The Pennsylvania State University
The Graduate School
Department of Horticulture
College of Agricultural Sciences
AN EVALUATION OF AN EXTENSIVE GREEN ROOF MAINTENANCE METHOD
A Thesis in
Horticulture
by
Richard C. Hoover
2010 Richard C. Hoover
Submitted in Partial Fulfillment
of the Requirements
for the Degree of
Master of Science
May 2010
The thesis of Richard C. Hoover was reviewed and approved* by the following:
Robert Berghage
Associate Professor or Horticulture Thesis Advisor
E. Jay Holcomb Professor of Floriculture
Robert Shannon Associate Professor of Agricultural Engineering
Richard Marini
Professor of Horticulture Head of the Department of Horticulture
*Signatures are on file in the Graduate School
ii
ABSTRACT
The work presented in this thesis evaluates a green roof maintenance procedure designed
specifically to increase the number of plants in genus Sedum growing in sparsely vegetated
portions of an extensive green roof. Over their life, green roofs may suffer from poor plant
coverage because of factors such as poor initial plant establishment, unfavorable growing
conditions, extreme weather, and poor maintenance. Through evapotranspiration, plants
contribute significantly to the stormwater and cooling benefits a green roof provides, so
maintaining thick plant coverage is important. Plant death also leaves behind areas of exposed
growing media which provide weed seeds a place to germinate. Sedum species are desired over
other species because of their ability to survive the stresses of living on a green roof thus
providing the benefits of evapotranspiration over the entire growing season. The procedure
proposed herein seeks to increase the number of Sedum plants with minimal inputs of labor, plant
material, and fertilizer, thereby decreasing long term maintenance costs. Through a series of
experiments outside on an experimental green roof, and in a climate controlled greenhouse, this
thesis demonstrates that Sedum plants on an existing green roof provide a large and viable source
of new Sedum cuttings that when dispersed on exposed green roof media will propagate
themselves into new Sedum plants. On the experimental green roof, control treatments, where no
cuttings were harvested and no cuttings were spread, had significantly fewer new plants per
square meter than all treated sections, where cuttings were harvested and spread. A reel mower
showed significantly more new plants per square meter than a string trimmer, while both tools
required the same amount of time to complete treatment. The greenhouse experiments found no
relationship between the rooting of sedum cuttings and irrigation regimes. Three of the four
species tested, S. album, S. sexangulare, S. spurium (John Creech) exhibited 100% rooting after
10 days, regardless of whether they were watered daily, every other day, twice in 10 days, or
iii
watered when they were spread and not watered again. S. kamtschaticum exhibited 100% rooting
eventually, but not within the 10 day study period. The greenhouse experiments also found that
all the Sedum species on the experimental green roof, S. acre (aureum), S. album, S.
kamtschaticum, S. hispanicum, S. rupestre, S. rupestre (Angelina), S. sarmentosum, S.
sexangulare, S. spurium (White Form), S. spurium (John Creech), and S. spurium (Fuldaglut),
were capable of producing viable cuttings.
iv
TABLE OF CONTENTS
LIST OF FIGURES ...................................................................................................................... vi
LIST OF TABLES........................................................................................................................ vii
ACKNOWLEDGEMENTS ......................................................................................................... viii
CHAPTER 1: LITERATURE REVIEW .................................................................................... 1
Chapter 1-1: Types of Green Roofs..................................................................................... 1 Chapter 1-2: Green Roof Benefits ....................................................................................... 4 Chapter 1-3: Structural Components Green Roofs............................................................. 8 Chapter 1-4: Green Roof Plants........................................................................................... 13 Chapter 1-5: Green Roof Planting Methods ....................................................................... 16 Chapter 1-6: Vegetation Effects on Green Roof Performance .......................................... 20 Chapter 1-7: Green Roof Maintenance ............................................................................... 23
CHAPTER 2: THESIS GOALS AND OBJECTIVES............................................................... 27
CHAPTER 3: OUTDOOR EXPERIMENT INTRODUCTION............................................... 29
Chapter 3-1: Goals and Objectives ...................................................................................... 30 Chapter 3-2: Hypotheses ...................................................................................................... 31
CHAPTER 4: OUTDOOR EXPERIMENT MATERIALS AND METHODS ....................... 32
Chapter 4-1: Materials.......................................................................................................... 32 Chapter 4-2: Experimental Design ...................................................................................... 35 Chapter 4-3: Methods........................................................................................................... 43
CHAPTER 5: OUTDOOR EXPERIMENT RESULTS AND DISSCUSSION ...................... 51
Chapter 5-1: Explanation of the Proposed Maintenance Method...................................... 51 Chapter 5-2: Efficacy of Treatment..................................................................................... 52 Chapter 5-3: Cutting and New Plant Species Compositions ............................................. 54 Chapter 5-4: Comparison of Harvesting Tools – Efficiency ............................................. 56 Chapter 5-5: Comparison of Harvesting Methods - Cutting Quality ................................ 58 Chapter 5-6: Other Observations from the Roof ................................................................ 60
CHAPTER 6 – GREENHOUSE EXPERIMENT INTRODUCTION...................................... 62
Chapter 6-1: Goals and Objectives ...................................................................................... 64 Chapter 6-2: Hypotheses ...................................................................................................... 65
CHAPTER 7 – GREENHOUSE MATERIALS AND METHODS ......................................... 66
Chapter 7-1: Materials.......................................................................................................... 66
v
Chapter 7-2: Methods – Growing a Cutting Stock ............................................................. 67 Chapter 7-3: Methods - Watering Regime and Greenhouse Cutting Viability Tests....... 68 Chapter 7-4: Methods – Outdoor Cutting Viability ........................................................... 70
CHAPTER 8: GREENHOUSE EXPERIMENT RESULTS AND DISCUSSION ................. 72
Chapter 8-1: Watering Regime and Greenhouse Cuttings Viability ................................. 72 Chapter 8-2: Outdoor Green Roof Cutting Viability Results ............................................ 75
CHAPTER 9: THESIS CONCLUSIONS ................................................................................... 77
WORKS CITED ........................................................................................................................... 87
Appendix A A Collection of Vegetation Maps of the Experimental Green Roof................... 90
Appendix B: A Collection of Vegetation Maps of the Experimental Green Roof from
2007 ....................................................................................................................................... 98
Appendix C: Analysis of Rooflite ® Green Roof Growing Media .......................................... 109
Appendix D: Tables and Graphs................................................................................................. 111
vi
LIST OF FIGURES
Figure 1-1: Cut away of a green roof structure........................................................................... 8
Figure 3-1: Photograph of the experimental green roof on 5/25/2009...................................... 29
Figure 4-1: Photograph of the short handled hand sheers used in this experiment. ................. 33
Figure 4-2: A diagram of the experimental layout. .................................................................... 36
Figure 4-3: A map showing the location of the Root Cellar building on the University
Park campus of The Pennsylvania State University (The Gould Foundation). ................ 37
Figure 4-4: Graphical representation of the rainfall between 5/1/2009 and 6/7/2009. Cuttings were harvested on 5/28/2009 and 6/1/2009. ........................................................ 49
Figure 4-5: Daily High and Low Temperatures from 5/1/2009 to 6/30/2009 .......................... 49
Figure 5-1: The mean number of new Sedum plants per square meter for each treatment. The error bars show standard error. Control sections showed significantly lower
new plant density than any of the treatments (p = 0.00). The string trimmer showed
significantly lower new plant density than the reel mower (p = 0.036). The new plant densities of hand shear and reel mower treatments were not significantly
different. ................................................................................................................................ 53
Figure 5-2: The estimated time required to cut the entire roof (397 m2). ................................. 57
Figure 8-1: Rooting data for S. album collected in the greenhouse. Thirty cuttings were sampled.................................................................................................................................. 72
Table 8-2: Rooting data for S. sexangulare collected in the greenhouse. Thirty cuttings
were sampled ........................................................................................................................ 73
Table 8-3: Rooting data for S. kamtschaticum collected in the greenhouse. Twelve
cuttings were sampled. ......................................................................................................... 73
Table 8-4: Rooting data for S. spurium (John Creech) collected in the greenhouse. Thirty cuttings were sampled. ......................................................................................................... 74
vii
LIST OF TABLES
Table 1-1 – Comparison of Extensive, Semi-Intensive, and Intensive Green Roofs. The
table is an excerpt from Green Roofs in Sustainable Landscape Design (Cantor
2008)...................................................................................................................................... 3
Table 5-1: Data table showing the mean number of new Sedum plants per square meter
for each treatment. Control sections showed significantly lower new plant density
than any of the treatments (p = 0.00). The string trimmer showed significantly lower
new plant density than the reel mower (p = 0.036). The new plant density of hand shears and reel mower treatments was not significantly different..................................... 53
Table 5-2: Comparison of the species composition of cuttings new plants from
replications 1 through 5........................................................................................................ 54
Table 5-4: Comparison of damaged cuttings taken by each harvesting tool. The
percentage of mangled cuttings is significantly different for all three treatments. .......... 59
Table 5-5: Comparison of the flowering cuttings taken by each harvesting tool. The reel mower and string trimmer produced significantly more flowering cuttings than hand
shears. .................................................................................................................................... 59
Table 8-1: The percentage and ratio of cuttings from the experimental green roof rooting
7 days after being spread on green roof growing media. ................................................... 75
viii
ACKNOWLEDGEMENTS
Thank you to the Information Technology Services department at Penn State for funding
my graduate studies. Special thanks go to Kathy Mayberry for alerting me of the graduate
assistantships at the ITS Help Desk. Thank you also to all the Penn State students, faculty, staff,
students, and retirees who had enough computer problems to keep me funded for three years.
Special thanks to Ed Snodgrass, John Shepley, and Emory Knoll Farms, Inc. for
employing me as a summer intern and for supplying plants for this research project. Thanks to
the EKF family for creating the hardest working, friendliest, and most educational work place
possible. I’ll always keep the tractors rolling slow and low because of Ed.
Dr. Rob Berghage, thank you for the freedom to work on this project, and encouraging
me to stay the course on the M.S. degree. Thank you Laurel Valley Farms for supplying growing
media used in the greenhouse experiments.
Thank you to Michele Marini for her statistical advice and willingness to meet at the last
minute.
1
CHAPTER 1: LITERATURE REVIEW
Chapter 1-1: Types of Green Roofs
A green roof can be defined simply as any roof that bears vegetation (Cantor, 2008).
Vegetation bearing roofs are seen throughout history from sod roofs used by settlers in the
American Great Plains in the mid to late 1800’s, in rooftop gardens that appeared in Italian cities
in the 1300s and 1500s, and even in 500 to 90 BC in the Hanging Gardens of Babylon (Vidmar,
et al. 2007). While they have existed in some form for much of human history, a modern green
roof carries with it a twist.
Modern green roofs are compound technologies with numerous constructed layers
underlying any visible vegetation. Depending on region and the goals of a specific project, green
roofs may be referred to as ecoroofs, living roofs, roof gardens, or brown roofs. Their appearance
may vary as much as their names imply. Some green roofs are beautiful and elaborate landscape
designs like those found on the ground; others may only support low-growing, ground-covering
plant species or moss; and some may be brown for significant portions of the year. Despite all
the potential variation, there are general definitions that distinguish different kinds of green roofs.
The German document Forschungsgesellschaft Landschaftsentwicklung Landschaftsbau,
(FLL) (Guidelines for the Planting, Execution, and Upkeep of Green-Roof Sites) defines three
categories of rooftop greening: Intensive greening; simple intensive greening; and extensive
greening (FLL, 2002). The FLL distinguishes the three categories of rooftop greening by plant
selection and maintenance requirements. Intensive and simple intensive greening may have
shrubs, grasses, and even trees, though simple intensive greening limits plant selection and
2
maintenance requirements comparatively. Extensive greening involves plants that create a
“virtual nature,” and require little external input for maintenance or propagation (FLL, 2002).
The three categories of green roofs commonly used in American literature are intensive,
semi-intensive, and extensive green roofs. Interestingly, American references distinguish the
three categories by the depth of growing media on the roof. Intensive green roofs are
characterized by having 30 cm (12 in) or more of growing media. Often the media has higher
organic matter content compared to extensive green roof growing media. Extensive green roofs
are characterized by shallower media depth, generally less than 15.2 cm (6 in) of largely
inorganic growing media (Cantor 2008, Snodgrass 2006). Semi-intensive green roofs fall in
between intensive and extensive green roofs in terms of media depth, plant selection, and
maintenance requirements. Table 1-1 summarizes the differences between intensive, semi-
intensive (simple intensive), and extensive green roofs (Berghage, Personal Communication).
3
Table 1-1 – Comparison of Extensive, Semi-Intensive, and Intensive Green Roofs. The table is
an excerpt from Green Roofs in Sustainable Landscape Design (Cantor 2008).
Characteristic Extensive Semi-Intensive Intensive
Depth of Growing Media
15.2 cm (6 in) or less
+/- 25% of 15.2 cm (6in)
15.2 cm (6 in) or more
Accessibility Often inaccessible May be partially
accessible Usually accessible
Fully Saturated Weight
Low 48.8 - 170.9 kg/m
2
(10 - 35 lb/ft2)
Varies 170.9 - 244.1 kg/m
2
(35 - 50 lb/ft2)
High 244.1 - 1464.7 kg/m
2 (50 - 300 lb/ft
2)
Plant Diversity Low Greater Greatest
Cost Low Varies High
Maintenance Minimal Varies Generally high
4
Chapter 1-2: Green Roof Benefits
Vegetating a rooftop replaces an impervious rooftop surface with a pervious one covered
with plants, therefore replacing some of the ecological function lost to development. Ecosystem
functions, including the evapotranspirative component of the hydrologic cycle provided by pre-
developed land, are partially restored. Many green roof benefits are derived from the ability of a
green roof to restore part or all of the evapotranspirative component of the hydrologic cycle
where development has reduced open space and vegetation (Berghage, et al. 2007). Storm water
mitigation, reductions in a building’s heating and cooling loads, extended roof life, improved
aesthetics, noise filtration, animal habitat, and air filtration are benefits provided by green roofs.
The benefits provided by rooftop greening can be valuable to a building’s owner, its occupants,
and the surrounding community. Some of the most important benefits are discussed in greater
detail in the following paragraphs.
According to Green Roofs for Healthy Cities, a green roof trade association, up to 75
percent of many cities are covered in impervious surfaces. Roofs in Toronto make up between 15
and 30 percent of the impervious surface of the city. In Portland, Oregon, roofs make up 25
square miles of the city surface (Snodgrass, 2006). When roofs are hard, impervious surfaces,
they clearly represent part of the storm water problems in developed areas. After a rain event
stormwater can cause combined sewer overflows (CSO) in municipalities with wholly or partially
combined sewage and stormwater systems. CSOs discharge untreated or partially treated sewage
into nearby waterways (Snodgrass, 2006). From 1997 to 2001, the rate of urban development
averaged 890,000 ha/year (2,400 ha/day) (Berghage, et al. 2007).
5
Storm Water Mitigation
After a rainfall event, green roofs mitigate storm water in three ways: Green roofs
increase the time of concentration, retain storm water, and provide some buffering against acid
precipitation (Jarrett et. al. 2006, Berghage et. al. 2007). Because of their storm water mitigation
benefits, green roofs can reduce the need for other water management features such as retention
basins, bioswales, and rain gardens that require additional land to construct. Highly developed
areas may not have available land to build other stormwater features, making green roofs an
attractive alternative.
Research and experience demonstrates that green roofs can retain 40-70 percent of annual
rainfall in northeastern North America, and northern Europe (Berghage, et al. 2007). A green
roof with 10 cm (4 in) of growing medium can retain 40-60 percent of the annual precipitation in
the northeastern United States, with nearly 90 percent retention of many summer storms
(Berghage et al. 2007). The same study shows that green roofs retain 100 percent of many
rainfall events of 15mm (0.6 in) or less. Water retention models developed at The Pennsylvania
State University indicate that green roofs will retain 48, 53, 78, 43, and 25 percent of rainfall on
the roof in May, June, July, October, and November, respectively in central Pennsylvania (Jarrett,
et al, 2006). The study pointed out that one drawback of retaining stormwater on a roof is the
stored water is not allowed to infiltrate the soil profile and recharge local supplies (Jarrett, et al.
2006). However, if water is being stored on a green roof in an urbanized area with mostly
impermeable surroundings, any water retained on the roof likely can not infiltrate the soil profile
anyway.
6
In addition to retaining storm water, green roofs improve some aspects of stormwater
runoff quality. Runoff from green roofs is clearly and consistently higher pH than runoff from
non-greened roofs (Berghage, et al. 2007). Acid buffering is particularly beneficial in places like
the Northeastern and Midwestern United States where rainfall pH is well below a pH of 5 and
frequently drops below pH of 4.5 (Berghage, et al. 2007.) The buffering capacity is due largely
to the acid buffering capacity of green roof growing media, but is also influenced by green roof
plants. The impact of plants and buffering capacity is discussed in Chapter 1-6.
Reduced Cooling Loads
The evapotranspirative effects of green roofs reduce the cooling loads for a building and
reduce the urban heat island effect. A study at the University of Technology, Berlin Germany,
found that daily potential evapotranspiration and real evapotranspiration from a green roof are
higher than from the natural landscape. The increased evapotranspiration can be attributed to the
urban heat island effect, lower humidity in urban areas, high global radiation, and higher wind
speeds on rooftops (Schmidt, 2006).
7
Extended Roof Life
Vegetation and substrate overlie the waterproofing membrane, thus protecting the
waterproofing membrane from solar exposure and ultraviolet degradation. The vegetation and
substrate also reduce the large temperature fluctuations common on black roofs where the
temperature of the roof can soar well above the ambient air temperature. The combination of
these effects causes green roofs to have up to three times the life expectancy of a non greened
roof (Vidmar, et al. 2007).
Aesthetics
Green roofs provide green space in place of hard, often unsightly roofs. Green space is
especially beneficial in highly urbanized areas where the rest of the landscape is often concrete,
asphalt, brick, and black rooftops. Some green roofs may not be visible to anyone, some may
only be seen from a distance, and others may be accessible to a building’s owners, tenants, or the
public. Regardless of its public visibility, a green roof can provide some of the green space that is
lost to development.
8
Chapter 1-3: Structural Components Green Roofs
Underneath the top layer of vegetation, a green roof is composed of a number of
constructed layers, each of which serves an important purpose in the overall performance of the
roof. Each layer is briefly discussed below. Some green roof systems may contain more layers
than described here, and some roofs may not contain all of the layers discussed. At a minimum, a
green roof should consist of a roof deck, a waterproofing membrane, drainage (either a layer or a
sloped roof), growing media, and vegetation.
Figure 1-1: Cut away of a green roof structure
9
Roof Deck
The roof deck provides structural support for the green roof. An extensive green roof
with 10.2 cm (4 in) or less media weighs between 48.8 – 170.9 kg/m2 (10-35 lb/ft2) when fully
saturated. Deeper and more complicated intensive roofs weigh between 244.1-464.7 kg/m2 (50-
300 lb/ft2) when fully saturated (Cantor, 2008). The green roof load must be added to snow load,
and other live and dead load requirements, so the roof deck and underlying building must be
strong and durable to support a green roof installation. Reinforced concrete, precast concrete
planks, and steel-concrete composites are most commonly used in green roof projects. Retrofit
projects may also encounter plywood or tongue in groove wooden decks (Cantor, 2008).
Waterproofing Membrane
Ultimately a green roof is a roof, and a roof must keep water out of a building.
Therefore, a waterproof membrane is essential to the success of any green roof. It is important to
install the waterproofing membrane correctly and detect any leaks before continuing construction
of the other layers. Once a green roof is complete, the waterproofing membrane is fully covered
by other layers including growing media and plants, thus making leak detection and repair
difficult (Snodgrass, 2006).
Depending on the underlying roof deck and local building codes, many different
waterproofing membranes can be used for a green roof. Common waterproofing membranes
include the following materials: thermoplastics, such as polyvinyl chloride (PVC) or thermal
polyolefin (TPO); ethylene propylene diene monomer (EPDM) rubber; liquid applied
polyurethane (PUR); asphalt; and bitumen (Cantor, 2008).
10
Insulation
The need for insulation depends on the climate and the goals of the green roof, so
insulation is not installed in all green roof projects. If decreasing heat gain or loss is a major
performance goal, then insulation may be installed either above or below the waterproofing
membrane. Common insulation materials include extruded polystyrene and polyisocyanurate.
Commonly, a layer of fiber board is installed on top of the insulation when the insulation overlies
the waterproofing membrane (Cantor, 2008).
Root Barrier
Some kind of root barrier is needed to protect the waterproofing membrane from root
penetrations. Sheets of thermoplastics such as PVC and high density polyethylene (HDPE) are
commonly used root barriers (Snodgrass, 2006). Copper foil and root-retardant chemicals can be
used, but they may be subject to code restrictions. In some cases, the root protection layer is
incorporated with other layers, such as the waterproofing membrane itself (Cantor, 2008). If the
waterproofing membrane or roof deck is an organic material, such as asphalt or bitumen, an
additional root barrier absolutely must be installed above the waterproofing membrane
(Snodgrass, 2006).
11
Drainage Layer
Standing water stresses plants and can lead to leaks in the waterproofing membrane, so
removing excess water is important to longevity of a green roof. A sloped roof may provide
enough gravity to drain excess water, but flat roofs need some kind of drainage layer to remove
gravitational water. The drainage layer can be constructed of highly permeable granular material,
such as crushed stone, or of geotextiles or other manufacture sheet materials. Some manufactured
drainage sheets consist of cup like structures that can be oriented to retain water, allowing the
roof to retain a greater volume of storm water (Cantor, 2008). Once the cups are full, excess
water will drain. Additional water retention increases the saturated weight of the roof and should
be taken into account during the design process.
Filter Fabric
A fine filter fabric layer is installed between the drainage layer below and the growing
media above. The fine fabric material prevents fine particulate matter from clogging the drainage
layer. The filter fabric also acts as a moisture distribution mat, redistributing localized pools of
excess water to drier areas of the roof. The filter fabric should be root permeable so plants can
access water in the drainage and water retention layers during drought (Snodgrass, 2006).
Growing Media
Green roof growing media may not resemble rich topsoil much at all. Extensive green
roof media must meet several criteria. Guidelines from the FLL dictate physical characteristics of
green roof media. The media must be porous; retain water, oxygen, and nutrients; and provide
stability for root systems. Though not dictated by FLL guidelines, media may also be engineered
12
to be lightweight in order to reduce total green roof load. The ideal extensive green roof substrate
should consist of 75 to 90 percent inorganic, weed-free material with some organic material.
Commonly used inorganic materials include expanded slate, expanded shale, expanded clay,
baked clay, volcanic pumice, scoria, sand, crushed brick, and crushed clay roofing tiles. Compost
should be used as the organic matter because soils contain silt and clay that can clog the filter
fabric (Snodgrass, 2006). The medium should also strongly resist compaction and deterioration
so it retains its depth over the lifetime of the roof.
The Pennsylvania State University Agricultural Analytical Testing Laboratory measured
the watering holding capacity of 39 multi course green roof media samples using the standard
extensive roof media test. The findings showed the average water holding capacity of the media
to be 43.1 percent with a low of 14.7 percent and a high of 65.2 percent (Berghage et. al. 2007).
Media depth is very important to plant survival. Though extensive green roofs have less
than 15.2 cm (6 in) of media, plant viability varies greatly between 5.1 cm and 15.2 cm (2 to 6 in)
of inorganic material. A single species of Sedum may survive well in 5.1 cm (2 in) of media, but
deeper rooted plants with greater nutrient requirements, such as grasses, will not survive well in
such a shallow environment (Snodgrass, 2006).
Vegetation
The top layer of a green roof consists of vegetation. Requirements for suitable green roof
vegetation are discussed further in the Chapter 1-4 Green Roofs Plants.
13
Chapter 1-4: Green Roof Plants
Given ample water, media depth, and maintenance, the plant palette for green roofs is
quite large and can include succulent species, herbaceous species, annuals, perennials, and even
shrubs and small trees. However, given the same geographic area, the environment on a green
roof, especially extensive green roofs, is more extreme than the environment on the ground.
Because a green roof is on top of a building, it is exposed to increased wind velocities, increased
sun exposure, increased heat, more frequent drought, and shallower growing media than a similar
location on the ground (Durham et. al. 2004) As a result, the plant selection, particularly for
extensive green roofs with shallow media and low maintenance requirements, is limited to plants
that can withstand the extreme environments on rooftops. Hardy succulent species are well suited
to the rigors of life on a green roof.
Hardy succulent species in the genus Sedum are commonly used green roof plants. The
genus Sedum is a member of the family Crassulaceae of the order Rosales. Other members of the
Crassulaceae include Crassula, and Sempervivum. Members of the Crassulaceae store water in
fleshy leaves, stems or roots, allowing them to survive drought (Stephenson, 1994). Sedums are
highly drought resistant, shallow rooted, and very easy to propagate, all of which suits them to
prolonged life on green roofs. The research in this thesis was performed on Sedum species, so
while there are many other plant species suitable for green roofs, they are not discussed further in
this review.
14
Like many drought tolerant species, many Sedums use specialized photosynthesis known
as Crassulacean Acidic Metabolism (CAM) (Stephenson, 1994). Plants using CAM
photosynthesis open their stomata at night when rates of transpiration are low. Carbon dioxide
absorbed through stomata overnight is temporarily stored as the organic acid malic acid. Malic
acid accumulates in the vacuoles of leaf cells, increasing their osmotic concentration, which
allows the plant to absorb and store water, leading to observed succulence. When temperatures
rise the next day and stomata close, malic acid moves from the vacuoles and is broken down to
release carbon dioxide within the cells. Using sunlight, CAM plants re-fix the released carbon
dioxide into the usual products of photosynthesis, sugar and starch (Ingram, et al. 2002).
Non-CAM plants open their stomata during the day when the sun is shining, temperatures
are warm, and rates of transpiration high. By keeping stomata closed during the day and open at
night, CAM plants reduce the amount of water lost through transpiration. This adaptation allows
CAM plants such as Sedum to tolerate large diurnal temperature changes and prolonged drought
(Stephenson, 1994). Some Sedum species, such as S. album and S. telphium can switch from C3
to CAM photosynthesis in response to water availability (Nagase and Thuring, 2006).
15
Sedum Propagation
In addition to their ability to grow in harsh conditions, Sedum species are easy to
propagate. In his book Sedum: Cultivated Stone Crops, Ray Stephenson, describes how easily
Sedum species can be propagated.
“For the majority of Sedum species, if pieces were merely ripped off the parent and rammed into any growing medium, some propagations would be successful.
To be even more successful, choose healthy, sterile stems (without flowers or
buds) and strip them of their lower leaves. When inserted into porous, sandy, or
gritty compost, these stems quickly form roots. Avoid keeping cuttings in full sun. Put them in a position where birds, cats, or rodents will not disturb plants
and labels. Water the cuttings after a week or so. Unlike mesophytes that relish
a watering-in process, Sedum cuttings are likely to rot if watered immediately. By waiting at least three days so that tissue damage has had time to callus, rot is
unlikely and small rootlets are likely to have formed to make use of the delayed
moisture. Finally, do not take cuttings of biennials or annuals after the longest day of the year.”
An excerpt from Sedum: Cultivated Stone Crops. (Stephenson, 1994).
Stephenson goes on to say that few plants are easier to propagate than Sedums, many of
which grow from a single leaf if given proper care. Note that Stephenson specifically mentions
that porous, sandy, or gritty compost should be used for successful propagations. As described
earlier, green roof media is course and porous, and should contain compost, suggesting green roof
media may be well-suited for Sedum propagation.
At Emory Knoll Farms, Inc., a nursery specializing in green roof plants, Sedum plugs are
propagated almost exclusively from cuttings. Sedum album and S. sexangulare, in particular, root
easily from tiny small cuttings. At Emory Knoll Farms, S. album and S. sexangulare cuttings are
harvested, then chopped into tiny cuttings and sprinkled onto growing media in a nursery plug.
The smaller cuttings are more desirable for these species. (Personal observation of author 2008).
16
Chapter 1-5: Green Roof Planting Methods
Green roof vegetation is established by five different methods, or sometimes by a
combination of methods: direct seeding; cutting propagation; planting nursery plugs; installation
of vegetated mats; and installation of vegetated modules. Larger nursery containers should not be
used, because the potting medium is rich in organic matter that is unsuitable for green roofs
(Snodgrass, 2006). Plants transplanted from containers to green roofs often die when the rich
potting media wicks excess moisture away from the roof (personal observation of author).
Seeding
As of 2006, no green roof in North America has been established solely by planting seeds
(Snodgrass, 2006) Seeding may require two to three years before a roof planting is fully
established, making seeding the slowest method of plant establishment. Sedum seed is small and
difficult to disperse evenly, especially when specific plant designs are specified. One gram of
Sedum seed can cover a 929 m2 (10,000 ft2) roof at recommended 215 seeds per m2 (20 seeds per
ft2). Evenly distributing one gram of seed over 929 m2 (10,000 ft2) is a daunting task to say the
least (Snodgrass, 2006).
In the summer of 2008, a mixed Sedum cutting bed was seeded at Emory Knoll farms.
The bed required watering three to four times a day in the middle of the hot Maryland summer.
This kind of attention makes seeding a rather impractical method of planting a roof.
17
Montrusso, et al. (2005) studied establishment rates of Sedum seed on green roof
platforms in Michigan. The study shows 100 percent Sedum coverage by the spring of the second
year of the study. It is important to note however, that the platforms were irrigated in 15 minute
cycles applying 0.38 cm of water with each cycle. Irrigation occurred three times a day from
days 1-36, twice a day from days 37-51, and once a day from days 52 to 91, with day 91 being the
end of the first growing season. Irrigation resumed once a day during the second growing season,
day 362.
Cutting Propagation
Spreading cuttings directly on the growing media is a viable method of establishing
Sedum and Delosperma species on a green roof. Distributing cuttings is the most common
planting method in Germany where green roofs are more common (Snodgrass, 2006). Cuttings
establish new plants faster than seeds and may not require supplemental irrigation depending on
weather conditions and time of year. Cuttings should be spread at a rate of 9-12 kg/100 m2 (18-
25 lb/1000ft2) (Snodgrass, Personal Communication.. 2/2010). Cuttings are more expensive than
seeds but cheaper and easier to install than nursery plugs (discussed below). However, there are
inherent risks in planting a roof entirely of cuttings. Compared to nursery plugs, cuttings require
more monitoring and can require more irrigation to ensure growth. Cuttings can easily be blown
or washed off a roof, or moved around by birds and rodents that might be present of a roof
(Stephenson, 1994). Using cuttings alone effectively reduces the plant selection to Sedum and
Delosperma species (Snodgrass, 2006).
18
Nursery Plugs
Nursery plugs are essentially small young shoots that are fully rooted in a small amount
of propagation media. Plugs are commonly established from cuttings, but can also start from
seed. Because plugs have established roots prior to planting in the roof, each individual plug is
more likely to survive than an individual cutting. Therefore the risk of total plant failure is
reduced as compared to using only cuttings. Planting plugs with established root systems allows
the installer to use a greater variety of plant species. Common green roof plugs come in 72 cell
plug trays and are 8.9 cm (3.5 in) deep. The typical planting density is 23 plugs per square meter
(2 plugs per ft2), which provides full coverage in 12 to 18 months. Doubling the planting density
increases the rate of coverage, but also doubles the cost of plant material and may double the
labor required to plant the roof. Assuming a planting density of 23 plugs per square meter (4
plugs per ft2), a single 72 plug tray will cover about 3.3 square meters (36 ft2) of roof area. The
use of plugs also extends the planting season from just after the last frost of spring until late
summer or early fall in mild climates and increases the plant palette to include species other than
Sedum and Delosperma (Snodgrass, 2006).
Vegetated Mats
Vegetated mats consist of a thin layer of mesh in which plants are pre-grown in a nursery
or other off-site location. The advantage of a vegetated mat is that the roof is instantly green after
installation. Vegetated mats are also useful on sloped roofs because the mat and established root
systems help reduce erosion by holding the growing medium in place.
19
Vegetated mats have a number of drawbacks compared to other installation methods.
Vegetated mats are bulky, heavy, and difficult and expensive to transport. Mats cannot survive
long distance transport in hot weather, so refrigerated trucks may be required to move a mat from
a nursery to its final location. Once a mat arrives at a job site, it must be unrolled immediately, or
else it will die, meaning there is little flexibility to adapt to unforeseen installation problems.
Another drawback of vegetated mats is that one square meter of roof space requires one square
meter of nursery space to grow the mat, making the mat more expensive to grow per square meter
of roof coverage than plugs. When rolled, the weight of all the vegetation is concentrated directly
under the mat. This means that, if placed on the roof when rolled, the mat could collapse the roof
because the weight of the mat is concentrated.
Despite the instant green appearance provided by a vegetated mat, a Swedish study found
that after 3 years, there is no difference in succulent cover on roofs planted from cuttings, plugs,
or vegetated mats. This means that any of the three planting methods should yield the same level
of plant coverage after 3 years. The advantage of a vegetated mat then is only realized in the first
two years of life for a green roof installation (Emilsson, 2008).
Pre-Vegetated Modules
Modules are plastic or metal containers, generally 0.37 to 1.5 m2 (4 to 16 ft2), which are
filled with growing media and plants. The depth of the tray can vary, but should be less than 10
cm (4 in) for extensive green roof applications. Modules are pre-grown at a nursery and placed
on top of a roof. Modules are a very expensive installation method, and share many advantages
and disadvantage with vegetated mats. Modules contain more growing media than vegetated
mats, so they are heavier by area (Snodgrass, 2006). The weight of modules increases their
20
associated shipping expenses and makes modules tiresome to move and install manually.
However, modules are easy to install without any horticulture knowledge, and can be removed
easily should the roof leak or a module die. Like vegetated mats, modules require one square
meter of nursery space to one square meter of roof space. Modules can cause weight loading
problems if they are stacked during installation or maintenance. Stacking modules concentrates
their weight at certain points instead of evenly dispersing the load across the roof. If modules
need to be stacked, the stacks should be kept small.
Chapter 1-6: Vegetation Effects on Green Roof Performance
A green roof must support 60 percent plant coverage before it is considered a finished
product that can be handed over from a contractor to a building owner (FLL, 2002). In addition
to the aesthetic value provided by plant coverage, plants have significant impacts on the ability of
a roof to provide all the benefits mentioned in Chapter 1-2. Roots hold the growing media in
place preventing erosion, affect heat loss from a building, and affect the stormwater retention
abilities of the roof. Keeping desirable plant coverage maintains the intended benefits for the life
of a green roof.
During establishment, rapid root growth helps prevent wind and water erosion by holding
the growing media in place on the roof (Rowe, et. al., 2006). Should all the growing media be
lost to erosion, the green roof would no longer support plant life and no longer be considered a
green roof.
21
Transpiration affects heat loss from a building, thereby affecting the cooling load for the
supporting building. Two types of heat loss are considered: latent heat loss and sensible heat loss.
Latent heat loss is heat lost due to evapotranspiration. Sensible heat loss is heat lost due to
convection and conduction. The latent heat loss is three to eight times higher than the sensible
heat loss from a green roof. There is no doubt that plant transpiration is partly responsible for the
latent heat loss from a green roof (Gaffin et al. 2006). Without a healthy plant population, a
green roof can not realize all its potential energy saving benefits.
It has been suggested that the stormwater retention benefits of a green roof are due to the
water holding capacity of the green roof growing medium with little contribution from plants.
However, experiments at The Pennsylvania State University using green roof modules planted
with succulent Delosperma nubigenum, Sedum spurium, and Sedum sexangulare suggest that
plants can contribute as much as 40 percent of the annual stormwater retention function of a green
roof (Berghage, et al. 2007). The contribution of plants to water retention appears to be due to
the effects of evapotranspiration on media water storage.
Plants use water for growth, metabolism, and cooling. To fulfill these life requirements,
plants must use their roots to remove water from the growing media, move it through vascular
tissue, and release it through pores called stomates. By removing water from the growing
medium, plants replenish the water storage capacity of the medium. Using lysimeters and
vegetated and non-vegetated green roof modules, Berghage, et al. (2007) found that vegetated
modules experienced more rapid water loss for the first 5 to 6 days following irrigation than non-
vegetated modules. After 5-6 days, the rate of water loss slows and is not statistically different
than the rate of water loss from unplanted modules.
22
Through the use of modeling, studies at The Pennsylvania State University suggest that
increasing media depth provides surprisingly little increase in the annual storm water retention of
a green roof. The model suggests that a green roof with 89 mm of growing media, planted with
Sedum spurium, providing 40 mm of water retention capacity will retain 45 to 55 percent of
annual rainfall volume in State College, Pennsylvania. In the same climate, providing only 3 mm
will retain 25 to 40 percent of annual rainfall (Jarrettt, et al. 2006). The model uses a low storage
capacity of 3 mm and high of 76 mm. The percent of annual rain retained varies from about 35 to
55 percent respectively. This suggests two things: first, that increasing the thickness of growing
medium does not greatly impact the annual rain water storage capacity of a roof, second, that
adding as little as 3mm of water storage capacity to a roof significantly reduces the annual
rainwater runoff from that roof. A roof with 3mm of storage capacity does not provide enough
media depth to support plant life, so such a roof in not considered a green roof (Jarrett, et al.
2006).
Selecting plants with large leaf area, high concentration of stomates, and high water
demands will increase the evapotranspirative potential of a green roof, which may increase the
economic benefits of green roofs. By increasing economic benefits green roofs may be more
attractive to building owners and communities (Compton, et. al. 2006). In order to increase
economic benefits, Compton and Whitlow argue that green roofs should be designed with the
goal of optimizing evapotranspirative benefits, rather than the goal of reducing maintenance. By
not allowing rain water to drain, using plants with high rates of evapotranspiration, and irrigating,
the authors created a zero stormwater discharge green roof system.
Theoretically a roof populated with plants that have large leaf area and high densities of
stomata will use more water for evapotranspiration compared to a roof populated with low
23
growing, succulent species with high tissue to volume ratios. However, irrigation is required to
grow the former, while the latter will survive periods of drought. A green roof only reaps the
benefits of evapotranspiration when the plants are alive, so low growing succulents seem to be a
better choice on non-irrigated roofs (Berghage, et al. 2007).
In a study testing the acid rain buffering capacity of green roof media, Berghage et. al
(2007) found that plants generally have impact the buffering capacity of a green roof. The study
used planted and unplanted green roof modules irrigated with deionized water adjusted with
sulfuric acid to pH 4. Leachate from planted modules was as much as 1 pH unit higher than the
leachate of unplanted modules. Eight of the nine plant species tested produced higher pH
leachate than leachate from unplanted modules. The pH difference between planted and
unplanted modules became more pronounced after about 50 to 100 days of the study.
Chapter 1-7: Green Roof Maintenance
The focus of this thesis is finding a suitable method of increasing Sedum coverage on a
green roof suffering from poor plant coverage. Ultimately this is a maintenance concern.
Maintenance during establishment is critical to the long term success of the roof, and
hand weeding and fertilization are required (Snodgrass, 2006). The FLL (2002) dictates plant
cover requirements be met before a green roof is handed over from the contractor to the owner.
Seeded or planted vegetation should have gone through a dormant phase, and where weather
permits, a period of drought or frost, generally this requires 12 to 15 months. Green roofs created
24
by seeding and planting of shoots of the genus Sedum should provide as uniform a plant stock as
possible and aim to provide 60 percent ground cover when the plants are in an uncut state. At
least 60 percent of the plant stock must consist of varieties contained in the seed mixture. The
population of shoots from plants in the genus Sedum must be no less than 75 percent of plant
stock and those shoots must have taken root. Fostered and alien vegetation can not be considered
part of the 60 percent ground cover. If plants in the fostered and alien category exceed 20 percent
of the ground cover, the site is deemed unsuitable for handover. Vegetation which has become
rampant and thus weakened by excessive watering and fertilization is not fit to be handed over.
The FLL (2002) instructs that the maturation period of a roof, the time between planting
and roughly 90 percent plant coverage, can take up to two years on extensive green roofs,
depending on the planting method and how far development has advanced. After maturation,
additional maintenance work is needed to maintain high percentages of plant coverage. Plant
care that should continue after the maturation period includes feeding with nutrients, removal of
alien coppice material and other unwanted vegetation, pruning and thinning, and infill seeding on
sizeable bare patches (FLL, 2002). The FLL dictates that as a rule, maintenance on extensive
green roofs consists of nothing more than one or two inspections per year.
The initial growing media should contain enough fertility to support the roof for a year.
After a year, slow release fertilizer should be applied at a rate of 5 gN/m2 on extensive roofs and
8 gN/m2 on intensive green roofs (FLL, 2002). 14-14-14 slow release fertilizers are commonly
used (Snodgrass, 2006). After 5 or 6 years, fertilization may not be needed at all, depending on
the health of the initial planting (Snodgrass, 2006). Fertilization should provide enough fertility
to support hardy succulent species, but not enough to promote weed growth. Over fertilization
25
leads to increases in weed coverage without increasing the coverage of desirable species such as
Sedums.
Fertilization alone could potentially increase the desired plant coverage on a green roof,
and it might be a welcome addition to the method described here. The addition of fertilizer will
increase materials and labor costs associated with any maintenance procedure. As noted in the
literature review, the rate of fertilization should provide enough nutrition to sustain Sedum
species, but not so much as to encourage weed populations. This balance may be difficult to
attain as there is a fine line between appropriate fertilization and excessive fertilization.
Fertilization also has the negative side effect of adding nutrients to any runoff that is discharged
from the roof. Because storm water mitigation is a primary goal of green roofs, adding nutrients
to the runoff is not desirable.
Aside from aesthetics, thick coverage by desired plant species is important for reducing
other maintenance requirements. Fully vegetated roofs are far more resistant to weed pressures.
Some weeds are inevitable in a green roof, because weed seeds are introduced in the growing
media, by wind, birds, rodents, and humans. When allowed to grow, weeds can out-compete
desired plants for nutrients and water during periods of optimal growing conditions then die
during periods of drought and stress leaving large empty patches of exposed media. The exposed
media then provides an excellent place for a new generation of weed seeds to germinate.
Exposed media and dead plants are also unsightly. Weeding should occur before the weeds are
allowed to set seed (Snodgrass, 2006).
26
To maintain the acid buffering capabilities of the roof system, Berghage, et al. (2007)
estimate that media pH should be evaluated every 13 and 19 years for clay-based and slate-based
media respectively. The study assumes no acidification of the media except that from acid rain
deposition. At that point, liming may be required to replenish the pH buffering capacity of the
media.
Maintenance unrelated to plants must occur over the life of a green roof. The FLL (2002)
states that maintenance work should include ensuring that roof outlets and drainage and watering
systems are in working order; removing dirt and deposits from inspection manholes, overhead
sprinklers, roof outlets, and drains; ensuring that surrounding fastenings and other structural
components are firmly in place and in good condition; and removing gravel deposits at joints and
on equipment should be conducted at intervals of several years.
27
CHAPTER 2: THESIS GOALS AND OBJECTIVES
This thesis proposed and evaluated a method of landscape maintenance designed
specifically to increase plant coverage on poorly vegetated portions of a green roof. Sedum
plants already established on the roof provided thousands of cuttings, which in turn provided
numerous propagules. Sedum cuttings were harvested manually from plants established on the
experimental roof using four tools: two pairs of hand shears, a manual reel mower, and a gasoline
powered string trimmer. Two pairs of hand shears, a pair with an extension pole and a pair
without an extension pole, were used because the pair with the extension pole proved too slow
and tiresome to continue using. Cuttings were then spread by hand on exposed media, watered
immediately, and allowed to root and grow for the remainder of the growing season.
Two experiments were used to evaluate the proposed method: The first experiment was
performed outside on an experimental green roof from May to October, 2009 at the University
Park campus of The Pennsylvania State University. The second experiment was conducted in a
greenhouse from April to October 2009 at the same institution.
The first objective of the thesis was to demonstrate that the proposed maintenance
method can be used to successfully increase the number of new Sedum plants on the green roof
by the end of one growing season. The second objective of the thesis was the evaluation of four
different tools for their usefulness for harvesting Sedum cuttings from a green roof. The third
objective was developing a watering regimen to reduce Sedum cutting death and improve rooting.
The fourth objective was to show that all the Sedum species present on the green roof were
28
capable of producing viable cuttings. By meeting these objectives, the proposed maintenance
method can be honed to supply effective and efficient Sedum coverage.
29
CHAPTER 3: OUTDOOR EXPERIMENT INTRODUCTION
Research suggests that full plant coverage is necessary to reap all the potential benefits of
a green roof. Realistically, not all green roofs will attain and maintain 100 percent plant
coverage. Poor plant establishment, poor plant choices, poor maintenance, poor growing
conditions, and extreme weather conditions can all lead to less than 100 percent coverage.
Figure 3-1 shows the green roof on which this experiment takes place before any treatments were
applied.
Figure 3-1: Photograph of the experimental green roof on 5/25/2009.
This photograph shows noticeable areas of exposed media between patches of establish
Sedum plants. Like other green roofs, poor plant coverage resulted from a combination of poor
plant establishment, poor plant selection, and little maintenance.
30
Chapter 3-1: Goals and Objectives
The goal of this experiment was the evaluation of a green roof maintenance method
introduced in Chapter 2 as a commercially viable method of increasing Sedum coverage on a
green roof.
The first objective of the experiment was showing that harvesting stem and leaf cuttings
and then spreading them on exposed media lead to the establishment of new Sedum plants. This
objective was met by determining and comparing the density of new Sedum plants in treated and
untreated areas of the roof at the end of the growing season.
The second objective of the study was evaluating four tools for efficiency in harvesting
Sedum cuttings. The four different tools used in the study include two different types of hand
shears, a manual reel mower, and a gasoline powered string trimmer. One pair of hand shears had
an extension pole and the other did not. Other tools may be suitable for harvesting cuttings, but
only four were tested in this experiment. The objective was met by timing how long it took the
author to harvest cuttings from one treatment and by comparing the quality of the cuttings
produced. Quality was measured by general appearance and health and by whether the cutting
was flowering.
The third objective was a comparison the species composition of the cuttings harvested at
the beginning of the season to the new plants counted at the end of the season. This objective was
31
met by noting the species of each cutting in a random sample of cuttings and noting the species of
each new plant counted.
Chapter 3-2: Hypotheses
For the purpose of these hypotheses, the act of harvesting and spreading cuttings was
considered a treatment. Therefore a treated section was one in which cuttings were harvested
with any tool and spread by hand. Control sections were those in which no cuttings were
harvested and no cuttings were dispersed. The specific hypotheses evaluated in this study were
the following:
1) Treated sections would show greater densities of new plants at the end of the
experiment.
2) There was a relationship between the composition of the cuttings harvested and the
composition of the new plants on the roof.
3) One of the four harvesting tools would yield higher quality cuttings than the others.
Criteria for evaluating cuttings are described in this report.
4) One of the harvesting tools would be more efficient than the others. Efficiency was
measured as time required to take cuttings from one treatment.
32
CHAPTER 4: OUTDOOR EXPERIMENT MATERIALS AND METHODS
Chapter 4-1: Materials
Harvesting Tools
Two sets of hand shears were used in this experiment. The Hound Dog® long handled,
hand actuated grass clippers were used to harvest cuttings in replications 1, 2, and 3. The Hound
Dog is a product of Ames True Temper ©, Camp Hill, PA 17011. The Hound Dog shears were
purchased inexpensively at Ollie’s Bargain Outlet of State College, Pennsylvania in May 2009,
and the product appeared to be discontinued by the writing of this thesis (www.hound-dog.com).
The long handled shears were used in an attempt to reduce worker fatigue caused by bending over
for an extended period of time, but this device was very slow, cumbersome, and tiresome to use.
In replications 4, 5, and 6 standard short handled shears were substituted for the Hound Dog. The
standard shears did not have a company name, but a photograph can be seen in Figure 4-1. The
short handled sheers were comparatively more maneuverable, faster, and less tiresome to use than
the Hound Dog ®, but they did require the worker to bend over while harvesting cuttings.
33
Figure 4-1 – Photograph of the short handled hand sheers used in this experiment.
The manual reel mower was an unknown brand in old, but working condition. The
blades were capable of cutting long grass, but it was not sharpened prior to use on the green roof.
The mower was capable of cutting all the Sedum species living on the roof. The mower was also
capable of cutting weeds present on the roof as long as the plant material was taller than 3.8 cm
(1.5 in).
The string trimmer used in this experiment was a Troy-Bilt® model TB10CS powered by
a two cycle gasoline motor. The string trimmer was produced by MTD Products, Inc. Cleveland,
Ohio.
34
A wet dry vacuum was used to collect cuttings. The vacuum was a Shop Vac® model
90P650A powered by 6.5 horsepower electric motor (120V 60 Hz 12 A). The vacuum was
manufactured by the Shop Vac Corporation, Williamsport, Pennsylvania.
Plant Material
The only plants used in this experiment were ones already established on the
experimental green roof. During the experiment, no plants were imported from an outside source.
The plants of interest were all Sedum species and included Sedum acre ‘aureum’, S. album, S.
rupestre ‘Angelina’, S. hispanicum, S. kamtschaticum, S. rupestre, S. sarmentosum, S.
sexangulare, S. spurium (John Creech), and S. spurium ‘fuldaglut’. All these species produced
viable cuttings and new plants.
35
Chapter 4-2: Experimental Design
The experiment was conducted using a randomized block design. There were four
treatments in the experiment, and each treatment was used one time within a replication. The
experiment was replicated five times. The four treatments were as follows:
Treatment 1: Control. No cuttings were harvested and no cuttings were spread
Treatment 2: Cuttings were harvested with hand shears. Two types of shears were used
because the first pair was too inefficient to continue using. The shears were grouped
together into one treatment. Harvested cuttings were spread by hand.
Treatment 3: Cuttings were harvested with a manual reel mower. Harvested cuttings
were spread by hand.
Treatment 4: Cuttings were harvested with a gasoline powered string trimmer.
Harvested cuttings were spread by hand.
36
4 - Reel Mow er 4 - String Trimmer 3 - String Trimmer 3 - Control
4 - Hand Sheers 4 - Control 3 - Hand Sheers 3 - Reel Mow er
5 - Control 5 - String Trimmer 2 - Control 2 - String Trimmer
5 - Reel Mow er 5 - Hand Sheers 2 - Reel Mow er 2 - Hand Sheers
6 - Control 6 - Hand Sheers 1 - String Trimmer 1 - Reel Mow er
6 - String Trimmer 6 - Reel Mow er 1 - Hand Sheers 1 - Control
North
Roof Area = 397 m2 (4273ft2)Eisenhower Auditorium
7.1
1 m
(23' 4
")6.9
3 m
(22' 9
")6.8
1 m
(22' 4
")
Tyson
Building9.55 m (31'4") 9.47m (31' 1")
Greenhouses
Figure 4-2: A diagram of the experimental layout.
The roof was already divided into six sections by cinder blocks that were part of the roof
installation. Each of the six sections served as one full replication of the experiment. Cinder
block divisions are represented by dotted lines. The replications are labeled 1 through 6 in a
counter clockwise fashion with replication 1 being the northwest corner of the roof. Each section
was divided into quarters with a grid of masonry twine. Masonry twine divisions are represented
by thin solid lines in Figure 4-2. Each treatment area was labeled with its section number and the
treatment it received (Section Number – Treatment), i.e. 1 – Control.
37
The six sections were not equally sized, so therefore the replications and the treatment
areas were not the same size. The size of each replication and each treatment area is found in
Appendix D: Figures and Tables.
Physical Characteristics of the Experimental Green Roof
Figure 4-3: A map showing the location of the Root Cellar building on the University Park
campus of The Pennsylvania State University (The Gould Foundation).
The experimental green roof was installed in 2006 and planted as a class project by the
2006 green roof class taught Dr. Robert Berghage at The Pennsylvania State University. The roof
was intended as a research roof. Because the root cellar building was mostly underground, the
38
roof was actually at the ground level, and it was viewable by the public. The root cellar building
was not heated. Sections 1 and 6 were installed with media 15 cm (6 in) deep. Sections 2
through 5 were installed with 10 cm (4 in) deep. In all sections, media was mounded to a depth
of about 20 cm (8 in) along the lines of cinder blocks. Mounding provided enough media depth
for the growth of some herbaceous species along the edges of the individual sections.
Weight loading was not an issue for this roof because the underlying building was
constructed of thick concrete capable of withstanding significant live and dead loads. Because
weight loading was not an issue, the media on this green roof was not comprised of lightweight
inorganic aggregate material common on commercial roofs. In an effort to reduce the total cost
of installation, inexpensive but heavy materials were used for growing media. The media used
was a custom mixture of roughly 75 percent sandstone gravel and 25 percent pre-consumer
compost from the composting facility at The Pennsylvania State University (Berghage, 2010
Personal Communication.).
Some of the original experiments on this roof tested plants for their survival on green
roofs. Many of the original plant species did not survive on the green roof in this climate.
Additionally, the roof was planted with distinct patterns of vegetation. The death of experimental
plants and the heavy patterns took their toll on the total plant coverage, leaving large areas of
exposed media with no plant coverage. Although many herbaceous plants died, most Sedum
species thrived, establishing areas of Sedum coverage. Vegetation maps from 2007 and 2009 are
included in Appendix A.
39
Cutting Data
A sample of 1500 cuttings was used to characterize the cuttings that were harvested and
spread during treatment. Cuttings were evaluated by species, quality, and length. Cutting weight
was not measured because rain fell prior to harvesting the cuttings. The water weight would have
significantly affected the total weight of any cutting sample.
Species
The species of each cutting was noted. If a cutting was unrecognizable, it was scored as
“other”. The data were used to determine species composition. Composition was reported as the
percentage of the total cuttings represented by a single species. For example, to find the percent
of S. album, the number of S. album cuttings was divided by the total number of cuttings. The
ratio was multiplied by 100 to yield a percentage.
Quality
Quality was partially subjective and scored as either “yes” or “no” based on two criteria:
flowering and damage. The quality of cuttings was used to evaluate harvesting tools.
Flowering: Flowering was scored yes if a cutting had buds or flowers.
Damage: Damage was scored yes if a cutting was broken, ripped, or otherwise severely
damaged by harvesting. Damage was subjective. A cutting marked as damaged was one that the
author would likely throw away if he were trying to achieve 100 percent successful propagations.
40
Length
Cutting length was measured in inches and rounded up to the nearest half inch. The
average cutting length was used as the standard cutting length for greenhouse experiment
explained in the greenhouse experiment section of this thesis.
New Plant Data
The only areas of the roof surveyed were those where exposed media comprised more
than 75 percent of a 1 square foot survey area. Areas with established plants were not surveyed.
New Plant Density (NPD)
NPD was a count of new plants per square meter of exposed media. Values were
collected by counting all new plants (NP) in a 1 square foot survey plot. Multiple survey plots
were used in each treatment. To outline the survey area, a 1 foot by 1 foot frame constructed of
PVC pipe was laid on an area of exposed media. Data were converted from NP/ft2 to NP/m2.
New Plant Species
The species of each new plant was noted. The data were used to determine the species
composition of the new plants. Composition was reported as the percentage of the total new
plants represented by a single species. For example, to find the percent of S. album, the number
of new S. album plants was divided by the total number of new plants. The ratio was multiplied
by 100 to yield a percentage.
41
Other Data
Time of Harvest (Efficiency)
The time required to harvest cuttings from each treatment is measured. Efficiency was
used as criteria to evaluate harvesting tools.
Vegetation Maps
Vegetation maps of each section were drawn prior to any treatments. The vegetation
maps for 2009 are found in Appendix A.
Experimental Timeline
Cuttings were harvested on two dates: 28 May, 2009 and 1 June, 2009. The experiment
concluded on 28 October, 2009.
The time of year when cuttings were harvested was important to the survival of new
plants. Late spring or early summer seemed to be the best time to harvest cuttings. In a personal
conversation, Ed Snodgrass, author of Green Roof Plants: A Resource and Planting Guide and
owner of Emory Knoll Farms Inc., noted that Sedum cuttings were more viable before they
flower. He noted that some German companies prune their stock beds all summer so they never
flower, thus providing viable cuttings all season. Most of the Sedum species on the experimental
roof flowered in summer or fall, not in the spring. In this experiment cuttings were spread
immediately following harvest. There was the added concern that spreading cuttings in the
middle of summer with high temperatures and sun exposure would kill cuttings before they had a
chance to root.
42
Other evidence suggested that spring was the best time to plant green roofs in order to
have the maximum survival rate over the first winter. A study at the Michigan State University
found that 81% of plants survived when planted in the spring, while just 23% survived when
planted in the fall (Getter, et. al., 2007). All this information suggested that an existing roof
should see the best results when cuttings were harvested and spread in the late spring or early
summer.
Finally, in his book Sedum: Cultivated Stone Crop, Ray Stephenson suggested not taking
cuttings of annuals or biennials after the summer solstice, which was the longest day of the year.
The summer solstice fell on 21 June, 2009 (Earth’s Seasons,
http://aa.usno.navy.mil/data/docs/EarthSeasons.php).
43
Chapter 4-3: Methods
The experiment consisted of four separate procedures carried out in order: Drawing a
vegetation map; harvesting cuttings; collecting cuttings; and spreading cuttings. The vegetation
map was drawn prior to any roof treatments and showed the location and species of established
plants as well as the location and relative size of areas without vegetation. Data were collected
during cutting harvest, after cutting collection, and at the conclusion of the experiment in
October. Each step is described in more detail below.
Drawing a Vegetation Map
Blunt wooden stakes (so not to pierce the waterproof membrane) and masonry twine
were used to create a grid of 0.91m by 0.91m (3ft by 3ft) squares across the entire roof. The
maps were drawn between 18 May, 2009 and 22 May, 2009. The intersection of the north-south
line of cinder blocks and the line of cinder blocks on the north side of each replication (the side
toward the greenhouses) was used as the origin for each grid. After the grid was created, a
vegetation map of each individual square was drawn. The edges of the vegetation patches were
drawn, and the species in each patch were labeled. Colored pencils were used to color the map
according to the color key included with the maps in Appendix A. Drawing each square
individually enhanced the detail and accuracy of the map because it allowed the author to better
estimate the size and orientation of individual patches of vegetation. The control treatments were
noted with CONTROL written in purple colored pencil.
44
Harvesting Cuttings
Cuttings from replications 1, 2, and 3 were harvested on 28 May, 2009. Cuttings from
sections 4, 5, and 6 were harvested on 1 June, 2009. The dates were chosen because they fall
after the last day of frost and before the summer solstice. Two dates were used because there was
not time to complete all 6 replications in one day.
Beginning in Section 1 and proceeding counterclockwise, all four treatments were
applied in one section before treatment began on the next section. In other words, each entire
replication of the experiment was completed before the next replication began. At the time of
cutting collection, many of the plants were budding or flowering. Although it was nearly
impossible to completely avoid harvesting cuttings from flowering plants, cuttings were harvested
preferentially from plants that were not flowering, and an attempt was made to avoid flowering
plants. The experimental layout in Figure 4-2 was used to locate each treatment. The individual
steps of the treatment are described in greater detail below.
Data Collected - Time
Time was measured while cuttings were being harvested. The time data from replications
1, 2, and 3 were collected on 28 May, 2009, and the time data from replications 4, 5, and 6 were
collected on 1 June, 2009.
A stop watch was used to measure the time required to harvest cuttings from each
treatment area. For each section, the section number and the tool used were noted. The sizes of
the treatment areas varied slightly. The correct area of each treatment is found in Appendix D:
45
Figures and Tables. Using the measured time and area of each treatment, the efficiency of each
tool was calculated as the time required to harvest cuttings per square meter of roof area.
Treatments
Treatment 1: Control
No cuttings were collected or spread.
Treatment 2: Hand Shears
Existing Sedum plants were cut approximately 3.8 cm (1.5 in) above the surface of the
growing media with hand shears. Cuttings from flowering parts were avoided, and foliage was
not completely removed from any individual plant. The cuttings were allowed to fall onto the
surface of the roof without any additional distribution. The Hound Dog long handled grass
clippers were used in replications (sections) 1, 2, and 3. The regular garden shears were used in
replications (sections) 4, 5, and 6.
Treatment 3: Reel Mower
The blades of the reel mower were set approximately 3.8 cm (1.5 in) above the surface of
the roof. If the experiment were repeated, blade height would need to be adjusted based on how
easily the mower can be pushed. In some instances, multiple passes over the same patch of
vegetation were necessary to harvest all the potential cuttings. Repeated passes were made at
different angles, approximately 45 degrees from the original angle of attack. Cuttings from
flowering parts were avoided, and all foliage was not removed from any individual plant. The
cuttings were allowed to fall onto the surface of the roof without any additional distribution.
46
Treatment 4: String Trimmer
Existing plants were cut about 3.8 cm (1.5 in) above the surface of the roof. Cuttings
from flowering parts were avoided and all foliage was not removed from any individual plant.
The cuttings were allowed to fall onto the surface of the roof without any additional distribution,
although the string trimmer did throw cuttings a few feet at times. The direction in which the
cuttings were thrown could be controlled by the angle at which the trimmer contacts a plant.
When harvesting the edge areas of string trimmer treatments, the trimmer was angled so it threw
cuttings back into the treatment area and away from other treatments.
Collecting Cuttings
Cuttings were collected from each treatment using a 6.5 horse power 16 gallon Shop
Vac® brand wet and dry vacuum, one treatment at a time. An attempt was made to collect all the
cuttings in the treatment area. Once all the cuttings were collected in a single treatment, the
cuttings were thoroughly mixed by reaching into the Shop Vac and stirring them by hand. A
random sample of cuttings (two or three large handfuls) was removed and placed into a brown
paper lunch bag. One full bag of cuttings was collected from each treatment for a total of twenty-
four bags full of cuttings. The remainder of the cuttings, which was the majority of the cuttings,
was spread immediately as described below.
47
Spreading Cuttings
Except for the cuttings that were sampled for data collection, all the cuttings collected in
the Shop Vac were immediately dispersed in the treatment area from which they were harvested.
The cuttings were spread by throwing them onto exposed media until all the collected cuttings
were dispersed. An attempt was made to evenly cover the bare patches and to avoid throwing
cuttings on existing plants.
Data Collected – Cutting Quality
Cutting data was collected immediately after cutting harvest. One hundred (100) random
cuttings were removed from each bag of cuttings. The following characteristics of each cutting
were noted: species; length in inches, rounded up to the nearest half inch; flowering, yes or no;
mangled, yes or no. If the cutting was unrecognizable, the species was counted as other. A
cutting was classified as flowering if it had flowers, buds, or a flower stem. A cutting was
classified as mangled if it was severely damaged or if it had partial leaves or damaged stems due
to the mechanical act of harvesting. Cuttings classified as mangled were ones that would likely
be thrown out if the goal were 100 percent propagation success.
After the data about the cuttings were collected, the cuttings were spread back into the
treatment area from which they were harvested. The cuttings were spread by throwing them onto
exposed media until all the collected cuttings were dispersed. An attempt was made to evenly
cover the bare patches and to avoid throwing cuttings on existing plants.
48
Care and Maintenance
Both of the cutting harvest days were followed by heavy rain. The first harvest was
followed immediately by heavy rain. The rain that fell on 28 May 2009 began immediately after
the cuttings were harvested and spread and continued over night. The cuttings were watered
again by hand on 1 June, 2009 immediately following the dispersion of cuttings harvested on that
date.
Figures 4-4 and 4-5 show daily rainfall and daily high and low temperature respectively
for the months of May and June 2009. The summer during which this experiment occurred was
relatively wet and cool, so drought stress was not much of an issue. The weather allowed weeds
to grow well. The weather data were collected by Weather Observatory in the Department of
Meteorology at The Pennsylvania State University (Syrett, Personal Communication., 2010). The
weather station was located on the University Park campus at the Walker Building.
0.000.200.400.600.801.001.201.401.601.802.00
5/26
/200
9
5/27
/200
9
5/28
/200
9
5/29
/200
9
5/30
/200
9
5/31
/200
9
6/1/
2009
6/2/
2009
6/3/
2009
6/4/
2009
6/5/
2009
6/6/
2009
6/7/
2009
Date
Ra
infa
ll (
cm
)
49
Figure 4-4: Graphical representation of the rainfall between 5/1/2009 and 6/7/2009. Cuttings
were harvested on 5/28/2009 and 6/1/2009.
-5.0
0.0
5.0
10.0
15.0
20.0
25.0
30.0
35.0
5/1/
2009
5/11
/200
9
5/21
/200
9
5/31
/200
9
6/10
/200
9
6/20
/200
9
6/30
/200
9
Date
De
gre
es
C
High Temperature
Low Temperature
Figure 4-5: Daily High and Low Temperatures from 5/1/2009 to 6/30/2009
Data Collection
New plant data were collected on two days in October: 1 October, 2009 and on 28
October, 2009. The first hard frost occurred on 14 October, 2009 which was in between the two
days of data collection. The Sedum species surveyed were hardy and undamaged by the frost.
New plant data were collected following the description in Chapter 4-2.
50
Statistical Analysis
New plant density data were analyzed using Minitab Statistical Software produced by
Minitab Inc., State College, Pennsylvania. A one way analysis of variance (ANOVA) was used
to determine the differences in mean new plant densities. A two sample T-test was used as a
secondary test to confirm that the mean new plant density was different between reel mower and
string trimmer treatments. An ANOVA was also used to determine whether there was a statistical
difference in the percentage of flowering and damaged cuttings between harvesting tools. Results
of the statistical tests are found in Appendix E: Statistical Output.
51
CHAPTER 5: OUTDOOR EXPERIMENT RESULTS AND DISSCUSSION
Chapter 5-1: Explanation of the Proposed Maintenance Method
The maintenance method proposed in this experiment was designed specifically to
propagate new Sedum plants on a green roof in places where growing media was exposed due to
poor plant coverage. The source of cuttings was Sedum plants already established on the same
green roof. Because Sedums were so easy to propagate from cuttings, cuttings provided a reliable
source of new plants. This method was designed to decrease long term maintenance costs by
minimizing the need for new plants, labor, and fertilizer.
Instead of using this method, new nursery plugs and cuttings could be purchased directly
from a plant nursery. However, there would be drawbacks to using nursery plants. The cost of
new nursery plants is one reason to avoid ordering them. At the time of this thesis, the price of
new Sedum plugs from Emory Knoll Farms started at $0.63 each. At the recommended planting
density of 23 plugs per m2 (2 plugs/ft2), the cost of new plugs was $14.49 per m2 ($1.26/ft2).
Cuttings of assorted Sedum species sold by Emory Knoll Farms were available in 18-25 pound
boxes for $300. The cost of cuttings per unit of roof area was approximately $3.20 per m2,
($0.30/ft2 or $300/1000 ft2) (Snodgrass, Personal Communication. 2009). Shipping costs were
not included in the plant prices and would vary depending the location of a project. Sedum plants
growing on a green roof could provide copious and free cuttings for a maintenance worker to
propagate.
52
Ordering new plants also involves advanced planning and scheduling on the part of the
maintenance person or company. Plugs and cuttings from a third party must be ordered in
advance and shipped to a project location to coordinate with installation. Once plants arrive on
site, they must be unpacked immediately and cared for until they are planted. If cuttings are
purchased, they must be spread almost immediately or else they will die. Planting can be delayed
due to weather conditions thus disrupting ongoing installations, and delays in planting greatly
increase the risk of plant loss. By avoiding the use of commercially available nursery plants, the
method proposed in this thesis eliminates the cost of plant material and shipping, reduces the risk
of plant loss due to weather delays, increases the flexibility of a maintenance schedule, and
reduces fossil fuel consumption related to shipping.
Chapter 5-2: Efficacy of Treatment
The process of harvesting and spreading cuttings described previously did result in the
establishment of new Sedum plants. The efficacy of treatment was confirmed by comparing
densities of new Sedum plants in treated and untreated sections at the end of the growing season.
Density was measured as the number of new Sedum plants per square meter of exposed media.
The mean new plant density was significantly greater in treated sections than in control
sections. Figure 5-1 and Table 5-1 compare new plant densities. Despite the fact the treatments
were replicated six times in six different roof sections, only five sections were sampled for new
plants.
53
a
bc
c
b
0
20
40
60
80
100
120
140
160
180
200
Control Hand Sheers Reel Mower String Trimmer
Treatment
Ne
w P
lan
t D
en
sit
y
(NP
/m2)
Figure 5-1: The mean number of new Sedum plants per square meter for each treatment. The
error bars show standard error. Control sections showed significantly lower new plant density than any of the treatments (p = 0.00). The string trimmer showed significantly lower new plant
density than the reel mower (p = 0.036). The new plant densities of hand shear and reel mower
treatments were not significantly different.
Table 5-1: Data table showing the mean number of new Sedum plants per square meter for each
treatment. Control sections showed significantly lower new plant density than any of the treatments (p = 0.00). The string trimmer showed significantly lower new plant density than the
reel mower (p = 0.036). The new plant density of hand shears and reel mower treatments was not
significantly different.
Treatment
Mean Density
NP/m2
(NP/ft2)
Area Sampled m2
(ft2)
Control 24 (2) 2.88 (31)
Hand Shears 121 (11) 2.04 (22)
Reel Mower 159 (15) 1.95 (21)
String Trimmer 96 (9) 2.60 (28)
54
Though no cuttings were spread, new plants established themselves in control sections by
unassisted, natural propagation. Sedum species are quite capable of slowly reproducing
vegetatively and by seed, so new plants were expected to gradually fill in areas of exposed media.
No attempt was made to restrict or control the natural spread of Sedum species in control sections.
Chapter 5-3: Cutting and New Plant Species Compositions
As hypothesized, there was a relationship between the species of the cuttings harvested in
May and the species of the new plants counted in October. Table 5-2 compares the species
composition of cuttings and the species composition of new plants.
Table 5-2: Comparison of the species composition of cuttings new plants from replications 1
through 5.
Species
% of Total
Cuttings
% of New Plants
(Control Sections)
% of New Plants
(Treated Sections)
S. acre 2% 0% 7%
S. album 38% 51% 35%
S. rupestre (Angelina) <1% 0% 0%
S. kamtschaticum <1% 0% 0%
S. hispanicum <1% 0% 0%
S. rupestre 1% 9% 3%
S. sarmentosum 8% 0% 8%
S. sexangulare 23% 31% 31%
S. spurium 20% 3% 13%
S. spurium (Fuldaglut) 8% 6% 3%
Other <1% <1% <1%
Sample Size 1351 68 808
As expected, Table 5-2 demonstrated that the maintenance method proposed in this thesis
was only effective at propagating plants that were present on a roof at the beginning of the
55
season. The percentages shown were measured as the number of cuttings of new plants of a
single species divided by the sample size. The ratio was then multiplied by 100. Control sections
were separated from treated sections when calculating the percentages of new plants reported in
Table 5-2. These data also demonstrated that all the species tested, except S. rupestre (Angelina)
S. hispanicum, and S. kamtschaticum could be relied upon to produce viable cuttings. Cuttings
from the three aforementioned species were tested for viability in the greenhouse portion of these
experiments, and they were found to produce viable cuttings.
The relatively large percentage of new S. acre plants compared to S. acre cuttings was
notable and likely due to the species’ ability to reseed itself. In another study, Sedum acre and
Sedum album were found to be the dominant species after one year on experimental green roofs
planted from seed, suggesting that these two species were relatively aggressive spreaders
(Monterusso, et al. 2005). This phenomenon was also reported by Durhman et al. (2004).
Durhman et al tested 25 succulent species and found that S. acre, S. album, and S. hispanicum
were among the 5 species with the fastest growth rate and greatest area coverage after one
growing season. It is likely that the low number of S. hispanicum and S. acre cuttings was likely
the result of those plants being very small when cuttings were harvested, making them difficult
plants from which to collect cuttings.
S. acre may be underrepresented in the new plant data shown in Table 5-2 due to
sampling difficulty. In areas of Sections 3 and 4 (Figure 4-2 or Appendix A), areas of exposed
media at the beginning of the season were almost fully covered with new S. acre plants at the end
of the season. These areas were not sampled because the number of new plants was too high to
count. New S. acre plants were tiny and numerous, making them very difficult to evaluate as new
and difficult to quantify, even in such a small survey area.
56
Sedum hispanicum has also been observed to have prolific reseeding abilities (Durhman
et al. 2007). Its reseeding ability may account for the large patches of S. hispanicum on this
experimental roof despite the very small number of S. hispanicum cuttings. S. hispanicum
flowers throughout June and July with seedlings appearing by early August.
Chapter 5-4: Comparison of Harvesting Tools – Efficiency
The tools used in this experiment were chosen for specific reasons. Portability and
mobility were important because access to a green roof may be limited to a ladder or a small
access staircase. Additionally, tools used on green roofs should not require an electrical outlet, as
it may be difficult to find a suitable outlet or long enough extension cord to allow use of the tool
over several hundred to several thousand square meters of roof area. Battery powered devices
may be useful, but a long battery life would be required. The three tools chosen, hand shears, a
manual reel mower, and a gasoline powered string trimmer met these preliminary criteria of green
roof suitability.
The time required to harvest cuttings from the 397 m2 experimental roof is displayed in
Figure 5-2 and Table 5-3. Time values were calculated by first measuring the time required to
complete a single treatment. Using the area of the treatment, the time was converted to time
required to treat 1 m2 (seconds/m2). The mean time required to treat 1 m2 was calculated for each
tool. The mean time was then multiplied by the area of the roof, 397 m2, to estimate the time
required to harvest cuttings from the entire roof. Table 5-3 displays time expressed as
hours:minutes:seconds (hh:mm:ss) to treat 1 m2 and the entire roof.
57
1:13:191:12:09
2:01:47
3:42:58
0:00:00
0:28:48
0:57:36
1:26:24
1:55:12
2:24:00
2:52:48
3:21:36
3:50:24
4:19:12
4:48:00
Hand Sheers (long) Hand Sheers
(short)
Reel Mower String Trimmer
Treatment
Tim
e (
hh
:mm
:ss)
Figure 5-2: The estimated time required to cut the entire roof (397 m2).
The times reported in Figure 5-2 only included the time required to cut parent plants.
The times did not include the time required to collect and spread any cuttings. In theory the tools
themselves could be used for cutting dispersal. The operator of hand shears could manually
disperse cuttings while continuing to take more cuttings. A simple flick of the wrist dispersed
cuttings while using hand shears. The reel mower and string trimmer both dispersed cuttings
away from the parent plant. Both the reel mower and the string trimmer also provided the
operator with some control over the direction and distance of cutting dispersal. The reel mower
tended to throw cuttings straight forward and backward along the axis on which it was rolling.
The distance of cutting dispersal could be controlled somewhat by the speed at which the mower
was pushed. The string trimmer tended to throw the cuttings to its sides in the direction that the
string rotated. The distance of cutting dispersal could be manipulated by adjusting the engine
throttle and string speed. With both the mower and the string trimmer, the operator gained a
sense of how to orient the tool to directionally disperse cuttings.
58
The difference in time to cut parent plants with a string trimmer and the reel mower was
negligible, and both were significantly faster than using either set of hand shears. It was
important to note however that, if a string trimmer were actually used on the entire roof, it would
likely need to be refueled at least once, adding an additional few minutes to the treatment time.
The author also observed that using hand shears caused more fatigue than using either the reel
mower or string trimmer. This meant that a harvester using hand shears would likely become less
efficient as fatigue set in. Therefore, it was reasonable to assume that the actual harvesting time
for the entire roof using a string trimmer or hand shears would be longer than the projected time
reported in Figure 5-2 and Table 5-3. The switch from the long handled Hound Dog® shears to
short handled shears was clearly justified by the difference in time required to harvest cuttings.
Chapter 5-5: Comparison of Harvesting Methods - Cutting Quality
In addition to being evaluated by efficiency, the different harvesting tools were evaluated
by the quality of the cuttings produced. Individual cutting quality was evaluated by whether the
cutting was visibly damaged or mangled and by whether it had buds or flowers. Cuttings
classified as either damaged or flowering were ones that the author would discard if 100 percent
propagation success were the objective of treatment. This did not necessarily mean that such
cuttings would not propagate successfully, but successful propagation was less likely. Table 5-4
shows the percentage of cuttings damaged by each tool. Table 5-5 shows the percentage of
cuttings that were flowering at the time of harvest. Flowering was used as a measure of cutting
quality because the harvester had the ability to select parent plants that were not flowering. The
control given by the harvesting tool affected how easily plants could be selected based on
59
flowering. As mentioned in the literature review, flowering cuttings were likely less viable than
non-flowering cuttings.
Table 5-4: Comparison of damaged cuttings taken by each harvesting tool. The percentage of
mangled cuttings is significantly different for all three treatments.
Treatment
Damaged
Cuttings
Total Cuttings
(Sample Size)
Percentage of
Damaged
Cuttings
Hand Sheers 8 551 1%
Reel Mower 45 499 9%
String Trimmer 85 450 19%
Table 5-5: Comparison of the flowering cuttings taken by each harvesting tool. The reel mower
and string trimmer produced significantly more flowering cuttings than hand shears.
Treatment
Flowering
Cuttings
Total Cuttings
(Sample Size)
Percentage of
Flowering
Cuttings
Hand Sheers 126 551 23%
Reel Mower 158 499 32%
String Trimmer 145 450 32%
Based on the percentages of damaged and flowering cuttings, hand shears yielded the
highest quality cuttings. Hand shears gave the harvester more blade control than the other tools,
and they cut plants in a less violent manner. Hand shears also allowed the harvester to straighten
stems before taking cuttings, which helped reduce damage. The added blade control allowed the
harvester to select plants that were not flowering. At the time of cutting harvest, a number of
species, particularly S. sexangulare were beginning to flower, so it was impossible to completely
60
avoid taking some cuttings that were flowering and still complete the treatment in a reasonable
time.
The difference in cutting quality between cuttings harvested with a reel mower and a
string trimmer could have affected the density of new plants shown in Figure 5-1 and Table 5-1.
Lower cutting quality should theoretically decrease cutting viability, thereby reducing new plant
density at the end of the season. The data showed that the string trimmer produced the lowest
quality cuttings and also returned a lower new plant density than the reel mower.
Chapter 5-6: Other Observations from the Roof
A large number of cuttings disappeared off the surface of the outdoor green roof by 22
June, 2009, three weeks after cuttings were harvested and spread. The cuttings may have died
because they dried out or rotted, or may have been moved by wind, animals, or rain. It was also
observed that roots on the new cuttings that were not growing down into the media by 6/22/2009
were shriveled and dead. These observations indicated that the first three weeks after the cuttings
were spread were critical to the survivorship of Sedum cuttings, and ultimately the success of the
maintenance method proposed in this thesis. These observations led to the greenhouse
experiments described later.
Robert Cameron, a Ph. D. student under Dr. Robert Berghage, ran a gasoline powered
rotary mower across the roof during the late summer. He set the blades as high as possible so that
they did not cut any of the Sedum foliage. His goal was cutting down the aforementioned
horsetail because it gave the roof a ragged appearance. Robert Cameron observed that the mower
in addition to cutting down the weeds, the mower also removed seed pods, dispersing Sedum
61
much of the seed produced that year (Cameron, Personal Communication., 8/2009). However,
Kathryn Sanford, another Ph. D. student under Dr. Robert Berghage, studying Sedum seed
viability, noted that the seed she harvested from this experimental roof demonstrated extremely
low viability (Personal Communication., 2/2010). Therefore, caution should be used if dispersing
seed to increase vegetative is the primary goal any future treatment. In addition to dispersing
seed, deadheading the plants has an aesthetic value. Seed pods may be considered unsightly by
some people because they are brown and twiggy in appearance, so there is subjective aesthetic
value to removing seed pods.
62
CHAPTER 6 – GREENHOUSE EXPERIMENT INTRODUCTION
A greenhouse experiment intended to evaluate the effects of irrigation on the rooting of
sedum cuttings, and to test the viability of sedum cuttings harvested from the experimental green
roof was conducted during the summer and fall of 2009. The experiments were conducted in a
computer controlled section of a greenhouse behind the Tyson Building on the University Park
campus. The greenhouse experiments were designed to closer analyze how sedum cuttings root
and survive during the first three weeks after being spread on green roof growing media.
When discussing sedum propagation, Ray Stephenson author of Sedum: Cultivated Stone
Crops wrote, “…Unlike mesophytes that relish a watering-in process, Sedum cuttings are likely
to rot if watered immediately. By waiting at least three days so that tissue damage has had time
to callus, rot is unlikely and small rootlets are likely to have formed to make use of the delayed
moisture” (Stephenson, 1994). If this statement held true for all the Sedum species used in the
outdoor experiment, then many of the cuttings might have rotted and died because they were
immediately soaked by heavy rain or manual irrigation after being dispersed. However, when
flats of sedum plugs were propagated at Emory Knoll Farms Inc., cuttings usually were not
allowed to callous before they were stuck in growing media and watered in. Propagation flats in
the nursery were filled with media and watered prior to cuttings being stuck. The flats were
watered again immediately after cuttings were stuck, and then watered daily for a few more days.
This treatment of sedum cuttings stands in contrast to Stephenson’s recommendations, but still
resulted in successful commercial propagation of Sedums.
63
Stephenson’s recommendation to allow three days before irrigating may be due to the
way in which he recommended propagating cuttings. He recommended sticking cutting into
growing media, whereas this maintenance method proposed setting the cuttings on the surface of
coarse growing media. When cuttings are stuck in growing media, they do not have the same
amount of aeration, which could be the cause for rotting. By laying the cuttings on top of the
growing media, they receive more oxygen because air can move freely around the entire cutting.
Despite the difference in propagation methods, there was merit to studying the effects of
irrigation on the rooting of sedum cuttings.
If the maintenance method put forth in this thesis were used commercially, watering a
green roof the same day cuttings were harvested would be logistically easier and more efficient
than allowing cuttings to callous for three days before irrigating. Immediate irrigation avoids
sending a worker to the same roof more than one time. It also meets the FLL recommendation
that only one or two maintenance visitations should be made to an extensive green roof each year.
However, if Stephenson is correct and sedum cuttings rot when they are not allowed to callous,
then irrigating immediately following cutting harvest may lead to complete failure of the
treatment. The cuttings in the outdoor experiment were watered immediately following
treatment. The difference between Stephenson’s recommended propagation method and the way
that sedum cuttings are propagated in the experiment is that Stephenson recommends sticking the
cutting into compost, whereas this method proposes laying the cutting on top of coarse, granular
media. There are differences in aeration and moisture between these techniques, so cutting laid
on top of green roof media may be less susceptible to rot because they have better flow.
64
A second greenhouse experiment tested the viability of Sedum species collected from the
green roof. Species that were completely non-existent or underrepresented in the cutting mix
found in the outdoor experiment were specifically chosen for the second greenhouse experiment.
Chapter 6-1: Goals and Objectives
The first goal of the first greenhouse experiment was to show that cuttings of some
Sedum species would survive immediate and prolonged irrigation without rotting. The objectives
of the first experiment were to determine if cuttings from four Sedum species would rot if they
were kept constantly moist for three weeks.
The goal of the second greenhouse experiment was to determine if all the species on the
green roof produced viable cuttings. Specifically the second experiment analyzed rooting of
cuttings from Sedum species that were either completely absent or poorly represented in the
cuttings mixture found in the outdoor experiment. The specific species of interest were S. acre
(aureum), S. album (flowered out), S. kamtschaticum, S. hispanicum, and S. rupestre (Angelina),
S. sarmentosum, and S. spurium (Fuldaglut).
65
Chapter 6-2: Hypotheses
1) Cuttings would rot when irrigated immediately after being spread on the surface
of green roof media. This is based on Ray Stephenson’s recommendations.
2) One hundred percent (100%) of the cuttings send out roots regardless of watering
regime.
3) All the Sedum species from the roof would produce viable cuttings, including S.
album plants that flowered vigorously all summer.
66
CHAPTER 7 – GREENHOUSE MATERIALS AND METHODS
Chapter 7-1: Materials
Pots
8 inch bulb pans from Dillen Products/Myers Industries, Inc., Middlefield, Ohio (Dillen Products,
www.dillen.com)
Inside diameter = 20.32 cm (8 in)
Outside diameter = 20.64 cm (8 1/8 in)
Depth 10.16 cm (4 in)
Volume 2.47 L (2.62 qt)
The 20.32 cm (8 in) size was chosen because it allowed an individual plug 324 cm2 (50.3
in2) of area. This was approximately the area given to each plug when a roof is planted with a
density of 23 plugs per m2 (2 plugs per ft2).
Growing Media
The growing media used was Rooflite® extensive green roof media supplied by Laurel Valley
Farms, Avondale, Pennsylvania. Full media specifications are included in Appendix C: Analysis
of Rooflite ® Green Roof Growing Media
67
Plant Material
Plants were taken out of 72 cell, 8.9 cm (3 in) deep plug trays. Plugs were supplied by
Emory Knoll Farms Inc., Street, Maryland. The four Sedum species used in the experiment were
S. album, S. sexangulare, S. kamtschaticum, and S. spurium (John Creech). These four species
were chosen because of their prevalence on the outdoor experimental green roof used is study.
The four species were also common green roof plants and were listed on Emory Knoll Farm’s list
of recommended green roof plants (www.greenroofplants.com). The plugs were grown to supply
cuttings for the first greenhouse experiment that tested watering regimes. The cuttings for the
second experiment were taken from established plants on the experimental green roof.
Chapter 7-2: Methods – Growing a Cutting Stock
Bulb Pan Preparations
Each bulb pan was filled to the top with media. Each pot was topdressed with a heaping
teaspoon (8 g) of Osmocote 14-14-14 fertilizer. This rate of fertilization fell within the medium
rate of 9-15 g per gallon prescribed on the back of the Osmocote bag. Fertilizer granules were
spread evenly across the surface of the growing medium by hand.
68
Planting - 4/14/2009
A single plug was inserted into the center of each bulb pan. Before being planted, the
root balls were gently rubbed between the fingers to unbind the roots. The pans were randomly
placed on greenhouse benches. After the plugs were planted, all pans were watered to runoff. All
plants were watered twice weekly until cuttings were harvested on 14 September, 2009. The
greenhouse was kept at approximately 21°/16° C (70° F/ 60° F) day/night temperatures.
Chapter 7-3: Methods - Watering Regime and Greenhouse Cutting Viability Tests
Experimental Design
The experiment was conducted as a randomized block design. Bulb plans were prepared
as described in Chapter 7-2, but no plugs were planted. The pans were fertilized in April and not
again before cuttings were spread in September, 2009. The non-vegetated bulb pans were
arranged into blocks of 16 pans (4 pans x 4 pans). Each of the four species listed in the plant
material section of Chapter 7-2 was randomly assigned to four pots. Each pot was then randomly
assigned one of the four watering regimes so that one pot of each species received each watering
regime. The watering treatments were daily watering, 1 day between watering, 7 days between
watering, and 14 days between watering. The entire experiment was replicated three times.
69
Harvest and Weigh Cuttings
Cuttings were harvested on 14 September, 2009 from the stock plants grown from
nursery plugs as described in Chapter 7-2. By the time plants were harvested, they were very
large and many had sent shoots out of their own pan. In some cases, shoots had begun to root in
adjacent pots.
Scissors were used to harvest 3.8 cm (1.5 in) cuttings from the tips of established plants.
Cuttings were separated by species and placed in brown paper lunch bags for transportation to a
lab where they were weighed. Ten cuttings of each species were weighed at a time.
Spread Cuttings
Ten cuttings of one species (S. album, S. sexangulare, and S. spurium (John Creech))
were placed on the surface of the growing media in a single bulb pan. Due to their size, only 4
cuttings of S. kamtschaticum were placed on the surface of the same sized bulb pan. All cuttings
were oriented in the same direction and were not touching. In two replications, cuttings were
spread on 14 September, 2009. In the third replication, cuttings were spread one week later, on
21 September, 2009.
Watering
All bulb pans were watered to saturation immediately after cuttings were spread. After
the first irrigation, one bulb pan of each species received each of the following watering
treatments: daily watering, watering every other day (1 day between watering , 1 DBW) ,
70
watering on spreading and at day 8 (7 DBW), and watering at planting with no additional
watering (10 DBW). All bulb pans were watered by hand to avoid cross watering. Water was
evenly distributed across the surface of the pan, so all cuttings received direct contact with
overhead water. Pans were watered to runoff.
Data Collected – Collected Daily
Cuttings were carefully monitored for the presence of roots. The ratio of cuttings
showing roots to total cuttings was noted daily. A pen or pencil was used to move the cuttings so
small roots could be seen. The cuttings were monitored for the presence of roots daily for 9 days.
Chapter 7-4: Methods – Outdoor Cutting Viability
Cuttings from the experimental green roof were propagated to test whether certain
species produced viable cuttings. The bulb pans used in this experiment were the same as the
ones used in Chapter 7-3. Media were prepared the same way as described in Chapter 7-3. The
species tested were S. acre (aureum), S. album (healthy, collected in October) S. album (flowered
out), S. hispanicum , S. kamtschaticum , and S. rupestre (Angelina), S. sarmentosum, S.
sexangulare (healthy, collected in October), and S. spurium (Fuldaglut).
Harvesting Cuttings
Scissors were used to harvest 3.8 cm (1.5 in) cuttings from the tips of established plants
on the outdoor experimental green roof. Cutting were separated by species and placed in a brown
71
paper lunch bag for transportation to a lab where they were weighed. Ten cuttings of each
species were weighed at a time.
Spreading Cuttings
For the watering regime experiment, 10 cuttings of S. album, S. sexangulare, and S.
spurium (John Creech) were placed on the surface of one 8 bulb pan. Due to their size, only 4
cuttings of S. kamtschaticum were placed on the surface of the same sized bulb pan. All cuttings
were oriented in the same direction and were not touching.
For the cutting viability test, 12 to 20 cuttings were placed on the surface of the bulb pan.
Again the exception was S. kamtschaticum, and only 4 cuttings of this species were placed on the
surface of a bulb pan. The specific number of cuttings of each species can be seen in Table 8-1.
Watering
Cuttings were watered twice a week. The greenhouse temperatures were raised to
27°/16° C (80°/60°F) day and night temperatures during this experiment. The temperature was
raised so the growing media could dry out between watering.
Data Collection
Cuttings were checked weekly for the presence of roots.
72
CHAPTER 8: GREENHOUSE EXPERIMENT RESULTS AND
DISCUSSION
Chapter 8-1: Watering Regime and Greenhouse Cuttings Viability
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
1 2 3 4 5 6 7 8 9 10
Days After Spreading
Pe
rce
nt
of
Ro
oti
ng
Cu
ttin
gs
Daily
1 DBW
7 DBW
10 DBW
Figure 8-1: Rooting data for S. album collected in the greenhouse. Thirty cuttings were sampled
73
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
1 2 3 4 5 6 7 8 9 10
Days After Spreading
Pe
rce
nt
of
Ro
oti
ng
Cu
ttin
gs
Daily
1 DBW
7 DBW
10 DBW
Table 8-2: Rooting data for S. sexangulare collected in the greenhouse. Thirty cuttings were
sampled
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
1 2 3 4 5 6 7 8 9 10
Days After Spreading
Pe
rce
nt
of
Ro
oti
ng
Cu
ttin
gs
Daily
1 DBW
7 DBW
10 DBW
Table 8-3: Rooting data for S. kamtschaticum collected in the greenhouse. Twelve cuttings were sampled.
74
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
1 2 3 4 5 6 7 8 9 10
Days After Spreading
Pe
rce
nt
of
Ro
oti
ng
Cu
ttin
gs
Daily
1 DBW
7 DBW
10 DBW
Table 8-4: Rooting data for S. spurium (John Creech) collected in the greenhouse. Thirty cuttings
were sampled.
There were no differences in rooting observed for S. album, S. sexangulare, and S.
spurium (John Creech). No cuttings rotted and died. The results suggest that when placed on
green roof growing media, there is no reason to allow Sedum cuttings to callous for three days
before irrigation.
S. kamtschaticum cuttings were much slower to send out roots. Once they appeared, the
roots of S. kamtschaticum were thicker and appeared to have more root hairs than the roots of the
other species. Eventually all the S. kamtschaticum cuttings did root, but not in the 10 days when
they were observed daily. The number of days to 100 percent S. kamtschaticum rooting was not
noted.
The first greenhouse experiment demonstrated that cuttings of the four Sedum species
tested did not rot and die under the conditions this experiment, which was intended to mimic
75
moderate conditions on an outdoor green roof. This finding suggested that irrigating a green roof
immediately after cuttings are spread on its surface will not result in cutting mortality due to rot.
This has important ramifications for the practical application of the maintenance method
proposed in the outdoor experiment potion of this thesis, because immediate irrigation allows a
single maintenance trip to be made, during which time cuttings are harvested, collected, and
watered.
Chapter 8-2: Outdoor Green Roof Cutting Viability Results
Table 8-1: The percentage and ratio of cuttings from the experimental green roof rooting 7 days after being spread on green roof growing media.
Species Percent of Rooting
Cuttings After 7 Days Ratio of Rooting
Cuttings After 7 Days
S. acre 100% 17/17
S. album (Healthy) 100% 12/12
S. album (Flowered Out) 100% 15/15
S. hispanicum 100% 15/15
S. kamtschaticum 75% 3/4
S. rupestre (Angelina) 100% 15/15
S. sarmentosum 100% 13/13
S. sexangulare (Healthy) 100% 20/20
S. spurium Fuldaglut 100% 15/15
Cuttings from S. album plants that had flowered all season were viable. Those cuttings
are labeled in Table 8-1 as S. album (Flowered Out).
These data showed that all the species of Sedum present on the outdoor green roof
produced viable cuttings. As mentioned in the outdoor experiment section, some species were
not well represented in the cutting or new plant mixes.
76
The data also show that S. album (Flowered Out) was capable of producing viable
cuttings after flowering all summer. The S. album parent plants appeared very tired and weak,
but they still produced viable cuttings. These data demonstrated that S. album still produced
viable cuttings after flowering, even though sedums tend to produce less viable cuttings after the
flower (Snodgrass, 2006.) cuttings could be harvested in the fall, after the parents plants have
flowered all summer, and the cuttings can be expected to root. S. album also has a tendency to
flower itself to death some years (Berghage, Personal Communication, 2009). Because the
cuttings are still viable, S. album plants that flowered vigorously could be cut back, the cutting
spread, and should re-establish the same patch of vegetation the following year.
77
CHAPTER 9: THESIS CONCLUSIONS
Efficacy of Treatment
Compared to control sections, all of the treated sections showed greater densities of new
Sedum plants, despite cutting losses due to drying out, rotting, and being removed by wind and
potentially animals. Therefore the maintenance method put forth in this thesis met its primary
goal of increasing the number of Sedum plants after one growing season.
Comparison of Harvesting Methods
By the measures of new plant density and efficiency, the reel mower was the most
effective tool tested in this experiment. In the same amount of time, the reel mower produced
fewer mangled cuttings and more new plants per square meter than the string trimmer. A reel
mower had the added advantage of not using fossil fuel or electricity, so it was arguably the
“greenest” tool used in this experiment. Considering that green roofs are a “green” technology,
the “greenness” of a reel mower may be important to some green roof owners, installers, and
maintainers.
If cutting quality were the primary concern of the harvester, then hand shears were the
best tools. Cutting quality may be the primary concern if cuttings were to be sold to third party or
if a very small stock of cuttings were available. In either case, the control and discretion provided
by hand shears made them a superior tool. However, hand shears were far too inefficient to be
78
useful on a 400 m2 roof or anything larger. Arguably, they may be too inefficient for use on a
roof half the size of the experimental roof.
Irrigation Regime
The greenhouse experiments demonstrated that irrigation can be applied immediately
after cuttings are harvested and spread. Therefore, irrigation should be included in future
applications of this maintenance method. By irrigation immediately after cuttings are harvested
and spread, a maintenance worker can complete the maintenance in a single trip to the roof. By
not waiting three days to allow cuttings to callous, this form of maintenance is logistically easier
to perform, less time consuming, and therefore less expensive.
Cutting Viability
All the sedum species on this green roof, S. acre (aureum), S. album, S. kamtschaticum,
S. hispanicum, S. rupestre, S. rupestre (Angelina), S. sarmentosum, S. sexangulare, S. spurium
(White Form), S. spurium (John Creech), and S. spurium (Fuldaglut), did produce viable cuttings.
This implies that this form of maintenance can be used on any sedum dominated extensive green
roof. Given that extensive green roofs are considered low maintenance installations (FLL, 2006,
Cantor 2008), this form of maintenance fits well within the maintenance guidelines of the FLL,
which suggests only one or two maintenance visits per year on extensive green roofs.
Limitations of the treatments
79
There was a major limitation in the method used in this experiment. Portions of a roof
with large areas of exposed media were the same areas with the smallest stock of new cuttings.
The same could be said of an entire roof. On a green roof showing very sparse vegetation, the
method used in this experiment would likely perform poorly as a means of increasing plant
coverage. Such a barren roof would need new plants from an outside source like a nursery
because there would be a miniscule stock of cuttings on the roof. There might be a percent area
coverage that acts as the balance point where this method would be effective and where new
plants must be imported from an outside source. Based on personal experience and the results of
this study, the author speculated that this method should be useful for attaining 100 percent
coverage if 60 percent or more the roof is already covered with sedum plants. At a lower percent
cover, outside plants would need to be purchased. This method would also prove ineffective on a
roof where Sedum species, or other easily propagated species are not the dominate form of
vegetation.
Fortunately, the method is easily adaptable to overcome the aforementioned drawback.
Constraining cuttings to the area from which they were harvested was adopted in this experiment
solely to provide quantitative data for comparing treatments. On a roof like the one used in this
experiment, cuttings could be easily harvested from one area where there is an abundance of
available cuttings, like section 3-3, collected in a container, then spread on another area of the
roof where there is an abundance of exposed media like section 1-1. Sections are those shown in
Figure 4-2.
Other Benefits of Treatment
80
Increasing the number of new plants was the goal of this treatment and that goal was met.
However, the author noted there were some other benefits to the form of maintenance proposed in
this thesis. The reel mower and string trimmer were also effective at cutting down weeds.
Depending on the weather conditions, cutting down weeds would kill some of them. This
experiment was conducted during a relatively wet and cool summer, so drought pressure never
became a serious issue, and many of the weeds, most notably horsetail, grew back after being cut
down. There were times when the weeds wilted, but they never reached the permanent wilting
point. It was worth noting that the greatest concentration of horsetail was in Sections 1 and 6,
shown in Figure 4-2. The growing medium in Sections 1 and 6 was about 15 cm (6 in) deep and
the largest areas of exposed media. Other sections of the roof were built with 10 cm (4 in) of
media depth, which means that Sections 1 and 6 retained more relatively water. Even if weeds
were not killed by being cut down, cutting them down at the right time could prevent them from
flowering and reseeding themselves.
In the fall a similar treatment could be used, for the purpose of harvesting cuttings, and
for the purposes of deadheading flowers, dispersing seed, and killing weeds. Robert Cameron, a
Ph. D. student under Dr. Robert Berghage, ran a gasoline powered rotary mower across the roof
during the late summer. He set the blades as high as possible so that they did not cut any of the
Sedum foliage. His goal was cutting down the aforementioned horsetail because it gave the roof a
ragged appearance. Anecdotal evidence suggested that the mower also removed seed pods,
dispersing any Sedum seed (Cameron, Personal Communication., 8/2009). Seed pods may be
considered unsightly by some people because they are brown and twiggy in appearance, so there
is subjective aesthetic value to removing seed pods. If the seed were viable, dispersing it should
lead to more new plants during subsequent growing seasons. However, Kathryn Sanford, another
Ph. D. student under Dr. Robert Berghage, studying Sedum seed viability, noted that the seed she
81
harvested from this experimental roof demonstrated extremely low viability (Personal
Communication., 2/2010). Therefore, caution should be used if dispersing seed to increase plant
density is the primary goal of any future treatment.
Further Research
This experiment raised a number of questions that make excellent topics of further
research. Interesting areas for further research include finding, adapting, or designing tools that
optimized harvesting efficiency; evaluating this method for use in the second or third growing
season of a new green roof plant establishment; determining the viability of Sedum cuttings after
they experience a hard frost; and evaluating how much foliage can be removed from a single
Sedum plant without killing it.
Other tools should certainly be evaluated for harvesting efficiency and cutting quality.
Some untested tools that may perform well in this method are long handled hedge trimmers or
gasoline powered rotary mowers.
Hedge trimmers generally have a long blade, which should effectively harvest a large
quantity of cuttings with a single pass. Hedge trimmers may provide similar cutting quality as
hand shears, because the harvester is provided similar blade control. A long handle will eliminate
the need to bend or kneel while harvesting cuttings, so efficiency should be high and worker
fatigue low. Gasoline powered, long handled hedge trimmers are commercially available from
Stihl ®, makers of power equipment. Stihl long handled hedge models HL100 and HL 100K
feature 20 inch blades and adjustable heads that allow the user to change the angle of the blade,
82
allowing it remain parallel to the surface of the green roof for optimal cutting harvest
(http://www.stihlusa.com/).
Gasoline powered rotary mowers also merit some consideration. A self propelled mower
will cover a lot of area very quickly and will certainly cut back large weeds and old flower stems
as noted. Research must determine if a rotary mower will actually produce viable cuttings. It is
possible that the powerful mower will simply damage cuttings and parent plants beyond viability.
Ideally any harvesting tool will combine a cutting mechanism and a collection
mechanism in one tool. This will allow a maintenance person to harvest cuttings from densely
vegetated areas of a roof and spread the cuttings on poorly vegetated areas without adding the
extra step of using a vacuum to collect the cuttings. A potential addition to a reel mower would
be a catcher designed to catch grass clippings as they are dispersed. A grass catcher may prove
effective for collecting Sedum cuttings also. Grass catchers are commercially available for many
manufacturers’ models of reel mowers.
A new tool could be designed that combines a hedge trimmer similar to the Stihl ®
product mentioned with a backpack vacuum or a cutting collection bin. The vacuum engine
could be reversed so the vacuum also acts as blower to easily disperse cuttings. If a pressurized
hose could be attached device, then a maintenance worker could harvest, collect, disperse, and
water cuttings in a single pass, saving time and labor costs. Such a tool has potential to produce
new intellectual property.
The maintenance method put forth in this thesis could potentially be used as a standard
operating procedure during the second or third year of a new green roof installation, or in
83
subsequent years if a roof experiences high plant mortality. The treatment has potential to
encourage one hundred percent plant coverage faster and at a lower cost than increasing planting
densities. If applied in the second growing season, pruning the previous year’s plants may
discourage flowering and encourage vegetative growth for a second season. The author notes that
plugs of S. album, S. kamtschaticum, S. sexangulare, and S. spurium (John Creech) from Emory
Knoll Farms Inc., planted in constructed boxes with 10 cm (4 in) of green roof media in his
backyard in May 2009 did not flower. Also, no plugs out of about 200 of the same species
flowered in a greenhouse on the University Park campus between April 2009 and February 2010.
In addition to potentially encouraging vegetative growth, pruning in the second year should yield
viable cuttings that will grow in the exposed media between the plugs installed the previous year.
A concern is that the plants may not be established well enough to withstand the stress of being
pruned during their second year on the roof.
An interesting study would compare the percent area coverage to three planting
treatments: Standard density, 23 plugs per m2 (2 plugs/ft2), and no treatment the second year;
double the standard density, 46 plugs per m2 (4 plugs/ft2) and no treatment the second year; and
23 plugs per m2 with treatment the second year. Plant area coverage would be measured upon
planting, at the end of the first growing season, at the beginning and end of the second growing
season, and at the beginning of a third growing season. The roof maintained by harvesting and
spreading cuttings may show greater plant coverage than the control. Labor and plant costs
should also be considered during this experiment.
A third interesting research topic is the viability of Sedum cuttings after a hard frost. The
author spread cuttings of S. acre and S. album in late October. The cuttings received a hard frost
and were even blanketed in an early snow. It appears that they survived the cold and began
84
rooting, but the experiment was not controlled. If it can be demonstrated that certain species
produce cuttings that are viable after a hard frost, then cuttings could be harvested and spread
earlier in the spring, before the last day of frost occurs. Treatment in this experiment occurred in
late May and early June, well after the last frost of the spring. If the treatment were applied
earlier in the season, fewer established plants would be flowering, so theoretically the cuttings
collected will be more viable. Also, the earlier cuttings are spread, the longer they have to
establish strong roots in their first growing season. Additionally, if cuttings can be harvested and
spread before the last frost, the maintenance person has more flexibility to perform this
maintenance procedure. Testing for cuttings survival after frost would also show that this for of
maintenance could also be used in the fall, and some cuttings could survive, root, and over winter.
While testing the viability of cuttings after a hard frost, the survivability of established
plants that are pruned then frosted should also be evaluated. This kind of treatment would be
ineffective if the parent plants die, even if new cuttings survive.
Finally, it may be worth evaluating what percentage of a Sedum plant can be taken as
cuttings without killing the plant. The author removed nearly all of the foliage on S. album, S.
kamtschaticum, S. sexangulare, S. spurium (John Creech) grown in a greenhouse for 6 months.
Plants were left un-watered for weeks with day/night temperatures around 27°/16° C (80°/60° F)
and had the added stress of a mealy bug population. Once watering began again, the plants
responded by sending out new shoots, even plants that appeared completely dead began to grow
again. This evidence is anecdotal because this was not a controlled experiment, but it suggests
that established Sedum plants can survive extensive abuse.
85
Practical Applications of this Research
The thesis showed that an green roof planted with sedum species provided cuttings
capable of increasing the number of individual sedum plants where the roof showed areas of
exposed media. Additionally, a method of harvesting and collecting cuttings with simple lawn
maintenance tools such as a reel mower and a string trimmer. The results have very practical
applications for green roof maintenance.
Due to poor maintenance, green roofs commercially installed green roofs will show
decreased plant coverage over time. So, some form of maintenance will be required to increased
and maintain the coverage of desired plants. As show in Table 1-1, extensive green roofs are
designed to be relatively low cost and low maintenance structures, so minimizing inputs of labor,
new plants, and fertilizers during maintenance is import for minimizing long term costs. The
form of maintenance put forth in this thesis requires relatively few man hours, and if combined
with some hand weeding would provide excellent Sedum coverage. Excellent Sedum coverage is
important to reducing weed populations because weed seeds tend to germinate only when media
is exposed.
For the best results, this form of maintenance should be used in the spring, before the
parent plants begin flowering. This will yield more viable cuttings, and provide the cuttings the
longest possible growing season in which to root and prepare to over-winter. If this form of
maintenance is used in the fall, some propagations would be successful, though fewer cuttings
should be expected to root and survive the winter as compared to cuttings spread in the spring and
summer. A study at the Michigan State University found that 81% of plants survived over winter
on green roof platforms when planted in the spring, while just 23% survived when planted in the
86
fall (Getter, et. al., 2007). All this information suggested that an existing roof should see the best
results when cuttings are harvested and spread in the late spring or early summer, and not the fall.
An advantage to a fall treatment is that flower heads can be removed, to give the roof a cleaner
appearance. Additionally, a second maintenance visit in the fall allows the maintenance workers
to fix any problems that may have occurred over the summer.
87
WORKS CITED
Berghage, R. D., Beattie, D., Jarrett, A. R., & Rezaei, F. (2007). Green Roof Water Use. National
Decentralized Water Resources Capacity Project,
Berghage, R. D., Beattie, D., & Negassi, A. (2007). Green Roof Capacity to Neutralize Acid
Runoff. National Decentralized Water Resources Capacity Project,
Cameron, R. D. (2009). Personal communication
Cantor, S. L. (2008). Green roofs in sustainable landscape design. New York, NY: W.W. Norton
and Company, Inc.
Compton, J., S., & Whitlow, T. H. (2006). A zero discharge roof system and species selection to
optimize evapotransiration and water retention. Greening Rooftops for Sustainable
Communities, Boston.
Dillen Products/Myers Industries, I. (2010). Dillen products - molded plastic products for
horticulture. Retrieved 3/4/2009www.dillen.com
Emilsson, T. (2008). Vegetation Development on Extensive Vegetated Green Roof: Influence of
Substrate Composition, Establishment Method, and Species Mix. Ecological Engineering,
33, 265.
Emory Knoll Farms, I. (2007). Green roof plants. Retrieved 3/4/2010, 2010, from
Getter, K., & Rowe, D. B. (2007). Effect of substrate depth and planting season on sedum plug
establishment on extensive green roofs. Greening Rooftops for Sustainable Communities,
Minneapolis.
88
Ingram, D. S., Vince-Prue, D., & Gregory, P. J. (2002). Science and the garden: The scientific
basis of practical horticulture. Oxford, England: Blackwell Publishing.
Jarrett, A. R., Hunt, W. F., & Berghage, R. D. (2007). Annual and Individual-Storm Green roof
Stormwater ResponseModels. National Decentralized Water Resources Capacity Project,
Montrusso, M., A., Rowe, D. B., & Rugh, C. L. (2005). Establishment and Persistence of Sedum
Species and Native Taxa for Green Roof Applications. Hortscience, 40(2), 391-396.
Rowe, D. B., Rugh, C. L., & Durhman, A. K. (2006). Assessment of substrate depth and
composition of green roof plant performance. Greening Rooftops for Sustainable
Communities, Boston.
Sanford, K. L. (February 2010). Personal communication
Schmidt, M. (2006) The evapotranspiration of greened roofs and facades. Greening Rooftops for
Sustainable Communities, Boston.
Snodgrass, E. (February 2010). Personal Communication.
Snodgrass, E., & Snodgrass, L. (2005). Green roof plants: A resource and planting guide.
Portland, OR: Timber Press.
Stephenson, R. (1994). Sedum: Cultivated stonecrop. Portland, OR: Timber Press.
Stihl Incorporated. (2010). Stihl. Retrieved 2/24/2009, 2010, from http://www.stihlusa.com/
Syrett, B. (11/24/2009). Weather observatory, department of meteorology, the Pennsylvania State
University
89
University Park Campus Maps with Building Indices (revised April 2009). Retrieved 2/23/2010,
2010, from http://www.campusmaps.psu.edu/print/pdf
Vidmar, J., Kelley, K., & Berghage, R. D. (2007). Background, Educational, and Promotional
Materials for Green Roofs: A series of Articles to Promote Understanding of the Benefits of
using green roofs. National Decentralized Water Resources Capacity Project,
90
Appendix A
A Collection of Vegetation Maps of the Experimental Green Roof
Maps were drawn between May 18th
to 22nd
2009
98
Appendix B:
A Collection of Vegetation Maps of the Experimental Green Roof from 2007
Maps were drawn by Kristen Casale
110
rooflite® extensive mc Specifications
rooflite® extensive mc is a growing medium for extensive green roofs in multi-course
construction. The material is a mixture of mineral light weight aggregates like HydRocks® and
premium organic components complying with the following requirements:
Particle Size Distribution
Proportion of silting components (d < 0.063 mm) Mass % < 15
Density Measurements
Bulk Density (dry weight basis) g/cm3
0.70 - 0.85
Bulk Density (dry weight basis) lb/ft3
44 - 53
Bulk Density (at max. water-holding capacity) g/cm3
1.15 - 1.35
Bulk Density (at max. water-holding capacity) lb/ft3
72 - 85
Water/Air Measurements
Total Pore Volume Vol. % > 65
Maximum water-holding capacity Vol. % 35 - 65
Air-filled porosity at max water-holding capacity Vol. % > 10
Water permeability (saturated hydraulic conductivity) cm/sec 0.001 – 0.12
Water permeability (saturated hydraulic conductivity) in/min 0.024 – 2.83
pH and Salt Content
pH (in CaCl2) 6.0 - 8.5
Soluble salts (water extract) g/L < 3.5
Soluble salts (gypsum extract) g/L < 2.5
Organic Measurements
Organic matter content g/L < 65
Nutrients
Phosphorus, P205 (CAL) mg/L < 200
Potassium, K2O (CAL) mg/L < 700
Magnesium, Mg (CaCl2) mg/L < 200
Nitrate + Ammonium (CaCl2) mg/L < 80
Supplier: Skyland USA LLC, phone: 1.877.268.0017, www.skylandusa.us All values are based on compacted materials according to laboratory standards and testing methods defined by the Forschungsgesellschaft Landschaftsentwicklung Landschaftsbau e.V. (FLL) Landscape Development and Landscaping Research Society, Guidelines for the Planning Construction and Maintenance of Green-Roofing, Green Roofing Guideline, 2008
Minitab® Statistical Software Minitab Inc
New Plant Data Statistical Output
One-way ANOVA: Plants/sq meter versus Treatment
Source DF SS HS F P Treatment 3 257836 85945 14.99 0.000 Error 98 561906 5734 Total 101 819742
S = 75.72 R-Sq = 31. 45% R-Sq{adj) 29.35%
Individual 95% CIs For Hean Based on Pooled StDev
Level N Hean StDev -+--- - --+------- -+- ------+-Control 31 23.61 25.99 (-----* Hand Shear 22 120.85 84.56 (__ * nn) Reel Hower 21 158.90 116.48 ---*-----) String Trimmer 28 96.49 66.71 ---*-
--------+- ---+--- --+-------
o 50 100 150
Pooled StDev 75.72
One-way ANOVA: Plants/sq meter versus Roof Section
Source DF SS HS F P Roof Section 4 15539 3885 0.47 0.759 Error 97 804203 8291 Total 101 819742
S = 91. 05 R-Sq 1.90% R-Sq(adj) 0.00%
Individual 95% CIs For Mean Based on Pooled stDev
Level N Mean StDev +------- -+- -----+.- -+-1 27 102.86 76.25 (n __ n*nnn) 2 33 101. 77 109.98 (-----*--- )
3 29 74.61 79.62 ( - *------){_n_4 9 84.92 85.68 *--------
5 4 91.50 99.63' * +- -+---- +- ----- ----+---_.o 50 100 150
Pooled StDev 91. 05
Two-Sample T-Test and CI: Plants/sq meter_1, TreatmenC1
Two-sample T for Plants/sq meter_1
SE Treatment_1 N Mean StDev Mean Reel Mower 21 159 116 25 String Trinuner 28 96.5 66.7 13
Difference = mu (Reel Mower) - mu ( Trimmer) Estimate for difference: 62.4 95% CI for difference: (4.4, 120.4) T-Test of difference 0 (vs not =): T-Value 2.20 P-Value 0.036 DF 29
Cutting Data Statistical Output
One-way ANOVA: Length versus Treatment
Source Treatment Error Total
S = 0.8542
DF SS MS F P 2 52.938 26.469 36.27 0.000
1498 1093.097 0.730 1500 1146.035
R-Sq 4.62% R-Sq(adj) 4.49%
Individual 95% CIs For Mean Based on Pooled StDev
Level N Mean StDev ----+---------+--------HS 551 1.7523 0.8696 RM 500 1. 3180 0.7328 -*---- ) W'vJ 450 1.6511 0.9547
-_._+- - - - - - -+
+-.--------+-(----*
*----) -----+-
1. 35 1. 50 1. 65 1. 80
Pooled StDev 0.8542
One-way ANOVA: Flowering versus Treatment
Source Treatment Error Total
S = 0.4496
DF SS MS F P 2 2.996 1. 498 7.41 0.001
1499 303.044 0.202 1501 306.040
R-Sq = 0.98% R-Sq(adj) 0.85%
Individual 95% CIs
Level HS RM \1M
N 552 500 450
Mean 0.2264 0.3160 0.3222
StDev 0.4189 0.4654 0.4678
Pooled StDev --+- ------+-
---*-------) (
0.200 -_._-----+---
0.250
Pooled StDev 0.4496
For Mean Based on
--+---------+
--*-------) (-- * -------)
--+------0.300 0.350
One-way ANOVA: Condition versus Treatment
Source DF SS Treatment 2 7.6983 Error 1499 116.8057 Total 1501 124.5040
S 0.2791 R-Sq = 6.18%
Level N Mean StDev HS 552 0.0127 0.1120 RM 500 0.0900 0.2865 WW 450 0.1889 0.3919
Pooled StDev 0.2791
MS F P 3.8492 49.40 0.000 0.0779
R-Sq(adj) 6.06%
Individual 95% CIs For Mean Based on pooled StDev --+---------+-- --+------(_nL_ )
(---*(_ *nn)
-+-------- +---------+- -----+0.000 0.060 0.120 0.180