Biological Flora of the British Isles: Salvia pratensis

This account presents information on all aspects of the biology of Salvia pratensis L. (Meadow Clary) 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, herbivores and disease, history and conservation. Salvia pratensis is an erect, rosette‐forming, perennial herb with a broad native distribution covering much of Europe—from the British Isles, Spain and Morocco in the west, across Europe into Asia, as far east as the Urals. In the British Isles, the species is nationally scarce, confined to a few south‐ to west‐facing sites with calcareous soils in Southern England and one site in Wales. It is predominately found in unimproved pasture, hay meadows and grassy verges, but can occur on the fringes of scrub or woodland. Although the species is abundant in central Europe, changes to land management since the mid‐20th century have resulted in fragmented and threatened populations in several European countries. It is cultivated as an ornamental, as is S. × sylvestris, the hybrid with S. nemorosa. Populations are typically gynodioecious, having both female (male‐sterile) and hermaphrodite individuals at variable proportions. The species has a mixed mating system and is self‐compatible via insect pollination, but predominantly outcrosses. Honeybees and bumblebees are abundant pollinators, but a diverse range of bee species and other insect species visit S. pratensis flowers. Inbreeding depression has been documented, presenting a conservation concern for small, fragmented populations. The species is the focus of conservation efforts and has been reintroduced to sites where it had become locally extinct in Britain. To sustain favourable habitat, site management should maintain low soil nutrient levels, and prevent scrub encroachment and the dominance of coarse grasses. The removal of sward by hay cutting or grazing after plants have flowered and set seed is advised, in addition to maintaining a degree of disturbance to provide bare patches of soil for seedling recruitment.

suggest a lower value of 32 native hectads. Discrepancy between these figures is due to difficulty in ascertaining the true origin of the populations from records alone. All other hectad records ( Figure 1) are naturalised sites resulting from the escape from garden cultivation, or introduction via grass seed contaminants (Rich et al., 1999). Rich et al. (1999) make particular reference to the late F I G U R E 1 The distribution of Salvia pratensis in the British Isles. Each dot represents at least one record in a 10-km square of the National Grid. Mapped by Colin Harrower, Biological Records Centre, Centre for Ecology and Hydrology, mainly from records collected by members of the Botanical Society of the British Isles, using Dr A. Morton's DMAP software 1800s and early 1900s when the species was introduced via the import of foreign grain. This is thought to explain the origin of a vast number of naturalised populations, commonly found on noncalcareous substrata.
The species is considered native in 10 vice-counties across Central and Southern Britain (Table 1), including East and West Kent, Oxfordshire, West Gloucestershire, Buckinghamshire, South Wiltshire, Surrey and Berkshire (Druce, 1886(Druce, , 1897(Druce, , 1926Grose, 1957;Hanbury & Marshall, 1899;Riddelsdell et al., 1948;Salmon, 1931). In addition to this, West Sussex and Monmouthshire contain populations which, based on the substrata, are most likely native (Rich et al., 1999;Wade, 1970), and possibly Middlesex from seeds washed down the Thames (Rich, 2000). The Cotswolds, particularly the vice-county of Oxfordshire, forms the stronghold for the species in Britain, where around 90% of British populations are found. In almost all of the vice-counties, at least one population was rerecorded up to 2003, with the exception of Berkshire where no populations have been recorded since 1955 (King, 2004;Rich et al., 1999).
Salvia pratensis belongs to the Temperate Element of the European flora (Preston & Hill, 1997). The species has a broad native distribution throughout Europe ( Figure 2). The southern edge of the species' native global distribution is Morocco, stretching across the Pyrenees to Turkey and across the Urals in Western Russia.
The northern limit of the species' native range is reached in the British Isles and the Netherlands in the west, across Germany and Poland, to the Baltic States and North Western Russia in the east (Hedge, 1972;Wigginton, 1999). Outlying and isolated occurrences are found in Sweden, Denmark and Finland where the species is naturalised (Gärdenfors, 2005). It has also been introduced to the United States of America, where it is naturalised in several states and included on noxious weed lists for 46 states (USDA National Plant Data Team, 2020).
In central and eastern Europe, S. pratensis can be found in abundance on roadside verges, in dry pastures and unimproved hay meadows (Hedge, 1972;Slavík, 2000). Ivanek (1984) describes S. pratensis forming over a quarter of the hay crop in Arrhenatherum elatius-associated grasslands in Poland.
The native distribution in Britain is limited to lowland hay meadows and chalk downland, and hence, the species is not commonly found at high altitudes, having a maximum altitude of around 300 m (Smith, 2019). In continental Europe, a maximum altitude of 1,920 m is recorded and, though it is not an alpine species, it is commonly found in abundance in alpine meadows (Hultén & Fries, 1986;Thompson, 1911).

| Climatic and topographical limitations
Salvia pratensis is a Southern Continental European species, generally found in warm temperate, dry environments (Hultén & Fries, 1986).
However, cultivated material is known to withstand cold winters (Yeo, 1995;see Section 5.3).
The species' restricted distribution in the British Isles is likely linked to specific habitat and microclimate requirements. British populations occur mainly on sunny, gentle-to-steep sloping ground, facing south to west, on free draining soils (Scott, 1989). The average annual rainfall from all 37 native hectads in the British Isles is 733 mm (Preston et al., 2002), a low total compared to the national average per hectad. This coincides with a low Ellenberg value for moisture of 3, indicating the species' preference for drier sites. The mean January temperature in native hectads is 3.7°C, fitting within the normal distribution for native plant species in the British Isles. The mean July temperature in hectads is 16.3°C, among the highest for species native to the British Isles. An Ellenberg value for light requirements of 8 indicates that it is found in situations where light is generally >60% relative illumination in the summer months (Hill et al., 2004). Sites with such climatic and topographical characteristics provide a warm, dry microclimate in which the species has managed to survive at the edge of its global range in the British Isles. Ellenberg (1988) stated the following climatic indicator values for Central Europe: a water value of 4, between dry and moist sites; a temperature value of 6, between fairly warm to warm conditions; a light value of 8, a light-loving plant; and a continentality value of 4, suboceanic.

| Substratum
In the British Isles, Salvia pratensis is a calcicole, found on nutrientpoor, calcareous, free draining, sandy soils derived from parent rock of oolitic limestone or chalk (Scott, 1989). Long-recorded populations generally considered native occur only on dry calcareous substratum, most commonly the Oxfordshire strata of Jurassic limestone, on soils where the rock is near to the surface (Lousley, 1950).
During a survey of British S. pratensis, Scott (1989) Salisbury (1952) that dry, nutrient-poor soil characteristics are of greater importance than a basic pH.

| COMMUNITIE S
In the British Isles, Salvia pratensis is predominately a grassland plant of lightly grazed or unmanaged open grasslands on weakly to strongly calcareous soils (Rich et al., 1999;Rodwell, 1991 et seq.;Scott, 1989) (Hegland et al., 2001). It is placed with grey dune plants in the class of Koelerio-Corynephoretea (Hegland et al., 2001;Schaminée et al., 1996). In Central and Eastern Europe, S. pratensis is frequently found on roadside verges, unimproved dry pastures and hay meadows, often in abundance (Hedge, 1972;Slavík, 2000). In Central Europe, Ellenberg (1988) includes S. pratensis as characteristic of more or less calcicole arid and slightly arid grasslands and characteristic of Arrhenatherum grasslands in the lowlands to submontane habitats.

| Grazing
Salvia pratensis is known to be extremely palatable to grazing animals. Sheep and deer will graze it heavily and cattle will remove aerial plant sections down to sward height (Scott, 1989). Lousley (1950) described a population in the North Downs, Kent, where rabbit grazing had prevented flower production.
Rabbits have been observed to bite through inflorescence stalks (Lousley, 1950), but among-site comparisons of grazing by Scott (1989) suggested that they avoid the leaves. Sites heavily grazed by sheep and cattle were found to have up to 100% of flowering stems removed. By surveying plants prior to and during grazing, Scott (1989) identified selective grazing by sheep: Open flowers were initially removed, followed by unopened flowers and then flowering stems. The leaves and basal rosette will be eaten by sheep last, but at one heavily grazed site, all above-ground plant matter was eaten within 3 days. Cattle will unselectively graze all plant parts, commonly leaving the basal rosette, unless at high intensity. Grazing of flowering shoots may promote the production of adpressed vegetative shoots (Smith, 2019), although vegetative spread is generally limited. Of the few populations with yearround grazing regimes in the Netherlands, Hegland et al. (2001) found that sites with more intense grazing had smaller proportions of juvenile or progressive plants than sites with seasonal grazing or mowing regimes. High-intensity grazing can destroy basal buds, significantly impacting the survival rate to a reproductive age (Hegland et al., 2001;Scott, 1989). Intense grazing, or grazing at the wrong time of year, is therefore likely to contribute to population decline. Yet, while grazing can greatly reduce the biomass of S. pratensis, seasonal grazing is beneficial in maintaining suitable habitat and promoting establishment through soil disturbance (see Section 11). Hegland et al. (2001) found that populations of greatest viability in the Netherlands were located at sites with the most diverse vegetation cover (by species richness) and the rarest species. Open areas with more bare soil and less vegetation cover were found to support more flowering S. pratensis plants. In addition, a higher proportion of species normally associated with nutrient-rich community structures were generally found to render the S. pratensis population less viable (having a greater proportion of older plants). Competition from nutrient-loving species may contribute to site unsuitability more than high nutrient levels alone (see Section 6.5.3).

| Gregariousness
Populations in the British Isles range from a single long-lived plant to many hundreds of plants (King, 2004;Scott, 1989). A demographic survey of 10 sites by Moughan (2012) (King, 2004;Scott, 1989). Within a vice-county, it is common that only one or two sites/populations are present (Table 1). Greater densities of plants and seedlings are found at sites that experience disturbance (Hegland et al., 2001). Populations surveyed in Britain were found to be more compact in growth form when growing at high density with up to 50 plants per square metre (Rich et al., 1999).

| Performance in various habitats
The level of disturbance and hence amount of open soil was the most significant parameter affecting population structure of S. pratensis in dry grasslands in the Netherlands (Hegland et al., 2001). Here, a positive correlation between the amount of bare ground and proportion of juvenile plants (those with 1-2 leaf pairs and a rosette ≤5 cm) was documented (Hegland et al., 2001). Disturbance and bare soil are required for increased recruitment from seed and a younger more dynamic population structure. Of 51 plots surveyed in the Netherlands, Hegland et al. (2001) observed seedlings in only seven plots, six of which were in late-mown grassland. A more progressive population structure was also observed in late-mown grassland. Seedling establishment is limited in tall undisturbed grassland which tends to have an accumulation of moss and litter at ground level (Scott, 1989).

| Effect of frost, drought etc.
Salvia pratensis is given a hardiness rating of 'H1' by Compton (2011), meaning it can withstand the severest of European continental climates down to −20°C and below. Its occurrence in arid calcareous grasslands in Europe and the presence of a tap root suggests a reasonable level of drought tolerance, but there are no specific data available.

| Morphology
Salvia pratensis is an erect, rosette-forming herb, with short basal internodes. The mode of branching is monopodial, or, when the apical meristem is lost, sympodial from dormant basal buds. A deep tap root forms over time, with spreading, fibrous and yellow-tinged rootlets. In a study investigating phenotypic variation in S. pratensis populations of various sizes in the Netherlands, Ouborg et al. (1991) measured 17 morphological characteristics of first-generation plants grown from wild seed in a greenhouse (14/10 hr light/dark; 20/15°C; n = c. 300 plants). After 8 months, the longest leaf was 27.6 cm.
The average number of leaves at 60 days was 11, increasing to 39 at 8 months when the average number of basal rosettes was three.
Larger populations were found to exhibit significantly higher phenotypic variation.

| Mycorrhiza
Salvia pratensis is known to form associations with arbuscular mycorrhizal fungi (Harley & Harley, 1987), including with the ubiquitous taxa Glomus claroideum, G. geosporum, G. intraradices and G. mosseae (Zaller et al., 2011). Glomus geosporum has been shown to significantly increase the shoot and root biomass of S. pratensis in controlled growth experiments in field soil mixed with 50% sand (Zaller et al., 2011).

| Perennation: Reproduction
Salvia pratensis is a hemicryptophyte, probably capable of living 30 years or more (Rich et al., 1999). One plant originating from the Dordogne, France, has been retained in cultivation for 26 years by T.
Rich. The species reproduces by seed, but there are suggestions that it can spread vegetatively to a limited extent; Šerá and Šerý (2004) assumed that lateral spread via vegetative propagation can occur up to 0.3 m per year. Overwintering occurs as vegetative shoots in rosettes. Ouborg (1993) found that it can take 4-5 years for an individual to form a rosette large enough to produce flowers.

| Chromosomes
A chromosome number for S. pratensis in the British Isles has not been established; however, several varied figures have been recorded from a broad area. The commonly cited and currently accepted chromosome number is 2n = 18 (Hedge, 1972;Sell & Murrell, 2009;Stace, 2019).
Contrary to these findings, however, the chromosome number of 2n = 16 has also been recorded in some cases including: seeds from cultivated S. pratensis grown at Rotterdam Botanic Garden in the Netherlands (Haque, 1981;Haque & Ghoshal, 1980); plants from the Serpukhovsky area of west Russia (Patudin et al., 1975) and in the Ukraine (Slavík, 2000). Plants from the Ukraine with 2n = 16 chromosomes may be the species Salvia dumetorum (Slavík, 2000), but this taxon is believed to be better placed within S. pratensis by some (Hedge, 1972). A chromosome number of 2n = 32 has also been recorded from cultivars grown at Hamburg Botanic Garden, Germany (Scheel, 1930) and from some natural Bulgarian plants (Markova & Ivanova, 1982). Chromosome counts of n = 9 and n = 16 suggest that polyploid variants may exist. This is not unexpected as members of Salvia sect. Plethiosphace are known to hybridise (Haque, 1981;Hedge, 1972;Scheel, 1930).  6.5.2 | Response to flooding Mommer et al. (2006) investigated the effects of flooding on a range of plant species found along the River Rhine in the Netherlands.
S. pratensis was found to exhibit a significant intolerance to complete submersion in water, decaying rapidly in both light and dark conditions. Median lethal time (where 50% of plants had died) was 11 days of submersion in full light, and when light was excluded, the plant was estimated to survive <5 days. The leaves were found to have an extremely low aerenchyma content of c. 5% before submersion and c. 2% after 11-day submersion with light, with no measurable material after 11-day submersion in darkness. Although the species is severely damaged by flooding, it is able to grow in higher elevated and dry habitats on the Rhine floodplain (above 13 m with a maximum flooding period of 5 days per growing season; Mommer et al., 2006), likely due to the presence of sandy, free-draining soils.

| Response to nutrients
Salvia pratensis is characteristic of low-nutrient soils and a decline in populations due to soil improvement has been recorded in the British Isles (Rich et al., 1999). An analysis of 11 sites in England showed that sites supporting species with higher Ellenberg indicator values for nitrogen had smaller S. pratensis populations with more regressive demographic structures (Moughan, 2012). In Central Europe, Ellenberg (1988) gave the species an indicator value for nitrogen of 4, meaning it is more often found on nitrogen deficient and average nitrogen soils than on eutrophic soils. In cultivation, however, it is known to grow well in nutrient-rich potting mixes (K. McGinn pers. obs.), suggesting that the species' intolerance of high nutrient levels may in part be an indirect effect of greater competition from surrounding vegetation (see Section 4.2). Scott (1989) performed soil analyses for six populations in the British Isles and found that concentrations of potassium (K + ) were low, ranging between 0.07 and 0.55 mEq/100 ml soil, whereas an adjacent area of improved grassland where S. pratensis was absent at one site had levels around 1.3 mEq/100 ml. Some sites, particularly Stuart Fawkes Nature Reserve, Gloucestershire, had high magnesium concentrations, but an overall relationship between the distribution of the species and high magnesium concentrations was not supported (Scott, 1989).

| Biochemical data
Salvia species, particularly S. officinalis, produce essential oils that are utilised in traditional medicines (see Section 10). Anačkov counted for over 50% of substances identified and the most prevalent essential oil was (E)-caryophyllene, forming over 25% of the total essential oil volume. Veličković et al. (2003) found that the most represented components were thymol (30%), (E)-caryophyllene (28%), p-cymene (9%) and caryophyllene oxide (4%). Extracts of callus tissue from S. pratensis have been shown to exhibit antimicrobial properties (Maslova et al., 2019). While S. pratensis appears to contain a lower amount of essential oil than other Salvia species, the antimicrobial activity of its essential oils has been found to be higher than that of S. officinalis (Mossi et al., 2011;Veličković et al., 2003). Harborne (1992) identified that delphinidin 3',5'-dimethyl etherbased malvidin pigments are responsible for flower colour in S.
pratensis. This group of pigments was found in eight other species of Lamiaceae originating from the Old World, all of which possess violet-blue flowers pollinated by bees (Harborne, 1992).

| PHENOLOGY
Salvia pratensis has a main flowering period of late May-early July in the British Isles (Rich et al., 1999). However, when inflorescences are removed early or mid-season by hay cutting or grazing, secondary flowering shoots are sometimes observed, extending the flowering period into early October. Scott (1989) found that 13% of plants cut in mid-June flowered again in mid-August. Fruiting occurs around 1 month after flowering, with seeds released from mid-July (Scott, 1989). Seedlings in the wild emerge from July onwards (Smith, 2019).

| Floral biology
Salvia pratensis is polycarpic and populations are typically gynodioecious, having both female (male-sterile) and hermaphrodite individuals, the latter being more abundant (see final paragraph in this section). It is, however, possible for individuals to have both female and hermaphrodite shoots (Smith, 2019). The species has a mixed mating system and is self-compatible, but insect pollination is required for within-flower selfing (Scott, 1989;van Treuren et al., 1993). Hermaphrodite flowers are protandrous (Ouborg & van Treuren, 1994;Scott, 1989 The stamens in Salvia flowers are modified to form a spoonshaped lower lever arm which functions to dispense pollen onto pollinators' bodies (Reith et al., 2007). A hole in the spoon-shaped lever allows access for a pollinator's proboscis and the force of the proboscis entering it is converted into an inwards and upwards movement, causing the thecae to be lowered onto the thorax or abdominal regions of the pollinator (Reith et al., 2007). Pollinators then access a reward of nectar from nectaries sited at the base of the corolla.
Honeybees ( (Table 2). Salvia pratensis is said to be less specialised in its pollinators relative to other Salvia species (Reith et al., 2006). Insects are known to be capable of collecting nectar or pollen without triggering the staminal lever mechanism and therefore, not pollinating the flower (Table 2; Reith et al., 2007). Nectar robbers (which pierce the corolla to obtain nectar) and nectar thieves (which forage normally but transfer little pollen due to a mismatch in floral and insect morphology) are documented (Reith et al., 2006). Scott (1989)  The authors suggested that this factor underpins the maintenance of sexual dimorphism in the species. There are suggestions that seed set may be higher for female relative to hermaphrodite flowers, but data collected by Zhang and Claßen-Bockhoff (2019) did not show significant differences. Scott (1989) found that 1%-9% of flowers in eight British populations were female, with higher frequencies observed in fields that were shaded or subject to herbicide treatment. Intermediate flowers were observed at frequencies of 1%-13%. In four Dutch populations, female flowers were observed at frequencies of 0%-4% in two small populations (<30 individuals) and 22%-24% in two large

| Hybrids
The hybrid between S. nemorosa and S. pratensis, S. × sylvestris, is common in Central and South-East Europe where the two species are sympatric. Salvia × sylvestris is also known to have arisen in cultivation (Stace, 2019;Stace et al., 2015). In the British Isles, this hybrid is naturalised on sand dunes at Burnham Overy Saithe, West Norfolk, where it was discovered in 1953 and rerecorded as still being present in 1997 (Stace, 2019;Stace et al., 2015). Earlier records of the hybrid as a casual and garden escape are known from Hertfordshire, West Norfolk and Jersey (Stace et al., 2015). Stace et al. (2015) describe this hybrid as being partially fertile and Hegi (1927) found that 60% of its nutlets contained an embryo.
In comparison to S. pratensis, S. nemorosa has a denser inflorescence, far smaller flowers and longer, coloured (pink to purple as opposed to green) bracts (Stace et al., 2015). Salvia × sylvestris is intermediate in these characters, but can be very close to S. nemorosa, having colourful (purple, blue, pink or white) bracts, at least the lower flowers longer than the corolla, and a corolla colour similar to the bracts (Stace et al., 2015). All three taxa are grown in gardens and S. × sylvestris has been known in cultivation for several centuries (Stace et al., 2015). Salvia pratensis and S. nemorosa have been the focus of artificial hybridisation and selection by plant breeders, resulting in an array of named cultivars of S. × sylvestris.
These cultivars have been bred to combine the larger flowers of S.
There are 33 cultivar names associated with S. × sylvestris that are As a result, S. × sylvestris has become a popular and widely available garden plant.
Salvia × digenea Borbás is a hybrid between S. amplexicaulis Lam. and either S. nemorosa or S. × sylvestris (the parentage is disputed; Stace et al., 2015). Along with S. × sylvestris, these taxa form a taxonomically complex group that is often treated under S. nemorosa in horticulture (Compton, 2011).

| Seed production and dispersal
Salvia pratensis produces nutlets with an average weight of 2.4 mg (1-3.3 mg; derived from 12 '1,000 seed weight' measurements; Royal Botanic Gardens Kew Seed Information Database, 2020). Each nutlet contains a single seed. Hagemann et al. (1967) (2017) The original authors used a synonym.

TA B L E 2 (Continued)
polycarpic, perennial plants, with a small amount of vegetative lateral growth representing a medial input into reproductive success.
At six British populations, Scott (1989) observed seed set to mostly exceed 90%, but reduced to 84% in plants where flowers had been 'robbed' and to 83% where plants produced flowers in mid-August when fewer pollinators were available. Plants were found to produce up to 1,500 seeds each, but high rates of loss due to mammal predation were observed. Haque and Ghoshal (1981) found that open-pollinated plants exhibited lower seed set compared to handpollinated plants.
Each flower can produce up to four nutlets which drop from the calyces once they mature and dry (Scott, 1989). Barochory (seed dispersal by gravity alone) is considered the chief dispersal route for S. pratensis and most seeds fall close to the parent plant (Diacon-Bolli et al., 2013;Ouborg et al., 1991;van Treuren et al., 1991).
However, a range of other seed dispersal mechanisms have been documented. Römermann et al. (2005) demonstrated the potential for epizoochory (seed dispersal on the exterior of animals) via an experimental approach using dry sheep and cow pelts and a mechanical shaking device. The experiments showed that 50% of S. pratensis nutlets tested remained attached to sheep's wool after 1 hr of shaking, but only 2% remained in cattle hair (n = 100 seeds each treatment; three replicate runs each). Such dispersal potential may be higher in wet conditions. The potential for epizoochory was also described by Bouman and Meeuse (1992), whereby the sticky mucus produced by glandular hairs on the calyces may promote the adherence to animal fur, with mature seeds remaining inside the calyces. This mode of dispersal has been documented for other Salvia species (Zona, 2017). The nutlets also produce thick mucilage upon wetting which may similarly aid in dispersal by adherence to animals, although direct field-based observational evidence is lacking for Salvia species (Scott, 1989;Zona, 2017). Hydrochory has been suggested as an explanation for the distribution of Dutch S. pratensis along the river system of the Rhine (Hegland et al., 2001) and may account for the Middlesex population (Rich, 2000). Ridley (1930) stated that S. pratensis nutlets may float in water for up to 8 days and experimental work by Zona (2017) showed that seeds of Salvia species can float for many hours. In addition, Bonn (2004) found that seeds of S. pratensis can survive passage through the digestive system of sheep, but at very low rates. Dispersal methods other than barochory would suggest that the distance of seed dispersal can be much greater than that of 1 m stated by some authors. Historical long-range human-mediated dispersal has also been documented via the transport of grass seed contaminants (Rich et al., 1999).

| Viability of seeds: Germination
Salvia pratensis seed is able to germinate immediately from nutlets after wetting. Both viable and non-viable seeds produce a thick layer of mucilage from expelled epicarp cell contents 10-15 min after wetting (Scott, 1989). For ripe seed sown within 10 days of harvest, Scott (1989) observed an average germination of 72% (62%-91%; n = 596 seeds, originating from four sites), with no significant variation between dark and light treatment. Seed subject to wetting, drying and rewetting on alternate days successfully germinated. Samples reached 50% of their germination total within the first 7 days and 95% after 12 days. Underripe seeds collected while green but subject to after-ripening for several days showed some viability, with an average germination of 33% (23%-50%; n = 360 seeds, originating from three sites).
In 12 lab-based tests conducted on agar at the Millennium Seed Bank, an average of 95% germination was achieved (77%-100%; a mean of 35 seeds per test; Royal Botanic Gardens Kew Seed Information Database, 2020). These tests were conducted in a range of conditions, from 15 to 26°C and mostly with 12/12 hr light/dark.
Germination took place over an average of 8 days (S. Miles, pers.
comm., 9/12/20). All tests were conducted following storage at −20°C, but dried to 15% relative humidity prior to storage, as per normal seed banking protocols for desiccation tolerant species (S. Miles, pers. comm., 9/12/20). Ouborg and van Treuren (1995) found that a minimum seed size of 1.3 mm was necessary for germination to occur; seeds smaller than this were non-viable. Out of eight seed size classes, an optimum size of 1.4-1.6 mm was suggested as seeds within this range produced seedlings with significantly longer roots and cotyledons (n = 75 seeds per size class, but only 25 for the largest size class). In the four Dutch populations surveyed, the proportion of non-viable seeds varied between 27% and 45% and this variation was not related to population size. Seeds resulting from self-fertilisation appear to be less viable; Ouborg and van Treuren (1994) found that such seeds weighed 12% less (mean ± SE: selfed = 2.29 ± 0.09 mg; crossed = 2.61 ± 0.08 mg) and had 29% lower germination (mean ± SE: selfed = 30 ± 3%; crossed = 42 ± 4%; n = 30 seeds per treatment).

Seeds of S. pratensis appear to be orthodox (desiccation tolerant)
and storage of dry seeds in hermetically sealed containers at cool temperatures is recommended (Royal Botanic Gardens Kew Seed Information Database, 2020;Young & Young, 1992). Germination tests following 20 years of storage at the Millennium Seed Bank showed no significant loss in viability (S. Miles,pers. comm.,9/12/20).
There are, nevertheless, accounts of limited germination following seed storage, but it is important to note that seed longevity will depend greatly on factors such as storage conditions. In a sample of 30 seeds 9-11 years old, Scott (1989) found no germination. Seeds sown after 1.5-year refrigeration (c. 4°C) were found to have 5% germination (n = 524), but this seed was selfed, from a single parent plant, with no seed excluded for size (J. Moughan, unpubl. data). No soil seed bank has been recorded for S. pratensis in Northern Europe (Thompson et al., 1997); however, seedlings have been observed to emerge just before the flowering season, suggesting that seed may be capable of lying dormant for at least 1 year (Scott, 1989).
Although S. pratensis seeds are likely orthodox, Chetverikova et al. (2008) included the species in a study on seed cryopreservation.
Germination on petri dishes was consistently high (>75%) following storage for 1 month via quick deep freezing (−196°C; cooling rate of 700°C per min), programmed two-stage deep freezing (−196°C; cooling first to −40°C at a rate of 1°C per min, followed by placement in liquid nitrogen), non-deep freezing (−10°C), in addition to a control (≥5°C). However, post-germination monitoring of 30 plants grown from each treatment showed that deep freezing caused significantly reduced length and width of rosette leaves.

| Seedling morphology
Germination is epigeal. Cotyledons have a slightly notched obtuse tip, cordate base, inconspicuous veins and a hairy petiole. The cotyledon margins are entire, and during the initial days, are slightly revolute. The first true leaves are hairy and more typical of adult leaves, with crenate margins and reticulate venation (Figure 3).

| Animal feeders or parasites
Salvia pratensis is extremely palatable to mammal grazers, including cattle, sheep, deer and horses (see Section 4.1). A total of 14 invertebrate species have been recorded as feeders on S. pratensis in the British Isles, including leaf miners, root borers and gall formers (Table 2). This total will be an underestimate as sawflies and molluscs are not listed to species level in Table 2. Scott (1989) noted considerable damage by insect larvae, most notably from the suborder Symphyta (sawflies), causing loss of up to 21% of flower whorls. Various Gastropoda feed on S. pratensis and can cause mortality in damp conditions (Rich et al., 1999;Scott, 1989).
In addition, 24 invertebrate species not present in the British Isles have been documented as feeding on S. pratensis ( Table 2).
The leaf beetle species Cassida canaliculata Laich is dependent on S. pratensis as a host, on which it is said to be monophagous in Germany and carries out all stages of its life cycle (Heisswolf et al., 2005).

| Plant parasites
In total, 10 fungal parasites have been recorded on S. pratensis in Europe, including rusts, smuts and mildews (Table 3). The anther smut Microbotryum betonicae is believed to be monophagous on S. pratensis (Ellis, 2017).

| Plant diseases
No data available.

| HIS TORY
The earliest known record of S. pratensis in the British Isles was in 1696 on the North Downs, Kent, where the plant still resides today (Druce, 1886;Marren, 2001;Pearman, 2017). The epithet, pratensis, originates from the species' preferred habitat of meadows. Although S. pratensis is not especially noted for its medicinal properties relative to S. officinalis, the common name, Meadow Clary, is indicative of its traditional use as a treatment for eye maladies; the mucilage from wetted seeds was used to rinse eyes and the word clary originates from the term 'clear-eye ' (Fisher, 1987). Medicinal uses as a gargle for sore throats and as a cleanser of the teeth have also been noted (Rich et al., 1999).

| CON S ERVATI ON AND MANAG EMENT
Salvia pratensis can be found in abundance in parts of its range in Central Europe (Hedge, 1972) and it does not appear on the European Red List of Vascular Plants (Bilz et al., 2011). The species is thus not of direct conservation concern internationally; however, it does grow in a declining habitat and is increasingly rare at the edge of its range.
In the British Isles, S. pratensis is nationally scarce, red-listed as near-threatened and a priority Biodiversity Action Plan species (Cheffings et al., 2005) (Rich et al., 1999). This decline followed intensified farming practices, particularly the ploughing of downland and altered management techniques of grassland communities (Scott, 1989). It is thought to have been widely introduced from Eastern European sources between 1880 and 1930s as a seed contaminant of grass and clover seed, but such accidental introductions declined thereafter following the establishment of seed cleaning regulations (Rich & Karran, 2006 and is a target species for conservation in floodplain grasslands, a priority habitat for Dutch conservation (Hegland et al., 2001;van der Meijden & Heukels, 1996;Ouborg & van Treuren, 1995). In Switzerland, it is classed as 'of least concern' on the country's red list (Moser et al., 2002). In Southern Germany, it is described as very common (Schmeil & Fitschen, 1996). In countries at the north and north-western extremities, it is rare. The naturalised populations in Sweden are listed as endangered, as these populations have become fragmented due to changes in roadside management and pasture farming (Gärdenfors, 2005).
Both British and Dutch S. pratensis populations are now generally only found in places protected by conservation efforts (Hegland et al., 2001;Ouborg & van Treuren, 1995;Rich et al., 1999). Since the 1950s, the number of British populations has remained almost constant, restricted to sites managed with more traditional, less intensive farming practices (Rich et al., 1999 (Ouborg & van Treuren, 1994;Ouborg et al., 1991;van Treuren et al., 1991). In the British Isles, Kay and John (1995) (Scott, 1989). The use of herbicides and artificial fertiliser can greatly damage individual plants and populations as a whole (King, 2004).
The removal of sward by hay cutting or grazing is advised, but it is important that this occurs after S. pratensis has flowered and set seed (in Britain, seed generally matures and drops in late July; Scott, 1989). Prolonged earlier cutting or grazing can have a very detrimental effect on S. pratensis populations by limiting reproduction (Hegland et al., 2001). Late mowing of experimental plots in October over 21 years in nutrient-poor limestone grasslands in Switzerland greatly benefitted S. pratensis, to the extent that it dominated plots (relative to plots that were mown in July, burned or without any biomass removal; Kahlert et al., 2005). Scott (1989) made a series of recommendations for site management. He stated that sites should be grazed by livestock at high intensity in spring, to reduce competition from vigorous grasses.
Stock should be removed at the beginning of May to allow the development of flowering shoots, then re-introduced at lower density from late July. For cattle, he suggested that grazing may be possible all year round, provided that the stocking density is low. Summer grazing every 4-5 years may be helpful to prevent the accumulation of unpalatable litter and reduce competition from surrounding vegetation. In addition, temporary stock proof enclosures during the flowering and fruiting period may be valuable, but they should not prevent rabbits from entering. For sites managed as hay meadows, a cut in late July or early August is recommended, which should allow around 50% of seed to ripen. Unripe seeds held on cut stems can after-ripen successfully to some degree (see Section 8.4), so hay should be left to dry in situ to maximise potential recruitment, but then be removed.
To provide a regeneration niche, site management should maintain a level of disturbance. Aftermath grazing, the grazing of hay meadows after hay has been cut, is successful in creating gaps in the sward and patches of loose soil in which seedling recruitment can occur (Smith, 2019). At Stuart Fawkes Nature Reserve in Gloucestershire, the introduction of hay cutting with subsequent cattle grazing has greatly increased the number of seedlings (Parker, 2009).
On road verges and at other sites where grazing is not feasible, Scott (1989) recommended a post-fruiting cut, with the clippings collected up, and wherever possible, manual scarification to provide bare soil for seedlings. Where more regular cutting is necessary for road safety, he suggested a cut in late-April and then again in late-July/early-August. At unmanaged sites, the short sward and scrapes created by rabbits may maintain suitable habitat for S.
pratensis, including niches for seedling establishment. Old anthills and worm casts may similarly aid seedling establishment at unmanaged sites. Craig Brough of the Royal Botanic Gardens, Kew, for facilitating access to library materials. We are also very grateful to the handling editor and four anonymous reviewers for providing valuable constructive comments on the manuscript.

PE E R R E V I E W
The peer review history for this article is available at https://publo ns.com/publo n/10.1111/1365-2745.13805.