR (b), number of flowering individuals (c), and estimated mortality (d) of Adenophora triphylla var. japonica over time. P = Perlite-only substrate, PR = Perlite and Reused soil mixture, R = Reused substrate. See the caption of Figure 6 and Figure 7 for the explanations for lines, points, ribbons, and asterisks in (a,b).
Figure 12. Changes in plant cover (a), estimated RGR (b), number of flowering individuals (c), and estimated mortality (d) of Adenophora triphylla var. japonica over time. P = Perlite-only substrate, PR = Perlite and Reused soil mixture, R = Reused substrate. See the caption of Figure 6 and Figure 7 for the explanations for lines, points, ribbons, and asterisks in (a,b).
Environments 12 00345 g012
Environments 12 00345 g013
Figure 13. Weed dry weight harvested on 5 September 2024. Error bars represent standard error. No weeds emerged at that time on substrate P. P = Perlite-only substrate, PR = Perlite and Reused soil mixture, R = Reused substrate.
Figure 13. Weed dry weight harvested on 5 September 2024. Error bars represent standard error. No weeds emerged at that time on substrate P. P = Perlite-only substrate, PR = Perlite and Reused soil mixture, R = Reused substrate.
Environments 12 00345 g013
Table 1. Habitat, distribution, and phenology of selected native species.
Table 1. Habitat, distribution, and phenology of selected native species.
SpeciesFamilyHabitatDistribution in JapanFlowering TimeReference
Imperata cylindricaPoaceaeOpen fields and embankmentsHokkaido to KyushuMay to JuneKitamura et al. [42]
Patrinia scabiosifoliaValerianaceaeSunny placeHokkaido to KyushuAugust to OctoberIwatsuki et al. [43]
Dianthus superbus var. longicalycinusCaryophyllaceaeGrasslandHonshu to KyushuJune to SeptemberKitamura and Murata [44]
Sanguisorba officinalisRosaceaeGrasslandHokkaido to KyushuJuly to OctoberKitamura and Murata [44]
Prunella vulgairis susp. asiaticaLamiaceaeSunny grasslandHokkaido to KyushuJune to AugustIwatsuki et al. [43]
Adenophora triphylla var. japonicaCampanulaceaeGrasslandHokkaido to KyushuJuly to NovemberIwatsuki et al. [43]
Table 2. Physicochemical properties of substrates sampled on 22 May 2024.
Table 2. Physicochemical properties of substrates sampled on 22 May 2024.
Electrical ConductivitypHNH4-NNO3-NP2O5Dry Bulk DensitySaturated Bulk DensityWHCC:N Ratio
mS/m mg/Lmg/Lmg/Lg/cm3g/cm3%
R17.845.631.617.110960.41.174.010.05
PR15.676.09.39.28710.31.069.310.93
P4.116.61.2<0.51.80.20.965.3
FLL guideline 6.0–8.5<80
as N		<80
as N		<200 35–65
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Iwata, T.; Shimoda, R.; Takahashi, T.; Umeki, K. Evaluating Native Grassland Species for Application in Extensive Green Roofs in Japan. Environments 2025, 12, 345. https://doi.org/10.3390/environments12100345

AMA Style

Iwata T, Shimoda R, Takahashi T, Umeki K. Evaluating Native Grassland Species for Application in Extensive Green Roofs in Japan. Environments. 2025; 12(10):345. https://doi.org/10.3390/environments12100345

Chicago/Turabian Style

Iwata, Tsukasa, Ryosuke Shimoda, Terumasa Takahashi, and Kiyoshi Umeki. 2025. "Evaluating Native Grassland Species for Application in Extensive Green Roofs in Japan" Environments 12, no. 10: 345. https://doi.org/10.3390/environments12100345

APA Style

Iwata, T., Shimoda, R., Takahashi, T., & Umeki, K. (2025). Evaluating Native Grassland Species for Application in Extensive Green Roofs in Japan. Environments, 12(10), 345. https://doi.org/10.3390/environments12100345

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop