2013 Pefromance of geophytes on extensive green roofs in the uk

This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution and sharing with colleagues.

Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party websites are prohibited.

In most cases authors are permitted to post their version of the article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information regarding Elsevier’s archiving and manuscript policies are encouraged to visit: http://www.elsevier.com/authorsrights

Author's personal copy Urban Forestry & Urban Greening 12 (2013) 509–521

Contents lists available at ScienceDirect

Urban Forestry & Urban Greening journal homepage: www.elsevier.com/locate/ufug

Performance of geophytes on extensive green roofs in the United Kingdom Ayako Nagase a,∗ , Nigel Dunnett b,1 a b

Chiba University, Graduate School of Engineering, Division of Design Science, 1-33 Yayoicho, Inage-ku, Chiba-shi, Chiba 263-8522, Japan University of Sheffield, Department of Landscape, Arts Tower, Western Bank, Sheffield S10 2TN, UK

a r t i c l e

i n f o

a b s t r a c t

Keywords: Aesthetic Biodiversity Covering plants Sedum Substrate depth Urban landscape

Dwarf geophytes have great potential for use on extensive green roofs because they often come from arid areas and can survive dry and hot summer in a dormant state. However, there has been little research regarding geophytes on green roofs. This experiment was conducted to study the performance of 26 species of geophytes on a green roof during 2005–2006 in Sheffield, UK. The geophytes were grown at two substrate depths (5 cm and 10 cm) of substrate on a green roof without irrigation. To investigate the susceptibility of geophytes to competition from a covering of permanent plants, the geophytes were grown with or without a surface vegetation layer of Sedum album. Overall, the growth, survival rate, regeneration and flowering of geophytes were more successful at a substrate depth of 10 cm than of 5 cm, probably because of improved moisture retention, fewer temperature fluctuations and the protection from digging by animals. The flowering period was limited to spring, therefore, it is recommended to combine with other plant species such as covering plants. Geophyte species did not compete much with S. album and Sedum cover had no significant effects on the growth, survival rate, regeneration and flowering of geophytes in most species. Iris bucharica, Muscari azureum, Tulipa clusiana var. chrysantha, Tulipa humilis, Tulipa tarda and Tulipa turkestanica had good performance at the substrate depth of 5 cm. In addition, Narcissus cyclamineus ‘February gold’ and Tulipa urumiensis exhibited a successful performance at the substrate depth of 10 cm. © 2013 Elsevier GmbH. All rights reserved.

Introduction

(Dunnett and Kingsbury, 2008). However, both Sedum roofs and biodiverse roofs often lack aesthetic appeal. Sedum spp. are usually evergreen, however, they only flower for a limited period and change little throughout a year (Kadas, 2006). Biodiverse roofs often lack the lush green appearance of green roofs and have an appearance similar to that of brownfields (Dunnett and Kingsbury, 2008).

The visual appearance may not be a concern if the roof is generally not visible and is installed primarily for its functional attributes such as storm water retention (Getter and Rowe, 2006). However, aesthetics may be important if green roofs are visible and actively used.

Geophytes are important plants species that can adapt to harsh environment found on extensive green roofs. They are plants with a swollen storage organ, such as true bulbs, corms, tubers and rhizomes (Raunkiaer, 1934; Mathew and Swindells, 1994). Bulb (e.g.

Narcissus, Tulipa and Lilium) has a short stem surrounded by fleshy scale leaves or leaf bases. Corm (e.g. Crocus, Colchicum and Gladiolus) consists of a swollen stem base covered with scale leaves.

Rhizome (e.g. Iris) has a continuously growing horizontal underground stem which puts out lateral shoots and adventitious roots at intervals. Tuber (e.g. Begonia, Anemone and Cyclamen) has a thickened underground part of a stem (Oxford dictionaries, 2010). The structures are different, but these plants act in the same manner,

Green roofs have gained global acceptance as a technology with potential to help mitigate the multifaceted and, complex environmental problems of urban centres (Clark et al., 2008). Extensive green roofs are characterized by their light weight, low maintenance requirements, little or no irrigation systems requirements and thin substrate depths (2–20 cm). They are widely used because they are easy to install on existing buildings without structural modifications and they are inexpensive (Oberndorfer et al., 2007).

The most commonly used species on extensive green roofs are Sedum spp. because they can tolerate extreme temperatures, high winds, low fertility and limited water supplies (Durhman et al., 2006, 2007; Van Woert et al., 2005). Recently, biodiverse roofs are often used for extensive green roofs. This type of extensive green roof is used to recreate conditions found in typical urban brownfield sites in order to enhance their potential biodiversity value

∗ Corresponding author. Tel.: +81 0432903113; fax: +81 0432903121.

E-mail addresses: a-nagase@faculty.chiba-u.jp (A. Nagase), n.dunnett@sheffield.ac.uk (N. Dunnett).

1 Tel.: +44 01142220611; fax: +44 01142754176.

1618-8667/$ – see front matter © 2013 Elsevier GmbH. All rights reserved. http://dx.doi.org/10.1016/j.ufug.2013.06.005

Author's personal copy 510

A. Nagase, N. Dunnett / Urban Forestry & Urban Greening 12 (2013) 509–521

i.e. these structures all play the role of storage organs that allow plants to retreat underground for long periods of dormancy (Garrett and Dusoir, 2004). There are several benefits of using geophytes on extensive green roofs. First, their ecological characteristics are appropriate for the green roof environment. Some geophytes are highly drought tolerant and may perform well on extensive green roofs without irrigation in the UK. These drought-tolerant geophytes often come from dry climates, such as South Africa, the Mediterranean basin and Central Asia (Kingsbury, 1996), where winters are wet and summers are hot and dry, with a short spring (Phillips and Rix, 1989). These plants can grow, flower and seed during cool moist seasons and disappear into the comparative cool of the earth during hot summers (Kingsbury, 1996). The growing season is short, and the plants use their stored energy to flower and quickly set seed during spring (Blamey and Blamey, 1979).

Second, they produce colourful flowers early in the season when many herbaceous plants have not started growing. The early flowering of geophytes provides colours and is also very useful for providing nectar sources at a time of year when little else may be flowering on a green roof. The first spring-flowering geophytes are a lifeline for overwintering insects in search of nectar after a long period of dormancy (Dunnett, 2004). Third, they require little maintenance and their storage organs often act as a means of propagation (vegetative reproduction). Geophytes grow and flower for a short time after planting. After planting, geophytes usually require little maintenance while some multiply rapidly when the growth conditions are suitable. For example, Muscari armeniacum proliferated rapidly after appearing spontaneously on an over 30years-old extensive grass roof in the UK (Dunnett and Kingsbury, 2008).

However, there are also disadvantages of using geophytes because the flowering periods of individual plants are relatively short and they become unsightly after flowering. In addition, geophyte species that are potentially suitable for extensive green roofs tend to exhibit winter to early summer growth. Thus, it is recommended that they are combined with other plants, such as like plants that cover the ground throughout the year. Mathew (1997) discussed the benefits of using creeping or carpeting plants with geophytes; the flower stems of geophytes receive some support and blooms are protected from soil splashes during heavy rain. However, it is necessary to avoid vigorous plants for cover plants with geophytes (Elliott, 1995). Previous studies have shown that the growth of geophytes was reduced because of competition with covering grass (Hughes, 1986). In a study of the competition between Allium vineale and Lolium perenne, emergence and growth of A. vineale were affected (Lezenby, 1961). These studies were conducted on the ground, but, it is predicted that vigorous covering plants may lead to nutrient removal and moisture stress (Hewson and Roberts, 1973).

Although there has been little research on how geophytes perform on extensive roofs, geophytes have been used on extensive green roofs. Allium spp. is one of the most commonly encountered geophytic genus on extensive green roofs (Dunnett and Kingsbury, 2008). Long-term research on extensive green roofs in Berlin showed that Allium schoenoprasum was the most dominant plant species throughout 20 years because of self-seeding (Köhler and Poll, 2010). Lilium auratum has been used in traditional thatched roofs in Japan for reinforcement and for its aesthetics.

Dwarf geophytes may be more appropriate for extensive green roofs because they are more drought tolerant than large hybrids (Glattstein, 2005; Snodgrass and Snodgrass, 2006). Short species may also be better at withstanding wind on green roofs, whereas tall or top-heavy flowers would not withstand on a windy site (Rees, 1992). Storage organs of dwarf geophytes are also small; therefore, they can tolerate a shallow planting and are expected to be better adapted to thin substrates.

The aim of this study was to identify appropriate geophyte species for extensive green roofs and to investigate how substrate depth and covering plants of Sedum spp. may affect the performance of geophytes on extensive green roofs. The effect of substrate depth was studied because it often limits the root growth and the availability of water and nutrients and it may be an important factor that affects plant performance on extensive green roofs (Dunnett et al., 2008; Olly et al., 2011; Rowe et al., 2012). Sedum spp. were used as covering plants because they are one of the most frequently used species for extensive green roofs. In addition, compared with other plants, Sedum spp. are expected to offer less competition to geophytes because they are low growing plants with shallow roots and require little water and nutrients.

Methods Experimental setup The experiment was initiated in December 2004 on the roof of a four-storey commercial building near the city centre in Sheffield, UK. The green roof was framed by timbers and consisted of root protection barriers, drainage layers (Floradrain FD 25/25E) and a commercial green roof substrate composed of crushed recycled brick and 10% organic material (Zinco sedum substrate and Zinco semi-intensive substrate 1:1, ≤7–15%, in which the granules measured <0.063 mm in diameter; salt content ≤2.5%; porosity 63–64%; pH 7.8–7.9; dry weight 940–980 kg/m3 ; saturated weight 1240–1360 kg/m3 ; maximum water capacity 25–42%; air content at maximum water capacity 22–38%; water permeability ≥0.064–0.1 cm/s) (Alumasc, 2006). Zinco substrate was obtained from Alumasc (Northamptonshire, UK). The substrate depth (5 cm and 10 cm) and covering plants (with and without Sedum album) were the variables. We tested substrate depths of 5 cm and 10 cm because a depth of at least 5 cm was necessary to cover geophytes.

It was estimated that the substrate depth of 5 cm depth was too thin to allow the sufficient growth of some geophytes; therefore, a substrate depth of10 cm depth was also tested. Half of plots were left without covering plants to test whether Sedum spp. affect the performance of geophytes on extensive green roofs. There were three replicates for each combination; hence, a total of 12 plots were arranged randomly (Fig. 1). These plots received similar length of sunlight. Each plot measured 60 cm × 145 cm and was divided into 30 subplots (12 cm × 24 cm) (Fig. 1). Each plot was framed by timbers, however, there were no partitions between each subplot. In each subplot, three individual geophytes from a single species were planted in a line. Twenty-six plant species were planted as underground storage organs on January 14, 2005. Storage organs were small; from 1.5 cm to 3.0 cm. Therefore, subplots provide enough space to grow geophytes. Name of plants and their characteristics are described in Table 1. Geophytes were obtained from Dutchbulbs (Manchester, UK). They are popular plants for typical gardens and they are easy to get and low price. These plants were expected to be well-adapted to extensive green roofs because they naturally in European or Asian alpine regions mainly with rocky or stony substrate and relatively low temperature. They tended to be short height, which is appropriate for green roof environment to resist strong wind. Twenty-six subplots were used out of 30 subplots. 4 subplots were left empty. The empty subplots were chosen in random. The geophytes were planted at a depth of 3 cm below the substrate surface using two different total substrate depths, i.e.

5 cm and 10 cm (Fig. 1). S. album seeds (0.5 g) were sown in Sedum cover plots on April 30, 2005 as covering plants. The seeds were obtained from Jelitto (Schwarmstedt, Germany). S. album seeds were too small to distribute over the plot; therefore, they were mixed with horticultural sand. It took 1 year for S. album to cover

Author's personal copy A. Nagase, N. Dunnett / Urban Forestry & Urban Greening 12 (2013) 509–521

511

Table 1 Characteristics of the geophyte species used in this study.

Family

Distribution

Typical habitat

Flowering season

Height

Allium flavum

Liliaceae

Dry hills

July

25–30 cm

Allium karataviense ‘Ivory queen’

Liliaceae

Semi-desert mountain

May

15 cm

Allium ostrowskianum

Liliaceae

Stony slopes at 3000–3800 m

May–June

15 cm

Allium unifolium

Liliaceae

Southern Europe from France to Greece Central Asia, especially Turkestan Central Asia, especially in Tien Shan, the Pamir Alai and the Ala Tau, as well as eastern Turkey and the Caucasus North California and south Oregon in the coast ranges

May–July

45 cm

Crocus sieberi ‘Tricolor’

Iridaceae

Crete and Greece

February–March

8 cm

Crocus tommasinianus

Iridaceae

Dalmatia, South Hungary, Yugoslavia and North Bulgaria

February–March

8 cm

Crocus vernus ‘Vanguard’

Iridaceae

February–June

10 cm

Iris bucharica

Iridaceae

Stony and grassy hills from 800 to 2400 m

March–April

30 cm

Iris danfordiae

Iridaceae

2000–3000 m, on bare, earthly hills

February–March

10 cm

Iris reticulata

Iridaceae

Europe from the Pyrenees eastwards to Poland and Russia, and south to Sicily and Yugoslavia Central Asia, especially the Pamir Alai, Tajikistan and northeast Afghanistan Central Turkey, in the Taurus, in west Malatya, Amasya and Gümüsane Russia, Turkey, Iran, South Transcaucasia and Iraq

Coast ranges, growing in the moist substrates in pine or mixed evergreen forest below 1200 m Short mountain turf or open woodland In woods and shady hillsides, especially on limestone, at ground 1000 m Mountains

March–May

20 cm

Ixioliron pallasii

Amaryllidaceae

Scree and bare stony places and among scrub from 600 m to 2700 m In fields and on hillsides from 200 to 2700 m

June

30 cm

Muscari azureum

Iridaceae

High elevation

March

15 cm

Narcissus cyclamineus ‘February gold’

Amaryllidaceae

Early March

30 cm

Puschkinia libanotica

Liliaceae

March–April

15 cm

Scilla siberica

Liliaceae

River banks and damp mountain pastures In scrub, in stony places and in meadows in mountains, at up to 3000 m Woods, scrub and among rocks, up to 2000 m

March–April

10 cm

Sparaxis tricolor Tulipa bakeri ‘Lilac wonder’ Tulipa clusiana var. chrysantha

Iridaceae Liliaceae Liliaceae

Cape province Fields and rocky places Stony mountain sides

June March April

25 cm 25 cm 20 cm

Tulipa hageri ‘Splendens’

Liliaceae

Cornfield and stony places

April

20 cm

Tulipa humilis

Liliaceae

Stony hillside

March

8–10 cm

Tulipa kolpakowskiana

Liliaceae

Rocky slopes up to 2000 m

May

15 cm

Tulipa linifolia

Liliaceae

Mountains

Mid-May

12 cm

Tulipa saxatilis Tulipa tarda

Liliaceae Liliaceae

Fields and rocky places Stony and rocky slopes

Early May April

15 cm 12 cm

Tulipa turkestanica

Liliaceae

Early May

15 cm

Tulipa urumiensis

Liliaceae

Stony slopes, by streams and on rock ledges from 1800 m to 2500 m Not known in the wild

April

10–15 cm

Adapted from Botschantzeva (1982) and Phillips and Rix (1989).

Western Asia from Turkey, Egypt and eastwards to western Siberia Caucasus and northwest Turkey Northwest Portugal and northwest Spain Caucasus, south Turkey, North Iraq and from Iran to Lebanon South Russia, Caucasus and Turkey southwards to Siberia, naturalized in east Europe South Africa Crete (Greece) Iran, near Shiraz eastwards to the Himalayas from Afghanistan to Kumaon, and naturalized in South Europe Greece, Crete, Bulgaria and west Turkey East Turkey, north Iraq and north west Iran Central Asia, especially the northern Tien Shan and southern Ala Tau Uzbekistan, north Iran and Afghanistan Crete (Greece) Central Asia, especially Tien Shan Central Asia, especially Tien Shan, the Pamir Alai, and north west China Iran

Author's personal copy 512

A. Nagase, N. Dunnett / Urban Forestry & Urban Greening 12 (2013) 509–521

Fig. 2. Overview of the study site.

Fig. 1. Layout of the study plots and the arrangement of geophytes within the plots and planting positions used for the geophytes at the substrate depths of 5 cm and 10 cm.

fully plots of both substrate depth of 5 cm and 10 cm. All of the plots received only rain water. An overview of the experimental site is shown in Fig. 2.

Data collection In the first year, the emergence rate (the percentage of plants exhibiting above-ground emergence) was measured 4 times between March 24 and April 28, 2005. In the second year, growth was measured 11 times (between March and July = 22 weeks). It was considered that recording the second year of growth would be important since the geophytes had overwintered on the roofs. The first year of growth was largely determined by the state of the bulb before planting (i.e. how it was grown prior to planting). However, growth in the second year was likely to be strongly determined by the actual growing conditions in situ. In addition, the covering plants may have had little effect in the first year because S. album developed after the growth of geophytes was completed in the first year. Therefore, the emergence rate was the only variable measured in the first year.

The parameters measured were as follows: leaf length (the longest leaf from the proximal to the apex); total leaf number (including small leaves more than 5 mm); flower height (the longest flower stem, from the proximal to the apex); proximal shoot number (number of shoots grown by vegetative reproduction); the period of above-ground growth (from emergence to completely drying out); flowering duration (from the opening of the buds to the end of flowering). Most emerged plants produced only one proximal shoot in the first year; however, the number increased in the second year because of vegetative reproduction.

Therefore, we treated the maximum proximal shoot number as an indicator of vegetative reproduction. These parameters were

chosen with the reference of previous studies to evaluate growth of geophytes (Lazenby, 1962; Ernst, 1979; Barkham, 1980a,b).

A few weeds were colonized in these experimental plots, however, they were removed when the measurements were carried out.

Mean monthly temperature and rainfall during 2005 and 2006 in Sheffield are shown in Fig. 3. Winter was mild in 2005, whereas the summer temperature was lower than the one of average. Annual precipitation in 2005 was higher than average. In 2006, winter was cold whereas summer was dry and warm. There was a small amount of rainfall during July in 2006.

Fig. 3. Changes in the mean monthly temperature and rainfall for Sheffield, UK during 2005 and 2006.

Source: Met office (2007).

Author's personal copy A. Nagase, N. Dunnett / Urban Forestry & Urban Greening 12 (2013) 509–521

513

and the substrate depth of 5 cm than with Sedum and substrate depth of 10 cm. These results suggest that covering plants had various types of effects on emergence; most species were not affected by the Sedum cover, some species competed with Sedum and some were encouraged by the Sedum cover.

The percentages of emergence differed among the species.

Although the substrate depth was important for emergence in the second year, the overall percentage of emergence in the first and second years tended to be similar. For example, I. bucharica, M. azureum, N. cyclamineus ‘February gold’, T. clusiana var. chrysantha, T. humilis, T. turkestanica and T. urumiensis even had high emergence rates at the substrate depth of 5 cm in the first and second years. By contrast, A. karataviense ‘Ivory queen’, C. sieberi ‘Tricolor’, C. tommasinianus, C. vernus ‘Vanguard’, I. danfordiae, S. siberica, T. hageri ‘Splendens’ and T. kolpakowskiana had low emergence rates in the first and second years.

Growth Fig. 4. Mean final percentages of emergence rates of all geophytes per plot during the second year (n = 3). Error bars represent the standard errors of the means.

Two-way ANOVA showed that the substrate depth had a significant effect on mean emergence (P < 0.05). Means with the same letter are not significantly different.

Data analysis T-Test was used to detect the effects of different substrate depths on emergence in the first year (Minitab Release 14). Twoway analysis of variance (ANOVA) was used to detect the effects of different substrate depths and covering plants (with and without) on the other parameters measured (Minitab Release 14).

This statistical analysis was carried out to compare the growth of plant within the same plant species in different treatments. If there were significant differences, the means were separated using Tukey test. However, the results of two-way ANOVA and Tukey test were sometimes contradictory, e.g. two-way ANOVA showed a significant difference, whereas Tukey test showed no significant difference. If this was the case, the result of two-way ANOVA was used to analyze the data. For all analyses, the threshold of significance was set at P < 0.05.

Results Emergence In the first year, the substrate depth had no significant effect on emergence. However, emergence was significantly higher with the deeper substrate during the second year (P < 0.05) (Fig. 4). This suggests that the substrate depth was more important for the survival rate over winter rather than emergence after a few months of planting. The covering plants and the interaction between the substrate depth and covering plants had no significant effects on emergence.

No significant difference was observed, however, emergence rate was higher with Sedum than without. This trend was more obvious at the shallow substrate.

In mean percentage of emergence of individual species, the results showed that most species exhibited better emergence with the deep substrate (Appendix 1 for first year, Table 2 for second year). In the second year, the substrate depth had a significant effect on five species (A. unifolium, N. cyclamineus ‘February gold’, S. siberica, S. tricolor and T. turkestanica). About half of the species had higher emergence with Sedum. The covering plants had a significant effect on the emergence of five species (A. flavum, A. ostrowskianum, I. reticulata, T. linifolia and T. turkestanica). Only T. linifolia had higher emergence without Sedum. A. flavum was the only species with a significant interaction effect between the two treatments. In this species, the percentage of emergence was much higher with Sedum

Growth was measured 11 times (from March to July = 22 weeks) and individual species exhibited growth peaks at different times.

Therefore, only the maximum leaf length, leaf number and flower height of individual species are shown in Table 3 (leaf length) and Appendix 2 (leaf number and flower height). Overall growth with the deeper substrate was better; these plants exhibited longer leaf length, higher number of leaves and greater flower height.

About half of the species were significantly affected by the substrate depth, and all of these species showed better growth with the deeper substrate. An exception was the leaf length of A. unifolium, which was longer at the substrate depth of 5 cm, whereas larger number of leaves was produced at the substrate depth of 10 cm.

The effect of covering plants appeared to vary among species.

Similar to the results of emergence, about half of the species performed better with Sedum. However, more species grew better without Sedum at substrate depth of 10 cm. The growth of two species, S. tricolor and T. turkestanica, was affected significantly by covering plants. Sparaxis tricolor grew better without Sedum, whereas the leaf growth of T. turkestanica was increased with Sedum.

Some species did not appear to exhibit sufficient growth, such as A. karataviense ‘Ivory queen’, C. siberi ‘Tricolor’, C. tommasinianus, C. vernus ‘Vanguard’, I. danfordiae and T. hageri ‘Splendens’.

Flower performance Similar to the results of emergence and growth, many species exhibited longer flowering periods with the deeper substrate (Table 4). The effect of the substrate was statistically significant for three species (M. azureum, N. cyclamineus ‘February gold’ and S. siberica). Again, the effect of covering plants differed among species.

However, most geophyte species were not significantly affected in duration of flowering by the Sedum cover. Only two species were significantly affected by the covering plants, i.e. T. kolpakowskiana and T. turkestanica. They showed longer flowering with the Sedum cover.

Table 5 shows the flowering periods of the individual species.

The flowering period was determined as the start of flowering (at least 1 plant of a given species showed flowering) until the end of flowering (no plants of a given species showed flowering) for all plots containing individual species. The combination of geophyte species produced over 4 months of flowering. Highest number of species showed flowering from the end of April to the beginning of May. In general, the flowering season of geophytes was not long, lasting less than 2 months. M. azureum, I. reticulata and P. libanotica had long flowering seasons because individuals flowered

Author's personal copy 514

A. Nagase, N. Dunnett / Urban Forestry & Urban Greening 12 (2013) 509–521

Table 2 Mean percentage of emergence of individual species per plot in response to substrate depth and Sedum cover in the second year (percentage, n = 3).

Substrate depth

5 cm

10 cm

Sedum cover Allium flavum Allium karataviense ‘Ivory queen’ Allium ostrowskianum Allium unifolium Crocus sieberi ‘Tricolor’ Crocus tommasinianus Crocus vernus ‘Vanguard’ Iris bucharica Iris danfordiae Iris reticulata Ixioliron pallasii Muscari azureum Narcissus cyclamineus ‘February gold’ Puschkinia libanotica Scilla siberica Sparaxis tricolor Tulipa bakeri ‘Lilac wonder’ Tulipa clusiana var. chrysantha Tulipa hageri ‘Splendens’ Tulipa humilis Tulipa kolpakowskiana Tulipa linifolia Tulipa saxatilis Tulipa tarda Tulipa turkestanica Tulipa urumiensis

88.89 0 77.78a 33.33b 22.22a 0a 22.22a 100 0a 44.44a 66.67a 100 88.89a 55.56a 33.33a 0b 44.44a 100 11.11a 88.89a 22.22a 33.33a 44.44a 66.67a 77.78ab 88.89a

Without Sedum cover

Sedum cover

22.22 11.11 44.44a 44.44ab 11.11a 11.11a 22.22a 100 22.22a 11.11ab 66.67a 100 66.67a 44.44a 0b 0b 22.22a 100 11.11a 77.78a 11.11a 66.67a 44.44a 77.78a 11.11b 100a

33.33 11.11 100a 88.89a 22.22a 0a 33.33a 100 0a 66.67b 55.56a 100 100a 66.67a 66.67a 66.67a 22.22a 100 33.00a 88.89a 44.44a 55.56a 44.44a 55.56a 100a 100a

S.E.

Probability

14.96 11.45 14.16 14.43 13.89 5.56 14.43 – 7.35 14.43 17.12 – 10.02 16.67 14.70 10.02 14.70 – 11.45 13.03 15.47 15.71 16.90 16.67 11.79 7.86

C: P < 0.05, D*C: P < 0.05 n.s.

C: P < 0.05 D: P < 0.01 n.s. n.s. n.s.

– n.s.

C: P < 0.01 n.s.

Without Sedum cover 33.33 33.33 66.67a 88.89a 22.22a 0a 11.11a 100 0a 11.11b 44.44a 100 100a 77.78a 44.44a 88.89a 11.11a 100 0a 77.78a 44.44a 88.89a 77.78a 55.56a 77.78ab 88.89a

D: P < 0.05 n.s.

D: P < 0.01 D: P < 0.01 n.s. n.s. n.s. n.s.

C: P < 0.01 n.s. n.s.

D: P < 0.01. C: P < 0.01 n.s.

Two-way ANOVA was used to compare values within a species. Means with the same letter are not significantly different. S.E. = standard error, P = probability, D = substrate depth regime, C = Sedum cover regime, D*C = interaction between the substrate depth regime and the Sedum cover regime, n.s. = not significant.

Table 3 Maximum leaf length of individual species per plot in response to the substrate depth and Sedum cover (cm, n = 3).

Substrate depth

5 cm

10 cm

Sedum cover

Without Sedum cover

Sedum cover

Without Sedum cover

S.E.

Probability

n.s. n.s. n.s.

D: P < 0.01 n.s. n.s. n.s.

D: P < 0.01 n.s.

D: P < 0.05 D*C: P < 0.05 D: P < 0.05 D:P < 0.01 n.s.

D: P < 0.01 D: P < 0.01 C: P < 0.05 D*C: P < 0.05 n.s. n.s. n.s. n.s.

D: P < 0.05 D: P < 0.05 n.s. n.s.

D: P < 0.01 C: P < 0.05 D: P < 0.01

Allium flavum Allium karataviense ‘Ivory queen’ Allium ostrowskianum Allium unifolium Crocus sieberi ‘Tricolor’ Crocus tommasinianus Crocus vernus ‘Vanguard’ Iris bucharica Iris danfordiae Iris reticulate Ixioliron pallasii Muscari azureum Narcissus cyclamineus ‘February gold’ Puschkinia libanotica Scilla siberica Sparaxis tricolor

May 22 June 6 April 27 April 27 March 29 March 17 April 27 June 6 March 29 May 4 June 6 May 22 May 4 April 27 May 22 May 22

4.34a 0a 4.46a 10.44a 1.00a 0a 0a 18.70c 0a 1.82b 1.36a 6.78a 0.33b 3.86a 1.14ab 0b

7.28a 1.89a 5.59a 11.22a 0.37a 0.24a 0a 18.36bc 0.7a 0b 13.42a 9.13a 6.08b 1.62a 0b 0b

16.88a 0a 2.28a 3.98a 0.30a 0a 2.57a 24.64ab 0a 11.22a 12.61a 10.47a 21.66a 3.14a 6.18a 2.88b

6.04a 0.22a 3.50a 4.21a 0.74a 0a 2.11a 25.50a 0a 2.68ab 6.06a 10.01a 18.90a 6.12a 4.63ab 8.17a

3.67 0.62 1.62 1.99 0.61 0.12 1.36 1.61 0.35 2.84 4.02 1.08 2.223 1.39 1.49 1.03

Tulipa bakeri ‘Lilac wonder’ Tulipa clusiana var. chrysantha Tulipa hageri ‘Splendens’ Tulipa humilis Tulipa kolpakowskiana Tulipa linifolia Tulipa saxatilis Tulipa tarda Tulipa turkestanica

April 27 April 27 April 13 April 13 April 13 April 27 March 29 April 27 April 27

7.77a 12.94a 1.68a 6.87a 2.14a 1.68b 5.40a 5.23a 13.66a

1.50a 13.81a 1.11a 5.66a 0a 3.28ab 3.77a 7.96a 1.67b

1.20a 13.99a 3.89a 5.88a 5.07a 5.09ab 3.03a 5.16a 18.67a

0.78a 14.21a 0a 7.50a 3.26a 7.57a 4.39a 5.37a 16.23a

1.99 1.64 1.41 1.46 1.50 1.45 1.88 1.73 2.67

Tulipa urumiensis

May 22

3.32b

10.48a

10.19a

1.38

5.87ab

The date indicates the day when the maximum leaf length was recorded for an individual species. Two-way ANOVA was used to compare values within a species. Means with the same letter do not differ significantly. S.E. = standard error, P = probability, D = substrate depth regime, C = Sedum cover regime, D*C = interaction between substrate depth regime and Sedum cover regime, n.s. = not significant.

Author's personal copy A. Nagase, N. Dunnett / Urban Forestry & Urban Greening 12 (2013) 509–521

515

Table 4 Mean flowering periods of individual species in response to the substrate depth and Sedum cover (week, n = 9).

Substrate depth

5 cm

Allium flavum Allium karataviense ‘Ivory queen’ Allium ostrowskianum Allium unifolium Crocus sieberi ‘Tricolor’ Crocus tommasinianus Crocus vernus ‘Vanguard’ Iris bucharica Iris danfordiae Iris reticulate Ixioliron pallasii Muscari azureum Narcissus cyclamineus ‘February gold’ Puschkinia libanotica Scilla siberica Sparaxis tricolor Tulipa bakeri ‘Lilac wonder’ Tulipa clusiana var. chrysantha Tulipa hageri ‘Splendens’ Tulipa humilis Tulipa kolpakowskiana Tulipa linifolia Tulipa saxatilis Tulipa tarda Tulipa turkestanica Tulipa urumiensis

10 m

Sedum cover

Without Sedum cover

Sedum cover

Without Sedum cover

1.56 0 0 0 0 0 0a 2.00a 0 0.44a 0.22a 2.67b 0b 0.67a 0a 0 0a 1.78a 0a 1.78a 0.44a 0.44a 0 1.33a 2.67a 1.56a

0.67 0 0 0.44 0 0 0a 1.78a 0 0a 0.22a 4.22ab 0.67b 0.44a 0a 0 0.22a 1.78a 0.22a 1.11a 0a 0.67a 0 1.56a 0.44b 1.56a

0.22 0 0 0.44 0 0 0a 2.22a 0 0.22a 0.22a 6.22a 3.11a 0.67a 0.89a 0 0.22a 1.78a 0a 1.11a 0.67a 0.67a 0 1.11a 2.89a 2.22a

0.89 0.67 0 0.22 0 0 0.22a 2.22a 0 0.44a 0.44a 5.56a 4.00a 1.78a 1.33a 0 0a 1.78a 0a 1.56a 0a 1.11a 0 1.33a 2.44a 1.78a

S.E.

Probability

0.45 0.24 – 0.24 – – 0.11 0.25 – 0.29 0.24 0.68 0.44 0.48 0.44 – 0.16 0.32 0.11 0.54 0.22 0.33 – 0.40 0.54 0.51

n.s. n.s.

– n.s.

– – n.s. n.s.

– n.s. n.s.

D: P < 0.01 D: P < 0.01 n.s.

D: P < 0.05 – n.s. n.s. n.s. n.s.

C: P < 0.05 n.s.

– n.s.

C: P < 0.05 n.s.

Two-way ANOVA was used to compare values within a species. Means with the same letter do not differ significantly. S.E. = standard error, P = probability, D = substrate depth regime, C = Sedum cover regime, D*C = interaction between the substrate depth regime and the Sedum cover regime, n.s. = not significant.

each other in succession, although the latter two species had a low percentage of flowering. Interestingly, late flowering species (e.g.

Allium spp., I. pallasii, S. tricolor and T. hageri ‘Splendens’) tended to flower for shorter periods. A. cernuum, A. ostrowskianum, C. sieberi ‘Tricolor’, C. tommasinianus, I. danfordiae, S. tricolor and T. saxatilis

Table 5 Flowering periods of individual species.

March Allium flavum Allium karataviense ‘Ivory queen’ Allium ostrowskianum Allium unifolium Crocus sieberi ‘Tricolor’ Crocus tommasinianus Crocus vernus ‘Vanguard’ Iris bucharica Iris danfordiae Iris reticulate Ixioliron pallasii Muscari azureum Narcissus cyclamineus ‘February gold’ Puschkinia libanotica Scilla siberica Sparaxis tricolor Tulipa bakeri ‘Lilac wonder’ Tulipa clusiana var. chrysantha Tulipa hageri ‘Splendens’ Tulipa humilis Tulipa kolpakowskiana Tulipa linifolia Tulipa saxatilis Tulipa tarda Tulipa turkestanica Tulipa urumiensis

April

May

June

No No

No

did not flower at all, therefore, these plants are not recommended for use on extensive green roofs in areas with a similar climate to Sheffield, UK. Vegetative reproduction In the mean proximal shoot number for individual species, the results showed that many species exhibited better vegetative reproduction with a deeper substrate (Table 6). The substrate depth had a significant effect on A. unifolium, N. cyclamineus ‘February gold’, S. siberica, S. tricolor and T. turkestanica. T. turkestanica appeared to be positively affected by the covering plants because this species produced significantly more shoots with Sedum. The effects of covering plants varied according to the substrate depth for some species. At the 5 cm depth, more species reproduced better with Sedum whereas the opposite effect was observed at the substrate depth of 10 cm. This effect was significantly for T. clusiana var. chrysantha. Good reproduction was observed in A. ostrowskianum, I. bucharica, N. cyclamineus ‘February gold’, T. clusiana var. chrysantha and T. urumiensis.

Discussion

No

Effect of the substrate depth

No

Gray highlighted areas indicate flowering periods.

The flowering period was defined as the start of flowering (at least 1 plant of a given species showed flowering) until the end of flowering (no plants of a given species showed flowering) in all plots of each species.

This study showed that the deeper substrates promoted greater emergence, growth, foliage and flower performance and vegetative reproduction in most geophyte species tested. Greater emergence with the deeper substrate in the second year suggested that the substrate depth might have an important effect on the survival rate over winter. It was found that some geophytes were dug out over winter. Thus, the deeper substrate is more likely to protect geophytes from digging by animals such as birds. There appears to be a common problem with birds removing plants, particularly plug plants, in their search for food on extensive green roofs (Emilsson

Author's personal copy 516

A. Nagase, N. Dunnett / Urban Forestry & Urban Greening 12 (2013) 509–521

Table 6 Mean number of proximal shoots produced by individual species (n = 9).

Substrate depth

5 cm

10 cm

S.E.

Probability

n.s. n.s n.s.

D: P < 0.05 n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s.

D: P < 0.01 n.s.

D: P < 0.01 D: P < 0.01 n.s.

D*C P < 0.01 n.s. n.s. n.s. n.s. n.s. n.s.

D: P < 0.05 C: P < 0.01 n.s.

Sedum cover

Without Sedum cover

Sedum cover

Without Sedum cover

Allium flavum Allium karataviense ‘Ivory queen’ Allium ostrowskianum Allium unifolium Crocus sieberi ‘Tricolor’ Crocus tommasinianus Crocus vernus ‘Vanguard’ Iris bucharica Iris danfordiae Iris reticulate Ixioliron pallasii Muscari azureum Narcissus cyclamineus ‘February gold’ Puschkinia libanotica Scilla siberica Sparaxis tricolor Tulipa bakeri ‘Lilac wonder’ Tulipa clusiana var. chrysantha

2.22a 0a 1.22a 0.89a 0.33a 0a 0.89a 5.00a 0.44a 0.56a 0.78a 1.00a 1.44b 0.78a 0.44a 0b 0.78a 2.56a

1.22a 0.22a 1.67a 0.67a 0.11a 0.11a 0.22a 4.33a 0a 0.22a 0.67a 1.33 1.33b 0.67a 0a 0b 0.33a 2.00a

0.67a 0.11a 2.89a 2.67b 0.33a 0a 0.67a 4.78a 0a 0.89a 0.56a 1.22 3.67a 1.00a 0.78a 0.89a 0.44a 2.11a

1.78a 0.33a 3.00a 2.56b 0.67a 0a 0.22a 4.89a 0a 0.22a 0.44a 1.78 3.67a 1.56a 0.89a 1.44a 0.33a 2.78a

0.67 0.15 0.77 0.40 0.33 0.06 0.39 0.46 0.15 0.24 0.19 0.23 0.32 0.34 0.28 0.18 0.31 0.21

Tulipa hageri ‘Splendens’ Tulipa humilis Tulipa kolpakowskiana Tulipa linifolia Tulipa saxatilis Tulipa tarda Tulipa turkestanica

0.11a 2.00a 0.56a 0.67a 0.67a 1.56a 2.11a

0.11a 2.33a 0.11a 1.00a 0.67a 1.33a 0.44b

0.67a 2.44a 1.44a 1.11a 0.67a 1.00a 3.00a

0a 1.89a 0.67a 1.89a 1.56a 1.00a 1.67ab

0.23 0.49 0.40 0.34 0.33 0.36 0.42

Tulipa urumiensis

2.11a

1.78a

2.33a

2.44a

0.41

Two-way ANOVA was used to compare values within a species. Means with the same letter do not differ significantly according to Tukey test. S.E. = standard error, P = probability, D = substrate depth regime, C = Sedum cover regime, D*C = interaction between the substrate depth regime and the Sedum cover regime, n.s. = not significant.

and Rolf, 2005). Although the soil temperature was not measured in this study, it was estimated that there were also fewer temperature fluctuations in the deeper substrate. Boivin et al. (2001) showed that the minimum daily temperatures were significantly lower at the substrate depth of 5 cm plots (−0.4 ◦ C) than at the substrate depths of 10 cm (0.9 ◦ C) or 15 cm (1.6 ◦ C) on extensive green roof in Quebec city Canada during October and November 1995. In another study, it was found that foliage formation by tulips, hyacinths, narcissi and irises was affected detrimentally when the substrate depth dropped below −1 ◦ C (Van der Valk, 1971). This result suggested that the low temperatures in shallow substrates might affect the survival rate and growth of geophytes. Previous studies of plant selection for green roofs showed that plants performed better with deeper substrates, although perennials such as forbs, grasses and Sedum spp. were used rather than geophytes (Dunnett, 2004; Dunnett and Nolan, 2004; Durhman et al., 2007).

In general, deeper substrates provided greater moisture retention and root protection from temperature fluctuations, while they also gave more vertical space for plant root growth before reaching the root barrier (Durhman et al., 2007). Moisture retention seems to be particularly important for plant growth, and Dunnett (2004) emphasized that the main constraint for perennial plant growth on extensive roofs was water availability rather than substrate depth alone. However, it is important to note that there are differences between geophyte species and other perennial species. Geophyte plants use their stored biomass and water for early shoot development; therefore, air humidity or contact with water has little effect during the early stages of growth, whereas they are more important during later growth and flowering (Boeken, 1991).

In this study, clonal growth appeared to be more important than seed reproduction. The seeds of Allium spp. germinated in the second year, although most seedlings disappeared over time. Clonal growth is commonly stimulated in geophytes under conditions in which there is a high assimilation rate (Van der Valk and Timmer,

1974). Plant growth is encouraged with deep substrate, and the assimilation rate is higher than that with shallow substrate, which leads to greater reproduction. However, some studies have shown that environmental stress, such as a shallow position in the substrate (Barkham, 1980a), or high density (Barkham, 1980b) can lead to a higher rate of vegetative reproduction in geophytes (Rees, 1972; Grime, 1977). Long-term research is necessary to determine how the rate of vegetative reproduction in geophytes changes over time on extensive green roofs.

Effect of covering plants Results showed that the S. album cover had no significant effects on emergence, growth, foliage and flower performance and reproduction in most species. This suggests that geophyte species did not compete much with Sedum. Sedum spp. require little water because they possess Crassulacean Acid Metabolism (CAM), where stomata open only at night to minimize the amount of water lost when carbon dioxide is converted to sugars during photosynthesis (Van Woert et al., 2005). They have a creeping habitat and fibrous roots, therefore, may not compete with geophytes for light and space. However, the effect of covering plants differed among species. Some species such as A. flavum, A. ostrowskianum, I. reticulata and T. turkestanica showed significantly better emergence with Sedum, whereas the emergence of T. linifolia was restricted with Sedum (Table 2). T. kolpakowskiana and T. turkestanica grew significantly better with Sedum, whereas S. tricolor grew better without Sedum (Table 3, Appendix 2). It is possible that the species that showed a negative response to Sedum spp., may be more sensitive to competition with other species.

Total emergence was higher with the Sedum cover at both substrate depths, although statistical analysis showed that the covering plants did not significantly affect emergence. In particular, the performance of geophytes was better with Sedum in a shallow

Author's personal copy A. Nagase, N. Dunnett / Urban Forestry & Urban Greening 12 (2013) 509–521

517

Table 7 Summary of the performance of individual species.

Allium flavum Allium karataviense ‘Ivory queen’ Allium ostrowskianum Allium unifolium Crocus sieberi ‘Tricolor’ Crocus tommasinianus Crocus vernus ‘Vanguard’ Iris bucharica Iris danfordiae Iris reticulata Ixioliron pallasii Muscari azureum Narcissus cyclamineus ‘February gold’ Puschkinia libanotica Scilla siberica Sparaxis tricolor Tulipa bakeri ‘Lilac wonder’ Tulipa clusiana var. chrysantha Tulipa hageri ‘Splendens’ Tulipa humilis Tulipa kolpakowskiana Tulipa linifolia Tulipa saxatilis Tulipa tarda Tulipa turkestanica Tulipa urumiensis

Emergence

Length of the period of above ground growth

Length of the flowering period

Reproduction

Medium Low High Medium Low Low Low High Low Low Medium High High Medium Medium Medium Low High Low High Medium Medium Medium Medium Medium High

Long Short Medium Long Short Short Short Long Short Medium Long Long Long Medium Short Short Short Long Short Long Short Medium Medium Medium Long Long

Short Short Short Short No No Short Long No Short Short Long Medium Short Short No Short Medium Short Medium Short Short No Medium Long Medium

Medium Low High Medium Low Low Low High Low Low Low Medium High Medium Low Low Low High Medium Medium Low Medium Low Medium Medium High

Potential for use on extensive green roofs

High

High High (10 cm is recommended)

High High

High High High (10 cm is recommended)

Emergence: mean percentage of emergent plants per plot, i.e., 0 ≤ low ≤ 30, 30 < medium ≤ 70, 70 < high. Length of the period of above ground growth (weeks): 0 ≤ short ≤ 3, 3 < medium ≤ 6, 6 < long. Length of the flowering periods (weeks): No = no flowering, 0 ≤ short ≤ 1, 1 < medium ≤ 2, 2 < long. Reproduction: mean number of proximal shoots, i.e.: 0 ≤ short ≤ 1, 1 < medium ≤ 2, 2 < high.

substrate, and this was probably because Sedum prevents moisture evaporation and digging by animals and provides some support to the geophytes. This is in agreement with the results of previous studies. Two previous studies showed that the substrate moisture levels in vegetated treatments with Sedum were typically higher than in unvegetated treatments or other planting treatments because its growth form impeded evaporation from the soil surface (Van Woert et al., 2005; Wolf and Lundholm, 2008). Butler and Orians (2011) showed that Sedum spp. enhanced the performance of neighbouring plants when there was a summer water deficit because they could reduce the plant size and make the plants less susceptible to subsequent drought. They also showed that cooling the soil could decrease the abiotic stress and that Sedum spp. may reduce water loss from the substrate.

In this study, half of plots were left without covering plants to test whether Sedum spp. affect the performance of geophytes on extensive green roofs. However, it does not mean that only geophytes are used and they would be barren most of the time except a short period with brief bust of geophytes in spring. The common expectation of roof greening is to have evergreen in the cold or dry season so as to maximize the environmental, ecological and landscape-aesthetic contributions of the vegetation. This study showed that geophytes were good candidates to combine with herbaceous plants for garden type of shallow substrate green roofs (e.g. semi-extensive green roofs).

Performance of individual species Table 7 shows the emergence, plant growth rate, length of the above-ground growth period and flowering periods and reproduction rate for the test geophytes. Mean period of above-ground growth for individual species is shown in Appendix 3. Geophytes grown on extensive green roofs should have high emergence and survival rates, adequate foliage for healthy growth, attractive flower growth, and ideally good vegetative reproduction. In this

study, I. bucharica, M. azureum, N. cyclamineus ‘February gold’, T. clusiana var. chrysantha, T. humilis, T. tarda, T. turkestanica, and T. urumiensis showed good potential for use on extensive green roofs.

In particular, I. bucharicsa, M. azureum, T. clusiana var. chrysantha, T. humilis, T. tarda and T. turkestanica showed good performance at the substrate depth of 5 cm, although they showed better growth, foliage and flowering periods with the substrate depth of 10 cm. These species are probably drought tolerant and can withstand the high temperature fluctuations that occur in shallow substrates. They could be very useful plants for extensive green roofs because only a limited number of species can survive, grow and flower at the substrate depth of 5 cm without irrigation.

However, N. cyclamineus ‘February gold’ and T. urumiensis may require a substrate depth of 10 cm. The geophyte species, namely A. karataviense ‘Ivory queen’, C. sieberi ‘Tricolor’, C. tommasinianus, C. vernus ‘Vanguard’, I. danfordiae, I. reticulata, S. siberica, S. tricolor, Tulipa bakeri ‘Lilac wonder’, T. hageri ‘Splendens’ and T. saxatil that exhibited low emergence, poor growth and no flowering are not recommended for use on extensive green roofs. Five species, i.e., C. sieberi ‘Tricolor’, C. tommasinianus, I. danfordiae, S. tricolor and T. saxatilis never flowered. There are several possible reasons for this failure. The geophytes might not have emerged and grown because of the low availability of soil water or nutrients. Moreover, they did not survive or persist their summer resting stage because the conditions that trigger development were not met.

In this study, planting was delayed until January because of the late green roof construction. Thus, the planting season may have affected the growth of some geophytes because the appropriate planting timing was October and November. Interestingly, the late flowering species were not very successful and had short flowering periods. This may have been due to a shortage of water before they developed because the rainfall was low during May 2005 and June 2006 (Fig. 3). Further study is required to confirm the reason for this failure.

Author's personal copy 518

A. Nagase, N. Dunnett / Urban Forestry & Urban Greening 12 (2013) 509–521

Application of geophytes on extensive green roofs Geophytes can be useful plants for adding aesthetic value and enhancing number of plant species to not only new green roofs but existing green roofs without any major changes to the planting design. Sedum green roofs are the most commonly used extensive green roofs, but they have less seasonal changes and dynamic changes compared with other vegetation types because of their limited flowering period and low structural diversity (Dunnett et al., 2008). Recently, the aesthetic value and higher species richness of green roofs have attracted considerable attentions. Hence, there is a demand for adding these improvements to existing Sedum green roofs. The seasonal conditions of the vegetation, soil coverage and colour affected preferences in a previous study of the aesthetic improvement of extensive green roofs (Dagenais et al., 2010). Geophytes may be useful for fulfilling these requirements. Geophytes may also be combined with biodiverse green roofs to add aesthetic value. Hands and Brown (2002) studied the visual preferences for the ecological rehabilitation of decommissioned industrial land in Canada. The study sites were commonly perceived as messy; however, the addition of colour in the form of bold flowering forbs significantly improved their visual preference. Although this study was conducted on the ground, this can also apply to biodiverse green roofs. It is important to reconcile aesthetics with ecology when considering biodiverse green roofs, rather than relying on habitat creation and restoration ecology theory, such as the exclusive use of naive species and the use of characteristics local plant communities (Dunnett, 2006). However, few scientific studies have investigated whether geophytes can provide nectar sources to wildlife and determined the species that can provide better nectar sources, although some books have shown that geophytes can act as nectar sources in early spring (Beresford-Kroeger, 2004; Waring, 2011). Unfortunately, the biodiversity value of geophytes was not measured in the present study. Therefore, further research is required to confirm the biodiversity value of geophytes.

In this study, all of them are non-native species in the UK. Sometimes, people think that native species are preferable for green roofs, especially for biodiverse roofs. However, many native geophytes in the UK (e.g. Hyacinthoides non-scripta, Allium ursinum) are found in relatively moist soil and shady places such as woodlands and they may be not able to tolerate for drought and wind on green roofs. Scilla verna is one of native dwarf geophyte species which may be able to survive on extensive green roofs. They are naturally found on sea coast. However, a few nurseries produce S. verna and they tend to be difficult to get this species. Therefore, it is also important to use non-native plant species to achieve green roof benefits. For example, annual meadows can be combined with geophytes to extend the flowering period for aesthetic reasons and to provide nectar sources. Nagase and Dunnett (2013) showed

5 cm Allium flavum Allium karataviense ‘Ivory queen’ Allium ostrowskianum Allium unifolium Crocus sieberi ‘Tricolor’ Crocus tommasinianus Crocus vernus ‘Vanguard’ Iris bucharica Iris danfordiae Iris reticulate

72.22a 5.56a 94.44a 72.22a 22.22a 0 16.67a 94.44a 5.56a 83.33a

that it was possible to create an annual plant meadow (mixture of native and non-native species) without irrigation using a substrate depth of 7 cm on an extensive green roof in Sheffield, UK. The geophytes start flowering early in spring, while annual plants take over them from late spring to autumn. However, geophytes may compete more with annual plants than Sedum spp. Further research is necessary to study how geophytes might perform with annual plant meadows on extensive green roofs.

Conclusion This study showed that some dwarf geophytes, such as I. bucharica, M. azureum, T. clusiana var. chrysantha, T. humilis T. tarda and T. turkestanica could perform well at substrate depth of 5 cm on an extensive green roof without irrigation. Overall, the performance of geophytes was better at the substrate depth of 10 cm depth than of 5 cm. However, it is possible to achieve sufficient growth and flowering with carefully chosen geophytes when using the substrate depth of 5 cm. For many existing buildings, thin substrate is ideal not to exceed the load-bearing capacity of the building roof slab. Most geophytes used in this study disappeared after May; therefore, it is important to combine them with other plant species such as covering plants of Sedum spp. Geophytes may be useful for adding aesthetic value and enhancing biodiversity for extensive green roofs including existing Sedum green roofs. In the present experiment, the green roof environment parameters were not measured, however, collection of data regarding continuous moisture and temperature data using a data logger in the substrate would have been helpful for analysing the determinants of plant growth. Thus, the present experiment needs to be continued to understand how the performance of geophytes might change over time. In particular, the loss of nutrients from the substrate over time may affect plant performance. A future study comparing the performance of geophytes with and without supplemental watering is recommended. It is also recommended that geophytes are studied in a climate-controlled greenhouse to identify the effect of important environmental factors such as temperature and watering.

Acknowledgements We express our appreciation to Almasc for providing the experimental materials, to career-support program for woman scientists in Chiba University for founding to proof our English, to Dr. Noel Kingsbury in University of Sheffield for his valuable advice, to Dr. Min-Sung Choi in University of Sheffield for helping to set up the experiment.

Appendix 1. Mean percentage of emergence of individual species per plot in response to the substrate depth during the first year (percentage, n = 3).

10 cm

S.E.

Probability

33.33b 5.56a 83.33a 27.78a 16.67a 0 16.67a 100a 5.56a 44.44b

11.15 5.56 7.50 10.86 9.58 – 9.04 3.93 5.56 10.65

D: P < 0.05 n.s. n.s.

D: P < 0.05 n.s. n.s. n.s. n.s.

D: P < 0.05

Author's personal copy A. Nagase, N. Dunnett / Urban Forestry & Urban Greening 12 (2013) 509–521

5 cm Ixioliron pallasii Muscari azureum Narcissus cyclamineus ‘February gold’ Puschkinia libanotica Scilla siberica Sparaxis tricolor Tulipa bakeri ‘Lilac wonder’ Tulipa clusiana var. chrysantha Tulipa hageri ‘Splendens’ Tulipa humilis Tulipa kolpakowskiana Tulipa linifolia Tulipa saxatilis Tulipa tarda Tulipa turkestanica Tulipa urumiensis

10 cm

61.11a 94.44a 100.00 55.56a 16.67a 66.67a 44.44a 94.44a 27.78a 72.22a 27.78a 55.56a 50.00a 66.67a 94.44a 100

72.22a 100.00a 100.00 22.22b 11.11a 66.67a 38.89a 94.44a 16.67a 66.67a 5.556a 77.78a 38.89a 66.67a 100a 100

519

S.E.

Probability

11.35 3.93 – 11.11 8.36 11.43 11.94 5.56 9.99 11.15 8.63 11.11 11.98 11.43 3.93 –

n.s. n.s.

– D: P < 0.05 n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s.

–

T-Test was used to compare values within a species. Means with the same letter are not significantly different. S.E. = standard error, P = probability, D = substrate depth regime, n.s. = not significant.

Appendix 2. Maximum leaf number and flower height of individual species per plot in response to the substrate depth and Sedum cover (n = 3).

Plant name

Maximum growth Date

5 cm

Sedum cover Allium flavum Allium karataviense ‘Ivory queen’ Allium ostrowskianum Allium unifolium Crocus sieberi ‘Tricolor’ Crocus tommasinianus Crocus vernus ‘Vanguard’ Iris bucharica Iris danfordiae Iris reticulata Ixioliron pallasii Muscari azureum Narcissus cyclamineus ‘February gold’

Puschkinia libanotica Scilla siberica Sparaxis tricolor

Tulipa bakeri ‘Lilac wonder’ Tulipa clusiana var. chrysantha Tulipa hageri ‘Splendens’ Tulipa humilis Tulipa kolpakowskiana

Leaf number Flower height (cm) Leaf number Flower height (cm) Leaf number Flower height (cm) Leaf number Flower height (cm) Leaf number Flower height (cm) Leaf number Flower height (cm) Leaf number Flower height (cm) Leaf number Flower height (cm) Leaf number Flower height (cm) Leaf number Flower height (cm) Leaf number Flower height (cm) Leaf number Flower height (cm) Leaf number Flower height (cm)

10 cm

Without Sedum cover

S.E.

Probability

1.57 3.75 0.14 1.12 1.02 – 2.24 2.74 2.32 – 0.06

n.s. n.s. n.s. n.s. n.s.

– D: P < 0.01 n.s. n.s.

– n.s.

– n.s. n.s. n.s.

D: P < 0.01 n.s.

– n.s. n.s.

D*C P < 0.05 n.s.

D: P < 0.05 D: P < 0.01 D: P < 0.01 D: P < 0.01

Sedum cover

Without Sedum cover

4.89a 3.59a 0a 0a 0.89a 0 10.44a 6.23a 2.56a 0 0a 0 1.00a 0a 32.22a 25.29ab 0a 0 2.00a 0a 2.89a 3.58a 3.56ab 7.27ab 8.11a 21.51a

2.89a 8.04a 0.22a 3.34a 1.67a 0 9.67a 1.49a 3.56a 0 0a 0 0.89a 1.44a 30.11a 26.64a 0a 0 0.67a 0a 1.67a 5.75a 4.78a 9.43a 8.56a 21.39a

0.63 1.41 1.24 2.89

2.00a 4.52a 2.00a 2.33a 5.44a

0.63 1.27 1.08 0.93 0.68

0 0.33a 0a 8.78a 17.19a 0a 0a 6.11a 7.40a 0.44ab 0a

– 0.60 0.74 0.96

0.89a 8.13a 0a 0a 3.56a 0 2.78a 0a 1.56a 0 0a 0 0.89a 0a 28.22a 17.81b 0a 0 0.44a 2.03a 0.33a 0.00a 2.11b 3.09b 0.78b 0b

4.22a 4.21a 0.3a 0a 2.67a 0 3.28a 4.90a 0a 0 0.11a 0 0.11a 0a 22.44a 19.05b 0.4a 0 0a 0a 3.67a 3.33a 3.33ab 5.37ab 2.11b 2.65b

Leaf number Flower height (cm) Leaf number Flower height (cm) Leaf number

May 22 July 10 June 6 May 22 April 27 – April 27 June 6 March 29 – March 17 – April 27 March 17 June 6 April 17 March 29 – May 4 April 13 June 6 June 6 May 22 April 27 4th May 27th April April 27 April 13 May 22 April 13 May 22

1.22a 2.27a 1.44a 0.89a 0b

0.78a 1.13a 0a 0a 0b

1.63a 0.78a 3.11a 1.61a 1.44b

Flower height (cm) Leaf number Flower height (cm) Leaf number Flower height (cm) Leaf number Flower height (cm) Leaf number Flower height (cm) Leaf number Flower height (cm)

– April 27 April 27 April 27 May 4 April 13 May 4 April 13 April 13 April 13 April 13

0 1.89a 0a 7.33a 14.04a 0.33a 0a 4.67a 3.66a 1.56ab 0a

0 0.78a 1.48a 6.78a 13.15a 0.33a 1.03a 5.78a 3.54a 0b 0a

0 0.22a 0a 7.67a 10.91a 0.78a 0a 6.22a 5.96a 3.44a 3.42a

0.81 0.72 3.33 1.96 0.22 – 0.57 0.68 1.03

0.33 0.52 1.47 1.75 0.89 0.87

n.s. n.s. n.s. n.s.

D: P < 0.01 C: P < 0.01 D*C: P < 0.01 – n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s.

C: P < 0.05 n.s.

Author's personal copy 520

A. Nagase, N. Dunnett / Urban Forestry & Urban Greening 12 (2013) 509–521

Plant name

Tulipa linifolia Tulipa saxatilis Tulipa tarda Tulipa turkestanica

Tulipa urumiensis

Maximum growth Date

Leaf number Flower height (cm) Leaf number Flower height (cm) Leaf number Flower height (cm) Leaf number Flower height (cm)

April 27 May 4 March 29 – April 27 April 27 April 27 April 13

Leaf number Flower height (cm)

May 22 May 4

5 cm

10 cm

S.E.

Probability

n.s. n.s. n.s.

– n.s. n.s.

C P < 0.01 D: P < 0.01 C: P < 0.05 n.s. n.s.

Sedum cover

Without Sedum cover

Sedum cover

Without Sedum cover

2.33a 2.23a 2.22a 0 4.33a 5.34a 5.00ab 13.03ab

2.56a 2.77a 2.00a 0 4.44a 7.07a 1.11b 1.89b

4.00a 3.67a 0.56a 0 4.00a 4.98a 5.44a 18.39a

6.56a 5.71a 0.56a 0 3.56a 4.66a 3.00ab 14.77a

1.47 1.73 0.81 – 1.41 1.60 1.11 3.00

5.11a 2.71a

1.89a 2.87a

6.22a 5.85a

6.11a 3.54a

1.34 1.50

The date indicates the day when the maximum flower height occurred in individual species. Two-way ANOVA is used to compare values within a species. Means with the same letter do not differ significantly from each other. S.E. = standard error, P = probability, D = substrate depth regime, C = Sedum cover regime, D*C = interaction between the substrate depth regime and the Sedum cover regime, n.s. = not significant.

Appendix 3. Mean period of above-ground growth for individual species (week).

5 cm Sedum cover Allium flavum

13.56a

Allium karataviense ‘Ivory queen’ Allium ostrowskianum Allium unifolium Crocus sieberi ‘Tricolor’ Crocus tommasinianus Crocus vernus ‘Vanguard’ Iris bucharica Iris danfordiae Iris reticulate Ixioliron pallasii Muscari azureum Narcissus cyclamineus ‘February gold’ Puschkinia libanotica Scilla siberica Sparaxis tricolor

10 cm Without Sedum cover

Sedum cover

S.E.

P

2.464

D*C P < 0.05 n.s. n.s.

D: P < 0.05 n.s. n.s. n.s. n.s. n.s. n.s. n.s.

D*C: P < 0.05 D: P < 0.05 n.s.

D: P < 0.05 D: P < 0.05 C: P < 0.05 D*C: P < 0.05. n.s.

D: P < 0.05 n.s. n.s. n.s. n.s. n.s. n.s.

D: P < 0.05 C: P < 0.05 n.s.

Without Sedum cover

4.00b

3.56b

5.33ab

0a 4.00a 4.44a 1.78a 0a 1.56a 16.00a 0a 2.22a 2.00a 13.56a 5.33b 4.89a 3.11ab 0c

1.11a 5.33a 4.44a 0.44a 0.22a 0.67a 14.67a 0.89a 0.67a 7.78a 15.78a 6.67b 2.89a 0b 0c

0.44a 7.56a 10.89ab 0.67a 0a 2.44a 16.00a 0a 5.56a 5.78a 15.78a 12.89a 4.89a 7.78a 6.22b

3.11a 7.56a 13.33a 2.00a 0a 1.56a 16.00a 0a 1.56a 4.00a 14.89a 13.57a 7.33a 4.89ab 13.11a

1.12 1.74 1.98 0.96 0.11 1.17 0.67 0.24 1.38 1.91 0.62 1.18 1.55 1.79 1.39

Tulipa bakeri ‘Lilac wonder’ Tulipa clusiana var. chrysantha Tulipa hageri ‘Splendens’ Tulipa humilis Tulipa kolpakowskiana Tulipa linifolia Tulipa saxatilis Tulipa tarda Tulipa turkestanica

0.89a 11.33a 1.33a 7.33a 2.67a 3.56a 2.89a 6.89a 8.89

1.11a 12.67a 2.00a 6.89a 0.67a 5.78a 2.22a 8.22a 1.11

1.78a 14.44a 3.11a 6.44a 5.78a 6.22a 3.78a 6.67a 14.22

0.67a 14.67a 0a 7.56a 3.56a 10.44a 6.44a 7.11a 10.89

0.97 0.98 1.36 1.40 1.65 1.92 1.58 2.23 1.76

Tulipa urumiensis

10.00

9.56

11.11

9.33

1.49

Two-way ANOVA is used to compare values within a species. Means with the same letter do not differ significantly from each other. S.E. = standard error, P = probability, D = substrate depth regime, C = Sedum cover regime, D*C = interaction between the substrate depth regime and the Sedum cover regime, n.s. = not significant.

References Alumasc exterior building products, 2006. Green Roof Substrates. www.alumasc-exteriors.co.uk (accessed 1st June 2006).

Barkham, J.P., 1980a. Population dynamics of the wild daffodil (Narcissus pseudonarcissus): clonal growth, seed reproduction, mortality, and the effects of density.

Journal of Ecology 68 (2), 607–633.

Barkham, J.P., 1980b. Population dynamics of the Wild daffodil (Narcissus pseudonarcissus): II. Changes in number of shoots and flowers, and the effect of bulb depth on growth and reproduction. Journal of Ecology 68 (2), 635–664.

Beresford-Kroeger, D., 2004. A Garden for Life: The Natural Approach to Designing, Planting and Maintaining a North Temperate Garden. University of Michigan Press, Michigan.

Blamey, P., Blamey, M., 1979. Bulbs. Fontana, London.

Boeken, B., 1991. Above-ground emergence in the desert tulip. Tulipa systola Stapf. in the Negev Desert of Israel. Functional Ecology 5, 705–712.

Boivin, M., Lamy, M., Gosselin, A., Dansereau, B., 2001. Effect of artificial substrate depth on freezing injury of six herbaceous perennials grown in a green roof system. Horticultural Technology 11 (3), 409–412.

Butler, C., Orians, C.M., 2011. Sedum cools soil and can improve neighboring plant performance during water deficit on a green roof. Ecological Engineering 37 (11), 1796–1803.

Botschantzeva, Z., 1982. Tulips: Taxonomy, Morphology, Cytology, Phytogeogr. CRC Press, Florida.

Clark, C., Adriaens, P., Talbot, F.B., 2008. Green roof valuation: a probabilistic economic analysis of environmental benefits. Environmental Science Technology 42, 2155–2161.

Dagenais, D., Gagnon, C., Pelletier, J.L., 2010. Development of design criteria to improve aesthetic appreciation of extensive green roofs. Acta Horticulturae 881, 703–708.

Dunnett, N.,2004. Options and constraints for plant use on green roofs in the UK. In: Proceeding of International Green Roof Congress, September 2004. International Green Roof Association, Nürtingen.

Author's personal copy A. Nagase, N. Dunnett / Urban Forestry & Urban Greening 12 (2013) 509–521 Dunnett, N., 2006. Green roofs for biodiversity: reconciling aesthetics with ecology.

In: Proceeding of Fourth North American Green Roofs Conference, Boston, May 2006. The Cardinal Group, Toronto.

Dunnett, N., Nolan, A., 2004. The effect of substrate depth and supplementary watering on the growth of nine herbaceous perennials in a semi-extensive green roof.

In: International Conference on Urban Horticulture, Acta Horticulture, vol. 643, pp. 305–310.

Dunnett, N., Kingsbury, N., 2008. Planting Green Roofs and Living Walls. Timber Press, Portland, OR. Dunnett, N., Nagase, A., Booth, R., Grime, P., 2008. Influence of vegetation composition on runoff in two simulated green roof experiments. Urban Ecosystems 11 (4), 385–398.

Durhman, A.K., Rowe, D.B., Rugh, C.L., 2006. Effect of watering regimen on chlorophyll fluorescence and growth of selected green roof plant taxa. HortScience 41 (7), 1623–1628.

Durhman, A.K., Rowe, D.B., Rugh, C.L., 2007. Effect of substrate depth on initial growth, coverage, and survival of 25 succulent green roof plant taxa. HortScience 42 (3), 588–595.

Elliott, J., 1995. Bulbs for the Rock Garden. Batsford, London.

Emilsson, T., Rolf, K., 2005. Comparison of establishment methods for extensive green roofs in southern Sweden. Urban Forestry and Urban Greening 3 (2), 103–111.

Ernst, W.H.O., 1979. Population biology of Allium ursinum in Northern Germany. The Journal of Ecology 67 (1), 347–362.

Garrett, F., Dusoir, R., 2004. Bulbous plants for use in designed landscapes. In: Hitchmough, J., Fieldhouse, K. (Eds.), Plant User Handbook. Blackwell Science, Oxford.

Getter, K.L., Rowe, B.D., 2006. The role of extensive green roofs in sustainable development. HortScience 41 (5), 1276–1285.

Glattstein, J., 2005. Bulbs for Garden Habitats. Timber Press, Portland, OR. Grime, J.P., 1977. Evidence for the existence for three primary strategies in plants and its relevance to ecological and evolutionary theory. American Naturalist 111 (982), 1169–1194.

Hands, D.E., Brown, R.D., 2002. Enhancing visual preference of ecological rehabilitation sites. Landscape and Urban Planning 58 (31), 57–70.

Hewson, R.T., Roberts, H.A., 1973. Some effects of weed competition on the growth of onions. Journal of Horticultural Science 48, 51–57.

Hughes, S., 1986. Growing bulbs in grassed areas. In: Horticultural Services Unit, Annual Reports 1986, Parks and Conservation Service., pp. 155–161.

Kadas, G., 2006. Rare invertebrates colonizing green roofs in London. Urban Habitats 4, 66–86.

Kingsbury, N., 1996. The New Perennial Garden. Frances Lincoln, London.

521

Köhler, M., Poll, P.H., 2010. Long-term performance of selected old Berlin greenroofs in comparison to younger extensive green roofs in Berlin. Ecological Engineering 36, 722–729.

Lazenby, A., 1962. Studies on Allium vineale L. III. Effect of depth of planting. The Journal of Ecology 50 (1), 97–109.

Mathew, B., 1997. Growing Bulbs. The Complete Practical Guide. B.T. Batsford, London.

Mathew, B., Swindells, P., 1994. The Gardener’s Guide to Bulbs. Mitchell Beazley, London.

Met office, 2007. Historic Station Data in Sheffield UK, Retrieved January 2007 from http://www.metoffice.gov.uk/climate/uk/stationdata/sheffielddata.txt Nagase, A., Dunnett, D., 2013. Establishment of an annual meadow on extensive green roofs in the UK. Landscape and Urban Planning 112, 50–62.

Oberndorfer, E., Lundholm, J., Bass, B., Coffman, R.R., Doshi, H., Dunnett, N., Gaffin, S., Köhler, M., Liu, K.K.Y., Rowe, B., 2007. Green roofs as urban ecosystems: ecological structures, functions, and services. BioScience 57 (10), 823–833.

Olly, L.M., Bates, A.J., Sadler, J.P., Mackay, R., 2011. An initial experimental assessment of the influence of substrate depth on floral assemblage for extensive green roofs.

Urban Forestry and Urban Greening 10, 311–316.

Oxford Dictionary of English, 2010. 3rd Revised. Oxford University Press, Oxford.

Phillips, R., Rix, M., 1989. Bulbs. Pan Original, London.

Raunkiaer, C., 1934. The Life-Form of Plants and Statistical Plant Geography.

Clarendon Press, Oxford.

Rees, A.R., 1972. The Growth of Bulbs. Applied Aspects of the Physiology of Ornamental Bulbous Crop Plants. Academic Press, London.

Rees, A.R., 1992. Ornamental Bulbs, Corms and Tubers. CAB International, Oxon.

Rowe, D.B., Getter, K.L., Durhman, A.K., 2012. Effect of green roof media depth on Crassulacean plant succession over seven years. Landscape and Urban Planning 104 (3/4), 310–319.

Snodgrass, E.C., Snodgrass, L.L., 2006. Green Roof Plants. Timber Press, Portland, OR. Van der Valk, G.G.M.,1971. Frost injury to flower bulb crops. In: Proceeding of International Symposium on Flower Bulbs. Centre for Publishing and Documentation, Wageningen, pp. 345–349.

Van der Valk, G.G.M., Timmer, M.J.G., 1974. Plant density in relation to tulip bulb growth. Scientia Horticulturae 2, 69–81.

Van Woert, N.D., Rowe, D.B., Andresen, J.A., Rugh, C.L., Xiao, L., 2005. Watering regime and green roof substrate design affect Sedum plant growth. HortScience 40 (3), 659–664.

Waring, A., 2011. Better Beginnings for Beekeepers. Writersprintshop, London.

Wolf, D., Lundholm, J.T., 2008. Water uptake in green roof microcosms: effects of plant species and water availability. Ecological Engineering 33, 179–186.