The extent of an external water layer around moss tissue influences CO2 assimilation. Experiments on the desiccation-tolerant moss Syntrichia ruralis assessed the real-time dependence of the carbon and oxygen isotopic compositions of CO2 and H2O in terms of moss water status and integrated isotope signals in cellulose.
As external (capillary) water, and then mesophyll water, evaporated from moss tissue, assimilation rate, relative water content and the stable isotope composition of tissue water (δ18OTW), and the CO2 and H2O fluxes, were analysed. After drying, carbon (δ13CC) and oxygen (δ18OC) cellulose compositions were determined.
During desiccation, assimilation and 13CO2 discrimination increased to a maximum and then declined; δ18OTW increased progressively by 8‰, indicative of evaporative isotopic enrichment. Experimental and meteorological data were combined to predict tissue hydration dynamics over one growing season. Nonsteady-state model predictions of δ18OTW were consistent with instantaneous measurements.
δ13CC values suggest that net assimilation occurs at 25% of maximum relative water content, while δ18OC data suggests that cellulose is synthesized during much higher relative water content conditions. This implies that carbon assimilation and cellulose synthesis (growth) may be temporally separated, with carbon reserves possibly contributing to desiccation tolerance and resumption of metabolism upon rehydration.
The physiological adaptations of vascular plants and bryophytes represent two contrasting solutions to life on land in facilitating CO2 uptake, while managing water loss. Vascular plants have internalized gas exchange (mesophyll) surfaces, protected and regulated by stomatal guard cells in the epidermis, to minimize water loss. By contrast, bryophytes are astomatous, nonvascular species that absorb and lose water very readily (Sveinbjörnsson & Oechel, 1992) and rely on the presence of an external water film to maintain tissue water potential close to 0 MPa (Proctor, 2000). During desiccation events, while most vascular crop plants can progressively acclimatize to water deficits using osmotic adjustment to water potential minima in the region of −2 MPa (Oliver et al., 2005), the water content of bryophytes commonly decreases to 5–10% of the dry mass and the internal water potential can fall below −100 MPa (Proctor et al., 1998, 2007).
While critical for hydration, the external water layer limits CO2 diffusion into chloroplasts, thus water content, photosynthetic assimilation and growth are intimately linked across bryophyte species (Rice & Giles, 1996). The extent of the external water layer is controlled by water availability, wind speed, air temperature and relative humidity but also by leaf structure, so that the optimal hydration level for photosynthetic assimilation is species-specific (Silvola, 1990; Rice & Giles, 1996; Williams & Flanagan, 1996; Meyer et al., 2008). As bryophytes are intimately linked to their environment, stable isotope analysis of water and CO2 molecules, either within plant compartments or in real time, provides markers for assessing the relationships between photosynthesis and external conditions, carbon assimilation and growth. Stable carbon and oxygen isotopes in organic matter will reflect the dominant conditions for net assimilation and cellulose synthesis, respectively, integrated over one or more growing seasons.
Carbon isotope discrimination (∆13C) is dominated by two fractionation processes: a biochemical fractionation during CO2 fixation by the photosynthetic carboxylase enzyme RuBisCO, c. 29‰ in C3 plants (O'Leary, 1988), and a diffusional fractionation during the bidirectional diffusion of CO2 between the atmosphere and the chloroplasts of up to 4.4‰. This largely explains why the carbon isotope composition of assimilates (and subsequently plant tissue) is 13C-depleted compared with atmospheric source CO2. The relative contributions of the biochemical and diffusional fractionations are dependent upon the ratio of chloroplastic (pc) and atmospheric (pa) CO2 partial pressures (Farquhar et al., 1982). A high pc/pa indicates that there is little diffusion resistance and corresponds to ∆13C values close to 29‰. Conversely, when pc/pa is low, the turnover of the internal CO2 pool is slow because of diffusion resistance, the proportion of 13CO2 fixed by RuBisCO increases and photosynthetic discrimination tends towards 4.4‰ (Farquhar et al., 1989).
In astomatous bryophytes, the external water layer is a critical determinant of diffusion resistance for CO2 and, consequently, the extent of discrimination against 13CO2. Previous real-time measurements on both liverworts and Sphagnum moss show that a reduction in the external water layer is associated with an increase in instantaneous discrimination against 13CO2, and assimilation rate (Rice & Giles, 1996; Williams & Flanagan, 1996; Meyer et al., 2008). As a proportion of the assimilated carbon is used to synthesize the structural carbohydrate cellulose, a major degradation-resistant component of bryophyte organic matter, the carbon isotope ratio of moss cellulose (δ13CC) should be a good proxy for the optimal hydration level for photosynthesis.
Furthermore, during cellulose synthesis, oxygen atoms from carbonyl groups exchange with oxygen atoms in water. As a result, the oxygen isotope ratio of cellulose (δ18OC) is enriched by c. 27‰ compared with water at the site of cellulose synthesis (Deniro & Epstein, 1979; Da Silveira et al., 1989; Sternberg et al., 2006), a value that has been confirmed in Sphagnum species (Zanazzi & Mora, 2005). Thus δ18OC is a good recorder of the isotopic composition of tissue water (δ18OTW) at the time of cellulose synthesis, that may, or may not, be simultaneous with carbon assimilation. Because the size of the external and internal water pools in moss tissue is limited and rain events are intermittent, fractionation processes during evaporation and water vapour diffusion, and thus δ18OTW, are dynamic with strong diurnal and seasonal variations. We therefore expect to infer the precise timing of cellulose synthesis from the oxygen isotope composition of bryophyte cellulose.
The aim of this study was to interpret δ13CC and δ18OC values from Syntrichia ruralis subsp. ruralis grown in natural conditions, in light of real-time experimental measurements of photosynthesis and stable isotope composition of atmospheric water vapour, CO2 and tissue water. Measurements were made over a time-course of tissue drying, from saturation to desiccation, to reveal the optimal moss water status for both carbon assimilation and growth, and infer the temporal juxtaposition of carbon assimilation and cellulose biosynthesis on both a diel and seasonal basis.
Materials and Methods
Syntrichia ruralis subsp. ruralis (Weber and Mohr; synonymous with Tortula ruralis (Hedwig)) was chosen for this study because it is a highly desiccation tolerant moss that can withstand the complete loss of free water (Oliver et al., 2000). Following desiccation, rehydration is almost instantaneous (30–90 s) and the recovery of some processes, such as photosynthesis and respiration, can be extremely rapid. In the case of photosystem II, recovery occurs within a few minutes (Proctor, 2001) and full recovery can occur within a few hours (Oliver et al., 2005), depending upon the rate at which the previous desiccation occurred (Oliver & Bewley, 1997).
Syntrichia turves 10–20 mm deep were collected from a single population growing on a roof during spring 2011 (H. Griffiths, Cambridge, UK). Moss samples were soaked for 24 h in deionized (DI) water and maintained at 20°C in ambient external light conditions. Immediately before experimentation, green shoots of Syntrichia (c. 10 mm in length) covering a basal area of 10 cm2 were cut from the turf, immersed briefly in DI and blotted to remove excess water but retain the external water directly associated with the leaflets.
Photosynthetic measurements and ‘on-line’ gas collections were carried out using the LI6400-XT open gas exchange system (Li-Cor, Lincoln, NE, USA). Moss samples covering 10 cm2 were placed within the Whole Plant Chamber (LI6400-17; Li-Cor) and illuminated using the RGB Light Source (LI6400-18; Li-Cor). Approximately 130 shoots of Syntrichia were used in each experiment and each shoot included c. 10 green leaflets. Pure CO2 was added through the CO2 injection unit of the LI6400-XT to provide inflowing air with a CO2 concentration of 400 ppm at a flow rate of 100 μmol s−1; photon flux density (PFD; waveband: 400–700 nm; 10% blue fraction (photon basis)) was maintained at 1000 μmol m−2 s−1 and chamber air pressure at 103.3 kPa. Leaf temperature was calculated within the Li-Cor using energy balance techniques and set to be maintained at 20.5°C.
Isotopic composition of tissue water (δ18OTW)
Syntrichia samples were maintained in the gas exchange cuvette for fixed periods (0, 30, 60, 90, 120, 150 min) or until photosynthesis ceased (max. 300 min). Absolute assimilation rate was recorded every minute. To normalize for variation in leaf area between samples, assimilation rates are presented as a percentage of the maximum value recorded (Amax). At the end of the measurement period the fresh mass of the tissue (FM) was determined and the tissue placed within a sealed vial (Exetainer; Labco, High Wycombe, UK). Any water remaining associated with the moss (both internal and external tissue water) was collected by cryogenic vacuum distillation (Ehleringer et al., 2000; West et al., 2006). The desiccated moss tissue was reweighed to establish dry mass (DM) and relative water content (RWC) was calculated as a percentage of the dry mass:
The tissue water was equilibrated with 1 ml of pure CO2 gas in a sealed vial for 48 h before direct measurement of the isotopic composition of the CO2 using an Isotope Ratio Mass Spectrometer (IRMS; SIRA, VG Isotech, modified by Pro-Vac Services Ltd, Crewe, UK) with laboratory standards interspersed throughout the unknown samples. The measured δ18O value was standardized relative to VSMOW using the known water standards BAS-LO (δ18O = −31.44‰ VSMOW), BAS-HI (δ18O = −8.73‰ VSMOW) and SWK (δ18O = −0.97‰ VSMOW). The measured oxygen isotopic composition of the tissue water (δ18OTW) was expressed relative to the DI source water (δ18OSW = −9.0 ± 0.2‰ VSMOW) according to:
Photosystem II fluorescence was also measured before and after each gas-trapping period using a Mini-Pam Photosynthesis Yield Analyzer (Walz, Effeltrich, Germany) and the quantum yield (Φ) calculated as:
where Fs is the light-adapted fluorescence before saturating pulse and is the light-adapted maximal fluorescence.
‘On-line’ measurements of heavy isotopes in CO2 (13CO2 and C18O16O) and H2O (H218O)
Air leaving the gas exchange cuvette was passed under positive pressure through a liquid nitrogen trap. After 20 min the CO2 and condensed water vapour was successively purified into glass collection vials (Loewers, Hapert, the Netherlands; Harwood et al., 1999). Air collections were repeated over 20-min periods until the moss ceased to be photosynthetically active owing to desiccation.
The carbon and oxygen isotopic compositions of the isolated CO2 and H2O were analysed as described earlier and photosynthetic discrimination calculated based on the isotopic composition of CO2 in air entering (δe) and exiting (δo) the chamber, following Evans et al. (1986):
where ξ = pe/(pe− po) and pe and po are the CO2 partial pressures in air entering and exiting the chamber, respectively.
Isotopic composition of cellulose (δ13CC and δ18OC)
About 5 g of Syntrichia material was removed from the experimental material before immersion in DI, dried to constant weight at 70°C and then ground into a homogeneous powder. Cellulose was extracted from a 0.5 g subsample of the dried organic matter using a method developed for the extraction of α-cellulose from wood (Loader et al., 1997), homogenized and freeze-dried before analysis. In preparation for δ18OC analyses, 0.1 mg of cellulose was transferred into individual pressed 5.25 × 3.20 mm silver capsules (Elemental Microanalysis Ltd, Okehampton, UK). Before δ13CC analyses, 1 mg of cellulose was transferred into individual pressed 5.25 × 3.20 mm tin capsules (Elemental Microanalysis Ltd). Both isotopic analyses were then completed at the Godwin Laboratory, University of Cambridge, UK, following standard procedures (Werner et al., 1996; Kornexl et al., 1999) using a Thermo Finnigan 253 Stable Isotope Ratio Mass Spectrometer (Thermo Fisher Scientific, Waltham, MA, USA) with High Temperature Conversion Elemental Analyser (TCEA; Thermo Fisher Scientific). δ18OC was calculated to the VSMOW scale using two laboratory standards (α-cellulose and CC31), and an international standard (NBS127), to calibrate the oxygen isotope ratio of high purity CO reference gas. The isodat operating software (Thermo Fisher Scientific) was used to calculate sample values relative to the reference gas value. Standards were placed at the beginning, middle and end of the run to correct for any drift during the run.
δ13CC was then used to calculate assimilation-weighted carbon isotope discrimination ∆13C:
where δ13Ca is the isotopic composition of atmospheric CO2 (δ13Ca = −8‰ VPDB).
Modelling isotopic composition of tissue water (δ18OTW) – laboratory conditions
In order to develop an appropriate model to describe δ18OTW over time, synthetic data were generated to predict δ18OTW as Syntrichia tissue dried, using the experimental starting conditions: a leaf temperature (TL) of 20.5°C, an 87% air relative humidity and an air pressure (Pa) of 103.3 kPa. Relative humidity normalized to the leaf temperature (hs) and the atmospheric water vapour mole fraction (wa) were then calculated. The relative humidity at the tissue water–air interface (hi) was assumed to be saturating (hi = 1) as long as the moss remained metabolically active. Indeed, even after the external water layer is lost and the internal water starts to evaporate, leaves shrink and their cell volumes reduce (Proctor et al., 1998), ensuring that while sufficient water remains for metabolism hi is maintained very close to unity. Proctor et al. (1998) experimentally determined the water potential (Ψ) of drying Syntrichia tissue, and Ψ can be used to calculate hi. However, these derived hi values were unsuitable to transfer into our model as the drying curves began from a RWC equivalent to only 200% (compared with 450% in our case) and thus the published data were not indicative of the relevant external water layer dynamics in our experiment and model.
Transpiration rate was calculated as:
where gt is the total conductance to water vapour from the evaporative site to the outside air, and wi is the water vapour mole fraction inside the leaf. The total conductance, gt, was assumed to follow one of two different regimes, dependent upon the presence or absence of the external water layer (Williams & Flanagan, 1996). Based on the Δ13C data, this transition was estimated at 110% RWC (see 'Experimental material' and 'Experimental conditions' sections). Analysis of the progression of RWC during drying (Fig. 1) suggested that gt remained constant in the presence of a water film (RWC > 110%) and decreased linearly down to zero at lower RWC, which is consistent with previous studies on different mosses (Williams & Flanagan, 1996).
The isotopic composition of water vapour inside the experimental chamber (δv) was estimated from the measured isotopic composition of transpired water (δE; Fig. 2d). Estimated δv was used to calculate both the steady-state maximum possible enrichment value of bulk tissue water (δM) and the nonsteady-state value of tissue water at multiple time-points (δL; Farquhar & Cernusak, 2005; Helliker & Griffiths, 2007):
where a+ is the temperature-dependent isotopic fractionation during liquid water and water vapour equilibrium (Majoube, 1971), αk is the isotopic fractionation during water vapour diffusion in air (Merlivat, 1970) and W (mol m−2) is the moss tissue water volume per leaf area.
Modelling isotopic composition of tissue water (δ18OTW) – environmental conditions
Diel and seasonal changes in the isotopic composition of Syntrichia tissue water (δL) were also modelled over one growing season using the same equation (Eqn 8), incorporating half-hourly measurements of air temperature (assumed to be equal to leaf temperature), air pressure and relative humidity recorded during 2010 at a Met-Log Automatic Weather Station (Instromet, North Walsham, UK) located in Cambridge, UK, c. 4.5 km SW of the Syntrichia growth site. Output from the isotope-enabled global circulation model IsoGSM (Yoshimura et al., 2008) was used to estimate the isotopic composition of precipitation (δP) and water vapour (δv) on a six-hourly basis at our Syntrichia collection site. The δP output of IsoGSM of −5‰ to −6‰ was similar to measured δP for Cambridge, UK, of −6.4‰ in 2004 (Reyes-García et al., 2008). After measured precipitation events, the tissue RWC was reset to a maximum of 300%, wet mass to 40 g and δL equal to δP, with conductance (gt) deemed to be proportional to wind speed when RWC > 110% (i.e. when an external water layer was present). The minimum tissue mass threshold was set at 0.4 g (i.e. 1% of wet mass). The assimilation rate of the Syntrichia tissue over time was modelled assuming that photosynthetic photon flux density (PPFD) was 2.02 times global radiation, that Syntrichia had a quantum yield of 0.002 mol (CO2) mol (photons)−1 and a saturating Amax of 0.5 μmol m−2 s−1 (values determined from laboratory measures), and that assimilation occurred when tissue RWC was between 110% and 300%. Mean values of δL over the growing season were calculated for night-time periods and for periods when the moss was at full turgor (RWC > 110%) in addition to an assimilatory flux-weighted measure (AδL).
Laboratory experiments with Syntrichia
The relative water content of Syntrichia (n = 7) declined rapidly over the first 50 min from a maximum of 500% to 200% RWC as the external water evaporated. A slower decline from 200% to 25% over the following 250 min occurred as the RWC tended towards a minimum internal water content required for continued metabolism (Fig. 1). An exponential relationship was fitted to the data (RWC = 450% × exp(−0.0105*t)) and used to estimate RWC in subsequent experiments as conditions remained consistent.
Assimilation rate (A) was dependent upon RWC, with both wet and dry conditions associated with a reduction in A (Fig. 2a). The mean value of A, as a proportion of Amax, increased between 450‰ and 300% RWC from c. 60% of Amax up to 90%. Assimilation remained over 80% of Amax until RWC fell below 100%. As RWC declined further, A fell rapidly to a minimum close to c. 20% of Amax. Therefore, the relative effect of low RWC on A was higher and more rapid than the equivalent response under high RWC.
The carbon isotope discrimination during photosynthesis (∆13C; (Eqn 4)) was dependent upon RWC (Fig. 2b). As RWC decreased from 400% to 120% dry mass, instantaneous ∆13C increased gradually from 6‰ to 14‰, before an abrupt increase between 120% and 110% RWC, supporting a ∆13C of 16‰. Following this maximum discrimination at 110% RWC, a rapid decline in ∆13C occurred as the water content further reduced. The 13CO2 discrimination data were divided into two phases – ‘wet’ (RWC ≥ 110%) and ‘dry’ (RWC < 110%) – and significant linear models were fitted to each phase, with the ‘dry’ gradient 11 times steeper than the ‘wet’ gradient (wet: RWC ≥ 110%, F = 195.5, df = 5, r2 = 0.97, P < 0.001; dry: RWC < 110%, F = 143.9, df = 2, r2 = 0.99, P = 0.05).
A significant linear relationship was also found between the oxygen isotope photosynthetic discrimination (∆18O; (Eqn 4)) and relative water content. As RWC decreased from 400% to 50%, the discrimination value increased linearly from 20‰ to 120‰ (Fig. 2d; F = 48.6, df = 8, r2 = 0.80, P < 0.001), at least partly reflecting the progressive enrichment of the tissue water (Fig. 2c).
As the RWC declined from 500% to 50%, the ∆18O of tissue water above source water increased from 3‰ to 10‰ (Fig. 2d), and this progressive evaporative enrichment was illustrated by a highly significant linear relationship (F = 48.6, df = 11, r2 = 0.80, P < 0.0001).
The δ18O value of water transpired by Syntrichia throughout the drying curve was also dependent upon RWC (Fig. 3), tending to increase as the moss dried out from −14‰VSMOW to a peak of −8‰VSMOW at an RWC of 180%, before declining again to −14‰VSMOW with further desiccation. Both steady-state (δM) and nonsteady-state (δL) models of the ∆18O of tissue water agreed well with the observations, except at the beginning of the drying curve where the nonsteady-state model provided a closer match to δTW than the steady-state model, because moss transpiration and associated isotopic effects were adjusting to the new environmental conditions.
As the Syntrichia dried out, the quantum efficiency of photosystem II decreased (Fig .4) from a maximum of 0.62, when the moss was maximally hydrated, along a plateau at 0.55 before a rapid decline below 100% RWC, a decline mirroring that observed in both assimilation rate (Fig. 2a) and Δ13C (Fig. 2b) at low RWC. Below 100% RWC the number of photosynthetically active samples remaining decreased from five to three and consequently the precision of the data decreased, although a dependence on the efficiency of PSII on RWC still continued to be apparent.
Isotopic composition of Syntrichia cellulose
The isotopic composition of Syntrichia cellulose was measured to be 21.0 ± 0.5‰VSMOW (n = 3) for δ18OC and −25.3 ± 0.1‰VPDB (n = 3) for δ13CC. This value corresponded to a flux-weighted ∆13C value of 17.8‰ (Eqn 5). The oxygen isotope composition of cellulose is dependent upon both the composition of source water (δ18OSW) and an enrichment fractionation of 27 ± 3‰ (Deniro & Epstein, 1979; Da Silveira et al., 1989; Zanazzi & Mora, 2005). Taking into account this enrichment fractionation and the measured δ18OC of 21.0‰VSMOW, δ18OSW is estimated to be c. −6‰VSMOW, similar to measured δP for Cambridge, UK, of −6.4‰VSMOW in 2004 (Reyes-García et al., 2008).
Leaf water model under field conditions
A 20-d time-series of the key meteorological parameters (rainfall, relative humidity), calculated tissue RWC, the isotopic composition of precipitation (δP), atmospheric water vapour (δV) and tissue water (δTW), and the modelled assimilation rate (A) over summer 2010 are shown in Fig. 5, and are a representation of data covering 1 yr Measured relative humidity had a diel range of c. 50%, tempered by sporadic rain events, while the IsoGSM estimates of δP and δV were consistently c. 5‰ and −17‰ respectively. Implementing the physiological dynamics established during the laboratory experiments (Fig. 1) in the context of the local conditions (as described in 'Materials and Methods' section), modelled RWC was assumed to peak after rain events, and then decline rapidly from saturation towards desiccation (Fig. 5c). The laboratory experiments showed that assimilation rate was highly sensitive to low RWC (Fig. 2a). Outside rain events, the modelled leaf water δL(Eqn 8) tracked the maximal leaf water enrichment δM ((Eqn 7); Fig. 5d), as with a small internal volume and a rapid turnover time the moss tissue reached steady-state conditions. Fluctuations in δL of over 80‰ occurred on a diel cycle, with significant evaporative enrichment in δL above δP predicted during the day, followed at night time, when relative humidity was close to 100%, by a depletion of δL to values approximately equal to δP. Assimilation rate fluctuated dependent upon the RWC of the tissue and the incident radiation intensity, with rapid increases in A, up to Amax, following precipitation (and hence tissue rehydration) during daylight hours, before declines in the assimilation rate when RWC fell below 110% (Fig. 5e). The seasonal mean value of AδL was 18.7 μmol CO2 m−2 s−1‰, compared with a mean δL of 4.1‰ at night and −2.0‰ when the moss was at full turgor. From these values of δL, the associated δ18OC could be estimated by adding the 27‰ fractionation factor associated with cellulose synthesis.
Under most environmental conditions a strong water potential gradient drives evaporation from hydrated moss tissue to the atmosphere. The only, rather limited, barriers to the evaporation of external water from moss tissue are any undulations in the surface topography (Oliver et al., 2000) or the overlap of adjacent leaflets. For desiccation-tolerant mosses growing without a continuous water supply, brief periods of hydration and metabolic activity must be exploited for net carbon gain and growth before the onset of desiccation. The initial rapid decline in S. ruralis relative water content (Fig. 1) was attributable to evaporating external water. Following the loss of external water, the internal water evaporated at a slower rate until only tightly bound metabolic water remained and photosynthesis ceased (Proctor et al., 1998). Despite this extreme desiccation, Syntrichia tissue remains capable of rapidly reinitiating photosynthesis when water becomes available (Oliver et al., 2000).
The peak in Δ13C at 110% RWC (Fig. 2b) was associated with the transition between the evaporation of external and internal tissue water. At this point CO2 diffusion limitation is minimal, yet sufficient cellular and metabolic water remains for diffusive supply to maximize photosynthetic reactions (Rice & Giles, 1996; Proctor et al., 1998; Meyer et al., 2008). The slow increase in discrimination from 8‰ to 16‰ between 400% and 110% RWC was a direct response of the reduction in diffusion limitation as the external water layer evaporated. Photosynthesis also responded to the reduction in diffusion limitation for RWC values above 350% and reached 90% of Amax below this threshold (Fig. 2a), suggesting that in the presence of an external water film A and Δ13C co-vary in a nonlinear fashion. The mean maximum discrimination of 16‰ was lower than the theoretical maximum value of 29‰ in C3 plants when fractionation by Rubisco is the limiting factor (O'Leary, 1988), but similar to the maximum values measured in liverworts and consistent with diffusion limitation in the absence of intercellular air spaces (Meyer et al., 2008).
Bryophytes assimilate carbon and grow more slowly than their vascular counterparts, but with the flexibility conferred by their poikilohydric characteristics they are able to exploit a wide range of niches and thrive in environments with irregular, unpredictable periods of free water availability (Proctor et al., 2007). This was exemplified in Syntrichia by the 200% range of RWC over which the carbon assimilation rate was maintained at c. 90% of Amax (Fig. 2a), although the absolute value of Amax (c. 0.5 μmol(CO2) m−2 s−1) was at least an order of magnitude below that measured in many vascular plants. At high RWC quantum efficiency was maximal (Fig. 4) but photosynthesis was only 60% of Amax. This suggested that the light-harvesting apparatus was fully operational, but that photosynthesis was limited by the internal conductance of CO2 to chloroplasts. By contrast, the three indicators of photosynthetic capability, namely assimilation rate, quantum efficiency and ∆13C, all declined extensively and rapidly when RWC was below 100%, as metabolic reactions were compromised because of desiccation. The quantum efficiency of Syntrichia showed a similar relationship with RWC to that observed in Sphagnum species in both Alaska (Murray et al., 1989) and New Zealand (Maseyk et al., 1999).
The linear increase in ∆H218OTW as RWC decreased (Fig. 2d) reflected the isotopic enrichment that occurs in evaporating water pools (Craig & Gordon, 1965). The fact that this linear relationship holds even at low RWC values, with no isotopic shift at the implied transition between the pools (110% RWC), may indicate that the internal and external tissue water pools are well mixed, although with little metabolic activity and associated fractionation at low RWC values the linear relationship could alternatively indicate that evaporation is the dominant fractionating factor. The fact that ∆C18O16O in CO2 also increased linearly with decreasing RWC (Fig. 2c) is most likely the consequence of rapid oxygen isotope exchange between CO2 and tissue water catalysed by carbonic anhydrase activity, even at the lowest range of RWC values.
The transpired water vapour was depleted in heavy isotopes compared with the tissue water (Fig. 3) because of both kinetic (H218O has a lower diffusivity than H216O) and equilibrium effects (H218O has a lower saturating vapour pressure than H216O). The output of both the steady-state (SS, δM) and nonsteady-state (NSS, δL) models predicted values close to the measured δ18OTW. However, particularly at high RWC, the NSS model was in closer agreement with the measurements than the SS model. This is because, in the absence of a constant water influx into the system, isotopic steady state was unlikely to be reached at each successive measurement point. Therefore, when environmental conditions are well characterized, the NSS model provides a suitable framework for the prediction of the isotopic composition of moss tissue water.
The stable isotopic composition of Syntrichia cellulose represents an integrated signal related to the conditions in which the moss was growing during years of development in the natural environment. The estimated source water input during organic matter synthesis of −6‰, calculated from δ18OC, fitted with the weighted annual isotopic composition of precipitation measured at Keyworth, Nottinghamshire, UK (c. 130 km NW of Syntrichia source), during 2005–2006 of −6.5‰ (Jones et al., 2007) and the measured isotopic composition of precipitation in Cambridge, UK, during 2004 of −6.4‰ (Reyes-García et al., 2008). Given the evaporative enrichment of tissue water during drying events (Fig. 2d), this implies that Syntrichia cellulose synthesis only seems to have occurred during times of saturation (i.e. following rainfall and perhaps in response to early morning dew), when, having been recharged yet undergone minimal evaporative enrichment, tissue water was isotopically similar to source water.
In contrast, δ13CC indicated that the majority of net carbon assimilation occurred during times of minimal diffusion resistance, with the estimated ∆13C value of 17.8‰ slightly higher than the maximum measured online carbon discrimination of 16‰, which occurred when the RWC had declined from saturation, to a substantially drier 110%. By this point the oxygen isotopic composition of the tissue water had undergone evaporative enrichment of 5‰ under laboratory conditions.
The contrasting δ13C and δ18O signals of the cellulose raise the intriguing possibility of temporal separation between optimal conditions for carbon assimilation (low RWC, minimal diffusive resistance, immediately before desiccation) and cellulose synthesis (high RWC, during or immediately after saturation). The optimal period of carbon assimilation is ended by one of two phenomena, both of which have been experimentally shown to cause a reduction in assimilation rate and Δ13C (Rice & Giles, 1996; Williams & Flanagan, 1996; Meyer et al., 2008). First, desiccation-related metabolic shut-down, would include the slowing, and eventual cessation, of assimilation and growth. Second, a precipitation (or other water saturating) event, that recharges internal and external water pools of the moss tissue. Saturation would increase the diffusion resistance (reducing A and Δ13C) while metabolic reactivation facilitates cellulose synthesis (from the carbon products assimilated when CO2 conductance was high) in the presence of water with an isotopic composition equal to that of precipitation. A consequence of the desiccation scenario would be the presence of a high concentration of soluble sugars during periods of metabolic shutdown, and indeed a pool of sucrose contributing up to 18% DW, has been measured in moss species, including S. ruralis (Smirnoff, 1992). Although osmotica do play a role in desiccation tolerance (Proctor et al., 2007), previous studies found that the concentration of sucrose was not correlated directly with the relative degree of desiccation tolerance in six moss species and that the sucrose concentration in Syntrichia ruralis varied little with water status (Smirnoff, 1992). Thus the temporal separation of assimilation and growth, supported by our data, may not rely completely on a significantly different concentration of sucrose reserves between the two phases and instead indicate that both reasonably high glucose concentrations and optimal turgor conditions are required for the upregulation of enzymes required to synthesize cellulose (e.g. cellulose synthase) and to provide the necessary cellular pressure to kick-start cambial activity. Further molecular investigations on the temporal phasing of the production of key enzymes responsible for cellulose synthesis would help confirm the underlying cause of this temporal separation of growth and assimilation.
The environmental feasibility of this ‘temporal separation’ hypothesis was investigated by using meteorological data to predict the changes in δ18OTW diurnally and seasonally (Fig. 5). The model output showed that the RWC of the moss decreased rapidly from the maxima at times of saturation to levels too low for net assimilation (A), with the extent of ∆13C proportional to A throughout these daytime declines in RWC. As most assimilation occurred under highly discriminating, rapidly drying conditions this model output is consistent with the measured δ13CC value.
In phase with the measured diel cycle in relative humidity, the model output (Fig. 5) indicated significant isotopic enrichment of leaf water (δL) during the day, followed by depletion to a level approximately equal to δP overnight. In addition, the small volume of leaf water and rapid turnover time resulted in δL being very similar to the steady-state value of δM. If the overnight increase in relative humidity provided sufficient moisture to activate metabolic machinery, there would be little or no light to stimulate photosynthetic machinery (and hence no assimilation). However, soluble sugars could be converted into cellulose with the carbonyl oxygen atoms exposed during the metabolic processes exchanging with water molecules (Sternberg et al., 2006) in equilibrium with, or, in the case of atmospheric water vapour, more isotopically depleted than, δP, as has been previously shown in Tillandsia usneoides (Helliker & Griffiths, 2007). The measured δ18OC value was more consistent with the modelled isotopic composition of tissue water during periods of full turgor than the more enriched flux-weighted tissue water composition (AδL), suggesting that cellulose synthesis was temporally independent from periods of assimilation.
From a physiological perspective it is important to consider that all daytime desiccation events will be preceded, however briefly, by conditions optimal for maximal assimilation. At this point, with a precipitous metabolic and photosynthetic decline looming (as measured by A and Φ), investment in growth is a high-risk strategy, whilst the maintenance of soluble sugars as both osmotica and respiratory substrates for survival and post-desiccation recovery are essential. Following resaturation, the pool of sugars remains, while assimilation is limited both by the need to reactivate photosynthetic machinery and the CO2 diffusion limitation; it is at this point we hypothesize an upregulation in the production of enzymes involved in cellulose synthesis. Thereafter, as the external liquid phase evaporates there is a gradual increase in CO2 conductance and assimilation and a gradual decrease in cell turgor. At this point, investment in the synthesis of cellulose is stopped and the sucrose pool is then replenished by maximum carbon assimilation rates. If the desiccation persists then an optimal concentration of compatible solutes will be synthesized to ensure adequate protection of the metabolic apparatus and DNA during the quiescent period until wet conditions return. As the growth rate of moss is slow it is likely that genetic studies will be required to confirm exactly when during the rewetting and drying cycle of mosses cellulose synthesis takes place (Roberts et al., 2012).
These experiments confirm the importance of RWC to the instantaneous stable isotope composition of carbon and oxygen in moss tissue, and to the integrated cellulose signal that reflects periods of net assimilation. While fundamentally an interesting phenomenon, the possible temporal separation of carbohydrate utilization discussed here has implications for the interpretation of peatland stable isotope palaeoclimate records, as the majority of both net assimilation of carbon and conversion into preserved organic matter will occur during periods of ‘optimal hydration’, a state that may differ for the two processes (CO2 assimilation and cellulose synthesis) and that will not necessarily reflect mean climatic conditions.
J.R. is funded by NERC (NE/G523539/1, AFI11_05). P.C. and D.A.H. contribute to the BAS ‘Polar Science for Planet Earth’ core research programme. Salary for L.W. was supported through a Natural Environment Research Council Advanced Fellowship (NE/G014418/1) and a Marie Curie Intra-European Career Development Fellowship. Research leading to the results has received funding from the European Community's Seventh Framework Programme (FP7/2007-2013) under agreement no. (237582). The INRA department EFPA is gratefully acknowledged for funding the sabbatical of J.O. in Cambridge that resulted in the development of this manuscript. We also thank Kei Yoshimura for kindly providing ISOGSM outputs.