Plant Syst Evol (2012) 298:931–936
DOI 10.1007/s00606-012-0602-z
ORIGINAL ARTICLE
Floral functional structure, sexual phases, flower visitors
and aspects of breeding system in Roella ciliata (Campanulaceae)
in a fragmented habitat
Christopher N. Cupido • Leanne J. Nelson
Received: 7 October 2011 / Accepted: 24 January 2012 / Published online: 16 March 2012
Ó Springer-Verlag 2012
Abstract Continued habitat destruction in lowland areas
of the Western Cape, South Africa, due to urbanization has
reduced Roella ciliata to a few isolated populations and it
is unknown whether this endemic shrublet will persist.
Here, we examine the reproductive biology of a small
population of R. ciliata occurring in a residential nature
reserve 30 km east of Cape Town to provide baseline
information for future studies. Data were recorded on the
functional structure of the flower, duration of the morphological gender phases, floral visitor profile and the
relationship between breeding system and habitat. Observations revealed that this species possesses reproductive
characteristics typical of Campanulaceae such as dichogamy, protandry and secondary pollen presentation via
pollen collecting hairs. Floral longevity ranges between 1
and 3 days. Anthesis occurs in synchronicity at 0700 hours
and individual flowers close between 0930 and 1500 hours
daily. The staminate phase lasts between 2 and 29 h and is
significantly influenced by the removal of pollen by floral
visitors, while the pistillate phase lasts between 21 and
69 h. R. ciliata maintains a generalized, entomophilous
pollination system although floral visitors are generally
scarce. Despite some intraspecific variation, the pollen/
ovule ratio indicates that R. ciliata has a facultatively
autogamous breeding system, in contrast to its floral attributes which might favour xenogamy. This reproductive
mechanism may ensure survival under these habitat
C. N. Cupido (&)
South African National Biodiversity Institute, Compton
Herbarium, Private bag X7, Claremont 7735, South Africa
e-mail: c.cupido@sanbi.org.za
L. J. Nelson
Biodiversity Program, Queensland Museum, South Brisbane,
Queensland, Australia
conditions and could represent a shift from outcrossing to
selfing at the cost of reducing genetic variation over time.
Keywords Campanulaceae Cape Floristic Region
Habitat fragmentation Roella ciliata Reproductive
biology
Introduction
The continued existence of an organism depends on its
ability to regenerate itself in a stable as well as a changing
natural environment. Failure of its reproductive process to
adapt to change could lead to its extinction. The use of
anthropogenic interventions to reduce the risk of extinction
or loss requires an understanding of the organism’s
reproductive biology. For flowering plants, this includes
floral biology, seed dispersal, germination ability, pollination and life span.
Studies on the reproductive biology of the Campanulaceae have focused almost entirely on Campanula L. (e.g.
Richardson and Stephenson 1989; Evanhoe and Galloway
2002; Giblin 2005) and Wahlenbergia Schrad. ex Roth
(e.g. Lloyd and Yates 1982; Anderson et al. 2000). Evidence from these studies suggests that the reproductive
biology of the family is complex and detailed studies of
individual taxa are required to understand the reproductive
characteristics (Anderson et al. 2000). Little is known
about the reproductive biology of the South African bellflower and how finely tuned it is to the diverse flower
structure, pollinators and habitat, and its significance in
systematics (Cupido 2009).
The South African endemic genus, Roella L. is largely
confined to the Cape Floristic Region (CFR) with one
species occurring along the east coast from the Eastern
123
932
C. N. Cupido, L. J. Nelson
Cape to KwaZulu-Natal. Species of Roella are summer
flowering dwarf shrubs or perennial herbs with an erect to
prostrate habit. In a checklist of Campanulaceae, Lammers
(2007) followed the treatment by Goldblatt and Manning
(2000) and Welman and Cupido (2003) in recognizing 20
species. More recently, Cupido (2010) described a new
species of Roella from Hermanus on the east coast of the
south-western Cape. Fire is critical in the growth and survival of Roella species in the CFR. Plants are most abundant and flower profusely within the first 3 years after fire.
When fire is absent for extended periods, plants begin to
disappear from the landscape.
In the midst of ongoing habitat destruction in lowland
areas of the Western Cape, South Africa, due to urbanization, R. ciliata L. has become restricted to a few small
populations which occur in vacant residential plots, small
nature reserves and farmland. Data on the reproductive
biology of this species are limited, and it is uncertain
whether these populations will persist in this fragmented
landscape where fire is often absent. Through developing a
detailed understanding of the reproductive biology of the
species, it is hoped that R. ciliata may be conserved despite
ongoing transformation of its natural habitat.
To this end in order to provide baseline information for
future studies on the effects of habitat fragmentation on
the reproductive biology of R. ciliata, we address the
following:
1.
2.
3.
4.
Functional structure in bud, at anthesis and senescence.
Duration of the morphological gender phases of the
flower.
The relationship between breeding system and habitat.
Visitor profile.
et al. 2006). Over 250 indigenous plant species occur in the
reserve amongst a considerable number of exotic species.
The population of the study species, R. ciliata, occurs
amongst grass on a gentle slope near the eastern border of
Bracken in granite-derived sandy to loamy soils. Adams
(2010) estimated a population size in excess of 150 individuals and reported that the last recorded fire occurred in
2006. R. ciliata is an erect or sprawling dwarf shrub up to
45 cm high. Its leaves are linear, ciliate with axillary
clusters of smaller leaves present. Its bell-shaped flowers
are blue with a dark ring or spot on the lobes, often edged
with white, and borne solitary on the tips of branches. This
species is endemic to the CFR, where it ranges from the
Cape Peninsula to Caledon on slopes or flats. A voucher of
the species was deposited at the Compton Herbarium
(NBG).
Field study
The population was initially visited in early January 2010
to confirm historical records of R. ciliata in the Brackenfell
area. Subsequent to this visit, the population was monitored
from 29 August 2010 to observe the onset of flowering. In
early October, 15 plants with five or more buds were
selected and tagged for the study. Over the ensuing weeks,
the population was visited at different times of the day to
ascertain the most suitable period for data collection. This
became necessary as closing flowers were noted during
morning and late afternoon visits, irrespective of the degree
of sunlight, creating difficulty in distinguishing newly
opening from senescing flowers. Data were collected during the period 0700–1500 hours.
Phenology and floral longevity
Material and methods
Site and species studied
Field studies were conducted at the Bracken Nature
Reserve, which is located in the residential area of Brackenfell, approximately 30 km east of Cape Town. Prior to
its proclamation in 1976, a portion of the land was utilized
as a granite quarry and later converted to a land-fill site
which was closed in 2000. The area is currently undergoing
rehabilitation by the City of Cape Town and forms part of
the 39-ha reserve which conserves two isolated remnants of
vegetation. The smaller of the two, a granite outcrop,
Perdekop, is approximately 4 ha in extent while the
remaining larger area is generally referred to as Bracken.
Floristically, the reserve falls within the CFR and its vegetation is described as the critically endangered, Cape Flats
Sand Fynbos and Swartland Granite Renosterveld (Rebelo
123
The flowering period was defined as the period between the
onset of anthesis of the first and senescence of the last
flower. Anthesis was recorded as the time the petals separated to expose the style to pollinators, following the
definition of Evanhoe and Galloway (2002). The scoring of
senescence at two stages, wilt and complete senescence,
proved difficult in practice since the turgidity of the style
could not be ascertained without damaging the floral parts.
Therefore, flowers were considered as senesced when the
wilted petals had completely enclosed the style, rendering
it inaccessible to floral visitors. Floral longevity was
determined by randomly selecting three flowers per plant,
and recording the date/time of anthesis and senescence.
Flower functional structure
The stages of flower development and function were
studied in bud immediately prior to anthesis, during
Floral functional structure, sexual phases, flower visitors and breeding system in Roella ciliata
anthesis, and when 1, 2 and 3 days old. Flowers were
collected and fixed in 70% ethanol.
Pollen/ovule ratio
To determine the pollen/ovule (P/O) ratio of R. ciliata,
three buds on 15 plants (N = 15) were randomly collected
and fixed in 70% ethanol. Pollen and ovule quantities were
estimated using a modification of the technique described
by Cruden (1977). The number of pollen grains per bud
was calculated by suspending mature anthers in 0.5 ml of
water. Anthers were ruptured and shaken to release the
pollen grains which were counted with a Neubauer haemacytometer. The number of ovules per ovary was determined by counting under a dissecting microscope. The P/O
ratio was calculated by dividing the total pollen count by
the total ovule count.
Duration of the morphological gender phases
The morphological gender of one randomly selected flower
on 15 plants was tracked every hour between 0700 and
1500 hours under consistent environmental conditions to
determine the duration of the staminate and pistillate
phases. The procedure was later replicated in 14 additional
plants increasing the sample size to 29. Flowers were
considered staminate from the time of anthesis to the full
expansion of the stigmatic lobes and pistillate from the
time of stigmatic lobe expansion to complete senescence.
The removal of [50% pollen by floral visitors during the
staminate phase was estimated by eye and recorded as
considerable. A one-way analysis of variance was performed to test the effects of considerable pollen removal on
the duration of the staminate phase. Statistical analyses
were conducted using Minitab 16 software (Minitab 2010).
Floral visitors
All plants in our sample were observed simultaneously for
floral visitors from 0700 to 1500 hours, by which time
most flowers had closed. Floral visitor diversity and
behaviour were recorded at 10-min intervals. Several
individuals were captured for identification.
Results
Flower structure and function
When the flower is in bud, the corolla is enclosed by five
broad leathery calyx lobes interlocked by their marginal
hairs and cilia. The apices of the calyx lobes taper to a
sharp point and extend beyond the corolla. The base of the
933
bud is surrounded by rigid leaf-like bracts. As the corolla
elongates and the calyx lobes begin to separate, its exposed
surface is protected by hispid hairs. Prior to anthesis, the
stamens occupy three-quarters of the space inside the
closed corolla. The stamens extend beyond and completely
surround a hollow style. The anthers enclose the portion of
the style with pollen-collecting hairs. The lower portion of
the style is covered by hispid hairs and is surrounded by the
filaments. Its upper flattened portion forms the unexpanded
stigmatic lobes with a gland on each side of its base.
Prior to and at anthesis, the flower is in the morphological male phase (Fig. 1a). Flowers are actinomorphic
and the corolla is campanulate. The corolla comprises a
short tube with an inner ring of hairs near its base and five
longer lobes with hairs on the midrib and at each sinus. The
corolla lobes are alternate with the calyx and the stamens.
The five stamens are inserted at the upper part of the
hypanthium between the corolla tube and the nectariferous
disc. The filament bases are arching and dilated with hairy
shoulders. They form a dome over a nectariferous disc by
interlocking the shoulder hairs with those on the tube base
and lower portion of the style. Inside the bud, the anthers
dehisce introrsely by longitudinal slits and pollen is shed
and deposited onto collecting hairs. As the style extends up
through the anthers, pollen is brushed up and presented to
potential insect visitors as a ‘club’. Soon after shedding
pollen, the anthers wither and remain at the lower part of
the flower. The inferior ovary is two-locular with numerous
ovules on axile placentae. The top of the ovary is covered
by a plug with a nectariferous disc.
The onset of the female phase occurs when the stigmatic
lobes separate and the pollen-collecting hairs retract
(Fig. 1b). At this stage the flower is ready for pollination.
Occasionally pollen remains present below the separated
lobes and may fall to the bottom of the flower as the collecting hairs retract. This phase is interrupted when the
petals wilt and enclose the style while the stigmatic lobes
remain expanded (Fig. 1c).
Phenology and floral longevity
The population of approximately 150 individuals flowered
from late October to late January (mid-spring to midsummer), and was considered to be at its optimal growth
period due to recent fire. Each day, floral anthesis occurred
in synchronicity across the population at 0700 hours and
closing took place between 0930 and 1500 hours. The
process of floral senescence occurred during the late
afternoon and/or overnight because all flowers had completely senesced by 0700 hours when observations began
the following day. Flowers that had not wilted by
1500 hours the day before but had wilted by 0700 hours
the following day were scored as 1 day. The longevity of
123
934
C. N. Cupido, L. J. Nelson
Fig. 1 Stages of floral development: a morphological male phase, b morphological female phase, c complete senescence
individual flowers ranged between 1 and 3 days with the
majority lasting 2 days. Early senescence (\3 days)
appears to be triggered by considerable pollen removal
([50%) within the first few hours after anthesis.
The duration of the morphological male phase across our
sample population (N = 29) varied between two and 29 h
(14.8 ± 11.3 h; mean ± SD). The majority of flowers had
a considerable amount of pollen removed by floral visitors
within 3 h of anthesis. These same flowers had a significantly lower mean staminate duration (11.04 ± 10.6 h)
than flowers which experienced little to no pollen removal
(21.6 ± 9.1 h; F (1, 27) = 6.83, p = 0.014). Onset of the
pistillate phase generally occurred within a few hours after
pollen dissemination; however, some flowers reached
female onset the following day. The pistillate duration for
all flowers ranged between 21 and 69 h (34.1 ± 12.7 h)
before complete senescence.
Pollen/ovule ratio
There was considerable intraspecific variation in pollen and
ovule quantities within the population (Fig. 2). Pollen grain
and ovule quantities per bud ranged from 6,250 to 86,250
(N = 15, 32,522 ± 1,654) and 116 to 462 (N = 15,
239 ± 58), respectively. Individual P/O ratios show corresponding variations, ranging from 32 to 285. Based on
the mean pollen and ovule counts across the population, the
P/O ratio of R. ciliata is estimated at 136 and falls within
the range of facultative autogamy (Cruden 1977). However, a few buds had relatively low P/O ratios approaching
those characteristic of obligate autogamy (Cruden 1977).
Floral visitors
Floral visitors were observed from the time the flowers
opened at 0700 hours. All flowers were entomophilous,
visited by insect species from multiple orders. Honey bees
123
300
Pollen-ovule ratio
Duration of the morphological gender phases
350
250
200
150
100
50
0
Population (N=15)
Fig. 2 Within-population variation in P/O ratio
(Apis mellifera capensis) were the most abundant and
active visitors; some individuals were observed visiting up
to ten flowers on the same plant in 1 min. Monkey beetles
(Peritrichia sp.) were common, generally remaining on the
same plant for long periods ([20 min) and possibly overnighting in closed flowers. Flies (Diptera) and blister beetles (Ceroctis capensis) were relatively infrequent and were
observed on fewer than five occasions. Floral visitors were
observed foraging for nectar, incidentally collecting pollen
on their bodies through contact with the pollen collecting
hairs. Foraging activity occurred primarily in the morning
hours (0800–1100 hours) prior to the flowers closing.
Discussion
Field observations of R. ciliata show that flowering extends
over a period of 12 weeks from October to January, after the
main flowering period for most bulbs and annuals in this
vegetation type. Flower colour and size, and plant height in
the population are relatively uniform. Individual flowers
begin closing daily between 2 and 8 h after opening
Floral functional structure, sexual phases, flower visitors and breeding system in Roella ciliata
irrespective of weather conditions and have an average
longevity of 2 days. R. ciliata has the typical functional
structure known elsewhere in the Campanulaceae: flowers
are dichogamous whereby the order of presentation of the
sexual phases is protandrous and pollen is presented secondarily via pollen-collecting hairs. At the end of summer,
plants become moribund while bearing unopened capsules.
New vegetative growth appears in late winter.
Our study demonstrated that the duration of the morphological male phase varies depending on the amount of
pollen removed by floral visitors. We found a relationship
between high levels of pollen removal and shorter staminate phases. On average, flowers which had a considerable
amount of pollen removed had approximately half the
staminate length of those which had little to no pollen
removed. The dissemination of pollen is believed to trigger
the onset of the morphological female phase, thereby
making pollinator activity an important factor in the longevity of R. ciliata’s flowers. These data are consistent
with several studies of protandrous species in the Campanulaceae family (Devlin and Stephenson 1984; Evanhoe
and Galloway 2002; Richardson and Stephenson 1989,
etc.) which show that gender phase durations are plastic in
response to pollen removal. The amount of pollen removed
by floral visitors may not be the only factor controlling the
duration of the gender phases. Studies in Campanula afra
Cav. (Nyman 1992) have shown that stimulation of pollencollecting hairs and the frequency thereof reduce the length
of the male phase. The effects of pollen removal and
stimulation of the pollen-collecting hairs on the length of
the staminate phase in R. ciliata needs to be determined, as
it is unclear whether they act independently or in combination. While our study did not examine the effects of
pollen deposition on the length of the pistillate phase, it is
plausible that the amount of pollen received by stigmas or
pollen hair stimulation influences the longevity of the
female phase and time of senescence.
Dichogamy and the functioning of pollen-collecting
hairs are among the mechanisms known to restrict selfpollination (Proctor and Yeo 1972; Nyman 1993) and
promote out-breeding in plants. Contrary to possessing
outcrossing attributes, the mean P/O ratio suggests that
R. ciliata is a facultative inbreeder characteristic of disturbed or early successional habitats according to the categories of Cruden (1977). There was some intraspecific
variation in the pollen and ovule numbers across our
sample population (Fig. 2). This trend has been identified
within individuals and among populations (Cruden 2000).
Such variation could be associated with levels of selfcompatibility among individuals as we noted that some P/O
ratios were approaching obligate autogamy. Cruden’s
generalizations are not always consistent with P/O ratios of
known breeding systems (Lord 1980; Arnold 1982; Böcher
935
and Phillips 1985; Anderson et al. 2001). Factors which
may influence variation in P/O ratios such as types/efficiency of pollen vector, differences among sets of stamens,
intra-inflorescence variability and flower type (unisexual or
bisexual) should be taken into consideration when calculated (Preston 1986).
Field observations reveal that R. ciliata maintains a
generalized, entomophilous pollination system. Honey bees
and monkey beetles are considered the primary pollinators
due to their relative frequency while blister beetles and flies
are thought to play a minor role as pollen vectors. Actinomorphy and an open floral design may facilitate generalization by enabling morphologically diverse taxa to fit
and access rewards. Flower colour, particularly the dark
spots on the petal lobes (beetle marks) may attract monkey
beetles as demonstrated by Van Kleunen et al. (2007) for
members of the Iridaceae. The capacity to exploit a range
of nonspecific pollinators may benefit this species because
it occurs in degraded areas where insect activity may be
unreliable. In the absence of suitable pollinators, the progressive flower closure by midday could be a mechanism to
prevent access by illegitimate visitors or to promote
autonomous self-pollination despite being protandrous and
possessing floral attributes which favour outcrossing.
Direct habitat loss has reduced R. ciliata to a few isolated
populations and it is unknown how the local pollinating
fauna has been affected. Changes to the abundance and
composition of insect pollinators can have negative impacts
on plant reproductive success (Jennerston 1988; Aizen and
Feinsinger 1994; Steffan-Dewenter and Tscharntke 1999;
Cunningham 2000; Donaldson et al. 2002). Ecological
factors which may influence plant sensitivity to habitat
disturbance are degree of pollinator specialization and
breeding system. Self-compatibility and a generalized pollination system may provide some resilience in a changing
environment (Johnson and Steiner 2000; Bond 1994; but
see Aguilar et al. 2006). It is also expected that insects
visiting surrounding home gardens may be reciprocally
available to the flowers of R. ciliata. However, more data on
the pollination ecology of this endemic shrublet are needed,
particularly in relation to ascertaining pollinator efficiency,
pollination rates and reproductive output.
The isolated and fragmented habitat, disturbance caused
by recent fire and unreliability of floral visitors to deliver
outcross pollen might meet the conditions for these plants
to shift to an autogamous mating system. The advantage of
inbreeding under these conditions is that it promotes
reproductive assurance albeit at the cost of reducing
genetic variability over time.
Acknowledgments We are grateful to the following individuals for
contributing to the study. Tshepo Mamabolo, Manager of Bracken
Nature Reserve, for the research permit and assistance locating the
123
936
plant population, Prof. Jeremy Klaasen and Danielle Cupido, Medical
Bioscience Department at the University of the Western Cape, for
assistance in the pollen counting, Dr Jonathan Collville, Applied
Biodiversity Research Division at the South African National Biodiversity Institute, who kindly identified the insects, and Ursula
Cupido for general field assistance.
References
Adams R (2010) An evaluation of the distribution and status of Roella
ciliata (Blue-eyed Roella) within the central district of the City
of Cape Town. Report to the Cape Peninsula University of
Technology
Aguilar R, Ashworth L, Galetto L, Aizen MA (2006) Plant
reproductive susceptibility to habitat fragmentation: review and
synthesis through a meta-analysis. Ecol Lett 9:968–980
Aizen MA, Feinsinger P (1994) Habitat fragmentation, native insect
pollinators and feral honey bees in Argentine ‘‘Chaco Serrano’’.
Ecol Appl 4:378–392
Anderson GJ, Bernardello G, Lopez P, Crawford DJ, Stuessy TF
(2000) Reproductive biology of Wahlenbergia (Campanulaceae)
endemic to Robinson Crusoe Island. Plant Syst Evol 223:109–
123
Anderson GJ, Bernardello G, Stuessy TF, Grawford DJ (2001)
Breeding system and pollination of selected plants endemic to
Juan Fernández Islands. Am J Bot 88:220–233
Arnold RM (1982) Pollination, predation, and seed set in Linaria
vulgaris (Scrophulariaceae). Am Midl Nat 107:360–369
Böcher J, Phillips M (1985) Aspects of the reproductive ecology of
Mimulus repens (Scrophulariaceae) at Lake Ellesmere, Canterbury, New Zealand. N Z J Bot 23:141–149
Bond W (1994) Do mutualisms matter? Assessing the impact of
pollinator and disperser disruption on plant extinction. Philos
Trans Biol Sci 344:83–90
Cruden RW (1977) Pollen ovule ratios: a conservative indicator of
breeding systems in flowering plants. Evolution 31:32–46
Cruden RW (2000) Pollen grains: why so many? Plant Syst Evol
222:143–165
Cunningham SA (2000) Depressed pollination in habitat fragments
causes low fruit set. Proc R Soc B Biol Sci 267:1149–1152
Cupido CN (2009) Systematic studies of the South African Campanulaceae sensu stricto with an emphasis on generic delimitation.
PhD thesis, University of Cape Town, Cape Town
Cupido CN (2010) A new species of Roella (Campanulaceae) from
Western Cape, South Africa. Edinb J Bot 67:425–430
Devlin B, Stephenson AG (1984) Factors that influence the duration
of the staminate and pistillate phases of Lobelia cardinalis
flowers. Bot Gaz 145:323–328
Donaldson J, Nänni I, Zachariades C, Kemper J (2002) Effects of
habitat fragmentation on pollinator diversity and plant reproductive success in Renosterveld shrublands of South Africa.
Conserv Biol 16:1267–1276
123
View publication stats
C. N. Cupido, L. J. Nelson
Evanhoe L, Galloway LF (2002) Floral longevity in Campanula
americana (Campanulaceae): a comparison of morphological
and functional gender phases. Am J Bot 89:587–591
Giblin DE (2005) Variation in floral longevity between populations of
Campanula rotundifolia (Campanulaceae) in response to fitness
accrual rate manipulation. Am J Bot 92:1714–1722
Goldblatt P, Manning JC (2000) Cape plants. A conspectus of the
Cape flora of South Africa. Strelitzia 9. National Botanical
Institute, Cape Town
Jennerston O (1988) Pollination in Dianthus deltoides (Caryophyllaceae): effects of habitat fragmentation on visitation and seed
set. Conserv Biol 2:359–366
Johnson SD, Steiner KE (2000) Generalization versus specialization
in plant pollination systems. Trends Ecol Evol 15:140–143
Lammers TG (2007) World checklist and bibliography of Campanulaceae. Royal Botanic Gardens, Kew
Lloyd DG, Yates JM (1982) Intrasexual selection and the segregation
of pollen and stigma in hermaphrodite plants, exemplified by
Wahlenbergia albomarginata (Campanulaceae). Evolution 36:
903–913
Lord EL (1980) Intra-inflorescence variability in pollen/ovule ratios
in the cleistogamous species Lamium amplexicaule (Labiatae).
Am J Bot 67:529–533
Minitab (2010) Statistical software, version 16. State College, PA.
http://www.minitab.com
Nyman Y (1992) Reproduction in Campanula afra (Campanulaceae):
mating system and the role of the pollen collecting hairs. Plant
Syst Evol 183:33–41
Nyman Y (1993) The pollen-collecting hairs of Campanula (Campanulaceae). II. Function and adaptive significance in relation to
pollination. Am J Bot 80:1437–1443
Preston RE (1986) Pollen-ovule ratios in the Cruciferae. Am J Bot
73:1732–1740
Proctor M, Yeo P (1972) The pollination of flowers. Taplinger, New
York
Rebelo AG, Boucher C, Helme N, Mucina L, Rutherford MC (2006)
Fynbos biome. In: Mucina L, Rutherford MC (eds) The
vegetation of South Africa, Lesotho and Swaziland, vol 19.
Strelitzia. South African National Biodivesity Institute, Pretoria,
pp 53–219
Richardson TE, Stephenson AG (1989) Pollen removal and pollen
deposition affect the duration of the staminate and pistillate
phases in Campanula rapunculoides. Am J Bot 76:532–538
Steffan-Dewenter I, Tscharntke T (1999) Effects of habitat isolation
on pollinator communities and seed set. Oecologia 121:432–440
Van Kleunen M, Nänni I, Donaldson JS, Manning JC (2007) The role
of beetle marks and flower colour on visitation by monkey
beetles (Hopliini) in the Greater Cape Floral Region, South
Africa. Ann Bot 100:1483–1489
Welman WG, Cupido CN (2003) Campanulaceae. In: Germishuizen
G, Meyer NL (eds) Plants of southern Africa: an annotated
checklist, vol 14. Strelitzia. National Botanical Institute, Pretoria, pp 336–346