Portal de Periódicos da CAPES

Coincya wrightii (O.E. Schulz) Stace ( Rhynchosinapis wrightii (O.E. Schulz) Dandy ex A.R. Clapham)

2000; Wiley; Volume: 88; Issue: 3 Linguagem: Inglês

10.1046/j.1365-2745.2000.00477.x

1365-2745

Stephen G. Compton, Roger S. Key,

Plant Taxonomy and Phylogenetics

A biennial or short-lived perennial herb. Stem up to 1.3 m high, pubescent, hairs appressed. Leaves pinnatifid-lyrate, up to about 45 cm in length and 13 cm in breadth, pubescent with ascending hairs. Flowering stems erect, branched, woody below in older plants. Inflorescence loose racemose, with generally fewer than 10 flowers open simultaneously. Petals large, up to 9 mm across and 10.5 mm long, with a 9.5-mm claw; petals and petal veins yellow. Sepals up to 8 mm long, erect until after anthesis, pubescent, yellow, with yellow-green veins. Pedicel ascending, mainly between 25 and 45° from vertical when bearing buds, and between 45 and 90° from vertical when flowers are open; siliqua mainly straight. Distal end of valves obtuse. Beak straight to curved; length of mature siliqua up to 60 mm, with beak up to 20 mm. Valves dehiscent, beak indehiscent. Seed length 1.2–1.9 mm. Mean air-dry mass of the seed is 1.52 mg (n = 20, range 0.85–2.53 mg). Tap root slender for its length. Smell strong, cabbage-like. Other vegetative and floral characteristics are given in Table 1. No recorded variation in flower colour or other discrete phenological characters, but some variation in flower size has been noted (Leadley & Heywood 1990). Coincya (Brassicaceae) was synonymised with Rhynchosinapis by Leadley & Heywood (1990) and currently includes six species and several subspecies. It is distinguished from Brassica and Sinapis by the combination of the upper stem leaves not clasping the stem, valves of the fruit having several prominent parallel veins, a slightly flattened beak, large petals and erect sepals (Rich 1991; Stace 1997). In contrast to these genera, Coincya is probably monophyletic and distantly related to other genera in the subtribe Brassicinae (Warwick & Black 1993). There are, however, few clear morphological characters (Leadley & Heywood 1990) and chloroplast DNA characters (Warwick & Black 1993) to distinguish the species belonging to Coincya. Coincya has a western European distribution, with a concentration of taxa in the Iberian Peninsula. Five of the six species have very local distributions and most are regarded by Leadley & Heywood (1990) as being relict taxa. The majority of the species are associated with either rocky slopes or maritime sands, though others occur in alpine meadows and ruderal habitats. Coincya wrightii (Lundy cabbage) –‘loose and sprawling, but bright in hue’ (Gosse 1874) – is a native British species endemic to the island of Lundy in the Bristol Channel, where it is confined to a 2500-m length of coastline. Coincya wrightii was first distinguished by O.E. Schulz (as Brassicella wrightii) in a publication by Wright (1936). Previously it was considered as a form of Coincya monensis (L.) W. Greuter & Burdet or Brassica oleracea L. The widely distributed and highly variable C. monensis differs from C. wrightii in longevity (it is generally annual), basal leaf shape and the extent of stem hairiness (it is generally less hairy) (Leadley & Heywood 1990). Lundy is situated in the Bristol Channel, some 18 km from Hartland Point, Devon (Fig. 1). It comprises a central plateau bounded by cliffs and steeply sloping ‘sidelands’, with a total area of around 430 ha and a highest point of 141 m a.s.l. Coincya wrightii has a current altitudinal range of around 2–120 m, from the splash zone to the top of the cliffs. It extends inland to a maximum of around 300 m in Millcombe. The distribution and abundance of Coincya wrightii on Lundy in 1997, mapped on a 100-m grid. Coincya wrightii occurs mainly along the southern and central parts of the eastern coast of Lundy, with a few plants also in Lametry Bay on the south coast (Fig. 1). In 1997, it was present within a rectangular area extending 2500 m N-S and 600 m E-W. The plant is widely distributed within this area, in 1997 being present in 60% of the 100 × 100 m squares which include at least some land. Early records (Gosse 1874; Wright 1936) implied that the distribution of Coincya wrightii was limited to the loose, weakly metamorphosed Devonian slates that form the extreme south-east of the island, which are replaced by Eocene granites from Ladies' Beach northwards. More recent records (Marren 1972; Irving 1984; Farrell 1993) have shown that plants occur as far north as Knight's Templar Rock and in some recent years more individuals were present on the granites than the slates. The apparent range expansion this century is probably not genuine, as most of the populations in the northern part of its range are on near-vertical cliffs that are partly or wholly hidden from above. Small temporary populations appear at the edge of the plateau and on some of the more gently sloping sidelands. Stock exclusion experiments have confirmed that these populations fail to flower owing to vertebrate grazing pressure (S.G. Compton & R.S. Key, personal observation). Stock have been maintained on the island for centuries, so this does not imply any recent contraction in range. While the range of the plants has been relatively stable during recent years, the size of the populations has apparently fluctuated widely, from a recorded low of 324 plants in flower in 1978 to over 10 000 in 1998 (Cassidi 1980; S.G. Compton, R.S. Key & R.J.D. Key, personal observations). Surveys have varied in intensity, however, and it seems likely that the earlier counts missed many plants growing in the more inaccessible locations. Furthermore, the proportion of plants that flower varies considerably between years, and counts based only on plants in flower can be misleading. The climate of Lundy is mild, with warmer winters and cooler summers than the mainland, reflecting its westerly and maritime location. Annual rainfall averaged 1150 mm during the period 1971–92 with the lowest monthly average of 52.2 mm in July (Gibson 1993). Coincya wrightii occurs typically in exposed, open situations, but persists under moderate shade from trees, Ulex europaeus, Prunus spinosa and the fern Pteridium aquilinum. It is found mainly on bays cut into the sea cliffs, but also colonizes steeply sloping sidelands, rocky outcrops and the sides of well-vegetated gulleys, most prominently in Millcombe. It rapidly colonizes land-slips, which are a regular feature on the slates in the south of its range. Coincya wrightii is favoured by, but is not restricted to, south- and east-facing slopes. The northern limit of the plant does not correspond closely with changes in geology, as its range extends throughout the first and slightly into the second of the two granite types distinguished by Dollar (1941). Differences between the granites are considered to be minor (Edmonds et al. 1979; Thorpe et al. 1990) and changes in coastal topography towards the north of the island, leading to easier access by grazing mammals, may be the factor limiting the distribution of the plant. Some apparently suitable areas nonetheless remain uncolonized. In the south, the factors limiting its distribution appear more clear cut, as exposure increases strikingly away from the east coast, which is sheltered from the Atlantic gales, and this is associated with general changes in the vegetation on the cliffs. Coincya wrightii typically occurs on moderate to steep slopes, where it is rarely subject to waterlogging. Soils on Lundy are well-drained, loamy podzols (Dawes 1979). They are characterized by the Soil Survey of England and Wales as belonging to the Moretonhampstead Series in the south, the Moorgate Association in the north and the Hexworthy Series on the cliffs (Dawes 1979; Edmonds et al. 1979). The soils show considerable local variation in acidity, with pH values ranging from 4.9 to 6.2 on the slates and 2.9 (beneath Rhododendron ponticum) to 6.5 on the granite (S. Richardson, personal communication). Soil nitrate values are highly variable, with no clear relationship to underlying strata, ranging from less than 1 to 140 mg L−1 soil filtrate (based on soil samples mixed with an equal volume of water, Richardson et al. 1998). A series of 25 irregularly shaped vegetation plots, averaging around 64 m2 in area, and sited so as to include as far as possible the full range of vegetation types occupied by the plant, contained 82 associated vascular plant species. Rubus fruticosus, Sedum anglicum, Teucrium scorodonia and Umbilicus rupestris were the most frequently occurring associates, with many of the other species recorded only rarely and/or at low cover values (Table 2). The association with R. fruticosus is likely to reflect the protection from grazing mammals afforded by its prickles. Within its narrow coastal range, the plant appears to occupy almost the full range of plant communities present. A classification of the 25 plots using Ward's Method (performed by SPSS 6.1; SPSS Inc., Chicago, USA) distinguished three main groupings (Fig. 2). A broadly similar classification was obtained when average linkage (between groups) was employed. The first group of vegetation plots distinguished by the classification (between plots 1 and 8 in Fig. 2) comprises communities found on inland granite outcrops, sea cliffs and a large, active rockslide on the slates above Victoria Beach. It is characterized by large areas of bare substrate, and a high species richness (Table 3). The second group of plots (between 9 and 17 in Fig. 2) include more heavily vegetated sea and quarry cliffs, and the deep valley of Millcombe. The highest average cover values of Coincya wrightii are recorded from this vegetation group, together with a relatively small coverage by bare ground. Pteridium aquilinum and the climbers Lonicera periclymenum and Hedera helix are also frequent. The third group of plots (between 18 and 25 in Fig. 2) were situated on the slate cliffs in the south of the plant's range, the relatively gently sloping sidelands, and a hillock dominated by Ulex europaeus. Coincya wrightii was never abundant in these plots, where grass cover was consistently 50% or higher. A classification of the plant communities on Lundy that include Coincya wrightii. The cliff communities with Coincya wrightii do not figure in the National Vegetation Classification (Rodwell 1993–2000), but fit adequately into the coastal extremes of communities W24 (Rubus fruticosus–Holcus lanatus underscrub) and W25 (Pteridium aquilinum–Rubus fruticosus underscrub), particularly W25b, the Teucrium scorodonia subcommunity. MC12a Festuca rubra–Hyacinthoides nonscripta cliff bluebell communities also appear to be represented (A.J.C. Malloch, personal communication). The communities containing Coincya wrightii are dynamic in terms of their turnover of species and changes in total vegetation cover. This is particularly true of the early successional communities developing on unstable shales (Fig. 3), and granite cliffs (Fig. 4), but is also evident in gorse-dominated inland populations (Fig. 5). Vegetation changes on the 1994 landslip above Victoria Beach: (a) changes in late-May/early-June percentage cover (dark area vegetated) between 1994 and 1997; and (b) percentage cover for different plant species within the vegetated area, based on visual estimates for each year; C = Coincya wrightii, G = combined grass species, D = Digitalis purpurea, Se = Sedum anglicum, T = Teucrium scorodonia, Sc = Scrophularia scorodonia, n values indicate the remaining number of species present. Vegetation changes on granite cliffs at Quarry Bay: (a) changes in the late-May/early-June percentage cover (dark area vegetated) between 1994 and 1997; and (b) percentage cover for different plant species within the vegetated area, based on visual estimates for each year; C =Coincya wrightii, G = combined grass species, R =Rubus fruticosus, n values indicate the remaining species present. Vegetation changes on an inland hillock just north of Millcombe: (a) changes in the late-May/early-June percentage cover (dark area vegetated) between 1994 and 1997; and (b) percentage cover for different plant species within the vegetated area, based on visual estimates for each year; C =Coincya wrightii, G = combined grass species, U =Ulex europaeus, n values indicate the remaining species present. Coincya wrightii is fairly tolerant of competition, persisting under Ulex europaeus and even seeding into deep litter beneath stands of Pteridium aquilinum. Seeding experiments have, however, shown that it cannot persist in the dense sward which characterizes some less steep areas of the sidelands. It also does not survive under the alien shrub Rhododendron ponticum, which poses a considerable threat to its survival. Rhododendron ponticum is increasing its range on the eastern sidelands of the island, where it forms almost monospecific stands and is colonizing the previously Pteridium aquilinum-covered sidelands, occupying crevices in rocky outcrops, and spreading down onto the cliff face from the upper sections of the sea cliff. Without control, it appears capable of encompassing most, if not all, of the habitat range of Coincya wrightii (Compton & Key 1998; Compton et al. 1998, 1999). Randall (1978) considered that Coincya wrightii occurs in areas ‘which are mostly inaccessible to grazing animals’, a conclusion borne out by recent surveys and stock exclosures, which have shown that ‘sink’ populations are present on the gentler slopes of the sidelands where reproduction is only possible once grazers are excluded. Some protection is afforded to plants growing within dense Rubus fruticosus, which can provide effective barriers to sheep and goats. There is no record of regular human utilization for food; the leaves are overpoweringly cabbage-scented. Most visitors pay no attention to the plant, despite its proximity to the main road on the island, and much of the population is found on steep and often inaccessible cliffs. Picking or uprooting is therefore not considered as a problem. The plant is rarely solitary, with populations typically comprising scattered individuals separated from their nearest neighbours by less than 5 m, and some plants contiguous. The most isolated individuals are found in atypical situations, such as on a fallen tree trunk in Millcombe, or on a rock exposure surrounded by a large, closed rhododendron canopy. There is no simple relationship between abundance, gregariousness and vegetation succession. In late May 1997, an early successional population on a 1994 land slip occupied 6.0% of the 252 0.25 m2 squares in an 18 × 14 grid. A mean of 1.27 ± 0.46 (SD) plants were present per occupied square (range = 1–2 plants, n = 19 plants of which 17 were flowering or had flowered). In an 8 × 8 × 0.25 m2 grid positioned on a more mature sward nearby, 40.0% of the 0.25 m2 squares were occupied. There was a mean of 2.09 ± 1.31 plants in those 0.25 m2 grid squares that were occupied (range = 0–5, n = 48 plants of which 24 were in flower). Subpopulations are generally separated by less than 100 m (Fig. 1), with the largest break around 400 m, separating the most northerly 10% or so of the plants from the remainder. The tallest individuals are found in the sheltered conditions of Millcombe, where they can reach over 1.3 m in height; they are shorter on the sea cliffs and inland rocky outcrops, but the plants are not markedly stunted even in the most exposed situations. Numbers of flowers produced are correlated with overall plant size, with some individuals flowering when less than 30 cm tall. Soil disturbances favour Coincya wrightii and it is an effective coloniser of open ground, which is produced and maintained within its range by the instability of the substrate in the south, and the steepness of the granite outcrops and sea cliffs further north. Rabbits, larger grazers, streams and human activities also contribute to small-scale sward disturbance. One large-scale land slippage, above Victoria Beach in May 1994, was immediately colonized by C. wrightii which had survived being carried down with the soil and rocks. Seedlings were already present by September the same year. Continuing disturbance in this area through to 1998 severely inhibited plant colonization, and C. wrightii has remained one of the most numerous colonists. A slightly smaller slippage occurred approximately 100 m away in March 1997, and C. wrightii was again amongst the first species to colonize. Lundy has a mild, maritime climate, where frosts are typically infrequent and mild, yet Coincya wrightii growing in gardens on the mainland can survive frosts as severe as − 15 °C. Adult plants appear to be drought tolerant, although the severe drought of 1995 did coincide with high late-summer mortality rates among mature plants in Millcombe. The mortalities nonetheless occurred only after a period of intense flowering activity had been completed, and may have been the result of the latter, rather than drought per se. Subsequent recruitment of young plants was exceptionally high, suggesting that drought may favour C. wrightii through reduced competition. Consistent with this pattern, high population levels were recorded in 1993, a year preceded by several warm and dry summers. Long slender tap root extending to almost 2 m. Stem erect, branched above, woody towards base, sparse to densely pubescent, covered with simple appressed hairs. Leaves deeply pinnatifid-lyrate with sinuations increasing in depth towards the petiole, the lower leaves broadly ovate with broadly triangular terminal lobe and rectangular lower lobes, the margins broadly sinuate with broad acute, antrorse teeth; the upper leaves becoming progressively broadly triangular, deeply divided into triangular to linear sinuate-dentate lobes with smaller acute antrorse teeth; often with an alternate pair of obovate to linear sinuate lobes separated from the main lamina adherent to the petiole. Stomatal densities on the lower side of leaves average 480 mm−2 (range 418–593 mm−2, n = 10); on the upper side 233 mm−2 (range 187–281 mm−2, n = 10). Examination of the roots of five plants from Lundy showed extensive colonization by ‘fine endophytes’ of the Glomus tenuis type. Also present were some coarse endophyte (typical VA) hyphae, but no arbuscules or vesicles were seen (D.J. Read, personal communication). Chamaephyte, also hemicryptophyte in years when winter conditions are severe. Longevity can reach at least five years (Wright 1936). Reproduction by seed only. Diploid, 2n = 24, as in most congeners (Wright 1936). None available. None for Coincya wrightii. Leaf wax and seed chemistry has been investigated in Iberian congeners (Vioque et al. 1995). Flowering typically commences in May, peaks in early June, and persists at low levels through to September or October. Peak flowering is later in small plants flowering for the first time. Dehisced fruits are recorded between June and October. Flowering within racemes is moderately well synchronized, typically few new flowers being produced subsequent to the first dehiscence. Synchronization between racemes is variable. Germination is not strictly seasonal, but most abundant in autumn. Some autumn-germinating plants produce flowers the following summer. Outbreeding, but self-compatible (Wright 1936; S.G. Compton, unpublished observations, but see Leadley & Heywood 1990). Two of the stamens have shorter filaments (with the top of the anthers the same height as the stigma) and four (2 × 2) have longer filaments with anthers which project beyond the petals and become reflexed as they mature. The shorter stamens mature slightly later. The flowers are conspicuous, bright yellow, with a well developed platform for insects created by the claw and limb of each petal at right angles. The smell is sweet. Lateral and median nectaries are well developed, with contrasting external morphology, possibly producing differing nectar concentrations (Eisikowitch 1981). The nectaries are partially protected by a calyx tube of about 8 mm, which initially restricts nectary-access to very small insects or those with long tongues. As the flower fully opens, alternate sepals bulge outwards at the base and the sepals detach from each other, allowing insects, including ants, access to the nectaries from the sides. Honey bees are absent from Lundy, but a wide variety of other bees, as well as flies, butterflies and moths, and beetles visit the flowers and are potential pollinators (Table 4). Wright (1936) suggested that the adults of Meligethes spp. (pollen beetles) were the major pollinators. Meligethes viridescens adults are by far the most frequent floral visitors, often becoming covered in pollen, but they may not be efficient pollinators. They also feed on the petals and flower buds, as well as pollen, and are minor pests of cruciferous crops such as oilseed rape (Winfield 1992). Leadley & Heywood (1990) hypothesized that two other beetles (Ceutorhynchus contractus var. pallipes and Psylliodes luridipennis) might also contribute to pollination, but neither species appears to visit the flowers (R.S. Key & S.G. Compton, unpublished observations). Analogy with similarly flowered crucifers suggests that wind pollination may also be significant (Free & Nuttall 1968). The range of C. wrightii does not overlap with any congeners, and no natural hybrids are known. It can be crossed experimentally with other diploid Coincya, the hybrids being vigorous (Da Sikka 1940; Harberd & McArthur 1972; Leadley & Heywood 1990). Large plants can produce several hundred fruits, each containing on average 18 seeds (Table 1). Around 95% of the seeds originate in the siliqua, where up to 42 seeds may be present, compared with a maximum of 3 seeds in the beak. The siliqua is dehiscent, throwing the seeds some distance from the plant, whereas those in the beak are released only as the fruit wall decays. Gómez-Campo (1977) regarded this two-part fruit structure as broadening Coincya's seed dispersal in both time and space, and interpreted a perceived expansion in the relative size of the beak among Iberian taxa (relative to the perceived primitive condition in Coincya wrightii) in terms of the development of a more ‘K-selected’ strategy. Seed dispersal can exceed 30 m, based on the presence of seeds in the soil throughout the 60 m width of bracken-covered sideland separating populations of C. wrightii on inland and sea cliffs (R. Straker, personal communication). Flocks of linnets (Acanthis cannabina) are attracted to the fruits, and manage to catch some of the seeds as they break them open. They may disperse the seeds more widely, if they do not destroy all those seeds that they eat. The absence of dispersal across the sea seems surprising, given that the beaks float, and seeds immersed in seawater for up to 7 days can remain viable (Rich 1999). Seeds collected directly from plants in July did not germinate immediately (a germination rate of less than 1% was recorded by day 31 when seeds were kept on moistened filter paper at room temperature). In contrast, a germination rate of 83.3% by day 31 was recorded for seeds collected by hand from the soil in January 1996 and maintained at 20 °C. Light is not required for germination. Soil germination trials indicate that germination is rapid under suitable conditions, with around 25% of those seeds that germinate doing so within 7 days and over 80% within 21 days at 20 °C (Fig. 6; R. Straker, personal communication). The longevity of seeds in the soil can be at least 3 years, based on annual photographic records showing the reappearance of an isolated group of plants in 1997 in an area where no flowering had occurred in the previous 4 years. Supporting evidence for a persistent seed bank is provided by viable seeds in soil samples collected from areas where no flowering had occurred for several years (Table 5). Coincya wrightii germination rates (mean ± S.E.) at 20 ± 2 °C. Methods followed Ter Heerdt et al. (1996). n (source sites) = 3, total number of germinating seeds = 360. Seedling development is illustrated in Fig. 7. Cotyledons are longitudinally folded around the radicle (orthoplocal) as in other Coincya species (Leadley & Heywood 1990). Seedling development in Coincya wrightii. Left to right: 4, 7, 14 and 28 days after germination. Sheep and ponies are currently farmed on the island and feral populations of rabbits, soay sheep, goats and sika deer are present. These may all eat Lundy Cabbage, but this is confirmed only for rabbits, goats and sheep. Despite its small area of distribution, Coincya wrightii supports a large and diverse invertebrate community (Table 6). Furthermore, and uniquely in the British Isles, C. wrightii is the only host-plant for an endemic insect species, the Lundy cabbage flea beetle, Psylliodes luridipennis (Cox 1998). It is also the main host of two other beetles of uncertain taxonomic status, the Lundy cabbage weevil, Ceutorhynchus contractus var. pallipes and the blue Lundy cabbage flea beetle, currently regarded as a flightless ‘form’ of Psylliodes napi. Plant mortalities owing to invertebrate damage have not been recorded, but insect densities are generally high, and occasionally plants are skeletonized by Pieris brassicae larvae. A survey of all leaves on 10 plants in May 1993 examined 706 leaves, of which 42.6% bore leaf mines caused by the larvae of beetles and flies, 84.4% had ‘shot holes’ caused by the feeding of adult beetles and 4.2% had petioles mined by beetle larvae. The seed-feeding weevil Ceutorhynchus assimilis may have the most significant effect on the reproduction of the plant, as it destroyed around 16% of the seeds in 148 fruits collected in 1993. Forty-six per cent of the fruits also showed signs of damage by exophytic invertebrates. Cassidi (1980) noted increased levels of leaf and fruit damage in the relatively sheltered areas around Millcombe, but so far as is known all the more common insect feeders are distributed throughout the range of their host, and the 1993 survey of seed damage found no significant differences between the Millcombe, Quarry Bay and Victoria Beach populations (16.0, 13.4 and 19.0%, respectively). Observed disease symptoms are similar in appearance to those produced by powdery mildew (Erysiphe cruciferarum Opiz ex Junell), downy mildew (Peronospora parasitica (Pers.) Fr.) and dark leaf (Alternaria brassicae (Berk.) Sacc. and A. brassicicola (Schw.) Wiltshire) on oil seed rape (A.D.A.S. 1981). Heavily mildew-diseased plants were present just above high tide level at Victoria Beach in 1994 and 1995. The possibility that Coincya wrightii was introduced to Lundy by man from the Mediterranean was raised by Pugsley (1936). The temporary ownership of the island by the Knights Templars in the late 12th century, with their strong mediterranean connections, might have been one way in which the plant could have been introduced. However, there is no evidence that they ever occupied Lundy, thanks to the refusal of the de Marisco family to give up the island (Langham 1994). Another possible source of introduction was ‘Algerine’ (Barbary) pirates who used the island as a harbour and shelter in the 17th century (Langham 1994). Contrary to the hypothesis of human introduction is the absence of records from elsewhere and the presence of beetle taxa apparently associated exclusively with the plant. These have been felt to confirm the plant's long association with Lundy (Wright 1933). It is of interest that it was the distinctiveness of the beetles which convinced Dr Elliston Wright that the plants were a separate species, subsequently confirmed by Professor O.E. Schulz (A.J. Willis, personal communication). It seems unlikely that Coincya wrightii diverged in situ on Lundy from a more widespread mainland ancestor, as it is not closest morphologically to the British form of C. monensis, the only congener native to Britain. Lundy became an island as postglacial sea levels rose, and was an estimated 10 times its current area about 9000 years bp (Schofield & Webster 1991). Wright (1935) concluded that the flora and fauna of Lundy are essentially oceanic, with no evidence of any effective postglacial land bridge to the mainland. Leadley & Heywood (1990) considered C. wrightii to be a relict taxon, and suggested that the very mild climate of Lundy might have allowed the island to act as a refuge during glaciations. However, Lundy would not have benefited from the ameliorating effects of the North Atlantic Drift during the last glacial period (Coope 1986), and there may have been extensive sea ice in the area, making survival of the plant seem highly unlikely. Alternatively, colonization of Lundy may have been from source populations in the Iberian Peninsula, as Coincya has a strong concentration of taxa in that area, and Leadley & Heywood (1990) noted that C. wrightii shares indumentum and leaf-shape features with forms of C. monensis in north-west Spain that are absent among other congeners. The plant may have been transported shortly after the retreat of the ice, at which time ocean currents with ‘slicks’ of fresh glacial meltwater flowed northwards along the west coast of Europe (Coope 1986). The conservation status of Coincya wrightii is ‘Endangered’ RDB1, and it is protected under section 8 of the Wildlife and the Countryside Act. Most of its distribution lies within a designated Site of Special Scientific Interest. Together with its host-specific herbivorous beetle Psylliodes luridipennis, it is listed in the UK Biodiversity Action Plan (UK Steering Group 1995). There is evidence of soil nitrogen enrichment adjacent to streams draining the south-east of the island, owing to fertilizer applications, but only a small proportion of the cabbage population is found in these areas (Richardson et al. 1998). Action plans have been prepared for the conservation of Lundy cabbage and its associated insects, and for the control of Rhododendron ponticum on Lundy (Compton & Key 1998; Compton et al. 1998). Thanks to Rosy Key, Lynne Farrell and Tim Rich, and to wardens Andrew Gibson, Emma Parkes and Liza Cole, for their help during the course of our research and to all the other residents of Lundy Island for making our visits possible and pleasurable. Help with insect determinations was provided by Drs C.M. Drake, P. Kirby and D.A. Sheppard. Professor D.J. Read provided the section on mycorrhiza and Dr J. Edmonds helped with morphological descriptions. Particular thanks to Professor A.J. Willis for his patience and persistence during the production of this manuscript. Our work on Lundy cabbage was made possible by the practical assistance and financial support of the English Nature Species Recovery Programme.