Relationships between productivity, number of shoots and number of species in bryophytes and vascular plants

Authors

  • A. Bergamini,

    Corresponding author
    1. Institut für Systematische Botanik, Universität Zürich, Zollikerstrasse 107, CH-8008 Zürich, Switzerland,
    Search for more papers by this author
  • D. Pauli,

    1. Institut für Systematische Botanik, Universität Zürich, Zollikerstrasse 107, CH-8008 Zürich, Switzerland,
    2. Institut für Umweltwissenschaften, Universität Zürich, Winterthurerstrasse 190, CH-8057 Zürich, Switzerland
    Search for more papers by this author
  • M. Peintinger,

    1. Institut für Systematische Botanik, Universität Zürich, Zollikerstrasse 107, CH-8008 Zürich, Switzerland,
    2. Institut für Umweltwissenschaften, Universität Zürich, Winterthurerstrasse 190, CH-8057 Zürich, Switzerland
    Search for more papers by this author
  • B. Schmid

    1. Institut für Systematische Botanik, Universität Zürich, Zollikerstrasse 107, CH-8008 Zürich, Switzerland,
    2. Institut für Umweltwissenschaften, Universität Zürich, Winterthurerstrasse 190, CH-8057 Zürich, Switzerland
    Search for more papers by this author

Ariel Bergamini, Institut für Systematische Botanik, Universität Zürich, Zollikerstrasse 107, CH-8008 Zürich, Switzerland (tel. + 41 1634 84 11; fax + 41 1634 84 03; e-mail bergamin@systbot.unizh.ch).

Summary

  • 1We measured species density, biomass and shoot density for both bryophytes and vascular plants in 90 small plots in 18 calcareous fens. In addition, we recorded leaf area index and litter mass of vascular plants. Our goals were: (a) to compare the relationship between biomass and species density for the two taxonomic groups, (b) to test whether biomass and species density of bryophytes and vascular plants are related to their shoot density, and (c) to assess the degree to which biomass, shoot and species density of bryophytes are correlated with characteristics of the vascular plant layer.
  • 2For bryophytes there was a positive linear relationship between biomass and species density. Vascular plant species density was not related to biomass. Furthermore, bryophyte biomass and species density were linearly and positively related to bryophyte shoot density. For vascular plants, only biomass but not species density was related to shoot density.
  • 3We concluded that a bryophyte favourability gradient existed along which biomass and shoot and species density increased. This gradient was attributed to positive interactions within dense bryophyte stands, high clonal fragmentation, absence of competitive hierarchies and to the limited ability of larger bryophyte species to replace small species along this favourability gradient.
  • 4Since species density for vascular plants varied independently from biomass and shoot density, there was no such favourability gradient as for bryophytes. Large size variation, predominantly negative interactions between species, and clonal integration of species (e.g. tussock-forming grasses and sedges) may be responsible for the different behaviour of the two taxonomic groups.
  • 5Bryophyte favourability decreased with increasing vascular plant biomass. Concerning light availability, we found highest bryophyte favourability at intermediate levels where the combination of radiation and moisture seems to be optimal for bryophytes. No relationship was found between bryophyte favourability and vascular plant shoot density and litter mass.

  • 6The negative relationship between bryophyte favourability and vascular plant biomass is important for bryophyte conservation. Stands of low vascular plant production are those with the potential for highest species richness, and should therefore receive conservation priority.

Introduction

Bryophytes are a major component of many plant communities in terms of both biomass and species diversity (e.g. Vitt & Pakarinen 1977; Rieley et al. 1979; Longton 1984; Russel 1990). For example, in wet habitats such as fens and bogs, their biomass may exceed that of vascular plants (e.g. Longton 1984; Wheeler & Shaw 1991). Above-ground biomass or ‘productivity’ has been considered to be one of the most important determinants of species richness (Grime 1973, 1979; Rosenzweig & Abramsky 1993), and while for vascular plants this relationship has received much attention since the studies of Grime (1973, 1979) and Al-Mufti et al. (1977), it is poorly understood for bryophytes with the notable exception of rheophytic communities (Muotka & Virtanen 1995; Virtanen et al. 2001). We therefore focus upon the relationship between species number, biomass and shoot density of both bryophytes and vascular plants in 18 Swiss montane calcareous fens.

Species richness per unit area (hereafter called species density following Magurran 1988) of vascular plants at the level of small plots (≤ 1 m2) is often reported to be highest at intermediate levels of biomass. This leads to a ‘hump-shaped’ relationship between species density and biomass (reviewed in Grace 1999). Although this pattern is perhaps not as widespread as was previously thought (Waide et al. 1999), there is good evidence to suggest that very high biomass values are antagonistic to high species densities (Marrs et al. 1996; Grace 1999). Furthermore, plot biomass depends on the number of individuals (or ramets for clonal plants) growing in a plot. Fisher et al. (1943) and Preston (1962), by linking the number of species to the number of individuals, proposed that the number of species should increase with the number of individuals by a probabilistic effect of drawing different numbers of individuals from a single pool. However, does this hold for systems, such as calcareous fens, where most plant species grow clonally and for bryophytes as well as for vascular plants?

Most bryophytes are ectohydric (Buch 1947), i.e. they neither possess roots nor an internal vascular system, and since water and nutrient uptake occurs over the whole shoot surface, their size is restricted. Also, since bryophytes cannot regulate water loss, they frequently dry out and enter a physiologically inactive state and, thus, are considered to be poikilohydric plants (Walter 1962). Physiological activity is prolonged where shoot density is high, because evaporative water loss is reduced (Proctor 1982) and growth is therefore often best in dense stands (Bates 1988; Økland & Økland 1996; but see also Zamfir & Goldberg 2000). Such a positive effect of density on plant performance is less common for vascular plants and may occur mostly in communities growing in rather harsh environments (Callaghan 1987; Bertness & Callaway 1994).

Many studies have examined interactions among bryophytes (recently reviewed by Rydin 1997) and effects of bryophytes upon seedling emergence (e.g. Zamfir 2000) but very few studies have considered how the performance of the bryophyte layer is dependent upon structural properties of the vascular plant layer (Watson 1960; Sveinbjörnsson & Oechel 1992; Økland 1994). Nevertheless, structural properties of the vascular plant layer, such as biomass, shoot density, leaf area index (LAI) and litter mass, seem likely to be important since bryophytes are much smaller than most vascular plants and interactions for light are thus asymmetric (Rydin 1997). Light absorption of the vascular plant layer depends mainly on the cumulative leaf area (Monsi & Saeki 1953) and, in addition to the effect of living tissues, a thick, persistent litter layer will inhibit the development of a vigorous bryophyte layer (Cornish 1954; Wheeler & Giller 1982; van Tooren et al. 1988), mainly by heavy shading (Sveinbjörnsson & Oechel 1992). Dense stands of vascular plants could, however, benefit bryophyte growth if they reduce evaporative water loss by increased shading. Such contrasting effects could produce a unimodal relationship between biomass and shoot density of bryophytes and variables related to vascular plant productivity. Moreover, if bryophyte species density is unimodally related to bryophyte biomass, then bryophyte species density will be lowest at intermediate vascular plant productivity.

We asked the following specific questions:

  • 1Is their biomass an adequate predictor of the species density of bryophytes and vascular plants, and what is the shape of each relationship?
  • 2Do biomass and species density of bryophytes and vascular plants depend upon their shoot density?
  • 3What is the relationship between the properties of the bryophyte community (biomass, shoot density and species density) and those of the vascular plant layer (biomass, shoot density, LAI and litter mass)? Which variable or combination of variables best explains bryophyte variation?

Methods

STUDY SITES AND METHODS

We studied 18 montane wetlands located in the pre-Alps of north-eastern Switzerland. The sites were randomly selected from the inventory of Swiss fenlands (BUWAL 1990). They are situated between 800 and 1400 m a.s.l. and distributed over an area of 3500 km2. The use of extensive management practices (mowing once a year in September) means that these montane wetlands can be considered to be semi-natural communities. Annual precipitation is relatively high throughout the study region (1500–2800 mm, Uttinger 1967). Sites have predominantly north to north-western aspects and range in size from 1 to 9.1 ha. The underlying bedrock consists of various calcareous sediments of tertiary and mesozoic age (Spicher 1972). Soil pH ranges from 5.4 to 7.2 (mean 6.2) and the vegetation belongs mainly to the Caricion davallianae alliance (vegetation classification after Ellenberg 1996).

Species richness of both bryophytes and vascular plants is high (Peintinger 1999; Bergamini et al. 2001). Graminoids, such as Carex davalliana Sm., C. panicea L., Molinia caerulea Moench and Festuca rubra L., and forbs, such as Potentilla erecta (L.) Räuschel, Trifolium pratense L., Lotus corniculatus L.s.l. and Succisa pratensis Moench, are common (Plattner 1996; Peintinger 1999), and the bryophytes are usually dominated by pleurocarpous mosses, with Calliergonella cuspidata (Hedw.) Loeske, Climacium dendroides (Hedw.) Web. & Mohr, Campylium stellatum (Hedw.) J. Lange & C. Jens, Bryum pseudotriquetrum (Hedw.) Schwaegr., Drepanocladus cossonii (Schimp.) Loeske and Plagiomnium elatum (B. & S.) T. Kop. being prominent components (Bergamini et al. 2001).

Data were collected during August 1997 (i.e. at peak vascular plant biomass) in five randomly selected plots, each of 20 × 20 cm (0.04 m2), at each site (total n = 90). Leaf area index (LAI) of vascular plants was used as an indicator of light availability for bryophytes and was measured, using a LAI-2000 Plant Canopy Analyser (LI-COR Inc., Lincoln, Nebraska, USA), at four points for each plot: one just above the vascular plant canopy and three within the canopy just above the bryophyte layer. LAI measurements are only available for 84 plots due to technical problems. We then clipped vascular plants just above the bryophyte layer and recorded the number of species and total shoot number (ramets rather than shoots were counted for clonal plants and, as these formed the majority of vascular plants, shoot number did not equal genet number). Next, all bryophytes were collected, including their brown parts, and also the litter of vascular plants. Litter was defined as unattached above-ground vascular plant material. A provisional list of bryophyte species was prepared in the field and species identity was subsequently checked in the laboratory after separation of litter and bryophyte material. The dry weight of bryophytes (hereafter ‘bryomass’), vascular plants and litter was recorded after drying for at least 24 h at 70 °C.

After drying, a random subsample (compasing 100–130 shoots) was taken from each bryophyte sample from each plot. Shoot number, including lateral branches of indeterminate growth (offshoots, equivalent to the main shoot) greater than 10 mm, was counted before weighing and extrapolating to the whole sample (to give bryophyte shoot density in each plot).

Peak biomass (excluding the bottom layer of approx. 3 cm) was used as a measure of vascular plant productivity. Although bryomass is the result of accumulation and decomposition of biomass over several years, there is good evidence that it is closely related to bryophyte productivity (Wielgolaski et al. 1981; see also van Tooren et al. 1988). This was confirmed by our analyses of the data of Rieley et al. (1979) and Pande & Singh (1988), both of which resulted in a highly significant positive linear regression of bryophyte production on bryomass (Radj.2 = 87.2%, P < 0.001, n = 14, and Radj.2 = 85.3%, P < 0.001, n = 18, respectively).

STATISTICAL ANALYSIS

We used regression models in which ‘site’ was included as block factor. Plot values were thus adjusted for differences between sites. Plots without LAI values (see above) were omitted from analyses of the effect of vascular plant variables on bryophyte variables, but the whole data set was used otherwise. Variables were log-transformed (indicated in tables) whenever necessary to obtain normally distributed residuals and/or to achieve homoscedastic distributions of points around regression lines. Both linear and quadratic terms were fitted but only linear regressions are presented where the quadratic term was not significant (P > 0.05). Irrespective of significance, the quadratic term is then included to assess whether the widely cited ‘hump-shaped’ relationship applies to bryophytes and vascular plants in this study.

To find the most parsimonious regression explaining variation in bryophyte variables (bryomass, species density or shoot density), we used the forward selection strategy after Payne et al. (1993) with the inratio set to 4 (criterion for including a variable in the multiple regression equation). Since bryomass and log (bryophyte shoot density) were unimodally related to LAI (see Results), the variables LAI and LAI2 were treated as if they were one variable in the forward selection. Data were evaluated for collinearity by the standard procedure provided by GENSTAT 5.0 program, release 3.2 (Payne et al. 1993).

As a guide to the fit of the regression models we used the adjusted R2 statistic after Payne et al. (1993), expressed as a percentage:

Radj.2 = 100 × (1 − (residual mean squares/total mean squares))

All analyses were carried out with the GENSTAT 5.0 program, release 3.2 (Payne et al. 1993).

Results

The bryophyte layer of the studied wetlands was generally well developed, forming a continuous cover with bare soil rarely visible. Bryophyte species density per plot varied from 3 to 18 (mean 8.5 ± 0.26, plot area = 0.04 m2) and bryomass from 0.04 g to 13.2 g (6.3 g ± 0.39). Vascular plant species density (20.3 ± 0.54; range 8–32) was higher than that of bryophytes. Mean biomass was also higher for vascular plants (15.9 g ± 1.27; range 4.6–92.0.g) but in 19 plots, distributed over 12 sites, bryomass exceeded vascular plant biomass.

Shoot densities of bryophytes and vascular plants were both very variable (8–45 946, mean 1421.4 ± 107.1 and 83–579, mean 248.7 ± 10.4, respectively). Litter mass was low, varying from 0.7 to 6.0 g per plot (mean 2.4 g ± 0.10).

There were significant (P ≤ 0.05) or marginally significant (P ≤ 0.1) effects of site in most regressions (Tables 1 and 2), indicating large-scale variability.

Table 1.  Linear or quadratic regressions for the relationships between species density, biomass and shoot density for bryophytes and vascular plants. Percentage variance accounted for by each regression is the adjusted R2 statistic. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001
Dependent variableSource of variationd.f.SSF-ratioRadj.2
(a) Bryophytes
Species densitySite17141.60  2.33** 
 Bryophyte biomass 1146.59 41.02*** 
 [Bryophyte biomass]2 1  0.05  0.01 
 Residuals70250.15 40.9%
Species densitySite17141.60  1.95* 
 Log [Bryophyte shoot density] 1 93.15 21.78*** 
 Residuals71303.65 29.3%
Log [Biomass]Site17  6.09 12.84*** 
 Log [Bryophyte shoot density] 1  9.53341.91*** 
 Residuals71  1.98 85.9%
(b) Vascular plants
Species densitySite171135.69  4.21*** 
 Log [Vascular plant biomass] 1  36.20  2.28 
 (Log [Vascular plant biomass])2 1  40.67  2.56 
 Residuals701109.93 39.2%
Species densitySite171135.69  4.01*** 
 Log [Vascular plant shoot density] 1   3.24  0.19 
 Residuals711183.56 36.1%
Log [Biomass]Site17   1.38  2.39** 
 Log [Vascular plant shoot density] 1   0.25  7.27** 
 Residuals71   2.41 25.1%
Table 2.  Linear or quadratic regressions for the relationships between properties of the vascular plant layer (biomass, shoot density, litter mass and LAI) and bryophyte variables (biomass, shoot density and species density). Percentage variance accounted for by each regression is the adjusted R2 statistic. †P ≤ 0.1, *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001
Dependent variableSource of variationd.f.SSF-ratioR2adj.
(a) Bryophyte biomassSite16318.82 2.41** 
 Log [Vascular plant biomass] 1209.0725.30*** 
 Residuals66545.39 36.1%
Bryophyte biomassSite16318.82 1.88* 
 Vascular plant shoot density 155.89 5.28* 
 Residuals66698.57 18.1%
Bryophyte biomassSite16318.82 2.13* 
 LAI 135.44 3.78 
 LAI2 1109.7911.71*** 
 Residuals65609.23 27.5%
Bryophyte biomassSite16318.82 1.75 
 Litter mass 11.11 1.10 
 Residuals66753.36 11.7%
(b) Bryophyte shoot densitySite1620859232 1.65 
 Log [Vascular plant biomass] 11621475520.53*** 
 Residuals6652133429 26.5%
Log [Bryophyte shoot density]Site164.90 1.78 
 Vascular plant shoot density 10.29 1.71 
 Residuals6611.34 13.7%
Log [Bryophyte shoot density]Site164.90 1.95* 
 LAI 10.15 0.92 
 LAI2 11.26 8.00** 
 Residuals6510.23 21.0%
Log [Bryophyte shoot density]Site164.90 1.74 
 Litter mass 10.01 0.04 
 Residuals6611.62 11.6%
(c) Log [Bryophyte species density]Site160.391 1.76 
 Log [Vascular plant biomass] 10.076 5.50* 
 Residuals660.914 16.8%
Bryophyte species densitySite16137.26 1.59 
 Vascular plant shoot density 119.67 3.65 
 Residuals66355.88 12.7%
Bryophyte species densitySite16137.260 1.51 
 LAI 10.001 0.00 
 LAI2 19.858 1.75 
 Residuals65375.549  8.9%
Bryophyte species densitySite16137.26 1.51 
 Litter mass 10.46 0.08 
 Residuals66375.10  8.0%

There was a positive relationship between species density and biomass (Table 1a, Fig. 1), but it was linear rather than unimodal (quadratic term not significant). In contrast, vascular plant species density was unrelated to biomass (Table 1b, Fig. 1).

Figure 1.

Relationships between species density, biomass and number of shoots for two plant groups (bryophytes and vascular plants). Dependent variables are adjusted for the effects of site (see Methods). Regression lines are only drawn for significant relationships (P ≤ 0.05).

Species density of bryophytes increased with increasing shoot density, even when four outlying plots with low densities (Fig. 1) were omitted, but species density was unrelated to shoot density in vascular plants (Table 1, Fig. 1).

Biomass was positively related to shoot density for both groups (Fig. 1), although the relationship for vascular plants should be interpreted with caution, since biomass was more variable at high densities than low densities. Omitting the three plots with the lowest biomass did not influence the outcome of the regression for bryophytes, which showed a much closer relationship than vascular plants (Radj.2 = 85.9% vs. 25.1%).

The relationship between bryomass and the vascular plant layer depended on the variable considered (Table 2a, Fig. 2a). Bryomass was negatively related to vascular plant biomass and vascular plant shoot density, but weakly unimodally related to LAI and unrelated to litter mass.

Figure 2.

Relationships between bryophyte variables (a, biomass; b, shoot density; and c, bryophyte species density) and properties of the vascular plant layer (vascular plant biomass, vascular plant shoot density, LAI, and litter mass). Dependent variables are adjusted for the effects of sites (see Methods). Regression lines are only drawn for significant (P ≤ 0.05) relationships.

Bryophyte shoot density was related only to vascular plant biomass (linear decrease) and LAI (unimodal, although this must be interpreted with caution, since shoot density was more variable at high and low light availability than at intermediate light availability (Table 2b, Fig. 2b)).

Bryophyte species density also declined with vascular plant biomass (Table 2c, Fig. 2c). Highest and lowest species densities were reached at intermediate light availability (Fig. 2C). Unlike for bryomass and bryophyte shoot density this relationship was therefore not significant.

For both bryomass and bryophyte species density, stepwise multiple regression showed that vascular plant biomass and light availability (LAI + LAI2) were the best explanatory variables, although in the opposite order (Table 3). Both regressions explained only about 40% of the variation of the dependent variables, indicating that there are other major sources of variation. Biomass was the only vascular plant variable significantly contributing to the explanation of bryophyte species density (Table 2c).

Table 3.  Best multiple regressions for the relationships between properties of the vascular plant layer and bryophyte biomass or shoot density, respectively. For bryophyte species density the most parsimonious regression was on site and log [vascular plant biomass]. Results of this regression are already presented in Table 2c. Percentage variance accounted for by each regression is the adjusted R2 statistic. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001
Dependent variableSource of variationd.f.SSF-ratioRadj.2
(a) Bryophyte biomassSite16318.82 2.72** 
 Log [Vascular plant biomass] 1209.0728.49*** 
 LAI 10.17 0.02 
 LAI2 175.5510.29** 
 Residuals64469.67 43.2%
(b) Bryophyte shoot densitySite1620859232 2.03* 
 LAI 1728422211.34*** 
 LAI2 11199615318.68*** 
 Log [Vascular plant biomass] 1796096812.39*** 
 Residuals6441106841 40.2%

Discussion

THE RELATIONSHIP BETWEEN SPECIES DENSITY, BIOMASS AND SHOOT DENSITY FOR BRYOPHYTES AND VASCULAR PLANTS

The two taxonomic groups showed different biomass-species density relationships: there was a positive linear relationship for bryophytes but no significant correlation in vascular plants. Neither showed the expected unimodal relationship between biomass and species density as predicted by Grime (1973), Tilman & Pacala (1993) and Huston (1994).

No comparable data are available for wetland bryophytes. Virtanen et al. (2001), studying the relationship between biomass and species density in rheophytic bryophyte assemblages, found three different patterns: a positive linear relationship in two streams (out of 14), a unimodal relationship in five streams, and no relation at all in seven streams. Unimodal relationships occurred only in streams with largely different microhabitats that allowed for small-scale community diversification.

Our results for vascular plants were consistent with the findings of some previous wetland studies (e.g. Vermeer & Verhoeven 1987; Moore & Keddy 1989), although others have detected a humped relationship in such habitats (e.g. Wheeler & Giller 1982; Moore et al. 1989; Garcia et al. 1993).

A comprehensive review by Waide et al. (1999) showed that only 30% of the studies investigating the biomass-species density relationship revealed a unimodal pattern and there is growing evidence that this depends upon the spatial scale that is studied (Moore & Keddy 1989; Waide et al. 1999; Weiher 1999; Gross et al. 2000; Virtanen et al. 2001). When studies within plant communities are considered, only 24% reveal a unimodal pattern, with 42% showing no relationship and 22% a positive one (Waide et al. 1999). The limited range of biomass values within a community may be too narrow to demonstrate an underlying unimodal relationship (Rosenzweig & Abramsky 1993; Grace 1999), consistent here with the two plots with the highest vascular plant biomass showing low species density but not with the bryophyte data (highest biomass in plots with highest number of species, Fig. 1). The influence of other variables (e.g. successional age or edaphic conditions of sites) may also affect productivity and thus the species density-biomass relationship (Gough et al. 2000; Loreau 2000).

Favourability gradients

The denser bryophytes grow, the more biomass they produce in total and the more species live together in small plots (see Fig. 1). Økland (1994) was the first to talk about such bryophyte ‘favourability gradients’ whereby the water content of the shoots rises as their density increases (Proctor 1982) and bryophytes are therefore able to remain photosynthetically active for a greater part of the growing season and show greater biomass production (Bates 1988). This positive density-dependent relationship appears to be a common phenomenon in bryophyte assemblages (e.g. Økland & Økland 1996; Økland 2000) and, in combination with high clonal fragmentation and the absence of competitive hierarchies (at least in comparable bryophyte communities in chalk grasslands, During & van Tooren 1988), this is probably responsible for the parallel increases in biomass, shoot density and species density.

In contrast, vascular plants showed no such favourability gradient. Although denser growing shoots produced more biomass, species density was not related to biomass or shoot density. Tilman & Pacala (1993) have explained decreases in vascular plant species density at high levels of productivity as a response to light limitation. Under these conditions, slow-growing species are outcompeted by a few highly competitive fast-growing species: according to an extension of the self-thinning law to thinning in mixed communities, larger species replace smaller species under more productive conditions (Bazzaz & Harper 1976; Schmid 1991; but see Stevens & Carson 1999b). Although this may be true for some vascular plant communities, it seems unlikely for bryophytes, whose shoot size is much more restricted by morphological and physiological constraints (cf. the tall herb Angelica sylvestris vs. the small grass Festuca rubra). Thus, under more productive conditions, the increase of bryomass per plot is likely to be mainly an effect of increased shoot density, leading to the very close relationship between these variables in bryophytes but not in vascular plants.

Effects of clonality

The domination of the vegetation in these wetlands by clonal plants (personal observation) may have important consequences for the relationship between species density and biomass. Specifically, the probability of two neighbouring shoots belonging to the same species is much higher than in communities dominated by non-clonal species, where species density and shoot density are often found to be closely related (Condit et al. 1996; Stevens & Carson 1999a). The lack of a relationship for vascular plants was therefore as expected.

Most bryophytes exhibit clonal growth patterns (During 1990) and experimental evidence is growing that physiological integration of ramets in ectohydric bryophytes may reach levels comparable with those of clonal vascular plants (Alpert 1989; Økland et al. 1997; Eckstein & Karlsson 1999). There was, nevertheless, a close relationship between bryophyte species density and shoot density, suggesting dominance of non-clonal plants (cf. Stevens & Carson 1999a). Light limitation in dense stands and therefore browning of shoots in the lower strata of the bryophyte canopy (van der Hoeven & During 1997), may lead to earlier physical disintegration of connections between ramets and parent plants in bryophytes than in many clonal vascular plants, such as tussock-forming grasses and sedges, so that physiological integration is unlikely to regulate bryophyte shoot density. Even in vascular plants, whether regulation of shoot density operates via physiological integration or via external factors, is still a matter of debate (Suzuki & Hutchings 1997; Meyer & Schmid 1999).

THE RELATIONSHIP BETWEEN THE BRYOPHYTE AND THE VASCULAR PLANT LAYER

Of all the measured variables of the vascular plant layer, biomass was the best single predictor of each of the three, intercorrelated bryophyte variables, with bryophyte favourability decreasing with increasing vascular plant biomass. There was a unimodal relationship between bryophyte favourability and light availability, most likely caused by a combination of optimal radiation and moisture at intermediate light levels, which leads to longer periods of photosynthetic activity for the ecto- and poikilohydric bryophytes and, thus, to their higher growth rates (Callaghan et al. 1978; Bates 1988; Økland & Økland 1996; Økland 2000). However, scatter in our data was high, indicating that important sources of variation were not included in our restricted set of explanatory variables. Marrs et al. (1996) proposed that such variable data sets should be characterized by boundary conditions rather than by regression analyses that obscure valuable information. This would include the observation that values of bryophyte biomass, shoot density and species density were particularly unpredictable at intermediate light levels, although favourability was consistently low at both ends of the gradient.

Litter mass of vascular plants accounted for only a small part of the bryophyte variation, possibly due to the low levels maintained by yearly mowing.

Although biomass is a rather crude descriptor of the structure of the vascular plant layer (Spehn et al. 2000) and thus of the environment experienced by bryophytes (Watson 1960; Sveinbjörnsson & Oechel 1992), the negative relationship is important for bryophyte conservation: stands of low vascular plant production are richer in bryophytes. Since vascular plant growth in fens is mainly controlled by the availability of phosphorus and nitrogen (Verhoeven & Schmitz 1991; Pauli 1998), it is crucial to curb supply of these nutrients in order to maintain bryophyte diversity.

Acknowledgements

We are grateful to the nature conservancy authorities of the cantons of Schwyz, Glarus, St Gallen and Appenzell-Ausserrhoden and to the land owners of the investigated sites who allowed us to conduct this study. We thank E. Urmi, P. Alpert, R. Økland and one anonymous referee for various comments which improved earlier versions of the manuscript considerably, M. Naegeli for helping with the very time-consuming separation of vascular plant litter and bryophytes, and M. Matthews for linguistic aid.

Ancillary