Short-Term Growth Dynamics of Spontaneous and Planted Vegetation on Subtropical Extensive Green Roof as Renaturalized Biotope

Abstract

Spontaneous vegetation within a managed green space is often regarded as unwelcoming and insignificant weeds. This perception is still deep-rooted among green-space managers and the general public worldwide; they are generally uncertain about the management needs after allowing these groups of flora to take root. The short-term growth dynamics of both spontaneous and planted vegetation should be analyzed, and a widely acceptable, feasible management plan to balance aesthetic and ecological functions should be formulated with the backing of data and analysis for such fast-growing flora in tropical and subtropical regions. A manicured, extensive green roof with only seven (two native, five exotic) plant species was transformed into a renaturalized biotope by replacing 15 native ferns and forb species over 15 months. After planting, a baseline plant survey was conducted, with 54 plant species representing spontaneous growth and 14 planted species alive (7 planted native species survived, plus 7 species planted prior to renaturalization revived). Three quarterly plant surveys recorded the cover-abundance of each species, and the growth dynamics of the planted and spontaneous plant species were evaluated over the first year of study. During each quarterly survey, the number of planted and spontaneous plant species remained stable (ranging from 14 to 16 species and 51 to 54 species, respectively), with a constant turnover of 11 to 12 die-out species and 11 to 12 newly colonized or revived species. Plant coverage of different plant forms fluctuated slightly (within 7%) in the quarterly surveys according to seasonal changes, except for ferns, which outperformed (12% increase in coverage in a year) all the other plant forms. The height of the planted vegetation fluctuated in a year, being shorter during the summer, while the height of spontaneous vegetation remained stable throughout the year, exhibiting resilience to scouring heat. The seasonal growth tendencies of both planted and spontaneous plants were illustrated in relation to their species ranks, and further hierarchical cluster analysis was conducted for the clustering of spontaneous species. Their differential growth patterns provided comprehensive information or supported decisions regarding plant selection and maintenance, which is a scientific novelty within this unexplored topic. Management recommendations based on the findings were suggested to fulfill both aesthetic and ecological needs. Species with stable and less stable growth patterns could be useful to meet maintenance efficiency and biodiversity enhancement needs, respectively. These findings provide insights to form guiding principles for choosing plant species for renaturalization projects.

1. Introduction

Skyrise greenery in cities is becoming prevalent, often in the form of manicured, neat, and tidy patches of groundcover or shrubs with limited biodiversity that intrinsically require higher resource and labor inputs for maintenance. Spontaneous vegetation within a managed green space is often regarded as neglected, unwelcoming, or insignificant weeds [1,2,3,4]. Although green-roof systems are nature-based solutions to provide environmental benefits, such as urban heat island mitigation and air quality improvement in urban areas, a continuous flow of resources and manpower is devoted to the management of these landscaped areas.

Extensive green roofs have received the most research attention; apart from the engineering design of green roofs, plants are the pivotal interface of sustainable and effective functioning landscapes as nature-based solutions. In many parts of the world, most extensive green roofs are manicured with neat and tidy patch plantings of short woody shrubs and monotonous herbaceous covers limited to a few plant species—in some extreme cases, only one to several grass (Poaceae) or sedum species, such as Zoysia japonica, Cynodon dactylon, and Sedum acre, are planted. This practice is perceived as prioritizing maintenance efficiency over ecological complexity [5]. In tropical and subtropical regions, without regular mowing, weeding, and other maintenance work, turfs on rooftops are rapidly colonized by spontaneous plant species. Some green roofs, whose owners cannot afford frequent and costly maintenance efforts, will be invaded by spontaneous plants, especially undesirable tree and shrub species, creating long-term structural damage to green-roof engineering layers and the roofs themselves, in some cases leading to water leakage.

Green roofs designed with locally occurring, curated vegetation cohorts, whether native or exotic, provide greater biodiversity than single- or several-species grass turf or herbaceous groundcover species with most exotic plants. Cook-Patton and Bauerle [6] reviewed the potential benefits of plant diversity on green roofs in their review paper. Many studies have proven that a high diversity of plant species contributes to benefits in terms of water retention, thermal properties, reduced nutrient leaching, pollutant removal, and ecological values [7,8,9,10,11]. Some scholars have opined that the flora in urban China is facing the risk of biological homogenization [2]. Choosing plant species to be planted first in the establishment period and setting optimal management strategies are the top concerns for maximizing green-roof performance, as it can be extremely costly and impractical to eradicate certain unwanted plant species. Many naturalized exotic species are observed to be aggressive colonizers that outcompete both stable native and exotic species. Without a thorough understanding of the growth dynamics of spontaneous and planted species after breaking up continuous vegetation cover, landscape designers and managers will not have confidence in adopting biodiverse designs and letting spontaneous plants grow, let alone implementing renaturalization projects [12].

The advocacy of the utilization of native flora has progressed in more developed countries, but it is rather limited in some other regions. A list of suitable native plants for extensive green roofs has not yet been developed, and the flexibility of adopting more native plant species has been limited further by the shortage of native plant stock supplies and experience in local and nearby nurseries. A conservative approach has been observed among landscape designers in regions where plant selection for extensive green roofs is undertaken, as has a conservative mindset among the public. For instance, in Hong Kong, Law et al. [13] identified a divergence in perceptions of the importance of native plants in urban green space among Hong Kong citizens.

Certain studies have advocated using weed species as design alternatives for green roofs [1,14]; these would be viable in temperate climate zones, where plant growth is slower than in hotter and wetter counterparts. A study in Germany in 2006 [1] examined plant dynamics when using spontaneously occurring species and combining them with ornamental plants in yearly intervals for 3 years. This study specifically tested 3 target species that have conspicuous flowers with and without competition from 2 matrix spontaneous species. This medium-term observation served the purpose well in temperate regions with slower plant growth, and only several target species could be tested with control groups and replicates. Scholars from Singapore documented the creation of a biodiverse green-roof habitat with about 90 selected plant species seeded, and they monitored the plots for 18 months with basic parameters, while plant-growth dynamics were not detailed [15]. Such a green-roof creation project, which focused on specially selected plant species, was fundamentally different from naturally occurring spontaneous species. The arrival of seed sources cannot be controlled by landscape managers. They are keen to know which colonizing species should be eradicated immediately and which species could be accommodated to enhance biodiversity.

Short-term plant dynamics of both naturally occurring spontaneous and planted vegetation on extensive green roofs in subtropical and tropical regions is an uncharted realm. This study aims to provide detailed documentation and analysis as foundational knowledge to formulate suitable landscape management guidelines for and after spontaneous plants being introduced or allowed to take root on extensive green roofs and other lawns (amenity turf) previously maintained with high resource input.

2. Materials and Methods

2.1. The Study Site

This study was conducted on a retrofitted low-pitched (10 degrees) (total area: 329 m²) green roof on a single-story office building in Kadoorie Farm and Botanic Garden in Hong Kong (22°25′ N, 114°07′ E) at an altitude of 133 m a.s.l., situated in the bottom of a valley in proximity to rich flora and fauna. The vicinity is regarded as a rural setting. The lowest point of the green roof is approximately 3 m from ground level and 5 m at the ridge. The north-facing pitched roof is exposed to maximum sunshine and wind, while the south-facing pitched roof is partly sheltered from nearby trees and has a slope to provide partial shading during the daytime. Hong Kong has a mean annual temperature of 23.6 °C (1988–2017) and a mean annual rainfall of 2398 mm (1981–2010), with around 80% of rainfall between May and September, a typical humid subtropical monsoon climate. In winter, it is dry with breeze and sunshine, and the temperature rarely falls below 10 °C. During the study period (May 2013–Nov 2015), the daily minimum and maximum mean temperature ranged from 14.2 and 26.6 °C, while the absolute daily minimum did not go lower than 5 °C at the site [16]. The auto sprinkler system irrigated the green roof twice a day, starting at 0300 and 1930, respectively, for 30 min each. The irrigation system operated daily regardless of the amount of rainfall, and there was no precipitation sensor or rain gauge. No fertilizer was applied before the green-roof transformation and during plant surveys.

2.2. A Manicured Extensive Green Roof

The retrofitted green roof was constructed on this old office building around 2008. This extensive green roof has a typical engineering design, with a waterproofing membrane over a concrete slab, a drainage layer, rock wool, and ~80 mm sandy loam. Seven plant species were originally installed before renaturalized planting. The original plant species included 1 native fern (Nephrolepis auriculata) and 5 exotic herbaceous ground covers (Zoysia japonica, Ophiopogon jaburan, Arachis duranensis, Tradescantia zebrina, Asparagus densiflorus) and 1 cultivar (Ophiopogon japonicus cv. ‘Nanus’) [17,18,19,20].

2.3. Transforming a Manicured Green Roof into a Biodiverse Habitat

The plant species chosen to renaturalize this extensive green roof are native to Hong Kong, their original habitats range from exposed to shaded environment. Ferns with ornamental value were selected to give a wild yet aesthetic impression. By breaking continuous vegetation cover, spontaneous colonizing plants could take root and play a part in forming a biodiverse habitat [21,22]. The transformation was undertaken in 4 phases from May 2013 to Sep 2014, which included the total uprooting and the removal of the original vegetation and the transplanting of potted pre-grown plants with intact root balls on the same day [18]. The above-mentioned 7 original plant species were removed as far as possible by workers manually, except the fern species Nephrolepis auriculata. Before this experiment, we anticipated that some species with extensive rhizomes and root tubers might resprout, as it was impractical to remove all perennating organs from the substrate [23]. Fifteen native species (12 native ferns and 3 native forbs species) of pre-grown plants (a total of 1016 pots of plants) from in-house nurseries were transplanted after they reached their life stage suitable for transplanting (see

Table 1. Planted quantity, planting date and location, characteristics, and survival rate of selected taxa planted on green-roof site from May 2013 to Sep 2014.

Species	Family	Number of Pots Planted	%	Planting Date	Planting Location	Life-Form Group	Habitat	Survival Rate by the End of Renaturalized Planting (Sep 2014)
Adiantum capillus-veneris	Adiantaceae	8	0.77%	May 2013	South-facing roof	Low-growing fern	Heavy shade, moist rocky cliff	0%
Adiantum flabellulatum	Adiantaceae	2	0.19%	May 2013	South-facing roof	Low-growing fern	Acidic sunny slope	0%
Adiatum malesianum	Adiantaceae	23	2.21%	May 2013	South-facing roof	Low-growing fern	Heavy shade, moist rocky cliff	0%
Asplenium prolongatum	Aspleniaceae	20	1.92%	May 2013	South-facing roof	Low-growing fern	Heavy shade, moist rocky cliff	0%
Ctenitis eatonii	Dryopteridaceae	27	2.59%	May 2013	South-facing roof	Medium height fern	Wet places in forests	0%
Cyclosorus parasiticus	Thelypteridaceae	256	24.54%	May 2013; mid-Sep 2013; late Sep 2014	Both roofs	Tall fern	Forest undergrowth, beside stream	37.50%
Desmodium heterocarpon	Fabaceae	12	1.15%	late Sep 2014	North-facing roof	Creeping forb	Sparse shrubland	100%
Elephantopus scaber	Asteraceae	40	3.84%	late Sep 2014	North-facing roof	Low-growing forb	Sunny open field, roadside	0%
Liriope spicata	Liliaceae	16	1.53%	late Sep 2014	South-facing roof	Medium height forb	Slope, forest undergrowth	0%
Macrothelypteris torresiana	Thelypteridaceae	115	11.03%	May 2013; mid-Sep 2013; early and late Sep 2014	Both roofs	Tall fern	Forest valley	100%
Pteris biaurita	Pteridaceae	22	2.11%	May 2013; mid-Sep 2013	Both roofs	Medium height fern	Forest undergrowth, sunny location	80%
Pteris ensiformis	Pteridaceae	446	42.76%	May 2013; mid-Sep 2013; late Sep 2014	Both roofs	Low-growing fern	Forest undergrowth, sunny location	100%
Pteris semipinnata	Pteridaceae	7	0.67%	May 2013; mid-Sep 2013	Both roofs	Low-growing fern	Forest undergrowth, sunny location	100%
Selaginella moellendorffii	Selaginellaceae	30	2.88%	May 2013	South-facing roof	Low-growing fern	Heavy shade, moist rocky cliff	0%
Sphenomeris chinensis	Lindsaeaceae	19	1.82%	May 2013; mid-Sep 2013	Both roofs	Low-growing fern	Forest margins	0%

). Detailed site observation had been recorded for roof aspect, intensity, and duration of sun exposure, in relationship to the surrounding local terrain, trees, and large shrubs nearby, which cast shade on the green-roof site. Plants were arranged according to the microclimate (e.g., tree shading, aspect) on the green roof, fitting their respective habitat preferences. For example, those species with the original habitat, “heavy shades”, were planted on the part of the green roof with tree shade and shorter sunlight exposure. Additionally, aesthetic concerns were taken into account. To let visitors enjoy the view of all vegetation, the species with a taller mature height were planted at the top of the roof ridge, and shorter species were arranged closer to the edges so that shorter species could be viewed on ground level. Figure 1 shows the hand-drawn planting design (bird-eye view) of such green-roof transformation. Figure 2 shows the commencement of the first phase of planting work in May 2013.

Starting from the first phase of renaturalization in May 2013, bare soil surface on the roof appeared when the original plant cover was removed, allowing seeds of spontaneous plants in contact with the substrate, and some germinated (Figure 3a shows the first phase of planting partially done in May 2013). Most prepared native plants [86%] were transplanted in the first 2 phases in May and Sep 2013. The remaining plants were transplanted in follow-up (3rd and 4th) phases in early and late Sep 2014. No major maintenance work had taken place on the studied roof for 1 year throughout the monitoring period. Figure 3b shows the appearance of the green roof 20 months after the major phases (1st and 2nd) of planting work ended (i.e., May 2015), with well-established ferns and spontaneous plants. Figure 4 is a methodology flowchart for this research.

2.4. Vegetation Surveys and Statistical Analysis

A baseline plant survey was carried out in Oct 2014, followed by quarterly surveys in Apr, July, and Nov 2015 to record and assess all plant species, including previously planted species and spontaneous plant species. Trained personnel took the following measurements and observations to ensure the quality and consistency of the data. For each species, average height and coverage percentage (for each plant species, coverage percentage on two 1 × 1 m quadrats were estimated according to the Domin-Krajina cover-abundance scale, quadrat locations were determined by randomly generated grid numbers, and the average figure was computed as the coverage percentage figure), the trend of coverage (through visual estimation classifying them into 3 classes: “Vigorous”, “Stable”, “Declining”) were recorded in baseline and subsequent 3 quarterly surveys. Additional information on the “Alive” or “Dead” status of each species was recorded in quarterly surveys, while their locational information was not recorded.

The Shannon-Wiener Diversity Index (H’) of species diversity was calculated after each quarterly survey as H’ = −Σ pi(ln pi), where pi=relative coverage abundance of each species [24], and the evenness index was calculated as J’ = H’/ln(S), where S = total number of species [25]. Smith and Wilson’s Index of Evenness (Evar) is an index of evenness based on the variance in abundance of species. It is independent of species richness and sensitive to both rare and common species in the community richness [26]. Evar is calculated as:

$$Evar=1-{\displaystyle \frac{2}{\pi }}arctan\left\{\sum _{s=1}^S{\left(\mathrm{l}\mathrm{n}(x_s\right)-\sum _{t=1}^S\ln \left(x_t\right)/S)}^2/S\right\}$$

Data from vegetation surveys was analyzed by comparing the above indices and the growing trend of relative abundance coverage was analyzed by hierarchical cluster analysis using Ward’s linkage method with SPSS software [27].

3. Results

Before the baseline survey in Sep 2014, 5 species (4 fern and 1 forb species) out of the 15 species planted had an 80–100% survival rate, one fern species (C. parasiticus) had 37.5% survival and 8 species (6 fern species and 2 forb species) died (Table 1). The growth of these species could have been affected by transplant shock and, in the long term, by the competition for resources on the extensive green roof. As revealed from this planting test, low-growing ferns and forbs that prefer shade and moist microhabitat (e.g., Adiantum spp., Asplenium prolongatum, Selaginella moellendorffii, Sphenomeris chinensis) could not survive on this exposed green roof with high solar insolation. While forbs like the Elephantopus scaber, which require intense sunlight, might be outcompeted by taller species [28].

3.1. Revolution of Planted and Spontaneous Vegetations in Terms of Coverage, Plant Height and Diversity

For the 7 species originally planted on the manicured green roof, one fern species had a dominating growth regime, while the other 6 species revived from remaining rhizomes and grew up to a coverage of 1.95% on average. The coverage of dominant fern N. auriculata was around 10% during renaturalization planting in Sep 2013 and grew continuously from 13% in Oct 2014 to 28% in Nov 2015. For 15 native fern and forb species planted during the renaturalization planting, 8 species died off, and 7 native species survived (Table 1) by Sep 2014. Adding the above two figures together, they accounted for “the number of planted species alive” shown in

Table 2. Selected diversity and evenness indices of all planted and spontaneous plant species over the period Oct 2014 to Nov 2015.

	Oct 2014	Apr 2015	Jul 2015	Nov 2015
Shannon–Weiner Diversity Index (H’)	3.805	3.632	3.486	3.416
Pielou’s Evenness Index (J’)	0.901	0.864	0.829	0.812
Smith and Wilson’s Index of Evenness (Evar)	0.542	0.528	0.739	0.712
Number of plant species alive	68	67	67	67
Number of planted species alive	14	14	15	16
Number of spontaneous species alive	54	53	52	51
Number of die-off spontaneous species compared with previous plant survey	--	12	11 *	12
Number of newly colonized (spontaneous) species compared with previous plant survey	--	11	5	9
Number of revived spontaneous species compared with previous plant survey	--	--	6	3

* Includes one species removed due to maintenance purposes.

, and it ranged narrowly between 14 to 16 species in the 3 quarterly surveys.

In the baseline survey in Oct 2014, there were a total of 68 plant species recorded on the roof. A total of 14 planted species (the 7 original species planted on the original manicured roof plus 7 native ferns and forbs planted in renaturalized planting) covered 45.4% (area-weighted average height of 35.4 cm; tallest: 60 cm, shortest: 5 cm). In Oct 2014, 54 spontaneous plant species covered 54.6% of the roof, with an area-weighted average height of 27.4 cm (tallest: 142 cm, shortest: 5 cm). Figure 5 shows the trend of plant height taken into account of plant coverage area. Planted species grew vigorously in springtime, declined over the summer, and revived slightly in early winter, while spontaneous species started at 27 cm in Oct 2014, grew steadily to 34.8 cm in spring, exhibited a stable height with a gentle peak (36.1 cm) in summer, and declined slightly in winter (33.3 cm). In Nov 2015, there were a total of 67 plant species. A total of 15 planted species covered 55.5% (area-weighted average height of 45 cm) of the roof, while 52 spontaneous species covered 44.5% (area-weighted average height of 33 cm). Compared with planted vegetation, which is mostly ferns, the height of spontaneous plants was less affected by seasonal effects.

It is worth noting that tall planted ferns might compete with certain shorter planted and spontaneous species, such as low-growing and sun-loving ferns and forbs, to sustain themselves on the roof. From an aesthetic point of view, spontaneous species may help to add seasonal interest. From an ecological point of view, spontaneous species may provide potential additional larval food and nectar sources.

Spontaneous species included trees, shrubs, and perennial and annual herbaceous plants. Table 2 shows the diversity and evenness indices of all planted and spontaneous plant species over the period Oct 2014 to Nov 2015. The diversity of plants continuously decreased throughout the whole monitoring period, from 3.805 to 3.416, and evenness J’ showed a gradual decrease in evenness. This means the relative abundance of species tended to be less even among species, and some species had become more abundant than others. The number of plant species alive over the period did not fluctuate much, starting at 68 and remaining at 67 in 3 subsequent surveys. There had been a total of 93 species that existed or once existed throughout the whole monitoring period, including both planted and colonizing species that died off.

The number of spontaneous species that appeared in the baseline survey (Oct 2014) was 54, and this remained stable at 51 in the last survey in Nov 2015. The number of spontaneous species that died off remained at 11 or 12 species throughout the three quarterly surveys. Newly colonized spontaneous species peaked in spring and early winter, but a relatively low number (5 species) were observed in summer. This could be explained by a relatively harsh site environment with strong evapotranspiration. Six species reappeared on the roof in July 2015. Four out of these six species were annual plants. They may likely be too small in size during the April 2015 survey to be discernible. The 3 species that reappeared in Nov 2015 were all small fleshy herbaceous plants—Torenia fournieri, Cajanus scarabaeoides, and Youngia japonica—which could hardly survive during intensive heat and sunshine, but they revived in winter instead.

3.2. Revolution of Planted and Spontaneous Vegetations Grouped under Different Plant Forms

Figure 6 shows the coverage percentage of all plant species grouped under 8 different plant forms (Annual Forb, Annual/Perennial Forb, Perennial Forb, Annual Graminoid, Perennial Graminoid, Fern, Shrub and Tree). Perennial forbs (37.4%) were the most abundant plant form on this green-roof site, followed by ferns (32.6%) and then annual forbs (17.7%) in October 2014, and the coverages evolved throughout the survey periods. Annual forbs successfully colonized the bare soil between planted ferns, and they flourished in the year 2014 and died off in the winter. Some annual forbs seeded themselves, and their next generation grew in the spring of 2015 and expanded slightly in the summer of 2015. The coverage of annual forb was 7% less in Nov 2015 than in the previous year, while coverage of ferns continuously expanded by nearly 14%, covering 46% of the green-roof area.

On this renaturalized extensive green roof with minimal maintenance input, it was observed that most spontaneous plants were annual and perennial forbs. The height of annual forbs started at 31 cm in Oct 2014, peaked at 39.4 cm in July 2015, and wilted back to 28.5 cm in Nov 2015. In general, most Annual/Perennial and Perennial forbs we transplanted on the roof are hardy ferns and forbs, which are tolerant under full sun with high evapotranspiration and yet shade-tolerant species. Annual/Perennial and Perennial forbs started at 27.9 cm in Oct 2014 and grew steadily to 32.8 cm in Nov 2015. Most fern species were planted species, with a total coverage of 32% in Oct 2014. Only 4 fern species appeared spontaneously, and they colonized less than 1% of coverage for each species throughout the whole monitoring period.

3.3. Species Abundance Distribution

The accumulative coverage abundance of all spontaneous species and plant species from Oct 2014 to Nov 2015 is shown in Figure 7a and Figure 8a, respectively, with the highest accumulative coverage species on the left and descending to the right. The spacing between lines reflected the coverage percentage of the given survey period. If lines overlapped, it implied that the coverage percentage of such species dropped to 0. It may die off completely, be dormant, or have its above-ground parts wilted. Equal spacing of lines implies stable coverage throughout the whole year. The cumulative abundance of spontaneous species ranged from 7.3% (Polygonum chinense) to 0.1% (Solanum nigrum and Symplocos paniculata). The coverage abundance of the majority of spontaneous plants fluctuated throughout the survey dates, except for 5 species, namely Oxalis debilis, Oxalis corniculata, Pilea microphylla, Phyllanthus tenellus, and Hedyotis diffusa. The most abundant spontaneous species, Polygonum chinense, only covered an average of 1.8%, and the least abundant spontaneous species, Solanum nigrum, covered an average of 0.05% during the surveyed period.

As shown in Figure 8a, Nephrolepis auriculata increased drastically from 13% to 28%, amounting to a cumulative coverage of 81%. This fern species had been either spreading aggressively via underground tubers or regrowing from the remaining rhizome left behind after the clearance. This fern could not be effectively removed by manual uprooting. Therefore, the decision to plant this species should be carefully deliberated. Cyclosorus parasiticus was the fern planted for renaturalization; its coverage decreased steadily from 8% to 3.4%; its survival rate was 37.5%. It was observed that this species did not perform well on sunny roofs with a high evaporation rate. Most other planted species had an average abundance ranging from 2.6% to 1.2%, significantly higher than the average abundance of spontaneous plants.

Figure 7b and Figure 8b are scatter diagrams showing species abundance distribution with accumulative coverage abundance for the species rank [29]. Species abundance distribution diagram often uses the logarithm scale [30]. However, the data for this study is best illustrated normally, as the variance between dominant and rare species in this site is smaller compared with mature habitats in other studies. Regression lines in 4 surveyed periods show the general trend of abundance through time. Among the 77 spontaneous species (which ever existed on this site), vigorous and dominant species generally had higher abundance earlier on and then gradually declined. The trend overturned when looking at less abundant species, showing that they colonized late and emerged late in the year. As for planted species, dominant planted species expanded at an increasing speed from Oct 2014 to Apr 2015 and then became stable in Jul and Nov 2015. Medium-abundant planted species were stable in coverage throughout the year. For the least abundant planted species, their coverage in Oct 2014 and Apr 2015 was slightly higher than in Jul and Nov 2015. The above is only a generalization of the trend. A great variance between species still exists and requires analysis from an additional perspective. Their seasonal growth pattern was further investigated by hierarchical cluster analysis in this study.

3.4. Seasonal Growth Pattern of Spontaneous Species

Figure 9 shows the 7 clusters of spontaneous species based on the pattern of coverage abundance in the year, using hierarchical cluster analysis. The seven clusters could be named after their traits, as shown in

Table 3. Traits of 7 clusters with number of species, range of average coverage percentage of spontaneous species, and typical pattern of coverage percentage.

Cluster Number Shown in Figure 9	Traits of Cluster	No. of Species	Range of Average Coverage % (Std Dev at Species Level)	Typical Pattern of Coverage % over the Year (14 October, 15 April, 15 July, 15 November)
1	Dominant, slight decline in summer	6	1.83–1.25% (0.195)	1.4/2.3/0.9/1.6
2	Thrive in spring, decline over winter	5	1.2–1% (0.083)	2.2/1/0.8/0.3
3	Sustain over the year, slight decline in winter	14	1.225–0.725% (0.130)	0.9/0.9/1.1/0.8
4	Decline in spring, regrow in summer	8	0.8–0.5% (0.122)	1/0.1/1.1/0.4
5	Late colonizer	14	0.775–0.2% (0.183)	0.2/0.1/0.1/0.9
6	Early colonizers, decline and die off	8	0.58–0.25% (0.125)	0.9/0.4/0.1/0.1
7	Minority, seasonal	22	0.5–0.025% (0.141)	0.1/0.3/0.3/0.2

. This table also illustrates the range of average coverage percentages of spontaneous species and the typical pattern of coverage over the year.

It is observed that most cluster groups of spontaneous species generally emerged in spring, declined slightly in summer due to intense sunshine, and then declined in winter. Average coverage of annual and annual/perennial species ranges from 0.8–0.45%, concentrating at the midway point of the coverage rank list. Spontaneous species with relatively higher coverage are mostly perennial, this might be due to the seasonal growth pattern of annual plants, while the perennial plants grow throughout the whole year. For cluster 1, these dominant spontaneous species, which emerged in the early phase and had an average coverage of 1.25% to 1.825%, are all perennial plants, with the only exception of an aggressive fast-growing annual forb Bidens alba. There is no special tendency observed in life cycle type for Clusters 2, 3, 5, and 7, while there is a higher proportion of annual species identified in both Clusters 4 and 6. Both Clusters 4 and 6 have the characteristics of declining over winter. It is not certain whether the species died or became dormant.

4. Discussion

4.1. Single Dominant Species Flourished, While Minorities Diminished, Others Remained Stable

Throughout the 1 year of this study, we observed that one single dominant fern species expanded steadily, while annual and perennial forbs (they are also minority species) declined in coverage, other species remained stable. Further study is needed to confirm the fate of minority species over a longer monitoring period. Our findings concurred with the findings by Hwang and others [31] based on an 18-month study in tropical Singapore that the notable contribution of coverage increased for specific dominant species while some others declined. It is also observed that spontaneous tree seedlings colonized from zero coverage to around 2%. Such a fast-growing regime is worth noting in warmer subtropical and tropical regions, where spontaneous trees could grow speedily. Unlike in temperate regions, spontaneous shrubs and trees are not common [32,33].

4.2. Annual Herbaceous Vegetation Fluctuated

Recent research found that there are about 20–30% of herbaceous species are annuals in South China [34], and our finding aligned with such findings. The numbers of annual species alive among herbaceous plant species alive at 4 respective surveys were 29%, 25%, 31%, and 21%. The numbers of annual species alive among herbaceous plant species that once existed on the roof were 22%, 19%, 24%, and 15% in 4 surveys. Our findings concurred with the study done in Qingdao City in northern China [35], that the number of spontaneous plant species peaked in the summer months. The coverage percentage of annuals decreased continuously from around 20% in Oct 2014 to 12% in Nov 2015, but their coverage did not peak in the summer months. Annual herbaceous vegetation may have been negatively affected by dominant fern species in terms of resources and changing site conditions. A competitive relationship between spontaneous trees and shrubs versus herbaceous groundcovers was observed in Singapore [31].

4.3. Floral Diversity Peaked and Leveled, While Dominant Species Expanded

Floral diversity peaked and then leveled with a slight decline during this 1-year short-term observation. Studies in Singapore on green roofs and rewilded lawns showed that floral diversity peaked in the 10th and 18th months, respectively [15,31]. Differences in site size could explain this differing pattern, as many studies proved that species richness is significantly correlated with patch area [15,36,37]. Although the absolute number of plant species alive on our studied green roof was maintained at 67, the slight decline in diversity metrics accounted for the increase in coverage abundance of dominant fern species. This implies that dominant species might have a negative influence on the ecological balance of this specific green roof in the long term, and this is applicable when other dominant species are continuously aggressively expanding.

4.4. Spontaneous Vegetation Being More Resilient than Planted Vegetation

Spontaneous vegetation was observed to be more resilient than planted vegetation and exhibited steady plant height throughout the year, whereas the height of planted vegetation decreased over the summer. Such differential plant performance to seasonal heat and longer sun exposure was not reported in previous studies. However, a study in the Mediterranean region [38] showed that the species richness and abundance of planted vegetation on those green roofs were strongly affected by variables like sun exposure substrate depth, while spontaneous vegetation exhibited a weak correlation with the above variables. Further long-term research is needed to ascertain whether seasonal heat shock persists among planted vegetation on green roofs.

4.5. Medium and Less Abundant Spontaneous and Planted Species Require Attention

Planted species ranked 3rd to 16th in cumulative abundance showed stable coverage, while only some spontaneous species exhibited stable coverage, and most of them had seasonal fluctuation, decline, or revival over the year. If landscape designers and managers are aiming to enhance biodiversity on green roofs, they could select and preserve those less stable spontaneous species grouped in Clusters 2, 4, 5, 6, and 7 in Table 3. These species help to diversify larval and nectar plants and add seasonal interest aesthetically. If green-space owners and managers put stable green-roof appearance as a priority, they can adopt planted species ranked 3rd to 16th and spontaneous species grouped in cluster 2.

4.6. Factors Determining Plant-Growth Patterns on Renaturalized Green Roof

There are three main factors determining the plant-growth pattern on this renaturalized green roof.

First, there is the availability of plant parts for sexual and asexual propagation and the timing of their arrival. The proximity of their mother plants and the availability of dispersal agents determined whether the species had a chance to colonize the site. Many previous studies [14,38,39] have suggested that the initial plant colonization is mostly dependent on the local surrounding seed rain, while abiotic filtering (e.g., substrate depth and exposure) has a strong impact on species diversity [40].

Second, plant physiological and life cycle traits, as compared with site abiotic factors, play a pivotal role. The matching of plant-growth requirements for specific species and adaptation to site factor changes, e.g., a change in irrigation regime and the removal of vegetation, is necessary. If a certain species is an annual plant, the removal of wilted plants before the seeds ripen may affect the amount of seed source.

Third, the growth dynamics of other plant species are constantly affecting the microhabitat of the site and the resource competition regime, such as sunshine, above- and below-ground space, water and nutrients, etc.

4.7. Management Recommendations for Balancing the Ecological, Aesthetic, and Functional Value of Extensive Green Roof

This study showed that some woody shrub and tree species and species with oversized plant forms colonized as spontaneous growth on this subtropical green-roof site. The structural safety and aesthetic value of the green roof could be undermined without proper maintenance input. If frontline gardeners cannot accurately identify the plant species, this could lead to delayed decision-making and possibly irreversible damage to structural safety or, to a lesser extent, water leakage. Thus, training for frontline gardeners and managers is essential. For those green roofs accessible by users, it is advised that owners and managers require further deliberation to determine the maximum height of plants to be allowed to grow to fit long-term management goals.

Table 4. Management recommendations for the extensive green roof after renaturalization for striking a balance of enhancing biodiversity and maintaining aesthetic and functional values.

Type of Plants with Following Traits	Management Recommendations	Reason(s)	Examples of Plant Species
Tree and shrub species	Remove all	Damage engineering structure of the extensive green roof	Celtis sinensis, Ficus subpisocarpa, Acacia confusa
Large-sized forb	Remove all	Shade out shorter plants	Alocasia macrorrhizos
Tall and/or fast-growing, easily germinated forb, ferns, graminoids	Remove all or most of them (aggressive forbs to be moved before/during flowering)	Shade out shorter plants Adversely affects the aesthetic value	Bidens alba, Laggera alata, Aster subulatus, Cyperus involucratus
Aggressive and fast spreading forb, ferns, graminoids with tubers	Remove all or most of them	Upset balance of species diversity	Nephrolepis auriculata, Polygonum chinense, Imperata cylindrica var. major
Mild forb, ferns, graminoids with tubers	Monitor and remove some or all, if necessary	May upset the balance of species diversity if conditions favor their growth	Pteris vittate, Desmodium heterocarpon
Thick rhizome species/fast-growing geophytes	Remove all	Damage engineering structure of green roof and can slowly spread to form large clumps	Curcuma spp.
Annual species with suitable mature height (except aggressive ones)	Allow them to self-seed over winter Remove dead stems in spring to early summer (only if dead stems are unaesthetic)	Preserve seed source and mimic the natural environment	Spilanthes paniculata, Drymaria cordata, Eclipta prostrata

outlines the management recommendations for minimal intervention and allowing maximum ecological diversity. These recommendations should apply to tropical and subtropical regions, where plant-growth speed is faster than in temperate regions.

4.8. Guiding Principles for Choosing and Maintaining Plant Species in Renaturalization Projects

Based on the above observation and analysis, below are the guiding principles for choosing plants to be propagated, installed, transplanted, and for maintenance:

• Landscape designers and managers should communicate the expected appearance of the concerned green roof and the amount and pattern of manpower required thoroughly. Both should understand all traits of the plant species chosen.
• A summary or pictorial guide for all introduced plant species should be available to all landscape managers and frontline gardeners for easy identification, and their means of propagation (e.g., estimated germination rate) and effective removal should be taught.
• If native fern species are planned to be adopted in the renaturalization project, those fern species should be selected with great care to avoid aggressive species undermining biodiversity. Additionally, their mature height should not be underestimated by referencing their sizes in pots in nurseries. Their mature height and size should be compatible with other species to avoid competition for sunshine and resources.
• Annual forb species could be invisible over the winter months and thrive in the summer months. If some of them (e.g., Bidens alba) are too aggressive against other species, they could be problematic to deal with after the seeds are ripe.
• To control the spread and coverage of certain target species in subtropical regions, mowing may not be the best management practice, unlike what was suggested for temperate regions [1]. Total uprooting of the plants is sometimes a practical solution under the constraints of manpower. Understanding the growth cycle and physical structure of plants will help in formulating optimal management strategies.

4.9. Limitations of the Study

As confined by the available manpower of the study, no further data are available to assess the long-term plant-growth dynamics. It would be assertive to draw arbitrary conclusions on the fate of respective clusters of spontaneous vegetation. Future studies for medium and long-term (e.g., 2–5 years, 5–10 years) plant-growth monitoring will shed light on the role of horticultural management in enhancing and maintaining the ecological, aesthetic, and functional value of naturalized biotope on green roofs under different accessibility and recreational needs. Other studies showed that spontaneous species contribute to 60% of the variance between roofs [31], which implies that systematic studies on green roofs with multiple sites or experimental plots under controlled variables will help to answer certain questions.

The coverage abundance and height data were collected for this study, it is rather impractical to identify whether the species died at the survey date, and further continuous observation over months is needed to confirm no revival of specific species.

The diversity of microbial activity and how it affected the activity in the health of green-roof plants are not accounted for in this study. They might increase plant drought tolerance, nutrient availability, pathogen protection, and substrate stabilization. Future research directions could investigate this aspect to formulate best practices for balancing microbial activity and refraining from applying additional fertilizer, which might lead to stormwater pollution.

Stormwater mitigation [41], water use and drought response [42], and other functional traits (e.g., substrate temperature reduction, nitrate retention, temperature, and precipitation) [14,43,44] were out of the scope of this paper; however, they will be essential topics in water-sensitive regions and to regions opt to maximize ecosystem services of green roofs.

5. Conclusions

Understanding the short-term growth dynamics of both spontaneous and planted vegetation to balance aesthetic and ecological functions in management plans on extensive green roofs is a prerequisite to designing and implementing renaturalization projects for enhancing urban biodiversity. This paper provided a detailed analysis of plant colonization and decline over a year. This filled the research gap for planning, designing, and maintaining a biodiverse biotope when integrating other research findings to inform a long-term effective management plan. Scientific novelty was clearly demonstrated with sound theoretical and practical significance.

This study examined and documented how a manicured green roof with merely 7 plant species transformed into a biodiverse habitat by replacing selected species with native flora and by colonizing spontaneous herbaceous plants. This study revealed that the species number of planted and spontaneous plant species remained stable for a year, with seasonal turnover of spontaneous and revived species. In terms of coverage abundance, dominant aggressive fern outperformed other plant forms together with a declining trend of annual and perennial forbs, implying a slight decrease in plant diversity. Seasonal growth tendencies of both planted and spontaneous plants were illustrated in relationship to their species rank and further hierarchical cluster analysis for clustering spontaneous species. Their differential growth patterns provided comprehensive information supporting the decision on plant selection and maintenance, which is the scientific novelty within this unexplored topic. We concluded by suggesting important guiding principles on plant selection and management recommendations based on the findings to meet both aesthetic and ecological needs. This provided valuable insights into balancing biodiversity and maintenance needs, which a recent paper [11] called for research attention. Future research could focus on plant colonization on non-irrigated extensive green roofs in tropical/subtropical regions to find out whether and how water limitation would affect the colonization and growth of spontaneous and planted vegetation. The perception of green-roof users and managers and the subsequent maintenance input are essential socio-economic aspects to consider, especially in urban areas. People will sometimes complain about weedy green spaces as eyesores, regardless of whether they are privately or publicly owned. This dimension of consideration should be thoroughly understood to build the confidence of landscape managers.