All clades across the Tree of Life resorbed nutrients during organ senescence, but the efficiency differed strongly among clades. Mosses, lichens, and lycophytes generally showed low REN, liverworts and conifers intermediate and monilophytes, eudicots, and monocots high. With reduced numbers of tracheophyte clades (only eudicots or angiosperms present), the pattern for REP was similar to REN but less clearly expressed. Within mosses, taxa with more efficient conductance also showed higher REN. Thus, the variation in nitrogen resorption efficiency in a subarctic flora broadly supports the hypothesis that the evolution of conducting tissues has aided plants to optimize their internal nutrient cycling in terrestrial environments. While we have most confidence for our results for RE as expressed on a stable secondary chemistry basis, derived with our novel application of FTIR–ATR, the general correspondence of the between-clade patterns for REsr and RE% show that the larger differences in RE are rather robust to methodological factors (interaction effect of clade x method type is not significant). Below we will discuss our findings in more detail with special focus on the types of conducting system that might support nutrient resorption of autotrophs across the Tree of Life.
Conductive systems as vehicles for nutrient resorption across autotrophic clades
The fact that all nontracheophyte cryptogam clades had distinctly low REN compared to seed plants is consistent with the pattern of increasing differentiation of conducting tissue, from lichens, liverworts, and moss clades with no or little conducting tissue to the high differentiation of phloem in seed plants, although the apparently lower REN of conifers compared to angiosperms might be the results of other traits regulating RE (see below). Though based on fewer clades, the trend for REP was similar, with angiosperms showing highest REP in comparison with mosses, lichens and, to a lesser extent, liverworts. Our results for monilophytes and lycophytes were surprising, even though the numbers of species represented in this subarctic flora were too low for any firm statements. Still, the lycophytes showed low N resorption, while monilophytes had particularly efficient N resorption, even higher than the REN% of 52% reported for a temperate fern (Killingbeck et al. 2002). The phloem of lycophytes (e.g., Lycopodium) differs from other vascular cryptogams in possessing plasmalemma-lined sieve area pores which are wide open, whereas the pores of certain monilophytes, for example, Equisetum and leptosporangiate ferns, are traversed by membranes of endoplasmic reticulum (ER). In sieve tube members of eudicots, ER may play an important role in phloem-loading processes (Behnke and Sjolund 1990). However, recent studies suggest ER facilitates the trafficking of proteins (P-proteins in angiosperms, refractive spherules in monilophytes) between SE and companion cells (CC; Behnke and Sjolund 1990; Van Bel 2003). Such proteins, absent in conifers, lycophytes, and mosses, are thought to be involved in sugar metabolism, transmembrane sugar transport, membrane water permeability and protein degradation (Behnke and Sjolund 1990; Van Bel 2003). Thus, we speculate that the absence of P-proteins or refractive spherules in conifers and lycophytes may partly explain their low RE. In addition, the evergreen habit of both conifers and lycophytes might compensate for insufficient RE, conserving nutrients by extending their mean residence time (Aerts 1995). Furthermore, the low N concentrations in green tissue found in these clades (Fig. 3) limit the extent of RE (Aerts 1996), given that there is always a pool of N that remains immobile during senescence (Killingbeck 1996). We expected that the evolutionary emergence of specialized parenchyma cells, i.e., “Strasburger cells” in conifers and CC in angiosperms (Behnke and Sjolund 1990), would enhance the longitudinal flow in the phloem due to increased porosity of the end walls, which would then facilitate RE compared to that in vascular cryptogams (Van Bel 2003). However, RE of monilophytes was equally high as RE in seed plants, possibly owed to the presence of single cytoplasmatic connections between SE and parenchyma cells in this clade (Behnke and Sjolund 1990). It is tempting to assign differences in RE to minor vein type structure and their related phloem loading types, from ancestral symplastic phloem loading (ferns, conifers) to apoplastic or mixed loading in evolutionarily young angiosperms (Van Bel 2003). However, in the dicots, all three minor vein types are found (Turgeon et al. 2001) and, moreover, symplastic phloem loading also occurs in evolutionarily young angiosperms such as Salix (Van Bel 2003). Gamalei (1991) suggested evolutionary specialization of phloem loading in relation to climate. Typical in arctic-alpine tundra is the closed minor vein type, related to apoplastic phloem loading (except trees: symplastic) as symplastic plasmodesmal transport is sensitive to chilling. However, interactions between minor vein type and transported sugar identity on phloem loading type (Van Bel 2003), which is hypothesized to influence hydraulic gradient strength (Holbrook and Zwieniecki 2005), may further complicate the implications for resorption. Analogous to studies on xylem (Roth and Mosbrugger 1996), stelar morphology from a primitive protostele in the earliest land plants (e.g., Cooksonia, Rhynia) to eustele (e.g., eudicots), may have implications for phloem transport properties possibly allowing for increasing values of RE from primitive to sophisticated phloem transport properties.
In conclusion, many open questions remain concerning evolution of SEs and their cell biology, pathways and modes of phloem loading, impact of environmental factors on phloem transport and possible adaptation of phloem loading modes to climatic conditions (Van Bel 2003). Yet answering these questions will provide the basis from which we can start to evaluate the factors influencing RE in tracheophytes.
Novel screening of nutrient resorption in higher cryptogam taxa
Our study is the first broad comparative study of nutrient resorption efficiency across and within basal cryptogam and seed plant clades. While for some bryophyte and lichen taxa, the subarctic flora did not support sufficient numbers of species for statistically sound comparison, we have quantified some consistent and logical patterns. Despite scattered evidence of N and P translocation for several moss species (e.g., Eckstein and Karlsson 1999), no study so far has investigated RE of a representative number of moss species in order to detect consistent taxonomic variation. Within mosses, translocation has been mainly linked to the endohydric Polytrichales (Collins and Oechel 1974; Reinhart and Thomas 1981) which feature leptoids (Hébant 1977), that is, conducting tissue, somewhat comparable with the sieve cells of higher plants. Also, this moss class alone features refractive spherules and callose, associated with plasmodesmata (Ligrone et al. 2000), which, in angiosperms, are associated with pores during sieve plate development (Behnke and Sjolund 1990). Indeed, REN% in the Polytrichaceae was higher than in other mosses, especially when compared to the ectohydric Sphagnum. However, though devoid of leptoids, transport of photoassimilates (Alpert 1989; Hakala and Sewón 1992) or N (Eckstein and Karlsson 1999) has also been reported for Dicranum, Grimmia and Hylocomium, respectively, most likely facilitated by an internal conducting strand of elongated parenchyma cells (Ligrone and Duckett 1994). Moreover, evidence has been found for conducting tissue in the Sphagnales differing from the Bryopsida only in lacking plastid – microtubules associations (Ligrone and Duckett 1998); this conducting tissue might explain P, C, and N translocation found for the peat moss Sphagnum (Rydin and Clymo 1989; Aldous 2002). We have to be aware that some interspecific variation in RE may be due to environmental factors. Furthermore, as not all mosses could be collected at the same time, species differences in seasonality of translocation (Skre et al. 1983), downwards for storage and upwards for growth, might have contributed to some of the observed variation. Also, cyanobacterial N fixation observed on feather mosses (Zackrisson et al. 2009) might have complicated the observed pattern. Indeed, REN of these mosses was higher than in most other species of the Bryopsida (data not shown).
This is, to our knowledge, the first study of RE of N and P in senescing photosynthetic parts of liverworts. With the exception of the complex thallose liverwort Marchantia (Rota and Maravolo 1975), little is known about translocation in this clade. Ligrone et al. (2000) suggested a microtubule-based translocation system for the marchantialean liverwort Asterella. However, liverworts seemed to generally resorb nutrients at rather high rates, comparable with those of conifers and monilophytes. A central strand of conducting tissue, which may lead to increased translocation of nutrients, has so far only been found for species in the liverwort orders Calobryales, Pallaviciniales (Hébant 1977), Pelliales and Marchantiales (Ligrone et al. 2000), of which the latter two were represented in our subarctic flora. The hypothesized link between central strands and RE is in need of explicit comparison across liverwort orders. Whether liverwort taxa representing different growth forms, (hairy) leafy, simple or complex thallose, show differences in RE, needs broader screening in other climatic regions as most species in this study belonged to the leafy liverwort order Jungermanniales, while all other orders were represented by only one (Ptilidiales) or very few species in the study area.
While several studies have reported nutrient resorption and translocation in nonsenescent tissues in lichens, nutrient resorption related to tissue senescence in lichens has hardly been studied at all. Translocation of N and P between fresh tissues has been demonstrated for Cladonia stellaris, Stereocaulon paschale, and Cladonia portentosa, respectively, (Hyvärinen and Crittenden 2000; Ellis et al. 2005; Kytöviita and Crittenden 2007), while for Caloplaca trachyphylla, transport of carbohydrates has been shown (Bench et al. 2002). Translocation in lichens, in which fungal hyphae provide the main structure, can most likely be compared with translocation in (ecto-) mycorrhizal mycelium. Out of three suggested transport mechanisms, diffusion, mass flow along turgor gradients or cytoplasmic streaming through pores of septa (Finlay 1992), the latter seems to be the most likely transport mechanism in lichens (Hyvärinen and Crittenden 2000). Assuming that a similar transport mechanism prevails in all fungal hyphae, we hypothesized that RE would not differ among lichens of different growth forms, when excluding N2-fixing lichens. Indeed, REN did not differ among lichen orders. However, REP% was higher in the Lecanorales and seemed to be negatively related to the N-fixing ability of lichens. The differences found in REP may be due to P in cyanobacteria. As cyanobacteria are located in cephalodia that are still present in dead lichen tissue, cyanobacterial P might not be accessible during resorption but is left behind in aging lichen tissue. Furthermore, algae in lichens are known to store P as polyphosphate in granules (Guschina et al. 2003), complicating the interpretation of the observed pattern. For N2-fixing lichens, with a constant input of readily available N, one would expect N resorption to be less important than in non-N2-fixing lichens, analogous to reduced REN of N2-fixing higher plants (Killingbeck 1996). However, N2-fixing lichens did not differ in REN from non-N2-fixing lichens, which may point to a similar transport mechanism across lichen species.
While in this study we concentrated on the relation of increasingly complex conducting tissue on RE, we acknowledge that other factors may also have important impacts on nutrient resorption. The influence of the environment on the poikilohydric bryophytes and lichens should not be underestimated since these cryptogams can take up nutrients via their whole surface. It is also known that nutrients can be lost from tissues through leaching. While this was shown to be of little significance to subarctic vascular plants (Freschet et al. 2010), we do not know the degree to which nutrients can be leached from different bryophytes and lichens. Also, the strength of the nutrient sink, for example, in roots of vascular plants, may be a factor contributing to nutrient resorption within the plants. Seasonality of translocation may play a role not only for mosses (Skre et al. 1983) but also for liverworts and lichens. Furthermore, while we know for a few moss species that nutrients can be stored in brown tissue (Skre et al. 1983), this has not been shown yet for liverworts and lichens. While all the above factors may have some influence on nutrient resorption efficiencies measured in broad species screenings, the fact that we found (mainly) coherent patterns among higher taxa, despite differences in sampling time and sampling from different environments, seems to strengthen the idea that conducting tissues indeed play an important role in nutrient resorption.
In conclusion, our results for an N-limited subarctic flora support the hypothesis that the progressive evolution of tissues to facilitate internal transport across the major clades of land plants is, by and large, coupled with their nutrient resorption efficiency during organ senescence. As such, this has led to a lesser dependency of plants on external nutrient supply. However, the generality of such evolutionary lines in conducting tissues and nutrient resorption efficiency needs to be tested more rigorously across different floras in different climatic regions with different levels of N versus P availability.
While many organism characters have been mapped explicitly onto the Tree of Life before, nutrient resorption may represent one of the first organismal processes to have been given this treatment. Under the assumption that actual nutrient resorption, which involves several interacting physiological and chemical processes (Gan 2007), is influenced by myriad genes (Gan 2007), our approach will be of great interest for phylogenetic analysis of other complex organismal processes as well.