seasonal variation in the seed banks of herbaceous species ... · 894 seasonal variation in seed...

Journal of Ecology (1979), 67, 893-921

SEASONAL VARIATION IN THE SEED BANKS OF HERBACEOUS SPECIES IN TEN CONTRASTING HABITATS

K. THOMPSON* AND J. P. GRIME

Unit of Comparative Plant Ecology (NERC), Department of Botany, The University, Sheffield S1O 2TN

SUMMARY (1) Measurements have been made of seasonal variation in the density and

composition of the reservoir of germinable seeds present in surface (0-3 cm) soil samples collected at 6-weekly intervals from ten ecologically-contrasted sites in the Sheffield region.

(2) The procedure was not designed to provide a complete assessment of the seed flora, and the methods were found to be ineffective in recovering germinable seeds of those species (e.g. Endymion non-scriptus, Viola riviniana, several Umbelli- ferae) in which there is only a brief interval between fulfilment of a chilling requirement and the onset of germination.

(3) The techniques adopted were particularly suitable for the detection of persistent seed banks (i.e. those in which some of the component seeds are at least 1 year old), and also allowed recognition of species in which there is a transient accumulation of detached germinable seeds during the summer.

(4) Comparison of the results obtained for populations of the same species in different types of habitat suggests that seasonal variation in seed number is a function of the species rather than of the environment.

(5) It is concluded that the major evolutionary force determining the nature of the seed bank is the selective advantage derived from mechanisms of seed dormancy and germination which allow seedlings to evade the potentially-dominating effects of established plants.

(6) From the data collected in this study, four types of seed bank (Types I-IV) have been recognized, and an attempt has been made to assess their ecological significance.

(7) The transient seed banks (Types I and II) are adapted to exploit the gaps created by seasonally-predictable damage and mortality in the vegetation, whilst the persistent seed bank (Type IV) confers the potential for regeneration in circumstances where disturbance of the established vegetation is temporally and/or spatially unpredictable. A second type of persistent seed bank (Type III) has characteristics intermediate between those of Types I and IV, and contains some seeds which germinate soon after release and others which are more persistent in the soil.

(8) A feature of the results was the lack of a general correspondence between the species-composition of the seed flora and that of the associated vegetation. At certain sites, substantial persistent seed banks were detected for species which were either extremely scarce or did not occur at all in the established vegetation.

(9) Both transient and persistent types of seed banks were represented at each of the ten sites; this is consistent with the hypothesis that complementary mechanisms of regeneration are involved in the maintenance of floristic diversity.

INTRODUCTION

The existence of a reservoir of buried viable seeds in the soils of arable fields and pastures in Britain has been recognized for a long time (Brenchley 1918; Chippindale & Milton

* Present address: Department of Botany, University of Durham, Durham DH1 3LE.

0022-0477/79/1100-0893$02.00 (?1979 Blackwell Scientific Publications

893

This content downloaded from 69.167.65.194 on Mon, 2 Dec 2013 13:21:57 PMAll use subject to JSTOR Terms and Conditions

http://www.jstor.org/page/info/about/policies/terms.jsp

894 Seasonal variation in seed banks

1934; Milton 1943; Champness & Morris 1948), and it is known that the seeds of certain herbaceous species-e.g. Ranunculus repens*, Chenopodium album, Juncus effusus, Agrostis tenuis and Poa annua-accumulate in the soil in very large numbers. Studies of the viable seed content of grasslands of known history (Brenchley 1918; Chippindale & Milton 1934) indicate that buried seeds of some arable weeds can retain viability in undisturbed soils for periods in excess of 50 years. This conclusion rests upon the assumption that arable weed seeds have not been introdtuced since the last cultivation, and to this extent the evidence is not conclusive. However, the seed-burial experiment conducted by Beal in North America and described by Darlington & Steinbauer (1961) provides unequivocal proof that the seeds of at least three species (Oenothera biennis, Rumex crispus and Verbascum blattaria) can survive 80 years of burial. These species are probably exceptional, and the results give little insight into the longevity of natural populations of buried seeds which are subject to processes (e.g. predation) excluded in Beal's experiment. In a study conducted under more natural conditions, Roberts & Dawkins (1967) have shown that in the absence of cultivation and fresh seed input the viable seed content of an agricultural sandy loam showed an exponential decay, with a half-life of about 3 years.

Early investigations concerning buried seed banks were not designed to examine the mechanisms by which losses occur through factors such as predation, soil pathogens and germination, and little attention was given to the processes whereby many seed banks are continually replenished by fresh seed input. In a more thorough analysis of the dynamics of the seed banks of selected species, Sarukhan (1974) has shown that even between taxonomically-related species exploiting the same habitat there may be considerable differences in the size and longevity of the buried seed populations. From his study it would appear that Ranunculus bulbosus, R. acris and R. repens constitute a series, in which R. bulbosus has the least and R. repens the most persistent seed bank. Although most studies of buried seeds have been devoted to annual weeds it is apparent that high densities of buried seeds are by no means confined to arable situations. In particular, Chippindale & Milton (1934) have shown that very large numbers of seeds of perennial grasses, sedges, rushes and shrubs may accumulate in the soils of pasture, heathland and mire, and Spray & Grime (in Grime 1977) have found large populations of buried seeds, mainly of small-seeded forbs, in derelict calcareous grasslands. These data must be borne in mind when an attempt is made to consider the ecological importance of buried seeds. Although the majority of investigations of buried seeds have been concerned with arable fields, it would be unwise to assume that the accumulation of a seed bank is merely a 'weedy' adaptation. Because studies in the British Isles have not so far extended to many types of natural vegetation, it is not yet possible to gain a general perspective with regard to the role of buried seeds. One particular aspect which remains virtually unexplored concerns the mechanisms whereby seeds enter and leave the seed bank. Nor has much effort been made to recognize the circumstances in which some plant populations survive in the absence of accumulations of buried seeds.

This paper describes two experimental studies designed to acquire basic information concerning the composition and dynamics of the seed bankst associated with a range of

* Nomenclature follows that of Clapham, Tutin & Warburg (1962). t Henceforth in this paper the term 'seed bank' refers to all the detached viable seeds of a species

at a specific time, and includes seeds present both above and below the soil surface. The experimental methods used in this investigation provide an estimate of the 'readily-germinable' component of the seed bank. The term 'seed' includes both seeds and fruits.



K. THOMPSON AND J. P. GRIME 895

vegetation types in northern England. The first experiment involves laboratory estimations of seasonal variation in the numbers of readily-germinable seeds present in ten ecologically-contrasted sites. In the second experiment an effort is made to relate these seasonal patterns to the periodicity of germination at one of the sites. The data collected in this study are drawn upon in the Discussion, firstly to produce a functional classification of seed banks, and secondly to examine the role of persistent and transient seed banks in the maintenance of floristic diversity.

EXPERIMENT 1

Materials and methods

General procedure The method consisted essentially of the collection of surface soil from ten sites at

intervals of approximately 6 weeks over the period October 1974 to October 1975. Readily-germinable seeds present in the samples were estimated by placing the soil in an environment known to be favourable for seed germination in a wide range of plant species, and identifying all emergent seedlings. It must be emphasized that this procedure was not designed to provide a complete assessment of the seed flora present at each site. The extremely laborious techniques which are required for accurate estimations of the numbers of seeds in soil samples have been described by Kropac (1966), Major & Pyott (1966) and Roberts (1970), and these are clearly impracticable for large-scale studies, especially those involving small-seeded species. Where, as here, the purpose of the investigation is merely to detect the presence of persistent seed banks or to recognize more transient contributions to the seed flora, it is not necessary to adopt such exhaustive methods, although, as explained in the Discussion, care must be taken in the interpretation of data based upon incomplete assessments. The methods used were basically those of Spray & Grime (in Grime 1977), who estimated seasonal variation in the germinable seed populations present in four derelict calcareous habitats over a 2-year period (1970-72).

In order to provide a basis for comparison of the seed flora with the established vegetation, the composition of the vegetation at each site was described.

Selection and description of field sites Location of the ten study sites is shown in Fig. 1. The composition of the established vegetation at each site was described by recording

the species present in each of one-hundred 10 x 10-cm quadrats located at random. Details are given in the Appendix.

A brief description of each of the ten sites is given below.

(a) Vegetation dominated by woody species Site I-Totley Wood (SK 325813): a fragment of semi-natural woodland situated on a Mill- stone Grit scarp dominated by rather stunted specimens of Quercus petraea and Betula pubescens, with a stream terrace at its base occupied by taller specimens of Quercus petraea with Acer pseudoplatanus and Fraxinus excelsior. The ground flora of the slope was composed principally of Deschampsia flexuosa, with local patches of Pteridium aquilinum. The ground flora of the greater part of the terrace was composed of Endymion non-scriptus, Galeobdolon luteum, Holcus mollis and Milium effusum. A more species-rich flora, with Poa trivialis, Ranunculus ficaria, Silene dioica and Urtica dioica, grew on the lower part of the terrace.




Site 2-Staveley Hedgerow (SK 423763): a hedge at Breck Farm, near Staveley, on the Coal Measures, with Crataegus monogyna, Sambucus nigra, Acer campestre, A. pseudoplatanus and Sorbus torminalis. The grass verge adjacent to the hedge (but not the area sampled) had been sprayed occasionally with herbicides prior to the period of study. The samples were taken from beneath the shade of the hedge in rather heterogeneous vegetation, including tall grasses (e.g. Agropyron repens and Arrhenatherum elatius) and woodland species (e.g. Hedera helix and Mercurialis perennis).

(b) Derelict herbaceous vegetation Site 3-Litton Mills (north-facing slope) (SK 155733): a species-rich, tall-herb vegetation on Carboniferous Limestone. The most frequently recorded species were Festuca rubra, Carex flacca, C. caryophyllea, Origanum vulgare and Helictotrichon pubescens. There was also a marsh element in the flora, comprising Angelica sylvestris, Cirsium palustre, Lotus uliginosus and Parnassia palustris. Grass litter was locally abundant, and elsewhere the ground surface was colonized by a dense cover of bryophytes. Occasional burning has been recorded at the site (Lloyd 1972), but no fires occurred during the period of the investigation.

Site 4-Litton Mills (south-facing slope) (SK 155732): the flora was species-rich, containing local areas of broken turf dominated by tussock grasses, e.g. Festuca ovina, Helictotrichon pratense and Koeleria cristata. Winter annuals occurred on ant hills and rock outcrops, and the latter were colonized by Sedum acre. The marsh element noted in the flora of the north- facing slope was scarcely represented. On the abundant exposures of bare soil and scree, large numbers of seedlings of grasses and winter annuals were observed in the late summer and autumn. Effects of severe moisture stress have been recorded during previous studies at the site (Grime & Curtis 1976), and extensive droughting occurred during the period of this investigation. A small part of the site was burned during the summer.

Site 5-North Anston (SK 526836): a gentle southwest-facing Magnesian Limestone slope, with rough grassland consisting of a mosaic of vegetation types, including areas dominated by Brachypodium pinnatum on the deeper soils and an open turf of Bromus erectus and Festuca ovina on the steeper slopes with shallower soils. Drought is probably the factor preventing Brachypodium pinnatum from invading these shallower soils (Lloyd 1972). The deeper soils were mull rendzinas, surface-leached in places, and colonized by less calcicolous species such as Agrostis tenuis and Betonica officinalis. In those parts of the site dominated by Brachypodium pinnatum, the soil between the tussocks was covered by dense litter and seedlings were rarely observed. Rock outcrops and ant hills occurred locally within the site, and were colonized by species such as Campanula rotundifolia, Helianthemum chamaecistus, Thymus drucei and Veronica chamaedrys. Seedlings of Campanula rotundifolia were observed in these areas throughout much of the year. Much of the surrounding grassland was burned in May 1975, but the actual site was largely untouched.

Site 6-Ladybower (SK 185878): a steeply-sloping (300) acidic roadside verge on the Millstone Grit, extending upslope for 30 m, with a mosaic of different vegetation types. About half the site was composed of rough grassland dominated by Deschampsia flexuosa and Festuca ovina, while the remainder was equally divided between Betula pubescens scrub, patches of Calluna vulgaris and patches of Vaccinium myrtillus. The site was subject to occasional grazing by stray sheep.

(c) Wetland vegetation Site 7-Fox House (SK 262807): this site was situated on the Millstone Grit, and was divided into two ecologically-distinct parts. The upper half comprised an area of waterlogged peat dominated by Juncus effusus and was rather species-poor. Wherever higher plants were absent there was an almost 100% cover of Sphagnum spp. The other half of the site coincided with mineral soils flushed by moving ground water, and supported a much more diverse flora, including Cardamine palustre, Epilobium palustre, Filipendula ulmaria and Holcus lanatus. All parts of the site were grazed lightly by sheep. Site 8-Orgreave Marsh (SK 437863): a small patch of relatively eutrophic mire developed on the Coal Measures and adjacent to a small lake. Willow scrub fringed the landward side. The parts more distant from the open water were colonized by tall stands of Epilobium hirsutum,



K. THoMpSON AND J. P. GRIME 897

while the areas which remained inundated for most of the year were dominated by Berula erecta, Juncus effusus, J. inflexus and Typha latifolia. In the summer the water table fell below soil level in the stands of Epilobium hirsutum, whilst in the parts nearer to the lake the depth of water varied from 2-3 cm in summer to 40 cm in winter. In consequence there were large areas of exposed mud during the summer, and these were rapidly colonized by seedlings, including large numbers of Epilobium hirsutum. Surprisingly, seedlings of Juncus spp. were scarcely observed during visits to the site.

(d) Disturbed vegetation Site 9-Orgreave Cinder Tip (SK 437861): immediately adjacent to Site 8, and consisting of bare cinders interspersed with areas bearing a thin covering of mineral soil. Closed vegetation, where it existed, was dominated by Agrostis stolonifera, A. tenuis, Holcus lanatus and Poa pratensis. Open areas supported a more ruderal vegetation, in which Matricaria matricarioides, Poa annua, Rumex crispus and Senecio squalidus were prominent. The site suffered occasional burning and continual disturbance by small boys. Very large numbers of Chamaenerion angustifolium seedlings were apparent in the autumn.

Site 10-Staveley Arable Field (SK 425763): situated approximately 200 m from Site 2, this site comprised part of an arable field, including a damp hollow, at Breck Farm, near Staveley. The field had a history of arable farming, and carried a wheat crop during most of the period of the investigation, although the damp hollow itself was not sown. The weed flora was dominated by the grass species Agrostis gigantea, A. stolonifera, Poa annua and P. trivialis, together with an understorey of typical arable weeds, e.g. Atriplex patula, Polygonum aviculare, P. persicaria, Stellaria media and Tripleurospermum maritimum. Lolium multiflorum and L. perenne were also represented, and were derived presumably from the field margins and adjacent pastures. A dense carpet of seedlings of Poa spp. and Lolium spp. was observed at the site in late summer and autumn.

Collection of soil samples On each sampling occasion one-hundred soil samples were collected from each site.

The majority of the sites were 100 x 10-m rectangles, and ten randomly-distributed soil samples were collected from each of ten 10 x 10-m subsections. At two sites (sites 2 and

6

M G 89

2 B ~~~~~~~cm 1 B S

4

CL

FIG. 1. Map showing location of experimental sites (1-10) in relation to geological boundaries. S, Sheffield; W, Warsop; B, Buxton; CL, Carboniferous Limestone; MG, Millstone Grit; CM, Coal Measures; ML, Magnesian Limestone; BS, Bunter Sandstone.




10), local topography necessitated slight variations on this pattern. Each soil sample was collected using a trowel, and was as far as possible a cylinder of 7-cm diameter and 3-cm depth, incorporating everything in the top 3 cm of the soil profile, including litter but excluding large twigs and stones.

Soil preparation The soil samples were spread out on plastic sheets and allowed to air-dry in the

laboratory with the aid, in the case of wetland soils, of an electric fan. When thoroughly dry the soil was sieved through a 1-cm-mesh sieve to remove the coarsest plant fragments and stones. The sieved soil samples were mixed thoroughly and then placed in a 2-cm- deep layer overlying 2 cm of coarse sand, in each of four 25 x 25-cm plastic trays.

Seed germination The forty plastic trays were arranged in four randomized blocks beneath banks of

warm-white fluorescent tubes, in a controlled environment providing a 16-h day at 20 ?C and an 8-h night at 15 'C. In previous laboratory tests (Grime & Jarvis 1975; Grime et al. 1980) this regime had been found to be suitable for the germination of seeds of a wide variety of native species. The trays were regularly watered from below with tap water, and the numbers of seedlings were recorded at frequent intervals. In initial tests the majority of seedlings appeared during the first 3 weeks, and germination beyond 5 weeks was found to be negligible. In subsequent experiments all seedlings were removed, identified and counted after a period of 36 days. At no time during the germination period was the soil disturbed, and seedlings were removed only at the end of the experiment. Previous work (e.g. Roberts 1969) has shown that the usual effect of disturbance is to cause more seeds to germinate.

Seedling identification With practice and the aid of Chancellor (1966) it became possible to identify virtually

all seedlings at an early stage of development. Seedlings which were still unidentifiable at the end of 36 days (principally grasses) were transplanted to pots and grown, where necessary, until flowering.

Results Before a site-by-site description of the results is attempted it may be helpful to list

several general features of the data. (1) The species from which numerous germinable seeds were recovered may be

broadly classified into two groups. The first group consisted mainly of grasses, and was made up of species in which the occurrence of germinable seeds was restricted to a well- defined peak, usually occurring during the late spring and summer (Fig. 2). The second group was composed of grasses and forbs in which germinable seeds were detected throughout the year. Within this latter group, a distinction may be drawn between species for which germinable seed numbers remained relatively constant throughout the year (examples are illustrated in Fig. 3), and those for which an essentially similar pattern was interrupted by rather erratic departures on particular sampling dates (Fig. 4). It seems reasonable to conclude that all of the patterns shown in Figs 3 and 4 indicate the presence of a large reservoir of germinable seeds throughout the year, and it is likely that the greater variation evident in Fig. 4 is merely a reflection of the more patchy distribution of seeds of these species within the sample area.



K. THOMPSON AND J. P. GRImE 899

(2) The patterns shown by species occurring at more than one site were generally very consistent between sites. Similarities are apparent between the results obtained for Chamaenerion angustifolium at three sites (Fig. 5), and a close correspondence was also observed between sets of data for other common species such as Agrostis tenuis, Holcus lanatus and Juncus effusus. This suggests that species differences may be of greater importance than environmental differences in determining the seasonal patterns exhibited by seed banks.

(3) The total number of seeds recorded varied enormously between sites. To an extent this variation could be related to seed size, since in general the most abundant seeds were also the smallest. Thus those sites with an abundance of small-seeded species such as Calluna vulgaris, Epilobium hirsutum and Juncus spp. tended to possess large seed banks.

(4) There was no close correspondence between the relative proportions of species in the germinable seed reservoir and the species-composition of the established vegetation at each site. There are several instances of species (e.g. Hydrocotyle vulgaris (site 7) and Berula erecta (site 8)) which were prominent in the vegetation but were not detected as seeds. Conversely, seeds of species such as Juncus effusus (site 1) and Digitalis purpurea (site 2) were recorded frequently at sites where they were not represented in the established vegetation.

The results obtained for each site have been summarized in a series of figures (Figs 6-7) in which for each species a histogram column shows the frequency of occurrence in the established vegetation, and the horizontal lines on the vertical bar indicate the total number of germinable seeds recovered on seven or eight sampling occasions distributed throughout the year. In some species seeds were not recovered at all or their appearance was restricted to certain of the sampling dates. As explained in the Discussion, this pattern may be due to the transient form of the seed bank or may arise from the localized

7 - (a)

6-

5-

4 -

3-

a 2

z I 0\ 0

4 _ __ _ _ _ _ I \

7p (b)

E z

4

3

2

0ON 7D J F M A M J J A SO0

FIG. 2. Two examples of species in which the occurrence of germinable seeds is restricted to a well-defined peak in late spring and summer. (a) Festuca ovina (site 4), (b) Lolium

perenne (site 10).



900 Seasonal variation in seed banks distribution of seeds within the habitat. Data are included for all species which occurred in 1000 or more of the quadrats (2000 in the case of the three calcareous grasslands, sites 3, 4 and 5) and for any other species which, although uncommon, was a significant contributor to the seed numbers. In these Figures seedling numbers are expressed as totals per four 25 x 25-cm trays of soil, and are displayed on a log scale.

This form of data-presentation allows a rapid assessment to be made of the contributions of each species to the vegetation and to the reservoir of germinable seeds, and also provides some measure of the seasonal variability in the size of the seed bank. The results obtained for the ten sites will now be considered in turn.

Site 1-Totley Wood (Fig. 6(a)) The lack of correspondence between the vegetation and the germinable seeds recovered

from this site is quite remarkable. Seeds of the four most frequent species recorded in the vegetation were particularly scarce. Field observations suggested that three of these plants (Deschampsiaflexuosa, Holcus mollis and Ranunculusficaria) produced little or no seed at this site during the period of study. The remaining species, Endymion non-scriptus,

7 (a)

6 -

5

4 _

3 0

0 ~ ~ ~ -

7 (b)

6 -

5 -

4 04

70 - ( c

0 _

0 _ 70 -(c)

60-

50

40 -

30

20 - 0

I0 0 I

0 N D J F M A M J J A S 0

FIG. 3. Three examples of species in which germinable seed numbers remain relatively constant throughout the year. (a) Thymus drucei (site 4), (b) Campanula rotundifolia

(site 5), (c) Holcus lanatus (site 9).




released large quantities of viable seeds in July, and a high density of seedlings appeared in the field in the early spring. Over the intervening period, however, germinable seeds were not detected in the soil samples.

The species with the most numerous germinable seed populations were Digitalis purpurea, Juncus effusus, Milium effusum, Poa annua and P. trivialis, and all were represented throughout the year. Juncus effusus, the species with the second largest number of germinable seeds, was not recorded in the vegetation at all, and Digitalis purpurea, Milium effusum and Poa annua were all infrequent. Only Poa trivialis showed significant representation in both the vegetation and the soil samples.

Site 2-Staveley Hedgerow (Fig. 6(b)) As in site 1, the most frequent species in the vegetation, in this instance Agropyron

repens and Mercurialis perennis, were virtually unrecorded as seeds. Although seeds of the next most frequently-occurring species in the vegetation, Arrhenatherum elatius, were detected, they occurred in much smaller numbers than those of Digitalis purpurea, Juncus effusus and Scrophularia nodosa, all of which were absent from the established vegetation. The highest numbers of germinable seeds were obtained for Chamaenerion angustifolium, a species which was moderately frequent in the vegetation. It is interesting to note that all of the species with comparatively large numbers of germinable seeds at this site were small-seeded. The total number of seeds recovered remained relatively constant throughout the sampling period, and the only species of frequent occurrence to exhibit a pronounced seasonal pattern in germinable seed number was C. angustifolium.

Site 3-Litton Mills, north-facing slope (Fig. 6(c)) By far the commonest contributor to the seed reservoir at this site was Origanum

vulgare; the number of seeds detected for this species exceeded the total for all other

14 - (a)

12 -

10 _

8 _

6 -

a 4

12

01~~~~~~ E 1

* 12-

O N D J F M A M J J A S O

FIG. 4. TWO examples of species in which relatively constant germinable seed numbers are interrupted by a few unusually high or low values on particular sampling dates.

(a) cirsium palustre (site 7), (b) Juncus inflexus (site 8).



902 Seasonal variation in seed banks 14 - 0 (a)

12-

00

8 -

6 -

4 -

2-

O I 1 1. 1 1 I1

8 -(b) cn

c' 6

0~~~~~~~~~~ a)

0 2 _ E z *

0 I I O~I I I 1 1- 1 1

01~~~~~~~~ lOO -(c)

-75-

50-

25

0~~~ O N D J F M A M J J A S O

FIG. 5. A comparison of seasonal variation in numbers of germinable seeds of Chamaenerion angustifolium recovered at three sites. (a) Staveley Hedgerow (site 2), (b) Orgreave Marsh

(site 8), (c) Orgreave Cinder Tip (site 9).

species. Holcus lanatus seeds were numerous, as to a rather lesser extent were seeds of Agrostis tenuis, Anthoxanthum odoratum and Plantago lanceolata. All these species were frequent in the vegetation at the site and none displayed a marked seasonal variation in seed numbers. No examples were found in which large numbers of germinable seeds coincided with the absence of the species from the vegetation. Among the species which were prominent in the vegetation, a considerable number (including Achillea millefolium, Angelica sylvestris, Arrhenatherum elatius, Carex caryophyllea, C. flacca, Centaurea nigra, Festuca rubra, Helictotrichon pubescens, Lotus corniculatus, Pimpinella major, P. saxifraga, Succisa pratensis and Valeriana officinalis) were scarcely represented as seeds. It is worth noting that this latter group consisted almost exclusively of relatively large-seeded species.

Site 4-Litton Mills, south-facing slope (Fig. 6(d)) Origanum vulgare again displayed high numbers of germinable seeds, but the most

numerous seeds were those of the small winter annuals Arabidopsis thaliana and Arenaria




serpyllifolia. Seeds of Saxifraga tridactylites, another winter annual, were also present in large numbers. Although locally frequent within the site these three species did not appear among the commonest vegetation components. Perennial species for which seeds were relatively abundant were Agrostis tenuis, Festuca ovina, Plantago lanceolata, Silene nutans and Thymus drucei. Seeds of Helictotrichon pratense were encountered in large numbers only in August and were absent on all other occasions except one. With the exception of Agrostis tenuis, all of the perennial species mentioned above were frequent in the vegetation. Arrhenatherum elatius, Briza media, Carex caryophyllea, Festuca rubra, Helianthe- mum chamaecistus, Koeleria cristata, Lotus corniculatus and Pimpinella saxifraga were all of common occurrence in the vegetation but were recorded as infrequent or absent as seeds. The numbers of germinable seeds of Agrostis tenuis, Holcus lanatus, Origanum vulgare, Plaitago lanceolata, Silene nutans and Thymus drucei showed little variation with season, but those of the winter annuals and the perennial grasses Festuca ovina and Helictotrichon pratense exhibited strong peaks in late summer. Because of the strong periodicity il these species, total seed numbers recovered from the site also displayed a pronounced seasonal pattern, reaching a maximum in late summer and falling to relatively low levels at other times.

Site 5-North Anston (Fig. 6(e)) This site had the smallest number of seeds recorded at any of the ten sites. The perennial

herbs Agrostis tenuis, Campanula rotundifolia and Plantago lanceolata made a large contribution to the seed total, and all were major constituents of the vegetation. Antho- xanthum odoratum and Holcus lanatus seeds were recorded in small but relatively constant numbers, and both species were moderately frequent in the vegetation. Seeds of Chamaenerion angustifolium were present in only small numbers, but nevertheless showed the marked seasonal fluctuation observed at other sites (sites 2, 8 and 9). Although the vegetation was dominated by Brachypodium pinnatum, seeds of this species were virtually absent. Similarly, there were very few germinable seeds of Betonica officinalis, Carex caryophyllea, Centaurea nigra, Dactylis glomerata, Festuca ovina and Lotus corniculatus, all species which were frequent in the vegetation. Bromus erectus seeds were absent except on one sampling occasion in late summer, and in this respect this species strongly resembled Helictotrichon pratense at site 4. Sieglingia decumbens showed an unusual pattern in that an apparently stable seed reservoir declined to zero during the spring of 1975, and was not renewed throughout the rest of the sampling period. (In passing, it is interesting to note that Urtica dioica (site 1), Stachys sylvatica (site 2) and Polygonum aviculare (site 10) displayed similar patterns.)

The numbers of seeds of the most numerous species, Agrostis tenuis and Campanula rotundifolia, showed little seasonal variation, while those of Plantago lanceolata declined in summer and increased in autumn. The total germinable seed numbers for the site showed a maximum in autumn and a minimum in spring.

Site 6-Ladybower (Fig. 7(a)) Calluna vulgaris, which was fifth in order of abundance in the vegetation at this site,

accounted for approximately three-quarters of all the seeds recorded during the sampling. Conversely, seeds of the two commonest species in the vegetation, Deschampsiaflexuosa and Vaccinium myrtillus, were recovered in extremely small numbers. Seeds of Festuca ovina were also very infrequent despite the abundance of this species in the vegetation. Agrostis canina, A. tenuis, Juncus effusus and Potentilla erecta were all relatively abundant as seeds, although of these species only P. erecta was of common occurrence in the



904 Seasonal variation in seed banks 90 - (a) Site I 90 - (b) Site 2

80 _1000? 80 _1000?

70 | '70;

90~~~~~~~~~

60 0-- a 0 0

100 a 00

40 4ItLLLllL2llu H X h00 L 0

30 t-i 0 70

20 'III 20~~~~~~~~0

FIG0 60 { fe

det--e 0emnal 0ed vria as.A ahofsvnsmln ae oee

90thin he site seed Sitbers 3 re rcd as lio st

50 - 2 80 100

70~~~~~~~~~

E~~~~~~~~~~~~~~~

20 - 20 - 0~~~~~~~~~~~~10 50~~~~~~~~~

40 1~~~~~~~~ 10 30 10

00

20~~~~~~~~~~~~~~~~~~~~~~~

60 - ~ ~ ~ ~ ~ ~

FIG. . Thefrequncy o occurenceof spciesas esablised plnts (istogams) nd a deace gemnbeses(etclbr)-tec fsvnsmln ae o ee in crtai speies,incldingthoe wih trnsiet sed baks r vey loalizd ditribtio

wihn 0h sie- ednmesaercre slne ueipsdo h etclbr




90 -(d) Site 4

80 - C :?< tQi<%?fi;|R ,gl 2tea>

80 _ 1000 ,

70 _ E

760 - 1

:50 _D

30

20 1 I0~~~~~~~~~~~~~~~~~~~~ 0

50 -~~~~~~~~~~~~~~~~

90 (- it

70

0 T T x

FIG. 6 (cont.)

2 50~~~~~~~~~~~~~~~~~~R 40~~~~~~~~i -Z

I0~~~~~~~~~~~~~~~~~~ 70 - C3IiC 2tt~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~)

C: E~~~80-Se?~

C, C,~~~FG.6(cn.




90 (a) Site 6

80 _ 002

70~~~~~~~~~~~~~~

60 -

ioo

o 50

40

20~~~~~~~~~~~~~~~

@~~~~~~~ t t 0 'PO X ,,XEb 30, !8,XBCX l

20 1100

90 - b it

80 -100

70-

10 -~~~~~~~~~~~~~~~~~~~~~~0

401 I I II I IT I 10

FIG. 7. The frequency of occurrence of species as established plants (histograms) and as detached germinable seeds (vertical bars). At each of eight sampling dates (or fewer in certain species, including those with transient seed banks or very localized distribution within the site) seed numbers are recorded as lines superimposed on the vertical bar.



K. THompSON AND J. P. GRimm 907 90 (c) Site 8

_ 80 t 1000

o ?

0 0~~~~~~~~~~~~~~~~~0

30 - -T

0-----

E E~

50 - ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ -

70~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 40 - ~ T

~~~~~~~~~~ 50 ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~~~~~~~~~~~~~~1

40 ~~~~400 30

0 H

'C- - - 0F

I t~~~~i.7 cn.




vegetation and Juncus effusus was not recorded at all. Nardus stricta produced a small peak in seed numbers during the late summer. Not surprisingly, the seasonal pattern in total seed numbers closely followed that of Calluna vulgaris, i.e. there was a large, relatively constant reservoir, with a single peak in mid-summer.

Site 7-Fox House (Fig. 7(b)) An extremely large number of seeds was recovered from this site and over 900 of

them belonged to Juncus effusus. The other main contributors were Agrostis tenuis, Anthoxanthum odoratum, Cirsium palustre, Deschampsia cespitosa, Galium palustre and Potentilla erecta, all of which were relatively frequent in the vegetation. No seeds were recovered of Hydrocotyle vulgaris, the commonest species in the vegetation. Carex nigra, Deschampsia flexuosa, Festuca rubra, Galium saxatile and Rumex acetosa all had low numbers of germinable seeds, despite being major components of the vegetation. The seasonal pattern in total seed numbers reflected that of the dominant constituent, Juncus effusus, i.e. a large stable seed population with a single depression in April.

Site 8-Orgreave Marsh (Fig. 7(c)) At this site the bank of germinable seeds was dominated by three small-seeded species,

Epilobium hirsutum, Juncus articulatus and J. effusus, all of which were of common occurrence in the vegetation. Also present in relatively large numbers as germinable seed were Alisma plantago-aquatica, Deschampsia cespitosa, Galium palustre, Juncus inflexus, Poa trivialis, Rumex sanguineus, Sagina procumbens, Chamaenerion angustifolium and Poa annua. The last three species were not recorded in the vegetation at the site, and the seeds were presumably derived from the surrounding wasteland (see the vegetation description for the adjoining site 9). Seeds of Cardaminepratensis, the most abundant species in the vegetation, were present in only small numbers, while those of Berula erecta, the next most abundant species, were not detected at all. The numbers of Deschampsia cespitosa seeds declined after the first sampling occasion, while those of Rumex sanguineus and Sagina procumbens declined slightly later in the year. Alisma plantago-aquatica numbers varied erratically throughout the year. All three Juncus spp. displayed minor seasonal variations, but were present in quite large numbers at all times. Epilobium hirsutum declined slightly from a peak in October 1974, but germinable seed was present throughout in large numbers. Chamaenerion angustifolium exhibited a seasonal pattern closely similar to that obtained at sites 2 and 9. The total number of seeds recorded at site 8 was very large, second only to site 7, and showed a clear seasonal pattern, with declining values in spring and early summer and a maximum in late autumn.

Site 9-Orgreave Cinder Tip (Fig. 7(d))

The total number of seeds recovered from this site was large, but fell far short of the enormous numbers at sites 7 and 8. Two small-seeded species, Chamaenerion angustifolium and Saginaprocumbens, were abundant as germinable seed at this site. There were in addition large numbers of seeds of Holcus lanatus and Juncus effusus. All of these species were present in the vegetation, although Sagina procumbens and Juncus effusus were restricted to local patches. At this site relatively small but constant numbers of seeds of Matricaria matricarioides and Trifolium repens were recovered. Two species, Agrostis tenuis and Poa annua, which at other sites had seeds present throughout the year, were recorded at site 9 only on certain occasions. This was almost certainly due to the patchy distribution of these species within the sampling area. Similarly, the erratic variation in the numbers of seeds of Juncus articulatus, J. effusus and J. inflexus was




probably a result of these species being restricted to a small waterlogged hollow within the sampling site. The pattern in total seed numbers showed a peak in autumn and declining values in early summer, and was strongly affected by the contribution of Chamaenerion angustifolium.

Site 10-Staveley Arable Field (Fig. 7(e)) Poa annua and P. trivialis dominated the seed flora at this site, and both species

exhibited a large peak in seed number in August. With the single exception of Matricaria matricarioides, all of the major contributors displayed some seasonal periodicity in seed numbers. Agrostis spp. (A. stolonifera and A. gigantea) showed a peak in October which was considerably later than the peaks exhibited by Poa annua and P. trivialis. The seeds of Lolium perenne were virtually restricted to one sampling occasion, in August, a pattern closely resembling that of Helictotrichon pratense (site 4) and Bromus erectus (site 5). Seed numbers in Juncus bufonius declined from a plateau in early summer but increased again in late summer. The total number of germinable seeds was more variable seasonally than at any of the other sites. Values fell to very low levels in spring and summer, but then rose to an abrupt peak during the autumn.

EXPERIMENT 2

Materials and methods

Choice of site As explained in the Introduction, the objective in this experiment was to examine the

relationships between seasonal variation in numbers of detached germinable seeds and the timing of germination under natural field conditions. In selecting a site for this investigation, a major concern was to include species representative of both the transient and persistent types of seed banks encountered in Experiment 1. An area conforming to this requirement was found to occur within the area of derelict herbaceous vegetation already sampled on the south-facing slope at Litton Mills (site 4). The range of commonly- occurring species included some which in Experiment 1 were found to have persistent seed banks (e.g. Silene nutans, Thymus drucei) and others (e.g. Festuca ovina, Helicto- trichon pratense) in which germinable seeds were recovered only for a relatively short time during the summer. An additional factor governing the choice of this site was the open character of the turf, which allowed relatively easy detection of small seedlings.

Procedure Seasonal variation in the germinable seed content of the habitat was estimated follow-

ing basically the same procedure as that described in Experiment 1. At approximately 6-weekly intervals from April 1976 to March 1977, twenty-five randomly-distributed soil samples were collected from a rectangular area of 25 x 10 m.

In order to monitor the appearance of seedlings in the field over the same period, five 20 x 20-cm permanent quadrats, randomly distributed within the 25 x 10-m area, were set up. On the occasion of each soil collection the five quadrats were examined, and, taking care not to disturb the soil surface, all seedlings were removed and identified.

Results The results are presented in graphical form in Figs 8 and 9. It is difficult to compare

directly the results from this experiment with those described for the same site in Experi- ment 1. This is because the sample area used in this second experiment was only a



910 Seasonal variation in seed banks quarter of the extent of the site used in Experiment 1. Some of the species encountered in the first experiment (e.g. Origanum vulgare) were not recorded, and this could be corre- lated with the fact that these plants were largely restricted to areas of the site excluded from the second experiment. Conversely, data for certain species, such as Briza media and Koeleria cristata, which were much more common in the area used for the second experiment than in the site as a whole, became more prominent in the results. The climate during the two experiments was very similar; in both, the summer period was unusually hot and dry.

When the numbers of germinable seeds are considered it is clear that, as in the previous experiment, the species may be classified broadly into two groups. Moreover, the patterns shown by individual species were closely similar to those exhibited by the same species in Experiment 1.

In Briza media, Festuca ovina, Helictotrichon pratense and Koeleria cristata (Fig. 8) a complete absence of germinable seeds during the spring and early summer was followed by the development of large peaks in the late summer and autumn. Seedlings of all four species were absent in the field during the summer, but large populations developed in

30 -(a)

20-

0

n 20 (b) U) F o

V, I 10

' 20 (c) 0

E =3 10 z

0I~

20i (d)

100

01 ~ ~ 0 A M J J A S 0 N D J F M

FIG. 8. Seasonal variation in numbers of detached germinable seeds (O) and in recently- germinated seedlings (0) in four species exhibiting transient peaks of germinable seeds during spring and early summer at site 4. (a) Festuca ovina, (b) Koeleria cristata, (c) Briza

media, (d) Helictotrichon pratense.



K. THOMPSON AND J. P. GRIME 911 40 (a)

30-

cn20 *-~ ~ ~ -

I 0 -0~ 0

0

E 40 -(b)

o 30 30

E

20 -

10

01~~~~~~~ A M J J A S O N 0 J F M

FIG. 9. Seasonal variation in numbers of detached germinable seeds (@) and recently- germinated seedlings (0) in two species with a persistent reservoir of germinable seeds.

(a) Thymus drucei, (b) Silene nutans.

autumn, the maxima in seedling numbers following shortly after the peaks in germinable seeds.

The species for which germinable seeds were recovered in relatively large numbers throughout the year were Silene nutans and Thymus drucei. The fluctuations in seed numbers apparent in the data for Silene nutans and Thymus drucei are probably a result of the patchy distribution of these species within the site. Despite this variability, however, the patterns are totally unlike those of the Festuca ovina group in that germinable seeds are present throughout the year and numbers never fall to zero. Seedlings of Silene nutans appeared at the site in small numbers both in early summer and autumn, while those of Thymus drucei were recorded only in the autumn (Fig. 9).

One species, Pimpinella saxifraga, showed a pattern totally unlike either of the two groups discussed above. Seeds of this species were not detected on any occasion, and seedlings were present on only one date, in March 1977. It is known (Grime et al. 1980) that before germination can take place in P. saxifraga a period of chilling is required. Once this requirement is satisfied this species can germinate even at low temperatures and in darkness. Hence it is not surprising that seeds escaped detection by the soil sampling, and germinated in the field plots at Litton in February or March.

DISCUSSION

Effectiveness of the methods It has been acknowledged already (p. 895) that the methods used in this study provide an

incomplete assessment of the size and composition of the seed banks under investigation. It is mainly for this reason that the estimates of seed numbers have not been expressed in




relation to a standard area of ground or volume of soil; such a presentation would inevitably encourage misleading comparisons with previous studies in which complete determinations have been attempted. The aims of this investigation were to classify seed banks of herbaceous species, and to analyse their functional significance in contrasted vegetation types, and it is our contention that progress towards these objectives can be made using methods which do not involve complete recovery of the seed flora.

The methods adopted appear to be particularly suitable for the detection of persistent seed banks. The large and rather constant numbers of seeds recovered from common species such as Calluna vulgaris, Juncus spp. and Origanum vulgare indicate that the methods of sampling and seed germination were suitable for the recognition of a persistent reservoir of readily-germinable seeds. Even for species which were not very common vegetation constituents (e.g. Milium effusum, site 1), the technique appears to have functioned efficiently. More variability was encountered where the population concerned was confined to a small area within the site (e.g. Juncus inflexus, site 9) or contained only a few flowering individuals (e.g. Alisma plantago-aquatica, site 8). Varia- bility of this type is not usually sufficient to obscure the presence of a persistent seed bank and could have been avoided only by removing an impracticably large number of samples.

The techniques used were also successful in detecting the very common pattern in which seeds were present in the habitat only for a short period of time during the summer or autumn (e.g. Helictotrichon pratense, site 3; Lolium perenne, site 10). In marked contrast, however, the methods proved to be quite unsuitable for the study of species such as Endymion non-scriptus (site 1), Pimpinella major (site 3), Pimpinella saxifraga (sites 3 and 4), Viola palustris (site 7) and Viola riviniana (site 3), in each of which most, if not all, of the freshly-released seeds exhibit innate dormancy and must experience a period of chilling before germination is possible (Thompson & Cox 1978; Grime et al. 1980). None of these species was recorded in significant numbers in the soil samples. This is perhaps surprising in view of the fact that at the field sites concerned a large output of seeds was observed in species such as Endymion non-scriptus and Pimpinella saxifraga. In theory, the failure to detect germinable seeds in these plants could have been due to the use of an unsuitable laboratory environment. However, from the results of a germination screening programme (Grime et al. 1980) it is apparent that the majority of these species, together with many other native plants of similar seed physiology, are unlikely to persist in an ungerminated state once the chilling requirement has been fulfilled. This conclusion is based upon the observation that, after chilling, seeds of most of these species germinate, in light and in darkness, over a wide range of temperatures and are capable to an unusual extent of rapid germination at low (< 10 ?C) temperatures. It would appear that in these plants there is a very short interval between the attainment of germinability and the onset of germination. The low recovery of seeds may be related therefore to the fact that practical limitations did not allow a frequency of sampling sufficient to impinge upon this transient 'germinable phase'.

A final consideration in evaluating the data is the extent to which results for a single year may be representative of long-term patterns. This is a problem particularly relevant to this study since the weather during much of 1975 was unusually hot and dry. In this connection it is helpful to refer to the investigation of Spray & Grime (in Grime 1977), in which the seed banks associated with four derelict limestone habitats were monitored. The methods in this study were very similar to those employed here, but the results refer to a 2-year period from 1970 to 1972. It is encouraging that very similar seasonal patterns




in seed numbers were found in the species common to the two studies (e.g. Arenaria serpyllifolia, Chamaenerion angustifolium, Festuca ovina, F. rubra, Helictotrichon pratense, Holcus lanatus, Koeleria cristata, Origanum vulgare, Potentilla erecta, Plantago lanceolata and Thymus drucei).

During the spring of 1975 the number of germinable seeds recovered for four species (Polygonum aviculare (site 10), Sieglingia decumbens (site 5), Stachys sylvatica (site 2) and Urtica dioica (site 1)) declined from a constant level to virtually zero and failed to recover in the following autumn. The disappearance of germinable seeds during the spring could have been due to the induction of dormancy by high temperature (Courtney 1968). However, since the seeds of three of the four species concerned are known to be stimu- lated to germinate by high temperatures and/or large diurnal fluctuations in temperature (Mason 1976; Thompson 1977), it is also possible that in these species losses from the seed bank during early 1975 were due to unusually high rates of germination induced by the exceptionally hot weather experienced in that year.

Interpretation of the results Species with low densities of germinable seeds

In Figs 6-7, twenty-two species may be identified in which a relatively high frequency in the established vegetation is associated with the absence of germinable seeds in the soil samples or with their appearance in very small numbers. As explained under the last heading, the seeds of certain of these plants have a chilling requirement and exist in a germinable state for only a brief period of time. In addition to these species, however, there are several in which the scarcity of seeds may be related to the low reproductive vigour of the populations under investigation. At the Totley Wood site low levels of flowering by Deschampsia flexuosa and Holcus mollis in shaded situations have been described by Al-Mufti et al. (1977). It may be significant also that several of the species for which seeds were extremely scarce (Brachypodium pinnatum (site 5), Carex caryophyllea (sites 3, 4 and 5), Hydrocotyle vulgaris (site 7), Ranunculus ficaria (site 1) and Succisa pratensis (site 3) are among the small number of common plants for which it has not been possible, despite repeated efforts over the period 1972-77, to make large-scale seed collections from populations established in the Sheffield region. However, this explanation cannot be applied in the case of Berula erecta (site 8) and Carex flacca (site 3), in both of which large quantities of viable seeds were produced at the sites during the period of study. In searching for an explanation for this phenomenon the possibility may be considered that these two relatively large-seeded species were subjected to exceptionally high rates of predation by small rodents.

The distinction between transient and persistent seed banks When the results of Experiments 1 and 2 are considered, it is apparent that the species

concerned occupy a wide spectrum with respect to their seed banks. The species at one extreme are characterized by the production of a transient seed bank, whilst those at the other are associated with persistent seed banks. For the purpose of this discussion a transient seed bank is defined as one in which none of the seed output persists in the habitat in a viable condition for more than 1 year. In a persistent seed bank some of the component seeds are more than 1 year old. Before further consideration of the results, an additional point of difference between transient and persistent seed banks may be established. Those workers who have split up the soil profile into horizons of different depth (Chippindale & Milton 1934), or who have excluded surface soil (Wesson &




Type I

I II I I I I I I I I I I I i I I I I * I I I I I I I Type 1-1

Type v

l l l l l l l l l l l l l l l l l l l l l l l l l

A M J J A S O N D J F M A M J J A S O N D J F M A

FIG. 10. Diagrammatic representation of the four types of seed bank encountered in this study. Shaded areas: seeds capable of germinating immediately after removal to suitable laboratory conditions. Unshaded areas: seeds viable but not capable of immediate germination. Type I: annual and perennial grasses of dry or disturbed habitats. Type II: annual and perennial herbs colonizing vegetation gaps in early spring. Type III: species mainly germinating in the autumn but maintaining a small persistent seed bank.

Type IV: annual and perennial herbs and shrubs with large persistent seed banks.

Wareing 1967), before estimating the germinable seed content have demonstrated con- clusively that buried seeds make up a very high proportion of persistent seed banks. Conversely, the seeds of those species which lack persistent seed banks tend not to become buried. This distinction immediately suggests an opportunity for analysis, in that it may be possible to recognize the characteristics of the seed which determine whether it shall germinate or enter a buried seed bank. As we suggest under the next heading, one ap- proach here is to compare the morphology and germination-physiology of 'transient' and 'persistent' seeds.

The functional significance of the various types of seed bank In attempting to assess the role of seed banks in vegetation dynamics or in the ecology

of individual species, it will be helpful to distinguish four main types encountered in this study. The essential features of the four types of seed bank are summarized in Fig. 10. From this figure it is apparent that Types I and II are transient, whilst Types III and IV are persistent. From the results of comparative laboratory experiments (Grime & Jarvis 1976; Mason 1976; Thompson 1977; Grime et al. 1980), it is possible to recognize some physiological and morphological characteristics of the seeds associated with the four types of seed banks.

Seed bank Type I-species with transient seed banks present during the summer. Included in this group are a number of common grasses in which the seeds are released during the late spring and summer, and germinate nearly synchronously in the cooler moist conditions of the autumn. Typical species encountered in this investigation include relatively large-seeded (i.e. >0.5 mg) perennials such as Arrhenatherum elatius, Bromus erectus, Dactylis glomerata, Festuca rubra, Helictotrichon pratense and Lolium perenne. Three smaller tussock grasses, Briza media, Festuca ovina and Koeleria cristata,




also fall into this group, and from our own field observations and the demographic studies of Davison (1971, 1977) and Watkinson (1978) it is clear that a number of annual grasses such as Bromus mollis, B. sterilis, Hordeum murinum, Lolium multiflorum and Vulpiafasciculata may be safely included in this category.

Features which distinguish the seeds associated with this type of seed bank are: (1) large size and/or elongated structure, including projections such as long awns; (2) lack of pronounced after-ripening or dormancy mechanisms; (3) ability to germinate over a wide range of temperatures, including, in many species, relatively low temperatures (c. 5 ?C); and (4) ability to germinate in the light and in continuous darkness. All of these features would appear to facilitate rapid germination soon after seed fall, and none is conducive to the persistence of ungerminated seeds in the habitat. Large seeds are ob- viously less likely than small ones to be washed into small cracks and fissures in the soil surface and to be buried by the activities of the soil fauna. Moreover, the wide range of temperatures suitable for germination in these species ensures that seedling establishment can commence even if the seed falls rather later in the year and experiences colder conditions. For seeds falling earlier than usual, lack of moisture rather than high temperature is likely to be the factor delaying germination. Finally, since germination in these species can take place in darkness, germination is not prevented even if the seed becomes buried, providing that moisture supply and temperatures are suitable.

The functional significance of the Type I seed bank appears to be related to exploitation of grasslands subjected to seasonally-predictable damage by drought. In southern Europe, disturbance of vegetation by drought occurs each year and is most severe in its effect upon grasses, many of which are shallow-rooted. In many grasslands there is additional seasonal damage resulting from trampling or defoliation by wild or domestic grazing animals. Under such conditions detached viable seeds of the majority of grasses are present in the habitat only during the dry season, and reach a minimum in the wet season, when germination results in the appearance of large numbers of seedlings in the areas of bare ground developed during the preceding summer. Type I seed banks are thus part of the regenerative mechanism whereby predictable mortalities in the population of established plants are replaced annually during a particularly favourable season.

The failure of the grasses in this category to develop persistent reserves of buried seeds in the soil is well-documented (Brenchley & Warrington 1930; Chippindale & Milton 1934; Milton 1939; Champness & Morris 1948), and is of profound importance to pasture management in Europe since it appears to play a crucial role in the process whereby at more northerly latitudes sown species such as Lolium perenne and L. multiflorum tend to be replaced by native grasses with persistent seed banks.

Seed bank Type II-species with transient seed banks present during the winter. The second major group of species which possess transient seed banks is that for which in the present study Pimpinella saxifraga is the best-documented example. From the results of laboratory germination studies (Grime et al. 1980), a pattern of winter dormancy followed by germination in the early spring may be predicted for a large number of species, some of which (Endymion non-scriptus, Heracleum sphondylium, Mercurialis perennis, Pimpinella major and P. saxifraga) were present in the vegetation examined in this investigation. However, germinable seeds of these species escaped detection in the present study and, as explained on p. 912, it seems likely that this was mainly due to the extremely short delay which occurs in these species between satisfaction of the chilling requirement and the onset of germination.



916 Seasonal variation in seed banks The data of Grime et al. (1980) indicate that species with transient seed banks present

in the winter resemble those forming Type I seed banks, in that the seeds are relatively large and are likely to escape burial in the soil. Seeds in this category germinate in darkness, and they are also capable to an exceptional extent of germination at low temperatures.

Seed bank Type II is strongly associated with species of northern or continental regions within the temperate zone. In these geographical areas, plant growth in the late autumn and winter is severely restricted by low temperatures, and for this reason gaps in herbaceous vegetation originating during the summer are likely to remain open until the following spring-and may be enlarged during the winter and early spring as a result of frost heaving, solifluction and water erosion. Under these conditions, it would appear that the most favourable period for seedling establishment will occur in the spring, and we conclude, therefore, that the Type II seed bank represents a specific adaptation delaying germination until the beginning of the growing season.

Seed banks Types III and IV-species with persistent seed banks. In this investigation, a large number of species from a wide range of habitats has been found to be capable of forming persistent seed banks. It would not be surprising if on closer inspection this group of plants was found to be rather heterogeneous with respect to the form and function of the seed bank. At this early stage of analysis it seems useful to recognize two main types of persistent seed banks.

In the first type (Type III, Fig. 10) many of the seeds germinate soon after they are released, and effectively this component of the seed output functions as a Type I seed bank. However, a proportion of the seeds fails to germinate directly and some of these become incorporated into a persistent seed bank. In the present study plants conforming to this pattern included the two small winter annuals Arabidopsis thaliana (site 4) and Arenaria serpyllifolia (site 4), and several grasses such as Poa annua (site 10) and Poa trivialis (site 10). A remarkable example of a Type III seed bank was evident at Orgreave Marsh (site 8), where marked differences could be inferred with respect to the fate of seeds released from Epilobium hirsutum. During the autumn, large numbers of freshly- dispersed seeds of this species were observed to germinate over a wide area of the marsh. From the seasonal sampling, however, it was apparent that a persistent reservoir of seeds was maintained in close proximity to the parent plants. This pattern differs substantially from that consistently observed in Chamaenerion angustifolium (sites 2, 8 and 9; Fig. 5), another commonly-occurring, wind-dispersed species with a large seed output. In C. angustifolium the release of some of the seeds is often somewhat delayed, and this may explain the fact that germinable seeds were recovered from samples collected in the following spring; there is, however, no evidence which would suggest that this species is capable of developing a persistent seed bank.

The second type of persistent seed bank which may be recognized (Type IV, Fig. 10) is that in which few of the seeds germinate in the period immediately following dispersal, and the species maintains a seed bank the size of which changes little with season and is large in relation to the annual production of seeds. Species which in this study exhibited well-defined Type IV seed banks included Calluna vulgaris (site 6), Juncus effusus (sites 1 and 7), Milium effusum (site 1) and Origanum vulgare (site 3).

Caution must be applied in observing a rigid distinction between seed bank Types III and IV, although from comparative laboratory experiments (Grime et al. 1980) there is evidence that between species representative of the two types there are some differences in




seed physiology. Clearly there is a strong possibility that in some species there will be variation from place to place and from year to year in the proportion of the seed output which enters the buried seed bank.

When the large number of species forming persistent seed banks in the sites examined in this study is considered as a whole, it is interesting to note that despite their diverse origin and ecology the species form a relatively coherent group in terms of their seed characteristics.

With very few exceptions the species display characteristics which are in marked contrast to those described for seed banks Types I and II. The majority have small seeds, and many of the more abundant species possess seeds which are exceedingly small. In laboratory experiments (Mason 1976; Thompson 1977; Grime et al. 1980), germination usually occurs over a rather restricted range of temperatures, and in particular tends to be inhibited at the cold end of the range. In many species germination in at least a proportion of the seeds is inhibited by continuous darkness. This suite of seed characteristics appears to be particularly relevant to the problem of understanding how persistent seeds become buried and are subsequently prevented from germinating during periods of extended dormancy in the soil.

Among the species forming persistent seed banks there are several in which the seeds are relatively large. In certain of these plants, notably Galium palustre, G. saxatile, Potentilla erecta and Trifolium repens, many of the seeds have mechanisms of innate dormancy which cause them to be incapable of germination in the period immediately following dispersal, and this characteristic probably functions as a delaying mechanism increasing the chance of burial.

In respect of seed burial and the enforcement of dormancy upon buried seeds, there are no doubt other mechanisms at work which we do not understand and which may be difficult to quantify. A subject which has not been considered in this study is the role of invertebrates and small mammals in seed burial. Mortimer (1974) has shown that animals may exert an important effect in accelerating seed burial.

Consideration of the mechanisms whereby seeds become buried in the soil forms only part of the analysis which is required to assess the significance of a persistent seed bank. It is equally vital to know how germination is subsequently initiated in the buried seeds, especially since the timing and location of germination are of crucial importance in determining the success of seedling establishment. Studies of the mechanisms by which buried seeds are caused to germinate have been reported briefly by Thompson, Grime & Mason (1977) and will form the basis of future papers.

Seed banks and the maintenance of floristic diversity

The results obtained in this study and their theoretical implications are relevant to the problems of understanding the mechanisms which frequently allow plant species of different biology to co-exist within a plant community. The data in Figs 6-7 provide numerous examples of vegetation in which the component species exhibit widely- contrasted methods of regeneration. In accordance with the concepts summarized by Grubb (1977) we may predict that at each site the identities of the species regenerating most successfully will change from year to year in accordance with fluctuations in climate and in factors such as the timing, distribution, form and severity of vegetation disturbance. One source of variety in regenerative response which is conspicuous in Figs 6-7 is related to differences in persistence of the component seed banks. At each of the sites




examined in this study persistent seed banks were apparent in certain species, and in some instances the plants concerned were relatively rare components of the established vegetation. It seems reasonable to expect that possession of a buried seed bank may allow many of these plants to become more abundant following years in which disturbance of the vegetation and the soil is exceptionally widespread or severe. However, in contrast to the plants with persistent seed banks, there are, at most of the sites, other species in which the seeds remain dormant and/or viable for only a short period of time, and the crop of seedlings which is produced is restricted characteristically either to the spring or the autumn. In marked contrast to the species with persistent seed banks, the regenerative success of these plants in any year will depend crucially upon (1) the output of seeds, and (2) the ability of the resulting offspring to exploit the particular opportunities for regeneration which characterize the year concerned.

Whilst temporal and spatial variation in opportunities for regeneration by seed appear to play a vital part in the maintenance of diversity in herbaceous vegetation, it is necessary to consider also the role of vegetative responses by established plants. The importance of the latter is suggested at certain of the sites in this study by the presence of species in which the seeds appear to be either exceedingly scarce or subject to high rates of predation.

ACKNOWLEDGMENTS

We are particularly grateful to Dr J. G. Hodgson for his expert assistance in the identification of seedlings in field and laboratory and we should also like to thank S. R. Band, A. M. Neal, A. V. Curtis and Miss Jane Rodman who participated in the fieldwork. We are also indebted to Mr M. Wareing who allowed us to work on his property and to the Natural Environment Research Council who provided financial support for this project.

REFERENCES

Al-Mufti, M. M., Sydes, C. L., Furness, S. B., Grime, J. P. & Band, S. R. (1977). A quantitative analysis of shoot phenology and dominance in herbaceous vegetation. Journal of Ecology, 65, 759-791.

Brenchley, W. E. (1918). Buried weed seeds. Journal of Agricultural Science, 9, 1-31. Brenchley, W. E. & Warrington, K. (1930). The weed seed population of arable soil I. Numerical

estimation of viable seeds and observations on their natural dormancy. Journal of Ecology, 18, 235-272.

Champness, S. S. & Morris, K. (1948). Populations of buried viable seeds in relation to contrasting pasture and soil types. Journal of Ecology, 36, 149-173.

Chancellor, R. J. (1966). The Identification of Weed Seedlings of Farm and Garden. Blackwell Scientific Publications, Oxford.

Chippindale, H. G. & Milton, W. E. J. (1934). On the viable seeds present in the soil beneath pastures. Journal of Ecology, 22, 508-531.

Clapham, A. R., Tutin, T. G. & Warburg, E. F. (1962). Flora of the British Isles, 2nd edn. Cambridge University Press, London.

Courtney, A. D. (1968). Seed dormancy and field emergence in Polygonum aviculare. Journal of applied Ecology, 5, 675-684.

Darlington, H. T. & Steinbauer, G. P. (1961). The eighty-year period for Dr. Beal's seed viability experiment. American Journal of Botany, 48, 321-325.

Davison, A. W. (1961). The ecology of Hordeum murinum L. II. The ruderal habitat. Journal of Ecology, 59, 493-506.

Davison, A. W. (1977). The ecology of Hordeum murinum L. III. Some effects of adverse climate. Journal of Ecology, 65, 523-530.

Grime, J. P. (1977). Interpretation of small-scale patterns in the distribution of plant species in space and time. Structure and Functioning of Plant Populations (Ed. by J. W. Woldendorp), pp. 101-124. North-Holland Publishing Company, Amsterdam.




Grime, J. P. & Curtis, A. V. (1976). The interaction of drought and mineral nutrient stress in calcareous grassland. Journal of Ecology, 64, 976-998.

Grime, J. P. & Jarvis, B. C. (1975). Shade avoidance and shade tolerance in flowering plants. II. Effects of light on the germination of species of contrasted ecology. Light as an Ecological Factor (Ed. by R. Bainbridge, G. C. Evans & 0. Rackham), pp. 525-532. Blackwell Scientific Publications, Oxford.

Grime, J. P., Mason, G., Curtis, A. V., Neal, A. M., Rodman, J. & Shaw, S. (1980). A comparative study of germination characteristics in a local flora. New Phytologist, 77.

Grubb, P. J. (1977). The maintenance of species-richness in plant communities: the importance of the regeneration niche. Biological Reviews, 52, 107-145.

Kropac, Z. (1966). Estimation of weed seeds in arable soil. Pedobiologia, 6, 105-128. Lloyd, P. S. (1972). The grassland vegetation of the Sheffield region. II. Classification of grassland

types. Journal of Ecology, 60, 739-776. Major, J. & Pyott, W. T. (1966). Buried viable seeds in two California bunch grass sites and their

bearing on the definition of a flora. Vegetatio, 13, 253-282. Mason, G. (1976). An investigation of the effect of temperature upon the germination and growth of

native species, using temperature-gradient techniques. Ph.D. thesis, University of Sheffield. Milton, W. E. J. (1939). The occurrence of buried viable seeds in soils at different elevations and in a

salt marsh. Journal of Ecology, 27, 149-159. Milton, W. E. J. (1943). The buried viable seed content of a midland calcareous clay soil. Empire

Journal of Experinental Agriculture, 11, 155-167. Mortimer, A. M. (1974). Studies of germination and establishment of selected species with special

reference to the fate of seeds. Ph.D. thesis, University College of North Wales. Roberts, H. A. (1970). Viable weed seeds in cultivated soils. Report of the National Vegetable Research

Station (1969), 25-38. Roberts, H. A. & Dawkins, P. A. (1967). Effect of cultivation on the numbers of viable weed seeds in

soil. Weed Research, 7, 290-301. Sarukhain, J. (1974). Studies on plant demography: Ranunculus repens L., R. bulbosus L. and R.

acris L. II. Reproductive strategies and seed population dynamics. Journal of Ecology, 62, 151-177.

Thompson, K. (1977). An ecological investigation of germination responses to diurnal fluctuations in temperature. Ph.D. thesis, University of Sheffield.

Thompson, K., Grime, J. P. & Mason, G. (1977). Seed germination in response to diurnal fluctuations of temperature. Nature, London, 67, 147-149.

Thompson, P. A. & Cox, S. A. (1978). Germination of the bluebell (Hyacinthoides non-scripta (L.) Chouard) in relation to its distribution and habitat. Annals of Botany, New Series, 42, 51-62.

Watkinson, A. R. (1978). The demography of a sand dune annual, Vulpiafasciculata. II. The dynamics of seed populations. Journal of Ecology, 66, 35-44.

Wesson, G. & Wareing, P. F. (1967). Light requirements of buried seeds. Nature, London, 213,600-601.

(Received 6 October 1978)

APPENDIX

List of the herbaceous species recorded as established plants at the ten field sites. At each site the frequency of occurrence of each species in 100 randomly-distributed 10 x 10-cm quadrats is recorded. Species present at the site but not recorded in the quadrats are represented by the symbol +. Species recorded only infrequently are omitted.

A provisional classification with respect to seed bank type (I-IV) is attempted for species making significant contributions either to the vegetation and/or to the seed flora.

* Figures in parentheses after the species names are average seed weights in mg based upon collection of seeds from native populations in the Sheffield area.

t For an explanation of seed bank types see Fig. 10. Seed bank types and seed weights are given only for those species which appear in Figs 6-7.

a = < 10 seeds recorded throughout the year at any one site. b = > 10 seeds recorded, but data too erratic to allow classification of seed bank.



920 Seasonal variation in seed banks Seed bank

Species type t Sites

1 2 3 4 5 6 7 8 9 10 Achillea millefolium (0-16)* a 37 5 12 3 Agropyron repens (2-02) a 58 4 Agrostis canina (0-05) IV 2 43 Agrostis stolonifera (0-02) a 20 11 57 Agrostis tenuis (0-06) IV 34 + 20 + 14 36 Alopecurus geniculatus (0-22) a 5 4 Angelica sylvestris (.1-5) a 32 2 Anthoxanthum odoratum (0-45) III-IV 33 6 28 Arrhenatherum elatius (2-39) I 23 29 17 1 9 Berula erecta (0-49) a 39 Betonica officinalis (1-37) a 29 Brachypodium pinnatum (2-85) a 72 Briza media (0-23) I 2 19 19 Bromus erectus (4-23) I 36 Calluna vulgaris (0-03) IV 18 Campanula rotundifolia (0-07) IV 13 10 28 Cardamine pratensis (0-6) a 12 41 Carex caryophyllea (1-19) a 63 35 24 Carexflacca (0-37) a 48 5 11 1 1 Carex nigra b 4 37 Carex otrubae (1-27) b 27 Centaurea nigra (2-55) b 25 2 27 Centaurea scabiosa 10 1 Cerastium fontanum (0-16) III-IV 4 7 Chamaenerion angustifolium (0-05) I + 4 + 11 Cirsium arvense 1 2 3 1 Cirsium palustre (2-0) IV 9 11 Dactylis glomerata (0-51) I 1 18 7 24 2 3 Deschampsia cespitosa (0-31) III + 7 26 3 Deschampsia flexuosa (0-46) a 20 93 15 Digitalis purpurea (0-07) III + Endymion non-scriptus (6-17) II 36 1 Epilobium hirsutum (0-05) III 29 Epilobium palustre 7 2 Equisetum arvense 1 3 Equisetum sp. 15 Festuca arundinacea 2 7 Festuca ovina (0-38) I 3 73 26 45 4 Festuca rubra (0-79) I 1 90 20 12 19 3 13 Filipendula ulmaria 6 3 Galium aparine 5 2 Galium palustre (0-91) IV 20 35 Galium saxatile (0-56) IV 9 32 Galium sterneri 14 5 Galium verum 13 4 3 Hedera helix 1 1 1 Helianthemum chamaecistus (0-32) a 23 12 Helictotrichon pratense (2-08) I 8 47 Helictotrichon pubescens (1.92) a 47 5 Heracleum sphondylium (5-52) II 3 2 6 2 Hieracium spp. 6 3 4 3 Hieracium pilosella 15 4 Holcus lanatus (0-32) III 41 10 3 1 50 Holcus mollis (0-19) a 31 3 1 Hydrocotyle vulgaris a 48 Hypericum hirsutum 4 2 Hypochaeris radicata 13 Juncus articulatus (0-02) IV 11 + Juncus effusus (0-01) IV 41 19 + Juncus inflexus (0-03) IV 9 +




Seed bank Species type t Sites

1 2 3 4 5 6 7 8 9 10 Koeleria cristata (0-3) I 35 Lathyrus montanus 1 3 + Leontodon autumnalis 1 3 Leontodon hispidus 1 15 17 Lolium perenne (1.79) I 11 Lotus corniculatus (1-67) a 26 23 33 2 4 Lotus uliginosus 2 3 Luzula campestris 5 3 6 Mercurialis perennis (2.17) II 1 29 Molinia caerulea 2 8 Nardus stricta (0.38) b 5 3 Origanum vulgare (0.10) IV 57 15 Pimpinella major (2-12) II 27 Pimpinella saxifraga (1.19) II 22 20 Plantago lanceolata (1.9) III-IV 13 31 37 1 Poa annua (0-26) III-IV + + 68 Poa pratensis (0-25) a 19 5 4 4 7 18 Poa trivialis (0-09) III 17 8 15 + 44 Polygonum aviculare (1.45) a 22 Potentilla erecta (0-58) IV 2 38 7 Potentilla sterilis 3 1 1 Pteridium aquilinum + 3 1 Ranunculus ficaria a 17 Ranunculus repens (2-32) a 1 3 20 4 Rumex acetosa (0.74) a 3 3 + 20 1 Scabiosa columbaria 2 10 Sieglingia decumbens (0-87) a 3 3 3 Silene dioica 3 1 Silene nutans (0-28) IV 1 11 Stachys sylvatica (1-4) a + 3 Succisa pratensis (1.54) a 21 Thymus drucei (0.11) IV 20 4 Trifolium medium 19 Trifolium pratense 2 2 Trifolium repens (0-56) IV 1 3 4 8 11 Tripleurospermum maritimum (0-29) b 7 Trisetum flavescens 11 2 12 Triticum aestivum 7 Tussilago farfara 7 Typha latifolia 7 Urtica dioica (0-20) IV 4 + Vaccinum myrtillus (0-26) a 74 5 Valeriana officinalis (0.95) a 46 Veronica chamaedrys 4 1 1 + Vicia angustifolia 11 Vicia cracca 1 4 2 Viola hirta 1 2 + Viola palustris (0-63) II 11 Viola riviniana (1.01) II + 20 3



seasonal variation in the seed banks of herbaceous species ... · 894 seasonal variation in seed...

Documents