Biological Flora of the British Isles: Campanula rotundifolia


  • Nomenclature of vascular plants follows Stace (2010) and, for non-British species, Flora Europaea.

Correspondence author. E-mail:


1. This account presents information on all aspects of the biology of Campanula rotundifolia L. that are relevant to understanding its ecological characteristics and behaviour. The main topics are presented within the standard framework of the Biological Flora of the British Isles: distribution, habitat, communities, responses to biotic factors, responses to environment, structure and physiology, phenology, floral and seed characters, herbivory and disease, history, and conservation.

2.Campanula rotundifolia is a slender, rhizomatous, prostrate to erect herb with long-stalked roundish basal leaves, linear stem leaves, and a blue, bell-shaped corolla. It is widespread in the British Isles though absent from parts of Ireland. Globally, C. rotundifolia has a circumpolar distribution extending from the Arctic Circle to northern Mexico and North Africa. It forms a polyploid complex with some characters linked to ploidy level. Populations in the British Isles are predominantly tetraploid and hexaploid, with occasional pentaploids. The distribution of these cytotypes in the British Isles has a strong spatial structure.

3.Campanula rotundifolia has a wide ecological amplitude, tolerating very dry conditions but also occasionally occurring in permanently saturated habitats, and it grows on a very wide range of soil types, from coarse sands through loams to heavy clays and pure peats. However, C. rotundifolia is rarely found on fertile lowland clays where competition from more vigorous species may limit growth. Campanula rotundifolia is found in a range of grassland, heath, mire, scrub and sand dune communities spanning a wide range of soil pH.

4.Campanula rotundifolia is perennial and spreads by seed and by rhizome. Plants generally overwinter as frost-resistant green rosettes. As the spring season progresses, plants make slow vegetative growth. Erect flowering stems develop from June onwards. Campanula rotundifolia is largely self-incompatible, and is insect pollinated, regularly producing large quantities of viable seed.

5. Although C. rotundifolia is a widespread and locally very common species there is some evidence from Britain and across Europe that it is in decline. These losses are likely to be due to a number of factors including agricultural intensification, reversion of grassland to scrub and woodland, disturbance, and atmospheric pollution.

Harebell, Bluebell (esp. Scotland), Bellflower. Campanulaceae. Campanula rotundifolia L. is a slender, rhizomatous perennial herb with a creeping, slender, elongated and much branched rootstock which produces adventitious buds. A white tap root is also occasionally present. All plant parts exude latex when broken. Stems slender and wiry, shortly creeping, then ascending to 15–60 cm, though prostrate, erect and lax forms also occur. Stems glabrous to sparsely pubescent, but all British plants are at least somewhat pilose, with short stiff white hairs which are most obvious on the lower part of the stem.

Basal and cauline leaves differ in morphology. Basal leaves cordate and reniform, sometimes crenate, c. 1 cm wide (total area 1–10 cm2) with long slender petioles. Sessile linear leaves on ground-hugging creeping stems. Cauline leaves on flowering stems, progressively narrower and more distant and stalkless upwards towards the flowers, spathulate to lanceolate, variable in number. Secondary veins slightly raised on leaf adaxial surfaces, leaf margins with hydathodes.

Inflorescence paniculate or racemose, with 1 – many flowers, very slender pedicels, sometimes stout. Corolla bell-shaped (varying from campanulate with a hemispherical base and ovary to conical with a conical ovary) with five lobes, rarely four, six or seven, 12–30(35) mm long, erect in bud, nodding when open. Usually blue, but varying in colour from white (rarely) through all shades of pale blue to deep blue and, in non-British material, violet blue, sometimes weathering to pale pink. Stamens 5, anthers cream, with cream to deep mauve pollen; style 1, stigmas 3. Flower size and number is sometimes used as a taxonomic character in the subsection Heterophylla and as a guide to polyploid level in C. rotundifolia s.s. and C. gieseckiana. However, although there may be differences, flower size and number vary enormously with environmental conditions.

Campanula rotundifolia forms a polyploid complex (2n = 34, 68 and 102) across Europe, 2n = 68 and 102 predominate in the British Isles, and 2n = 34 has not been recorded in the British Isles since the 1960s, despite searching the original locations. Two subspecies have recently been described in the British Isles based on morphology and cytotype:

  • subsp. rotundifolia. Stem-leaves narrow and acute; flowers few to many, mostly <20 mm; capsules obconical; 2n = 68. Suitable places throughout the British Isles’

  • subsp. montana (Syme) P.D. Sell. Stem leaves broader and subacute to obtuse; flowers 1-few, mostly >20 mm; capsules broadly obconical, squat; 2n = 102. Mostly in uplands; Ireland, western Scotland, Isle of Man and extreme south-west England (Stace 2010), possibly endemic.

However, in our experience, 2n = 68 is absent in Ireland and rarely co-occurs with 2= 102. Typical plants of the cytotypes 2n = 68 and 102 (tetraploid and hexaploid, respectively) can be distinguished morphologically but separation is not consistently possible as the 2n = 102 cytotype is very variable in form (see VI (A)). This description uses information from: Böcher (1960); Gadella (1964); Clapham, Tutin & Moore (1987); McAllister (1972); Sell & Murrell (2006); Grime, Hodgson & Hunt (2007); Poland & Clement (2009); Stace (2010).

Campanula L. is the largest genus of the Campanulaceae. The British Isles are comparatively poor in Campanula species, with 17 species recorded (Stace 2010), compared with 144 in Europe (Tutin et al. 1976) and about 85 species in Croatia, the circum-Adriatic and west Balkan regions (Kovačić 2004). Within the Britain Isles, C. rotundifolia is readily distinguished by its morphology from other native Campanula species. The classification of Campanula taxa is the subject of ongoing debate, and is largely unresolved (e.g. Eddie et al. 2003; Shulkina, Gaskin & Eddie 2003; Kovačić 2004; Roquet et al. 2008). Within this genus, the taxonomy and delimitation of C. rotundifolia is also difficult as it is an extremely variable species, forming a complex polyploid series. Indeed, Hultén (1971) described this as ‘An extremely complicated complex, where different authors rarely arrive at the same conclusions …’. Many variants have been described as distinct species and there is a considerable number of subspecies. Much of the variation is continuous and the taxonomic issues are not settled (e.g. Kovanda 1970, 1977; Tutin et al. 1976; Kovačić & Nikolic 2006). In addition to the recent distinction of two subspecies in the British Isles already mentioned, particular difficulties arise in distinguishing C. rotundifolia from the very similar C. gieseckiana (Greenland, Iceland, N. Norway), C.  intercedens Witas. (N. America), C. latisepala Hultén (W. N. America) and, perhaps especially, the contiguous C. langsdorffiana Fisch. (Siberia). In Europe, as infraspecific taxa, Flora Europaea lists montane subsp. polymorpha (Witasek) Tacik, subsp. sudetica (Hruby) Soó and var. alpicola Hayek, which are tetraploid, as is the morphologically similar C. groenlandica Berlin. However, in North America, Shetler’s (1982) survey of the Nearctic harebells concluded that they all belonged to one species, C. rotundifolia L. The extent of confusion is exemplified by a study of floral morphometrics by Kovačić & Nikolic (2006) of western Balkan lineages of endemic Campanula L. which concluded that ‘the heterophyllous taxa cannot be successfully separated only on the basis of their floral variables, while they are often so polymorphous and variable that they cannot be recognised even on the basis of entire plant morphology.’

The current taxonomic difficulties are clearly associated with evolutionary processes and further work is needed to gain a fuller understanding of the taxonomic divisions indicated above. In this work, we rely on the original naming by authors of their species, and on their determinations of the ploidy of their material. Within the British Isles, the cytotype of studied material is rarely defined, but if the location of the experimental study is known, it may often be deduced.

Campanula rotundifolia is found in a range of acidic to calcareous grasslands, heathlands, mires and dunes, as well as on maritime cliffs and in woodland. It is widespread and locally common but there is some evidence for recent decline in abundance and range.

I. Geographical and altitudinal distribution

The distribution of Campanula rotundifolia in the British Isles (Fig. 1) differs from that of any other taxon (Perring & Walters 1962; Preston, Pearman & Dines 2002). As highlighted in this map, the species is absent from several southern and central counties of Ireland (Scannell & Synott 1987), and the rarity of the species in Ireland is most striking given the high frequency in Great Britain, where absences in the West Country, the Wash area and inland north-west Scotland are also notable. The intensively cultivated arable land around the Wash was largely reclaimed from the sea or drained in historic times and there are few habitats suitable for C. rotundifolia, but land-use history and intensive cultivation do not account for the other gaps in the distribution of the species.

Figure 1.

 The distribution of Campanula rotundifolia in the British Isles. Each dot represents at least one record in a 10-km square of the National grid. Native: (inline image) 1970 onwards, (inline image) pre 1970; introduced (inline image) 1970 onwards, (inline image) pre 1970. Mapped by Colin Harrower, the Biological Records Centre, Centre for Ecology and Hydrology, Wallingford, mainly from data collected by members of the Botanical Society of the British Isles, using Dr A. Morton’s DMAP software.

The distribution of cytotypes of C. rotundifolia in the British Isles has a strong spatial structure and little overlap (Fig. 2). Hexaploids (2n = 102) have an oceanic westerly and northerly distribution, while tetraploids (2n = 68) are more easterly and southern. Rarely, hexaploids occur in isolated upland areas within regions otherwise occupied by tetraploids. Occasional pentaploids have been found where populations are mixed. Although there have been some past records of diploids (2n = 34) occurring in the Cheviots, West Norfolk and Cambridge area (Kovanda 1970) and tetraploids occurring in NW Ireland (Böcher 1960), subsequent intensive searches have failed to relocate them and they are excluded from this map. It is notable that the gaps in the species distribution in NW Scotland and SW England occur at the boundaries between hexaploid and tetraploid cytotypes.

Figure 2.

 Distribution of tetraploid (2n =  68; inline image, inline image, inline image), pentaploid (2n = 85; inline image, inline image, inline image) and hexaploid (2n = 102; inline image, inline image, inline image) cytotypes of Campanula rotundifolia in the British Isles. Data from McAllister (1972) and J. Shepherd and J. Wilson (unpubl. data.) mapped on a 10-km square basis, using Dr A. Morton’s DMAP software.

On a broader scale, the global distribution of C. rotundifolia has been described as being Circumpolar Boreo-temperate (Preston & Hill 1997). Hultén’s (1971) map (Fig. 3) highlights two main types of C. rotundifolia with overlapping latitudinal distributions; the Linnaean type – much branched, small flowered and common in the central lowlands of Europe, and a larger-flowered type predominating in the north and also occurring in some mountainous regions, including the Alps and Scottish mountains. In Europe, diploids and tetraploids are widespread (Böcher 1960; Kovanda 1970; Laane, Croff & Wahlstrøm 1983) and hexaploids rare, while hexaploids are found most extensively in the British Isles, and occasionally in France (Kovanda 1970) and Spain (McAllister 1972). The Nearctic distribution, summarized by Shetler (1982), is predominantly tetraploid, with occasional isolated northern or montane hexaploids, and frequent northern diploids in Greenland.

Figure 3.

 World-wide distribution of Campanula rotundifolia (Hultén & Fries 1986). Cross-hatched areas indicate regions of common occurrence, dots indicate isolated, fairly exact occurrences. Reproduced by permission of Koeltz Scientific Books.

There seems to be a fair measure of agreement that the diploids fall into two categories (Böcher 1960; Löve & Löve 1965). First, there are the arctic diploids usually named C. gieseckiana ssp. gieseckiana found in Greenland (Böcher 1960, 1966) and with populations on Mount Washington, New Hampshire (Löve & Löve 1965), Spitzbergen (Flovik 1940), Northern Norway (Laane, Croff & Wahlstrøm 1983) and Northern Russia (Gadella 1964). Second, there are diploids with a more southern continental distribution usually referred to as the diploid cytotype of C. rotundifolia s.s. These are found largely in Central Europe in areas which were between the Arctic and Alpine ice sheets (Kovanda 1970, 1977), but indistinguishable plants have also been found on the shores of Lake Baikal in Eastern Siberia (H. A. McAllister, unpubl. data) and specimens differing primarily in having papillose ovaries occur in Southern Europe (Tutin et al. 1976). Other closely related diploids exist whose relationship with the above taxa is unclear.

In the British Isles, the altitudinal range of C. rotundifolia is large, with records from sea level up to 1160 m (Breadalbane, Mid Perth; Preston, Pearman & Dines 2002). Elsewhere, there are records from much higher altitudes (e.g. 2700 m on Mt Olympus, Greece; Blionis, Halley & Vokou 2001). In the British Isles, within their respective geographic ranges, tetraploids have been identified from sea level to over 900 m (in the southern Scottish Highlands on Beinn Dourain near Bridge of Orchy, and in Glen Falloch), and the hexaploid from sea level to over 1000 m on Ben Alder (McAllister 1972).

II. Habitat

(A) Climatic and topographical limitations

Campanula rotundifolia has a wide ecological amplitude and all of the British Isles falls within the climatic range of the species.


In the British Isles, the January and July mean temperatures of grid squares containing C. rotundifolia are 3.2 and 14.4 °C respectively and the mean annual precipitation is 1104 mm (Hill, Preston & Roy 2004). The northernmost record for C. rotundifolia (2n = 34, subspecies gieseckiana) is 78o07′ in Colebukta, Svalbard (average July temperature 5.9 °C; Alsos, Spjelkavik & Engelskjen 2003), but although there was flowering in late August, no germinable seed was set, and no seed bank has been detected in any of the four populations observed on Svalbard, which persist vegetatively. This is likely to be due to the thermal requirements for seed production not being met, although lack of pollinators may also be a limiting factor (Alsos, Spjelkavik & Engelskjen 2003). Records of flowering in different seasons indicate a range of first flowering dates from mid July to mid August (Engelskjøn, Lund & Alsos 2003), not dissimilar to the British Isles. Further south, in northern Norway, at 69° N with a July temperature of 10.9 °C, C. rotundifolia var. rotundifolia (2= 68) produces seedlings, and flowering plants of 2n = 34 C. gieseckiana var. gieseckiana have been recorded further north on the mainland (Laane, Croff & Wahlstrøm 1983).

Rainfall and humidity

Campanula rotundifolia is often considered a plant of moderately dry soils (Clapham, Tutin & Moore 1987), and it can tolerate very dry conditions on light sandy soils or dry crevices in cliffs or walls. It can also be found occasionally in permanently saturated habitats such as flushed rock ledges or in saturated Sphagnum turfs.

Exposure to wind

Apart from the summer flowering stems, the vegetative growth of C. rotundifolia is ground-hugging, and the species is found in both sheltered and exposed habitats, although it is often of more compact form in the latter. It appears tolerant of extremely exposed conditions, as it is frequently found on exposed cliff ledges (also indicating a degree of salt tolerance, although this is at variance with Hill et al. (1999)) and mountainous areas, probably largely due to lack of competition in these habitats. It is also found in more sheltered situations where the habitat is sufficiently open.

Light flux

This species has an Ellenberg value of 7, being generally found in well-lit locations, though occasionally in shade (Hill et al. 1999) on some southern cliffs and in semi-shaded situations (e.g. on well drained roadside banks, beneath trees), but it never occurs in deep shade.

(B) Substratum

Parent material

Campanula rotundifolia tolerates a very wide range of soil types, from limestone and chalk coarse sands through loams to heavy clays and (rarely) pure peats, but is most common on well drained, nutrient poor (but often lime-rich) stony or sandy soils or peaty soils. Occurrence on fertile soils where competition with other species is normally strong, is usually only when the habitat is kept open by other factors, such as erosion, land management, or toxic concentrations of heavy metals.

Soil pH

A wide range of soil pH values is tolerated by both British cytotypes. McAllister (1972) found that tetraploids occurred on soils with pH values from 3.65 to 8, though occasional chlorosis was observed on very wet, alkaline sites. Hexaploids grew on soils with a similar pH range from 3.8 to 8. However, despite extensive sampling, very few hexaploid plants were found on soils with intermediate pH values between 5.6 and 6.8.

In Norway, Wielgolaski (2001) notes that high pH may be related to delayed flowering.

III. Communities

Given its broad ecological amplitude it is no surprise that C. rotundifolia occurs in a wide range of plant communities in Britain (Rodwell 1991a,b, 1992; 2000). The majority of these are essentially grassland communities, although they may fall into a different classification according to underlying substratum or the presence of shrub species. Thompson et al. (1996) noted the complete absence of the species from improved grassland in central England, and this preference for nutrient-poor habitats is reflected in Rodwell’s classifications. However, despite this, in cultivation the species grows well in fertile conditions and the natural occurrence on nutrient poor sites may indicate intolerance of competition. Communities in which C. rotundifolia is found are summarised in Table 1.

Table 1.   British plant communities classification within which Campanula rotundifolia is found (Rodwell 1991a,b, 1992, 2000)
Habitat typeCommunities
Calcicolous grasslandCG1, CG2, CG3, CG4, CG5, CG6, CG7, CG8, CG9, CG10, CG11, CG12, CG13, CG14
Calcifugous grasslandU1, U4, U5, U10, U13, U14, U15, U16, U17, U19, U20, U22, U23, U24
Mesotrophic grasslandMG1, MG2, MG3
HeathlandH1, H10, H12, H16, H18
Sand duneSD7, SD8, SD9, SD11, SD12
Maritime cliffMC9
WoodlandW11, W19, W20, W23


Campanula rotundifolia has been recorded in all 14 calcicolous grassland (CG) communities. In many of these C. rotundifolia is an associate or preferential species, but in Sesleria albicans-Galium sterneri grassland (CG9), Festuca ovina–Agrostis capillaris–Thymus praecox grassland (CG10) and Dryas octopetala–Silene acaulis ledge community (CG14) C. rotundifolia is a constant. Although cover is generally low it can be up to a Domin score of 5 (11–25% cover) in CG9. All of the communities in which C. rotundifolia is constant are confined to northern England, Wales and Scotland. CG9 has a variable sward which can be open or closed, very short or taller and tussocky, whereas CG10 frequently has a closed sward, cropped closely by grazing. CG14 is a very different community with a variable composition of mixed dwarf shrubs, sedges and herbs on ledges of rock outcrops (Rodwell 1992).

Campanula rotundifolia is also recorded in 14 out of 24 calcifugous grassland and montane communities. In all of these, C. rotundifolia is a preferential or associate species reaching a maximum frequency of III (40–60%) and Domin score of 1–4 in the Pteridium aquilinum–Galium saxatile community (U20) (Rodwell 1992).

Campanula rotundifolia is recorded in only three mesotrophic grassland (MG) communities [MG1 Arrhenatherum elatius grassland (MG1), Arrhenatherum elatius–Filipendula ulmaria tall-herb grassland (MG2) and Anthoxanthum odoratum-Geranium sylvaticum grassland (MG3)]. This is not surprising given its poor competitive ability in nutrient rich conditions (although if soils are contaminated with metals it may be more abundant – see V). Of these, it is most frequent in MG2 grassland where it has a frequency of III and abundance of 1–4 (Domin scale). In this grassland, the tall herbs and grasses form an uneven canopy below which smaller species such as C. rotundifolia can be locally abundant (Rodwell 1992).

Heathland and mire

In addition to grasslands, C. rotundifolia is also found in five heathland communities: Calluna vulgaris–Festuca ovina heath (H1), Calluna vulgaris-Erica cinerea heath (H10), Calluna vulgaris–Vaccinium myrtillus heath (H12), Calluna vulgaris–Arctostaphylos uva-ursi heath (H16) and Vaccinium myrtillus–Deschampsia flexuosa heath (H18). In heathlands C. rotundifolia will typically be found only occasionally in areas of more open Calluna vulgaris cover together with species such as Galium saxatile, Potentilla erecta, and Cerastium fontanum. Campanula rotundifolia is also recorded in one mire community, Carex demissa–Saxifraga aizoides mire (M11). This community, restricted to upland areas in the north and west of the UK, has a diverse and variable hummocky sward with exposed silt and rock debris (Rodwell 1991b).

Maritime and sand dune

There are several coastal communities in which C. rotundifolia is found, including five sand dune communities and a maritime cliff community. All of the sand dune communities are fixed or semi-fixed, but C. rotundifolia occurs on both acidic and calcareous sands. On calcareous sands C. rotundifolia occurs in the semi-fixed dune community Ammophila arenaria–Festuca rubra (SD7) as an associate species with a frequency of I (<20%), but as this dune grassland becomes further stabilized it develops into Festuca rubra–Galium verum fixed-dune grassland (SD8) where the frequency of C. rotundifolia increases up to a frequency of IV (60–80%). In the absence of grazing this community can develop into Ammophila arenaria–Arrhenatherum elatius dune grassland (SD9) where C. rotundifolia also occurs but is less common. On acid sands, C. rotundifolia can be found in two fixed dune communities, at low frequency in the lichen-rich community Carex arenaria–Cornicularia aculeata dune community (SD11) and up to a frequency of III in the grassland Carex arenaria–Festuca ovina–Agrostis capillaris dune grassland (SD12). Campanula rotundifolia is also found in the maritime cliff community Festuca rubra–Holcus lanatus maritime grassland (MC9; Rodwell 2000).


Although not a common species of woodlands, C. rotundifolia does occur in two open woodland communities: in the understorey of Quercus petraea–Betula pubescens–Oxalis acetosella woodland (W11), an open and grassy community, and Juniperus communis ssp. communis–Oxalis acetosella woodland (W19) which has a discontinuous herbaceous layer. It also occurs in the herb rich scrub community Salix lapponum–Luzula sylvatica scrub (W20) and Ulex europaeus–Rubus fruticosus scrub (W23) where the grassland between U. europaeus forms a continuous cover with a close affinity to U4 grassland (Rodwell 1991a).

Mainland European communities

Within mainland Europe C. rotundifolia occurs in similar communities to those in Great Britain. In Germany, Oberdorfer (1978) describes C. rotundifolia as being common in the Festuca-Brometea order (base-rich dry grasslands). Within the Festucetalia vallesiacae (continental steppe grasslands dominated by Stipa spp.) C. rotundifolia is sparse but it reaches higher levels of abundance within the Mesobromion alliance (chalk and limestone grasslands). It is also found, though at lower constancy, throughout the Koelerio-Phleion phleoidis and Xerobromion alliances. The Elynetea and Seslerietea orders describe subalpine and alpine base-rich grasslands. Here C. rotundifolia is largely absent from the high altitude communities where C. scheuchzeri replaces it. In the subalpine Seslerion communities C. rotundifolia is commonly found. Campanula rotundifolia is largely absent from dry open Sedo-Scleranthetea order, only occurring in the Diantho gratianopolitani-Festucetum pallentis and the Artemisio lednicensis-Melicetum ciliatae communities. In both of these communities C. rotundifolia is found with calcareous grassland species.

The class Nardo-Callunetea includes well-drained acid grasslands and dwarf-shrub heaths. Here C. rotundifolia is absent from the highest altitude community (Aveno-Nardetum – altitudinal range 1750–2250 m) but occurs in the other four alliances (Nardetum alpigenum, Violo-Nardetum, Leontodonto helvetici-Nardetum, and Lycopodio-Nardetum). The Violion caninae order describes the acid grasslands extending from lowland areas to the lower montane zone. Here C. rotundifolia is commonly found in all associations except the Polygalo-Nardetum alliance. Within the Vaccinio-Genistetalia heathland communities C. rotundifolia is common but it occurs only very rarely in Juncus squarrosus communities (Oberdorfer 1978).

VI. Response to biotic factors

Grazing and trampling

The upright flowering stems are grazed by animals, and observations of sheep grazing in Iceland found that plants were grazed on up to 15% of occasions in a forest enclosure (Thorhallsdottir & Thorsteinsson 1993). The flowering stems are brittle and easily damaged by trampling. Thus, where there is much animal activity, flowering and seed set are poor and the ability of the plant to spread vegetatively is likely to be increasingly important and this habit gives the species a competitive advantage following disturbance (Hillier 1984). Beneficial effects of grazing were shown in Sweden, where Lindborg, Cousins & Eriksson (2005) found that C. rotundifolia was most frequent in plots which had been continuously grazed open grassland for 50 years rather than in grassland plots which had been abandoned and restored at different periods.


Disturbed open ground, such as that created by rabbit scrapes and mole hills, provides important locations for the establishment of seedlings, which are poor competitors in closed swards. Although often found in close association with ants, and ants have been seen regularly visiting the nectaries, King (1977) found that this species was similarly abundant on and around ant hills.

Plant-soil feedbacks

Studies of plant-soil feedbacks through soil organic matter, soil nutrient availability and the composition of soil microbial communities in a number of grassland species in the Netherlands indicated that identity of previous 2-year-old monocultures had no significant effect on biomass of C. rotundifolia (Bezemer et al. 2006). However, Agrawal et al. (2005) found that growth of C. rotundifolia was reduced by 10% when plants were grown in soil containing a microbial filtrate from previous culture of this species.

Plant behaviour under competition

The growth strategy of C. rotundifolia is intermediate between being a stress-tolerator and a C-S-R strategist (where competition is restricted by both stress and disturbance; Grime, Hodgson & Hunt 2007). It is commonly a subordinate species and does not compete well with other, more vigorous species. It is intolerant of high intensities of disturbance, and restricted to relatively unproductive vegetation.

There is some uncertainty concerning the foraging plasticity of this species. An experimental assessment showed that C. rotundifolia had some ability to preferentially allocate above- and below-ground dry matter to areas of highest light and nutrients respectively (Campbell, Grime & Mackey 1991). This high precision foraging was suggested as typical of subordinate species. However, in contrast to these observations, Wijesinghe et  al. (2001), using different methods, found that C. rotundifolia showed no precision in root placement in relation to nutrient patches.

Hovstad & Ohlson (2008) found that seedling establishment of C. rotundifolia was facilitated by intermediate levels of litter, possibly mediated by positive effects on mycorrhiza or reduced water stress. The ability to forage selectively for light (Campbell, Grime & Mackey 1991) may also be related to this facilitation. At higher levels of litter, seedling emergence was reduced as a result of reduced light levels at the soil surface and an inhibitory effect of litter extract on seedling emergence. This inhibitory effect of litter extract was attributed to phytotoxic polyphenols but results suggested that chemical effects of litter were weaker than physical effects (Hovstad & Ohlson 2008).

Booth & Grime (2003) evaluated the effects of different levels of intra-specific diversity on limestone grassland communities at Cressbrookdale, North Derbyshire, incorporating C. rotundifolia as one of the subordinate species in a microcosm study. After five growing seasons, higher levels of species diversity occurred in the most genetically diverse communities. In a continuation of this study, Whitlock et al. (2007), evaluated the genetic diversity of several of the component species, including Campanula, and every living shoot of this species was analysed. In contrast to the dominant species, C. rotundifolia lost more genotypes from the genetically diverse communities. Diversity in fitness of the dominant species may be essential to protect subordinate species such as C. rotundifolia from exclusion and thus for conservation purposes it is important to ensure that populations incorporate a wide range of intraspecific variation. When grown with different genotypes of Carex caryophyllea and Koeleria macrantha under fertilised and unfertilised conditions C. rotundifolia showed a very variable response which not only depended on the fertility treatment but also on interaction with both of the neighbour species genotypes (Fridley, Grime & Bilton 2007).

Aside from the work described above, there has been little investigation into the competitive ability of C. rotundifolia within European habitats. In a competition experiment investigating the effect of the invasive species Hesperis matronalis in North America on native species, Hwang & Lauenroth (2008) found that the growth of C. rotundifolia was consistently suppressed in competition with H. matronalis at all neighbour densities, nitrogen and watering treatments due to its lower relative growth rate.

V. Response to environment

(A) Gregariousness

Campanula rotundifolia very rarely forms pure stands or dominates communities. Perhaps the only situation in which this occurs is on rock ledges and very shallow rocky soils where its tap root provides a competitive advantage. In these situations in the British Isles it often grows with the rhizomatous Festuca rubra, perhaps the commonest associate of C. rotundifolia in the British Isles. Otherwise, C. rotundifolia is usually a subordinate species, growing through grassland on soils of low fertility, sometimes singly but more usually scattered throughout the grassland.

(B) Performance in various habitats

Campanula rotundifolia usually occurs in habitats with high light flux and little or no shading. It is frequently found in short grassy vegetation and tends only to be found in taller vegetation where conditions are open with large amounts of bare soil, such as in sand dunes. Generally C. rotundifolia is found on relatively infertile soils and it therefore rarely occurs with species typical of fertile lowland clay or loam soils unless other conditions such as exposure, lack of soil depth or metal toxicity prevent proper development of vigorous species. Flowering frequency is affected by habitat: in open grassland, 96% of individuals flowered, compared with 37% in forest (Lindborg, Cousins & Eriksson 2005).

Metal tolerance

Campanula rotundifolia occurs in plant communities on contaminated soils, such as mine sites. It is moderately tolerant of heavy metals in soil, but not dependent on their presence (Baker et al. 2010). The species is described as an ‘elective pseudometallophyte’– occurring both on contaminated soils and on normal soils in the same region (Antonovics, Bradshaw & Turner 1971). However, Schwanitz & Hahn (1954 cited in Antonovics, Bradshaw & Turner 1971) and Cook, Lefebvre & McNeilly (1972), described it as zinc tolerant and noted that metal tolerant ecotypes were of small habit; Shimwell & Laurie (1972) found that it accumulated zinc to 100–2100 ppb and lead to 15–250 ppb at mine sites in the Pennines. Other authors also record accumulation of cadmium, copper and fluoride (Cooke et al. 1976; Baker 1981).

Atmospheric pollution

Open-top chamber studies have shown varying responses of C. rotundifolia to ozone. A study of episodic ozone exposure (Hayes et al. 2006) found no response to a 10-week summer treatment with peak ozone concentrations of 100 ppb, either immediately or after over-wintering, whereas Thwaites et al. (2006) found that it died out after one season’s treatment at average accumulated exposure over a 40 ppb threshold (AOT40) of 15.7 ppm (although it also declined in ambient controls) and Rämöet al. (2007) recorded diminished and delayed flowering when exposed to ozone, but no effect of CO2 treatment (350/455 ppm) on the date of flowering. In terms of perennial grassland species, these are relatively short term studies and there may be differences between relatively transient effects and longer-term cumulative effects. Furthermore, there may be within-species variation in tolerance and response to pollutants, as there is in survival and performance of C. rotundifolia genotypes in other microcosm studies (Whitlock et al. 2007). In heavily grazed or mown grassland where flowering is rare and the plant spreads vegetatively, and in areas of dense vegetation where seedling establishment is difficult, the opportunities for adaptation may be less than in locations where flowering is more abundant and there is bare ground for seedling establishment.

Using canonical correspondence analysis Stevens et al. (2010, 2011) found a negative correlation between the abundance of C. rotundifolia in acid grasslands and levels of atmospheric nitrogen deposition in Europe (nitrogen deposition range 2.4–43.5 kg N ha−1 year−1). However, in a controlled experiment in the same grassland community Carroll et al. (2003) failed to find any significant response of C. rotundifolia to nitrogen addition after 5 years of treatment but it is possible that there was insufficient time for effects to be seen. Van den Berg et al. (2010) surveyed permanent quadrats in calcareous grassland in 1990–1993 and again in 2006–2009. They found that the frequency of C. rotundifolia had declined by 25% in the area of highest nitrogen deposition, and the correlation between decline and nitrogen deposition was highly significant.

(C) Effect of frost, drought, fire etc.

The temperature range of C. rotundifolia is very broad and the occurrence of C. rotundifolia at northerly latitudes is testament to its frost hardiness. It will often flower late in the year, up to the first hard frost.

Though, at least in the British Isles, C. rotundifolia is usually thought of as a plant of dry places (Clapham, Tutin & Moore 1987), there are indications that the species is excluded from some habitats by water stress and Hill et al. (1999) record it as being intermediate between being a dry and moist site indicator (Ellenberg value 4). Droughted plants often have a stunted erect habit and produce small flowers with protruding stigmas (McAllister 1972), so it is possible to identify drought stressed plants in the field (these characters are not maintained in cultivation). The main types of dry habitat colonized by C. rotundifolia are poor grasslands on light sandy soils and dry crevices in cliffs and dry-stone walls. In all such situations the plant’s long tap root may enable it to access deeper water supplies than its shallower rooted competitors such as grasses. Such habitats become very dry in long summer droughts and the vegetation frequently becomes brown and parched. In all such situations examined, the C. rotundifolia plants remained green though often wilted.

The species’ ability to cope with drought was highlighted in a long-term drought experiment on shallow calcareous soil (Buxton, UK), in which C. rotundifolia showed little response to repeated summer drought and winter warming (Fridley et al. 2011) and persisted as a subordinate under a wide range of environmental conditions.

In tests of the ecological effects of heather burning in Scottish moorland Mallik & Gimingham (1985) found that C. rotundifolia can regenerate from underground perennating organs which escape the effects of fire.

VI. Structure and physiology

(A) Morphology

Campanula rotundifolia is morphologically highly variable. Observations on reference material of known ploidy level show that the more delicate plants with paniculate inflorescences, thin pedicels, small (<20 mm long) trumpet-shaped flowers, acute linear stem leaves, and few transitional leaves will generally prove to be tetraploid. Hexaploid plants are morphologically more variable: they may be short plants with stiff thick stems bearing few large or broad flowers on relatively thick pedicels with hemispherical corolla bases and ovaries, or alternatively, tall plants with large flowers (over 25 mm long, usually numerous). Both of these morphotypes tend to have blunt more or less spathulate stem leaves. However, while plants with the latter morphologies are fairly reliably hexaploid, plants of the former delicate morphology may also be hexaploid.

Pollen size and stomatal dimensions increase with increasing ploidy, though with overlap between the different cytotypes (e.g. Gadella 1964; Kovanda 1970; Hubac 1972; McAllister 1972; Laane, Croff & Wahlstrøm 1983). The possibility of other distinguishing (though microscopic) features has been raised by Laane, Croff & Wahlstrøm (1983), who, in a comparison of Norwegian diploids and tetraploids, found that the milky sap of tetraploids contained 0.5 μm spherical particles, while diploids did not, and that the fluorescence characters of these two cytotypes differed. These characters require further examination across a wider range of material.

Plant developmental stages

The first rosette-forming post-cotyledonary leaves have long petioles and are cordate and typical of the basal leaves of the species. Under favourable conditions the buds in the axils of the first one to three rosette leaves produce axillary rosettes while buds in the axils of later rosette leaves usually produce flowering stems. The rosettes bear only the long-petioled cordate leaves while the flowering stems nearly always bear only linear leaves with spathulate intermediate leaves towards the base of the stem (Fig. 4).

Figure 4.

 Glasshouse-grown tetraploid seedling, of Campanula rotundifolia, aged 6 months. Root length reduced by 60% in this image.

The main axes of the rosettes elongate very little. Buds in the axils of the rosette leaves develop to flowering stems, or occasionally rhizomes with long internodes. Leaf development on rhizomes seems to be controlled by light. A rhizome in darkness deep in the soil or crevice will only produce scale leaves, while one near the surface or appressed to the inside of a somewhat translucent pot will produce normal basal leaves on long petioles.

The tap root quickly develops into a carrot-like white main root. In many cases it is this alone which gives rise to the rhizomes, none arising from the aerial shoots. Though the tap root is usually long lived, the primary rosette rarely lives for more than 2 years, the growth of the plant being continued by the rhizomes and secondary rosettes. Secondary major roots sometimes arise from the original shoot or a rhizome branch.

The rhizomes may be branched or unbranched. The thicker stronger branches usually give rise directly to flowering stems while the thinner weaker rhizome branches give rise to rosettes. Such rosettes behave in exactly the same way as the primary rosette, producing axillary flowering shoots. Further rhizome branches are usually produced from the axils of rhizome scale leaves, not as adventitious shoots from a main root, and may be up to 0.5 m long. Roots are rarely borne on the young parts of a rhizome system, unless the rhizome is severed, only occurring on the older mature parts. No subsequently produced roots have the thickness of the primary root so seedlings are almost always distinguishable from plants derived from rhizome fragments. Initially rhizomes are usually positively geotropic to plagiotropic, growing downwards into the soil. Later they usually become plagiotropic to negatively geotropic, rising towards the surface to produce either a rosette of basal leaves or a flowering shoot.

Frequently, the rosette leaves wither by flowering time, new rosette leaves appearing in the autumn. In winter the aerial shoots die back so that the rhizome-system growing points terminate either in rosettes with petiolate cordate leaves or tufts of linear-spathulate leaves which will develop to flowering shoots in the following summer. Green leaves are usually present above the soil surface and no distinct overwintering buds are produced.

There is considerable variation in the extent of rhizome development. Most hexaploids have relatively short stout rhizomes. However, within both tetraploids and hexaploids the variation is ecologically correlated, cliff plants tending to have relatively short rhizomes while grassland and especially sand dune plants tend to have strongly developed rhizome systems.


Stomata are present on both surfaces of the leaves and vary in number and size according to cytotype, light and position on the plant. As with most species, there are more stomata on the lower side of the leaves, with means of 400 and 185 stomata mm−2 on the abaxial and adaxial surfaces respectively (Rea 1921). Stomata are also present on stems, and were reported as being ‘more abundant on stems than on rosette leaves’ by Stober (1917). Measurements of stomata by Kovanda (1970) showed that those of 2n = 34 C. rotundifolia tended to be smaller (mean length 28.4 μm) than 2n = 68 plants (mean 34.3 μm), although there was considerable overlap in size, and variation in size within a single leaf.

(B) Mycorrhiza

Like most herbaceous plants, Campanula rotundifolia is vesicular-arbuscular mycorrhizal (AM) (Harley & Harley 1987). Studies of growth response to mycorrhizal colonization have shown that some mycorrhizal fungi are more effective than others, and that a range of direct effects on treated plants and their offspring may occur.

Using seed from Finnish Lapland, Kytöviita, Vestberg & Tuom (2003) showed a four-fold increase in mass of C. rotundifolia seedlings inoculated with Glomus claroideum compared with uninoculated plants, or those inoculated with other Glomus species. Nuortila, Kytöviita & Tuomi (2004), also using seed from Finnish Lapland and mixed spore inoculum of different mycorrhizal species, investigated mycorrhizal effects on plant fitness through two generations in low nutrient conditions. They found that mycorrhizal mother plants were smaller and had fewer flowers than non-mycorrhizal plants, but had a higher root: shoot ratio and higher amounts and concentrations of phosphorus in shoots. Seedling offspring of mycorrhizal mother plants had higher relative growth rates than those of non-mycorrhizal mother plants. Pollination studies indicated that mycorrhizal plants were more self-compatible than non-mycorrhizal. By contrast Wijesinghe et al. (2001) found that biomass of inoculated plants was greater than non-inoculated, due largely to the greater number of flowering stems.

Working in Canada, in Alberta grasslands, Cahill et al. (2008) found that reducing mycorrhizal populations, by addition of benomyl, increased the number of flowering stems of C. rotundifolia from 8 to 18 stems. Several other, but not all, species behaved similarly. However, it was notable that AM colonization recorded in the field was very low (4.8% without benomyl, reduced to 3.4% with the addition of benomyl). The authors hypothesised that the increase in flowering may have been due to reduced competition by normally highly competitive mycorrhizal grasses, highlighting the difficulties of community ecology.

The effects of mycorrhizal infection in field situations may depend upon whether the plants are part of established mycorrhizal networks or not: isolated seedlings may benefit, whereas networked ones may not. Kytoviita, Vestberg & Tuom (2003) found that when grown in proximity to established Sibbaldia procumbens, the presence of mycorrhizal networks had no effect on growth of C. rotundifolia seedlings (although significant mycorrhizal colonization occurred), but when grown without Sibbaldia, seedlings inoculated with Glomus claroideum were about five times larger than non-inoculated plants, or plants inoculated with other Glomus species. Mycorrhizal colonization was highest with this inoculation treatment (72%), but significant levels of colonization (>40%) were achieved with several of the other treatments. Whether the difference in effects was due to competition with Sibbaldia in a constrained environment is not clear.

(C) Perennation: reproduction

Campanula rotundifolia is a perennial and spreads by seed and by rhizome. Although glasshouse-grown seedlings can easily flower in their first year, this is unlikely in most field conditions in the British Isles. The considerable distinctive variation in floral morphology enables clonal clumps to be distinguished in the wild, and clumps two to four metres across may be seen in grasslands, probably representing many years’ growth.

Flowering is very variable. In a common garden study of 93 clones from the British Isles (Wilson unpubl. data.), the mean number of flowering stems of plants in their second year of growth was 29, with a range of one to 96 stems between clones. Such vigour is rarely seen in the wild, where competition, grazing and trampling take their toll. With first flowering in July, continuing in many habitats until October, seed set occurs from August onwards. The capsule pedicels are brittle and in exposed or trampled locations, the capsules may break off before they are mature. Seed dispersal distances are short. Thus in grasslands, unless suitable microsites have been created (e.g. mole hill, or rabbit scrape), perennation is more likely by rhizome, whereas in more open situations, both seed and rhizome spread may be possible, but the pollination self incompatibility may act as a limit to the production of viable seed in isolated plants.

(D) Chromosomes

The base number is = 17 of the cytotype series 2n = 34, 68, 102; however this is thought to be secondary in origin and the mechanism of evolution of this number is obscure (Böcher 1960; Kovanda 1970). Enzyme analysis of 2n = 34 C. rotundifolia subsp. gieseckiana showed duplication of loci, indicating polyploidy (Alsos & Engelskjøn 2002). Examination of metaphase chromosomes suggested that Norwegian tetraploids are autopolyploid (Laane, Croff & Wahlstrøm 1983). Prepared, mitotic chromosomes are very small (1–1.5 μm). β chromosomes have been recorded in the three cytotypes (Kovanda 1970; McAllister 1972; Laane, Croff & Wahlstrøm 1983).

The 2C DNA amount of 2n = 68 C. rotundifolia from northern England is 5.3 pg (Mowforth 1986 cited in Bennett & Smith 1991), and, reporting from Slovenia, Vidic et al. (2009) found the 2C DNA amount of 2n = 34 C. rotundifolia was 2.62 Gbp (equivalent to 2.69 pg).

(E) Physiological data

Response to shade

As already discussed, this species is characteristic of open habitats and rarely occurs in shade, however, plants in semi-shade can become quite large and luxuriant. Morgan & Smith (1979) found that C. rotundifolia showed changes in growth habit in response to increased far-red light similar to other open-habitat species.

Water relations

Campanula rotundifolia avoids water stress by habitat selection, rather than tolerating it, and its taproot is advantageous in many habitats. Wolf & Lundholm (2008) included C. rotundifolia in tests of green roof microcosms which simulated summer conditions in Nova Scotia. In tests of different watering treatments, only succulents survived the dry treatment (water supplied every 24 days) but C. rotundifolia survived intermediate and wet treatments (watering every 11 and 4 days respectively).

Photosynthesis and respiration

Campanula rotundifolia has the leaf anatomy characteristic of a C3 plant, although no studies on its photosynthesis appear in the literature. Likewise, there is little available information on respiration. Respiratory fluxes of leaves and stems of C. rotundifolia were similar, and higher than those of reproductive structures (McCutchan & Monson 2001a). McCutchan & Monson (2001b) found that night-time respiration was not coupled to leaf carbohydrate supply.

(F) Biochemical data

Secondary plant compounds

Although starch is the principal storage polysaccharide of the majority of plant species, the Campanulales are one of the few dicotyledon orders which accumulate fructans (which are stored in vacuoles; Meier & Reid 1982). A survey of 130 herbaceous species in the Sheffield area (Hendry 1987; Brocklebank & Hendry 1989) found fructan accumulation in 20 species, including C. rotundifolia. Fructan, like starch, was mostly stored in roots and for this species the maximum concentration (74.9 mg g−1 fresh mass) was recorded in July and was the highest of the dicotyledonous accumulators, but less than that of the bulb-forming monocots. Although very low levels of starch were found (<2.0 mg g−1 fresh mass), this species also stored sucrose, in shoots, with maximum concentrations in January, showing higher concentrations than any of the other species assayed (Brocklebank & Hendry 1989). The ecological significance of fructan storage is not certain (Hendry 1987), although supporting the ability to remain green over-winter and a role in early spring growth are suggested as relevant to this species (Brocklebank & Hendry 1989). Vergauwen, Van den Ende & Van Laere (2000), working with C. rapunculoides, found fructans in roots, stems, petioles and floral parts and suggested that fructans stored in petals contributed to the osmotic changes required for rapid expansion of petals.

Little is known about the biochemical pathways linked to the development of flower colour in C. rotundifolia. However, luteolin glucosides have been identified in C. rotundifolia, in common with other species of Campanula (Justesen, Andersen & Brandt 1997). These flavones are known to function as co-pigments, enhancing the colour of anthocyanins.

Campanula rotundifolia contains biologically active hydrophilic water soluble alkaloids which inhibit glucosidase (Nash et al. 1997). Significant inhibition of bovine brain α-glucosidase, β-glucosidase, β-galactosidase, β-N-acetyl-galactosaminidase and β-N-acetyl-glucosaminidase was detected. This was due largely to a high concentration (up to 2% dry mass of leaves and stems) of the pyrrolidine alkaloid DMDP (2R,5R-dihydroxymethyl-3R,4R-dihydroxy-pyrrolidine). Such alkaloids are of considerable interest for therapeutic purposes (Watson et al. 2001) and DMDP has been patented as a potential nematocide (Birch et al. 1993). DMDP extracted from other plant species has been shown to deter feeding insects and interfere with many aspects of nematode activity (Birch et al. 1993). Whether such activity occurs in nature has not been tested, but alkaloids such as DMDP can be released into soil by producer plants, or ingested and taken up by other unrelated species. Because of their inhibition of glycosidases, they may act as a plant defence, making C. rotundifolia indigestible (Nash, Watson & Asano 1996).

Laticifers are a characteristic taxonomic feature of the Campanulales and associated taxa (Lammers 1992). Latex is present in all plant parts, and is assumed to have a protective role. Although not demonstrated in Campanula, Oppel, Dussourd & Garimella (2009), working with Lobelia cardinalis (also Campanulaceae) showed that caterpillars ‘trenched’ leaves to deactivate plant defences before feeding distal to the cuts, and that this prevented outflow of viscous latex and alkaloids held within the latex matrix. They further showed that latex was not a static defence, but that caterpillars feeding on the proximal side of trenches would find latex delivered to mouthparts from pressurized laticifers. Not only do such latexes provide protection from browsing, but they also act as a self-healing mechanism. The latexes of C. latifolia and C. glomerata coagulate (through evaporation and formation of amide bonds) after 2–7 s, whereas those of Vinca, Euphorbia and Ficus take up to 20 min (Bauer, Nellesen & Speck 2010). The latex of C. rotundifolia becomes sticky and viscous within one minute (Wilson, unpubl. data.).

VII. Phenology

Plants generally overwinter as frost-resistant green rosettes, with the characteristic cordate leaf morphology. As the spring season progresses, plants make slow vegetative growth, spreading by ground-hugging stems through surrounding vegetation. Flowering is relatively late, and erect flowering stems with sessile linear leaves develop from June onwards, with first flowers opening typically in July. Dates of first flowering (FFD) in the British Isles were reported in the Royal Meteorological Society’s Phenological Reports throughout the 58 year period from 1891 to 1948 (Jeffree 1960). The mean FFD of C. rotundifolia for the whole of this period was day 190.8 (9–10 July). Although it shows regional variation in FFD by meteorological district, this variation (10 days) was the smallest of the eleven species in the long-term study. The earliest flowering districts were S. Ireland, S.E England and E. England, and the latest were N. Ireland and W., E. and N. Scotland. Elsewhere, in a three year study in Norway at 61° N, Wielgolaski (1999) recorded a considerably earlier mean first flowering date (19 June). In a common garden study in Utrecht, using plants from different European localities Gadella (1964) found that diploids commenced flowering ‘late’ (between 1 and 23 July), hexaploids sometimes flowered early (11 June), but mostly late (30 June–20 July), whereas tetraploids showed the widest span of first flowering dates (from 30 May to 17 July).

Each flowering stem produces flowers over several weeks. In the British Isles, peak flowering is in late July-August, though occasional flowers may occur as late as November.

VIII. Floral and seed characters

(A) Floral biology

Campanulas are insect pollinated. Their floral structure facilitates cross-pollination and impedes self-pollination and there is also self-incompatibility. The main features of the floral structure were first described by Sprengel (1793), with later additions by many authors (see Shetler 1979). Flower structure and function in C. rotundifolia are similar to those most frequently found in the Campanulaceae, described in Campanula by Knuth (1909). The details of flower development and pollination in C. americana L. were described by Barnes (1885), and the mechanism in C. rotundifolia appears to be identical. Flowers are protandrous, releasing pollen before the stigma is receptive. The pollen is usually shed just before the flowers begin to open.

The pollen grains of C. rotundifolia are spheroidal and porate. Like the majority of species in the genus, these pores are arranged around the circumference of the grain (Nowicke, Shetler & Morin 1992). Data from Geslot & Medus (1971) indicate that the number of pores ranges from 3 to 6, with a tendency for higher pore numbers in hexaploid than tetraploid material. The surface is covered by short spinules, interspersed with a distinctive pattern of ridges (Dunbar 1973). Sizes reported by different authors are similar but vary according to preparation and sampling methods. Hubac (1972) determined pollen size at anthesis; diploid pollen was predominantly in 31.3–33.1 μm size classes, whereas tetraploid and hexaploid pollen occupied 39.3–40 μm and 43.5–45.2 μm size classes respectively.

In the flower bud, the anthers are appressed to the style which bears numerous epidermal hairs (‘pollen collecting hairs’) at the level of the anthers. Before the bud opens, the anthers dehisce and shed their pollen onto these stylar hairs. As the bud opens the anthers wither and the style with attached pollen elongates and is exposed. The stigmatic surface is still completely enclosed at this stage. In a flower in this condition nectar is secreted by the nectaries on top of the ovary and an insect probing for this between the expanded fringed bases of the filaments would be likely to collect pollen on its body by coming into contact with the stylar hairs.

As the flower ages, the stylar hairs invaginate. The remaining pollen is therefore no longer held by the hairs and, in the wild, is removed by the action of wind and rain. In the protected environment of a greenhouse the somewhat sticky pollen remains on the style but the slightest touch usually causes much of it to fall off. At about this time the three lobes of the stigma, which have previously been appressed together, begin to curl backwards from the tip of the style exposing the inner stigmatic surface. At first the lobes are patent, the stigmatic surfaces facing the mouth of the flower, but later they become recurved and may touch the style. Once the stigma lobes have recurved, pollen could fall onto them from the style and, by curling back to touch the style, the stigma lobes might come into direct contact with the pollen. However, in the wild, and frequently even under glass, all the pollen has usually dispersed before the stigmatic lobes become recurved.

Flowers visited by insects move quickly from the male stage to the female, and to seed set, whereas flowers which are not visited remain in each stage for much longer. Nyman (1993) showed that the stylar hairs (‘pollen collecting hairs’) were an important mechanism controlling stigma spreading and promoting cross-pollination. Tactile stimulation of the hairs shortened the length of the male phase of the flower and accelerated stigma maturation. In the US, Bingham & Orthner (1998) found that stigma receptivity was of greater duration in alpine than in foothills plants, and Giblin (2005) comparing alpine and montane populations, found that they varied in their floral longevity, associated with an extended female phase in alpine flowers, which were pollen-limited.

It has often been suggested on mechanistic grounds that if cross pollination does not occur then the recurving of the stigmatic lobes self pollinates the flower (Knuth 1909; Clapham, Tutin & Moore 1987). It has even been suggested that fertilization may occur through the stylar hairs. However, Gadella (1964) could find no evidence that either of these processes, which were put forward purely on the basis of flower structure, lead to fertilization. Many observers (summarised by Shetler 1979) have attempted to determine whether self-pollination occurred and have shown that flowers rarely set seed without insect visitors.

In addition to the flower structure not being conducive to self-pollination, most of those who have worked on the breeding system of C. rotundifolia have come to the conclusion that all three cytotypes are almost totally self-incompatible (Block 1964; Gadella 1964; Kovanda 1970). However, Bielawska (1964, 1968) reported that all Polish tetraploids examined were somewhat self-compatible and in certain cases highly so. Furthermore, McAllister (1972) found that of 15 tetraploid plants of British origin tested, two set viable seed. Both of these were from isolated habitats where self-compatibility could be of selective advantage. British hexaploids showed a higher degree of self-compatibility, but seedlings had poor survival and showed anomalous structures.

According to Knuth (1909), a very wide range of insects including Coleoptera, Diptera, Hymenoptera and Lepidoptera visit the flowers of C. rotundifolia. Bees, however, are the most prominent visitors as might be expected of a large blue bell-shaped flower. Near Malham Tarn, Yorkshire, Goyder (1983) observed high numbers of ‘foreign’ pollen grains on stigmas of C. rotundifolia, comprising 12–26% of the pollen present. These pollen grains were predominantly Potentilla, Cirsium and Rhinanthus, which did not correspond with adjacent flower species within a 5 m radius. Other species studied (Geranium robertianum, Potentilla erecta, Euphrasia confusa and Veronica chamaedrys) showed much smaller numbers of foreign grains, which were predominantly from Cirsium, suggesting that bees visiting Campanula were generalists.

In the Netherlands, the oligolectic bee Melitta haemorrhoidalis Fabricius was an important pollinator, increasing seed set considerably in comparison with other floral visitors (Hoffmann 2005). In Switzerland, C. rotundifolia is an important or exclusive source of pollen for several species of solitary bees (Müller et al. 2006). Depending on bee species, pollen from between 7 (for Chelostoma campanularum Kirby) and 66 (for Melitta haemorrhoidalis) intact flowers are required to rear a single larva. The oligolectic bee Chelostoma rapunculi Lepeletier is a specialist of Campanula species and its larvae fail to thrive when there is no Campanula pollen available (Praz, Mueller & Dorn 2008). In Germany, this species has a foraging distance of about 200 m (Gathmann & Tscharntke 2002). In the absence of other pollinators and with self-incompatibility, the structure of vegetation and presence of suitable nesting habitat for pollinators in proximity to food sources will be important in seed set and species survival.

In an evaluation of the behaviour of pollen collecting bumblebees (Bombus spp.) in Switzerland, Cresswell & Robertson (1994) found that pollen was a more valued resource than nectar, and that bees could evaluate pollen availability of a flower without landing.

Observations from Sweden, found that in bees’ visual spectrum, the floral colouration of Campanula species is mimicked by flowers of the red helleborine (Cephalanthera rubra) (Nilsson 1983). Males of two Chelostoma species, which normally visit Campanula, were found to incorporate Cephalanthera in their patrols, facilitating pollination of this species.

(B) Hybrids

Gadella (1964, 1966) performed many interspecific crosses within the genus Campanula. Within subsection Heterophylla he reported that the diploid C. cochleariifolia was not crossable with any of the cytotypes of C. rotundifolia. In contrast, Bielawska (1964) reported limited interfertility between C. cochleariifolia and tetraploid C. rotundifolia, although fewer than half the progeny were the expected triploids, the majority being tetraploid, presumably resulting from unreduced gametes of C. cochleariifolia. Crosses between lowland and montane C. rotundifolia and between these and ‘C. polymorpha’ and C. scheuchzeri were also made (Bielawska 1964) and no breeding barriers found among them. Without giving any data, Böcher (1966) states that arctic and temperate diploids crossed ‘without any difficulty’ as did tetraploids from both regions. He also successfully crossed diploids and tetraploids. Thus the arctic C. gieseckiana and temperate C. rotundifolia are reported to be fully interfertile. McAllister (1972) performed crosses within and between different cytotypes and species. He crossed tetraploids of C. rotundifolia s.l. (i.e. including C. intercedens from North America and C. asturica from North Spain) among themselves and also with the morphologically distinct C. scheuchzeri and C. cantabrica and found no evidence of significant breeding barriers among them. It was considered that the morphologically distinct species retain their identity through ecological and geographical separation.

All crosses among British hexaploids, even from the most distant localities (Durness in north-west Scotland and The Lizard in Cornwall) were successful. Crosses between British hexaploids and a hexaploid from the Picos de Europa (probably a distinct species) in northern Spain showed a low degree of incompatibility in that one cross resulted in seed giving a low percentage germination and its reciprocal in seedlings with very poor root development. Crosses between British plants and an Alaskan hexaploid (probably C. latisepala) were unsuccessful.

Natural tetraploid-hexaploid crosses may occur as the two highest altitude plants collected on Ben Lawers, Perthshire, Scotland were pentaploid (2n = 85), presumably first generation crosses between hexaploids and tetraploids and indicating the former, or as yet undetected, presence on the mountain of hexaploids. An aneuploid with 2n = 88 was found in Glendaruel, Argyll, Scotland near an isolated hexaploid population in an area otherwise occupied by tetraploids. This aneuploid is presumably the result of a backcross between a pentaploid and a hexaploid. To investigate the likelihood of naturally produced pentaploids, a number of artificial hybridizations were made. Ten tetraploid-hexaploid crosses among plants from the British Isles yielded viable offspring and plants of five progenies were confirmed to be the expected pentaploids with 2= 85. However, capsules resulting from successful crosses always contained a high proportion of thin shrunken seed. The pentaploid progenies showed no obvious lack of vigour and some individuals produced pollen, though always in reduced quantities. Gadella (1964) similarly obtained pentaploids with both tetraploids and hexaploids as female parent, but he stated that all his pentaploids lacked pollen.

Artificially produced pentaploids were backcrossed to both parents, some crosses succeeding in both directions. They were also selfed and crossed with other pentaploids, and again some of each type of pollination yielded seed. In the backcrosses to tetraploids the F2 plants had intermediate chromosome numbers (2= 71–77), though these were often difficult to determine exactly, and intermediate numbers were also found in backcrosses to the hexaploid parents.

(C) Seed production and dispersal

Campanula rotundifolia regularly produces large quantities of viable seed in the wild. In the British Isles seed ripens between mid-August and early October. Observations in localities throughout the country show that most of the seed has been dispersed and the capsules reduced to skeletons of veins by early December though viable seed has been collected in sheltered localities as late as 14 December.

The fruit of C. rotundifolia is a dry capsule containing up to 200 seeds [typically 10–100 seeds per flower (Grime, Hodgson & Hunt 2007)]. The seeds are small, brown, smooth and spindle-shaped, from 0.69 to 1.218 mm in length and 0.383–0.527 mm in breadth. Seed mass varies from 56 to 81 μg in the tetraploids and 70–115 μg in the hexaploid (McAllister 1972). This is consistent with Thompson & Grime’s (1983) average seed mass of 70 μg (probably tetraploid).

Immature capsules contain white ovules which turn deep crimson while the capsule is still green. If collected at the crimson stage and allowed to dry within the capsule, the seeds are viable and will germinate. The seeds turn light brown on drying and the capsule gradually dries to a straw colour. The pedicel bends over after flowering so that the capsule is inverted at maturity when it opens by three valves at its base. The seeds therefore cannot fall out without some movement of the capsule. In nature the seeds are dispersed passively by the shaking of the capsule in the wind, by grazing, or rain wash. Although not specifically adapted for wind dispersal, the small light seeds may be carried by wind and have been germinated from surface debris of glaciers in Colorado (Bonde 1969). Likewise, they could be carried across exposed rock faces and open ground.

There is no doubt that most seed is deposited in the immediate vicinity of the parent plant, trapped by surrounding vegetation. Soons & Ozinga (2005) calculated that the median and 99-percentile distances for dispersal by shaking were 0.07 and 0.35 m, respectively.

(D) Viability of seeds: germination and longevity

Trudgill, Squire & Thompson (2000), re-evaluating data from Grime et al. (1981), found that the thermal constant for 50% germination was 52 degree-days above a threshold temperature of 10.7 °C. Thus in the wild, germination occurs mostly in the warmer months.

In intensive studies of seedling occurrence and soil seed banks in calcareous grassland in Derbyshire, Cresswell (1982) and Hillier (1984) found that germinable seed was present in the soil throughout the year but germination was restricted to the warmer months, especially on the colder N-facing slope. More germination was recorded in artificially created gaps than in undisturbed turf and survival for 3 years only occurred in the gaps (Hillier 1984). In more extensive studies of over two hundred sites, McAllister (1972) found few seedlings, noting that the absence of competition and the presence of a more or less permanently moist surface soil seemed essential for the development of the slow growing seedlings.

The presence of a seed bank in the soil has been identified by a number of authors (Thompson & Grime 1979; Hillier 1984; Cresswell & Robertson 1994). On a site where established plants were abundant, its seed persisted in the seedbank, with little seasonal variation (Thompson & Grime 1979). However, in a large-scale survey, there were mixed findings on seed bank type for C. rotundifolia, although the majority of observations show either a transient (less than 1 year) or short term persistent (more than 1 year but less than 5 years) seed bank (Thompson, Bakker & Bekker 1997). Seed density was very variable ranging from five to over 1000 m−2. Nevertheless, persistence may be long term, as in Sweden Milberg (1995) found that although Campanula rotundifolia plants disappeared from ungrazed grassland plots which had reverted to woodland during an 18 year exclusion of grazing, viable seed was still present in the soil, whereas seed of many of the other original grassland species had vanished.

Young seedling-derived plants are particularly vigorous. When well established vigorous plants in pots were contaminated by self-sown seedlings in the greenhouse these seedlings out-competed the established plants and replaced them. The seedlings could be identified, even after 3 years, by their vigorous white thick tap roots. Similar behaviour in the wild would lead to patches of the species containing a mixture of clones, a behaviour which would be of advantage to a self-incompatible species like C. rotundifolia. Although many wild stands seem to consist of single clones as judged from morphological uniformity, many others are clearly composed of several genotypes, individual genotypes often being recognisable by distinctions in flower colours and shape. The seedlings show the classic morphology of the Campanula genus (Shulkina, Gaskin & Eddie 2003). They lack an epicotyl and the first internodes are very short, as they are rosette-forming. Development of the adult plant from a seedling is described by McAllister (1972) and the first stage in germination is usually the emergence of the radicle, which reaches several millimetres in length before the cotyledons appear from within the testa. The testa is very soon shed. The seed gives rise to a typical dicotyledonous seedling with two heart-shaped to oblong cotyledons (Fig. 5).

Figure 5.

 Development of tetraploid seedlings of Campanula rotundifolia, showing cotyledon and first rosette leaves: (a) 2 weeks; (b) 3 weeks; and (c) 5 weeks.

XI. Herbivory and disease

(A) Animal feeders or parasites

Cattle and other grazing animals eat the upright stems of C. rotundifolia in their relatively unselective grazing.

Despite the presence of biologically active alkaloids, and latex, there are a number of arthropoda that feed on or are parasites of C. rotundifolia including mites, weevils, leaf-miner flies, gall midges, plant bugs, aphids, moths and thrips (Table 2). Many species are restricted to Campanulaceae for food plants and several species, such as the mite Eriophyes campanulae are restricted to only two species of Campanulaceae, including C. rotundifolia (DBIF 2011).

Table 2.   Invertebrates that use Campanula rotundifolia as a food plant (DBIF 2011)
AcariEriophyidaeEriophyes campanulae Lindroth
Eriophyes schmardai (Nalepa)
ColeopteraCurculionidaeMiarus campanulae (L.)
Miarus graminis (Gyllenhal)
DipteraAgromyzidaeAmauromyza gyrans (Fallen)
Ophiomyia heringi Stary
Phytomyza campanulae Hendel
CecidomyiidaeContarinia campanulae (Kieffer)
Dasineura campanulae Rubsaamen
Geocrypta campanulae (Muller)
Geocrypta trachelii (Wachtl)
HemipteraMiridaeStrongylocoris leucocephalus (L.)
AphididaeDactynotus campanulae (Kaltenbach)
Dysaphis sorbi (Kaltenbach) II
LepidopteraGeometridaeEupithecia subfuscata (Haworth)
NoctuidaePolymixis xanthomista (Hubner)
Polymixis xanthomista (Hubner) ssp. statices (Gregson)
Standfussiana lucernea (L.)
Xestia ashworthii (Doubleday)
ThysanopteraThripidaeThrips vulgatissimus Haliday

(B) Plant parasites

There is no evidence in the published literature of any association between C. rotundifolia and holoparasitic or hemiparasitic plant species. The hemiparasite Euphrasia officinalis agg. occurs in acid grassland communities with C. rotundifolia but to the authors’ knowledge no association between the two species has been reported.

(C) Plant diseases

Ellis & Ellis (1997) cite 56 species of fungi which infect C. rotundifolia, of which three specifically infect C. rotundifolia: Leptotrochila radicans (Rob in Desm.) P.Karsten (Ascomycota, Helotiales), Coleosporium tussilaginis (Pers) Kelb (Basidiomycota, Pucciniales), and Puccinia campanulae Carmich (Basidiomycota, Pucciniales). A full list of fungal associates is available from the Ecological Flora of the British Isles database (Fitter & Peat 1994). The majority are found on dead stems and leaves. In Britain both tetraploids and hexaploids have been found in the wild infected with the rust fungi C. tussilaginis and with P. campanulae.

X. History

Within the British Isles, the first published record is Gerarde’s (1597) well illustrated account of Campanula rotundifolia, which is described as growing ‘wilde in most places of England, especially upon barren sandie heathes, and such like grounds’.

Fossil C. rotundifolia or Campanula-type seed and pollen occur in the palynological record, though they are never abundant. Fossil C. rotundifolia seed has been found in deposits dated to c. 29 000–33 500 BP in Sourlie, western Scotland (Bos et al. 2004), 37 400 years BP in Leicestershire (Bell et al. 1972) and 42 000 BP in Worcestershire (Coope et al. 1961). Although not identifiable to species, Campanula-type pollen occurs in post glacial deposits (Late Devensian) in Britain, primarily in association with species-rich grassland communities. For instance, in Scotland, pollen has been found in samples dated to 11 115 years BP from the Abernethy Forest, Inverness-shire (Birks & Mathewes 1978), from 11 355 and before 10 200 BP from Fife (Whittington, Edwards & Caseldine 1991; Whittington, Edwards & Hall 2001). Ince (1983) recorded C. rotundifolia in cores collected in North Wales (Llyn Llydaw and Cwm Cywion), dated to about 10 000 BP and Whittington, Edwards & Hall (2001) report a single grain of Campanula-type pollen from the Isle of Lewis in strata between 9250 and 10 250 BP.

XI. Conservation

Although C. rotundifolia is a widespread and locally very common species there is some evidence from the UK and across Europe that it is in decline. The New Atlas of the British and Irish Flora (Preston, Pearman & Dines 2002) reports no significant change in the range of C. rotundifolia between records from pre-1970; 1970 to 1986 and 1987 and 1999. There are however, some instances where it has been recorded in either the first or second time periods but was not present in later ones. This is particularly true in central England, south-west England and north Scotland. The majority of losses in the south-west of England appear to between the period pre-1970 and 1970 to 1986 whereas losses in the north of Scotland tend to be between the 1970 to 1986 and 1986 to 1999 time periods. These losses may be particularly damaging to hexaploid populations.

The Botanical Society of the British Isles (BSBI) Local Change Survey (Braithwaite, Ellis & Preston 2006) analysed the changing distribution of plants in Great Britain between two time periods, 1987–1988 and 2003–2004. This survey records a relative change of −11% and a change factor (which can be considered a relative measure of change at the edge of the species’ distribution) of −21%. Between 1998 and 2007 The Countryside Survey also reported a negative trend. Here, a change index (relative change in frequency between 1998 and 2007) of −0.39 was reported (Carey et al. 2008). Further evidence of recent decline can be found in the Plantlife Common Plant survey which shows an average reduction in the Domin score recorded for C. rotundifolia of 0.5 across 38 plots containing it that have been recorded annually between 2000 and 2008 (Plantlife, unpubl. data.).

These losses are likely to be due to a number of factors: agricultural intensification, especially increased use of artificial fertilisers; reduced land management leading to reversion of grassland to scrub and woodland, and disturbance, but there is some evidence linking decline in occurrence and cover of C. rotundifolia with atmospheric nitrogen deposition (Stevens et al. 2010, 2011). The species’ intolerance of competition with vigorous grasses appears to be an important factor (Stevens et al. 2006) and it is thought to be vulnerable to N deposition in Britain in both acid grasslands and heathland (Maskell et al. 2010).

As this species commences flowering in mid July, with seed set and ripening from August onwards, mid-summer mowing of roadside verges may prevent regeneration and spread from seed.

Restoration schemes often attempt to introduce C. rotundifolia and present a number of challenges, related directly or indirectly to agricultural intensification and deposition of atmospheric nitrogen (Walker et al. 2004b) and after-effects of fertilizer application (Smits, Willems & Bobbink 2008). Soil fauna development is also important, De Deyn et al. (2003) testing the role of soil invertebrates in grassland succession, found that soil fauna from secondary grassland succession suppressed early successional dominant plant species and enhanced the abundance of subordinate species and later successional species, including C. rotundifolia. A meta-analysis of the success of restoration schemes in lowland Britain (Pywell et al. 2003), found that C. rotundifolia performed consistently poorly and (Walker et al. 2004a,b) identified C. rotundifolia as being one of the species at high biological risk from the introduction of non-local genotypes in land restoration in the UK. This is especially so, given the strong spatial structuring of tetraploids and hexaploids in the UK that we highlight, and in this context the possibilities of minority cytotype exclusion (e.g. Husband 2000) require evaluation.


C.J. Stevens is funded by a Leverhulme Early Career Fellowship. Julia Wilson was funded by the Centre for Ecology & Hydrology, through project NEC04202. H.A. McAllister acknowledges supervision by Dr. D. Briggs during his PhD and receipt of a Carnegie scholarship during the first year of his research. The authors are grateful to the CEH Biological Records Centre for the provision of Fig. 1 and to Plantlife for providing Common Plant Survey data. Thanks to A.J. Davy, C.D. Preston, M.C.F. Proctor, D.T. Streeter and M.B. Usher for helpful comments on the manuscript and C.D. Preston for drawing Fig. 2.