Portal de Periódicos da CAPES

Geranium purpureum Vill.

2004; Wiley; Volume: 92; Issue: 4 Linguagem: Inglês

10.1111/j.0022-0477.2004.00909.x

1365-2745

Richard Tofts,

Botany, Ecology, and Taxonomy Studies

Geraniaceae, Geranium L.; subgenus Robertium Picard; section Ruberta Dumortier. Little-Robin. A leafy annual (occasionally also reported as biennial), 10–35 cm (occasionally more), usually branched from the base, bright or dark green, sometimes slightly reddish-tinged, fragile, with a strong disagreeable smell. Stems decumbent or ascending, ± clothed with dense hairs below, often nearly glabrous above but very variable in the indumentum. Leaves palmate, bright green, lower mostly with five leaflets, often narrow, with scattered adpressed hairs on both sides, on long petioles; leaflets 1.5–8.5 cm, ± ovate, deeply pinnatisect, the segments pinnately lobed; upper leaves mostly ternate, smaller, on shorter petioles. One to five extended internodes produced on the main axis with the first flowering branches arising at the upper nodes of this portion and the later ones arising from the lower nodes of the extended stem and upper nodes of the rosette. Pedicels 2–27 mm, ascending after flowering, mostly straight, pilose. Flowers in pairs (a cymule) but individual flowers sometimes aborted. Sepals five, ovate, mucronate or shortly aristate, ± pilose and glandular, erect and somewhat connivent, near apex, without ridges, calyx not angled. Petals five, 5–9.5 mm (including claw), lamina 2.5–3.5 × 1.5–2.5 mm, purplish-pink. Undehisced anthers yellow, pollen yellow. Mericarps 2.3–3 mm, reticulately ridged, usually with 3–5 pronounced overlapping collars or ridges at apex, glabrous or hairy. Mean mass of mericarp ranges from 1.61 mg to 2.05 mg (Cornish plants examined by the author) or 2.5 mg (reported by Yeo 1973). Seed ovoid, smooth, brown, 2.0 × 1.2 mm (Butcher 1961). Two subspecies have been reported from Britain although they are not separated by some authorities. Ssp. purpureum. Erect to ascending growth form (although may be more or less prostrate in exposed situations), generally on rocky limestone soils. Ssp. forsteri (Wilmott) H. G. Baker. Prostrate, tips of shoots ascending, growing on shingle beaches. This subspecies will adopt an erect habit if grown in crowded or shaded conditions, but the stem has an upright zig-zag growth form rather than the single vertical stem of ssp. purpureum (Pratt 2004). Little-Robin is a native plant of hedgebanks, walls, cliffs, shingle, railway ballast, grassland and disturbed habitats. It is much rarer than G. robertianum (Herb Robert) which is similar to G. purpureum but has larger petals, orange or purple anthers and mericarps with 0–1(−2) deep collar-like ridges at the apex. Great Britain represents the most northerly area in the range of Geranium purpureum where it reaches 52° N. The British distribution is shown on Fig. 1. Wigginton (1999) reported it from Cornwall, Devon, Somerset and Hampshire. Preston et al. (2002) showed records from the Channel Islands, southern Ireland (see below) and, as an alien, from the Isle of Man and a location by the south coast close to the Kent–Sussex border. Geranium purpureum has recently been reported from Lewes railway station (D. Streeter, pers. comm., reported in Briggs et al. 2001). It is thought to be contracting in its British range although the Cornish population has increased considerably in the past 20 years. Older records are known from Dorset (Baker 1955), Gloucestershire and Carmarthenshire (Perring & Farrell 1983). Geranium purpureum is thought to be extinct in the Isles of Scilly (Lousley 1971). In Ireland, G. purpureum has been reported from Cork and Waterford (O'Mahony 1985) although plants at the latter locality appear to show some characteristics associated with G. robertianum. It usually occurs in lowland situations close to the sea, although inland localities may be up to around 100 m. Geranium purpureum ssp. forsteri currently appears to be restricted to Hampshire (Wigginton 1999). Pratt (2004) does, however, suggest that Sussex plants may be referable to ssp. forsteri. The British distribution of the two subspecies is shown by Baker (1955). A distribution map of ssp. forsteri is given by Perring & Sell (1968). The distribution of G. purpureum in the British Isles. (○) Pre-1950; (•) 1950 onwards. Each dot represents at least one record in a 10-km square of the National Grid. Mapped by Henry Arnold, Biological Records Centre, Centre for Ecology and Hydrology, mainly from records collected by members of the Botanical Society of the British Isles. Outside the British Isles, G. purpureum is widespread in southern and western Europe. It is considered to have been expanding its range for some years in northern Switzerland (Huber 1992). It occurs eastwards through the Mediterranean to Turkey, Iran and parts of Africa (Fig. 2). Some continental botanists consider G. purpureum to be a subspecies of G. robertianum. Approximate world distribution of G. purpureum, based mainly on Hultén & Fries (1986) and Baker (1957) but also using other literature sources referred to in the text. Base map copyright © Houghton Mifflin Co. In Italy, G. purpureum occurs at altitudes of up to 1200 m in calcareous, sunny, rocky places whilst the closely allied G. robertianum occurs in shady places at up to 1600 m (Pignatti 1982). In Greece, G. purpureum is described as a lowland species being restricted to altitudes below 1200 m (Strid 1986) in contrast with G. robertianum which occurs up to 2100 m. In Cyprus, G. purpureum is present in damp, shaded rocky places in maquis at altitudes of up to around 1400 m (Meikle 1977) but G. robertianum is lacking. In Madeira where it is more common than G. robertianum, G. purpureum is widespread from sea level to 1000 m, occurring usually in shade on grassy banks and slopes, in woods, along paths and levadas (Press & Short 1994). Geranium purpureum is also known from the Azores, Balearics, Corsica, Sardinia and Sicily (Fl. Eur. 2) and the Canaries (Yeo 1973). In Iran, G. purpureum occurs in northern and north-western parts at altitudes of up to 1500 m (Parsa 1951). It is also noted from Syria, Lycia (south-west Turkey) and Lebanon (Boissier 1867). In Palestine, where plants agree best with G. purpureum var. modestum (Jord.) Rouy, it is fairly common in batha (a dwarf shrub formation) and maquis (Zohary 1972). It is reported from parts of Russia including Crimea and the Caucasus, especially along the coastlines (Shishkin & Bobrov 1974) where it occurs in open and dry parts, in contrast with G. robertianum which is present in shady places. It has also been noted from North Africa and close to the Red Sea and Indian Ocean coasts. It is naturalized in New Zealand and Australia where it grows on the central-western slopes area of New South Wales (Hnatiuk 1990) and also in South America where it occurs on the east and west coasts and the Juan Fernandez islands (Baker 1957). Its distribution is given as Mediterranean-Atlantic by Preston & Hill (1997). A plant of warm, open places. The recently expanding Cornish range of G. purpureum has tentatively been associated with recent mild winters (Preston et al. 2002) and the distribution of G. purpureum in Britain suggests that it is influenced by winter cold. Plants are restricted to locations where the January mean temperature is at least 4 °C and are most common in those limited areas of Britain where mean January temperatures equal or exceed 5 °C (temperature map in Hill et al. 1991). There is no strong relationship between the British distribution of G. purpureum and summer temperature, although Baker (1957) did consider that the British distribution of G. purpureum was explained by the double effect of a high temperature requirement for flowering and fruiting and the frost sensitivity of the species. The prevalence of G. purpureum in southern Europe, where it appears to be much more frequent than G. robertianum, suggests that it is favoured in Mediterranean climates, where the summer drought exerts a controlling influence on the vegetation, by its faster growth rate and annual life history. Baker (1957) found in comparative studies that G. purpureum was better suited to the Mediterranean climate of California than was G. robertianum but that in Leeds the reverse was found. Geranium purpureum occurs in rocky or stony places, on cliffs, dry hills and hedgebanks or in stabilized areas at the rear of shingle beaches (Wigginton 1999). Artificial substrates such as railway ballast are sometimes colonized (see, e.g. Huber 1992; Yeo 2003). It is often found in a rich soil (Baker 1955; see also VI(E)). Geranium purpureum ssp. purpureum and G. purpureum ssp. forsteri have both been recorded from sand dunes (Radley 1994). In Ireland, it occupies open sunny areas on old walls and roadsides, especially near the sea. Six of the Irish records listed by O'Mahony (1985) were on or at the base of limestone walls, with single records from a limestone quarry, sandstone wall, hedgebank and a pathway. Of the 11 sites reported by Curtis & McGough (1988), 10 were from Cork City and four of those were lost as a result of building development and road widening during the period 1970–88. In Cornwall (see Table 1), G. purpureum was found to be growing in soils of pH from 5.8 to 7.8. Hill et al. (1999) give G. purpureum a pH indicator value of 6 (between moderately acid and weakly acid, to weakly basic), an identical value to that given for G. robertianum. Hill et al. (1999) give G. purpureum a soil moisture value of 3 (a dry site indicator) compared with a value of 6 (an indicator of sites between moist and damp) for G. robertianum. Where Geranium purpureum grows on stabilized areas behind shingle beaches, it occurs in a very open community with Atriplex species, Geranium robertianum, Glaucium flavum, Lathyrus japonicus and Silene uniflora. Where it grows on cliffs, dry hills, hedgebanks, rocky and stony places, associates include Centranthus ruber, Dactylis glomerata, Galium aparine, G. mollugo, Geranium robertianum, Hedera helix, Lonicera periclymenum and Rubus fruticosus agg. (Wigginton 1999). Owing to its scarcity, G. purpureum does not feature as an associate in any of the NVC communities (Rodwell 1991a,b, 1992, 1995, 2000). Quadrat data are given for Cornish populations of G. purpureum in Table 1. Here, G. purpureum was most frequently encountered on the edge of scrub or in rough grassland. All locations had been subject to some degree of disturbance through clearance of vegetation or movement of soil. Associates occurring at a frequency of II (i.e. 21–40%) or more were, in order of decreasing frequency, Galium mollugo, Hedera helix, Dactylis glomerata, Rubus fruticosus agg., Galium aparine, Arrhenatherum elatius, Prunus spinosa, Brachypodium sylvaticum, Plantago lanceolata, Anisantha sterilis, Bromus hordeaceus ssp. hordeaceus, Umbilicus rupestris, Chaerophyllum temulentum, Elytrigia repens, Festuca rubra, Geranium robertianum, Holcus lanatus, Leucanthemum vulgare and Urtica dioica. Baker (1957) noted the following species occasionally growing in association with G. purpureum ssp. forsteri: Agrostis stolonifera, Anagallis arvensis, Arrhenatherum elatius, Atriplex prostrata, Beta vulgaris ssp. maritima, Cerastium diffusum, Cirsium arvense, C. vulgare, Convolvulus arvensis, Crepis capillaris, Elytrigia repens, Galium aparine, Geranium molle, Glaucium flavum, Holcus lanatus, Poa annua, Rumex crispus, Senecio jacobaea, S. vulgaris, Solanum dulcamara var. marinum, Suaeda maritima, Taraxacum agg. and Tripleurospermum maritimum. In mainland Europe, Geranium purpureum is in some localities restricted to ground under trees where it avoids the effects of grazing by stock (Herrera 1991). In Crete it grows in calcareous woodland, olive groves, gorge beds and among rocks (Chilton & Turland 1997). In the Agri Irini Gorge (Crete), it occurs both in dry, open ground associated with species such as Anagallis arvensis ssp. foemina, Briza maxima and Sedum album and also seasonally damper, slightly more shaded ground, associated with species such as Allium subhirsutum, Anthemis arvensis, Biscutella didyma, Geranium pusillum, Leontodon tuberosus, Orlaya kochii, Psoralea bituminosa, Thymus capitatus and Veronica cymbalaria. Poldini (1989) reported G. purpureum from Asplenietum trichomano-rutae-murariae subassociation calaminthetosum (Potentillion caulescentis) as a differential species with Calamintha nepetoides and Euphorbia fragifera, Micromerio-Euphorbietum wulfenii (Centaureo-Campanulion), Alysso alyssoidis-Sedetum albi and Frangulo rupestris-Prunetum mahaleb (Berberidion). Although Geranium plants are described as 'rabbit-proof' by Mabberley (1998), Herrera (1991) found that 90% of experimental G. purpureum transplants exposed to vertebrate herbivores (cattle, red deer Cervus elephas, fallow deer Dama dama and rabbits Oryctolagus cuniculus) were consumed well before reaching reproductive size whereas 60% of transplants protected from grazing set at least one fruit. Mean seed production was 0.9 per plant in the former case and 7.4 in the latter. In a separate experiment that did not involve transplanting, only 20% of unprotected plants survived grazing to produce a mean of 3.0 seeds and a dry biomass of 69 mg per plant whilst 75% of protected plants survived to produce a mean of 30.6 seeds and a biomass of 388 mg per plant (Herrera 1991). Highly gregarious in some locations such as Doñana, Spain, where G. purpureum grows under Juniperus turbinata (Herrera 1991) at an average density of 3.0 plants m−2. In Cornwall, most 1 × 1 m quadrats recorded for the present study contained only single G. purpureum plants but, on occasion, Domin cover values of up to 5 (11–25% cover) were recorded as a result of localized aggregations of plants. See VI E(i) Light, below. No evidence of damage to the cotyledons of seedlings was noted when exposed to a frost of −3 °C, although winter temperature does appear to affect the distribution of G. purpureum in Britain (see II(A)). None of the plants grown by Baker (1957) was frost-hardy. Laciniate leaves were produced in the rosette stage in a cold autumn or spring, whereas an especially cool summer period led to the production of laciniate leaves in the inflorescence and tiny flower buds that did not open (Baker 1957). Drought resistance does not appear to have been investigated in any detail. Geranium purpureum is able to flourish in seasonally very dry environments but this seems to result from the timing of germination and growth to avoid the effects of drought rather than by tolerating it. Plants of Geranium purpureum show various morphological differences, particularly with regard to growth habit (erect or prostrate) and degree of hairiness. Yeo (1973) reported that, under certain conditions, G. purpureum exhibits a scrambling habit with flowering stems about 1 m in length supported by the reflexed petioles which remain turgid for some time after the lamina has died. Flowering branches arise from the axils of the uppermost rosette leaves, beginning with an internode bearing a pair of opposite leaves at the top. In the axils of these leaves arises a further pair of branches and, between these, a pair of flowers (a cymule). Flowering finally ceases when neither of the two axillary buds flanking a cymule grows out. The leaves tend to become gradually reduced in size and complexity of lobing further up the stem. Yeo (1973) gave a detailed description of the petiole of G. purpureum and compared its anatomy with that of allied species including G. robertianum. Morphological information about G. purpureum is limited, but much that applies to G. robertianum (Tofts 2004) is likely to be applicable to the closely related G. purpureum. The upper leaf surface lacks stomata. Stomata occur on the lower (abaxial) leaf surface at a density of 41–164 mm−2 (mean = 107.4 mm−2, SD = 36.7 mm−2). No information, although likely to be associated often with VA mycorrhiza as is the case in the closely related G. robertianum (Tofts 2004). A winter annual which behaves as a ruderal species (Herrera 1991); reportedly it may grow as a biennial in Britain (Fl. Br. Is.). Seed sown during the autumn in Britain begins to germinate shortly afterwards. Figure 3 shows the proportion of seed germinating over a 78-day period from sowing in early September. Plants grown from seed sown in spring during the present study invariably flowered during the same year (see VI E(i), below). Cumulative proportion of seeds germinating over a period of 78 days after sowing outdoors in Gloucestershire, UK, on 9 September 2000. Bars show one standard error. 2n = 32 (Fl. Br. Is.). Hill et al. (1999) give G. purpureum a light value of 7 (plants generally in well-lit places but also occurring in partial shade). This contrasts with a value of 5 (semi-shade plants, rarely in full light, but generally with more than 10% illumination when trees are in leaf) for G. robertianum. Significant differences in G. purpureum biomass were recorded for plants grown in a garden experiment under three different light regimes (full sunlight, semi-shade (25% sunlight) and shade (12.5% sunlight)). Plants were grown from seed in February 2001 and harvested five months later. Mean dry biomass per plant under each treatment differed from that under all other treatments (anova, P < 0.001) with mean biomass values (g) of 1.12, 2.30 and 1.86, respectively. These figures compare with mean biomass values (g) of 2.02, 2.50 and 2.21, respectively, for G. robertianum grown at the same time under identical conditions (Board 2001). Over the period of this experiment, full sunlight resulted in a proportionately greater reduction in biomass (compared with the maximum under partial shade) of G. purpureum than of G. robertianum contrary to what might be expected from the data given by Hill et al. (1999). The effects of light treatment on leaf number and flowering in this experiment are discussed in VII. Baker (1957) found that plants of the prostrate G. purpureum ssp. forsteri would exhibit an erect habit if grown in shaded conditions. Such plants resembled ssp. purpureum apart from the broader leaf segments, poor development of vegetative anthocyanin and reduced flowering (see also introductory notes regarding the differences between the two subspecies). The internodes and petioles were comparatively extended and the basal leaves persisted into the flowering stage. Reported as frost sensitive (see II(A)). Reported to be often designated as nitrophilous by Herrera (1991), G. purpureum is, however, given a nitrogen indicator value of only 3 (associated with more or less infertile soils) by Hill et al. (1999). This contrasts with a value of 6 for G. robertianum (associated with soils of between intermediate and rich fertility). The 19 associated species recorded at a frequency of II or more (see III) have nitrogen indicator values of 4–8 (median = 6, mean = 5.7) and it therefore seems possible that the value of 3 given to G. purpureum is too low. Little information available, but likely to be similar to G. robertianum (see Tofts 2004). Bate-Smith (1973) reported the presence of the flavonoids quercetin and kaempferol in G. purpureum. Geranium purpureum flowers from May to September. Baker (1957) found seeds of G. purpureum ssp. forsteri that germinated in Leeds in the autumn flowered the following summer at almost the same time as plants from seed sown during the February of the year of flowering. In Spain, G. purpureum is a winter annual. By January, plants have two or three tiny leaves and the cotyledons are still attached. Peak flowering occurs in March and seed dispersal and senescence in May (Herrera 1991). Baker (1955) noted that, in cultivation, G. purpureum ssp. forsteri flowers about 2 weeks later than G. purpureum ssp. purpureum. Comparison of G. purpureum and G. robertianum plants grown under identical conditions over a 5-month period showed that the former develops much more swiftly, although total dry biomass may be significantly less (Board 2001). Figure 4 shows the rate of leaf development in G. purpureum compared with G. robertianum under three light treatments. At the end of the experiment, plants of G. purpureum grown under shade or semi-shade did not differ significantly in their number of leaves (mean number = 64.7) whilst those grown in full sunlight produced significantly fewer leaves (mean = 39.3). Plants of G. purpureum grown from seed sown in February 2001 were all in flower by 24 June 2001 irrespective of light treatment. This contrasts with G. robertianum where only 8% of plants were in flower by the same time and only 17% of plants by the end of the experiment on 21 July 2001. In a separate garden experiment, seeds of G. purpureum sown in November completed their life cycle and had died by the following July. By contrast, 94% of plants of G. robertianum sown as seed at the same time were still alive in July (R. Tofts, unpublished data). For plants grown in Leeds, Baker (1957) found that G. purpureum from various cultures had finished flowering by the end of July whilst G. robertianum was still flowering strongly at that time. Mean number of G. purpureum leaves in plants grown in full sunlight ('sun'), 25% ambient light ('semi-shade') and 12.5% ambient light ('shade') over a period of approximately 5 months. Twenty replicates per treatment. Bars show one standard error. Data from G. robertianum (taken from Tofts 2004, Figure 4) are shown for comparative purposes. Similar to Geranium robertianum (see Tofts 2004), G. purpureum can facultatively self-pollinate. Yeo (1973) stated that the flowers of G. purpureum are more adapted to self-pollination than those of G. robertianum owing to the shortness of the lamina, the keel on the claw being so feebly developed that distinct nectar passages are not formed and the flowers being homogamous. The pollen grain to ovule ratio is approximately 60. Almost all flowers set fruit and on average there are 4.6 seeds per schizocarp (Herrera 1991). The pollen grains are 62–93 µm diameter, spherical and opaque. The hybrid G. robertianum × G. purpureum might occur where both parents are present on maritime shingle and has been reported from a hedgebank in east Cork. It is possible that G. purpureum ssp. forsteri arose from this cross (Stace 1997). Baker (1955) suggested introgression from G. robertianum in Gloucestershire (Clifton Down), Carmarthenshire (Newcastle Emlyn), Torquay area (Bere Ferrers), north Cornwall and Dungarvan (Ireland). In open pollinated experimental mixtures of G. purpureum and G. robertianum, Baker (1957) found no evidence of hybrids in the progeny although hybrids have been synthesized in both directions that seem indistinguishable from G. robertianum apart from their complete sterility (Stace 1975). Five mericarps (Fig. 5) are produced per schizocarp, although one or more may be abortive, especially late in the season. The mericarps typically have dense wrinkle-like ridges and 2–3(−4) deep collar-like ridges at the apex, contrasting with sparse fine ridges and 0–1(−2) deep collar-like ridges at the apex for G. robertianum (Stace 1997), although some variation in these characters is evident (compare with Figure 5 in Tofts 2004). In plants grown in Leeds, Baker (1957) found that many flowers produced fewer than five viable seeds. Herrera (1991) reported the average number of seeds per fruit to be 4.6 ± 0.18 (SE). Plants protected from vertebrate grazing by Juniperus turbinata were found to produce 0–18 fruits (mean = 4.0, median = 3), whilst unprotected plants produced 0–6 (mean = 0.6, median = 0). Seed dispersal occurs by ballistic discharge over a distance of 0–160 cm, with 50% of seed landing ≤ 44 cm from the parent plant (Herrera 1991). Each fruit bears a rigid awn 13–15 mm long and an elastic sticky thread, both fused to the central axis of the fruit. On ballistic release, both mericarp and thread are carried together so that seeds often become stuck to the spot where they land. Yeo (1973), however, reported one plant in the progeny of seed from Oporto that discharged mericarps without the threads attached although it appeared normal in other respects. It has been conjectured that the discharge of mericarps with attached strands may facilitate horizontal and upward dispersal which may, in turn, assist colonization of cliffs, walls and banks (Yeo 1973). 1–11: Abaxial views of mericarps from various populations, hairs omitted for clarity. 1. Between Trevelver and Carlyon, Cornwall. SW957752. 2. Near Tregenna Farm, Cornwall. SW965747. 3. Near Penmayne, Cornwall. SW953757. 4. Little Petherick, Cornwall. SW920721. 5. Near Tregonce, Cornwall. SW929742. 6. Near Padstow, Cornwall. SW922747. 7. Near Penquean, Cornwall. SW950737. 8. Little Petherick, Cornwall. SW920720. 9. Near Penquean, Cornwall. SW950737. 10. Omalos, W. Crete. 11. Agri Irini Gorge, W. Crete. 12–15: Stages in seedling germination and development. Seed collected from Cretan plants (Agri Irini gorge), dried and sown in garden conditions 4 months after collection showed germination of 48–50%. Seed from the same batch (kept in dry and dark conditions at room temperature) sown 12 months later showed germination of 25–32%. O'Mahony (1985) found seeds of G. purpureum to remain viable for at least seven years. No scarification is required for germination. The apparently thicker fruit coat of G. purpureum compared with G. robertianum may be responsible for the seemingly poorer germination of the former species (Baker 1957). The germination of G. purpureum seed was found to be slightly lower than that of G. robertianum (57–70%) when kept under identical conditions in the present study. No ecological differences between plants with glabrous or hairy seeds were noted by Baker (1957). Yeo (1973) found that plants grown from seed sown in March flowered in June of the same year, but plants from seed sown in June or July did not flower until March or April of the following year. The stages in seedling development are shown in Fig. 5. Germination is epigeal. Reproduces entirely by seed, normally as an annual although a biennial life history has been reported occasionally. Vertebrate herbivory (coupled with limited powers of dispersal) is believed to cause the aggregated distribution of Geranium purpureum in parts of Spain (Herrera 1991). Spur-thighed tortoises Testudo graeca L. ingest and defecate viable seeds of annuals including Geranium species and have been suggested as potential dispersal agents by Herrera (1991). No data are available on invertebrates associated with G. purpureum, although they are likely to be similar to those associated with G. robertianum (see Tofts 2004). The greater rarity of G. purpureum may, however, result in comparatively fewer encounters with parasites. No data, but may be similar to those associated with G. robertianum (see Tofts 2004). Baker (1957) found G. purpureum seedlings to be more susceptible to 'damping-off' than those of G. robertianum. No material of G. purpureum is recorded by Godwin (Godw. Hist.). The date of the first British record is unclear, but Baker (1955) cites herbarium material of ssp. purpureum (Cockbush Common, West Sussex) dating from 1833 and material of ssp. forsteri (Stoke's Bay, South Hampshire) dating from 1829. Geranium purpureum was formally described in 1785. Before this date, the species is likely to have been passed over as a form of G. robertianum. Geranium purpureum is classified in the lower risk category as Nationally Scarce in the British Red Data Book (Wigginton 1999). Geranium purpureum ssp. forsteri is classified under the high threat category in the Biodiversity Action Plan for Hampshire (Hants BAP). It is not threatened in Europe. I am indebted to Mrs Margo Board for making available the results of a comparative experiment using Geranium robertianum and G. purpureum. The staff in the Fielding-Druce herbarium (Oxford) kindly made much of the literature available to me and Nina Nijhof-Vaags provided assistance with some of the German literature. I am very grateful to Professor Arthur Willis and the Associate Editors for improvements to an earlier version of this manuscript and for assistance with several of the references. Jonathan Silvertown and Verina Waights kindly allowed access to facilities at the Open University.