Physiological and photosynthetic plasticity in the amphibious , freshwater plant , Littorella uniflora , during the transition from aquatic to dry terrestrial environments

The physiological and photosynthetic responses of Littorella uniflora (L.) Ascherson, an amphibious macrophyte of isoetid life form, to rapid and prolonged emersion onto dry land, was studied at a reservoir. Water relations were little affected in the short term, but declining water potential and turgor pressure indicated water stress after flowering. High leaf lacunal CO2 concentrations suggested continued CO2 uptake from sediments. In contrast, a switch from Crassulacean acid metabolism (CAM) to C3 photosynthesis was indicated by much lower levels of DH + (down minus dusk titratable acidity) and phosphoenolpyruvate carboxylase (PEPC) activity in new terrestrial leaves, 7‐8fold higher activity of ribulose bisphosphate carboxylase oxygenase (Rubisco), and increased chlorophyll and soluble protein contents. Accumulated nitrate and amino acid pools were depleted, whereas storage of carbohydrates as soluble sugars, fructan and starch increased. Plant carbon and nitrogen isotope ratios (d 13 C and d 15 N) declined, perhaps reflecting changes in C fixation processes, N metabolism, and source C and N. In leaves of plants grown half-emersed for an extended period, contrasting activities of PEPC and Rubisco were found in submersed and emersed portions. Overall, L. uniflora showed considerable phenotypic plasticity, yet seemed to remain poised for resubmersion; these characteristics could be adaptive in the unpredictable water margin habitat.


INTRODUCTION
Amphibious plants form a significant part of temperate and tropical herbaceous floras (e.g.Sand-Jensen & Frost-Christensen 1999), and are an important component of aquatic and wetland ecosystems (see Sculthorpe 1967;Submersed, aquatic L. uniflora shows both physical and biochemical CO 2 -concentrating mechanisms.Thus, CO 2 uptake is predominantly via the roots from CO 2 -enriched sediments; CO 2 diffuses in the gas phase through the lacunal system to the leaves (Söndergaard & Sand-Jensen 1979;Boston et al. 1987a,b;Robe & Griffiths 1990, 1998).In addition, daytime CO 2 supply is augmented by Crassulacean acid metabolism, a pathway traditionally associated with water conservation in plants in arid habitats (Winter & Smith 1996).CAM involves night-time fixation of CO 2 by phosphoenolpyruvate carboxylase (PEPC), storage as malic acid, and regeneration in the light for refixation by ribulose bisphosphate carboxylase oxygenase (Rubisco) and the C 3 pathway (Smith, Boston & Adams 1985;Madsen 1987a;Robe & Griffiths 1990).In the leaf lacunae, which are lined with photosynthetic cells (e.g.see Robe & Griffiths 1998), CO 2 concentrations range from 0•42 to 2•7 mol m -3 (1-6%; Madsen 1987a,b;Robe & Griffiths 1988, 1990, 1992), sufficient to saturate photosynthesis in vitro and suppress the oxygenase activity of Rubisco.
Emergent L. uniflora collected from close to the water's edge, with leaves similar to those of aquatic plants, and incubated with roots submersed, are photosynthetically very similar to aquatic plants; shoots perform CAM, and CO 2 uptake is via the roots (Farmer & Spence 1985;Nielsen, Gacia & Sand-Jensen 1991;Nielsen & Sand-Jensen 1997).However, the very different leaves of terrestrial L. uniflora growing on dry land do not exhibit CAM (very low overnight acidification and PEPC activity) provided relative humidity is low (Aulio 1986;Groenhof, Smirnoff & Bryant et al. 1988), and in the isoetid Isoetes howellii CAM is lost from the leaf tips as they emerge (Keeley & Busch 1984).There is also one report that terrestrial leaves of L. uniflora have a twofold higher Rubisco activity than aquatic leaves (Beer et al. 1991).
As there was no clear picture of how L. uniflora responds to emersion when growing in situ at a single location, and, in particular, no information on how photosynthetic characteristics change in relation to environmental conditions and leaf form, we monitored lacunal CO 2 concentrations, titratable acidity (a measure of CAM), and PEPC and Rubisco activities as part of our study at a reservoir (Robe & Griffiths 1998).The effects of emersion on water relations, and reserves of soluble carbohydrate and nitrogen which may buffer against changing environmental conditions (e.g.Millard 1988;Chapin, Schulze & Mooney 1990;Staswick 1994), were also investigated.Plant carbon isotope ratios (d 13 C) were determined, as although they provide much insight into CO 2 sources and photosynthetic processes in terrestrial species (O'Leary 1988;Farquhar, Ehleringer & Hubick 1989), their application in aquatic macrophyte studies is still in its early stages (see Keeley & Sandquist 1992).Plant nitrogen isotope ratios (d 15 N) which, with a better understanding of the complex underlying factors, have the potential to inform about N sources and metabolism (Handley & Raven 1992), were also determined as there were no data for an amphibious macrophyte.Our aim was to obtain an integrated picture of 1042 W.E. Robe & H. Griffiths physiological and photosynthetic responses of L. uniflora to rapid and prolonged emersion.

Field site, plants and programme of sampling
This study was carried out during the warm and dry summer and early autumn of 1995 at Thirlmere Reservoir, Cumbria, UK (grid ref NY 323133).Environmental conditions and the appearance of the plants on the three sampling dates are illustrated in Table 1 and Fig. 1 of Robe & Griffiths (1998).Plants were sampled (i) in early June, when they were in very shallow water just before emersion (the water surface was a few centimetres above the leaf tips); these plants are referred to as 'aquatic' L. uniflora, (ii) in early July, 3-4 four weeks after emersion ('flowering' L. uniflora), and (iii) in early September after 3 months on dry land ('seed-bearing' plants).Particular care was taken to sample plants from areas of similar plant density and sediment type on each occasions, and to take plants of similar appearance within each area.For dawn and dusk measurements, adjacent clumps were sampled.The sampling areas were protected from grazing by wildfowl with a layer of fine, open-work plastic netting supported 0•4 m above the plants by wooden posts.

Sediment and plant water status
For determination of sediment water content, sediments taken from around the roots of the plants were weighed and dried to constant weight at 60 °C.For fresh weight (FW) to dry weight (DW) ratios and percentage water content of shoots, roots and stems, plant material was gently washed, blotted dry and dried to constant weight at 60 °C.For measurement of osmotic potential, shoots and roots were frozen in liquid nitrogen at midday, the cell sap was then expressed from the thawed material and centrifuged in a microfuge for 2 min.Determinations were made by the cryoscopic method using an Osmomat 030 (Gonotec, Berlin, Germany).For determination of water potential, plants were brought back to the laboratory at dusk, undisturbed, in large clumps of their own sediments and maintained under controlled conditions (temperature 18 °C, photosynthetic photon flux density in the 400-700 NM range (PPFD) 200 mmol m -2 s -1 , natural photoperiod) for 1 day.Whole shoots were cut from the stem at midday, quickly and carefully dried, sealed at the base with petroleum jelly, and then weighed and submersed in a graded series of mannitol concentrations in small test tubes.The water potential of the shoots was taken as that of the solution in which there was no change in fresh weight after 45 min (see Meidner & Sheriff 1976).

Leaf lacunal CO 2 concentrations, [CO 2 ] i
Plants were brought back to the laboratory at dusk, undisturbed in large clumps of their own sediment, and main-tained for 1 day under controlled conditions as described above.Leaf lacunal CO 2 concentrations, [CO 2 ] i , were determined during a 4 h period around midday.Mature healthy leaves were carefully removed, sealed with a thin coating of petroleum jelly, and inserted into a short section of rubber tubing connected to an infra-red gas analyser (ADC 225.MK3, Hoddesdon, Hertfordshire, UK).The tubing was briefly flushed with CO 2 -free air, then sealed and interlacunal gases expelled by pressure; the lacunal gases were then flushed through the infra-red gas analyser with CO 2free air (see Robe & Griffiths 1988).Lacunal volume was determined by two methods: (a) as the difference in the volume of water displaced by the leaves before and after expelling lacunal gases, and (b) from a comparison of total and lacunal surface areas estimated from photographs of leaf transverse sections, using the technique of stereology (Steer 1981).Both methods gave very similar results.

Titratable acidity
Whole shoots were frozen in liquid nitrogen at dusk and dawn.At dusk, plants were sampled in situ at Thirlmere.For dawn sampling, large clumps of plants in their own sediment were taken from the site after dusk, transported to a garden, and kept in the open air until dawn.On removal from the liquid nitrogen, cell sap was expressed from the thawed material and titrated against NaOH with phenolphthalein as indicator (see Robe & Griffiths 1988 for details)

Rubisco and PEPC assays
Plants growing in situ were sampled at midday.Photon flux density, temperature and relative humidity around midday on the three sampling days are shown in Table 1 of Robe & Griffiths (1998).Youngest mature leaves were frozen in liquid nitrogen within 2-3 s of removal from the shoots.Both enzymes were assayed spectrophotometrically.The method used for extraction and assay of Rubisco was based on that described by Besford (1984) with adaptations from Lilley & Walker (1974), Ward & Keys (1989), Quick et al. (1991) and Sharkey, Savitch & Butz (1991).Only the distal halves of the leaves, minus the tips, were used for the assays (0•4 and 0•1 g of submersed and terrestrial leaf tissue, respectively).Leaves were taken from the liquid nitrogen, cut and weighed while still frozen and dropped into 2 cm 3 of ice-cold buffer (100 mol m -3 HEPES adjusted to pH 7•8 with KOH, 5 mol m -3 DTT, 0•2 mol m -3 EGTA and 0•06 g Polyclar AT, flushed with CO 2 -free nitrogen) in a prefrozen mortar and pestle standing on ice.The tissue was very quickly homogenized with a little acid-washed sand and centrifuged at 16 000 g (at 2-4 °C) for 30 s.The extract was then divided into two.Half was snap-frozen (in four 110 mm 3 aliquots contained in Eppendorf vials) and stored in liquid nitrogen for measurement of initial activity.The enzyme in the remaining half was activated on ice with 20 mol m -3 KHCO 3 -and 25 mol m -3 MgCl 2 -.After 45 min, the activated extract was also snap-frozen, as above.The enzyme was assayed by the continuous measurement of 3-Phenotypic plasticity in amphibious Littorella uniflora 1043 PGA-dependent NADH oxidation in a coupled enzyme system.The freshly prepared reaction mixture (890 mm 3 in cuvettes of 10 mm path length) contained 100 mol m -3 HEPES-KOH adjusted to pH 7•8, 20 mol m -3 MgCl 2 , 5 mol m -3 DTT, 0•2 mol m -3 EGTA, 0•25 mol m -3 NADH, 3•5 mol m -3 ATP, 3•5 mol m -3 phosphocreatine, 10 units of 3-phosphoglycerate phosphokinase, 6 units of glyceraldehyde 3-phosphate dehydrogense, 16 units of creatine phosphokinase, 25 mol m -3 KHCO 3 -and 0•4 mol m -3 ribulose bisphosphate (RuBP).The solution was allowed to stand for a few minutes to reach 20 °C and the reaction was started with the addition of 110 mm 3 of freshly thawed extract.Measurements were made at 340 nm using a Unicam 8700 series spectrophotometer (Unicam Ltd, Cambridge, UK), and the reaction was linear for at least 1•5-2 min, sometimes 3 min.Each extract was assayed three times for both initial and total (fully activated) activity.Control assays without RuBP gave a very low activity, and this was subtracted when calculating the rate of CO 2 fixation, using an absorption coefficient of 0•63 mol -1 mm -1 , according to Bergmeyer (1983).
The assay for phosphoenolpyruvate carboxylase (PEPC) activity was based on that of Borland & Griffiths (1992).Distal leaf tissue (0•2 g, weighed frozen as above) was quickly homogenized in 2 cm 3 of ice-cold extraction buffer identical to that used for the Rubisco assay, but with the addition of 5 mol m -3 MgCl 2. The crude homogenate was centrifuged at 16 000 g for 30 s and de-salted, to remove malate, by passage through a pre-packed column (bed volume 9•1 cm 3 , bed height 5 cm 3 ) of Sephadex G25-M (Pharmacia Biotech Ltd, Milton Keynes, UK) held at a temperature of 2-4 °C, and the eluate used immediately for enzyme analysis.The reaction mixture (890 mm 3 in cuvettes of 10 mm path length) contained 100 mol m -3 Bicine, pH 8•2, 5 mol m -3 MgCl 2 , 0•16 mol m -3 NADH, 10 units malic dehydrogensase, 10 mol m -3 KHCO 3 and 2 mol m -3 PEP.The mixture was allowed to warm to 20 °C and the reaction was initiated by the addition of 110 mm 3 of extract.Measurements were made at 340 nm and the reaction was linear for at least 6 min.Control assays without PEP showed a very low activity which was subtracted when calculating the rate of CO 2 fixation as described above.Soluble protein content of the extracts was determined by the method of Bradford (1976).All reagents and enzymes for both assays were purchased from Sigma.

Chlorophyll
Leaf tissue for chlorophyll determinations was also frozen in situ in liquid nitrogen.Chlorophyll was extracted from 50 mg of thawed tissue (distal half of leaves minus the tip) in 80% acetone, and the absorbance of centrifuged samples measured at 665 and 649 nm (Vernon & Seely 1966).

Carbohydrates
Whole plants (shoots, stem and as much of the root system as could be obtained) were frozen in liquid nitrogen at dawn and dusk as described for titratable acidity above and stored at -80 °C.Plants were taken from the same area of the site on each date and adjacent clumps were sampled for dawn and dusk determinations.Ethanol-soluble neutral sugars, warm water-soluble fructans, and starch were successively extracted using methods closely based on those described by Borland & Farrar (1985) and Farrar (1980).Whole shoots (minus any senescing leaves), roots and stems were thawed, weighed and homogenized in 95% ethanol.The extract was then incubated at 80 °C for 30 min, centrifuged at 20 000 g for 10 min and the supernatant removed and stored at 4 °C.The pellet was re-suspended and the extraction repeated twice more.The supernatants were combined and made up to a final volume of 25 cm 3 .The pellets were dried at 40 °C overnight and re-suspended in acetate buffer at pH 4•5.The extract was incubated at 30 °C for 8 h, centrifuged as described above and the supernatant removed and stored.The pellet was re-suspended and the extraction repeated twice more, the final volume of combined supernatants being made up to 25 cm 3 .The pellet was finally re-suspended in 6 cm 3 acetate buffer pH 4•5 containing in excess of 100 units of amyloglucosidase (22 500 units g -1 from Rhizopus; Sigma) and incubated at 45 °C for 24 h.The extract was centrifuged at 20 000 g and the supernatant removed for assay.Aliquots of the ethanol, acetate buffer and amyloglucosidase extracts were assayed for carbohydrate by the method of Dubois et al. (1956).

Soluble nitrogen
Whole plants, all from the same area of the site, were frozen in liquid nitrogen in situ at Thirlmere at midday.Shoots, roots and stems were stored individually at -80 °C.Nitrate and free amino and soluble protein were extracted and determined spectrophotometrically as described by Robe & Griffiths (1994).

Stable isotope ratios (d 13 C and d 15 N), and total C and N analysis
Plant material, and sediments from around the roots, were dried at 60 °C and ground to a fine powder.Analysis of 13 C and 15 N natural abundance, and total C and N content, was carried out at Merlewood Research Station, Grange-over-Sands, Cumbria, using a Europa 20-20 isotope ratio mass spectrometer (Europa Scientific, Crewe, UK). 13 C and 15 N are stable isotopes of C and N which occur naturally in low abundance (1•1% in the case of 13 C and 0•366% in the case of N). d 13 C and d 15 N denote parts per thousand deviations, ‰, from the ratios 13 C: 12 C and 15 N: 14 N in a standard (a secondary standard referred to the original Pee Dee limestone in the case of C, and atmospheric N 2 in the case of N), and are calculated as: d (‰) = ((R sample /R standard ) -1) ¥ 1000, where R is the ratio 13 C: 12 C or 15 N: 14 N. Small variations in natural abundance (fractionations) occur during physical, chemical and enzymatic processes (Farquhar et al. 1989;Griffiths 1998).

Experimental partial emersion
Submersed L. uniflora, still rooted in large clumps of sediment, were removed from Thirlmere in May and maintained in tanks, with the water level at the tips of the original aquatic leaves (water surface 5 cm above the sediments).The tanks were kept outdoors in a semi-shaded position.New longer (10 cm), thinner leaves grew with their distal halves emersed and lower halves submersed.Measurements were made in September.

Water status of sediments and plants
Sediments dried out, particularly in the first 3-4 weeks after emersion (Fig. 1a).L. uniflora also lost water most rapidly just after emersion.Thus, fresh weight to dry weight ratios (Fig. 1b) and water content (Fig. 1c) declined.Osmotic potentials in shoots and roots (Fig. 1d) increased slightly just after emersion, but decreased substantially after flowering.Shoot water potential (Fig. 1d) declined slightly in the first 3-4 weeks after emersion, but more quickly after flowering.Shoot turgor pressure (calculated as water potential minus osmotic potential; Fig. 1d) fell most quickly just after emersion.

Leaf lacunal CO 2 concentration, CAM and activities of PEPC and Rubisco
Leaf lacunal CO 2 concentration, [CO 2 ] i , increased after emersion (Fig. 2a).[CO 2 ] i was 1•7 mol m -3 , equivalent to 3•8% in air, in the lacunae of aquatic L. uniflora, 3•1 mol m -3 , equivalent to 6•9% CO 2 in air, in the lacunae of the flowering stage, and 7•0 mol m -3 , equivalent to 15•7%, in the seed-bearing stage.In contrast, CAM activity was lost after emersion (Fig. 2b).In leaves of CAM species, the concentration of cell sap titratable acidity (H + ) reflects the overnight storage of CO 2 as malic acid (see Winter & Smith 1996).Shoots of submersed, aquatic L. uniflora showed a relatively high dawn H + and low dusk H + (Fig. 2b), and the DH + (dawn minus dusk titratable acidity) was 58 mmol H + g -1 FW, signifying a moderate level of CAM (see Robe & Griffiths 1990, 1992).After only 3-4 days of emersion, CAM activity in aquatic leaves was reduced, by 70%, to 17 mmol H + g -1 FW (not shown).The new terrestrial leaves of flowering L. uniflora showed no CAM, with a DH + of only 0•9 mmol H + g -1 FW.In seed-bearing L. uniflora, the DH + , although increased slightly to 6•1 mmol H + g -1 FW, also indicated negligible CAM.
As shown in Fig. 2(c), the pattern of activity of maximum extractable PEPC mirrored DH + , being relatively high in shoots of aquatic L uniflora and very low in leaves of the flowering and seed-bearing stages.Both initial activity of Rubisco (determined immediately after homogenization) and total activity (after full activation in the presence of activators CO 2 and Mg 2+ ) were determined.Initial and total activities of the Rubisco from L. uniflora increased 7-8-fold aquatic, flowering and seed-bearing L. uniflora, respectively.
The chlorophyll (a and b) content of the distal halves of youngest mature leaves, minus the tip (the tissue used for PEPC and Rubisco assays) was 43% higher in flowering than in aquatic L. uniflora (Fig. 2d).Soluble protein concentration was also 3•5-fold higher, but declined after flowering (Fig. 2d).Activities of PEPC and Rubisco in L. uniflora expressed on chlorophyll and protein bases exhibited similar seasonal trends to those found when the data are expressed on a fresh weight basis, for example Rubisco total activity was 1•25, 5•31 and 2•38 mmol CO 2 mg -1 chlorophyll min -1 and 0•15, 0•37 and 0•35 mmol CO 2 mg -1 soluble protein min -1 in aquatic, flowering and seed-bearing stages, respectively.
Figure 3 depicts DH + , and the activities of PEPC and Rubisco, in the upper and lower halves of the partially emersed leaves of L. uniflora grown outdoors in tanks for 4 months (see Materials and Methods).The submersed portions showed moderate levels of CAM activity but low Rubisco activity.In contrast, the emersed portions, exposed to the air, showed little CAM activity but high levels of Rubisco activity.The activation state of Rubisco was 54% in the submersed halves and 74% in the emersed halves.

Carbohydrate
The concentrations of neutral sugars, fructans and starch, at dawn and dusk, increased after emersion, particularly in roots and stems, but at different rates and to different degrees (Fig. 4).In the first 3-4 weeks, neutral sugars (Fig. 4a) accumulated most rapidly, being 8-and 2•3-fold higher in roots and stems, respectively, at dawn in flowering L. uniflora (e.g.60% of total C compared with 14% in Mean for 10 plants with standard deviation (SD).There was a significant difference between fresh weight to dry weight ratios in June and July, and in July and September; P < 0•001).(c) Water content of shoots (), roots (᭝) and stems (᭹) expressed as a percentage calculated from fresh weight to dry weight ratios.Mean for 10 plants.(d) Shoot osmotic potential (), water potential (ᮀ), and turgor pressure (᭹) calculated as osmotic potential minus water potential, and osmotic potential of roots (᭡).Mean for four plants at each sampling date, shown ± SD for osmotic potentials.The osmotic potential of shoots, but not roots, was significantly higher in flowering compared with aquatic L. uniflora (P < 0•001).In both shoots and roots, osmotic potentials in seed-bearing plants were significantly lower than in the flowering stage (P < 0•001).In some cases, the error bar is masked by the symbol.
after emersion, but declined again after flowering.The activation state of the enzyme (the proportion of initial activity to total activity expressed as a percentage) was 62% in aquatic L. uniflora, 55% in the flowering stage and 66% in the seed-bearing stage.The ratios of (fully activated) Rubisco to PEPC activities were 0•45, 32•2 and 10•0 in Mean values for 20-25 leaves from 6-7 plants with standard deviations (SD).The [CO 2 ] i of flowering L. uniflora was significantly higher than that of the aquatic stage (P = 0•003), while that of seed-bearing L. uniflora was significantly higher than that of the flowering stage (P < 0•001).(b) Dawn (᭹) and dusk (᭺) titratable acidity of five shoots ± SD.Titratable acidity at dawn in seed-bearing L. uniflora was significantly higher than in the flowering stage (P < 0•001) (c) Activities of PEPC (᭝) and Rubisco enzymes.The data for Rubisco show both initial activity, immediately after extraction (᭹), and total activity, after full activation in the presence of Mg 2+ and CO 2 ().Results, shown ± SD, are for 4-6 extracts using the distal half (minus the tip) of 6-10 youngest mature leaves.Initial and total activities were significantly different at each sampling date (P < 0•001).(d) Chlorophyll (a + b) content, and (e) soluble protein concentration, of the distal halves (minus the tip) of youngest mature leaves.Mean ± SD of five extracts.In some cases, the error bar is masked by the symbol.
Figure 3. CAM activity (as DH + ; dawn minus dusk titratable acidity) and activities of PEPC and Rubisco in partially emersed leaves of Littorella uniflora from Thirlmere, grown in tanks in a garden.Submersed L. uniflora, still rooted in large clumps of sediment, were removed from Thirlmere in May, and maintained in tanks with the water level at the tips of the original aquatic leaves (water surface 5 cm above the sediment).The tanks were kept outdoors in a semi-shaded position.New longer (10 cm), thinner leaves grew with their distal halves emersed and lower halves submersed.Both halves of the leaves were green and healthy.Measurements for upper and lower portions of six leaves combined were made in September.roots).In contrast, fructans (Fig. 4b) and starch (Fig. 4c) built up in concentration mainly after flowering.Thus, the starch concentrations of roots and stems were 4•3 and 6•5 times higher, respectively, at dawn in seed-bearing, compared with flowering L. uniflora (e.g.52% of total C com-pared with 14% in roots).For June and July samplings, carbohydrate concentrations were generally higher at dusk than at dawn (Fig. 4a-c).In September, however, concentrations of carbohydrates in stems and roots were lower at dusk than at dawn.

Soluble nitrogen
Nitrate concentrations in shoots, roots and stems (Fig. 5a) declined sharply in the first 3-4 weeks after emersion, and were at almost undetectable levels in these tissues of seedbearing L. uniflora.Free amino acid concentrations also declined most rapidly in the first 3-4 weeks (Fig. 5b).In aquatic L. uniflora, nitrate and free amino acid nitrogen comprised 7•4 and 4%, respectively, of total nitrogen in shoots, 3 and 4% in stems, and 12•5 and 8% in roots.

Plant water relations
Emersion had little immediate effect on plant water relations.Water and osmotic potentials remained at the high end of the range for aquatic and succulent terrestrial species (Milburn 1979;Larcher 1983), and there was only a small drop in turgor.The fact that L. uniflora has a well developed vascular system (Hostrup & Wiegleb 1991) and that water was probably still freely available in the sediments perhaps buffered against the full dehydrating effects of emersion; there was a strong positive correlation between turgor pressure and soil water content (r = 0•996, P = 0•01).However, there may also have been changes in cuticle thickness and/or permeability, and control of water loss at the membrane level (e.g.Chrispeels & Maurel 1994).It seems that water stress was not experienced until after flowering, when leaf water potential decreased more quickly (Fig. 1d).Osmotic adjustment in leaves and particularly roots, suggested by the decrease in osmotic potentials (Morgan 1984), did not prevent loss of water and turgor.Water stress may have been a cause of the abrupt drop in growth after flowering (Fig. 5a; Robe & Griffiths 1998).

Lacunal CO 2 concentrations
The increase in [CO 2 ] i after emersion (Fig. 2a) suggests that sediments may have been an important source of CO 2 for the fully terrestrial form of L. uniflora growing on dry land at Thirlmere (see also Nielsen et al. 1991).As the sediments became aerobic, carbon breakdown may have increased, and CO 2 could continue to diffuse into the root lacunae and via the stems to the reduced lacunal system in the leaves, eventually escaping through stomates or cuticle.The relative importance of sediment and atmospheric CO 2 supply for terrestrial L. uniflora might depend on sediment type; sediments with a higher proportion of gravel probably contain less CO 2 (e.g.Robe & Griffiths 1988).The residual lacunal system could allow gas exchange with the sediments to continue on re-submersion, so that a period of reliance on bulk water CO 2 uptake and anaerobic metabolism in the roots (e.g.ap Rees et al. 1987; see also Jackson & Armstrong 1999) would be avoided.

CAM and C 3 photosynthesis: PEPC and Rubisco activity
In aquatic L. uniflora carrying out moderate levels of CAM, PEPC activity, assayed in the absence of malate, exceeded Rubisco activity (Fig. 2c), and the ratio of fully activated Rubisco activity to PEPC activity was 0•45, similar to that in terrestrial CAM species (see Dittrich, Campbell & Black 1973), although lower than previously reported for L. uniflora (Farmer, Maberly & Bowes 1986;Baattrup-Pedersen & Madsen 1999) and other aquatic CAM species (Keeley 1999).PEPC activity was in excess of that predicted from the rate of overnight CO 2 fixation calculated from DH + (by 22-fold; 1•5 compared with 0•069 mmol CO 2 g -1 FW min -1 ) Phenotypic plasticity in amphibious Littorella uniflora 1049 as also reported for terrestrial CAM species (see Borland & Griffiths 1992).In aquatic L. uniflora, Rubisco total activity (0•677 mmol CO 2 g -1 FW min -1 or 40•6 mmol CO 2 g -1 FW h -1 ) closely matched maximum rates of photosynthesis in thin tissue slices incubated with rapid stirring and saturating CO 2 and PPFD (35-40 mmol O 2 g -1 FW h -1 ; see Robe & Griffiths 1992).Rates of photosynthesis in intact plants incubated under near-natural conditions (50-300 mmol m -2 s -1 incident PPFD) are lower (4-to 12-fold: Robe & Griffiths 1990), probably due to a combination of low incident and within-leaf light intensities (Robe & Griffiths 1990).
The plasticity of photosynthesis in L. uniflora is illustrated by the rapid loss of CAM from newly emersed, nonsenescent leaves, its absence from new terrestrial leaves which show several-fold higher activity of Rubisco, and the contrasting activities of PEPC and Rubisco in the two halves of partially emersed leaves.In contrast, permanently submersed L. uniflora show gradual seasonal changes in CAM activity and photosynthetic capacity (Robe & Griffiths 1992).For the plants growing at Thirlmere, the switch from CAM to C 3 underlay rapid new leaf production culminating in flowering only 3-4 weeks after emersion (Robe & Griffiths 1998), so could be important for seed output in a habitat where re-submersion is unpredictable.Photosynthetic pathway plasticity has been discovered in several other amphibious species, thought to be recently evolved from terrestrial ancestors, which are CAM or C 3 + C 4 when submersed and C 3 when emersed (Keeley 1998a(Keeley , 1999)), or C 3 when submersed and C 4 when emersed (Ueno 1996), or which show several-fold higher Rubisco activity in emergent tissues (Maberly & Spence 1989).
The loss of CCMs and switches to C 3 in amphibious species on emergence into the aerial environment have been related to the disappearance of diffusional limitation of CO 2 (e.g.Aulio 1986;Raven et al. 1988;Nielsen et al. 1991;Keeley 1996;Raven & Spicer 1996;Keeley 1999).However, there is no information on how emersion was sensed by L uniflora or on how photosynthetic pathway changes were triggered.One possibility is that localized decreases in intercellular CO 2 , for example around cells near the epidermis, were sensed in some way (see e.g.Assman 1999;Matsuda, Bozzo & Colman 1997;Kaplan & Reinhold 1999).Alternatively, water loss might have been detected in the small whole-leaf reductions in turgor (Fig. 1d) or as reductions in cell volume, or increased osmotic strength (e.g.Bray 1993;Shinozaki & Yamaguchi-Shinozaki 1997), particularly in epidermal cells.Loss of water as a signal of emersion could perhaps most easily explain the absence of CAM in the upper halves of partially emersed leaves (Fig. 3), the rapid loss of CAM from emersed aquatic leaves of L. uniflora (see Results), and also Isoetes howellii (see Keeley 1996), and the presence of CAM in terrestrial L. uniflora under conditions of high humidity (Farmer & Spence 1985;Aulio 1986).Signal transduction may have involved plant growth substances such as ABA, and photosynthetic enzyme expression could be modulated by other environmental and internal factors, such as photoperiod, temperature, light quality, ethylene, O 2 , and nutrient status (see Cushman & Bohnert 1997;Trewavas & Malhó 1997).In terrestrial CAM species, water stress induces CAM by initiating transcription of PEPC (Cushman & Bohnert 1997) and fluctuations in the content of Rubisco small subunit mRNA (Michalowski et al. 1989), raising intriguing questions about how signal perception and transduction, and control of gene expression could differ in L. uniflora.
The Rubiscos of the aquatic macrophytes examined so far have a relatively high K m for CO 2 (low affinity for CO 2 ; Yeoh, Badger & Watson 1981), and are optimized to operate at high CO 2 concentrations (see Raven, Osborne & Johnston 1985).Preliminary measurements indicate that the Rubisco of submersed L. uniflora may have a K m (CO 2 ) up to fivefold higher than that of Plantago major (unpublished results).However, a high-K m (CO 2 ) Rubisco would be a disadvantage in the aerial environment because of the higher levels of O 2 fixation and photorespiration.Perhaps in amphibious plants such as L. uniflora, regulation of the activation state of Rubisco (see Fig. 2c, Fig. 3), could compensate for variations in CO 2 supply and also allow acclimatization to changes in light intensity or content of Rubisco protein (see Mott, Snyder & Woodrow 1997;Hammond et al. 1998).

Carbohydrate storage
In aquatic L. uniflora, starch seemed to be the main overnight store of C and source of energy and C for biosynthesis.After emersion, considerably more excess photosynthate was available, and this generally accumulated as sugars.Sugars stored in the vacuole may have been associated with osmotic adjustment (see Stewart & Larher 1980;Morgan 1984) as water stress began to develop after flowering.After flowering, accumulation of fructans and starch in stems and roots increased.Fructans, polymers of fructose, stored in the vacuole, seem to be involved in water uptake and retention, and cell expansion (Hendry 1993;Albrecht, Biemelt & Baumgartner 1997;Vijn & Smeekens 1999), and may have physiological importance for an amphibious species such as L. uniflora during water deficit and re-submersion.Storage carbohydrates could also provide interim energy and CO 2 supplies (see ap Rees et al. 1987) during the rapid re-growth of aquatic leaves which can follow resubmersion (unpublished results).

Carbon isotope ratios
The d 13 C values for L. uniflora were similar to previous reports for this species (-25‰ for submersed plants; Keeley & Sandquist 1992) and other isoetids (-21 to -33‰; Osmond et al. 1981), and were within the wide range recorded for aquatic and amphibious macrophytes (-13•4 to -50•7‰; Keeley & Sandquist 1992).The d 13 C of sediment C was also similar to that previously recorded for terrestrial soils (Schleser & Jayasekera 1985).

W.E. Robe & H. Griffiths
Plant d 13 C reflects the d 13 C of the source CO 2 , the fractionation of isotopes which occurs during Rubisco and PEPC carboxylation, and processes of diffusion, dissolution and hydration, and is mainly determined by the rate-limiting step (O'Leary 1988;Farquhar et al. 1989).In terrestrial species using atmospheric CO 2 (d 13 C -7 to -9‰), a switch from CAM (in which fixation by Rubisco occurs in a relatively closed system) to C 3 (fixation of CO 2 taken up via stomata) can result in a large shift to more negative d 13 C values (e.g.-14 to -30‰) as discrimination by Rubisco against the heavier isotope of C is expressed.However, in L. uniflora, emersion and loss of CAM resulted in a very small (3-4‰) decline in d 13 C.A similar small difference between the d 13 C of submersed and emersed Isoetes howellii (-26 and -29‰, respectively) was found by Keeley & Sandquist (1992).In I. howelli, a seasonal pool species which seems to take up little, if any, sediment CO 2 (Keeley 1998b), a change in the d 13 C of source CO 2 , from -15 to -20‰ for submersed plants (lacunal CO 2 from respiration and malate decarboxylation), to -7‰ for emergent plants (atmospheric CO 2 ), combined with substantially increased discrimination by Rubisco, could explain the slightly more negative leaf d 13 C.However, L. uniflora at Thirlmere seems to have continued to take up sediment CO 2 with a d 13 C of around -23•5‰ (CO 2 in the interstitial water of sediments; unpublished results).The more negative d 13 C of emersed plants may therefore indicate changes in the d 13 C of lacunal CO 2 due to cessation of PEPC fixation and malate decarboxylation, increased rates of daytime C fixation and a small increase in discrimination by Rubisco in a slightly more open system.

Nitrogen storage
Aquatic L. uniflora contained high concentrations of nitrate and free amino acids, similar to those found in spring in permanently submersed plants in natural habitats (see Robe & Griffiths 1994).The disappearance of a large part of these soluble N reserves in the first 3-4 weeks of emersion suggests that they may have been used for new growth.The low N content of senesced and old leaves compared with young leaves at all three sampling dates (72% lower; Fig. 6) also suggests recycling of functional and structural protein (see Stoddart & Thomas 1982;Millard 1988), although volatilization and/or leaching may account for some of the loss (Feller & Fischer 1994).The Rubisco protein which was not activated in vivo (34-45%; Fig. 2c) may also constitute a store of N (e.g.see Stitt & Schulze 1994).Although plant N content declined after emersion, %N was above the 1•3% critical value (minimum N for maximum growth) determined for aquatic macrophytes by Gerloff & Krombholz (1966), and well within the range for terrestrial macrophytes growing with a wide range of N supply (0•68-4•0% ;Gebauer, Rehder & Wollenweber 1988) in all non-senesced tissues except roots of seed-bearing L. uniflora.
The d 15 N signals found in L. uniflora will be a function of both the d 15 N of the source N, and the fractionation which has occurred during acquisition and metabolism (Robinson, Handley & Scrimgeour 1998).The d 15 N of soil total N is probably not a good approximation of plant available pools (aquatic L. uniflora is known to utilize both NO 3 -and NH 4 + ; Schuurkes, Kok & Den Hartog 1986; Robe & Griffiths 1994), which are thought to turn over quickly and have a rapidly changing d 15 N (Hogberg 1997).However, methods to accurately extract and analyse the d 15 N of these pools are not available (Handley & Scrimgeour 1997).The varying d 15 N of L. uniflora could partly reflect changes of source and fractionation during uptake by mycorrhizal fungi (Hogberg 1987); mycorrhiza have been reported in roots of L. uniflora (Söndergaard & Laegaard 1977).Intra-plant d 15 N differences could be the result of changes in the proportion of root and shoot reduction (Evans et al. 1996), which in L. uniflora do occur in response to NO 3 -supply and light intensity (Robe & Griffiths 1994).Fractionation during N incorporation into amino acids and proteins, mixing of different pools (as a result of translocation), and losses during senescence (e.g. of NH 3 ) may also have affected intra-plant d 15 N (see Hogberg 1997;Robinson, Handley & Scrimgeour 1998).

CONCLUSION
In L. uniflora, marked physiological and photosynthetic changes accompanied emersion onto dry land at a reservoir, and these findings provide a clearer picture of the extent of phenotypic plasticity in this amphibious species.The switch from CAM to C 3 photosynthesis seemed an important component of the rapid new leaf growth which preceded flowering only 3-4 weeks after emersion (Robe & Griffiths 1998).At the same time, the accumulation of Phenotypic plasticity in amphibious Littorella uniflora 1051 carbohydrate reserves, low water and osmotic potentials and possibly continued uptake of CO 2 from sediments suggested a plant poised for re-submersion.These characteristics seem to be adaptive in a habitat where water level fluctuations are large and unpredictable.Since plasticity has a genetic basis and can evolve rapidly (Bradshaw 1965;Pigliucci, Cammell & Schmitt 1999), the population of L. uniflora at Thirlmere may be more responsive to emersion than those in surrounding tarns which are almost permanently submersed.

© 1054 Figure 1 .
Figure1.Sediment water content and plant water status for Littorella uniflora at Thirlmere Reservoir, Cumbria, at three sampling dates in 1995.(a) Water content of sediments around roots of L. uniflora expressed as a percentage calculated from fresh and dry weight measurements.(b) Fresh to dry weight ratios of shoots (), roots (᭝) and stems (᭹) of L. uniflora.Mean for 10 plants with standard deviation (SD).There was a significant difference between fresh weight to dry weight ratios in June and July, and in July and September; P < 0•001).(c) Water content of shoots (), roots (᭝) and stems (᭹) expressed as a percentage calculated from fresh weight to dry weight ratios.Mean for 10 plants.(d) Shoot osmotic potential (), water potential (ᮀ), and turgor pressure (᭹) calculated as osmotic potential minus water potential, and osmotic potential of roots (᭡).Mean for four plants at each sampling date, shown ± SD for osmotic potentials.The osmotic potential of shoots, but not roots, was significantly higher in flowering compared with aquatic L. uniflora (P < 0•001).In both shoots and roots, osmotic potentials in seed-bearing plants were significantly lower than in the flowering stage (P < 0•001).In some cases, the error bar is masked by the symbol.

Figure 2 .
Figure2.Lacunal CO 2 concentration, CAM, activities of PEPC and Rubisco, and chlorophyll and soluble protein content of shoots of Littorella uniflora at Thirlmere Reservoir, Cumbria, at three sampling dates in 1995.(a) Lacunal CO 2 concentration, [CO 2 ] i, determined during a 4 h period around midday.Mean values for 20-25 leaves from 6-7 plants with standard deviations (SD).The [CO 2 ] i of flowering L. uniflora was significantly higher than that of the aquatic stage (P = 0•003), while that of seed-bearing L. uniflora was significantly higher than that of the flowering stage (P < 0•001).(b) Dawn (᭹) and dusk (᭺) titratable acidity of five shoots ± SD.Titratable acidity at dawn in seed-bearing L. uniflora was significantly higher than in the flowering stage (P < 0•001) (c) Activities of PEPC (᭝) and Rubisco enzymes.The data for Rubisco show both initial activity, immediately after extraction (᭹), and total activity, after full activation in the presence of Mg 2+ and CO 2 ().Results, shown ± SD, are for 4-6 extracts using the distal half (minus the tip) of 6-10 youngest mature leaves.Initial and total activities were significantly different at each sampling date (P < 0•001).(d) Chlorophyll (a + b) content, and (e) soluble protein concentration, of the distal halves (minus the tip) of youngest mature leaves.Mean ± SD of five extracts.In some cases, the error bar is masked by the symbol.

Figure 6
Figure 6 shows d 13 C and total C, and d 15 N and total N in sediments and L. uniflora.The d 13 C of plant tissues ranged from -23•9 to -29•1‰ and declined (became more nega-

Figure 4 .
Figure 4. Soluble carbohydrate content of shoots, roots and stems of Littorella uniflora at dawn and dusk at Thirlmere Reservoir, Cumbria, at three sampling dates in 1995.(a) Ethanol (95%)-soluble neutral sugars, i.e. sucrose, fructose, glucose and fructans with a low degree of polymerization.(b) Water (i.e.buffer)-soluble fructans.(c) Starch.Means for whole shoots (dawn and dusk ᮀ), roots (dawn ᭡ and dusk ᭝) and stems (dawn ᭹ and dusk ᭺) of three plants are shown with standard deviation.Data for shoots of seed-bearing L. uniflora do not include the seeds.Only starch in shoots was significantly higher at dusk than at dawn in flowering compared with aquatic L. uniflora (P = 0•04).Neutral sugars in shoots and all carbohydrates in roots and stems were significantly lower (P = 0•05 -0•001) at dusk than at dawn in seed-bearing compared with flowering L. uniflora.In some cases, the error bar is masked by the symbol.

Figure 5 .
Figure 5. Soluble nitrogen content of shoots, roots and stems of Littorella uniflora at Thirlmere Reservoir, Cumbria, at three sampling dates in 1995.(a) Nitrate concentration in shoots (), roots (᭡) and stems (᭺).(b) Free amino acid concentration in shoots (), roots (᭡) and stems (᭺).Results, shown with standard deviation, are mean values for whole shoots, stems and roots of five plants sampled at midday.The differences between amino acids concentrations in aquatic and flowering L. uniflora and in flowering and seed-bearing plants were significant (P < 0•001) for shoots, roots and stems.Data for shoots of seedbearing L. uniflora do not include the seeds.In some cases, the error bar is masked by the symbol.

Figure 6 .
Figure 6.Carbon and nitrogen isotope ratios (d 13 C and d 15 N expressed as ‰, see Materials and Methods) together with C and N contents expressed as a percentage of dry weight, for plants of Littorella uniflora and sediments at Thirlmere Reservoir, Cumbria.Data are for 6-8 plants.The material collected in early June was combined.For shoots, roots and stems of flowering and seed-bearing plants, the results are mean values.For leaves of different ages, the data are for leaves from several plants combined, similarly for ramets, flowers and seeds.Data for sediments are for a sample from 20-30 g taken from round the roots.The d 13 C and %C of shoots of seed-bearing L. uniflora were significantly higher than those of the flowering stage (P < 0•001), and shoot and root d 13 C were significantly different in both flowering and seed-bearing plants (P < 0•001).Seed-bearing L. uniflora showed significantly lower %N in shoots, roots and stems compared with the flowering stage (P < 0•001).The d 15 N of shoots and roots of seed-bearing L. uniflora did not differ significantly from that of the flowering stage (P > 0•05); however, d 15 N of stems was significantly lower (P = 0•002) in the seed-bearing stage.