Variation in decomposability among and within main cryptogam taxa
Our study was the first to reveal consistent patterns of variation in litter decomposability of a wide range of non-vascular cryptogams when incubated in a standard environment. While there were several significant interactions between litter incubation environment or incubation period and species group identity on mass loss rates, the general pattern was that of a strong influence of cryptogam taxonomic identity on litter mass loss.
Both rapid and slow lichen mass loss for non-Cladonia (Wetmore 1982) and Cladonia (Moore 1984), respectively, have been reported in comparison to vascular plants, emphasizing the importance of taxonomic identity. Despite expected antibacterial activity (e.g. Vartia 1973), lichen decomposition rates in our study were generally comparable with those of vascular plants. Stark & Hyvärinen (2003) suggested that the microbial community growing underneath lichens is well adapted to the lichen secondary metabolites, utilizing them as a C source, which might also explain the high turnover rates in our study. Even certain Cladoniaceae, a less degradable group compared with non-Cladonia lichens, reached mass losses of up to 50% after 2 years of incubation. Crittenden & Kershaw (1978) proposed that N2-fixing lichens exhibited higher N contents and that decomposition rates of N2-fixers should therefore be greater than those of non-N2-fixing lichens. In our study, N2-fixing and non-N2-fixing lichens displayed similar high decomposition rates in spite of N in N2-fixing lichens indeed being significantly higher than in non-N2-fixing lichens. N2-fixing lichens, as opposed to non-N2-fixing lichens, decomposed equally well in contrasting litter beds, suggesting that at higher N concentrations the microbial community is not limited by substrate N leading to fast decomposition decoupled from substrate N (Coulson & Butterfield 1978).
Bryophytes comprising Sphagnum, non-Sphagnum mosses and liverworts showed consistently low decomposability compared with vascular plants. This has been shown earlier for Sphagnum and single bryophyte species (Hobbie 1996; Aerts, Verhoeven, & Whigham 1999; Liu, Fox, & Xu 2000) but ours is the first explicit and comprehensive test of the general low degradability of bryophyte litter. Within the Polytrichaceae, mass loss ranged from 10% to 35%, emphasizing that even closely related species can diverge strongly in decomposability. The overall lower decomposition rates found for Sphagnum have been attributed to structural, lignin-like and soluble phenolic compounds (Verhoeven & Liefveld 1997), both of which are also found in non-Sphagnum mosses (Erickson & Miksche 1974; Zinsmeister & Mues 1990; Ligrone et al. 2008), and decomposition-inhibiting bacteria associated with Sphagnum (Opelt, Berg, & Berg 2007). While most liverworts decomposed slowly too, as related presumably to high contents of secondary metabolites (Asakawa 1994), a few species reached decomposition rates higher than those of any mosses in our study. Despite their suggested antibacterial activity (Zhu et al. 2006), oil bodies, known to contain large amounts of terpenoids and aromatic compounds (Asakawa 2004), do not seem powerful enough to inhibit decomposition. This may be due to volatilization of most oil body content within days or weeks after collection. The interaction effect of time and species among liverworts implies a wide heterogeneity of species responses to decomposition over time, possibly due to subgroups within the liverwort group as indicated by chemosystematic differences between, for instance, the Jungermannidae (e.g. Lophozia lycopodioides) and Marchantiidae (Marchantia alpestris) (Asakawa 2004). While we were able to reveal the importance of species identity on litter mass loss, little is known about the intraspecific variability in mass loss of species growing at contrasting sites or in different geographic regions, with possible feedbacks on chemical composition (cf. Bakken 1995) and, consequently, litter decomposability. Furthermore, little is known about mass loss rates of cryptogam species on longer time scales which could deviate considerably from our shorter-term results.
Species-environment interactions on mass loss
Mass loss across and within groups was significantly affected by species (or group) identity, litter bed environment (at species level) and their interactions. This contrasts with earlier studies investigating predominantly Sphagnum where no significant interaction was found (Belyea 1996; Turetsky et al. 2008). The deviating results may be due to the strongly contrasting environments (birch forest vs. mire) used in our study or to the species chosen. Decomposition rates for Cladonia species were highest in the nutrient-poor birch forest litter environment where the species naturally occurred and microbial communities might therefore be well adapted (Stark & Hyvärinen 2003). In the nutrient-rich birch forest environment where Cladonia species are naturally absent in the Abisko region, accordingly lower mass loss rates were found. Decomposition rates of Cladonia species were also lower in the peatland environment. Correspondingly, Tolonen (1971) found subfossil remnants of lichens in peat cores. Wetmore (1982) suggested thallus structure and firmness of the fungal cortex as determinants for lichen decomposition, while usnic acid content, known for its antibiotic effects (Cocchietto et al. 2002), did not determine mass loss of the species in his study. Fumarprotocetraric acid, present in Cladonia rangiferina and C. arbuscula, is known to be more toxic at lower pH values (Gardner & Mueller 1981), and possibly also negatively affects decomposition rates. Mass loss rates of Peltigera aphthosa and Nephroma arcticum, both free of fumarprotocetraric acid and with significantly greater N content compared with the Cladoniaceae, were equally high under all environmental conditions, possibly due to local microbial increases which is due to high N content (Coulson & Butterfield 1978). Crittenden & Kershaw (1978) proposed leaching and structural damage as possible pathways for N which, in lichens, consisted to a large degree of amino acid N (Solberg 1970). In N-limited environments as common in the (High) Arctic (Shaver & Chapin 1995), amino acids are readily taken up by mosses (Krab et al. 2008), lichens (Dahlman et al. 2004) and vascular plants (Chapin, Moilanen, & Kielland 1993). Regarding non-Sphagnum mosses, the regression of N versus mass loss proved to be significant, with Racomitrium lanuginosum exhibiting extremely low N values (see also Pakarinen & Vitt 1974). In contrast, Turetsky et al. (2008) reported k values of both Sphagnum and non-Sphagnum mosses to be positively related to the ratio of metabolic to structural carbohydrates, but not to N. As a peatland hummock-builder, R. lanuginosum displayed equally low decomposition rates independent of site which might not only be due to low N values, but also to high values of structural carbohydrates as found in hummock-building Sphagnum species (Turetsky et al. 2008). In contrast, most of the other non-Sphagnum mosses showed a tendency towards increased mass loss rates in the peatland environment. Leaching in the wet mire environment could be a minor cause for enhanced mass loss rates compared with the birch forest but the low soluble content in mosses (Pakarinen & Vitt 1974; Turetsky 2003) cannot account for more than 1–10% of total mass loss. Maybe more importantly, low moisture could have inhibited moss decomposition (cf. Flanagan & Veum 1974; Meentemeyer 1978), especially in the nutrient-rich birch forest litter bed where bryophyte cover was thin and might have dried out during warm periods. Furthermore, the habitat-specific soil fauna can be expected to influence decomposition rates to various extents in the contrasting habitats of peat and mineral sites (Coulson & Butterfield 1978). Except for N2-fixing lichens and R. lanuginosum, where mass loss did not vary across habitats, environment influenced decomposition rates of non-N2-fixing lichens and bryophytes as shown in previous investigations (Coulson & Butterfield 1978; Wetmore 1982; Belyea 1996; Coxson & Curteanu 2002). This complicates predictions of changing decomposition patterns based on species shifts at the landscape level. Future studies should include bacterial, fungal and soil fauna communities of the various ecosystems and investigate a wider range of cryptogam species, including both Sphagnum species and liverworts, in order to unravel the underlying mechanisms of species–environment interactions on cryptogam mass loss.
The relation of initial litter chemistry and decomposability
Principal component analysis based on infrared spectra showed a clear separation of lichens, liverworts, mosses (including Sphagnum) and vascular plants. Relations of overall lichen or bryophyte mass loss to macronutrients and pH were weak while at cryptogam group level, R2 improved considerably. The same pattern, although less pronounced, was found for relations of mass loss to infrared spectra. In both analyses, R2 for liverworts decreased from family to order while R2 for the genus Sphagnum was high. Even with increasing numbers of principal components, regressions were only valid up to class level (Jungermanniopsida) suggesting that Marchantiopsida (Marchantia alpestris) showed a different decomposition pattern. Taxonomic identity might be important when predicting mass loss, possibly due to differences in carbohydrate partitioning (structural vs. metabolic), as has been suggested for true mosses versus Sphagnum species (Turetsky et al. 2008). Differences in chemical components, their allocation and respective mass loss might also be related to habitat, as hydric species with a constant nutrient input have been found to show higher protein levels, both higher or lower metabolic carbohydrate content and within the genus Sphagnum, higher allocation to photosynthetic tissue compared with species from drier habitats (Pakarinen & Vitt 1974; Rice 1995; Davey 1999). Furthermore, availability and type of conductive tissue (Hébant 1977) might influence mass loss, enabling redistribution of easily decomposable components (e.g. Sveinbjörnsson & Oechel 1991; Hakala & Sewón 1992). The lower R2 for non-Sphagnum mosses suggests that indeed further division of this large and heterogeneous group is likely to improve mass loss predictions.
Due to considerable overlap of peaks characterizing plant compounds, only indications of possible cryptogam mass loss predictors can be given. Next steps should include analysis of bryophyte and lichen components with conventional laboratory assays, thereby linking wavelengths to chemical compounds. At this stage, we considered compounds as positive or negative mass loss predictors in analogy with predictors known for vascular plants (Palm & Rowland 1997). Metabolic carbohydrates, lipids and proteins as easily decomposable components should be positively related to mass loss, and structural and non-soluble aromatic components negatively. Soluble phenolics, however, may either serve as energy source or inhibit decomposition (Palm & Rowland 1997). N2-fixing lichens showed mainly positive predictors of mass loss, i.e. metabolic carbohydrates, lipids and amino acids (proteins, peptides). The latter were reflected in the significant relation of mass loss to N, emphasizing the importance of N-limitation in the (sub) Arctic (Shaver & Chapin 1995). In contrast, non-N2-fixing lichens revealed also negative predictors, i.e. aromatic compounds (see regression coefficients between 14.01 and 14.99 μm; but note the positive relation at 9.26, 9.82 μm). Mass loss was further positively determined by N, Ca, K and pH. While the influence of other components (e.g. lichen acids) on pH cannot be excluded, it correlated well with the sum of Ca, Mg and K for vascular plants (Cornelissen et al. 2006) and lichens (P < 0.001, R2 = 0.53; see Appendix S3). Ca, often present as Ca oxalate crystals (Brown 1987), might positively influence decomposition via the dietary needs of the decomposer community (Swift, Heal, & Anderson 1979; Nicolai 1988) or provide an indirect positive link to decomposition since both Ca and K were significantly correlated to N (R2Ca = 0.54, P < 0.001, excluding Umbilicaria; R2K =0.61, P < 0.001). Similar interconnections have been found for vascular plants (Swift, Heal, & Anderson 1979).
Since bryophytes generally had low decomposition rates, the chemistry underlying low degradability is of particular interest. Mass losses among Sphagnum species as well as among non-Sphagnum mosses were negatively related to structural carbohydrates (cf. Turetsky et al. 2008) and aromatic compounds, the latter indicating polyphenols substituting lignin in bryophytes (Erickson & Miksche 1974). The positive relation of K to mass loss among Sphagnum and non-Sphagnum mosses may reflect the mainly intracellular location of K in green shoots where N, P and metabolic activity are highest (Pakarinen & Vitt 1974; Brown & Wells 1990). Indeed, proteins (N) and lipids (P), and proteins and metabolic carbohydrates were related to mass loss of non-Sphagnum and Sphagnum mosses, respectively. Extracellularly located Ca is known to increase in older, less metabolically active tissue (Vitt & Pakarinen 1987) explaining indirectly its negative impact on Sphagnum decomposition. This Ca accumulation is mainly attributed to death of tissue providing additional exchange sites (Brown & Wells 1990), which are especially numerous in Sphagnum with its high cation exchange capacities (Clymo 1963). Mg, a positive predictor of mass loss for non-Sphagnum mosses, might be indirectly linked to decomposition, being the cofactor of the N-rich photosynthetic enzyme Rubisco.
For the liverwort family Scapaniaceae, mass loss was positively related to (N in) proteins, while aromatic compounds (as in non-Sphagnum mosses) related both positively and negatively. Structural polyphenolics in liverworts (Erickson & Miksche 1974) are likely to negatively influence decomposition while soluble aromatic liverwort compounds, known to be biologically active (Asakawa 1994), may be antimicrobial in soil or serve as energy source, as suggested for lichens (Stark & Hyvärinen 2003). The positive relation between Na and mass loss might be linked to Na requirements of the decomposer community (Swift, Heal, & Anderson 1979). Na availability in liverwort tissue may reflect environmental conditions or as fungi show high Na levels (Swift, Heal, & Anderson 1979), relates to basidiomycetous infections, repeatedly found in jungermannialean liverworts (Duckett, Russell, & Ligrone 2006).
The infrared measure was not only faster and easier to conduct, but smaller sample amounts were needed which is of major importance when processing minute samples of liverworts. However, measurements of macronutrients and pH might be easier to achieve and could be used to predict mass loss of cryptogam groups. If FTIR-ATR is not available, including analytical determinations of structural and metabolic carbohydrates, proteins and aromatic compounds (both soluble phenolics and lignin-like compounds) should improve predictions. To investigate if and how initial chemistry of the generally less degradable bryophytes relates to final-stage decomposition, screening multiple species over longer time periods will be necessary.