• conservation;
  • demography;
  • ecological restoration;
  • population viability;
  • seedling recruitment


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Appendix
  • 1
    We need to understand how threatened plant species respond to natural or management-induced habitat changes to conserve them successfully. Because long-term demographic studies are not feasible for large numbers of species, there is a clear need for simple short-term methods to assess demographic responses.
  • 2
    The population structure of 23 populations of the endangered perennial Salviapratensis was studied in relation to the vegetation and management of dry floodplain grasslands along Dutch rivers. The aims of the study were to (i) evaluate the suitability of single censuses of population stage structures as a tool for quick assessments of the viability of plant populations; (ii) test whether viable populations of S. pratensis indicate sites of higher conservation value; (iii) obtain information on the viability of the remaining populations; and (iv) provide advice for optimal habitat management.
  • 3
    Using cluster analysis, we distinguished three different types of populations: (i) ‘dynamic’ populations, characterized by a large proportion of young individuals (seedlings, juveniles and immatures); (ii) ‘normal’ populations, with a relatively higher proportion of adults but still a considerable number of young individuals; and (iii) ‘regressive’ populations, in which adult stages, especially large flowering individuals, dominated and rejuvenation hardly occurred. The three population types differed with respect to population size and total plant density, which were highest in dynamic, intermediate in normal and lowest in regressive populations.
  • 4
    Both the structure and composition of the surrounding vegetation were associated with the type of population found. The percentage of bare soil surface (indicating an open vegetation structure) was positively related to recruitment of S. pratensis. The dynamic populations occurred in a species-rich vegetation, comprising species of nutrient-poor soils and characteristic of floodplain grasslands. The regressive populations occurred more often in species-poor vegetation, comprising mainly species of nutrient-rich conditions.
  • 5
    Salvia population structure was correlated with management. We observed that those populations with a late mowing regime had higher proportions of young stages and larger population sizes. Although we expected that recently established populations on the young river dunes formed during ecological restoration projects would be dynamic, most sampled plots were of the regressive type, which suggests that either site conditions were not yet optimal or Allee effects limited further expansion of the small founder populations.
  • 6
    We conclude that conservation of the remaining populations of S. pratensis in Dutch dry floodplain grasslands will be best achieved by late mowing with hay removal.


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Appendix

The evaluation of changes in habitat quality or ecological conditions often occurs on the basis of vegetation monitoring. In addition, changes in frequencies or abundances of rare species with a narrow ecological tolerance are frequently used as indicators for local ecological change. Nevertheless, observed changes in abundance are only rarely linked to demographic data. In the case of plant conservation, it is frequently assumed that rare species occurring in specific plant communities will benefit from management measures that maintain the community. However, because the demographic responses of endangered or target species to this community-based management are often largely unknown, the outcome is generally unpredictable. Counts of flowering plants yield very little information on population viability and its relation to different management regimes. Particularly when there is no time or money for detailed demographic monitoring of individual plants over a period of years (Menges & Gordon 1996), information on population structure can reveal important information (Oostermeijer, van’t Veer & den Nijs 1994).

A relatively simple way to describe a species’ population structure is by determination of the relative densities and/or proportions of the different ‘age states’ (cf. Gatsuk et al. 1980; Rabotnov 1985) or ‘life stages’. This information can be used to describe the demographic status and future outlook of each population and to relate this to environmental features, such as the surrounding vegetation structure and composition, soil conditions and management. The method was used to assess population viability of Gentiana pneumonanthe in Dutch and Norwegian wet heathlands and hay meadows (Oostermeijer, van’t Veer & den Nijs 1994; Oostermeijer et al. 1996) and recently on Succisa pratensis as an indicator species of Swiss calcareous fens (Bühler & Schmid 2001). In several studies on other (rare) plant species, the concept has proved useful to describe the successional stage or demographic viability of populations (Morgan 1997; Gusewell, Buttler & Klotzli 1998; Hutchings, Mendoza & Havers 1998; Jones 1998; Dietz, Fischer & Schmid 1999; Schmidt & Jensen 2000).

The main aim of the present study was to analyse the population stage structure of the rare perennial plant Salvia pratensis L. in order to (i) learn more about the interactions between different life stages and habitat and identify factors that are correlated with demographic viability; (ii) test whether viable populations of S. pratensis are also indicative of sites with higher conservation value in terms of vegetation composition and species richness; and (iii) evaluate further the suitability of using the population stage structure as an indicator of the demographic viability of a rare plant species.

We hypothesized that populations of S. pratensis can be classified into different types that vary in the relative proportions of young life stages and that can be related to vegetation structure. Furthermore, we expected that viable (i.e. growing or stable) populations of S. pratensis occur in more species-rich vegetation types of higher conservation value. Finally, we hoped to identify management regimes that positively affect the demographic viability of S. pratensis in order to provide advice on the best short-term conservation management.

Materials and methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Appendix

study species

Salvia pratensis (Lamiaceae) is a long-lived perennial that can grow to 1 m. Germination occurs mainly in autumn (Muller 1978), and there is no persistent seed bank (Thompson, Bakker & Bekker 1997). There is a critical rosette size for reproduction, which is usually reached 4–5 years after germination (Ouborg & van Treuren 1994, 1995). Plants normally produce one to five flower stalks with violet-blue flowers, but we have observed individuals with up to 50 flower stalks. The species is pollinated primarily by bumblebees (Bombus spp.) and has relatively high outcrossing rates of 40–80%, even in small populations (van Treuren et al. 1993). Seed dispersal may be aided by the presence of glandular hairs on the calyx, which promote adherence to grazing animals (Weeda et al. 1988). On the basis of its distribution along rivers and its recent colonization of newly formed river dunes, water dispersal may also occur. The strong smell and taste of the plants seem to deter grazing animals (personal observations of the authors).

The species occurs mainly in central and southern Europe, and its north-western distribution limit passes through the Netherlands, although outlying populations are also found in Denmark and Sweden. Its distribution also extends into Great Britain, where S. pratensis is rare and protected (Rich 1999). In the Netherlands, S. pratensis is rare (Ouborg & van Treuren 1995), red-listed and legally protected (van der Meijden 1996). The Dutch Ministry of Agriculture, Nature Management and Fisheries lists it as a ‘target’ species of dry floodplain grasslands and dikes (Bal et al. 1995), which means that specific efforts should be made to maintain this species in its characteristic habitat.

habitat characteristics and management

Salvia pratensis is characteristic of semi-natural grassland communities. Phytosociologically, it is a diagnostic species of the Medicagini–Avenetum pubescentis association, belonging to the class Koelerio–Corynephoretea, the dry grasslands on nutrient-poor sandy soils (Schaminée, Stortelder & Weeda 1996). Although the plant is occasionally found in calcareous grasslands in the south of the country, its main habitat type in the Netherlands is formed by the dry floodplain grasslands situated mainly along the Rhine and the Meuse system. It has been estimated that 89% of those grasslands have been lost between 1968 and 1992 (van der Zee 1992). As a result, S. pratensis currently has a highly fragmented habitat, consisting generally of very small nature reserves of a few hectares one to several kilometres apart. In these reserves, the traditional management, such as mowing and haymaking or seasonal grazing, has usually been maintained since the 1950s. However, some areas are now maintained by low-intensity year-round grazing by Konik horses and/or selected ‘hardy’ cattle breeds.

classification of life stages

Based on field observations and data in Ouborg (1993), we distinguished a number of life stages (Fig. 1).


Figure 1. Schematic drawings of the five life stages distinguished in the populations of S. pratensis. S = seedling; J = juvenile; I = immature; V = vegetative adult; G = generative adult (with a small and a large size class). The height of the largest plant is c. 1 m, the seedling is c. 1–1·5 cm in diameter.

Download figure to PowerPoint

  • 1
    Seedlings (S) had cotyledons and mostly one leaf pair. Because seeds primarily germinate in autumn, we observed few seedlings during this study.
  • 2
    Juveniles (J) had one to two leaf pairs, and a rosette with a diameter of approximately 5 cm.
  • 3
    Immatures (I) were a transition between the juvenile and vegetative adult state. We classified individuals with three leaf pairs as immature. The rosettes of immatures were 7–10 cm in diameter.
  • 4
    Vegetative adults (V) had a large rosette of > 10 cm in diameter, with at least four leaf pairs. Occasionally, we observed plants with an extra side rosette. Leaves were considerably thicker than those of the younger states, and the leaves below were sometimes withered.
  • 5
    Generative adults (G) normally had one to five flowering stalks in addition to the rosettes, although we have observed plants with up to 50. Salvia pratensis produces no vegetative stems. In this life stage, we distinguished two size classes: small generative adults (Gs), with one to two flowering stalks, and large generative adults (Gl), with more than two flowering stalks. We assumed that the larger individuals were older.

determination of stage structure and vegetation analysis

We analysed 23 populations of S. pratensis along the larger Dutch rivers, primarily situated in nature reserves owned by the Dutch State Forestry Service and Society for the Preservation of Nature. Besides information from these organizations, we used records from the national floristic database (FLORBASE) managed by the non-governmental organization FLORON. Unfortunately, S. pratensis had disappeared from many of the localities that we visited. It became evident that the small number of populations and the large variation between sites would preclude an approach involving statistical comparison of a number of pre-selected sites with different management regimes. Hence, we used all populations in river forelands that we could find and get access to.

In each population, depending on size, area and heterogeneity, two or three plots (occasionally one in very small populations) of 1–4 m2 were analysed, yielding a total of 51 samples. In each plot, the number of individuals in each of the different life stages of S. pratensis was counted to estimate their relative proportions and mean densities. Additionally, the vegetation structure was described by recording the following parameters: percentage cover of litter, bryophytes, herb layer, the percentage of bare soil surface and the height of the herb layer. All vascular plants plus their percentage cover were recorded. Additionally, four small soil samples (total 10 g) were taken from the top 10 cm of the soil profile in each plot. The samples were placed together in a plastic bag, dried in an oven at 65 °C, mixed thoroughly and analysed at the soil chemical laboratory of the IBED Institute for the following variables: NO2, NO3, PO43−, Ca2+, K+ (all in µg g soil−1), electric conductivity (µS cm−1) and pH(H2O).


We obtained information on the management regime of each population through personal observation and interviews with site managers. Some areas were managed traditionally as hay meadows by means of mowing and haymaking once or twice each year. Others were seasonally grazed by domestic cattle, and another group was managed by year-round low-intensity grazing by Galloway cattle and/or Konik horses. None of the sites was artificially fertilized. We classified the management regimes observed as follows.

  • 1
    Early mowing, between 15 June and 15 July, usually early in this period, and some sites were cut a second time later in the season.
  • 2
    Late mowing, after 15 July.
  • 3
    Seasonal grazing from spring to autumn by domestic cattle or horses.
  • 4
    Year-round low-intensity grazing.

data analysis

Data consisting of individual population samples (one to four plots per population) were used to analyse relationships between the percentage of different life stages and vegetation parameters. In order to draw conclusions at the population level, we pooled individual samples (plots) per site and determined the life stage structure of the total population. For statistical analyses at the population level, we calculated the population means of the environmental parameters from the plot measurements.

The population structure data were used to analyse the extent of the existence of significantly different groups of populations, both at the sample and population level. We used a K-means clustering for this purpose (Hartigan 1975). This method clusters samples in a pre-selected number of groups by maximizing between- relative to within-group variation. A few subjective corrections on this clustering were made (see the Results), the consequences of which were tested with an anova.

The relationships between the relative proportion (%) of each life stage and the structure of the surrounding vegetation were investigated using multiple regression analysis (Sokal & Rohlf 1981). Residuals were checked for normality using the Kolmogorov–Smirnov one-sample test with the Lilliefors option (Wilkinson 1989).

The vegetation data (species cover, omitting the scores for S. pratensis) were subjected to ordination with a detrended correspondence analysis (DCA; Hill 1979) with the computer package canoco (ter Braak 1992). This analysis aims at finding ordination axes along which species’ scores show maximum dispersion. The dispersion is a measure of the amount of variance accounted for by the axes. We plotted the population types of S. pratensis on the DCA axes to investigate if there were differences in vegetation composition between the types. Such vegetation differences were tested with a Monte Carlo test after a canonical correspondence analysis (CCA; ter Braak 1992). Spearman rank correlation coefficients were calculated between the axis scores, the population size of S. pratensis, the ratio of seedlings plus juveniles to generative adults [hereafter abbreviated (S + J)/G ratio] and the soil parameters. Furthermore, a synoptic vegetation table was generated to identify specific species (or species groups) that were positively or negatively associated with the different population types.

Associations between management regime and the observed population types were tested with a χ2-test on a contingency table (Fowler & Cohen 1990). Furthermore, the differences in population structure variables between management types were tested with Kruskal–Wallis type anovas (abbreviated as K-W tests).


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Appendix

population structure and population types

Three groups of population types emerged from the K-means clustering (Tables 1 and 2). Pre-selection of two, four or five groups for the clustering led to small groups, with only a few samples, combined with very large groups. Only when we pre-selected three groups was the number of samples in each group large enough for a meaningful interpretation. A subjective correction was performed on a few samples (Wageningse waard, Randerwaard). Classification of these populations was difficult due to the small number of individuals. An analysis of the statistical effects of the corrections is also presented in Table 1. At the population level, there were originally no significant differences between the types with regard to the relative proportion of immatures. After the corrections, this changed to the adult vegetative stage.

Table 1.  Analysis of variance statistics of the K-means clustering of population stage structures (proportions of each life stage) into three groups at the sample level and the population level. The last two columns give the F-ratio and the P-value for the corrections that were made a posteriori (see text). For the cluster analysis, the small and large generative plants were pooled
Sample level
Seedlings0·00220·02448 2·42  0·099 6·68  0·003
Juveniles1·63120·4254892·09< 0·00149·92< 0·001
Immatures0·25720·64048 9·62< 0·001 9·77< 0·001
Adult vegetatives0·42520·6314816·17< 0·001 5·00  0·011
Generatives2·36620·7634874·43< 0·00133·34< 0·001
Population level
Seedlings0·00420·00520 7·10  0·005 7·11  0·005
Juveniles0·42120·2392017·59< 0·00117·32< 0·001
Immatures0·02520·23220 1·06  0·364 6·72  0·006
Vegetatives0·16120·23220 6·93  0·005 1·29  0·299
Generatives0·48020·2102022·80< 0·00118·84< 0·001
Table 2.  A summary of the sampled plots and the clustering of S. pratensis into three structure types. Corrected samples are indicated by showing the original cluster type
LocationPlotS+J/G ratio*Patch typeOverall typePopulation sizeManagement
  • *

    S+J/G ratio = ratio of seedlings + juveniles to generative adults.

  • D = dynamic; N = normal; R = regressive.

Millingerwaard1 0·86RD 100Year-round grazing
Neerijnen1 2·67DD 250Early mowing
 2 0·67D   
 3 0·29N   
Luistenbuul1 0·77ND 250Early mowing
 2 3·50D   
 3 0·57N   
Hengforderwaard1 1·31DD 300Seasonal grazing
 2 2·20D   
Koekkoekse waard1 0·88ND3500Late mowing
 2 1·57D   
 3 1·60D   
 4 0·00N   
Bijland1 0·47ND3800Late mowing
 2 1·50D   
 3 1·87D   
Wageningse waard1 0·00RN   3Early mowing
 2 1·00N (D)   
Rammelwaard1 0·33NN   6Late mowing
Poederooijen2 0·18NN  20Early mowing
Staartjeswaard1 0·13NN  30Seasonal grazing
Koornwaard1 0·50NN  35Seasonal grazing
 2 0·00N   
Wilperwaard1 0·20NN 100Seasonal grazing
 2 0·25N   
 3 0·00N   
Meijnerswijk1 0·00RN 150Year-round grazing
 2 0·25N   
 3 0·00R   
 4 0·00N   
Winsense waard1 0·17NN 300Early mowing
 2 0·07N   
Vreugderijkerwaard1 0·25NN 350Seasonal grazing
 2 0·00N   
Hurwenen1 0·00RN 350Early mowing
 2 0·00N   
 3 0·00N   
 4 0·39N   
Ravenswaard1 0·47NN 350Early mowing
 2 0·20N   
Cortenoever 21 0·25NN 700Seasonal grazing
 2 0·09N   
Olsterwaard1 0·00RR   2Early mowing
Randerwaard1 0·00R (N)R   6Early mowing
Loevenstein1 0·00RR  60Year-round grazing
 2 0·00R   
 3 0·00R   
Velperwaard1 0·00RR 100Early mowing
Cortenoever 11 0·00NR 700Early mowing
 2 0·00R   

The average stage structure of the three types is shown in Fig. 2a,b. The analysis on plot data hardly differed from the analysis on populations. Although the K-means clustering was performed on samples in which large and small generative adults were pooled, we show the proportions separately. The classification of the three population types could be interpreted as follows.


Figure 2. Stage structure (proportions of life stages) of the three population types of S. pratensis that emerged from the K-means clustering, based on (a) plots and (b) pooled population data (see text). (c) The median percentage bare soil and the median cover of bryophytes and litter in the vegetation as grouped per population type. Error bars in (a) and (b) represent standard deviations.

Download figure to PowerPoint

Populations of group 1 had a relatively high proportion of young life stages, primarily juveniles. Consequently, the proportion of adult plants was low and most of the generative plants were small. This type was characterized by a relatively high (S + J)/G ratio, with a mean of 3·4 and a range of 0·67–22·0. On average, the proportion of young stages was 0·60. This cluster is very similar to the ‘invasive’ type described by Oostermeijer, van’t Veer & den Nijs (1994) for Gentianapneumonanthe. Because there is no clear vegetation succession involved in the habitat of S. pratensis, we named this cluster ‘dynamic’ instead of ‘invasive’.

The second cluster comprised populations with lower proportions of seedlings and juveniles than group 1, and relatively higher proportions of immatures, vegetatives and small generatives. Large generative plants were also scarce in this type. The (S + J)/G ratio had a mean of 0·27 and ranged from 0 to 0·77. The average proportion of young life stages was 0·36. This type is similar to the ‘stable’ or ‘normal’ type described by Oostermeijer, van’t Veer & den Nijs (1994).

Populations of the third cluster consisted primarily of later life stages, such as adult vegetatives and particularly large generative plants. In total, two juveniles were found in populations of this type so, except for those two cases, the (S + J)/G ratio was 0. The mean proportion of young life stages was only 0·1. Following Oostermeijer, van’t Veer & den Nijs (1994), we named this type ‘regressive’.

The three population types had significantly different (S + J)/G ratios (K-W test, χ2 = 37·0, d.f. = 2, P = 0·001) and proportions of young life stages (K-W test, χ2 = 36·2, d.f. = 2, P = 0·001). Total plant density per population was also distinctly different. Medians ranged from 12·1 plants m−2 in the dynamic cluster, through 6·5 plants m−2 in the normal, to 2·3 plants m−2 in the regressive cluster (K-W test, χ2 = 13·3, d.f. = 2, P = 0·001). The same tendency was found for population size, where the medians were 300 individuals for both the dynamic and normal clusters, and 100 for the regressive type (K-W test, χ2 = 6·4, d.f. = 2, P = 0·05). None of the populations of the latter cluster consisted of more than 100 individuals (Table 2). Additionally, there was a significant positive correlation between population size and the proportion of young plants (rs = 0·292, n = 23, P = 0·05), suggesting that larger populations were in general more dynamic.

relationships between life stages and vegetation structure

We found no significant differences between the medians of vegetation structure parameters between the population types, although Fig. 1c suggests that the percentage of bare soil was quite different. Probably owing to large variation within the types, this difference was not statistically significant (K-W test, χ2 = 4·4, d.f. = 2, P = 0·112 for bare soil, P > 0·2 for the other parameters).

The multiple regression analysis with the vegetation structure parameters as independent variables and the proportions of the life stages as dependent variables yielded some significant correlations (Table 3). At both the plot and population levels, there was a significant negative correlation between bryophyte cover and the percentage of generatives. At the population level, a positive correlation also existed between the percentages of juveniles and generatives, the (S + J)/G ratio and the percentage of bare soil.

Table 3.  Statistics of the multiple regression models with the percentage of each life stage as dependent, and the vegetation structure parameters as independent variables. Residuals were tested for normality. Regressions were calculated both at the level of samples (plots) and on averaged samples per population. The left-hand side gives the standardized regression coefficients (β) and the t-test for their deviation from zero. The right-hand side gives the anova table for the multiple regression model. The results are shown only for the significant models
 βtPAnalysis of variance
Sample level
Percentage generatives (R2 = 0·086), not transformed
Constant 0·000 9·4460·000Regression 1131·8994·6080·037
Bryophytes−0·293−2·1470·037Residual49 28·627  
Population level
S + J/G ratio (R2 = 0·172), log-transformed
Constant 0·000 0·0520·959Regression 1  0·1074·3600·049
Bare soil 0·415 2·0880·049Residual21  0·025  
Percentage seedlings (R2 = 0·225), log-transformed
Constant 0·000−1·7280·099Regression 2  0·0002·8950·079
Bare soil 0·306 2·3620·028Residual20  0·000  
Herb cover 0·306 1·7150·102     
Percentage juveniles (R2 = 0·179), log-transformed
Constant 0·000 0·3160·755Regression 1  0·0144·5740·044
Bare soil 0·423 2·1390·044Residual21  0·003  
Percentage generatives (R2 = 0·434), not transformed
Constant 0·000 4·9170·000Regression 2622·2587·6660·003
Bryophytes−0·414−2·3840·027Residual20 81·170  
Bare soil 0·626 3·6040·002     

There were no differences in any of the soil parameters between the population types (K-W tests, P = 0·897); neither did these variables show any significant correlation with other parameters, such as population size or (S + J)/G ratio.

population structure in relation to vegetation composition

The DCA ordination explained a significant proportion of the variation in the species composition of the vegetation (total inertia = 6·47; axis 1, gradient length = 4, eigenvalue = 0·54, percentage variance explained = 8·4; axis 2, 3·4, 0·39, 6·0). We plotted the samples, differentiated per population type, on DCA axes 1 and 2 (Fig. 3). There were significant differences in vegetation composition between the three population types (Monte Carlo test; P = 0·01). However, these differences could not be explained by any significant correlation coefficients of the environmental variables with the DCA axes. Only litter showed a significantly positive correlation with DCA axis 1 (rs = 0·321, n = 51, P = 0·05). The (S + J)/G ratio and population size of S. pratensis were significantly negatively correlated with DCA axis 1 (rs = −0·475 and −0·478, respectively, n = 51, P = 0·001). These relationships might be explained by other factors, such as differences in management.


Figure 3. Ordination diagram (DCA axes 1 and 2) showing the distribution of the individual samples (plots) labelled according to the population type of S. pratensis and the position of a selection of the most informative co-occurring plant species. Legend for species names (following van der Meijden 1996): Rhinmino, Rhinanthus minor; Thalminu, Thalictrum minus; Centscab, Centaurea scabiosa; Knauarve, Knautia arvensis; Veroaust, Veronica austriaca ssp. teucrium; Orobcary, Orobanche caryophyllacea; Helipube, Helictotrichon pubescens; Galiveru, Galium verum ssp. verum; Rumethyr, Rumex thyrsiflorus; Euphcypa, Euphorbia cyparissias; Eryncamp, Eryngium campestre; Medifalc, Medicago falcata; Geradiss, Geranium dissectum; Urtidioi, Urtica dioica; Festrubr, Festuca rubra; Agrostol, Agrostis stolonifera; Hyporadi, Hypochaeris radicata; Arreelat, Arrenatherum elatius; Rhinangu, Rhinanthus angustifolius; Sangmino, Sanguisorba minor; Achimill, Achillea millefolium; Crepbien, Crepis biennis; Persmacu, Persicaria maculosa; Equiarve, Equisetum arvense; Cynocris, Cynosurus cristatus; Holclana, Holcus lanatus; Artevulg, Artemisia vulgaris; Phleprat, Phleum pratense; Planmajo, Plantago major; Elyrepe, Elytrigia repens; Ephesul, Euphorbia esula; Cirsarve, Cirsium arvense; Antsylv, Anthriscus sylvestris; Hypeperf, Hypericum perforatum; Poteanse, Potentilla anserina.

Download figure to PowerPoint

At the species level (nomenclature follows van der Meijden 1996), one can qualitatively investigate which species occur with higher presence and mean abundance in the different population types. By investigating the synoptic vegetation table (Appendix 1), and the positions of species in the ordination diagram (Fig. 3), associations between the population types and the species composition of the surrounding vegetation could be evaluated. Veronica austriaca ssp. teucrium, Thalictrum minus and Orobanche caryophyllacea, typical species of the dry floodplain grasslands and also rare in the Netherlands, occurred only at sites where S. pratensis had a dynamic population structure. Other species, found on the left-hand side of the ordination diagram, only occurred in samples with dynamic and/or normal S. pratensis populations, such as Centaurea scabiosa, Euphorbia cyparissias and Rhinanthus angustifolius. Rumex thyrsiflorus, Medicago falcata and Eryngium campestre clearly had their highest frequency in the dynamic type, intermediate in the normal and lowest in the regressive type. In the regressive type there were several species of nutrient-rich conditions, such as Holcus lanatus, Poa annua and Crepis biennis, which were not found in the dynamic Salvia populations. In normal and regressive populations we typically found species of nutrient-rich conditions, such as Cirsium arvense, Urtica dioica, Medicago lupulina, Leucanthemum vulgare and Cynodon dactylon.

There was a marginally significant difference in the number of species (per plot or population) between the population types observed in S. pratensis (K-W test, χ2 = 4·99, d.f. = 2, P = 0·083). The number of species was largest (c. 24) in dynamic and normal, and lowest (c. 19) in regressive populations.

effects of management

The different management methods (mowing or grazing) had a significant effect on the vegetation. This was investigated using the same method as for population types in relation to the vegetation (Monte Carlo test, P = 0·001). After removing the influence of management in canoco, there was still a significant association between population type and vegetation composition (Monte Carlo test, P = 0·014). In other words, vegetation differed in composition among the population types, and this was not merely a result of differences in management.

There was a significant association between population type (at the population level) and the four different management regimes (χ2 = 13·6, d.f. = 6, P = 0·05). First, populations that were mown late had significantly more dynamic populations than expected (individual χ2 = 6·8). Secondly, sites with year-round grazing had more regressive populations than expected (individual χ2 = 4·5), although there were actually too few sites with this management type to draw useful conclusions. No differences in the frequencies of specific population types were found between mowing or grazing.

There were significant differences between the proportions of several life stages between the management types (Fig. 4). Seedlings were mainly found in populations in late-mown grasslands (K-W test, χ2 = 18·3, d.f. = 3, P = 0·001). Of the seven plots in which seedlings were observed, six were in mown areas and only one in a grazed site. Four of those six plots with seedlings were situated in sites that were mown late. However, there was no difference in the proportion of juvenile plants between management types (P > 0·20). Immature and adult vegetative plants were relatively more frequent in the seasonally grazed sites and less frequent on sites that were grazed year-round (K-W test, immatures χ2 = 8·6, d.f. = 3, P = 0·035; vegetative adults χ2 = 12·7, d.f. = 3, P = 0·005). The proportion of large flowering plants was highest in the populations that were grazed year-round or mown early and very low under the two other types of management (K-W test, χ2 = 19·6, d.f. = 3, P < 0·001). The late mowing regime was associated with a significantly larger population size (K-W test, χ2 = 14·6, d.f. = 3, P = 0·002). This was partly caused by the fact that the two largest populations, Bijland (3800 individuals) and Koekoeksche waard (3500), were both mown late.


Figure 4. Population stage structure of S. pratensis in relation to the four different types of habitat management. Error bars represent standard deviations.

Download figure to PowerPoint


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Appendix

The observed division of S. pratensis populations into three types was based on the relative proportions of the different life stages. However, besides the stage structure, there were certain other features that distinguished the population types. These included differences in population size, total density of S. pratensis individuals, and vegetation structure (mainly the percentage of bare soil cover) and composition (species characteristic of nutrient-poor vs. nutrient-rich soils). The type of habitat management also seemed to have an effect on the population structure of S. pratensis.

The data support our hypothesis that it was possible to distinguish different types of population structure for S. pratensis. Comparing our findings with the literature, we named the three population clusters dynamic (invasive; sensuRabotnov 1985; Oostermeijer, van’t Veer & den Nijs 1994), normal and regressive (Rabotnov 1985; Oostermeijer, van’t Veer & den Nijs 1994). With long-term demographic studies and matrix projection models, Oostermeijer et al. (1996) could relate the three population types found in the wet heathland species Gentiana pneumonanthe to the actual population growth rates (measured as the finite rate of increase, λ). Dynamic (invasive) populations increased (λ > 1), normal populations were stable (λ≈ 1) and regressive populations declined (λ < 1). It is possible that the three population types of S. pratensis show similar growth rates, although this remains to be confirmed.

In contrast to the predominance of regressive populations in Gentiana pneumonanthe (Oostermeijer, van’t Veer & den Nijs 1994), most of the Salvia populations were either of the dynamic or the normal type. This suggests that the management of the remaining sites was generally favourable. However, the small populations under an early mowing regime currently depend on the survival of a few large, and probably rather old, flowering plants, and therefore risk local extinction if the current conditions do not change. In those reserves, it is necessary to either change the mowing time to a later date, after seed shedding of Salvia and most other species has occurred, or change to seasonal grazing. Nevertheless, in such small populations, the response to improved management in the form of seedling recruitment could be impaired by inbreeding depression in combination with low seed production resulting from the Allee effect (Oostermeijer 2000).

Bare soil was the most important vegetation structure parameter, suggesting that the presence of open ground is important for the performance of the population. The proportions of seedlings and juveniles and the S + J/G ratio were positively correlated with the percentage of bare soil. Open patches in the vegetation are thus apparently important as ‘safe sites’ for germination and seedling survival. This is consistent with field observations on recruitment of S. pratensis in British populations (Rich 1999) and with studies on the recruitment of various other plant species in regularly disturbed areas (Holderegger 1996) and grasslands (Johnson & Thomas 1978; Fowler 1988; Rusch 1992; Krenová & Leps 1996). According to Spackova, Kotorova & Leps (1998), removal of the vegetation and/or bryophyte layer is important for the seedling recruitment of many wet meadow species. In general, it has been suggested (Tilman 1993) that lack of seedling recruitment may be one of the major causes of declines in species diversity in grasslands.

It is interesting that the percentage of generative individuals showed the same positive correlation with bare soil surface. Given that there are also relatively high percentages of young plants, this indicates that a higher proportion of adults is flowering in open patches. Hence, the vegetation structure seems to affect the most important phases in the demography of this species: germination, seedling establishment and flowering.

We hypothesized that viable populations would be found in vegetation with a higher conservation value, which would be reflected in a higher species diversity or the occurrence of other rare species. To some degree, the species composition of the surrounding vegetation was associated with the viability of S. pratensis populations. Consistent with our hypothesis, species diversity was higher in dynamic and normal populations, although the differences were small and only marginally significant. However, other (rare) species of the Medicagini–Avenetum pubescentis, the characteristic plant community of dry river grasslands (Schaminée et al. 1996), were indeed restricted to sites with dynamic Salvia populations, whereas species typical of nutrient-rich conditions were associated with regressive populations. Under the latter conditions, population size may still be rather large, and the remaining flowering Salvia individuals are often tall and bear many flowering spikes and hence appear quite vital. Both types of information clearly give a false impression of population viability. By assessing the population stage structure, more useful information can be obtained in a simple way.

Vegetation management is one of the most important and interesting aspects of applied vegetation science. In this study, the most striking results were the differences in structure between populations of S. pratensis with a late mowing regime and the early mown or the grazed populations. Late mowing was performed in three populations, which together had more plots with a dynamic structure than were expected by chance. Two of those populations were the largest in the Netherlands and the majority of the few seedlings observed was found here. In contrast, many of the regressive populations were mown early, although there were several early mown populations with a normal or dynamic structure. This result suggests that late mowing is and has been the most favourable management for S. pratensis. Other authors also propose late mowing as a method for recovering typical dry floodplain grassland vegetations from nutrient-rich situations (van Eck, van Zuijen & Sykora 1997). However, early mowing of river dykes (in June) has been recommended more recently (Liebrand 1999).

The difference in recruitment between the late and early mown populations is due to the timing of seed shedding. In the early mown populations, seeds are generally not ripe at the time of mowing. The seeds produced after regrowth later in the season might be less viable, for instance because of reduced pollinator visitation, higher selfing rates and subsequent inbreeding depression (van Treuren et al. 1993; Ouborg & van Treuren 1994). One population where mowing was performed relatively early, but after the seed was shed, was classified as dynamic. A less rigid mowing schedule, which takes the temporal and spatial variation among populations into account, seems the best way to maintain or promote the remaining populations of S. pratensis. Another argument in favour of late mowing is that it opens up the vegetation canopy in the autumn period (Olff et al. 1994), which is when S. pratensis germinates.

Many ecological restoration projects are in progress in the Dutch floodplains. This will eventually result in new habitats and will enlarge the remaining habitats for many characteristic plants and animals. The dominant management regime in these areas, year-round grazing by free-ranging animals, will definitely require larger areas to avoid too high a grazing pressure on the remaining rare plant species. Disturbance by year-round grazing is more unpredictable and dynamic than the traditional mowing and haymaking (Bokdam & Gleichman 2000). On the basis of the few populations included in this study, it is difficult to draw conclusions concerning the effects of year-round grazing on the population structure of S. pratensis, the more so because this type of management is relatively recent and populations have only recently established. Three populations were situated in ecological restoration areas with year-round grazing, and of the nine plots analysed in those sites, one was dynamic, two were normal and six were regressive. At the same time, all populations were relatively small. Hence, contrary to our expectations, very few of the colonization events seem to have yielded dynamic (rapidly) expanding populations. Although the small population sizes may merely be a consequence of rather recent colonization, it could also be that the regressive structure of these small founder populations is caused by problems with reproduction (seed quantity) and inbreeding depression (seed quality) as a result of the Allee effect (Ouborg & van Treuren 1994, 1995; Groom 1998; Oostermeijer 2000). Additional studies are needed to provide evidence for this hypothesis.

Bakker (1989) doubts whether species-rich grasslands can be restored with the management regimes associated with ecological restoration of intensively used agricultural areas. This management generally involves removal of the nutrient-rich topsoil of pastures or fields, followed by year-round grazing by hardy cattle breeds or horses. Nowadays, most populations of S. pratensis occur in the few semi-natural grasslands that are still under traditional management, such as mowing and haymaking or seasonal grazing by domestic cattle. Our findings suggest that year-round grazing of ecologically restored areas has not yielded populations of sufficient vitality to compensate for any local extinctions that would be caused by unfavourable management of remaining sites. Hence, maintaining the quality of the remaining reserves will be an important management task until any restoration projects have recreated sufficiently large new habitat patches in the immediate locality.


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Appendix

We thank all reserve managers for their co-operation and assistance, and Ella de Hullu of the State Forestry Service (Staatsbosbeheer) and Bart van Tooren and Roel Douwes of the Society for the Preservation of Nature (Natuurmonumenten) for research permits, information on research locations and useful ideas and suggestions. The floristic research organization FLORON kindly provided population coordinates from their databases. Thanks are due to Patrick Meirmans for statistical advice and useful comments on the manuscript, and to Eva Johanna for additional help.


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Appendix
  • Bakker, J.P. (1989) Nature Management by Grazing or Cutting. Kluwer Academic Publishing, Dordrecht, the Netherlands.
  • Bal, D., Beije, H.M., Hoogeveen, Y.R., Jansen, S.R.J. & Van Der Reest, P.J. (1995) Handboek natuurdoeltypen in Nederland. IKC-Natuurbeheer, Wageningen, the Netherlands.
  • Bokdam, J. & Gleichman, J.M. (2000) Effects of grazing by free-ranging cattle on vegetation dynamics in a continental north-west European heathland. Journal of Applied Ecology, 37, 415431.
  • Ter Braak, C.J.F. (1992) canoco: A FORTRAN Program for Canonical Community Ordination. Microcomputer Power, Ithaca, NY.
  • Bühler, C. & Schmid, B. (2001) The influence of management regime and altitude on the population structure of Succisa pratensis: implications for vegetation monitoring. Journal of Applied Ecology, 38, 689698.
  • Dietz, H., Fischer, M. & Schmid, B. (1999) Demographic and genetic invasion history of a 9-year-old roadside population of Bunias orientalis L. (Brassicaceae). Oecologia, 120, 225234.
  • Van Eck, J.M.C., Van Zuijen, M.P. & Sykora, K.V. (1997) De effecten van beheer op de vegetatie van de Bijlanddijk en het Helicopterveldje. Stratiotes, 14, 318.
  • Fowler, N.C. (1988) What is a safe site? Neighbour, litter, germination date and patch effects. Ecology, 69, 947961.
  • Fowler, J. & Cohen, L. (1990) Practical Statistics for Field Biology. Wiley and Sons Inc., New York, NY.
  • Gatsuk, L.E., Smirnova, O.V., Vorontzova, L.I., Zaugolnova, L.B. & Zhukova, L.A. (1980) Age states of plants of various growth forms – a review. Journal of Ecology, 68, 675696.
  • Groom, M.J. (1998) Allee effects limit population viability of an annual plant. American Naturalist, 151, 487496.
  • Gusewell, S., Buttler, A. & Klotzli, F. (1998) Short-term and long-term effects of mowing on the vegetation of two calcareous fens. Journal of Vegetation Science, 9, 861872.
  • Hartigan, J.A. (1975) Clustering Algorithms. Wiley and Sons Inc., New York.
  • Hill, M.O. (1979) decorana– A FORTRAN Program for Detrended Correspondence Analysis and Reciprocal Averaging. Cornell University, New York, NY.
  • Holderegger, R. (1996) Reproduction of the rare monocarpic species Saxifraga mutata L. Botanical Journal of the Linnean Society, 122, 301313.
  • Hutchings, M.J., Mendoza, A. & Havers, W. (1998) Demographic properties of an outlier population of Orchis militaris L. (Orchidaceae) in England. Botanical Journal of the Linnean Society, 126, 95107.
  • Johnson, C.D. & Thomas, A.G. (1978) Recruitment and survival of seedlings of a perennial Hieracium species in a patchy environment. Canadian Journal of Botany, 50, 572580.
  • Jones, P.S. (1998) Aspects of the population biology of Liparis loeselii (L.) Rich. var. ovata Ridd. ex Godfery (Orchidaceae) in the dune slacks of South Wales, UK. Botanical Journal of the Linnean Society, 126, 123139.
  • Krenoá, Z. & Leps, J. (1996) Regeneration of a Gentiana pneumonanthe population in an oligotrophic wet meadow. Journal of Vegetation Science, 7, 107112.
  • Liebrand, C.I.J.M. (1999) Restoration of species-rich grasslands on reconstructed river dikes. PhD Thesis. Agricultural University, Wageningen, the Netherlands.
  • Van Der Meijden, R. (1996) Heukels’ Flora van Nederland. Wolters-Noordhoff, Groningen, the Netherlands.
  • Menges, E.S. & Gordon, D.R. (1996) Three levels of monitoring intensity for rare plant species. Natural Areas Journal, 16, 227237.
  • Morgan, J.W. (1997) The effect of grassland gap size on establishment, growth and flowering of the endangered Rutidosis leptorrhynchoides (Asteraceae). Journal of Applied Ecology, 34, 566576.
  • Muller, F.M. (1978) Seedlings of the North-Western European Lowland. Centre for Agricultural Publishing and Documentation, Wageningen, Netherlands.
  • Olff, H., Pegtel, D.M., Van Groenendael, J.M. & Bakker, J.P. (1994) Germination strategies during grassland succession. Journal of Ecology, 82, 6977.
  • Oostermeijer, J.G.B. (2000) Population viability analysis of the rare Gentiana pneumonanthe: importance of genetics, demography, and reproductive biology. Genetics, Demography and Viability of Fragmented Populations (eds A.G.Young & G.M.Clarke), Vol. 4, pp. 313334. Cambridge University Press, Cambridge, UK.
  • Oostermeijer, J.G.B., Brugman, M.L., De Boer, E.R. & Den Nijs, J.C.M. (1996) Temporal and spatial variation in the demography of Gentiana pneumonanthe, a rare perennial herb. Journal of Ecology, 84, 153166.
  • Oostermeijer, J.G.B., Van’t Veer, R. & Den Nijs, J.C.M. (1994) Population structure of the rare long-lived perennial Gentiana pneumonanthe in relation to vegetation and management in the Netherlands. Journal of Applied Ecology, 31, 428438.
  • Ouborg, N.J. (1993) On the relative contribution of genetic erosion to the chance of population extinction. PhD Thesis. University of Utrecht, Utrecht, the Netherlands.
  • Ouborg, N.J. & Van Treuren, R. (1994) The significance of genetic erosion in the process of extinction. 4. Inbreeding load and heterosis in relation to population size in the mint Salvia pratensis. Evolution, 48, 9951008.
  • Ouborg, N.J. & Van Treuren, R. (1995) Variation in fitness-related characters among small and large populations of Salvia pratensis. Journal of Ecology, 83, 369380.
  • Rabotnov, T.A. (1985) Dynamics of plant coenotic populations. Handbook of Vegetation Science. Vol. 3. The Population Structure of Vegetation (ed. J.White), pp. 121143. Junk Publishers, Dordrecht, the Netherlands.
  • Rich, T.C.G. (1999) Conservation of Britain’s biodiversity: Salvia pratensis. Watsonia, 22, 405412.
  • Rusch, G. (1992) Spatial pattern of seedling recruitment at two different scales in a limestone grassland. Oikos, 65, 433442.
  • Schaminée, J.H.J., Stortelder, A.H.F. & Weeda, E.J. (1996) De Vegetatie Van Nederland, 3. Opulus Press, Uppsala, Sweden.
  • Schmidt, K. & Jensen, K. (2000) Genetic structure and AFLP variation of remnant populations in the rare plant Pedicularis palustris (Scrophulariaceae) and its relation to population size and reproductive components. American Journal of Botany, 87, 678689.
  • Sokal, R.R. & Rohlf, F.J. (1981) Biometry: The Principles and Practice of Statistics in Biological Research, 2nd edn. W.H. Freeman, San Francisco, CA.
  • Spackova, I., Kotorova, I. & Leps, J. (1998) Sensitivity of seedling recruitment to moss, litter and dominant removal in an oligotrophic wet meadow. Folia Geobotanica, 33, 1730.
  • Thompson, K., Bakker, J.P. & Bekker, R.M. (1997) Soil Seed Banks of North-West Europe: Methodology, Density and Longevity. Cambridge University Press, Cambridge, UK.
  • Tilman, D. (1993) Species richness of experimental productivity gradients: how important is colonization limitation? Ecology, 74, 21792191.
  • Van Treuren, R., Bijlsma, R., Ouborg, N.J. & Van Delden, W. (1993) The effects of population size and plant-density on outcrossing rates in locally endangered Salvia pratensis. Evolution, 47, 10941104.
  • Weeda, E.J., Westra, R., Westra, Ch. & Westra, T. (1988) Nederlandse Ecologische Flora, wilde planten en hun relaties, Vol. 3. IVN/VARA/VEWIN, Amsterdam/Hilversum/Rijswijk, the Netherlands.
  • Wilkinson, L. (1989) systat: The System for Statistics. Evanston, IL.
  • Van Der Zee, F. (1992) Botanische Samenstelling, Ecologie En Erosiebestendigheid Van Rivierdijkvegetaties. Wageningen Agricultural University, Wageningen, the Netherlands.


  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Appendix
Table Appendix.  Synoptic vegetation table, showing presence (percentage of relevés) and mean abundance (Dafor scale) per species in each Salvia population type. Dafor classes: I, < 1%; II, 1–2%; III, 3–5%; IV, 6–10%; V, > 10% mean cover
SpeciesDynamic n = 10Normal n = 29Regressive n = 12SpeciesDynamic n = 10Normal n = 29Regressive n = 12
Veronica austriaca ssp. teucrium40 III    Festuca rubra100 V72 IV100 IV
Thalictrum minus20V    Rumex thyrsiflorus100IV41IV 50IV
Silene latifolia ssp. alba20I    Arrhenatherum elatius100IV52V 92V
Draba muralis20I    Dactylis glomerata 90II52III100III
Myosotis ramosissima20I    Taraxacum officinale s.l. 80III41III100III
Orobanche caryophyllacea20I    Medicago falcata 70IV34III 33IV
Ranunculus spp.10II    Trisetum flavescens 70III45IV 83IV
Sedum sexangulare10I    Eryngium campestre 70III28III 17V
Lepidium draba10I    Helictotrichon pubescens 60V31III 67III
Erodium cicutarium10I    Plantago lanceolata 60III59IV 67IV
Geranium dissectum10I    Poa pratensis 50II34III 58IV
Linaria vulgaris10I    Cerastium arvense 50II17II 25II
Anisantha sterilis40I10 II  Trifolium dubium 50IV21IV 42IV
Centaurea scabiosa20IV 7V  Trifolium pratense 50III28IV 58IV
Rhinanthus angustifolius20IV 3IV  Anthoxanthum odouratum 40IV38IV 42III
Ononis repens ssp. spinosa20IV 3III  Ranunculus bulbosus 40III21III 33II
Arenaria serpyllifolia20II10II  Ranunculus acris 40III48III 50III
Euphorbia cyparissias20II 3V  Lolium perenne 40III52IV100III
Ornithogalum umbellatum20I14II  Elytrigia repens 40III28IV 83V
Valerianella locusta20I 3II  Senecio jacobea 40II45III 75III
Erophila verna10I10I  Hypochaeris radicata 40II 7I 17III
Arabidopsis thaliana10I 3I  Bromus hordeaceus 40II14II 25III
Rumex crispus  14I  Cerastium fontanum 40II55II 83II
Agrimonia eupatoria   3III  Tragopogon pratensis 40II21III  8I
Vicia cracca   3III  Veronica arvensis 40I38I 42I
Carex arenaria   3II  Achillea millefolium 40I45II100II
Origanum vulgare   3II  Galium mollugo 30IV17III 17I
Primula veris   3II  Galium verum 30IV31III 25IV
Melilotus spp.   3I  Trifolium repens 30III31III 58IV
Stellaria media   3I  Poa trivialis 30III34IV100III
Medicago lupulina  31IV67 IIIRhinanthus minor 30III 3V  8I
Cirsium arvense  28II67IIIEquisetum arvense 30II 7II 33IV
Rumex acetosa  14III25ICentaurea jacea 20IV24III 17II
Cynodon dactylon  14III17IIIKnautia arvensis 20III10II  8II
Leucanthemum vulgare  14III 8IIBellis perennis 20III17I  8II
Trifolium campestre  14III 8IVAllium vineale 20II10I 50I
Alopecurus pratensis   7V17IIIPhleum pratense 20II 7III 25III
Agrostis capillaris   7IV17IIPlantago media 20II10II 17IV
Carum carvi   3V 8IVLotus corniculatus 20II10II 25V
Persicaria maculosa   3III 8IIVeronica chamaedrys 20II 7III  8III
Campanula rotundifolia   3II17IGlechoma hederacea 20II 7III 58II
Lathyrus pratensis   3II 8IVPotentilla reptans 20I 7II 33II
Lysimachia nummularia   3II 8IIITanacetum vulgare 20I 3II  8II
Leontodon hispidus   3II 8IICarex hirta 10III 3I  8II
Tussilago farfara   3I17IEuphorbia esula 10II 3II  8II
Tripleurospermum maritimum   3I 8IIPeucedanum carvifolia 10II10II  8II
Hieracium pilosella 10II 7II  8IIIDanthonia decumbens20III   8I
Convolvulus arvensis 10II17III  8IHypericum perforatum20II   8I
Cerastium glomeratum   3I 8IILuzula campestris 10II14I 17I
Holcus lanatus    50IVSanguisorba minor 10II 7IV  8IV
Rubus fruticosus s.l.    25VVicia sativa 10I 3I 17I
Crepis biennis    25IVGeranium molle 10I10II 25I
Artemisia vulgaris    17IBriza media 10I 3II  8II
Poa annua    17IPimpinella saxifraga 10I10II  8I
Cynosurus cristatus     8IIIHeracleum sphondylium 10I10III 25III
Urtica dioica     8IIIRanunculus repens 10I 3II  8I
Rumex obtusifolius     8III       
Agrostis spp.     8III       
Quercus robur     8II       
Agrostis stolonifera     8II       
Anthriscus sylvestris     8I       
Potentilla anserina     8I       
Populus nigra     8I       
Rumex acetosella     8I       
Plantago major     8I       
Capsella bursa-pastoris     8I       
Daucus carota     8I       
Solanum nigrum     8I       
Calamagrostis epigejos33II   8I       
Chaerophyllum temulum33I   8II