Comparative Cryptogam Ecology: A Review of Bryophyte and Lichen Traits that Drive Biogeochemistry

Abstract

Background

Recent decades have seen a major surge in the study of interspecific variation in functional traits in comparative plant ecology, as a tool to understanding and predicting ecosystem functions and their responses to environmental change. However, this research has been biased almost exclusively towards vascular plants. Very little is known about the role and applicability of functional traits of non-vascular cryptogams, particularly bryophytes and lichens, with respect to biogeochemical cycling. Yet these organisms are paramount determinants of biogeochemistry in several biomes, particularly cold biomes and tropical rainforests, where they: (1) contribute substantially to above-ground biomass (lichens, bryophytes); (2) host nitrogen-fixing bacteria, providing major soil N input (lichens, bryophytes); (3) control soil chemistry and nutrition through the accumulation of recalcitrant polyphenols (bryophytes) and through their control over soil and vegetation hydrology and temperatures; (4) both promote erosion (rock weathering by lichens) and prevent it (biological crusts in deserts); (5) provide a staple food to mammals such as reindeer (lichens) and arthropodes, with important feedbacks to soils and biota; and (6) both facilitate and compete with vascular plants.

Approach

Here we review current knowledge about interspecific variation in cryptogam traits with respect to biogeochemical cycling and discuss to what extent traits and measuring protocols needed for bryophytes and lichens correspond with those applied to vascular plants. We also propose and discuss several new or recently introduced traits that may help us understand and predict the control of cryptogams over several aspects of the biogeochemistry of ecosystems.

Conclusions

Whilst many methodological challenges lie ahead, comparative cryptogam ecology has the potential to meet some of the important challenges of understanding and predicting the biogeochemical and climate consequences of large-scale environmental changes driving shifts in the cryptogam components of vegetation composition.

INTRODUCTION

The quantification of interspecific variation in functional traits has been a powerful tool in comparative plant ecology in recent decades. It has enabled us to detect and test important trade-offs and relationships in plant design and function at regional and global scales (e.g. Grime et al., 1997; Craine et al., 2001; Díaz et al., 2004; Wright et al., 2004). It is now aiding the scaling from sets of species or traits to ecosystem properties and functions and to ecosystem responses to environmental change (e.g. MacGillivray et al., 1995; McIntyre et al., 1999; Eviner and Chapin, 2003; Garnier et al., 2004). Emphasis is gradually and partly shifting from traits that predict or control the functioning of the plants themselves, i.e. ‘response traits’, to traits that relate to the effects that plants impose on their environment, i.e. ‘effect traits’ sensu Lavorel and Garnier (2002; see also Cornelissen et al., 2003 and Table 1). However, these substantial advances in trait research have been founded almost exclusively on vascular plants, having all but ignored autotrophic non-vascular cryptogams such as bryophytes and lichens. This may be due to (historically grown) unfamiliarity of the comparative plant ecologist community with cryptogams, with taxonomic identification problems and methodological hurdles such as cultivation difficulties (but see Duckett et al., 2004). And yet these organisms are abundant in diverse biomes worldwide, ranging from very cold to very warm and from very dry to very wet (e.g. Seaward, 1977; Werger and During, 1989; Longton, 1997; Bates, 2000; Tan and Pócs, 2000; Belnap and Lange, 2001; Gradstein et al., 2001). They are particularly important determinants of ecosystem functioning in peatlands throughout the world (Gorham, 1991), in several other ecosystems in cold biomes (Longton, 1997), as epiphytes or epiphylls in in tropical and temperate ecosystems (e.g. Rieley et al., 1979; Pócs, 1980; Rogers, 1989; Knops et al., 1996; Clark et al., 1998; Gradstein et al., 2001) and as components of biological crusts in arid and semiarid biomes worldwide (Belnap and Lange, 2001). In these biomes, for instance (1) they contribute substantially to above-ground productive biomass (Wielgolaski et al., 1975; Veneklaas et al., 1990; Wolf, 1993; Lange et al., 1998; Belnap and Lange, 2001); (2) they may host nitrogen-fixing bacteria, providing major soil N input (lichens: Forman, 1975; Crittenden, 1983; Nash, 1996; Kurina and Vitousek, 2001; bryophytes: During and Van Tooren, 1990; Solheim et al., 1996) or (in the case of Sphagnum) host methanotrophic bacteria aiding carbon supply for photosynthesis (Raghoebarsing et al., 2005); (3) they control soil and vegetation hydrology and temperatures (Pócs, 1980; Clymo and Hayward, 1982; Veneklaas et al., 1990; Beringer et al., 2001; Van der Wal and Brooker, 2004); (4) they control soil chemistry and nutrition through the accumulation of recalcitrant polyphenols (Clymo and Hayward, 1982; Verhoeven and Toth, 1995) or through their effects on extracellular enzymes involved in nutrient mineralization (Sedia and Ehrenfeld, 2006); (5) they help form biological crusts on bare, dry substrates thus preventing soil erosion (Belnap and Lange, 2001; Kürschner, 2004); (6) they (lichens) are a potentially important agent of rock weathering, mobilizing minerals into ecosystems (Adamo and Violante, 2000; Landeweert et al., 2001); (7) they provide a staple food to mammals such as reindeer or caribou (lichens: Helle and Aspi, 1983; Danell et al., 1994) and to arthropodes (Hesbacher et al., 1995; Hodkinson et al., 1996), with important feedbacks to soils and biota (Van der Wal and Brooker, 2004); and (8) they both facilitate and inhibit or compete with vascular plants (Rundel, 1978; Keizer et al., 1985; During and Van Tooren, 1990; Equihua and Usher, 1993; Malmer et al., 1994; Van Breemen, 1995; Callaway et al., 2001; Sedia and Ehrenfeld, 2005; Dorrepaal et al., 2006; Gough, 2006), and can thereby affect vegetation succession (Sedia and Ehrenfeld, 2003).

Table 1.

Details for a selection of key traits discussed with respect to their link with biogeochemical cycling. See the main text for further details

	Response trait	Effect trait	Literature data available?	New protocol needed?	Methodological remarks
Tissue [N] and [P]	Yes	Yes	Yes	Adjustments to Cornelissen et al. (2003)	Use entire bryophyte gametophytes or lichens without the rhizoids or rhizines
Secondary metabolites	Yes	Yes	Mostly qualitative	No	Group by function; metabolomic approach is promising
Oil bodies	Yes?	?	Qualitative data only	Yes, for quantitative data	Only applies to liverworts
Tissue pH	?	Yes	No	No	See Cornelissen et al. (2006)
Acidifying potential		Yes	No	No, but adjustments needed	See Kooijman and Bakker (1994) and text
Allelopathic/nutrient immobilization capacity		Yes	Few	Yes	Bio-assays (see text)
N2 fixing capacity	Yes		Few	No	See Solheim et al. (1996), Crittenden (1973); selection and calibration among methods needed
Organic N uptake potential	Yes		Few	No	See Dahlman et al. (2004), Forsum et al. (2006)
Infection by mycorrhizal fungi	Yes		Few	No	See references in Read et al. (2000)
Inherent relative growth rate	Yes	?	Few	No, but calibration of data for lichens versus bryophytes is needed	See Furness and Grime (1982) for bryophytes, but see size effects; selection and calibration of many methods for lichens needed (see text)
Tissue dry matter content	Yes	Yes	No?	Adaptation from Garnier et al. (2001)	Water storage in non-photosynthetic cells needs to be addressed
Fluorescence	Yes		Few	No	Calibration of fluorescence data with actual gas exchange measurements at different light, temperature and moisture conditions may be needed for selected species
Resistance to herbivores	Yes	?	Few	No, but standardization and comparison between herbivore preferences	Bio-assays, e.g. ‘cafeteria’ experiments (see text)
Litter decomposability		Yes	Few	Yes	Multispecies simultaneous screening using litterbags in outdoor incubation beds may be promising (see Cornelissen, 1996)

Why bother with comparative cryptogam ecology?

Advances in comparative cryptogam ecology would be particularly timely, since bryophytes and lichens are thought to be facing great changes in their abundance, biomass and composition as a consequence of global environmental changes (Bates and Farmer, 1992; Callaghan et al., 2004). For instance, Sphagnum performance and Sphagnum-rich peatlands in general may be affected by climate warming (Gorham, 1991; Hobbie et al., 1999; Dorrepaal et al., 2004). Lichens and probably also non-Sphagnum bryophytes may decline in response to vascular plant expansion as induced by climate warming in some of the cold biomes (Hobbie et al., 1999; Cornelissen et al., 2001; Van Wijk et al., 2004). Climate warming may indirectly increase soil N and/or P availability through faster soil nutrient mineralization, while anthropogenic atmospheric N inputs are already having similar, albeit more drastic, effects on cryptogams. Lichens from nutrient-limited ecosystems tend to decline in response to increasing N and/or P availability (e.g. see review by Cornelissen et al., 2001), while some other species have been found to benefit (Gaio-Oliveira et al., 2004; Palmqvist and Dahlman, 2006). For bryophytes the response appears to depend on the species or functional group (Bates, 2000; Gordon et al., 2001; Van Wijk et al., 2003). Furthermore, bryophytes are sensitive to climate-induced changes in ecosystem hydrology (Bubier, 1995; Weltzin et al., 2001; Heijmans et al., 2004; Bates et al., 2005). Given the above key controls of cryptogams over ecosystem functions, important feedbacks to large-scale carbon and nutrient cycling, climate and hydrology can be anticipated to accompany climate-induced changes in cryptogam communities (Chapin et al., 2000; Beringer et al., 2001). However, there is still much uncertainty about the direction and magnitude of change for the various functional groups of cryptogams. Interspecific trait research will help to formulate and test predictions about such changes and their feedbacks. In this review we focus on traits linked to biogeochemical cycling in the stricter sense. Although ecosystem hydrology, irradiance and climate have strong impacts on biogeochemistry (see Fig. 1, left section), we shall exclude from our discussion those traits that are principally linked to the latter factors rather than directly to biogeochemistry. These exclusions involve many of the morphological traits that have been well documented for bryophytes and lichens, often as identification aids in taxonomic studies. We also exclude from our discussion morphological and life history traits related to vegetative and sexual reproduction, dispersal and diaspore survival in the soil (Fig. 1, bottom), which tend to have more indirect relationships with biogeochemistry, or relationships on longer-term timescales incorporating environmental disturbances and succession.

Fig. 1.

Conceptual diagram showing the relationships (arrows) between a selection of ‘soft traits’ (in italics) and ‘hard traits’ (lower case) of bryophytes and lichens and the ecosystem properties or functions (capitals) that they predict or affect. The traits pointing towards ecosystem properties or functions are effect traits, with the exception of those (response traits) pointing to the ‘regional species pool’ and ‘ecosystem disturbance response’. There are many more relationships than the ones shown here, including relationships among ecosystem functions and properties, but the arrows indicate some key ones of particular interest for this paper. This paper focuses on the right hand section of the diagram, although many indirect and long-term linkages also exist with traits and functions in the left and bottom sections (see text). See the main text for details and further biogeochemistry-related traits.

We arbitrarily distinguish between ‘soft traits’ and ‘hard traits’ (Hodgson et al., 1999), even though these are on a sliding scale. Soft traits are easy and inexpensive to measure for large numbers of plants and samples and at the same time have reasonable predictive power of other, hard plant traits or even of important ecosystem processes and responses themselves (Lavorel and Garnier, 2002; Cornelissen et al., 2003; see also Fig. 1).

It is important to note that, even for some of the traits for which ample data are available, these are often qualitative rather than quantitative (e.g. secondary metabolites, colours) and their relationships to ecosystem functions have been poorly tested, if at all. While not ignoring response traits, we shall give particular emphasis to effect traits and their relationships with ecosystem properties and functions (Fig. 1), as well as their possible impacts on vascular plant performance. It is clear that these fundamentally different organisms require partly different traits and methods than vascular plants. Thus, we shall discuss both some of the traits commonly used in vascular plant studies and novel traits (to be) tailor-made for cryptogam functioning and impacts. We shall give some pointers to some of the important theoretical issues and methodological problems (see Table 1) that need to be addressed. The most obvious methodological issue is that the performance of many cryptogam species, particularly bryophytes, is highly density-dependent, intraspecific facilitation being the norm rather than the exception (Van der Hoeven and During, 1997; Mulder et al., 2001; Pedersen et al., 2001; Rixen and Mulder, 2005). Photosynthesis and other metabolic processes of these poikilohydric organisms depend strongly on their hydration status, which in many species is a positive function of the densities of their in situ turfs. Thus, some cross-species cryptogam trait screening tests may have to be performed on turfs with standardized plant or thallus densities, or densities representative of natural intraspecific aggregation, rather than on isolated plants or thalli. This would be an essential difference with screening tests on vascular plants, for which isolated plants are generally used in laboratory tests, and free-standing plants where possible for trait measurements on plants in the field.

THE TRAITS

Tissue chemistry traits: nutrients, secondary metabolites, oil bodies, tissue pH

Among the traits that have been studied relatively well for cryptogams are those involving tissue chemistry, notably concentrations of the mineral nutrients nitrogen (N) and phosphorus (P), and the quality and (partly) the quantity of secondary metabolites. Tissue N, most of which is a component of the photosynthesis enzyme Rubisco, and to some extent P, involved in energy-demanding chemical transformations, are consistent predictors of interspecific variation in photosynthetic capacity and relative growth rate (RGR), as has been demonstrated comprehensively for wide-ranging vascular plants (Field and Mooney, 1986; Lambers and Poorter, 1992; Cornelissen et al., 1997; Wright et al., 2004). One would expect that tissue [N] and perhaps also [P] will have some predictive power of interspecific variation in photosynthetic capacity and RGR of cryptogams, too. However, in a large set of lichen species, including antarctic, boreal, temperate and subtropical members, Palmqvist et al. (2002) demonstrated that the best predictor of interspecific variation in photosynthetic capacity was chlorophyll a concentration, followed by [N] locked up in chlorophyll a (as opposed to in Rubisco). Total thallus [N] played a significant but smaller predictive role in their dataset. Interestingly, lichens with green algal photobionts on average had lower photosynthetic capacities than lichens with cyanobacterial photobionts, with tripartite lichens assuming an intermediate position in this respect.

Tissue nutrients also directly and indirectly affect important non-productive components of the ecosystem carbon budget, notably herbivory and decomposition (Fig. 1; see also below).

While there are many data for tissue N and P of smaller species sets of bryophytes and lichens in the literature (e.g. Chapin and Shaver, 1988 for bryophytes; Kershaw, 1985 and Palmqvist et al., 2002 for lichens), the challenge is to assemble these and calibrate them to correct for differences in methodology, seasonal variability and soil nutrient availability. For supplementary new N and P analyses, Cornelissen et al. (2003) provide a protocol that can also be applied to cryptogams, although whole photosynthetically active shoots or thalli without rhizoids or rhizines should be used instead of leaves.

Both lichens and bryophytes tend to invest strongly in secondary metabolites, and many of the compounds act as chemical, and in some cases structural, defences against herbivores or pathogens (lichens: Rundel, 1978; Lawrey, 1980; Gauslaa, 2005; bryophytes: Clymo and Hayward, 1982; Frahm, 2004; Glime, 2006), or as chemical screens to minimize damage of tissues and DNA due to UV-B radiation (Rozema et al., 2002; Arróniz-Crespo et al., 2004; Nybakken et al., 2004). Compared to screening for nutrient contents, it will be more difficult to assemble, process and apply the huge body of information available on the many different secondary metabolites detected in bryophytes and lichens, many of these metabolites being taxonomically very specific (e.g. Hegnauer, 1962; Markham and Porter, 1978; Zinsmeister and Mues, 1990; Elix, 1996; Mues, 2000; Asakawa, 2004). While one step is to group these compounds by molecular type and function (e.g. Fahselt, 1994 for lichens), it will be a challenge to deal with the common lack of quantitative data and the unknown functions and effects of many of these compounds. We shall probably have to apply shortcuts to assess the potential effects of the spectrum of secondary chemistry of each species in a larger species set. One way forward may be to screen different species relatively rapidly for their overall biochemical profiles or ‘metabolic fingerprints’, e.g. by applying Fourier transform-infrared spectroscopy or nuclear magnetic resonance spectroscopy (Dunn and Ellis, 2005; Krishnan et al., 2005).

Another way could be to add standard amounts of tissue extract to the growth medium of standard test organisms in multispecies cryptogam bio-assays, for instance to the small aquatic crustaceans Artemia, and measure their mortality (cf. Almeida-Cortez et al., 1999 for vascular plants). Glime (2006) applied a similar approach to test for phenolic defences of different populations of the aquatic moss Fontinalis antipyretrica, by comparing consumption by an aquatic isopod (Asellus militaris). Similarly, Davidson et al. (1990) compared the acceptability of three different mosses by herbivorous slugs experimentally. Surprisingly, while the gametophytes were either not or were hardly eaten by the slugs, the animals did eat immature sporophytes, probably because the sporophytes hardly contained any phenolic acids in contrast to the gametophytes. Gardner and Mueller (1981) added a range of eight different lichen acids to laboratory cultures of the moss Funaria hygrometrica at different concentrations and at a range of pH values. They reported a wide range of effects on spore germination and sporeling growth. These effects were predominantly inhibitory while a few were stimulatory. Such data are potentially useful, but still need to be translated to real concentrations of different lichen acids in different lichen species. The latter point was addressed in a study on field herbivory by the slug Pallifera varia (Lawrey, 1980, 1983). Lichen acids in the gut content and faeces of field-collected individuals of this lichen-feeding slug were compared with lichen acids measured in the main lichen species in the slug's habitat. A clear interspecific palatability ranking was found, with some lichen species (e.g. Xanthopermelia cumberlandia, Huilia albocaerulescens) being consistently avoided while at the other end the crustose lichen Aspicilia gibbosa was fed upon preferentially. Garmo (1986) screened different lichen species for chemistry and digestibility in the reindeer gut, which has a gut flora particularly efficient at lichen digestion. Using a particularly promising novel assay, Solhaug and Gauslaa (2001) and Gauslaa (2005) managed to extract C-rich secondary compounds non-destructively from a range of lichen species by means of acetone-scrubbing. Scrubbed lichen species belonging to Parmeliaceae, which are rich in C-based defences, became significantly more palatable to the herbivorous snail Cepaea hortensis when compared to non-scrubbed controls of the same species (Gauslaa, 2005). Lichens belonging to Physciaceae and Teloschistales, which are low in C-rich secondary chemistry, did not show such a response to acetone-scrubbing.

It is obvious from the above examples that many methodological issues need to be addressed in the bioassay approach, for instance whether to compare the cryptogams themselves or chemical extracts from them (Gauslaa, 2005); which extraction methods to use for which group of compounds; and the representativeness of the various possible test organisms. Perhaps results for a range of contrasting test organisms need to be compared in order to establish meaningful rankings of cryptogam species according to chemical defences, and their repercussions for the allelopathic and nutrient immobilization potentials of different cryptogam species.

A third way to test for cryptogam defences might be to apply soft traits with sufficient predictive power of overall secondary metabolite composition and abundance. In this respect, the oil bodies of liverworts may bear some promise. Oil bodies are true membrane-bound organelles that occur in the great majority of liverwort species but not in other bryophytes (Crandall-Stotler and Stotler, 2000). They contain high concentrations of a diversity of terpenoids and aromatic compounds such as phenolics, which have been linked tentatively to protection from herbivores, pathogens, cold and/or UV-B radiation (Crandall-Stotler and Stotler, 2000; Asakawa, 2004). Thus, measurements of oil body volume or projected area per unit volume, or area of the liverwort gametophyte could reveal useful information about overall investments in secondary metabolites, with implications for resistance of the species to environmental hazards. On the downside, oil bodies have to be measured in fresh plants, since they tend to disintegrate rapidly in dried specimens. Also, ontogeny, seasonality and sex of the gametophyte (male, female, sterile) may all affect oil body make-up (Asakawa, 2004), so standardized and representative sampling is required. Furthermore, it needs to be tested to what extent differentiation into the morphological types of oil bodies as defined by Crandall-Stotler and Stotler (2000) corresponds with differentiation in their chemical content.

Tissue pH has recently been shown to be an interesting new soft trait for use in interspecific comparisons (Cornelissen et al., 2006) since it integrates important information, particularly about the secondary chemistry (e.g. organic acids) and basic cation concentrations. Cornelissen et al. (2006) demonstrated for a wide range of subarctic vascular species that there was consistent variation in foliar tissue pH between functional types; foliar pH was a promising predictor of tissue digestibility, and litter pH was a predictor of litter decomposability across this species set (see Fig. 1). This easily measured species trait may also have important predictive power for biogeochemical processes in bryophytes. Indeed, Sphagnum (peat moss) is a spectacular example of a bryophyte that, through its constitutive chemistry, both acidifies its direct environment and inhibits decomposition and mineralization (Clymo and Hayward, 1982; Glime et al., 1982; Gagnon and Glime, 1992; Kooijman and Bakker, 1994; Vitt, 2000; see Fig. 1). Such Sphagnum-driven acidification and nutrient immobilization are paramount drivers of (vascular) plant community change and ecosystem functioning. There is also some evidence that non-Sphagnum bryophytes have some acidifying potential (Glime et al., 1982), whilst Scorpidium scorpioides from a wetland in the Netherlands increased the pH of simulated rainwater (Kooijman and Bakker, 1994). However, we still know very little about interspecific variation of tissue pH or acidifying potential, about the correspondence between these two parameters, or about the way and degree to which such interspecific variation is modified by the effects of various environmental factors on pH or acidifying potential of given species. For instance, both the pH and the acidifying potential may vary with the cation availability in the soil solution and the extent to which these cations have replaced H⁺ ions on the exchange sites in bryophyte cell walls. We therefore advocate standardized screening tests of tissue pH, as well as acidifying capacity itself, on wide ranges of bryophyte species and perhaps, by way of exploration, lichens. We are currently attempting such screening on multiple subarctic and alpine species, with promising preliminary results (Soudzilovskaia et al., unpubl. res.).

Free bivalent nitrogen fixing capacity

Nitrogen is a principal limiting environmental factor to many bryophytes and lichens, which tend to be at their most abundant in less productive sites where they experience less competition from vascular plants (e.g. Crittenden, 1983; Longton, 1988). Both bryophytes and lichens feature several adaptations to optimize their N uptake capacity. The best known adaptation is the symbiotic obligate relationship of many lichen species with Cyanobacteria (e.g. Nostoc, Anabaena), where the latter fix atmospheric N₂. Nitrogen inputs via this pathway can be important to a range of ecosystems (see Kershaw, 1985 and Nash, 1996 for reviews, and Matzek and Vitousek, 2003). Bryophytes of wide-ranging taxa from wide-ranging ecosystems have been found to host Cyanobacteria (Henriksson et al., 1987; review by During and Van Tooren, 1990; Matzek and Vitousek, 2003), although the degree of host specificity varies greatly. In the liverwort Blasia and the hornwort Anthoceros these associations are consistent and internal (see also Duckett et al., 2004) and are probably critical in the nitrogen economy of the liverworts. In contrast, many different moss species host and protect N₂ fixing Cyanobacteria, particularly in sheltered spots between or on their leaves (e.g. Dalton and Chatfield, 1987). However, these associations appear facultative to different degrees, but particularly so in Sphagnum, while host specificity is also not clear-cut (During and Van Tooren, 1990). Moreover, it is uncertain to what degree these associations result in a greater amount of N₂ fixed per unit ground area compared to free-living Cyanobacteria, making it difficult to interpret the value of this association as an effect trait. In several cases, though, bryophyte–cyanobacterial associations appear to be rather host-specific and abundant, and contribute importantly to ecosystem N inputs, at least in cold, northern biomes (Solheim et al., 1996; Zackrisson et al., 2004). This is probably also true for the important peatland moss Sphagnum, which may host both photosynthetic and heterotrophic N₂-fixing bacteria (During and Van Tooren, 1990). More hard and standardized data on N₂-fixing capacity as a trait are needed for lichens, but even more so for bryophytes. While good field and laboratory methods are available from previous studies, the challenge is to calibrate data among these studies. For new studies we tentatively propose that measurements should be made of maximum rates (mg fixed N m⁻² d⁻¹) for monospecific patches, measured either in the field or in the lab, irrespective of the environmental conditions under which the maximum rates are found. This will not be easy, given the reported strong depency of fixation rates on environmental factors, particularly hydrology, temperature and light, at least in a high-arctic study (Zielke et al., 2002). These and perhaps further factors may also explain the relation between fixation capacity and the successional phase of boreal forests (Zackrisson et al., 2004). A potential soft trait that may to some extent predict N₂ fixing capacity is the percentage of heterocysts among cyanobacterial cells as seen in microscopic thallus cross-sections (Kershaw, 1985), since heterocysts are the actual sites of N₂ fixation.

Other cryptogam traits related to nutrient acquisition capacity

Several further traits have emerged recently as potential components of the nutrient acquisition strategies of cryptogams in ecosystems where inorganic mineral nutrients are sparsely available.

First, cryptogam species may differ in their affinity for ammonium versus nitrate uptake (e.g. Bates, 2000; Paulissen et al., 2004 for bryophytes; Dahlman et al., 2002 for lichens) and ammonium can even be toxic to certain nitrate-specialized bryophytes (Paulissen et al., 2004). Such affinities may serve as adaptations to the predominant forms in which the two anions are available in their natural habitats (e.g. Forsum et al., 2006). Generally, acidic soils tend to be ammonium donors, while nitrate tends to be the predominant inorganic N form in higher-pH soils. Nitrate is more mobile in the soil and therefore easier to take up by plants, while ammonium is usually bound to a cation exchange complex. However, ammonium does not need to be reduced within the plant in order to be converted to amino acids, whereas nitrate has to be reduced first by the enzyme nitrate reductase, which requires energy (Schlesinger, 1997). Ammonium may enter the ecosystem also via wet, occult and dry deposition. Epiphytic lichens may convert a significant part of this ammonium into nitrate or organic nitrogen (Oyarzún et al., 2004). Multispecies data on preferential uptake of ammonium versus nitrate would provide important information about the role of different bryophytes, and perhaps also lichens (Dahlman et al., 2002), in ecosystems varying in soil fertility and pH. Such preferences may be screened for experimentally by supplying different species with nitrate versus ammonium and measuring their growth (Paulissen et al., 2004); however, this is labour-intensive. The method by which nutrients are supplied needs to be carefully considered and standardized, in view of the abscence of true roots for uptake in cryptogams.

Second, Kielland (1997) revealed substantial potential (laboratory) uptake of N as amino acids both by the peat moss Sphagnum rubellum and by the lichen Cetraria richardsonii, both arctic species from tundra soils where amino acids are more abundant than ammonium or nitrate in the soil solution. However, Kielland tracked amino acids from which only N was labelled as ¹⁵N, leaving the possibility that part of these amino acids were first mineralized before uptake. Given the known importance of this N uptake pathway in a wide range of vascular plant species and types, we recommend systematically screening cryptogam species for this trait as well. The protocol followed by Kielland (1997) provides a good starting point for this. Double-labelling, i.e with both ¹⁵N and ¹³C, is a preferable method, since the ratio between the two isotopes in the tissues gives more unambiguous information about how much of the uptake is due to unmineralized amino acids. Dahlman et al. (2004) recently applied double labelling to screen 14 northern lichen species for their capacity for uptake of a range of amino acids. They demonstrated that these species generally had a substantial amino acid uptake potential (albeit lower than their ammonium uptake potential), which probably contributes to their N nutrition in the field. Recently, the northern moss Hylocomium splendens has been shown to take up significant amounts of amino acids from throughfall precipitation in situ in the forest (Forsum et al., 2006). We are currently screening for amino acid uptake capacity of a range of northern bryophyte species using a similar double-labelling method, with promising preliminary results (E.J. Krab et al., Institute of Ecological Science, Vrije Universiteit, The Netherlands, unpubl. res.).

Third, During and Van Tooren (1990) and Read et al. (2000) have reviewed multiple evidence that endophytic fungi are found inside the gametophytes of members of several families of liverworts and in some hornworts, and speculated that some of these fungi may actually be symbiotic, i.e. the fungi facilitating phosphorus (and/or nitrogen) uptake by the bryophyte and extracting photosynthates from it in exchange. Read et al. (2000) deduced this from the fact that the same fungal taxa are known to form symbiotic mycorrhizal associations with the roots of vascular plants. Turnau et al. (1999) found some evidence that the thallose liverworts Conocephalum conicum and Pellia endiviifolia used neighbouring vascular plants as an inoculum or transport medium for the arbuscular mycorrhizal fungi that these liverworts associate with. Fungi involved in ericoid, ecto-mycorrhizal, vesicular-arbuscular and orchid mycorrhizae with vascular plants have been detected in certain liverwort or hornwort taxa as well (During and Van Tooren, 1990; Read et al., 2000). If further investigation were to reveal experimentally that mycorrhizal fungi do indeed aid P and/or N uptake of these bryophytes to a substantial degree (and if so, how to distinguish the symbiotic ones from parasitic ones), they could make a promising contribution to our knowledge on biogeochemistry.

Fourth, the fungal symbionts of lichens have been shown to promote chemical weathering of rock, and thereby to mobilize nutrients, principally through their release of lichen acids (Adamo and Violante, 2000). Given the ubiquitous occurrence of lichen–rock associations worldwide, experimentally screening different rock-dwelling lichens for lichen acid chemistry, and the effectiveness of the latter in chemical weathering, could enhance our predictive power of lichen impacts on biogeochemistry. Since the products of rock weathering serve many other organisms as well as lichens over longer time scales, rock weathering capacity may be interpreted principally as an effect trait.

Fifth, Bates (1994) demonstrated that bryophytes also differ in the efficiency of both the uptake and conservation of inorganic mineral nutrients. He grew the (presumably non-mycorrhizal – see above) mosses Pseudoscleropodium purum and Brachythecium rutabulum both under nutrient-deficient conditions and after an experimental pulse of NPK fertilizer. Pseudoscleropodium purum, which in nature depends on periodic nutrient inputs from atmospheric sources, exhibited efficient nutrient uptake and conservation, as well as faster growth in response to a NPK pulse. In contrast, B. rutabulum did not respond to such a nutrient pulse. This was linked to its more reliable access to soil nutrients via rhizoids in its natural habitats (Bates, 1994). Indeed, B. rutabulum was shown to grow faster than P. purum at continously high nutrient availability (Furness and Grime, 1982). Thus, it seems that the ability of different moss species to utilize pulse-like versus continuous nutrient supplies for growth-related processes is a hard trait, worth measuring because of its link to ecological strategy.

Sixth, certain bryophytes possess water-conducting cells and simple vascular structures, for instance in the stems and rhizoids of some acrocarpous mosses (e.g. Polytrichum), enabling them to take up and transport soil water, and presumably dissolved nutrients, more efficiently than other non-vascular cryptogams (Proctor, 2000). Qualitative information on such structures, including drawings, may often be found in floras and the illustrations therein. Ligrone et al. (2000) clearly distinguished the water-conductive tissues in different bryophyte taxa. However, there is, to our knowledge, no direct quantitative evidence that these tissues enhance nutrient uptake importantly. The same holds for other morphological adaptations that enhance water uptake rate after desiccation (Proctor, 2000), or water retention capacity, which may also indirectly affect nutrient fluxes. Such hydrology-related traits are important in themselves, since the water status of these poikilohydric organisms strongly determines their ecophysiology and climate responses.

Finally, epiphytic bryophytes and lichens have been found to be effective at intercepting nutrients from precipitation or throughfall water, thus acting as a filter between atmosphere, tree and soil (Rieley et al., 1979; Nadkarni, 1984, 1986; Knops et al., 1996; Hölscher et al., 2003). While the function of such filters has been studied in several biomes, the contributions of different component species of epiphytic communities, and their traits, with respect to nutrient interception and filtering are largely unknown. This could be a worthwhile subject for multispecies screening tests.

Cryptogam traits related to nutrient conservation

The functional significance of high nutrient use efficiency (NUE) as an adaptive strategy in infertile ecosystems has been investigated widely for vascular plants (Aerts and Chapin, 2000). Important contributors to NUE are the mean residence time (MRT) of nutrients in green tissues, as determined by leaf life span, and nutrient resorption efficiency (NRE), the latter being the percentage of leaf N or P retranslocated from senescing parts to other plant parts (Van Heerwaarden et al., 2003). Using complementary methods including ¹⁵N labeling, Eckstein et al. (1999) and Eckstein (2000) showed that the mosses Hylocomium splendens and Polytrichum commune combined long MRT with rather high NRE. Combined with the finding that the recycled nitrogen was mostly retranslocated to young, green segments, these papers together showed a high NUE for these two species. Bates (2000) gave a few further examples of internal nutrient cycling in bryophytes. Recently, another ¹⁵N-labelling study revealed retranslocation of N from older to new and growing thalli of the Scottish heathland lichen Cladonia portentosa (Ellis et al., 2005). These are important findings with respect to the role of cryptogams in nutrient-poor ecosystems. In a related study, basal parts of podetia of C. portentosa were labelled with ³³P, and subsequently its upward translocation to the young apices of the podetia could be demonstrated (Hyvarinen and Crittenden, 2000). Although these findings demonstrate nutrient recycling both in bryophytes and lichens, it is not known whether these are general patterns for these cryptogams, whether there are important interspecific differences in tissue MRT and NRE, or whether these could be predicted based on functional types or habitat types. While adding and tracking ¹⁵N labels in different bryophyte segments is a worthwhile method, it is hard to apply to multiple species over relatively short time scales. For interspecific screening of NRE, a less direct but still informative method would be to measure N or P concentrations of young leaves or thalli together with those of recently senesced ones, as long as the senesced tissues are (in all aspects other than age) broadly representative of the young tissues they are compared with (see Van Heerwaarden et al., 2003 for methodological issues).

Cryptogam traits related to ecosystem carbon gains

Potential (inherent) relative growth rate (RGR) is a key functional trait for vascular plants. Fast growth generally gives competitive advantage over other plants and/or the ability to exploit gaps after environmental disturbances, and therefore it tends to be associated with fertile habitats of low or high disturbance likelihoods (Grime and Hunt, 1975; Grime, 2001). Based on an empirical ‘hard’ dataset, the same appears to hold true for temperate bryophytes (Furness and Grime, 1982; Grime et al., 1990; Bates, 1994). However, comparing bryophyte RGRs among different datasets is difficult, given the strong dependence of RGR on hydration, temperature and other environmental factors (Clymo and Hayward, 1982; Furness and Grime, 1992; Bates, 1994). Several alternative methods for quantifying growth potential of bryophytes, including non-destructive ones, have been used (Russell, 1988), but cross-calibration of such datasets is almost impossible. An additional complication is that in bryophytes absolute growth itself is often rather independent of initial size, which makes the calculated relative growth rate (RGR) highly dependent on the initial size of the plant. This makes standardizing according to size an important issue.

The problem seems to concern length growth in particular, but growth in dry weight may show the same pattern (Rincon, 1988).

There have also been several RGR studies on lichens (e.g. Kärenlampi, 1971; Armstrong and Smith, 1996; Sundberg et al., 1997; Cooper and Wookey, 2001; Sundberg et al., 2001); however, using different methodologies. It is commonly assumed that thallus mass and size increase is mostly due to the growth of fungal hyphae (Hale, 1973; Palmqvist, 2000). Obtaining RGR data for large numbers of species, standardized or calibrated for methodology, is a major challenge; see Hale (1973) for a detailed review of methodological issues with respect to lichen RGR. Rogers (1990) made a worthwhile attempt at standardizing RGR using data for 34 species from the literature. Smaller differences among species in this dataset are probably strongly affected by methodology, environmental setting and the basis of expression, for instance (one-dimensional) relative length increment per time unit may not scale linearly with (two-dimensional) area increment or (essentially three-dimensional) mass increment when plotted across species. However, the orders-of-magnitude difference between slow-growing and fast-growing species will probably be meaningful in this dataset.

Based on evidence from vascular plants (Poorter et al., 1990), the maximum rate of photosynthesis may be a good correlate of potential relative growth rate when compared across species, both for bryophytes and lichens, although differences in the often substantial respiration rates for lichens (Lange et al., 1998; Palmqvist et al., 2002) may interfere with such a correlation. There have been rather numerous studies on photosynthetic rates of both bryophyte (e.g. Clymo and Hayward, 1982; Sonesson et al., 1992; Proctor, 2000) and lichen species (e.g. Nash et al., 1983; Kershaw, 1985; Lange et al., 1998; Palmqvist et al., 2002). These rates are highly influenced by environmental conditions during measurement in the various studies, e.g. light regime, hydration and temperature (Clymo and Hayward, 1982; Kershaw, 1985; Sonesson et al., 1992; Kappen et al., 1996; Lange et al., 1996; Tuba et al., 1996; Renhorn et al., 1997; Riis and Sand-Jensen, 1997; Proctor, 2000), CO₂ concentrations in water hosting aquatic bryophytes (Riis and Sand-Jensen, 1997), as well as by ontogeny. Still, for many of the species studied it should be possible to obtain a crude but useful estimate for the maximum photosynthetic rate (P_max) either from (available or new) empirical measurements or by means of models from which P_max can be calculated from available data. A good, albeit labour-intensive example of a systematic way to obtain P_max for several bryophyte species was reported by Skre and Oechel (1981), who screened five boreal species for P_max by measuring photosynthetic rates under a range of environmental conditions. Although bryophyte hydration status, temperature and light regimes and soil nutrient availability all affected photosyntetic rates greatly, and seasonal and year-to-year variation were substantial, they were able to derive P_max values this way. The acrocarpous moss Polytrichum commune had the highest P_max (2·65 mg CO₂ g⁻¹ h⁻¹), the two pleurocarpous mosses, Hylocomium spendens and Pleurozium schreberi had intermediate P_max values (1·39 and 1·20 CO₂ g⁻¹ h⁻¹, respectively), whislt both Sphagnum nemoreum and S. subsecundum had a particularly low P_max (0·25 and 0·57 CO₂ g⁻¹ h⁻¹, respectively). It would be interesting to test whether these species are representative of more general differentiation among acrocarps, pleurocarps and Sphagnum with respect to P_max.

An obvious shortcut for future multispecies screening would be to find reliable soft traits that predict interspecific (mass-based) RGR variation with some precision. Internal tissue N concentration, and perhaps also P concentration, of the complete vegetative bryophyte shoot (gametophyte) or the entire lichen thallus are among the obvious candidates (see above and Fig. 1). Specific leaf area (SLA, fresh leaf area per unit leaf mass) has been the most popular soft trait to do that job in vascular plant studies (e.g. Lambers and Poorter, 1992; Hunt and Cornelissen, 1997). However, for bryophytes, which often consist of single layers of poikilohydric cells with little investment in structure, this trait may not be very informative. Thallose species do not have leaves at all. We propose that the ratio between cell volume (at full turgor) and cell wall volume (or the ratio between cell surface and cell wall surface in cross-sections or flat projections) may be a proxy for RGR when tested across species. We speculate that this ratio may be approximated by the soft trait, tissue dry matter content (dry mass/saturated mass) in analogy with leaf dry matter content in vascular plants (Garnier et al., 2001), but there are no empirical data as yet to test this. However, a major complication is that saturation of certain species, e.g. Sphagnum spp., means that specialized (hyaline) cells for water uptake contribute disproportionately (up to 90 %) to the saturated weight (Clymo and Hayward, 1982), while these cells do not contribute to photosynthesis and growth in dry mass terms. Methods would be needed to subtract this water mass from the total fresh mass. Dilks and Proctor (1979) used water potentials to partition the total water pool in bryophytes into symplast water, apoplast water and external capillary water, the latter including water stored in hyaline cells of Sphagnum. However, these relationships are species-dependent (Proctor et al., 1998). It would be too labour-intensive to derive relationships between water content and water potential for each species in order to determine meaningful tissue dry matter contents as a proxy for RGR.

For lichens, which do not have any leaf-like structures and are fundamentally different in structure from plants anyway, SLA would not qualify as a trait either, but the ratio of total ‘algal’ (autotroph) volume (or area in cross-section) over fungal volume (or area in cross-section) might be tried as a soft proxy for RGR. Essentially, relatively high ‘algal’ volumes, incorporating both green algae and cyanobacteria, would be linked with photosynthetically productive thalli enabling faster growth, while thalli relatively rich in fungi would be expected to be linked with structure and protection, and slower growth. However, this trait may not be easy to quantify for a range of species. A promising and empirically tested alternative trait is the thallus chlorophyll a concentration. This has been demonstrated to correlate with the efficiency (e) of light irradiance conversion into biomass (Palmqvist, 2000; Sundberg et al., 2001) as well as to photosynthetic capacity (Palmqvist et al., 2002) and to RGR itself (Gaio-Oliveira et al., 2006). As for bryophytes, a speculative (inversed) cross-species link between thallus dry matter content and growth rate has to our knowledge not been tested empirically.

Nowadays, a convenient method to estimate photosynthetic activity of bryophytes and lichens in the field is provided by measurements of chlorophyll fluorescence (Wasley et al., 2006). The actual photosynthetic activity at ambient light can be estimated fairly accurately using the fluorescence parameter ΔF/F_m′ (Schroeter et al., 1992; Rice and Schneider, 2004), although the underlying assumption, that the relationship between this parameter and the rate of photosynthesis is linear, does not always hold (Green et al., 1998). Yet this method has been used successfully in screening a range of bryophyte species (Marschall and Proctor, 2004) and may be the priority candidate for multiple-species screening for photosynthetic capacity.

A recent experimental study has shown that the epiparasitic, non-photosynthetic liverwort Cryptothallus mirabilis benefits from the photosynthesis of neighbouring birch trees by importing carbon via the ectomycorrhizal fungi that connect the liverwort to the trees (Bidartondo et al., 2003). However, such parasitism is unlikely to contribute substantially to larger-scale carbon cycling, which makes it a low priority candidate for systematic screening.

Traits related to carbon and nutrient losses

On the vegetation losses side of the ecosystem carbon balance, herbivory on cryptogams is potentially important. We know that certain invertebrate herbivores specialize on non-vascular cryptogams (Gerson, 1973; Gerson and Seaward, 1977; Hodkinson et al., 1996), while reindeer (caribou) have a strong preference for lichens, particularly in winter (Richardson and Young, 1977; Helle and Aspi, 1983; Danell et al., 1994). However, we have already discussed the relatively great investments of many cryptogams in anti-herbivore defences, such as lichen acids and phenols, which minimize the actual carbon and nutrient losses to invertebrate and (reindeer and caribou excluded) vertebrate herbivory in nature.

Litter decomposition is another important source of ecosystem carbon loss and nutrient turnover. Various studies have shown that the litter of bryophytes generally tends to be resistant to decomposition, certainly when compared with vascular plants (Smith and Walton, 1986; Hobbie, 1996; Liu et al., 2000; Cornelissen et al., 2004) and Sphagnum peat-mosses are particularly resistant (Coulson and Butterfield, 1978; Clymo and Hayward, 1982; Scheffer et al., 2001; Dorrepaal et al., 2005). However, very little is known about variation in litter decomposability of different bryophyte species and types, or about the traits that provide the best proxies for decomposability. This should be high on the research agenda, in view of the vulnerability of bryophytes to increased atmospheric nitrogen deposition (Limpens and Berendse, 2003) and changes in ecosystem hydrology that accompany climate changes (see Introduction). A decline in bryophytes overall, or in their species composition, could have major repercussions for soil carbon and nutrient turnover. However, larger, multispecies comparative studies of bryophyte litter decomposability, such as have been carried out for vascular plants (e.g. Cornelissen, 1996) are absent from the literature as yet. From the few available studies it seems that substantial variation in litter decomposability occurs, even within the same genus. For instance, Johnson and Damman (1993) compared mass loss rates of several Sphagnum species cross-transplanted between peatland hummocks and hollows. While all Sphagnum species showed slow litter decomposition, S. fuscum decomposed much more slowly than S. cuspidatum, S. lindbergii or S. angustifolium, both in hummocks and in hollows. Litter bag experiments with Calliergonella cuspidata in Dutch chalk grasslands suggested considerably higher rates of decomposition, however, especially on N-exposed slopes (Van Tooren, 1988).

Even less is known about lichen decomposition, apart from several comparative studies on a few species at a time (Wetmore, 1982; Greenfield, 1993; McCune and Daly, 1994; Esseen and Renhorn, 1998). Together these studies suggest that lichens differ widely in decomposability, but that on average they tend to be more decomposable than bryophytes (see also Guzman et al., 1990; Coxson and Curteanu, 2002; Holub and Lajtha, 2003; but see Sedia and Ehrenfeld, 2006). However, Cladina stellaris was shown to be recalcitrant to decomposition, at least in comparison with litter of vascular plants in the same boreal community (Moore, 1984). The available studies are hard to compare because of differences in methodology and environment (cf. Preston et al., 2006) and are unlikely to reveal general patterns at the level of larger taxa or functional types. We are currently testing litter decomposabilities of 29 subarctic bryophyte and 16 lichen species of diverse types and habitats (S. I. Lang et al., unpubl. res.), essentially following the ‘litter incubation bed’ approach in which many species can be screened simultaneously in a standardized set-up (Cornelissen, 1996). One of the main methodological challenges in this type of screening is to obtain real litter, since cryptogam litter is notoriously difficult to distinguish and separate from live tissues. In bryophytes, stems may also have much longer life spans than the leaves attached to them, while the feasibility of separating these in a meaningful way is doubtful. One alternative method that may deserve developing further is to calculate bryophyte litter mass loss rates from annual litter production rates, estimated from shoot growth analysis and litter accumulation (Nakatsubo et al., 1997). This method may, however, be too labour-intensive for multispecies application, while standardizing the environment for different species would be a major challenge too. Another methodological challenge for some lichen species is that it may even be questionable whether they ever produce litter; the algal and fungal components might senesce and die at different rates and these processes may be determined largely by other environmental factors, such as being cut off from the sunlight by plants overgrowing them.

CONCLUSIONS

Comparative cryptogam ecology has the potential to meet some of the important challenges of understanding and predicting the biogeochemical and climate consequences of large-scale environmental changes driving shifts in the cryptogam components of vegetation composition. Relevant data for certain biogeochemistry-related traits are available for bryophyte and lichen species in the literature, including those involved in cryptogam nutrition and nutrient recycling, their anti-herbivore defences, and their potentials for carbon gain and losses. However, these data tend to be either too qualitative, based on too few species to detect patterns useful for scaling-up, or too mixed in terms of the methodologies or environmental conditions in the different studies. New multispecies, standardized trait tests using common protocols may be a way forward. We have discussed both the promises and problems of screening for several traits already well established in vascular plant research, and some new trait tests tailor-made to screen for impacts of cryptogam species on biogeochemistry. The expression of many of these traits depends importantly on the level of aggregation among individuals of the same species, which partly also determines their hydration state. This creates the dilemma of whether to standardize the multispecies tests by using totally standardized aggregation, or to screen them at densities closer to the natural environments of each species. Other methodological challenges include how to propagate and grow many different species in laboratories, growth chambers or greenhouses, and how to distinguish between living, senescing and dead parts of bryophytes and lichens.

The traits and tests present a starting point with respect to testing for the roles of different cryptogam species and types in ecosystem biogeochemistry. In the long run, traits related to environmental disturbances and vegetation succession, e.g. dispersal and regenerative traits (Fig. 1, bottom), will play a role too. Similarly, as discussed above, traits related to the water status of cryptogams have key impacts on ecosystem hydrology, ecosystem flammability and soil cohesion (Fig. 1, left), which themselves are critical factors in biogeochemical cycling. Even cryptogam traits related to vegetation albedo (Fig. 1, left) and regional energy budgets may, via their impacts on climate, ultimately feed back to biogeochemistry.