Blomqvist, M.M.
Citation
Blomqvist, M. M. (2005, February 3). Restoration of plant species diversity of ditch banks :
ecological constraints and opportunities. Retrieved from https://hdl.handle.net/1887/592
Version:
Not Applicable (or Unknown)
License:
Licence agreement concerning inclusion of doctoral thesis in the
Institutional Repository of the University of Leiden
Downloaded from:
https://hdl.handle.net/1887/592
Chapter 2
Declining plant species richness of grassland ditch banks –
a problem of colonisation or extinction?
M. M. Blomqvist, P. Vos, P. G. L. Klinkhamer & W. J. ter Keurs
Abstract
Small-scale landscape elements, such as ditch banks are an important remaining source of biodiversity in many agricultural landscapes, including the Western Peat District in the Netherlands. Unfortunately, plant species richness is declining even in these habitats. To understand the factors threatening biodiversity, we studied demographic traits (occupancy, trend, colonisation and extinction) for a large number of plant species, in a 25 year long data set. We developed a method to investigate the relative importance of colonisation and extinction for species increase and decrease in multi-species assemblages. We show that colonisation has been more important for determining species trends than extinction. Decreasing species were small and characterised by low nutrient tolerance and high light requirements, indicating that competitive ability influences species trends. The mechanism by which high nutrient levels reduce plant diversity appears to be closely related to colonisation (germination and seedling establishment). Local management should therefore continue to focus on nutrient reduction and the creation of regeneration sites. Yet, these measures will be insufficient for restoring species richness since isolation also hampers species increase. Therefore, to maximise the effects of local management, additional regional management solutions are required to improve seed dispersal, for example, from nature reserves.
Introduction
Restoration of soil conditions has often proved less efficient at restoring plant species richness than once hoped for (Olff and Bakker 1991; Olff et al. 1994a; Pfadenhauer & Klötzli 1996; Bakker & Berendse 1999; Berendse et al. 1999). A growing body of evidence suggests that this problem is often related to the lack of a seed bank and insufficient dispersal (Marshall and Hopkins 1990; Strykstra et al. 1998; Prins et al. 1998; Bakker & Berendse 1999). As can be expected from metapopulation and island biogeography theories (MacArthur & Wilson 1967; Hanski 1998), both extinction and colonisation can be important. Still, only few empirical studies have looked at the relative importance of colonisation and extinction for plant species increase and decline in multi-species assemblages (but see Ouborg 1993; Husband & Barrett 1996). From a conservation and restoration perspective such information would be valuable, as it could help to focus restoration efforts on the processes and ecological factors, which are most critical for a large number of species.
Much of the biodiversity of semi-natural vegetation has historically been created by low-intensity farming practices (Bignal and McCracken 1996). The intensification of agricultural practices during the last century has led to a rapid decline in species richness (Erlich 1988; World Resource Institute 1994; McNeely et al. 1995). Often, much of the former biodiversity in the agricultural landscape is maintained in small-scale landscape elements such as ditch banks, field margins, hedgerows and road verges (Geertsema 2002).
The Western Peat District in the Netherlands belongs to the most intensively exploited regions in Europe, with dairy farming as the most important activity since the fifteenth century (van der Linden 1982; van der Molen 1982). An extensive network of ditches and ditch banks cuts through the landscape, covering a total length 300 000 - 400 000 km (Higler 1994). Similar landscapes can also be found, for example, in the U.K. (Kruk 1991).
reports on the development of the species richness of these ditch banks indicate that even in this habitat, floristic richness is decreasing (Provincie Zuid-Holland 1998).
Previous research on these ditch bank systems has shown that they are greatly influenced by local management (Melman 1991). Simultaneous comparison of the vegetation composition at different sites, revealed that low levels of fertilisation, low ditch cleaning and mud dressing frequencies, and extensive grazing and mowing regimes were positively associated with high species richness (van Strien 1991). These new management recommendations have been locally applied during the past decade. Yet the species richness does not seem to increase (M. Kruk, personal communication). The question is, whether this is related to problems with colonisation (isolation, lack of propagules, poor germination), or to factors causing extinction (remaining nutrients, competition)?
Here, we use long-term data from ditch banks on dairy farms in the Western Peat District in the Netherlands to quantify the changes in plant species richness that have taken place during the last 25 years. Our aim is to determine which (ecological) factors and processes are related to the increase and decrease of different plant species, particularly the relative importance of colonisation and extinction for species trends. We further investigate the relationship between these demographic parameters and plant ecological characteristics. This way, attributes of threatened and dominant species groups can be identified, ultimately allowing for more specific restoration measures.
Methods
The vegetation database and selected data
We used data from the vegetation database of the Province of South-Holland (the ‘Information System for Vegetation’ (ISV) database) in the Netherlands, containing vegetation data collected by trained vegetation analysts from 1976 onwards. The major part of the data in the database comes from the agricultural landscape (Provincie Zuid-Holland 1993). The most important sampled habitats include grasslands, ditch banks and ditches. In this study, we focus on ditch banks in the agricultural landscape.
Ditch bank vegetation data were collected during a period of 25 years between 1976 and 1999 in six sampling episodes or periods (Table 1). The number of sampled ditch bank plots varied within each sampling period, with a total of 6004 plots. Presence of each species was always recorded in 50 m long relevees, the width of which varies with the width of the ditch bank. Basic data used were in a binary form: presence/absence per species per plot and per sampling episode.
A total of 401 taxa (recorded at genus, species or subspecies level) were recorded between 1976 and 1999 (nomenclature following van der Meijden (1996)). We focus on terrestrial herbaceous plants (dicots and monocots, 308 species, Appendix 1). For statistical reasons, the rarest species (average occupancy < 0.005 or 0.5%, n = 151) were excluded from further analyses. For the remaining 157 species, we calculated four demographic parameters: occupancy, trend in occupancy, colonisation and extinction values (based on measured appearance and disappearance values respectively). Next, we investigated how these parameters are related to each other. Finally, we examined the relationship between ecological plant characteristics and the demographic parameters. Species demography: estimating occupancy and trends
We calculated the mean occupancy by taking the sum of all occupied plots, expressed as a fraction of all sampled plots. Since the mean of the sampling year of all plots was close to the median year (1989.6 and 1988 respectively), this is a good estimate also for species showing a strong trend during the total sampling period. The mean occupancy values were logit transformed
before analyses. Trend values for each species were calculated using the data from the six sampling periods. Because sampling periods covered more than one year, an ‘middle’ year was taken for each period (Table 1). For each species, trends in occupancy were estimated through a logistic regression of numbers of occupied and unoccupied plots over time (i.e. the ‘middle’ year), using the statistical package R, version 1.2.2 (Hornik 2001). The significance of the slope was calculated with a quasi-binomial model, allowing possible over-dispersion to be modelled, thus resulting in a
Table 1. Timing and number of plots in the six sampling periods. Years refer to the time span covered per sampling period. Data is in presence/absence form, as totals of occupied and unoccupied plots per species per period. The weighted mean sampling year is indicated in the last row.
Occupancy and trend
more realistic estimate of the significance than the binomial model (McCullagh & Nelder 1989).
Species demography: estimating colonisation and extinction values
To estimate colonisation and extinction values, those plots were selected which had been sampled in any two consecutive periods. This resulted in five sampling events with ‘paired’ plots (Table 2) and a total of 1963 sampled pairs of plots. Based on presence/absence data, four transitions are possible: remaining absent (0-0), remaining present (1-1), appearing (0-1) and disappearing (1-0). We express the related numbers of plots with n0-0, n1-1, n0-1 and n1-0 , respectively. Occupancy p at t=0 can be expressed as
(n1-1 + n1-0) / ntotal; occupancy q at t=1 can be expressed as (n1-1 + n0-1) / ntotal.
Table 2. Example of measured data on four possible transitions: Lychnis flos-cuculi. For each species the dynamics values refer to number of plots in each dynamics category per sampling period. Data is in presence/absence form, as totals per dynamics category per period. p is the occupancy at t = 0: n1-1 + n1-0; q is the occupancy at t = 1: n1-1 + n0-1;
mean pq is the average occupancy at t = 0 and t = 1. Dynamics data (Lychnis flos-cuculi)
Period Years Remai-ning absent Remai-ning present Appea-rance Dis- appea-rance Total no. of plots p q Mean pq t0 t1 n0-0 n1-1 n0-1 n1-0 ntotal t0 = 1 t1 = 1 1 - 2 76-83 (79.5) - 84-92 (88) 591 90 49 143 873 233 139 186 2 - 3 82-92 (88) - 93-94 (93.5) 138 34 19 26 247 60 53 56.5 3 - 4 93-94 (93.5) - 95-96 (95.5) 159 27 9 11 206 38 36 37 4 - 5 95-96 (95.5) - 97-98 (97.5) 239 37 14 15 305 52 51 51.5 5 - 6 97-98 (97.5) - 99 298 30 13 21 362 51 43 47 Total 1976-1999 1353 218 104 216 1963 434 322 378
In a metapopulation context, the aim is to explain or predict the occupancy of an individual species as a state of equilibrium between colonisation and extinction. In that approach, rates of colonisation are usually expressed as the number of plots with observed colonisation as a fraction of the total number of unoccupied plots at t = 0: n0-1 /
(n0-1 + n0-0). Extinction rates are expressed as a fraction of the total number of occupied
plots at t = 0: n1-0 / (n1-1 + n1-0), (Fig. 1a). As a mathematical consequence of this
calculus, there is a strong correlation between colonisation or extinction rates and occupancy.
Fig. 1. Graphic representation of the colonisation/extinction data of the selected 157 species. (a) The ‘traditional’ way of estimating colonisation and extinction rates by dividing the measured colonisation by the number of empty plots and the extinction by the number of occupied plots (p). (b) Measured appearance (n0-1)
and disappearance (n1-0)
plotted against mean occupancy (mean of occupied plots at t=0 (p) and t=1 (q) (= mean pq). Each species is represented by two points in the graph (one for appearance, one for disappearance). Theoretically expected appearance and disappearance values (assuming independence of occupancy in a plot at t=0 and t=1) indicated as a curve. (c) Measured appearance and disappearance rates can be expressed as fractions of the theoretically expected rates. C and E can be seen as relative colonisation and
extinction values, mathematically independent
of occupancy. 95% confidence interval expressing the accuracy of an
To achieve this, we opted for a statistically descriptive type of analysis, by determining relative colonisation and extinction values rather than extracting absolute values for colonisation and extinction rates. Let us assume that the occupancy q of a plot at t = 1 is completely independent of occupancy p at t = 0 (i.e. that occupancy is purely determined by chance each time). We can now express the four possible transitions in a different way, based on probabilities. The theoretically expected number of plots remaining empty n0-0 then equals (1-p)(1-q)ntotal, the number of plots remaining occupied n1-1 equals
pqntotal, appearance n0-1 equals (1-p)qntotal and disappearance n1-0 equals p(1-q)ntotal.
Theoretically, based purely on chance, we thus expect, that for a species with 0.5 occupancy (mean pq), we have a 25% chance for each transition. We can now express the measured appearance and disappearance values (Fig. 1b, data points) as fractions of the theoretically expected rates (Fig. 1b, curve):
Colonisation value
(
p
)
qn
totaln
C
−
=
−1
10 and Extinction value
(
q
)
n
totalp
n
E
−
=
−1
0 1We thus use the values C and E as a descriptive measure for the relative degree of ‘dynamics’, rather than measures that are ecologically meaningful in a metapopulation context. High values (close to 1) can be expected for highly dynamic species, (e.g. annuals), low values can be expected for persistent, long-living species (Fig. 1c). We used Pearson’s correlation (SPSS 8.0) to test for significant correlations between occupancy, trends, colonisation and extinction values.
We must bear in mind that the appearance or colonisation encompasses both immigration (‘true’ colonisation) and regeneration from propagules already present at the site, as well as germination and establishment. Disappearance or extinction means that no above-ground parts were visible anymore, while dormant seeds or below-ground organs may still be present. Moreover, it should be clear that the measured sums in each of these categories represent net result of all changes in between the two sampling events t=0 and t=1 (thus including possible ‘rescue effects’ and possible multiple appearance and disappearance events). Possible errors are discussed in Appendix 2.
Ecological plant characteristics and demography
Table 3. Description of A. Demographic and B. Ecological traits: n = number of records, Range = range or no. of classes, SD = standard deviation, Var = variance, S = source. Ellenberg statistics calculated without indifferent species. All demographic variables and the ecological variable mean plant height are interval type variables; all other ecological variables are nominal type.
n Range Mean SD Var Description S A. Demographic variables
Occupancy 157 0.5-96.5 17.60 23.51 552.49 % of occupied plots 1 Trend 157 -0.19-0.21 -0.008 5.01 0.0025 slope logistic regression 1 Colonisation value (appearance) 156 0.16-1.01 0.614 0.21 0.0456 C = measured col / (1-p)*q 1 Extinction value (disappearance) 157 0.18-1.02 0.661 0.17 0.0292 E = measured ext / p*(1-q) 1 B. Ecological variables Occupancy (occ) 157 3 17.60 23.51 552.49 0.5-5%, 5-30%, 30-97% 1 Minimum light requirements (L) 155 4 7.05 0.88 0.771 indifferent (X), shade tolerant (4-6), intermediate (7), light demanding (8-9) 2
Moisture (F) 144 4 7.47 1.82 3.327 indifferent (X), dry (4-6), moist (7-8), wet (9-11) 2 Nutrient requirements (N) 139 4 5.80 1.95 3.800 indifferent (X), oligo-trophic (1-4), mesooligo-trophic (5-6), eutrophic (7-9) 2
Acidity (R) 157 3 6.15 1.57 2.460 indifferent (X), acid-neutral (2-6), acid- neutral-alkaline (7-9)
2
Begin flowering time (fl)
157 4 5.48 1.16 1.354 January-April, May, June, July-August
3 Mean plant
height (hgt)
157 4.5-225 61.10 37.74 1424.1 low – high 3 Seed bank index
(sb) 144 3 2.17 0.88 0.774 transient (1), short-term persistent (< 5 years) (2), long-term persistent (≥ 5 years)(3)
4
Germination
time (ger) 142 5 . . . immediately, spring, summer, autumn spring, late 3 Selfing (self) 157 2 . . . selfing - not selfing 3 Dispersal (disp) 143 6 . . . water, wind, uspecialised,
agricultural, animal, mixed (water + animal & water + wind)
5
Life duration (life)
157 2 . . . annual, perennial 3 1) Calculated on the basis of data from ISV-vegetation database of the Province of South-Holland
2) Ellenberg et al. (1992); Wiertz (1992); Hill et al. (1999) 3) CBS (1997)
Since the ecological variables were not measured directly for our study species in our area, and due to possible co-dependence of many ecological variables, we opted for a more exploratory analysis of the data rather than testing specific a priori hypotheses. We performed multiple linear regressions with ‘best subset regression’. This method selects and compares all possible multiple linear regression models for a given set of independent variables (unlike other regression types, which select only one model). Interaction or quadratic terms can also be inserted into the regression model. The fit of the model is evaluated with Mallow’s Cp, which should be smaller or
approximately equal to the number of degrees of freedom in the model (Miller 1990). If several significant models appear, selection of the ‘best’ model(s) must be made on the basis of best professional judgement (biological knowledge). We used the statistical package GENSTAT version 4.1 for these analyses.
In our analyses, we used trend, colonisation value C and extinction value E as separate dependent variables. The Ellenberg values for light (L), moisture (F) and nitrogen (N) and pH (R) were converted to nominal variables, including ‘indifference’ as a class. Germination time (CBS, 1997) was converted into 5 classes (Table 3). Where two germination times were cited, we selected the earliest possible time after seed set. As depicted in Fig. 1, occupancy can be seen as a demographic variable (the ‘end-result’) or as an ecological species trait, in the sense that it indicates the general agreement between the preferences of a species and the environment. We decided to investigate to what extent possible differences in the three other demographic parameters could be attributed to differences in occupancy. We therefore included three occupancy classes (0.005-0.05, 0.05-0.30, 0.30-0.97) representing approximately equally sized classes of rare, intermediate and common species, in the analysis. All in all, 12 independent variables were included in the analyses (Table 3B).
Results
Species demography
Many more (n = 41) species showed a significant (p < 0.05) decrease in occupancy than an increase (n = 10, Table 4), indicating an overall decline in species richness. Common species show more positive trends than rare species (Fig. 2a). However, the explained variance in trend was low (r2 = 2.8%, p < 0.0001, n = 157) indicating that species with similar occupancies vary greatly in their trends. Occupancy was not related to colonisation values (Fig. 2b, r2 = 0.54%, p = 0.363, n = 156). Since common species are
also often more abundant (Gaston et al. 2000) and thus probably more resilient to extinction, we also calculated the correlation coefficient between occupancy and extinction. The explained variance was low (Fig. 2c, r2 = 7.9%, p < 0.0001, n = 157).
Table 4. Categories of species mean occupancy values and trends. We divided the mean occupancy values into six classes, so that species with similar occupancies could easily be recognised. Whether or not the occupancy increased or decrease significantly during the study period (trend) was tested for each species with quasi-binomial logistic regression. (contd. on opposite page)
Occupancy
Rare Intermediate
Trend 0.005-0.015 (0.5-1.5%) 0.015-0.05 (1.5-5%) 0.05-0.15 (5-15%) negative Eriophorum angustifolium*
Carex panicea* Caltha palustris* Cynosurus cristatus* Prunella vulgaris* Hydrocotyle vulgaris* Leontodon autumnalis* Agrostis canina* Bellis perennis* Achillea ptarmica* Plantago lanceolata* Cirsium palustre* Juncus subnodulosus* Epilobium palustre* Apium nodiflorum*
Veronica chamaedrys Mentha arvensis* Carex nigra Veronica beccabunga Sium latifolium* Equisetum palustre* Potentilla anglica* Filipendula ulmaria* Vicia cracca*
Carex acutiformis Deschampsia cespitosa* Trifolium pratense*
Carex ovalis Stellaria palustris* Potentilla anserina*
Potentilla palustris Valeriana officinalis Cicuta virosa
Hypericum tetrapterum* Bolboschoenus maritimus* Festuca pratensis
<-0.03 Senecio aquaticus Carex riparia Lythrum salicaria*
-0.03 Typha angustifolia Bromus hordeaceus Alisma plantago-aquatica
Festuca arundinacea Sagittaria sagittifolia Agrostis capillaris Trifolium dubium Catabrosa aquatica Phragmites australis
Equisetum arvense Symphytum officinale Lathyrus pratensis Iris pseudacorus Angelica sylvestris Persicaria maculosa Juncus conglomeratus Butomus umbellatus Geranium dissectum Plantago major Lysimachia vulgaris Rumex crispus Persicaria lapathifolia
0.00
Rorippa palustris Arrhenatherum elatius Rumex conglomeratus Stachys palustris Lysimachia thyrsiflora Persicaria mitis Epilobium parviflorum Senecio vulgaris Sparganium emersum Lycopus europaeus Eupatorium cannabinum Carex paniculata Dactylis glomerata Myosotis arvensis Peucedanum palustre Rumex obtusifolius
+0.03 Anthriscus sylvestris Atriplex prostrata Myosotis laxa (subsp. cespitosa)
>+0.03 Rorippa sylvestris Solanum dulcamara* Epilobium hirsutum*
Calystegia sepium Sonchus asper Poa annua Galium aparine Capsella bursa-pastoris Cardamine hirsuta*
Lamium purpureum var. purp. Ranunculus ficaria Epilobium tetragonum Carex pseudocyperus Cardamine flexuosa
Epilobium ciliatum Sonchus oleraceus Heracleum sphondylium Holcus mollis Persicaria minor Cirsium vulgare Sonchus arvensis Bidens frondosa
Table 4. contd. The lowest significant values (at p < 0.05 level) occurred at a slope of ± 0.03. In order to facilitate the interpretation of this table, we selected this as the separation value between trend classes, resulting in three trend classes. Species are in order of decreasing trend within each column. Species showing significant trends are indicated with an asterisk, * p < 0.05.
Occupancy
Intermediate (contd.) Common
Trend 0.15-0.30 (15-30%) 0.30-0.70 (30-70%) 0.70-0.97 (70-97%) negative Eleocharis palustris* Berula erecta* Carex disticha* Triglochin palustris* Anthoxanthum odoratum* Carex acuta* Mentha aquatica*
Lychnis flos-cuculi* Lotus pedunculatus* Rumex hydrolapathum* Oenanthe fistulosa*
<-0.03 Equisetum fluviatile* Sparganium erectum*
-0.03 Festuca rubra Trifolium repens
Lysimachia nummularia Ranunculus acris Sagina procumbens Stellaria media Acorus calamus Bidens cernua Cirsium arvense Ranunculus flammula Carex hirta Galium palustre Poa pratensis Ranunculus sceleratus
0.00
Persicaria amphibia Scutellaria galericulata Juncus effusus
Oenanthe aquatica Juncus articulatus Holcus lanatus Urtica dioica Rumex acetosa Glyceria fluitans
Phleum pratense Cerastium fontanum Persicaria hydropiper
+0.03 Polygonum aviculare Alopecurus geniculatus Ranunculus repens
>+0.03 Lolium perenne Cardamine pratensis
Alopecurus pratensis* Glechoma hederacea*
Taraxacum officinale Agrostis stolonifera
Stellaria uliginosa* Glyceria maxima* Rorippa amphibia* Poa trivialis
Elytrigia repens* Phalaris arundinacea*
This indicates that other factors than occupancy must be more important for explaining the trends.
We found that species with positive trends had higher colonisation values (C) than species with negative trends (Fig. 2d, r2 = 22.4%, p < 0.0001, n = 156). On the other
hand, ditch bank plant species with different trends are more similar in their extinction (E) patterns (Fig. 2e, r2 = 0.25%, p = 0.534, n = 157). In addition, we found that the
variance of the estimated colonisation value (0.0456) was much larger than that of the extinction value (0.0292) (Table 3A). Species thus differ more in their colonisation than in their extinction behaviour. All in all, this indicates that, during the last decades, colonisation has been a more important determinant of plant increase and decrease in ditch banks of dairy farms than extinction.
Colonisation and extinction in ditch bank plants were strongly positively correlated (Fig. 2f, r2 = 66.9%, p < 0.0001, n = 156). Good colonisers thus also tended to
have higher extinction values. As our data-set included both annuals and perennials, we wanted to know whether this correlation could be due to the presence of annuals which were expected to have both high colonisation and extinction values. The correlation, however, remained strong both within perennials (r = 0.776, p < 0.0001, n = 130) and within annuals (r = 0.889, p < 0.0001, n = 26). Furthermore, as can be expected, trend was highly correlated with the difference between colonisation and extinction (Fig. 2g, r2
= 53.0%, p < 0.0001, n = 156).
Ecological plant characteristics and demography
Using ‘best subset regression’ we were able to extract and compare all possible models to explain trend, colonisation and extinction, by including one or more of the 12 plant characteristics as independent variables. With only one independent variable included in the models (Table 5), several models were significant, but on the whole the explained variance (adjusted R2) was low and the difference between Mallow’s C
p-values and df
high, indicating a relatively low overall fit (Appendix 3).
Trend. Two alternative models (T1 and T2), with six variables were extracted, both with R2 ≈ 46% (Table 6). Oligotrophic, light demanding, short, outcrossing and rare
species are decreasing most (Table 7). Eutrophic, shade tolerant, tall, selfing and common species are increasing. In T1, germination time was also included, while moisture was included in T2. Species germinating in the summer and species preferring wet habitats decreased, while immediately germinating species and species preferring dry habitats increased. Due to the similarity of explained variances, no statistically based choice can be made between the two models. Pearson’s Chi-square test of moisture and germination time indicated that these two variables were not independent (χ2 = 26.482, p
Table 5. Results from the ‘best subset regression’ analysis. The eight best models with one dependent variable
explaining trend, colonisation and extinction are listed. See also Table 3.
Ecological df Trend Colonisation value C Extinction value E
trait R2
adj Cp p(F) R2adj Cp p(F) R2adj Cp p(F)
Occupancy 3 7.63 83.46 0.00 - - - 6.21 59.14 0.01 Light 3 5.63 87.74 0.01 - - - - - - Moisture 4 3.22 93.15 0.08 7.62 45.20 0.01 7.83 56.68 0.01 Nutrients 4 26.66 43.24 0.00 13.17 35.88 0.00 - - -
Acidity 3 - - - - - -
Colonisation. Three different models (C1-3) were extracted each with four terms and R2 ≈ 32% (Table 6). After forcing the four main ecological terms into the
models, one or more interaction terms were found for each model. By far the highest explained variance was found for C1 (R2 = 42%) with two interaction terms (Fig. 3). The
alternative models C2 and C3 had explained variances just above 30%. We therefore decided to adopt model C1 as the ‘best’ model for the colonisation. The lowest colonisation values were shown by oligotrophic species with mixed dispersal and a transient seed bank (Table 7). Eutrophic species with agricultural dispersal and a long-term persistent seed bank had the highest colonisation values. Species with inlong-termediate occupancy had lower colonisation values than common and rare species. See Fig. 3 for a graphic illustration of the interaction effects for nutrients and seed bank and seed bank and dispersal.
Table 6. Results from the ‘best subset regression’ analysis with multiple independent variables explaining the trend, colonisation and extinction values. The best models with the largest number of terms indicated. Explained variance (adjusted R2). Mallow’s C
p
-values and degrees of freedom (df) indicated. Cells contain p--values based on F-statistics; ntrend = 121, nC = 120, nE = 121. N = nutrients, sb = seed bank type, germ =
germination time. See also Table 3.
Ecological trait Trend Colonisation
value C Extinction value E
T1 T2 C1 C2 C3 E1 Occupancy 0.00 0.00 0.03 - 0.02 0.00 Light 0.00 0.00 - - - 0.01 Moisture - 0.03 - - - - Nutrients 0.00 0.00 0.00 1 0.01 2 0.00 3 - Acidity - - - - - - Begin flowering time - - - - - - Height 0.02 0.00 - - - 0.01 Seed bank - - 0.00 1 0.01 2 0.00 3 0.01 Germination time 0.02 - - - 0.05 3 0.02 Selfing 0.01 0.02 - 0.03 - - Dispersal vector - - 0.02 1 0.02 - 0.00 Life duration - - - - model statistics R2adj 46.56 46.08 32.87 31.84 30.68 39.42 Cp 11.11 11.03 11.37 11.99 13.83 12.14 df 14 13 13 12 12 17
significant interaction terms (when other terms forced):
1 (a) N*sb and (b) sb*disp: R2
adj model = 42.29, Cp = 17.20, df = 27, (a) 0.013, n = 142,(b) p(F) =
0.036, n = 134.
2 N*sb: R2
adj model = 34.41, Cp = 36.41, df = 18, p(F) = 0.036, n = 142. 3 sb*germ: R2
adj model = 31.33, Cp = 13.10, df = 20, p(F) = 0.040, n = 128 or N*sb: R2adj model =
Extinction. One model (E1) with six explanatory terms was extracted (R2 =
39%, Table 6). The lowest extinction rates occurred in common species with mixed dispersal, transient seed bank type and late spring germination, while the highest extinction rates were found in rare species dispersing via agricultural practices, with a long-term persistent seed bank and spring germination (Table 7). Tall and light demanding species also showed low extinction rates, while low-growing and shade tolerant species also tended to have high extinction rates.
Table 7. A summary of important ecological traits associated with different trend, colonisation and extinction values. For trend the results from two models (T1 and T2) are indicated. The ecological trait values are ordered from more negative to more positive trends and from low to high colonisation and extinction rates, see also Fig 3. and Appendix 2.
Trend
(models T1 and T2) negative < positive
Nutrients oligotrophic < indifferent < mesotrophic < eutrophic
Occupancy intermediate < rare<common
Light light demanding < intermediate < shade tolerant
Selfing not selfing or unknown < selfing
Height short < tall
and
Germination time(T1) summer < autumn < late spring < spring < immediate or
Moisture(T2) wet < indifferent < moist < dry
Colonisation value (model C1)
low < high
Seed bank transient < short-term persistent < long-term persistent
Nutrients oligotrophic < indifferent < mesotrophic < eutrophic
Dispersal mixed < unspecialised < water <animals < wind < agricultural
Occupancy intermediate < commom < rare
Nutrients * seed bank transient and indifferent or mesotrophic; short-term persistent and oligo- or mesotrophic < short-term persistent and eutrophic < long-term persistent and meso- or eutrophic
Seed bank * dispersal transient and water or wind or mixed dispersal < short- or long-term persistent and wind dispersal < transient or long-long-term persistent and agricultural dispersal
Extinction value
(model E1) low < high
Dispersal mixed < unspecialised < water <animals < wind < agricultural
Occupancy common < intermediate < rare
Seed bank transient < short-term persistent < long-term persistent
Light light demanding < intermediate < shade tolerant
Height tall < short
Discussion
Species demography
Our study supports the alarming reports that species diversity of ditch bank vegetation has been deteriorating (Table 4; Provincie Zuid-Holland 1998). Common species became more common, while intermediate species [including many ‘characteristic species’ for the Western Peat District, such as Caltha palustris and Lychnis flos-cuculi, (Westhoff and Weeda, 1984)] became rarer. The decline in intermediate species is particularly worrying; since these species have few opportunities to migrate back to the pastures, they are likely to become the new threatened species in the future.
Variation among species trends may result from differences in extinction, colonisation, or a combination of both. Previous studies have shown that colonisation and extinction may be important bottlenecks hampering restoration of plant species richness (see e.g. Ouborg 1993; Strykstra, et al. 1998; Bakker & Berendse 1999). Yet, to our knowledge, no study has attempted to estimate the relative importance of these two processes for a large number of species. For ditch banks, we have now been able to demonstrate that the process of colonisation is relatively more important than extinction for determining species increase and decline in the community (Figs. 2d, e). Not only does this information increase our understanding of vegetation dynamics and diversity, it also gives us a clear indication where we should focus our resources if we wish to combat current species decline as efficiently as possible.
Trends explained by colonisation or extinction
Next to a small effect of occupancy, the factors nutrients, light, height, selfing, and germination time (T1) or moisture (T2) were significantly associated with differences in trend (Tables 6, 7, Appendix 3). Our results thus clearly show that, overall, competition (particularly nutrient tolerance, R2 = 26.7%) still plays an important part in species increase and decline in this habitat. Negative relationships between nutrient availability and species richness have been documented before (e.g. Mountford et al. 1993; Tallowin 1996). However, our models also show that the effect is mainly through colonisation (Table 7, Appendix 3): eu- and mesotrophic species have higher colonisation values, while extinction values are similar for all species. This is in accordance with the findings of Hodgson (1986), who showed that once established, some oligotrophic perennials can persist for a long time after the environment has become unfavourable for establishment.
can be expected to most optimally make use of temporary gaps in productive environments (Olff et al. 1994b).
Tilman (1993) also found that more species went extinct under nutrient rich conditions. In our study, species differing in nutrient requirements did not show large differences in extinction values. In our relatively nutrient-rich habitat, a selection has probably already eliminated most species with low nutrient tolerance. Mean height and light requirements, however, do regulate species trends via their effect on extinction (Table 7). High extinction values and decreasing trends are associated with small species. Small species are apparently outcompeted by taller ones.
In addition, our results show that competition and establishment only explains part of the trend. We also found that outcrossing species have a higher chance of decreasing than selfing species (Appendix 3). This points to pollen-limited seed set, either caused by lack of suitable source populations or by lack of pollinators (Steffan-Dewenter & Tscharntke 1999). In both cases, it indicates that isolation is also a problem (Geertsema 2002), requiring management solutions on a regional level.
Colonisation, extinction and seed traits
Surprisingly, we found no effect on the trend of seed traits (seed bank type, dispersal vector; Table 6, Appendix 3). This corresponds with the results of Thompson (1994), yet contradicts those of Hodgson & Grime (1990) who found considerable differences in the seed bank and dispersal between decreasing and increasing species. However, since we were able to investigate colonisation and extinction separately from the trend, we can see that seed traits actually do have strong effects on both colonisation and extinction.
We found no net effect on the trend of dispersal vector (Tables 5, 6). Yet, water and animal dispersed species tend to decrease most, while ‘agricultural species’ are increasing (Appendix 3). The success of ‘agricultural species’ is logical in a habitat where mowing machinery and cattle are likely to regularly transport seeds (Strykstra et al. 1997). The low colonisation ability of water dispersed species is more surprising, as ditches might be expected to function as dispersal corridors (but see van Dorp et al. 1997; Geertsema 2002). As many target species belong to this group (e.g. Caltha palustris, Filipendula ulmaria, Lysimachia thyrsiflora, Lysimachia vulgaris), we would like to know more about the problems with water dispersal. Low colonisation could be a reflection of problems in the establishment phase, because water dispersed species with a transient seeds have lower colonisation values than species with long-term persistent seeds (Fig. 3). A similar pattern can be seen in wind dispersers. We thus see that successful colonisation not only depends on good dispersal abilities (regional processes), but is also dependent on the suitability of local site conditions for establishment.
Implications for management
The correlation between colonisation and trend shows that, during the 25-year study period, colonisation (rather than extinction) has been a more important determinant of plant increase and decrease. Future management efforts should thus primarily focus on enhancing this process: dispersal and / or establishment should be aided. Nonetheless, since the correlation between trend and the difference between colonisation and extinction is even stronger than that between trend and colonisation alone, extinction remains relevant.
Many of the selected ecological factors in the regression models are related to competition (nutrient tolerance, light requirements, height). Others relate more to the process of establishment (germination and seed bank). We also found support for the importance of regional processes (selfing and dispersal). On the basis of this, we suggest the following management strategies:
Continued reduction of nutrients. Further efforts to reduce the nutrient levels of ditch banks are necessary to encourage species diversity since the levels are clearly still too high for many species. And there are many ditch banks, which show that it is possible successfully to combine production on the fields with species-rich ditch bank vegetation.
Creation of gaps. Because germination and seedling establishment appear to be important bottlenecks, the creation of gaps could, for many species, provide the window of opportunity they need to establish.
Utilization of the seed bank. High colonisation values are associated with a long-term persistent seed bank. Still, many such species are decreasing, probably because they prefer less nutrient rich conditions. Sod cutting, which uncovers the seed bank and removes nutrients might be an efficient way to promote these species, although care should be taken not to deplete the seed bank.
Regional measures to promote dispersal and seed set. The importance of selfing and dispersal for species demography show that isolation is a problem and that a regional management focus is necessary. In order to maximise the effect of ‘nature-friendly’ ditch bank management, it would be useful to start by focusing these efforts in species rich areas, e.g. in the vicinity of nature reserves. Here, this management has the greatest chance of succeeding. Successful examples, in turn, would encourage other farmers and gradually the custom would spread, increasing the biodiversity of the landscape as a whole.
Acknowledgements
We especially want to thank Adrie van Heerden and Iman Zorge from Provincie Zuid-Holland for granting us access to their vegetation database. Evert Meelis, Wil Tamis, Marjolein Dutmer and Hans Metz from the University of Leiden provided valuable assistance with many statistical analyses. We also want to thank Eddy van der Meijden from the University of Leiden, Jan Bakker from the University of Groningen and two referees for comments on earlier versions of the manuscript.
Reference list
Bakker, J. P. & Berendse, F. 1999. Constraints in the restoration of ecological diversity in grassland and heathland communities. Trends Ecol. Evol. 14: 63-68.
Berendse, F., Oomes, M. J. M., Altena, H. J. & Elberse, W. Th. 1999. Experiments on the restoration of species-rich meadows in The Netherlands. Biol. Conserv. 62: 59-65. Bignal, E. M. & McCracken, D. I. 1996. Low-intensity farming systems in the conservation of the
countryside. J. Appl. Ecol. 33: 413-424.
Bunce, R. G. H. & Hallam, C. J. 1998. The ecological significance of linear features in agricultural landscapes in Britain. In: Bunce, R. G. H., Ryszkowski, L. & Paoletti, M. G. (eds.) Landscape ecology and agroecosystems, pp. 11-19. Lewis Publishers, Boca Raton, Florida, USA.
CBS 1997. Register Biodiversiteit (BIOBASE). Centraal Bureau voor de Statistiek, Voorburg / Heerlen, NL.
Ellenberg, H., Weber, H. E., Düll, R., Wirth, V., Werner, W. & Paulissen, D. 1992. Zeigerwerte von Pflanzen in Mitteleuropa. Scripta Geobotanica 18: 1-258.
Erlich, P. R. 1988. The loss of diversity: causes and consequences. In: Wilson, E. O. (ed.) Biodiversity, pp. 21-27. Natonal Academy Press, Washington D.C., USA.
Gaston, K. J., Blackburn, T. M., Greenwoods, J. J. D., Gregory, R. D., Quinn, R. M. & Lawton, J. H. 2000. Abundance-occupancy relationships. J. Appl. Ecol. 37 (Suppl. 1): 39-59. Geertsema, W. 2002. Plant survival in dynamic habitat networks in agricultural landscapes. PhD
thesis. Wageningen University, Wageningen, NL.
Grime, J. P., Hodgson, J. G. & Hunt, R. 1988. Comparative Plant Ecology. Unwin Hyman, London, UK.
Hanski, I. 1998. Metapopulation dynamics. Nature 396: 41-49.
Higler L. W. G., 1994. Sloten. In: Beije, H. M., Higler, L. W. G., Opdam, P. F. M.van rossum, T. A. W. & Verkaar, H. J. P. A.(eds.) Bos- en Natuurbeher in Nederland. Deel 1: Levensgemeenschappen, pp. 89-97. Backhuys Publishers, Leiden, NL.
Hill, M. O., Mountford, J. O., Roy, D. B. & Bunce, R. G. H. 1999. Ellenberg's indicator values for British plants. Technical Annex, Institute of Terrestrial Ecology / Department of Environment, Transport and Regions (NERC), Grange-over-Sands, UK.
Hodgson, J. G. 1986. Commonness and rarity in plants with special reference to the Sheffield flora. Part II: the relative importance of climate, soils and land use. Biol. Conserv. 36: 253-274.
Hodgson, J. G. & Grime, J. P. 1990. The role of dispersal mechanisms, regenerative strategies and seed banks in the vegetation dynamics of the British landscape. In: Bunce, R. G. H. & Howard, D. C. (eds.) Species dispersal in agricultural habitats, pp. 65-81. Belhaven Press, London, UK.
Hornik, K. 2001. Statistical computation and graphics system R. Available at: http://www.ci.tuwien.ac.at/~hornik/R/.
Husband, B. C. & Barrett, S. C. H. 1996. A metapopulation perspective in plant population biology. J. Ecol. 84: 461-469.
KNMI 2001. Langjarige gemiddelden 1971-2000. Available at: http://www.knmi.nl/voorl/maand/normalen71_001.html
Kruk, M. 1991. Somerset Levels & Moors: een veenweidegebied in Engeland. Veenweide 4: 21-22.
MacArthur, R. H. & Wilson, E. O. 1967. The theory of island biogeography. Princeton University Press, Princeton, USA.
Marshall, E. J. P. & Hopkins, A. 1990. Plant species composition and dispersal in agricultural land. In: Bunce, R. G. H. & Howard, D. C. (eds.) Species dispersal in agricultural habitats pp. 98-116. Belhaven Press, London, UK.
McCullagh, P. & Nelder, J. A. 1989. Generalized linear models. Chapman and Hall, London, UK. McNeely, J. A., Gadgil, M., Levèque, C., Padoch, C., Redford, K., 1995. Human influences on
biodiversity. In: Heywood, V. H. & Watson, R. T. (eds.) Global biodiversity assessment, pp. 711-821. UNEP, Cambridge University Press, Cambridge, UK.
Melman, D. 1991. Slootkanten in het veenweidegebied: mogelijkheden voor behoud en ontwikkeling van natuur in agrarisch grasland. PhD thesis, Leiden University, Leiden, NL (with English summary).
Melman, D. & van Strien, A. 1993. Ditch banks as a conservation focus in intensively exploited peat farmland. In: Vos, C. & Opdam, P. (eds.) Landscape ecology of a stressed environment, pp. 122-141. Chapman & Hall, London, UK.
Miller, A. J. 1990. Subset selection in regression. Chapman & Hall, London, UK.
Olff, H. & Bakker, J. P. 1991. Long-term dynamics of standing crop and species composition after the cessation of fertilizer application to mown grassland. J. Appl Ecol 28: 1040-1052. Olff, H., Berendse, F. & de Visser, W. 1994a. Changes in nitrogen mineralization, tissue nutrient
concentrations and biomass compartmentation after cessation of fertilizer application to mown grassland. J. Ecol 82: 611-620.
Olff, H., Pegtel, D. M., van Groenendael, J. M. & Bakker, J. P. 1994b. Germination strategies during grassland succession. J. Ecol 82: 69-77.
Ouborg, N. J. 1993. Isolation, population size and extinction: the classical and metapopulation approaches applied to vascular plants along the Dutch Rhine-system. Oikos 66, 298-308. Pfadenhauer, J. & Klötzli, F. 1996. Restoration experiments in middle European wet terrestrial
ecosystems: an overview. Vegetatio 126: 101-115.
Prins, A. H., Dijkstra, G. A. & Bekker, R. M. 1998. Feasibility of target communities in a Dutch brook valley system. Acta Bot. Neerl. 47: 71-88.
Provincie Zuid-Holland 1993. De vegetatie van Zuid-Holland 1976-1991. De wilde plantengroei in het bijzonder van het agrarisch cultuurlandschap. Deel 1: Beleidsrapport; Deel 2: Onderzoeksrapport en Kaartenbijlage. Provincie Zuid-Holland, Dienst Ruimte en Groen, Den Haag, NL.
Provincie Holland 1998. Staat van de natuur: een graadmeter voor de natuur in Zuid-Holland 1998. Provincie Zuid-Zuid-Holland, Dienst Ruimte, Groen en Gemeenten / Dienst water en Milieu, Rotterdam, NL.
Ruthsatz, B., Haber, W., 1982. The significance of small-scale landscape elements in rural areas as refuges for endangered plant species. In: Tjallingii, S. P. & de Veer, A.A. (eds.) Perspectives in Landscape Ecology, pp. 117-124. Pudoc, Wageningen.
Schaminée, J. H. J., van Duuren, L. & de Bakker, A. J. 1992. Europese en mondiale verspreiding van Nederlandse vaatplanten. Gorteria 18: 57-96 (with English abstract).
Steffan-Dewenter, I. & Tscharntke, T. 1999. Effects of habitat isolation on pollinator communities and seed set. Oecologia 121: 432-440.
Strykstra, R. J., Bekker, R. M. & Bakker, J. P. 1998. Assessment of dispersule availability: its practical use in restoration management. Acta Bot. Neerl. 47: 57-70.
Strykstra, R. J., Verweij, G. L. & Bakker, J. P. 1997. Seed dispersal by mowing machinery in a Dutch brook valley system. Acta Bot. Neerl. 46: 387-401.
Tallowin, J. R. B. 1996. Effects of inorganic fertilizers on flower-rich hay meadows: a review using a case study on the Somerset Levels, UK. Grassland and Forage Abstracts 66: 147-152.
Tamis, W. L. M., van 't Zelfde, M., van Elk, R. & Witte, J. P. M. 2000. Modellering van de kansrijkdom van het biologisch herstel van natte en vochtige vegetaties. CLM & RIZA, Lelystad, NL (with English summary).
Thompson, K. 1994. Predicting the fate of temperate species in response to human disturbance and global change. In: Boyle, T. J. B. & Boyle, C. E. B. (eds.) Biodiversity, Temperate Ecosystems and Global Change, pp. 61-76. Springen Verlag Berlin Heidelberg, Germany.
Thompson, K., Bakker, J. P. & Bekker, R. M. 1997. The 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: 2179-2191.
van der Linden, H., 1982. History of the reclamation of the western fenlands and the organizations to keep them drained. In: de Bakker, H. & van den Berg, M. W. (eds.) Proceedings of the symposium on peat lands below sea level, pp. 42-73. ILRI Publication 30, Wageningen, NL.
van der Meijden, R. 1996. Heukels' Flora van Nederland. Wolters-Noordhoff bv., Groningen, NL. van der Molen, W. H., 1982. Water management in the western Netherlands. In: de Bakker, H. &
van Dorp, D. 1996. Seed dispersal in agricultural habitats and the restoration of species-rich meadows. PhD thesis, Wageningen Agricultural University, Wageningen, NL.
van Dorp, D., Schippers, P. & van Groenendael, J. M. 1997. Migration rates of grassland plants along corridors in fragmented landscapes assessed with a cellular automation model. Landscape Ecology 12: 39-50.
van Harmelen, W., van Heerden, A. & Kruk, M. 1997. Waardevolle oeverplanten voor het agrarisch natuurbeheer: een hulpmiddel bij de herkenning. Provincie Zuid-Holland, Den Hâneker, Dienst Landelijk Gebied, Voorburg, NL.
van Strien, A. 1991. Maintenance of plant species diversity on dairy farms. PhD thesis, Leiden University, Leiden, NL.
van Strien, A. J., van der Linden, J., Melman, Th. C. P. & Noordervliet, M. A. W. 1989. Factors affecting the vegetation of ditch banks in peat areas in the western Netherlands. J. Appl. Ecol. 26, 989-1004.
Verkaar, D., van Duuren, L. & Schaminée, J. H. J. 1992. De internationale betekenis van Nederland voor hogere planten op basis van biogeografische gegevens. De Levende Natuur 2: 34-39.
Westhoff, V. & Weeda, E. 1984. De achteruitgang van de Nederlandse flora sinds het begin van deze eeuw. Natuur en Milieu 6: 8-17.
Wiertz, J. 1992. Schatting van ontbrekende vocht- en stikstofindicatiegetallen van Ellenberg (1979). Instituut voor Bos- en Natuuronderzoek (IBN-DLO), Wageningen, NL.
Appendix 1. List of the 308 terrestrial herbaceous plant species present in ditch banks in the research area. A. 157 more ‘common’ species and B. 151 very rare species. % occ = % occupancy, n-v = nature-value (Clausman & van Wijngaarden 1984). Mean n-v of all species = 34.5. ‘Target’ species in this thesis are defined according to two criteria: 1) species with a nature-value ≥ 35 and 2) those species present on a list of 25 valuable ditch bank species (van Harmelen et al. 1997). These ‘target’ species are indicated in bold letter type.
A. 157 species with occupancy > 0.5%
Species n-v % occ Species n-v % occ
Achillea ptarmica 42 1.17 Deschampsia cespitosa 27 4.66
Acorus calamus 33 15.66 Eleocharis palustris 29 20.40
Agrostis canina 39 1.62 Elytrigia repens 10 35.06
Agrostis capillaris 29 9.94 Epilobium ciliatum 33 1.42 Agrostis stolonifera 18 96.47 Epilobium hirsutum 23 11.89 Alisma plantago-aquatica 29 10.41 Epilobium palustre 47 2.45
Alopecurus aequalis 46 0.67 Epilobium parviflorum 27 8.96
Alopecurus geniculatus 23 53.15 Epilobium tetragonum 33 7.48 Alopecurus pratensis 21 35.69 Equisetum arvense 17 3.65 Angelica sylvestris 29 3.16 Equisetum fluviatile 36 21.32
Anthoxanthum odoratum 27 22.17 Equisetum palustre 25 14.11 Anthriscus sylvestris 24 1.17 Eriophorum angustifolium 45 0.52
Apium nodiflorum 36 8.23 Eupatorium cannabinum 31 0.95
Arrhenatherum elatius 30 1.72 Festuca arundinacea 36 1.40
Atriplex prostrata 27 2.07 Festuca pratensis 25 10.64 Bellis perennis 23 5.15 Festuca rubra 22 24.13 Berula erecta 30 20.49 Filipendula ulmaria 31 2.88
Bidens cernua 31 32.33 Galium aparine 19 0.70 Bidens frondosa 27 0.97 Galium palustre 35 62.86
Bolboschoenus maritimus 34 3.83 Geranium dissectum 38 1.98
Bromus hordeaceus 21 4.11 Glechoma hederacea 19 73.62
Butomus umbellatus 42 13.54 Glyceria fluitans 24 84.23
Caltha palustris 36 6.28 Glyceria maxima 22 90.67
Calystegia sepium 21 1.27 Heracleum sphondylium 32 0.70 Capsella bursa-pastoris 16 2.38 Holcus lanatus 17 93.37
Cardamine flexuosa 50 2.43 Holcus mollis 35 1.37
Cardamine hirsuta 34 7.11 Hydrocotyle vulgaris 40 5.78
Cardamine pratensis 22 71.00 Hypericum tetrapterum 48 0.88
Carex acuta 34 23.50 Iris pseudacorus 40 11.46
Carex acutiformis 39 0.57 Juncus articulatus 30 59.56
Carex disticha 43 19.57 Juncus conglomeratus 44 4.55
Carex hirta 28 27.86 Juncus effusus 24 43.87
Carex nigra 45 8.86 Juncus subnodulosus 45 0.98
Carex ovalis 43 0.85 Lamium purpureum var. purpureum 24 0.60
Carex panicea 50 0.82 Lathyrus pratensis 32 2.56
Carex paniculata 44 1.82 Leontodon autumnalis 25 0.82
Carex pseudocyperus 50 0.82 Lolium perenne 12 40.19
Carex riparia 36 4.75 Lotus pedunculatus 40 34.44
Catabrosa aquatica 33 4.95 Lychnis flos-cuculi 44 19.72
Cerastium fontanum 21 57.23 Lycopus europaeus 29 13.11
Cicuta virosa 45 5.71 Lysimachia nummularia 36 21.52
Cirsium arvense 18 16.54 Lysimachia thyrsiflora 45 11.84
Cirsium palustre 37 9.99 Lysimachia vulgaris 32 3.30
Cirsium vulgare 21 1.02 Lythrum salicaria 31 7.89
Cynosurus cristatus 34 0.58 Mentha aquatica 31 16.39
A. 157 species with occupancy > 0.5% contd.
Species n-v % occ Species n-v % occ
Myosotis arvensis 29 0.80 Rumex conglomeratus 34 12.97
Myosotis laxa (subsp. cespitosa) 43 5.08 Rumex crispus 19 8.86
Oenanthe aquatica 37 18.14 Rumex hydrolapathum 36 18.87
Oenanthe fistulosa 37 53.63 Rumex obtusifolius 23 6.78
Persicaria amphibia 22 36.51 Sagina procumbens 23 27.18 Persicaria hydropiper 27 71.64 Sagittaria sagittifolia 41 4.21
Persicaria lapathifolia 20 2.10 Scutellaria galericulata 34 25.40 Persicaria maculosa 18 5.45 Senecio aquaticus 43 0.67
Persicaria minor 42 0.77 Senecio vulgaris 17 1.48
Persicaria mitis 39 2.80 Sium latifolium 36 4.03
Peucedanum palustre 38 4.25 Solanum dulcamara 23 1.98
Phalaris arundinacea 20 70.49 Sonchus arvensis 21 0.65 Phleum pratense 19 20.97 Sonchus asper 23 1.90 Phragmites australis 18 5.48 Sonchus oleraceus 20 0.67 Plantago lanceolata 23 2.03 Sparganium emersum 43 2.23
Plantago major 18 10.11 Sparganium erectum 28 35.39
Poa annua 10 12.61 Stachys palustris 30 4.76
Poa pratensis 20 27.05 Stellaria media 11 42.37 Poa trivialis 12 94.97 Stellaria palustris 58 4.36
Polygonum aviculare 16 19.29 Stellaria uliginosa 36 47.72
Potentilla anglica 58 0.98 Symphytum officinale 31 5.25
Potentilla anserina 21 13.97 Taraxacum officinale 12 46.04
Potentilla palustris 41 1.08 Trifolium dubium 28 0.53
Prunella vulgaris 31 2.37 Trifolium pratense 22 9.88
Ranunculus acris 22 32.98 Trifolium repens 13 66.74 Ranunculus ficaria 4.03 Triglochin palustris 37 20.29
Ranunculus flammula 43 40.31 Typha angustifolia 32 0.53
Ranunculus repens 14 96.14 Urtica dioica 17 19.70 Ranunculus sceleratus 27 48.87 Valeriana officinalis 30 3.35 Rorippa amphibia 32 59.46 Veronica beccabunga 39 1.12
Rorippa palustris 28 5.21 Veronica chamaedrys 38 0.58
Rorippa sylvestris 30 1.37 Vicia cracca 25 6.18
Rumex acetosa 23 50.27
B. 151 very rare species with occupancy < 0.5%
Species n-v % occ Species n-v % occ
Achillea millefolium 22 0.47 Carex otrubae 38 0.15
Ajuga reptans 46 0.02 Carex rostrata 44 0.12
Alisma gramineum 50 0.25 Carex spicata 45 0.02
Alisma lanceolatum 50 0.23 Carex vesicaria 52 0.10
Alliaria petiolata 35 0.02 Centaurea jacea 35 0.05
Anisantha sterilis 33 0.02 Cerastium glomeratum 37 0.22
Arenaria serpyllifolia 37 0.02 Chamerion angustifolium 28 0.02
Aster tripolium 40 0.05 Chenopodium album 18 0.43
Atriplex patula 26 0.18 Chenopodium ficifolium 24 0.03
Barbarea vulgaris 44 0.02 Chenopodium glaucum 36 0.02
Brassica napus 0.08 Chenopodium polyspermum 32 0.47
Brassica nigra 38 0.02 Chenopodium rubrum 29 0.13
Brassica rapa 0.18 Cirsium dissectum 57 0.27
Calamagrostis canescens 44 0.40 Convolvulus arvensis 29 0.08
Calla palustris 63 0.07 Conyza canadensis 30 0.10
Callitriche stagnalis 53 0.02 Cornus sanguinea 0.03
Carduus crispus 34 0.17 Coronopus didymus 39 0.02
Carex echinata 53 0.08 Coronopus squamatus 35 0.02
Carex elata 51 0.08 Crataegus monogyna 29 0.13
B. 151 very rare species with occupancy < 0.5% contd.
Species n-v % occ Species n-v % occ
Crepis paludosa 59 0.02 Poa angustifolia 0.20
Cyperus fuscus 62 0.17 Poa nemoralis 45 0.02
Draba muralis 53 0.03 Poa palustris 44 0.15
Eleocharis palustris 47 0.15 Potentilla erecta 43 0.28
Epilobium montanum 34 0.17 Potentilla reptans 30 0.40
Epilobium obscurum 51 0.12 Ranunculus circinatus 38 0.08
Epilobium roseum 51 0.02 Ranunculus lingua 57 0.07
Erodium cicutarium 37 0.02 Ranunculus sardous 48 0.03
Erysimum cheiranthoides 33 0.03 Raphanus sativus 0.02
Euphorbia esula 40 0.02 Rhinanthus angustifolius 44 0.03
Euphorbia helioscopia 27 0.03 Rorippa austriaca 0.02 Fallopia convolvulus 24 0.03 Rumex acetosella 25 0.15
Festuca rubra 0.02 Rumex sanguineus 46 0.12
Galeopsis speciosa 47 0.02 Sagina nodosa 53 0.03
Galium mollugo 29 0.05 Schoenoplectus lacustris 42 0.05
Galium uliginosum 52 0.20 Schoenoplectus tabernaemontani 42 0.05
Geranium molle 33 0.12 Scrophularia nodosa 33 0.05
Geranium pusillum 38 0.05 Scrophularia umbrosa 48 0.03
Geranium robertianum 37 0.02 Senecio erucifolius 41 0.05
Glyceria notata 46 0.02 Senecio viscosus 36 0.02
Gnaphalium uliginosum 37 0.50 Sisymbrium altissimum 0.02
Hordeum secalinum 44 0.02 Solanum nigrum 23 0.38
Hypericum dubium 48 0.03 Stellaria aquatica 44 0.07
Hypericum perforatum 31 0.02 Stellaria graminea 34 0.38 Hypochaeris radicata 30 0.15 Stratiotes aloides 50 0.07
Isolepis setacea 47 0.03 Taraxacum celticum 0.02
Juncus acutiflorus 50 0.02 Taraxacum palustre 58 0.03
Juncus bulbosus 0.03 Tephroseris palustris 46 0.15
Juncus compressus 44 0.17 Thalictrum flavum 48 0.02
Juncus inflexus 34 0.13 Thlaspi arvense 29 0.02 Lactuca serriola 30 0.07 Trifolium fragiferum 44 0.05
Lamium album 20 0.40 Trifolium hybridum 44 0.02
Lamium purpureum 24 0.07 Tripleurospermum maritimum 29 0.07 Lapsana communis 26 0.15 Trisetum flavescens 40 0.02
Leersia oryzoides 62 0.28 Triticum aestivum 0.02
Leontodon saxatilis 40 0.32 Tussilago farfara 19 0.13
Lepidium ruderale 50 0.02 Typha latifolia 27 0.43
Leucanthemum vulgare 39 0.28 Urtica urens 23 0.05
Linaria vulgaris 28 0.03 Valeriana dioica 56 0.18
Lolium multiflorum 0.13 Valerianella locusta 48 0.02
Luzula campestris 35 0.05 Veronica agrestis 44 0.02
Luzula multiflora 40 0.22 Veronica arvensis 31 0.40
Matricaria discoidea 16 0.50 Veronica catenata 34 0.20 Matricaria recutita 24 0.18 Veronica hederifolia 44 0.10
Medicago lupulina 28 0.10 Veronica persica 29 0.02
Melilotus officinalis 36 0.02 Veronica polita 41 0.02
Menyanthes trifoliata 53 0.27 Veronica scutellata 54 0.03
Mimulus guttatus 0.02 Veronica serpyllifolia 40 0.12
Molinia caerulea 34 0.08 Vicia hirsuta 35 0.05
Montia fontana 0.35 Vicia sativa 0.05
Myosotis discolor 50 0.07 Vicia sativa subsp. nigra 32 0.02
Myosotis ramosissima 41 0.02 Vicia sativa subsp. sativa 40 0.02
Osmunda regalis 50 0.02 Vicia sepium 39 0.33
Papaver rhoeas 32 0.02 Vicia tetrasperma 52 0.02
Pastinaca sativa 34 0.02 Viola palustris 52 0.43
Appendix 2. Reliability of demographic variables
In this appendix we discuss three potential sources of errors.
1. The validity of colonisation and extinction values. We wanted to know if the estimated colonisation and extinction values (Fig. 1c) represent ecologically meaningful differences between species. Using a simple simulation technique, based on random samples of 1963 plots out of populations characterised by C or E and p and q, we estimated the amount of variance due to sampling error in the total set of species to be around 0.007. Subtracting this from the variances (0.0456 and 0.0292, respectively) indicates that considerable rest-variance remains. In other words, for the whole group of species, we are dealing with significant ‘real’ differences in colonisation and extinction values between species.
2. The accuracy of individual colonisation and extinction values. For individual species, we have indicated the estimated binomial confidence interval (CI) for an individual colonisation or extinction value of 0.6 in Fig. 1c. This shows that the colonisation/extinction values for a single species are more reliable for species with ‘intermediate’ occupancies than for ‘rare’ or ‘common’ species. To test whether this inaccuracy could have affected our results, we also performed all correlations without the rarest species (occupancy < 0.015 excluded, n=22). The results remained approximately the same (except for the correlation between occupancy and extinction). Therefore, this inaccuracy did not seem to have any important effect on our conclusions regarding the relationship between the demographic factors.
3. Possible effects of time-lags. In order to include as many (rare) species as possible, we used the sum of measured appearance and disappearance values per species (see ‘totals’, Table 2) rather than the average values, neglecting the possible effects of differing time-lags between t=0 and t=1. Since short time-lags appear only in the last sampling periods, this could constitute a source of systematic bias in species showing a strong trend. We recalculated colonisation and extinction values correcting for the time-lag. Though the results indicate a general underestimation of both C and E (corrected values exceeded estimated values from 2% up to 8 %), the errors showed no relationship with the trend. Thus this error is not likely to have constituted any source of a systematic bias.
