key: cord-0042859-t3qxscbp
authors: Losvik, Mary H.
title: Plant species diversity in an old, traditionally managed hay meadow compared to abandoned hay meadows in southwest Norway
date: 2008-06-28
journal: Nord
DOI: 10.1111/j.1756-1051.1999.tb01231.x
sha: a7865107a35f3c6c1c716b1364efcd8f19493efb
doc_id: 42859
cord_uid: t3qxscbp

A chronosequence, representing a successional series, was used for the comparison of a hay meadow site managed in an old traditional way for at least a hundred years, and hay meadow sites abandoned for about 10, 20 and 30 years, respectively. Old traditional management included grazing early and late in the growing season, mowing in August and light or no fertilizing. The tree cover was the most important factor deciding the composition of vegetation. Time since abandonment was not completely correlated to tree cover, as some plots had a dense canopy and others were situated in the open. The total species number decreased with number of years since abandonment in plots > 0.001 m(2)and <100 m(2). The highest species number in 1 m(2) plots was recorded in the managed site, with 38 species of phanerogams. Fourtyeight % of the indicators of traditional management present in the managed site was recorded in the site which had been abandoned for 30 years. Frequency — log area curves made it possible to group species according to persistence in the sward. As a result, a group of functional indicators of rare hay meadows in the region was distinguished.

At small scales, temperate grasslands are the most species-rich plant communities in the world Shmida & Ellner 1984) . Grasslands were once widespread in western Europe, but are now greatly reduced in extent as a result from changes in agricultural land-use practices (Keymer & Leach 1990; Willems 1990) . The grasslands may have high species numbers, up to well above 60 per m2 (Willems 1978 (Willems , 1982 Kull & Zobel 1991) . Old, traditionally managed hay meadows are rare in western Norway at present (0vstedal 1985; Lundekvam & Gauslaa 1986; Losvik 1988a Losvik , 1988b Losvik , 1993a Rosef 1996) . Earlier they were common elements of the agricultural landscape, but most of them are now either cultivated in a modern way or abandoned (Losvik 1996a ). The few old, traditionally managed hay meadows which are left, are mostly situated on phyllittic soils and have high species diversity. No investigation in the region has so far studied what happen to the species diversity and species composition when such meadows are abandoned.

In this study the species have been grouped into indicators of traditional management, common (hay meadow) species and additional species, the latter includes forest species. The species group indicators of traditional management was preliminary defined in Losvik (1993b) , and later (Losvik 1996b ) was shown to comprise species occurring in less than 60 YO of 548 plots (1-16 m2) of hay meadow vegetation analysed in western Norway by 6 different authors 1972-1991. These species are often low-statured, creeping, slender andor with rosettes, and mostly stresstolerators (Grime 1974 (Grime , 1979 ) and hemicryptophytes, annuals or biennials (Raunkiaer 1934) . The number of the indicators is usually low in abandoned sites (see data in Losvik 1981; Persson 1984; Austad & Skogen 1990; Borgegaard & Persson 1990) . Modern management, including medium or heavy fertilizing, also reduces species number in general (see e. g. During & Willems 1984; Willems et al. 1993) , and especially the indicators of traditional management (Losvik 1988b (Losvik , 1993a (Losvik , 1996b . As a consequence, indicators of traditional management are becoming increasingly more rare in the region. Common species are defined as being more frequent than indicators of traditional management in the agricultural landscape, tolerating medium quantities of fertilizer. These groups of species is considered to be more useful in choosing areas for hay meadow conservation, than the plain use of total number of species, as it reveals direct information on the number of target species, namely the indicators of traditional management in the sites.

The changes in plant species richness during succession have been studied e. g. by Nicholson & Monk (1974) , Bazzaz (1975) , Prach (1985) and Symonides (1985) . Both increase and decrease in total plant species diversity have been found in these studies, and the situation may be further complicated by environmental gradients (Peet 1978; Prach 1986) . As the species diversity depends on the spatial scale considered (e. g. Kwiatkowska & Symonides 1986 ; van der Maarel 1988a) it may be better described by the species-area relationship than by single species numbers.

The aim of the study was to compare the vegetation and ecology of 4 sites which formed a chronosequence, representing a successional series from a species-rich, old, traditionally managed hay-meadow to sites which had been abandoned for about 10, 20 and 30 years, respectively. Themes of special interest were: how important are the ecological factors years since abandonment and cover of tree canopy in deciding the composition of the vegetation, what differences in species diversity, as measured by species-area curves, are there between the sites, for how long are species which are characteristic for the agricultural landscape able to persist during overgrowing in such sites, and finally, what forest species and additional species are most important in the sites of this series of abandoned haymeadows.

The 4 study sites are located in a hill-side at Gjuvsland, Varaldsey in The Hardanger fjord, western Norway, generally sloping 25" towards south-east, from about 100 m.a.s.1. down to the shore. The hill-side was formerly common land, used as out-lying hay meadows. They were mown in August and grazed in spring and in autumn by sheep and cattle from 6 farms at Gjuvsland. In 1906 the hill-side was divided between the farms into 6 parts which each was fenced. Four of these parts constituted the study sites.

The bedrock is volcanic supercrustal rocks covered by scree material containing phyllite (Foslie 1955; see also Kolderup 1960; Askvik 1976 ). The soil is probably enriched with Ca through seepage from limestone bedrock occurring above the hay meadows. The mean precipitation at the nearest meteorological station Rosendal is 1799 mm year' (Ferland 1993) . The mean temperature in January is 0.5" C, in July 15.5" C, and mean yearly temperature is 7.2" C (at the nearest meteorological station measuring temperature, Omastrand, Aune 1993) .

Traditional management includes grazing by sheep for about 14 days in MayIJune and in September/October each year. Until about 50 years ago, mixed grazing by cattle and sheep was common, but later sheep alone were used. Stocking rates are about 300 sheepgrazingdaysha and year (including lambs), less in dry years. After the spring grazing period, dried leftovers of manure and twigs of trees are removed by raking. The grass is mown in August, using a small reaper and different forms of scythes. August may be rainy (Semme 1954) , but the farmers usually awaits a period of dry weather to allow drying of the grass on the ground. The grass is turned on the spot twice a day till it is dry, allowing seeds to be spread evenly thoughout the grassland. Usually no commercial fertilizers or wintermanure are applied, but some nutrients are recycled through urine and faeces during grazing, particularly when mixed grazing, involving higher grazing intensity than at present, was practised (Brelin 1979) . The main features of these management practices were common in western Norwegian out-lying hay meadows till about 1945 (Byrkjeland 1958) . As long as the whole hillside was managed, there was a scattered treelayer of pollards (mostly Fraxinus excelsior) or coppiced trees (Corylus avellana and Alnus incana). These trees and characteristic forest herbs and grasses often grew near outcrops or heaps of rocks, cleared away from the meadows, and functioned as centres of natural afforestation as management ceased. At the study site this has resulted in a mosaic-like pattern of patches with (1) a high, dense sward of grasses and herbs, lacking a tree layer, (2) a scattered field layer and a more or less closed canopy of trees, and (3) a low dense field layer in areas in the open which are situated close to small outcrops, narrow paths, ant-heaps and springs.

Site A is managed in the old, traditional way (A. A. Gjuvsland, pers. comm.) . But since about 1990 the number of sheep has been reduced to about on third. This reduction has, at least partly, been compensated for by longer grazing periods. The low-lying part (about %) of site A was lightly fertilized (about 100 kg fertilized ha.year, type varied) in June from about 1950 till 1990 (in the 60ties at most 200 kg fertilizer/ha.year). The soil is rich in minerals (base saturation: 54-76 %), and poor in organic matter (C: 2.3-3.8 %), nitrogen (0.3 1-0.47 %) and phosporous (0.2-0.5 mg/100 g dry matter) with pH 5.1-5.6 (Losvik 1988a) . The site has a rather high species diversity (Losvik 1985 , 33-38 species of phanerogams/l6 m2), compared to other grasslands in south-west Norway (0vstedal 1985; Lundekvam & Gauslaa 1986; Losvik 1988a Losvik , 1988b Losvik , 1991 Losvik , 1993a Hundt & Vevle 1990 ). Species such as Cynocurus cristatus, Briza media, Trifolium dubium and Linum catharticum, which are rare in western Norwegian grasslands at present (Losvik 1996b) , are abundant in the site. An assembly of uncommon grassland species of which usually only a few are recorded from each hay meadow site in western Norway, such as Leucanthemum vulgare. Euphrasia stricta. Rhinanthus minor, Polygala vulgaris, Centaurea jacea, Pimpinella saxifraga, Carum carvi, Danthonia decumbens, Lotus corniculatus and Platanthera chlorantha, are here found together in an area of about 1 ha. The traditional management has lowered the content of N and P in the soil and resulted in high species diversity, as have also been recorded in other studies (e. g. Kowalsky 1964; van der Maarel 1971; Silvertown 1980; Zoller & Bischof 1980; Collins & Barber 1985; de Leeuw & Bakker 1986 ). At the time of the investigation, the managed site had a scattered tree layer of Fraxinus excelsior, Alnus incana, Taxus baccata and old individuals of planted Prunus domesticus. Sites B, C and D were abandoned 1985 , 1975 and 1965 . Site C was grazed by sheep the whole summer for some years after the mowing ceased. In site D light grazing might have occurred for some years after abandonment, as sheep from neighbouring parts went trough broken fences. Site B was lightly fertilized in the 1950ties.

It is assumed that the vegetation was much the same over the whole hillside at the time when the management was the same, especially as the management probably had a tradition of a hundred years at least (A. A. Gjuvsland, pers. comm.) . Ecological factors such as as-pect and bedrock were mainly equal over the whole hillside. Site C is generally steeper than the other sites and slopes about 30". According to intervjues the tree structure of the 4 sites seem to have been quite similar, with less than 20 % tree cover of scattered trees.

The nomenclature follows Lid 1994.

The size of sites A-D was 1, 0.8, 0.4 and 1.2 ha respectively. The whole area of each site was divided into 10 equally-sized subsites, in order to assure that the data set should represent well the vegetation of the site. Steep, stony or otherwise inaccessible areas of the sites were excluded. Within each subsite the southwestern comer of one square plot of 100 cmz was chosen at random. Four subplots of 2 x 5 cm, placed in the comers of this plot were analysed in addition to a 20 x 50 cm plot, a 1 x 1 m plot, a 2 x 5 m plot and a 10 x 10 m plot, each made by extending the former plot towards south-east and north-east. Thus the plots within each subsite were nested, while all the ten 100 m2 plots were distinctly separated, each in its subsite. Presence of all rooted phanerogamic species of the field layer were recorded in 1995 or 1996. Shade was estimated as the cover of the tree layer in each 100 m2 plot. The number of years since abandonment, in addition to records of former management, was obtained by interviewing each farmer. Main ecological gradients in the data set and length of the gradients were assessed by Detrended Correspondence Analysis, DCA (Hill 1979; Hill & Gauch 1980) applied to presence-absence data for 137 species (species occurring in < 5 plots were omitted) in 40 quadrates of 100 m2 each, using CANOCO version 3.12 (ter Braak 1987a (ter Braak , 1990 . Canonical Correspondence Analysis, CCA (ter Braak 1986 (ter Braak , 1987b ) was used to evaluate the importance of the environmental factors: years since abandonment and cover of trees. Standard options were used. Monte Carlo test (ter Braak 1990) with 99 permutations was used to assess the significance of first canonical axis and overall significance.

Specieslog area curves, frequencieslog area curves, and the significance of differences in species richness (t-test) were constructed or calculated using EXCEL sistence in the sites and to frequency in the different plot sizes of the frequencylog area curves.

DCA indicated a gradient of shade along the first axis (eigenvalue 0.37, 2.8 SD) and of nutrients along the second axis (eigenvalue 0.15,2.2 SD). The indicators of traditional management and the common species as expected were situated in the light part of the first axis gradient, while the additional species, comprising forest species and species common along edges and roadsides, were mainly in the opposite direction (Fig. 1 a) . When the plots were classified according to tree cover, the gradient of shade along the first axis was clearly demonstrated (Fig. lb) . Grouping of the plots according to sites showed small variations along the gradients in sites A, B and C compared to site D (Fig. lc) . The plots within each site were more similar to each other than to any plot outside the site along axis 1 and 2. CCA demonstrated that tree cover (Cover) as expected was positively correlated with the first axis (Fig. 2 , eigenvalue 0.28). Years since abandonment (Abandon) was positively correlated to both the first and the second axis. Axis 1 accounted for 21.9 % of the species data, axis 2 accounted for only 4 %. Significance (p) of the first axis was 0.01, and an overall test also gave significant results.

Of the 154 species recorded in sites A-D, 35 % were absent in the plots of site D, 40 % were absent in site A (Appendix 1, 2) . Seven % of the species was only recorded in A, 12 % only in D. The speciesarea curves of site A for the total number of species, indicators of traditional management and common species were nearly linear, but the B-D curves increased exponentially with log area. Construction of log species log area curves made the total species number curve of site D close to linear, but the other curves then turned into power function curves. None of the curves showed any tendency to flatten out at the investigated size ranges.

The mean total species number for plot sizes >0.001 m2 and < 100 m2 was higher in the traditionally managed site A than in the abandoned sites (Fig. 3a , p < 0.001). Sites B and C had intermediate mean species numbers, while site D had a lower number of species than sites A-C in these plot sizes (p < 0.01). The highest species number in 1 m2 plots was recorded in site A with 38 species (mean 32, SD 3) and in the 100 m2 plots in site C with 60 species (mean 50, SD 7), followed by site A with 56 species (mean 49, SD 5). When all species in the 10 plots of each site were added ('plot' size 0.1 ha) however, site B had the highest species number (Appendix 1,2). The number of indicators of traditional . Site A had the highest and site D the lowest mean number of common species in plot sizes from 0.01 m2 to 100 m2 (Fig. 3c) . The differences in mean number of additional species between sites were negligible for plot sizes up to 1 m2 (Fig. 3d) , but for larger plot sizes the mean number was lower in site A than in sites C-D (p < 0.01) and for the largest plots it was also lower in site A than in site B (p < 0.01).

The species were grouped according to presenceifrequency of each species in all plot sizes in sites A-D (Appendix 1, 2). Group 1 mostly comprised indicators of traditional management and common species, species which are more or less dependent on high light availability. The frequency of species of group la tended to be much lower in sites B-D for all plot sizes than in site A, if they were present there at all (Fig. 4a) . Species of group lb (Fig. 4 b) persisted rather well as long as there were openings in the tree canopy. Other agricultural landscape species were rather shade tolerant (group lc, Fig. 4 c) , and some species even increased in frequency after abandonment (group Id, Fig. 4d ). Group 2a comprised border species which benefit from the lack of management in addition to the high light availability in

the early successional phases (Fig. 5a) , and group 2b comprised forest species (Fig. Sb) . The rest of the species occurred too scattered in the sites to be grouped.

The ordination results confirmed that low tree cover was important for the indicators of traditional management and the common species, while the additional species may tolerate well both abandonment and a high tree cover, even if the species may be present also in the traditionally managed hay meadows. With the assumption that the vegetation in sites B-D was about the same as in site A when they were managed, DCA indicated that open abandoned areas gradually became poorer in nutrients with time while areas with a closed tree canopy became richer (see Figs l b and lc, plots B1, D7, D9-10). Austad & Skogen (1990) recorded both rather poor Betula pendula forest and rich forest types with e. g. Ulmus glabra in abandoned meadows in a hillside in Sogn. In Losvik (1981) it is concluded that successional trends in abandoned hay meadows take different courses according to nutrient content in the soil. mown, soil analysis will be necessary to prove that the differences in vegetation resulted from differences in soil nutrients. A theory would be that litter of trees, here mainly ash (Fraxinus excelsior), decomposes more rapidly than the grass sward of the open areas, releasing more nutrients to the soil in the studied time span of abandonment. Bearing these statements in mind, an interesting interpretation of the CCA result may be that tree cover, according to the distribution of species in the diagram, seemed to be positively correlated to nutrients, while abandonment in general seemed to be negatively correlated to nutrients. In an abandoned area, tree cover, and with it the area which is in shade will increase with time. But the increase is not evenly distributed in space, as some areas are open for a considerable time span, and as such may be poorer in nutrients than areas with a tree canopy. This implies that the composition of the vegetation in an abandoned area is dependent not only on time since abandonment, but also on the local extent of the tree cover. The occurrence of seedlings of ash (Fraxinus excelsior) in at least 7 out of 10 plots of 0.1 m2 in sites B -D (Fig. 5b) indicate that the open areas of these sites is really in a transitional phase, during which the content of nutrients in the soil may be lower than in areas with a closed canopy. The rate of increase in species number with increase in plot size is an appropriate measure of richness, instead of using number of species in one plot size (Kilburn 1966; van der Maarel 1988a; Leps & Stursa 1989) . But this rate ought to be constant along the curve, and as the species-area curves differed too much in shape in this study, only differences in species numbers between sites at the different plot sizes were compared. Singh et al. (1996) curves were close to linear, while in sites B-D, with a large turnover of species, the curves were exponential. The form of the curves is dependent on density of individuals and plant unit areas. In the traditionally managed hay meadow the species were mostly small with individuals of the species quite evenly distributed. The much smaller variance in plots of site A compared to site D (Fig. lc) demonstrated that in site A the vegetation was more homogenous (van der Maarel & Sykes 1993) . Wherever the analysis would have started, comparatively many species would have been added by increasing the area (Fig. 3a) . In the abandoned meadows, the plant unit area was probably larger and there was a tendency for a few species to dominate in the plots. Both invaders and persistent agricultural landscape species had a contemporary scattered presence in the plots. Thus the linear log area curves may indicate homogenous stands, while exponential specieslog area curves may indicate succession or disturbance. Continuos increase of the curve at and even above the investigated plot sizes is recorded e. g. by Hopkins (1955) and Barkman (1989) . Both tree cover and management was important in deciding the species richness in the investigated chronosequence. In a wooded meadow in Estonia Kull & Zobel (1991) also found the highest species richness where tree canopy cover was lowest (see also Pausas 1994) , and highest species richness in sites with the most regular long-term mowing as compared to cases of cessation of mowing. For semi-natural grasslands van der Maarel (1988a) argued that grazing animals and mowing imply disturbance and to some extent stress, which enable more species to coexist than on a similar area without grazing or mowing. The reasons for the high mean species richness in site A compared to the other sites may be complex (see e. g. Giller 1984 ), but the management itself undoubtedly plays a major role in providing light and gaps for all the species at the right time and place throughout the growing season. In the abandoned sites less light penetrated into the lowest part of the field layer and this resulted in exclusion of low-growing species (Kull & Zobel 1991) . The species may either disappear after a gradual reduction in population size, or there may be a collapse in the occurrence of the species. Some of the species in this study (Appendix 1, group la) may have experienced such rapid (in less than 10 years) disappearance. The species diversity decline in sites B-D in 10-20 years was 21% -57 % in plots of 0.01 m2 -10 m2 (Table 1) . However, when the whole investigated area in each site was considered, the total number of species increased in all the abandoned sites (Appendix 1, 2) . This clearly demonstrated that use of total species number is very dependent on plot size and that it may be rather useless in choosing hay meadows for conservation. High species turn-over rates during the first 50-60 years of a sere was reported e. g. by Houssard et al. (1980) . Much more rapid decline in species diversity, 70% in 10-15 years, was reported from chalk grassland by Willems (1990) . In time sites B-D will turn into deciduous woodland with a closed canopy and the indicators of traditional management and common grassland species will even- 1.7) 19(4.5) 5 O( 6.8) 3 5( 9.8) tually be lost. The fact that the mean total number of species in plots of site A was higher for areas > 0.001m2 and < 100 m2 than in the other sites, and at the same time total species number of the whole investigated area in site A was smaller than in the other sites, showed that in site A most species were represented by a large number of individuals which were quite evenly distributed over the whole area of the site. This can easily be seen from Appendix 1, as it appears that in site A 28 species are present in all 10 plots of 100 m2 each, while in sites B-D the numbers are 1, 19 and 4, respectively.

On the other hand, the number of indicators of traditional management was highest in site A, both when total number in investigated plots and when mean number at all investigated plot sizes were considered (Appendix 1 and 2, Fig. 3b ). The usefulness of the group indicators of traditional management as functional indicators in choosing hay meadows in the region for conservation is thus demonstrated (see also McIntyre & Lavorel 1994). Common grassland species are quite frequent in the studied abandoned sites. They thrive when biomass is no longer removed, resulting in increased nitrogen levels in the soil. The additional species occur very scattered in the whole chronosequence, and are thus generally infrequent in small plots. In larger plots the difference, mainly between site A and site D become visible. In site D more forest species have established than in the other sites. As site C had been abandoned for a longer time than site B, more forest species had established there, and at the same time more light demanding species persisted (Appendix 1,2) probably a result from the period with grazing after cessation of mowing, and thus species richness was higher there than in site B. Similar species overlap, occurring when competitive exclusion has not yet had time to drive subordinate species to extinction is advocated also by Palmer (1994) . Higher species richness in grazed areas as compared to abandoned grassland was reported by RegnCll (1980) . The groups of species used in the present study differs from guilds sensu Pianka (1983) in that no species can be a member of different groups, moreover it is no rea-son to believe that species within one group interact more intensively, mainly by competition, than species from different groups (van der Maarel 1988b). The species of group 1 differed in their ability to persist in the abandoned hay meadows. The groups la and lb comprised some of the most rare hay meadow species in western Norway at present. In abandoned hay meadows they are often recorded close to or at refuges like outcrops or heaps of stones on dry or shallow ground in the open, where the field layer is low and scattered. Some of them are annuals which are dependent on gaps in the sward for germination of their seeds. For example Pimpinella saxifraga is known to be very sensitive to increased density of the sward (Grubb 1990) . These species will presumably not be able to survive the closing up of the canopy where they grow at present, and thus they face extinction at their sites in the near future. At sites in western Norway where these species have been recorded during the last 20 years, they probably will experience a collapse extinction as indicated by the differences in the frequencyarea curves. Hay meadows which have been abandoned less than 20 years ago are getting increasingly rarer, and so are the rare species within these groups. Other species in these groups are as vulnerable to abandonment as the rare species, but are generally more common because they tolerate the low or medium quantities of fertilizers used in the westem Norwegian small scale and part time farming system. Thus a smaller group of species, tolerating neither fertilizing nor abandonment may be delimited and used as indicators in choosing hay meadows for conservation (Appendix 1, 2: *). These species occur in less than 23 % of analysed hay meadow plots in western Norway published 1972 (Losvik 1996b , while indicators of traditional management occurred in less than 58 % of the plots. Several rare hay meadow species, not recorded in the plots of this study, such as Botrychiurn lunaria, Dianthus deltoides, Galiurn verurn. Gymnadenia conopsea and Lychnis viscaria, will have to be added to complete this list of indicators of rare hay meadows.

Generally the species which gradually decreased in frequency with time since abandonment or increased in early successional stages (Groups Ic and Id) were plants of some height and thus good light competitors in this situation. The groups comprised indicators of traditional management which are more common than species of groups la and lb, as a result from their tolerance of shade and/or light fertilization. Unusual long stalks, bringing them up towards the top of the field layer were observed in individuals of Potentilla erecta and Lotus corniculatus in the studied abandoned sites. Many border and forest species do in fact occur even in the traditionally managed hay meadow site. These species are usually growing close to stone heaps and below trees. They constitute a species pool and are able to expand as soon as the hay meadow is abandoned. The existence of a forest and border line species pool in managed hay meadows must be an old feature, as the whole of these landscapes were used very intensively under the old traditional management regimes, and thus the present forest species must have had refuges somewhere in the managed landscape.

The investigation shows that in order to preserve the hay meadow diversity, the continuity in management is crucial. Even with short periods of abandonment there is a risk of loosing species. The longer the period of abandonment the more species become extinct. With continuously smaller populations and eventually increasing distance between the fragmented area o f the populations, more and more species run the risk of becoming extinct in the site. Therefore it is very important to try to retain the management in areas with traditional landscape diversity and typical population structure. 

Frequency of species which were present in < 4 out of 10 plots of 100 m*each in the investigated sites. T: indicator of traditional management, * indicator of rare hay meadows. C: common grassland species, F: forest species. A: additional species, Abb.: Abbreviations of species names. 

Mixed grazing with sheep and cattle compared with single grazing

Husdyrbruket i Hordaland gjennom hundrad Ir. -Hordaland landbruksselskap

Effects of disturbance and diversity in mixed-grass prairie

Sheep-grazing with different foraging efficiencies in a Dutch mixed grassland

Diversity models applied to a chalk grassland

Nedbarsnormaler, normalperiode 1961-1990. -Rapport DNMI 39/93. Klima

Kisdistriktet 0lve-Varaldsey og bergverkdriftens historie

Community structure and the niche

Vegetation classification by reference to strategies

Demographic studies on the perennials of chalk grassland

DECORANA-a FORTRAN program for Detrended Correspondence Analysis and Resiprocal Averaging. -Ecology and Systematics departement

Detrendent correspondence analysis: an improved ordination technique

Hordalands berggrunnsgeologi

Temperaturnormaler, normalperiode 1961 -1990. -Rapport DNMI 02/93. Klima

Restoration of a deciduous woodland in western Norway formerly used for fodder production: effects on tree canopy and field layer

A critical evaluation of minimum area concepts

Plant species diversity in old-field successional ecosystems in southern Illinois

Calcareous grassland -a limited resource in Britain

Analysis of the species-area relation

Hardanger and Sogn

Ecology of mountain grasslands and use by elk

High species richness in an Estonian wooded meadow

Spatial distribution of species diversity indices and their correlation with

Species-area curve, life history strategies, and succession: a field test of relationships

University of Bergen, Bergen. -1988b. Phytosociology and ecology of old hay meadows in Hordaland, western Norway in relation to management

Phytosociology and ecology of mown grasslands in western Norway

How environmental and disturbance factors influence species composititon in temperate Australian grasslands

Microsoft Corporation 1992. Microsoft Excel. Brukerhhdbok 1. -Microsoft Corporation

Plant species diversity in old-field succession in the Giorgia Piedmont

Variation in species richness: towards a unification of hypotheses

Species richness patterns in the understorey of Pyrenean Pinus sylvestris forest

Forest vegetation of the Colorado front range: patterns of species diversity

Prediction of man's impact on plant species diversity

Vegetation development after the exclusion of grazing cattle in a meadow area in the south of Sweden

Evolutionary ecology

Succession of vegetation in abandoned fields in Finland

The life forms of plants and statistical plant geography

A numerical study of successions in an abandoned, damp calcareous meadow in S Sweden

Apen eng og beitemark i drift, Voss kommune, Hordaland -vegetasjon og skologi. -Cand. scient. thesis

Coexistence of plant species with similar niches

The dynamics of a grassland ecosystem: botanical equilibrium in the Park grass experiment

Plant species richness and species-area relations in a shortgrass steppe in Colorado

Floristic richness, diversity, dominance and species evenness in old-field successional ecosystems

Geography of Norwegian Agriculture. -Publ. norw.univ. school of econ. & business adm. Geographical series 3A

CANOCO -a FORTRAN program for canonical community ordination by (partial) (detrended) (canonical) correspondence analysis, principal components analysis and redundancy analysis (version 2. I). -Agricultural Mathematics group, Wageningen. -1987b. The analysis of vegetation-environment relationships by canonical correspondence analysis. -Vegetatio -1990. Update notes: CANOCO version 3.10. -Agricultural Mathematics group, Wageningen. van der Maarel, E. 1971. Plant species diversity in relation to management

Small-scale plant species turnover in a limestone grassland: the carousel model and some comments on the niche consept

Composition and species diversity of pine-wiregrass savannas of the Green Swamp, North Carolina

Observations on North-West Europaean limestone grassland communities. 111. Phytosociological and ecological notes on chalk grasslands of Southern England

Changes in chalk-grassland structure and species richness resulting from selective nutrient additions

Stufen der Kulturintensitat und ihr Einfluss auf Artenzahl und ArtengeGge der Vegetation

The vegetation of LindAs and Austrheim, western Norway

Acknowledgements -The project was supported by 0.

Kishorekumar for the assistance with the input of data, and to three unknown referees for comments on the manuscript.