Does the C₄ plant Trianthema portulacastrum (Aizoaceae) exhibit weakly expressed crassulacean acid metabolism (CAM)?

Keywords: CAM evolution, CO₂ assimilation, C₄ photosynthesis, facultative CAM, Kranz anatomy, Sesuvioideae, stem photosynthesis, Trianthema portulacastrum.

C₄ photosynthesis and crassulacean acid metabolism (CAM) are modifications of the ancestral C₃ photosynthetic pathway that have evolved independently in a wide range of families and genera (Smith and Winter 1996; Sage 2016). There are approximately 8100 known C₄ species, which are dispersed across 418 genera in 19 angiosperm families (Sage and Sultmanis 2016), whereas the roughly 16 000 CAM taxa are in 400+ genera from 37 families of angiosperms, gymnosperms, ferns and lycopsids (Smith and Winter 1996; Yang et al. 2015; Winter et al. 2020a).

In both C₄ and CAM plants, the structure and function of photosynthetic organs has been modified from the C₃ template creating CO₂-concentrating mechanisms that suppress photorespiration. In C₄ plants which, like C₃ plants, exhibit net CO₂ uptake during the light only, suppression of photorespiration has led to enhanced photosynthetic CO₂ uptake in warm climates at reduced water cost. Water-use efficiency is further enhanced in CAM plants as water loss by transpiration is reduced when CO₂ is taken up during the cool of the night and stomata close during most of the day (Winter et al. 2005). While the CAM cycle is associated with nocturnal increases and diurnal decreases in the organic acid content of chloroplast-containing tissues, 24-h fluctuations in H⁺ are typically absent from C₃ and C₄ tissues (Winter 2019). Despite the relatively frequent appearance of C₄ and CAM across the phylogeny of plants, only four families of vascular terrestrial plants contain C₄ species as well as CAM species: Aizoaceae, Asteraceae, Euphorbiaceae, and Portulacaceae (Smith and Winter 1996; Edwards and Ogburn 2012; Sage 2016).

Portulaca L. (Portulacaceae, Caryophyllales) is the only genus of land plants in which C₄ and CAM are known to co-occur in the same plant and, indeed, in the same leaf (Koch and Kennedy 1980). The genus is currently considered to contain 150 species (POWO 2020), overwhelmingly annual succulent-leaved herbs, allocated across six clades (Ocampo et al. 2013). Five clades contain C₄ taxa and one contains species with C₃-C₄ intermediate characteristics. CAM has been reported in nine species from five of the six clades (Koch and Kennedy 1980; Kraybill and Martin 1996; Guralnick and Jackson 2001; Guralnick et al. 2002; Holtum et al. 2017a; Winter and Holtum 2017; Winter 2019) including the clade with C₃-C₄ intermediates (Winter et al. 2019b). In all cases reported to-date, the expression of CAM in leaves of Portulaca with C₄ photosynthesis has been facultative; CAM is expressed in leaves that have been water-stressed but is absent from leaves of well-watered plants. In the case of P. oleracea, in which CAM expression in leaves is facultative but its expression in stems is constitutive with a facultative component, the transcript abundance of genes putatively associated with C₄ and CAM has been studied in well-watered and in drought-stressed plants (Ferrari et al. 2020).

Within the Aizoaceae, a family within the order Caryophyllales of close to 1800 species and 119 genera (Hernández-Ledesma et al. 2015; Klak et al. 2017; POWO 2020), C₃ and CAM photosynthesis are present in three subfamilies, Mesembryanthemoideae, Ruschioideae and Sesuvioideae (Ting 1989; Smith and Winter 1996; Winter et al. 2019a). C₄ photosynthesis is known in 31 species of the sub-family Sesuvioideae (Bohley et al. 2017) where it occurs in the genera Trianthema (21 of ~30 species), Sesuvium (5 of 13 species) and Zaleya (all species). Within the Sesuvioideae, CAM has been reported in two species, the pantropical, predominately C₃-exhibiting Sesuvium portulacastrum (Ting 1989; Winter et al. 2019a) and in S. maritimum (Martin et al. 1982).

Here we report the presence of low-level CAM-type nocturnal acidification in a C₄ member of the Sesuvioideae, Trianthema portulacastrum. Curved patterns of nocturnal net CO₂ loss indicated temporarily enhanced refixation of CO₂ at night, consistent with low-level CAM in this species.

A salt-tolerant annual that has Kranz anatomy in leaves but not in stems, Trianthema portulacastrum L. is a fast-growing, succulent-leaved herb that germinates during the wet season in the seasonally dry tropics of Panama, forming prostrate mats or clumps up to 50+ cm in diameter, before it senesces and dies during the dry season. Of uncertain geographical origin, it is widespread in the tropics and subtropics of western and south-east Asia, Africa, America and Australia (ALA 2020) where it inhabits beach dunes, moist or seasonally-dry open wetlands including alkaline flats, clay pans and playa lakes. It is often a weed of disturbed areas including gardens, irrigated soils and road-sides.

Seedlings of T. portulacastrum were obtained from the Sarigua National Park, Azuero Peninsula, Republic of Panama (8.013348°N, –80.485658°W), where the species coexists with Sesuvium portulacastrum (L.) L. (Winter et al. 2019a).

In October 2018, the second half of the rainy season, 10 seedlings (diameter of ~5 cm each) were planted in a mixture of 50% (v/v) forest soil: 50% (v/v) sea-sand (Novey, Panama) in a 1.2 × 1.2 × 0.38 m raised garden box (Hummert International) at the Smithsonian Tropical Research Institute, Santa Cruz Experimental Research Facility, Gamboa, Republic of Panama (9.120027°N, –79.702017°W) (Fig. 1). Plants were exposed to full natural solar radiation and received natural rainfall. From mid-December 2018 onwards, samples of leaves and stems were taken at dusk (17:30 – 18:00 hours) and dawn (07:00 hours) for fresh and dry mass determinations and for acid analysis at weekly intervals until the plants senesced and died during the dry season at the end of February 2019. Self-sown seeds germinated following the beginning of the wet-season rains in 2019. Starting at the end of May 2019, once again leaves and stems were sampled at dusk and dawn for fresh mass, dry mass and acid analysis at weekly to biweekly intervals. Samples were weighed for fresh mass determination and stored in liquid nitrogen to be processed for determination of dry mass and measurements of titratable acidity as described below.

Fig. 1.  (a) A potted Trianthema portulacastrum. Note the prostrate habit, slightly fleshy leaves and succulent stems. The outer diameter of the pot is 20 cm. (b) Trianthema portulacastrum growing outside in a 1.2 × 1.2 × 0.38 m raised garden box at the Smithsonian Tropical Research Institute, Santa Cruz Experimental Research Facility, Gamboa, Republic of Panama. Photo taken 15 December 2018.

Throughout 2018 and 2019, experiments were conducted with potted plants to study titratable acidity changes and gas-exchange responses during wet–dry-wet cycles. Plants were grown in terracotta pots (0.5 or 3.0 L) containing 2/3 (v/v) potting mix (Miracle-Gro Lawn Products) and 1/3 quartz sand (Novey). Plants were maintained at the Tupper Centre of the Smithsonian Tropical Research Institute, Panama City (8.962938°N, –79.702017°W) under either full solar radiation or underneath rain-shelters at 70% natural light. Growth conditions are specified in the corresponding figure legends. As in the outdoor experiments, for acidity measurements mature leaves and stems excised from plants at dusk and dawn were weighed for fresh mass determination and frozen in liquid nitrogen. Prior to freezing, leaf area was determined using a LI-3100 leaf area meter (LI-COR Biosciences).

Samples were transferred from liquid nitrogen to a freeze-drier (Freezone 4.5, Labconco). After 72 h samples were reweighed for dry mass determination. Freeze-dried tissue was extracted sequentially in boiling 50% ethanol and in water (Winter and Holtum 2017). Extracts were titrated with 5 mM KOH to pH 6.5

Twenty-four hour CO₂ exchange patterns by intact shoots of whole plants or intact attached branches were studied during wet–dry-wet cycles. In two of the three experiments shown in the results section, the shoot of a small seedling was enclosed inside a Perspex cuvette (internal dimensions 11 × 11 × 16 cm) that rested on the 0.5 L terracotta pot in which the plant grew. The roots and the pot remained outside the cuvette. After initial daily irrigation with water, 9–11 day drought treatments were imposed by withholding irrigation until net CO₂ uptake during the light period had decreased markedly, after which plants were rewatered daily. In another experiment, an intact attached branch of a plant was enclosed into a Walz GWK-3M gas exchange cuvette (Walz GmbH).

The gas-exchange cuvettes, pots and plants were located inside a controlled environment chamber (GC15, EGC) operating under 12 h light (28°C)/12 h dark (22°C) cycles. Illumination was supplied by a LED grow light (SS-GU300-w, Sunshine Systems). Photon flux density is specified in the corresponding figure legends. The cuvette was supplied with ambient air sourced 16 m above ground level and passed through a 2 m³ buffer at a flow rate of 2.3 L min^–1. Net CO₂ exchange was measured in a flow-through gas-exchange system consisting of Walz components and a LI-6252 CO₂ analyser (LI-COR Biosciences) (Holtum and Winter 2003).

Cross-sections of fully expanded healthy leaves (middle between base and tip, and between edge and midrib) and mature stems were prepared either by hand using double edge breakable razor blades or, in some cases, tissue was fixed in 4% (v/v) glutaraldehyde for 48 h at 7°C before sectioning in a cryostat microtome (CM1860, Leica). Chlorophyll auto-fluorescence (682 nm) was examined using a confocal microscope (FV3000, Olympus) with an integrated camera system.

Variation in dusk/dawn titratable acidity levels was determined in outdoor-grown plants over ~15 months. The study period included one full wet season and two dry seasons (Fig. 2a–c). Consistent with the annual life-form of T. portulacastrum, plants died during the latter part of each dry season. Reduced cloud cover during the dry season led to higher daily photon flux densities (Fig. 2a). As is typical for tropical rainforest climates, seasonal variations in temperature at the study site were small (Fig. 2b).

Fig. 2.  Seasonal changes in photon flux density (a), maximum (upper blue line), minimum (lower red line) and mean (middle green line) temperature (b), rainfall (c), variation of titratable acidity in leaves and stems at dusk (red dots) and dawn (blue squares) (d–g), and variation in fresh mass: dry mass ratio at dusk (red dots) and dawn (blue squares) in leaves (h) and stems (i) of Trianthema portulacastrum. Plants were grown in a raised garden box in a 50% soil : 50% sea-sand mixture. Measurements were taken between December 2018 and March 2020. Plants died during the 2019 dry season (February) and re-grew from previously produced seeds 2 months later at the onset of the rainy season. Acidity levels are expressed per unit fresh mass (leaves, d; stems, e) and dry mass (leaves, f; stems, g). Acidity and FM/DM values are means ± s.d. (n = 5; for leaves, each sample comprised 6–9 leaves). Samples were derived from different plants. Error bars are absent if they are smaller than symbols. In (d–i) * indicate significant increases of titratable acidity and FM/DM between dusk and dawn values (one-tailed t-test, P ≤ 0.05).

Throughout the course of the experiment, and irrespective of season, leaf H⁺ values, when expressed on a fresh mass basis, were significantly greater at dawn than at dusk for 15 of 30 measurements (Fig. 2d). The frequency of significantly greater dawn levels increased to 25 when tissue acidity was expressed on a dry mass basis (Fig. 2f), thereby correcting for diurnal decreases in leaf water content and thus reduced fresh mass: dry mass ratios (Fig. 2h). Titratable acidity of stems was consistently greater at dawn than at dusk on 27 of 29 days when acidity was expressed per unit of fresh mass, and on all days when expressed per unit of dry mass (Fig. 2e, g). On the days with significant overnight acidification, the fold-increase (dawn: dusk ratio of H⁺ per unit of dry mass) was greater in stems (2.03 ± 0.48, mean ± s.d., n = 29) than in leaves (1.55 ± 0.19, n = 25) (P = <0.001). There was no obvious seasonal change in the degree of nocturnal acidification of both leaves and stems (Fig. 2f, g). The H⁺ values were similar during the wet and dry seasons.

Linear regressions showed that nocturnal acidification of leaves and stems correlated positively with integrated PFD of the previous day (Fig. 3). The relationship was significant for leaves (P = <0.01) and close to significant for stems (P = <0.06).

Fig. 3.  Relationship between nocturnal increase in titratable acidity and integrated photon flux density of the previous day in leaves (a) and stems (b) of Trianthema portulacastrum. ΔH⁺ values were calculated from the data in Fig. 2f, g and refer to days with significantly greater acidity at dawn than at dusk. (a) y = 1.077x + 2.1188; (b) y = 1.1681x + 31.378.

In a pot experiment under full solar radiation at STRI’s Tupper Centre during the 2018 dry season, plants were exposed to a wet–dry-wet cycle. Small but significant nocturnal acidification per unit dry mass was detected in leaves of drought-stressed plants, whereas nocturnal acidification was consistently observed in stems of both watered and drought-stressed plants irrespective of whether the data was expressed on a dry mass or fresh mass basis (data not shown).

Sixty-eight days of CO₂ gas-exchange were monitored for three separate plants. Figures 4 and 5 depict the results for two plants, the shoots of which developed from seedlings inside the gas-exchange cuvette. One plant was exposed to a single watering-drought-rewatering cycle and a second to two watering-drought-rewatering cycles.

Fig. 4.  Fifteen days of net CO₂ exchange by the shoot of a potted Trianthema portulacastrum. Watering was withheld on day 3 (red arrow) and recommenced on day 13 (blue arrow). Shaded areas represent the 12 h dark periods. Photon flux density was 700 μmol m^–2 s^–1 at the level of the shoot. On the last day of the experiment, total leaf area was 68 cm² and leaf and stem dry masses were 0.41 and 0.22 g respectively.

Fig. 5.  Thirty-eight days of net CO₂ exchange by the shoot of a potted Trianthema portulacastrum. During this period, the plant was subjected to two drying/re-watering cycles. Watering was withheld on days 7 and 20 (red arrows) and recommenced on days 17 and 30 (blue arrows). Shaded areas represent the 12 h dark periods. Photon flux density was 700 μmol m^–2 s^–1. On the last day of the experiment, total leaf area was 21.6 cm².

Under both well-watered and drought conditions, net CO₂ uptake was restricted to the light during all treatments. In the experiment depicted in Figure 4, CO₂ efflux was initially relatively constant throughout the night following an initial slight overshoot and period of equilibration to night temperature (22°C) which was 6°C below day temperature (28°C). Following cessation of watering on day 3, a small curvature in the nocturnal CO₂ release pattern was detectable with lower rates in the middle than at the beginning and end of the dark period. CO₂ uptake during the day and CO₂ loss at night continued to increase as the shoot continued to grow utilising water in the pot. From day 9 onwards water-deficit stress developed as indicated by the progressive daily reduction in both CO₂ uptake during the day and CO₂ loss at night. Daytime CO₂ exchange developed a prominent morning peak of uptake. At all times, nocturnal CO₂ exchange remained below the CO₂ compensation point. Following rewatering on day 13, a recovery of CO₂ uptake in the light to pre-drought rates was observed within 3 days. Concomitant with the recovery in rates of daytime CO₂ uptake, nocturnal respiratory CO₂ loss increased to rates greater than observed before drought. The respiratory trace also developed a pronounced curvature that persisted.

The gas-exchange trace for the plant that was exposed to two watering-drought-rewatering cycles (Fig. 5) was similar to that for the plant exposed to a single cycle (Fig. 4) except that the absolute rates of CO₂ uptake were lower as it was a smaller plant. In Figure 5, the curvature in the nocturnal respiratory trace that was noted in particular following rewatering of the plant monitored in Figure 4, was again observed throughout most of the 38 day experiment, especially during the entire second watering-drought-rewatering cycle.

The observations for a branch of a third plant subjected to a single watering-drought-rewatering cycle are provided in Figure 6. Net CO₂ uptake in the light started to decline almost immediately when watering was stopped. Again, nocturnal net CO₂ exchange never crossed the compensation line to become positive, but during all 15 dark periods of the experiment a distinct temporal reduction of nocturnal net CO₂ loss occurred.

Fig. 6.  Fifteen days of net CO₂ exchange by a branch of a potted Trianthema portulacastrum. Watering was withheld on day 2 (red arrow) and recommenced on day 12 (blue arrow). Shaded areas represent the 12 h dark periods. Photon flux density was 700 μmol m^–2 s^–1.

The leaves contain a central Kranz-type core of chlorophyll-rich bundle-sheath (BS) and mesophyll cells that encircle the leaf vascular bundles (atriplicoid pattern) (Fig. 7a; Muhaidat et al. 2007). In the BS cells the chloroplasts are arranged centripetally. The Kranz cells are surrounded by chloroplast-sparse parenchyma cells one cell deep on the upper (adaxial) surface and two cells deep on the lower (abaxial) surface. The vascular bundles, chloroplast-dense Kranz cells and associated air spaces together constitute ~38% of the leaf volume, the chloroplast-sparse outer parenchyma cells and associated air spaces together constitute 30% of the leaf volume, and the epidermis/cuticle constitutes the remainder.

Fig. 7.  Hand cross-sections of a leaf (a) and a stem (b), and a microtome cross-section through the stem cortex (c) of Trianthema portulacastrum. In (a) note in the centre of the leaf the atriplicoid Kranz anatomy with abundant red-fluorescing chloroplasts. In the outer abaxial and adaxial layers of large parenchyma cells (P), red-fluorescing chloroplasts are present but sparse. In (b) note the red-fluorescing chloroplasts in the outer pith. In (c) note the even distribution of red-fluorescing chloroplasts across the stem cortex.

In contrast to the leaves, Kranz anatomy is absent from the stems of T. portulacastrum. The stems contain a band of vascular tissue encircling a central parenchymatous pith (Fig. 7b). The pith cells near the vascular band contain chloroplasts. The vascular tissue is surrounded on the outside by a cortex, ~9–11 cells deep, of parenchyma cells that contain chloroplasts that are relatively evenly distributed across the tissue (Fig. 7c).

Stomatal densities were 98 ± 17 mm^–2 and 89 ± 4 mm^–2 (mean ± s.d., n = 5) for the upper and lower leaf surfaces respectively. Stem stomatal densities were 18 ± 4 mm^–2 for the sun exposed upper surfaces and 16 ± 7 mm^–2 for the lower soil facing surfaces.

The C₄ pathway is the principal mode of carbon acquisition in T. portulacastrum. However, T. portulacastrum is not 100% C₄. Measurements of titratable acidity and CO₂ gas exchange strongly suggest that this species also exhibits low-level CAM photosynthesis. Increases in H⁺ content from dusk to dawn, a defining feature of CAM photosynthesis, were consistently observed in stems, and very frequently in leaves. Nocturnal acidification was detected in stems of field-grown plants on a regular basis, irrespective of whether acidity was expressed on a fresh mass or dry mass basis, whereas in leaves the lower degree of acid accumulation was most obvious when H⁺ was expressed on a dry mass basis because the day-night differences in leaf water content result in underestimates of nocturnal acidification when expressed on a fresh mass basis. Overall, nocturnal increases in H⁺ were much smaller than those of archetypal CAM species (e.g. many cacti and agaves) but in the range of other species that operate at the low end of the C₃ – full CAM or C₄ – full CAM spectrum (Table 1). Titratable acidity measurements in leaf and stem extracts are highly sensitive, can resolve differences of as low as 1 µmol H⁺ g^–1 fresh mass between dusk and dawn samples, and are well suited to reliably identify small nocturnal H⁺ increases such as those in T. portulacastrum. Nocturnal increases of H⁺ are most likely the result of CAM-type CO₂ dark fixation, where the uptake of 1 CO₂ typically leads to the accumulation of 1 malate anion + 2 H⁺. We are not aware of studies that have demonstrated significant nocturnal H⁺ accumulation in photosynthetic tissues that engage 100% in the C₃ or C₄ pathway.

Table 1.  Nocturnal acidification in Trianthema portulacastrum as compared to species with weak and strong CAM
Values for Trianthema portulacastrum are the means from all days with significant nocturnal acidification in Fig. 2. For Sesuvium portulacastrum, corresponding means were calculated from data in figs 2 and 3 of Winter et al. (2019a). Among the species with weakly expressed acidification, Bulbophyllum putidum had a CAM-type δ¹³C value of –15.2‰, and Zamioculcas zamiifolia and Oncidium panamense showed net CO₂ dark fixation. Data for Zamioculcas zamiifolia and Coleus scutellarioides were obtained from plants exposed to water deficit stress, a treatment that resulted in enhanced CAM expression. ND, not determined

Nocturnal net CO₂ exchange of CAM-exhibiting species reflects the balance between uptake of atmospheric CO₂ via PEP carboxylase and production of CO₂ by mitochondrial respiration. In archetypal CAM species, dark CO₂ fixation greatly outweighs respiration, leading to a highly positive carbon balance at night. Another feature of CAM-gas exchange is that rates of dark CO₂ fixation rarely reach a steady-state: they typically increase in the first half of the dark period and decline in the second half (Osmond 1978). In species with weakly expressed CAM, dark CO₂ fixation is enhanced but often does not exceed rates of CO₂ loss. Hence the nocturnal carbon balance may remain negative. Nonetheless, in analogy to the nocturnal increases and decreases of rates of net CO₂ dark fixation observed in species with strong CAM, enhanced dark CO₂ fixation in species with very weakly expressed CAM is still reflected by a characteristic temporary decrease of net CO₂ loss in the course of the night.

Here we present the results of three independent CO₂ exchange experiments with three different plants that spanned a total of 68 day-night cycles and generated 23380 measuring points at 4 min intervals. We tested whether T. portulacastrum has the capacity for net dark CO₂ fixation in the well-watered state, a result which would have been unequivocal evidence of constitutive CAM. We also examined whether T. portulacastrum has the capacity to reversibly induce or upregulate net dark CO₂ dark fixation in response to water-deficit stress, a result which would have been unequivocal evidence of facultative CAM. In both well-watered and drought-stressed plants, at no time did shoot or branch CO₂ exchange at night exceed the compensation point and become positive. However, the patterns of nocturnal CO₂ exchange were clearly different from plants that engage 100% in C₃ or C₄ (where the nocturnal CO₂ exchange trace is essentially a straight negative line under constant night-time temperature conditions as employed in the current study of gas-exchange). In T. portulacastrum curved patterns of net CO₂ loss were consistently observed, indicating a temporal reduction of net CO₂ loss likely related to enhanced CO₂ uptake. This curved shape of nocturnal CO₂ exchange has been observed for other species with low levels of CAM expression (e.g. Platycerium veitchii, Holtum and Winter 1999; Portulaca digyna, Holtum et al. 2017a; Sesuvium portulacastrum, Winter et al. 2019a). Taken together, the two observations, small nocturnal increases in titratable acidity and the small temporal reductions of nocturnal net CO₂ loss strongly suggest that a weakly expressed CAM cycle is present in T. portulacastrum. Utilising δ¹³C of plant carbon as an indicator of CAM use in T. portulacastrum is not possible because C₄ and CAM photosynthesis lead to similar carbon isotopic signatures.

Overall, considering the large dataset of seasonal changes in H⁺, nocturnal acidification in leaves and stems of the annual T. portulacastrum appears predominately constitutive with no or little evidence of stress-associated upregulation. The constitutive nature of low-level CAM in T. portulacastrum is supported by the positive relationship between overnight H⁺ increase and PFD during the previous day (significant in leaves and almost significant in stems), a feature well documented for cacti (Nobel 1982). The nature of CAM in the annual T. portulacastrum therefore not only differs in pattern and extent from known annual C₄ Portulaca species but also differs from other well-studied CAM annuals such as Mesembryanthemum crystallinum (Aizoaceae; Winter and von Willert 1972) and Calandrinia polyandra (Montiaceae; Winter and Holtum 2011). In M. crystallinum and C. polyandra the environmentally-triggered switch from C₃ to CAM involves net nocturnal CO₂ uptake during the CAM phase, assisting the extension of life-cycle into the dry season and enabling succulent leaves and branches to persist, reproductive material to develop and seeds to fill after the soil dries out (Winter et al. 1978; Winter and Ziegler 1992). In contrast, nocturnal acidification in T. portulacastrum is not restricted to the dry season. Furthermore, nocturnal CO₂ uptake in T. portulacastrum merely suffices to reduce CO₂ loss but does not lead to net dark CO₂ fixation.

Trianthema portulacastrum is only the tenth vascular C₄ or C₃-C₄ plant species recorded to also exhibit CAM cycle activity. The other C₄-with-CAM species are all Portulaca: P. australis (Winter and Holtum 2017), P. cryptopetala (Winter et al. 2019b), P. cyclophylla (Holtum et al. 2017a), P. digyna (Holtum et al. 2017b), P. grandiflora (Ku et al. 1981; Guralnick and Jackson 2001), P. molokiniensis (Winter et al. 2019b), P. oleracea (Koch and Kennedy 1980), P. pilosa (Winter and Holtum 2017) and P. umbraticola (Winter 2019). In an attempt to explain why CAM and C₄ photosynthesis were not known in a given species (with the exception of Portulaca spp.) despite sharing many similar metabolic steps and a tendency to cluster in similar regions of the angiosperm evolutionary tree (Edwards and Ogburn 2012), Sage (2002) suggested that each pathway requires a distinct leaf anatomy to function efficiently; C₄ evolution requires enhanced BS size in C₃ ancestors whereas CAM evolution requires succulence. Suggesting that the starting anatomies were mutually exclusive, Sage (2002) further postulated that the outcome of any evolutionary selection pressure may be predetermined, resulting in an incompatibility between the two types of photosynthesis that does not permit co-existence. Almost two decades later, the Sage-hypothesis holds for the most part. There are still no species known that exhibit C₄ in conjunction with the high rates of nocturnal CO₂ uptake typical of plants with pronounced CAM such as an Agave or Opuntia. In those species where CAM photosynthesis is known to co-exist with C₄, the rates of nocturnal CO₂ uptake are typically low and the two photosynthetic pathways appear to occur in anatomically distinct regions of the leaf, rather than in the same cells. The only evidence for both pathways functioning in the same cell is a report, based on an inability to immunochemically detect Rubisco in the large putatively CAM parenchyma cells, that malate is transported in the light from the CAM cells to Kranz cells in leaves of stressed P. oleracea (Lara et al. 2004). The postulated metabolic scheme did not consider the more parsimonious exchange of CO₂, rather than malate, between cell types, and it was not clear how PEP would be made available for CO₂ dark fixation in the parenchyma cells. Among C₄ and CAM researchers there is currently considerable interest in understanding traits that might favour the evolution of one photosynthetic pathway over the other (Edwards and Ogburn 2012). There is also much to be learned from localising the components that might enable the co-existence of the two photosynthetic pathways in leaves of species such as Trianthema and Portulaca (Ferrari et al. 2020).

If low-level CAM in the leaves of T. portulacastrum occurs in the layers of large parenchyma cells that lie above and below the central C₄ cell layers, one would expect to observe chloroplasts in the cells, but perhaps at low density as the degree of CAM is low. A low density of chloroplasts detected in parenchyma cells of leaves from plants used in this study supports this hypothesis (Fig. 7a). However, in several studies, leaf parenchyma cells of T. portulacastrum have been reported to lack chloroplasts: Kienholz (1926) reported that the parenchyma tissues he defined as water-storage cells were achlorophyllous; Muhaidat and McKown (2013) stated that the parenchyma is non-chlorenchymatous but showed a photograph in which the abaxial parenchyma cells clearly contained chloroplasts; and Muhaidat et al. (2007) also considered the parenchyma cells non-chlorenchymatous and showed a photograph of a leaf cross-section with very poorly developed parenchyma layer. In contrast, similar to our observations, Bohley et al. (2015) reported that the spongy parenchyma tissue contains a few chloroplasts. In all likelihood, as with the development of CAM, the development of the water-tissue-type large parenchyma cells and their chloroplast complement may depend upon growth conditions, specifically irradiance. In contrast to the observational variation reported for chloroplasts in leaf parenchyma, chloroplasts are abundant in the cortex and pith cells of stems (Fig. 7b, c).

CAM appears unevenly distributed among the Aizoaceae. It is widespread in the sub-families Mesembryanthemoideae and Ruschioideae (Smith and Winter 1996; Winter 2019) yet there is no clear evidence of CAM in the sub-families Acrosanthoideae and Azooideae. Only two species with CAM-type nocturnal acidification have been described among the ~50 species in the sub-family Sesuvioideae, in the predominately C₃ S. portulacastrum (Winter et al. 2019a) and now in the predominately C₄ T. portulacastrum. There are no reports of CAM in the C₄ members of Sesuvium, or in the only C₃ Trianthema, T. ceratosepala (Bohley et al. 2015). The other genera in the subfamily Sesuvioideae, Anisostigma and Tribulocarpus, exhibit C₃-type carbon isotope signatures. In both T. portulacastrum and S. portulacastrum, the expression of CAM features is at the very low end of the CAM spectrum in which, depending upon species and environmental conditions, the contribution of CAM to total carbon gain can vary from values of <1% to close to 100% (Winter 2019), with archetypal CAM species such as cacti gaining close to 100% of their carbon via CAM (Nobel 1988). Despite differences in their principal photosynthetic pathways, T. portulacastrum and S. portulacastrum exhibit similarities in morphology and habit. Both are prostrate annuals of the seasonally dry tropics with succulent or semi-succulent leaves that are attached to conspicuous green fleshy stems. The more salt-tolerant C₃ S. portulacastrum is a plant of coastal habitats whereas the C₄ T. portulacastrum is more an opportunistic weed of drier and disturbed landscapes. Although both are successful species in that they have pantropical distributions, it would be inappropriate to ascribe their wide distributions to ecological advantages endowed by the ability to express dual pathways of photosynthesis as both are salt-tolerant, produce prolific quantities of small seed, and in each the contribution of CAM to carbon gain is small.

The relationships between the origins of C₄ and CAM in the Sesuvioideae are uncertain. More than one interpretation is possible for the origin of C₄ (Bohley et al. 2015) and information on CAM presence or absence is as yet insufficient for hypothesis testing. A maximum likelihood optimisation of a phylogeny based upon molecular and trait characters favoured the hypothesis of a single early gain of C₄ photosynthesis in a C₃ ancestor of the Trianthema/Sesuvium/Zaleya lineage and two subsequent C₄→C₃ reversions, one in Sesuvium and one in Trianthema. An alternative scenario of C₄ evolution that excluded the possibility of C₄→C₃ reversions requires C₄ to have evolved from C₃ ancestors multiple times (at least twice in Sesuvium and at least three times in Trianthema). A situation similar to the latter alternative appears to have occurred in Portulaca, except that CAM is ancestral to C₄ and CAM continued to be expressed after C₄ had evolved (Christin et al. 2014).

Since fewer than 5% of the close to 180 species of Trianthema and Portulaca have been surveyed for CAM, and several unscreened species exhibit succulent leaves and/or fleshy stems that are similar to T. portulacastrum and P. oleracea (Hartmann et al. 2011; Bohley et al. 2015), it would be reasonable to expect that there are more species with features of CAM to be discovered within both genera. There also seems to be a strong possibility of CAM occurrence in C₄ species of the genus Sesuvium. We note that it is generally assumed, because of the absence of Kranz anatomy, that the dominant photosynthetic pathway in the stems of Portulaca and Trianthema with C₄ (or C₃-C₄ intermediate) leaf anatomy is C₃. However, clearly in some species these stems are not 100% C₃ and evidently have evolved some degree of CAM. To our knowledge, the relative contributions of C₃ and CAM to daytime and night-time carbon gain in water-stressed and non-water-stressed stem tissues has yet to be demonstrated experimentally. The presence of stomata, although at low density, suggests some CO₂ exchange with the atmosphere. Separate measurements of diel leaf- and stem-CO₂ exchange and studies of stem stomatal functioning are clearly warranted.

A spate of recent reports of CAM-type acidification in small fleshy-stemmed herbs with mildly succulent leaves such as Calandrinia (Montiaceae; Winter and Holtum 2011; Hancock et al. 2019), Coleus (Lamiaceae; Winter et al. 2020a), Portulaca (Portulacaceae; Holtum et al. 2017a; Winter 2019; Winter et al. 2019b), Sesuvium (Aizoaceae; Winter et al. 2019a) and now Trianthema, have not only increased our knowledge of CAM diversity in a numeric sense but highlight a hitherto unrecognised aspect of CAM diversity. Although CAM is well known to be associated with small and massive perennial stem-succulents (Nobel 1988; Smith and Winter 1996) it appears that annual herbs with fleshy stems may be larger contributors to CAM biodiversity than previously envisaged. Generally, when screening these plants for CAM by means of acid titrations or gas exchange, only leaves are sampled, but not stems and petioles. It is probable that more intense screening of the angiosperm phylogeny for CAM in stems and petioles, particularly in annual and bi-annual taxa, will uncover more species with CAM. Moreover, in herbs in which the expression of CAM in leaves may be facultative, i.e. absent in leaves of a well-watered plant, stem- or petiole-screening may detect a permanent CAM signal. We do not know the principal function of low-level CAM in stems and petioles. Reducing water-loss in T. portulacastrum is likely to be of minor importance as the stems have few stomata. It would be logical if CAM acted principally to reduce carbon loss by recycling respiratory CO₂.

The possible contribution of low level CAM to the overall carbon budget of the C₄ plant T. portulacastrum can be assessed by estimating nocturnal carbon loss in the absence of the temporary reduction in net CO₂ efflux, i.e. by assuming that after the initial overshoot of net CO₂ loss at the very beginning of the dark period, the rate of net CO₂ loss remained at the same level as was observed at the end of the dark period. For the branch experiment of Fig. 6, where the temporal reductions in nocturnal net CO₂ release were particularly pronounced, low level CAM contributed only up to ~2% to 24 h carbon gain (= carbon gain in the light minus carbon loss at night) in the well-watered and re-watered states. But under severe water-deficit stress (Fig. 6, day 11), when CO₂ uptake in the light was substantially reduced, the temporal reduction of nocturnal net CO₂ loss by 20% almost doubled 24-h-CO₂ gain, potentially aiding survival.

The authors declare no conflicts of interest.